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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
99ed8fcd-8fb3-47e5-b57e-167a3950ccdd | weakly-supervised-estimation-of-shadow | 1811.08164 | null | https://arxiv.org/abs/1811.08164v3 | https://arxiv.org/pdf/1811.08164v3.pdf | Weakly Supervised Estimation of Shadow Confidence Maps in Fetal Ultrasound Imaging | Detecting acoustic shadows in ultrasound images is important in many clinical and engineering applications. Real-time feedback of acoustic shadows can guide sonographers to a standardized diagnostic viewing plane with minimal artifacts and can provide additional information for other automatic image analysis algorithms. However, automatically detecting shadow regions using learning-based algorithms is challenging because pixel-wise ground truth annotation of acoustic shadows is subjective and time consuming. In this paper we propose a weakly supervised method for automatic confidence estimation of acoustic shadow regions. Our method is able to generate a dense shadow-focused confidence map. In our method, a shadow-seg module is built to learn general shadow features for shadow segmentation, based on global image-level annotations as well as a small number of coarse pixel-wise shadow annotations. A transfer function is introduced to extend the obtained binary shadow segmentation to a reference confidence map. Additionally, a confidence estimation network is proposed to learn the mapping between input images and the reference confidence maps. This network is able to predict shadow confidence maps directly from input images during inference. We use evaluation metrics such as DICE, inter-class correlation and etc. to verify the effectiveness of our method. Our method is more consistent than human annotation, and outperforms the state-of-the-art quantitatively in shadow segmentation and qualitatively in confidence estimation of shadow regions. We further demonstrate the applicability of our method by integrating shadow confidence maps into tasks such as ultrasound image classification, multi-view image fusion and automated biometric measurements. | ['Bernhard Kainz', 'Ozan Oktay', 'Nicolas Toussaint', 'Matthew Sinclair', 'Jo Schlemper', 'Alberto Gomez', 'Veronika Zimmer', 'Daniel Rueckert', 'Qingjie Meng', 'Martin Rajchl', 'Julia Schnabel', 'Jacqueline Matthew', 'Benjamin Hou', 'James Housden'] | 2018-11-20 | null | null | null | null | ['shadow-confidence-maps-in-ultrasound-imaging'] | ['medical'] | [ 6.95193350e-01 5.91000974e-01 3.01970303e-01 -8.02613258e-01
-1.15911603e+00 -5.03551245e-01 1.60133243e-01 4.97916788e-01
-3.75403941e-01 3.63825172e-01 -1.70524597e-01 -2.47033060e-01
1.94497630e-02 -5.60180247e-01 -5.65453351e-01 -8.45716834e-01
-3.70084727e-03 7.78233290e-01 9.55140829e-01 3.28939736e-01
1.01679288e-01 4.54541802e-01 -1.30653644e+00 1.82760730e-01
1.04092586e+00 1.28973675e+00 5.92521727e-01 1.06047130e+00
4.06056866e-02 2.74469346e-01 -6.92069590e-01 -1.82807267e-01
3.33421864e-02 -4.94564116e-01 -4.71423328e-01 -1.52346775e-01
5.28277040e-01 -1.74937323e-01 3.72104973e-01 9.28326011e-01
7.41558850e-01 -9.23853740e-02 8.84283483e-01 -9.03974652e-01
1.23322152e-01 5.08115232e-01 -2.89078265e-01 -1.58400357e-01
4.41417605e-01 -1.82578918e-02 6.10792935e-01 -8.06032956e-01
5.29186368e-01 7.45071054e-01 9.38371360e-01 2.41807520e-01
-9.48304892e-01 -5.98579586e-01 -3.66589218e-01 1.55486152e-01
-1.28200817e+00 -1.82117850e-01 5.95711112e-01 -4.26226974e-01
7.65937343e-02 7.00299084e-01 7.47694075e-01 4.39027995e-01
4.81546074e-01 5.85243344e-01 1.55163431e+00 -4.75332558e-01
3.03214937e-01 4.36798722e-01 -2.27521807e-01 1.37442505e+00
5.65359853e-02 -5.53575121e-02 -5.09324074e-01 -1.24932058e-01
8.13326955e-01 -2.33317390e-01 -6.41060412e-01 -4.90020126e-01
-9.24428403e-01 5.50571620e-01 3.38509500e-01 3.39555591e-01
-2.02410564e-01 2.79812157e-01 1.84454620e-01 -2.86349744e-01
4.57978785e-01 4.80005324e-01 -1.75959960e-01 3.37645411e-02
-1.35179281e+00 -3.96544009e-01 1.12213528e+00 4.31291819e-01
7.11695075e-01 -3.04986775e-01 -2.37109333e-01 6.38262331e-01
3.60532612e-01 8.93654585e-01 2.75103182e-01 -1.08824325e+00
-2.48370215e-01 2.15498954e-01 1.34217124e-02 -1.23738003e+00
-6.33187056e-01 -4.09398854e-01 -6.43510997e-01 4.69788462e-01
4.21097934e-01 1.57637581e-01 -1.19054151e+00 1.26572752e+00
3.85698199e-01 7.29905963e-01 -2.17489228e-01 1.09317970e+00
1.10274661e+00 4.66700196e-01 -3.90035301e-01 -4.26167041e-01
1.32755435e+00 -6.34845793e-01 -7.68520772e-01 5.32367416e-02
2.89090514e-01 -8.58111322e-01 1.13344312e+00 5.89782715e-01
-1.04721749e+00 -4.56210971e-01 -1.12674141e+00 4.33140606e-01
-4.18312363e-02 4.01803792e-01 2.16231689e-01 1.08492553e+00
-1.09686410e+00 5.57145238e-01 -1.07883286e+00 -2.26754490e-02
2.70355016e-01 4.89700258e-01 -1.93383142e-01 5.24511486e-02
-9.13407445e-01 9.72689748e-01 -1.28361350e-02 3.43401551e-01
-7.85587966e-01 -8.28837752e-01 -1.02838302e+00 -1.07443951e-01
6.49078369e-01 -3.57969344e-01 1.23354459e+00 -7.14212894e-01
-1.61135149e+00 7.79800177e-01 -1.17470041e-01 -3.07807744e-01
6.38534427e-01 6.64625466e-02 -1.55923396e-01 6.24440253e-01
1.53140947e-01 3.65247875e-01 9.46524858e-01 -1.75599873e+00
-6.50690377e-01 -5.35038188e-02 -2.45912850e-01 1.03411809e-01
6.85375407e-02 -3.33772898e-01 -7.58286476e-01 -4.51636046e-01
5.00882566e-01 -1.02929449e+00 -1.42646372e-01 3.94409657e-01
-2.44418755e-01 1.09587774e-01 7.93801725e-01 -7.76185393e-01
1.06233716e+00 -1.72802055e+00 -2.29366645e-01 8.37989271e-01
4.47111040e-01 5.01877442e-02 2.63569236e-01 -3.66803885e-01
4.82863724e-01 -2.47004494e-01 -7.29366004e-01 -4.83176351e-01
-2.55184084e-01 3.68556023e-01 1.10554203e-01 6.21779442e-01
-3.12510252e-01 6.59460127e-01 -1.08385789e+00 -1.00165236e+00
6.31362736e-01 3.68006945e-01 -1.64965436e-01 4.66443717e-01
7.03373030e-02 6.43887222e-01 -1.30417496e-01 7.71415412e-01
7.83370793e-01 -1.13084428e-01 2.21858218e-01 -7.16625571e-01
-3.73585261e-02 -2.00639382e-01 -1.26330173e+00 1.70113552e+00
-1.09557223e+00 8.60250354e-01 3.52671862e-01 -7.61886835e-01
7.37044215e-01 3.48410457e-01 5.17772555e-01 -3.51446629e-01
2.17556506e-01 5.01562357e-01 -6.01840168e-02 -6.39165938e-01
2.71359146e-01 -2.01823607e-01 -2.12241281e-02 5.30523419e-01
-6.72348514e-02 -9.13724124e-01 -1.92189991e-01 1.90345436e-01
9.33280647e-01 6.17980072e-03 2.09463686e-01 -3.54940563e-01
7.02272058e-01 -3.76900524e-01 4.24013436e-01 1.03276908e+00
1.60819031e-02 1.11328745e+00 2.54621834e-01 -7.55678862e-02
-6.14275098e-01 -1.16062772e+00 -4.78880852e-01 7.09118545e-01
7.16758668e-01 -1.97972924e-01 -9.59218562e-01 -1.00108337e+00
-1.78634539e-01 5.66655695e-01 -7.26758480e-01 1.26798749e-01
-6.05758250e-01 -4.82724786e-01 5.41108131e-01 5.44568479e-01
1.87830776e-01 -8.39140892e-01 -9.56936836e-01 9.08216611e-02
-3.83407325e-01 -1.12752557e+00 -4.88595784e-01 2.79994577e-01
-5.40677786e-01 -1.12979293e+00 -7.32262969e-01 -6.15375519e-01
8.91123891e-01 -1.93579733e-01 1.19008994e+00 2.56143600e-01
-6.94795489e-01 6.54913902e-01 -3.54734838e-01 -5.38215816e-01
-8.42645526e-01 -5.03172636e-01 -7.09022209e-02 5.86366514e-03
-6.38795197e-01 -3.95763457e-01 -1.02184546e+00 7.17213511e-01
-8.92673492e-01 9.28778350e-02 6.92739844e-01 8.53777766e-01
7.43055522e-01 -2.12500870e-01 2.23985925e-01 -1.19243228e+00
4.15628344e-01 1.10401593e-01 -8.12467337e-01 3.78156483e-01
-8.80736709e-01 3.32670882e-02 2.44435266e-01 -1.23601288e-01
-1.24344134e+00 2.86536932e-01 -3.57507110e-01 -2.63238370e-01
-1.32619098e-01 3.41914833e-01 2.78377295e-01 -4.86506611e-01
6.67172134e-01 8.99846852e-02 5.63953146e-02 -8.91507417e-03
2.54828185e-01 6.09397352e-01 8.74317527e-01 -3.80677104e-01
5.78379989e-01 6.55624449e-01 1.26210690e-01 -7.07779527e-01
-7.40381956e-01 -7.29162693e-01 -6.99445844e-01 -7.04464495e-01
9.03779030e-01 -3.50846052e-01 -8.32446992e-01 1.25896081e-01
-1.08793867e+00 -4.51335043e-01 -2.02799350e-01 4.88078564e-01
-5.79515398e-01 6.34030342e-01 -3.49672854e-01 -8.53337348e-01
-3.11211675e-01 -1.43710995e+00 1.42125976e+00 2.60769248e-01
-1.51870130e-02 -1.03866363e+00 1.33292824e-01 3.33492249e-01
2.20156461e-01 3.13190728e-01 4.81029928e-01 -2.93403566e-01
-5.04391849e-01 -2.22500041e-01 -2.50375092e-01 5.84463120e-01
3.35127205e-01 -1.92287505e-01 -1.29462647e+00 8.69837031e-02
1.71407629e-02 -1.73183426e-01 9.50686872e-01 9.18767989e-01
1.57939517e+00 8.52042809e-02 -4.16145712e-01 5.39238274e-01
1.26936984e+00 1.25051379e-01 3.72621715e-01 -4.78517473e-01
5.80217659e-01 5.84766626e-01 1.10239220e+00 2.01642916e-01
1.13749057e-01 6.72925472e-01 4.62767839e-01 -6.35234773e-01
-3.67475837e-01 2.26895466e-01 -1.55352116e-01 5.95785916e-01
-2.63841569e-01 -2.99911976e-01 -1.02583945e+00 2.49355510e-01
-1.71207643e+00 -3.91201079e-01 -1.82234555e-01 2.22511864e+00
9.41481471e-01 2.08331734e-01 -4.63052064e-01 1.14281960e-01
5.08210480e-01 -2.21162558e-01 -1.45545989e-01 -2.19432220e-01
2.70283133e-01 4.78521585e-01 5.56941986e-01 1.02885532e+00
-1.05512381e+00 5.96149921e-01 5.93564320e+00 1.03667545e+00
-1.17465246e+00 4.62486386e-01 6.95742428e-01 4.78767276e-01
-3.96835983e-01 -3.06606770e-01 -3.46083581e-01 3.50119799e-01
5.40085495e-01 5.90319216e-01 -2.85677463e-01 5.53890765e-01
1.46243811e-01 -9.23131883e-01 -9.29240942e-01 8.94283593e-01
4.82192844e-01 -1.54285431e+00 -5.87046206e-01 -3.15448761e-01
8.60276639e-01 -4.06381875e-01 -8.14348906e-02 -3.35079253e-01
-7.72941262e-02 -9.50444460e-01 5.08760154e-01 7.74651706e-01
1.26108682e+00 -4.09610212e-01 1.32613349e+00 6.14403933e-03
-1.14790106e+00 4.58246142e-01 6.83592409e-02 6.64606512e-01
2.52352118e-01 1.01110601e+00 -1.83605397e+00 4.48469937e-01
5.52084506e-01 -1.13947864e-03 -4.90127623e-01 1.10637689e+00
-3.83305758e-01 6.74007952e-01 -5.43159842e-01 -1.06678449e-01
4.45299447e-02 5.42107876e-03 4.04074579e-01 1.66776884e+00
2.52263278e-01 1.98399290e-01 1.32588834e-01 7.18189180e-01
2.95108259e-01 2.14756243e-02 -2.10064024e-01 6.05106950e-01
2.48194739e-01 1.67982042e+00 -1.55837893e+00 -4.64438498e-01
4.99290004e-02 1.10010862e+00 -4.50146079e-01 4.43995595e-02
-9.63893294e-01 -3.37536007e-01 -2.11801946e-01 2.67791390e-01
-6.38004169e-02 -4.07349430e-02 -4.68510389e-01 -6.14166081e-01
-1.70620859e-01 -3.34827840e-01 1.10333487e-01 -9.16909993e-01
-8.03815246e-01 6.98959470e-01 5.53001501e-02 -1.27843893e+00
-4.11469936e-01 -5.23684800e-01 -6.74376905e-01 6.33742154e-01
-1.33814418e+00 -1.06754637e+00 -9.70634878e-01 2.70522416e-01
4.95961457e-01 2.05824345e-01 8.78685236e-01 1.23839892e-01
2.34767213e-01 6.75548255e-01 -6.76163957e-02 -3.44562642e-02
8.77588987e-01 -1.68678224e+00 -2.51902878e-01 6.14717364e-01
1.22431256e-01 1.85219809e-01 9.54071581e-01 -7.47341394e-01
-9.19214845e-01 -7.03631699e-01 1.91701740e-01 -4.32463884e-01
3.73365343e-01 -3.08051229e-01 -8.37350845e-01 1.50407963e-02
-1.16326138e-01 3.85091603e-01 5.90333879e-01 -2.36887023e-01
2.12353617e-01 -3.64128858e-01 -1.29169190e+00 7.86921978e-02
5.57049394e-01 -3.21118236e-01 -8.80462304e-02 4.01754051e-01
5.33029795e-01 -1.06166673e+00 -9.21743095e-01 7.95361638e-01
9.10227060e-01 -1.10915470e+00 8.56310368e-01 4.71016824e-01
2.42107064e-01 -3.61958891e-01 1.61458462e-01 -1.16158164e+00
3.90980691e-01 -3.33997935e-01 1.74560949e-01 8.59257519e-01
5.78649938e-01 -4.42857176e-01 1.01438439e+00 2.20830441e-01
-6.18468881e-01 -8.89055252e-01 -1.11930108e+00 -2.94251770e-01
-6.84754670e-01 -8.80119145e-01 -7.22066984e-02 4.78687078e-01
-3.13088477e-01 -3.10137391e-01 -6.23720288e-02 6.67600811e-01
8.04308772e-01 3.42908621e-01 5.02572119e-01 -1.22391236e+00
-5.35145462e-01 -1.65451825e-01 -4.45593208e-01 -8.08166206e-01
-9.38275978e-02 -5.82097411e-01 9.21503425e-01 -1.60089242e+00
2.25995943e-01 -1.04662919e+00 -2.38385692e-01 2.89262652e-01
-1.21045351e-01 9.81792390e-01 -1.22597411e-01 -7.87104368e-02
-6.03593588e-01 1.57367334e-01 1.50460994e+00 7.20700324e-02
-1.33208483e-01 6.97971880e-01 -2.12026704e-02 1.11720288e+00
4.99565512e-01 -5.49385667e-01 -2.86087602e-01 1.48005903e-01
1.48252010e-01 5.80736279e-01 4.68372583e-01 -1.06661582e+00
3.69062573e-01 7.55453706e-02 3.10301244e-01 -7.18895912e-01
6.53099179e-01 -9.11180496e-01 -2.43439794e-01 6.80349767e-01
-1.79021239e-01 -6.21312022e-01 -2.41566226e-02 6.94175899e-01
-1.18332908e-01 -6.46448433e-01 8.88082504e-01 -8.42049271e-02
-3.29076290e-01 -1.44740403e-01 -3.26438457e-01 -2.41372034e-01
6.31495833e-01 -3.12397182e-01 2.16920286e-01 -7.04889894e-01
-1.09733510e+00 4.53288965e-02 1.29786864e-01 -8.98280367e-02
1.03369331e+00 -7.77508616e-01 -6.17756009e-01 2.15699017e-01
3.78318541e-02 1.16309024e-01 2.16177836e-01 1.32020974e+00
-9.07520413e-01 6.87858388e-02 2.07746312e-01 -1.32786489e+00
-1.67124617e+00 -2.44439736e-01 6.35625124e-01 -1.10356621e-01
-4.53587294e-01 1.07551157e+00 4.73757356e-01 -4.80219692e-01
3.35964441e-01 -8.72480512e-01 -5.19868806e-02 -2.07562104e-01
2.31179625e-01 2.62854487e-01 3.88006091e-01 -2.48109758e-01
-3.98443848e-01 9.15852666e-01 4.27268177e-01 -4.17466849e-01
8.76628935e-01 -1.82124585e-01 6.13214634e-02 4.69652534e-01
9.36437309e-01 3.47279251e-01 -1.24066186e+00 -4.06257510e-02
-2.74752676e-01 -5.28591335e-01 2.57402360e-01 -1.13657749e+00
-1.09163380e+00 8.91366899e-01 1.22539508e+00 1.34076789e-01
1.23612547e+00 4.48165417e-01 5.16504467e-01 9.09477770e-02
2.09504262e-01 -1.03562105e+00 4.35906619e-01 -1.60935387e-01
8.24163258e-01 -1.55956209e+00 2.59830207e-01 -9.25652206e-01
-6.36663556e-01 1.15025067e+00 3.22321743e-01 1.45470142e-01
6.87237144e-01 8.97775233e-01 2.89146692e-01 -1.19811341e-01
-1.47435237e-02 -1.23029046e-01 8.29028845e-01 6.81765854e-01
2.65316337e-01 2.01670885e-01 -3.89059007e-01 3.68505538e-01
-9.54715908e-02 -2.95735478e-01 6.58414900e-01 5.75770736e-01
-7.52089262e-01 -8.35410953e-01 -6.65786922e-01 4.68387902e-01
-3.91830385e-01 -4.49677333e-02 -3.30252647e-02 5.58130026e-01
5.40442944e-01 8.69368553e-01 4.34260368e-02 -3.25197548e-01
6.21456578e-02 -3.95904750e-01 6.73501730e-01 -6.88905478e-01
-5.34265041e-01 4.39113587e-01 4.65467684e-02 -6.35149121e-01
-4.97264922e-01 -6.05980515e-01 -1.65057862e+00 4.04839367e-01
-7.63826132e-01 2.24308416e-01 1.16928387e+00 9.50270712e-01
-3.56645823e-01 7.90000558e-01 5.56776643e-01 -9.65339720e-01
2.79898476e-02 -8.48594368e-01 -4.24695104e-01 1.26535639e-01
3.24588507e-01 -5.31349480e-01 -3.25293094e-01 2.19641894e-01] | [14.353878021240234, -2.2300291061401367] |
cfac85fc-23f3-47b2-b8f4-139bd91059e3 | multi-label-image-classification-with-1 | 2107.11626 | null | https://arxiv.org/abs/2107.11626v1 | https://arxiv.org/pdf/2107.11626v1.pdf | Multi-Label Image Classification with Contrastive Learning | Recently, as an effective way of learning latent representations, contrastive learning has been increasingly popular and successful in various domains. The success of constrastive learning in single-label classifications motivates us to leverage this learning framework to enhance distinctiveness for better performance in multi-label image classification. In this paper, we show that a direct application of contrastive learning can hardly improve in multi-label cases. Accordingly, we propose a novel framework for multi-label classification with contrastive learning in a fully supervised setting, which learns multiple representations of an image under the context of different labels. This facilities a simple yet intuitive adaption of contrastive learning into our model to boost its performance in multi-label image classification. Extensive experiments on two benchmark datasets show that the proposed framework achieves state-of-the-art performance in the comparison with the advanced methods in multi-label classification. | ['Jianfei Cai', 'Dinh Phung', 'Ethan Zhao', 'Son D. Dao'] | 2021-07-24 | null | null | null | null | ['multi-label-image-classification'] | ['computer-vision'] | [ 6.54421687e-01 -3.36272180e-01 -6.12248898e-01 -4.89669442e-01
-1.06872880e+00 -5.86192012e-01 6.62459970e-01 3.19570869e-01
-3.07282239e-01 4.60632831e-01 -2.42992446e-01 -4.03521359e-02
-1.99198455e-01 -3.03182989e-01 -3.53947461e-01 -1.03432870e+00
2.83149719e-01 1.32657677e-01 -1.40681639e-01 6.20165318e-02
3.11296582e-01 3.30833644e-01 -1.63527346e+00 6.06087565e-01
4.98325706e-01 1.05472434e+00 8.22808072e-02 3.42934072e-01
-7.52777159e-02 1.03204966e+00 -3.00866038e-01 -3.82402658e-01
1.71631411e-01 -4.63357478e-01 -1.07561910e+00 4.53364462e-01
9.31988001e-01 1.29928738e-01 3.11202351e-02 1.14413679e+00
2.76254147e-01 -9.18178633e-02 1.08620703e+00 -1.38009965e+00
-7.34189153e-01 2.70437688e-01 -1.03698778e+00 7.56020918e-02
1.09227061e-01 -4.34927166e-01 1.33000302e+00 -1.05529404e+00
3.64636958e-01 1.25219131e+00 7.61726618e-01 5.04519761e-01
-1.36212015e+00 -8.26290369e-01 3.52683008e-01 1.52135924e-01
-1.32615805e+00 -2.46359229e-01 8.44828308e-01 -3.28015774e-01
2.77176738e-01 3.37881923e-01 5.88386096e-02 1.13366461e+00
2.58569449e-01 1.14935291e+00 1.83348823e+00 -7.13940203e-01
-1.94413751e-01 3.20887059e-01 3.50069582e-01 9.14608240e-01
-6.50518686e-02 -2.33080462e-02 -3.46220523e-01 -3.14249843e-01
3.83694261e-01 4.77340668e-01 1.30849071e-02 -6.09742939e-01
-1.00413716e+00 9.95981693e-01 4.57211256e-01 3.50097150e-01
1.65715180e-02 1.63204804e-01 6.81431472e-01 4.84372169e-01
9.15063560e-01 3.78337771e-01 -3.49476039e-01 4.76651400e-01
-9.22968984e-01 9.84116569e-02 1.74711704e-01 8.12311828e-01
8.07599545e-01 -2.22138271e-01 -1.80914462e-01 1.24663568e+00
3.06772113e-01 2.29764879e-01 7.54048407e-01 -6.33134723e-01
2.53931224e-01 7.43640184e-01 -3.16158503e-01 -7.88621962e-01
-6.96204841e-01 -7.50656188e-01 -1.00022793e+00 2.62749910e-01
1.84574515e-01 2.93363899e-01 -7.30358064e-01 1.53887928e+00
2.07290590e-01 6.26936436e-01 1.26968935e-01 3.75569463e-01
7.34916747e-01 3.67506057e-01 3.36550862e-01 -4.94229257e-01
1.39437032e+00 -1.49904299e+00 -6.04589939e-01 -9.48068425e-02
1.01351035e+00 -8.13948274e-01 8.67946684e-01 3.67803842e-01
-6.00167632e-01 -6.91166937e-01 -1.06378806e+00 1.01912871e-01
-3.93516630e-01 1.82679698e-01 6.86730564e-01 6.38118207e-01
-6.11841500e-01 5.24461269e-01 -2.22219363e-01 -3.07342112e-01
6.70980811e-01 1.70331433e-01 -6.30881131e-01 -2.55193144e-01
-7.96308339e-01 7.96355665e-01 4.47500885e-01 -2.71458358e-01
-6.93875551e-01 -4.16181147e-01 -7.90566802e-01 -1.50925606e-01
5.26956856e-01 -3.03813517e-01 1.23151743e+00 -1.05656588e+00
-1.12031996e+00 1.45537031e+00 1.61296993e-01 -8.80056173e-02
3.76236141e-01 -4.12705131e-02 -4.87852037e-01 2.88458377e-01
4.34573174e-01 6.17461741e-01 1.09521222e+00 -1.62114346e+00
-9.78762686e-01 -3.62152994e-01 1.18301392e-01 2.21445337e-01
-5.14378488e-01 7.20488727e-02 8.64734575e-02 -8.43527019e-01
2.53677458e-01 -1.30191922e+00 -8.14236552e-02 -2.42743760e-01
-2.86092877e-01 -6.28421009e-01 8.16439331e-01 2.39463337e-02
1.06887817e+00 -2.19658899e+00 9.79001671e-02 -1.29802870e-02
3.34698975e-01 1.04733348e-01 -2.65020400e-01 3.82742643e-01
-4.33837622e-01 1.16452850e-01 -1.12291850e-01 -6.10963523e-01
-3.73940468e-02 7.90166184e-02 -2.22977564e-01 7.03269362e-01
1.50056943e-01 1.11039507e+00 -1.03977132e+00 -8.36272359e-01
7.23669678e-02 8.67080614e-02 -1.15130790e-01 2.03731745e-01
2.54792392e-01 5.66739738e-01 -3.19661409e-01 9.04468775e-01
6.87208474e-01 -8.02082956e-01 4.75289166e-01 -2.26475239e-01
1.09305926e-01 -3.96095872e-01 -1.02837694e+00 1.67962027e+00
-6.55908048e-01 2.99172938e-01 -2.20790222e-01 -1.24768484e+00
6.62526190e-01 4.07395273e-01 6.01245165e-01 -5.52052736e-01
1.30675092e-01 3.65832567e-01 -4.82575655e-01 -3.49535644e-01
3.97326767e-01 -4.70205933e-01 -3.16086531e-01 7.08768964e-01
2.86495000e-01 1.07625574e-01 5.92015982e-02 3.35790440e-02
5.20695448e-01 7.27447420e-02 8.62270296e-01 -3.80088717e-01
6.28621340e-01 -4.07324195e-01 3.55981231e-01 7.12867379e-01
-3.16730291e-01 4.55234021e-01 1.11631438e-01 -4.84374136e-01
-6.88795686e-01 -7.41106689e-01 -4.61532176e-01 1.78573954e+00
2.86257058e-01 -3.88866544e-01 -3.49678904e-01 -1.47107315e+00
9.45544150e-03 1.52312964e-01 -7.73106337e-01 3.57636041e-03
-5.04469395e-01 -9.65494156e-01 6.46596432e-01 4.32901025e-01
5.06555259e-01 -8.60409856e-01 -3.11680496e-01 3.62703167e-02
-5.45134321e-02 -1.01149189e+00 -3.56725454e-01 6.35069072e-01
-8.70340407e-01 -1.18133307e+00 -8.53712261e-01 -1.29926550e+00
7.58512199e-01 8.43072355e-01 9.42948818e-01 1.55922562e-01
-1.96519181e-01 4.03190196e-01 -4.51252013e-01 -7.71521255e-02
-5.93351483e-01 2.35139057e-01 -5.48372418e-02 4.27137524e-01
1.09352134e-01 -3.39182824e-01 -3.00089210e-01 5.18351793e-01
-1.20351160e+00 2.97062527e-02 6.97735667e-01 1.26750326e+00
6.16563201e-01 4.10626382e-02 9.13494170e-01 -1.49910164e+00
4.43553716e-01 -6.83175027e-01 -2.75234342e-01 6.53102398e-01
-1.10473382e+00 1.47975340e-01 6.40540302e-01 -5.41101754e-01
-8.68729472e-01 1.82335868e-01 1.10306963e-01 -3.37406963e-01
-2.23859504e-01 3.80847275e-01 1.62459195e-01 -4.64654475e-01
4.63890672e-01 1.78378224e-01 -1.81029156e-01 -6.67899072e-01
6.16751015e-01 8.80643964e-01 2.15740070e-01 -4.36085403e-01
6.19802952e-01 5.25928855e-01 4.55796242e-01 -2.28441566e-01
-1.49263871e+00 -9.55914795e-01 -9.38713670e-01 -1.65131256e-01
6.61493003e-01 -1.04944074e+00 -5.80639005e-01 6.44702256e-01
-6.59002483e-01 6.79970682e-02 1.44691318e-01 1.24383688e-01
-5.95934927e-01 5.56833446e-01 -6.60942554e-01 -4.12282407e-01
-1.32117063e-01 -1.21340084e+00 1.23548079e+00 2.56810337e-01
3.64452638e-02 -1.38594759e+00 9.02396515e-02 6.04767144e-01
6.87282309e-02 6.82640001e-02 1.04052329e+00 -9.10828829e-01
-1.20917253e-01 -2.68363208e-02 -4.27114516e-01 3.83447707e-01
4.55994934e-01 -5.14038026e-01 -1.39490747e+00 -6.98435307e-01
1.86886378e-02 -9.58764851e-01 1.12745535e+00 -9.91608053e-02
1.18361127e+00 -9.50433612e-02 -5.84909618e-01 4.80261475e-01
1.76898301e+00 -1.87086120e-01 1.02916718e-01 4.00514662e-01
1.00038195e+00 5.37601113e-01 9.40169930e-01 3.14683557e-01
3.29058409e-01 9.68443811e-01 5.22226870e-01 -3.47602427e-01
-4.87340800e-02 5.97293451e-02 -1.02155045e-01 8.53683293e-01
8.47258791e-02 4.90905829e-02 -6.53714657e-01 2.28114590e-01
-1.93809462e+00 -9.47814107e-01 4.38474258e-03 1.80899560e+00
8.59144330e-01 -9.78374258e-02 1.71444118e-01 1.41149044e-01
8.52536142e-01 4.67625916e-01 -3.41961294e-01 -2.53933758e-01
-1.55252695e-01 9.05813873e-02 4.84373659e-01 1.37018204e-01
-1.80953872e+00 8.65927041e-01 6.74037981e+00 1.24221325e+00
-1.20070755e+00 4.94905829e-01 8.07053447e-01 2.14539006e-01
5.72225973e-02 -3.35554093e-01 -1.03200030e+00 2.70757675e-01
8.28842700e-01 1.75178498e-01 -8.09197724e-02 9.32802081e-01
-4.12326902e-01 1.94825798e-01 -1.13938916e+00 1.24346805e+00
5.08458436e-01 -1.22295189e+00 1.40296612e-02 1.00356713e-01
1.22037709e+00 -3.97017032e-01 6.28806770e-01 4.94501650e-01
-3.72842886e-02 -9.94498789e-01 5.27027488e-01 7.66726583e-02
9.32258368e-01 -7.99089909e-01 6.47339046e-01 3.17860931e-01
-1.34057379e+00 -4.91535068e-01 -2.62016088e-01 2.19472367e-02
-1.27186596e-01 3.15573603e-01 -7.64486492e-01 8.39697361e-01
1.50524929e-01 1.02079082e+00 -1.09016454e+00 7.85622954e-01
-1.71988364e-02 4.05649692e-01 4.59408075e-01 2.78393745e-01
5.28479099e-01 8.27024505e-03 -1.51121775e-02 1.41972351e+00
-6.31279871e-02 -3.39492291e-01 6.71661198e-01 1.17406458e-01
-3.58915031e-01 5.76893568e-01 -6.78890526e-01 2.52633572e-01
2.39429563e-01 1.64908504e+00 -9.69940722e-01 -4.04668570e-01
-8.00247908e-01 1.04379117e+00 6.95375741e-01 1.81726992e-01
-8.25924277e-01 9.09787714e-02 2.47356117e-01 -3.39784741e-01
2.30292886e-01 2.15272114e-01 -8.59109759e-02 -1.06947017e+00
-2.58322507e-01 -1.05340815e+00 8.23893070e-01 -3.15413773e-01
-1.64824176e+00 5.39800406e-01 3.63022052e-02 -1.63028729e+00
-2.51192540e-01 -6.98410988e-01 -6.70305565e-02 3.56209815e-01
-2.04878902e+00 -1.78504133e+00 -1.15452960e-01 5.29942214e-01
7.36643314e-01 -4.84812617e-01 1.16235399e+00 4.52567786e-01
-3.39395374e-01 9.41989243e-01 6.13374949e-01 -6.34447113e-02
1.18906581e+00 -1.40838492e+00 -1.40716642e-01 3.91476572e-01
6.05816841e-01 3.23286176e-01 1.71334550e-01 -2.13607147e-01
-7.88481355e-01 -1.30068278e+00 7.17167258e-01 -3.64420384e-01
5.05714417e-01 -2.24740833e-01 -7.51690745e-01 7.23442018e-01
2.59804338e-01 3.53990674e-01 1.24455559e+00 2.30866864e-01
-9.03324783e-01 -1.72431380e-01 -1.08652020e+00 3.12112629e-01
7.58320212e-01 -8.81549001e-01 -3.33624005e-01 5.76705098e-01
5.14577568e-01 -6.63531721e-02 -9.98297691e-01 5.33037245e-01
6.47389472e-01 -7.95984447e-01 1.05101287e+00 -5.90962112e-01
3.53370756e-01 -8.28447118e-02 -3.28843266e-01 -1.32033408e+00
-6.32874846e-01 -1.96059495e-01 8.67550150e-02 1.23651743e+00
7.58230090e-02 -5.39255023e-01 4.58000779e-01 -1.71090260e-01
-1.51583506e-02 -1.01883137e+00 -7.45590627e-01 -7.88689017e-01
3.66484910e-01 1.18144728e-01 2.32344463e-01 1.43522441e+00
-4.53959107e-02 5.91735959e-01 -7.75393367e-01 3.24396715e-02
7.93084502e-01 5.83821058e-01 3.55161667e-01 -1.54847431e+00
-2.42618755e-01 -3.92819613e-01 -4.48046237e-01 -1.10547996e+00
7.82646000e-01 -1.32328212e+00 -1.66498974e-01 -1.02835941e+00
8.55450153e-01 -7.12498367e-01 -1.01588905e+00 4.96698529e-01
-6.15807116e-01 9.28724110e-01 3.57637137e-01 5.49698114e-01
-1.15228438e+00 1.40904441e-01 1.26340199e+00 -4.93853837e-01
3.45752537e-01 1.04765721e-01 -8.42236638e-01 6.44083619e-01
6.14034057e-01 -7.78078258e-01 -3.68931085e-01 7.40489662e-02
8.84023309e-02 -1.74214840e-01 1.40381828e-01 -7.86329448e-01
-8.49231780e-02 -1.81543201e-01 2.02194855e-01 -3.74890774e-01
2.89026201e-01 -7.98462510e-01 -6.48214668e-02 3.47762108e-01
-7.42944777e-01 8.27362482e-03 -1.52531385e-01 7.13681698e-01
-4.23601091e-01 -5.79115510e-01 1.06397033e+00 -2.24886522e-01
-9.03511107e-01 2.70431399e-01 -7.26954117e-02 -5.16643673e-02
1.06017625e+00 4.53423150e-02 -3.21931481e-01 4.01734971e-02
-6.43740535e-01 -1.39203072e-02 4.32799011e-01 5.11735380e-01
3.88140827e-01 -1.67337227e+00 -6.93703830e-01 1.37727305e-01
6.94286883e-01 -6.84453011e-01 2.52796352e-01 7.65182495e-01
3.02941240e-02 4.17754769e-01 -2.73451626e-01 -7.51996338e-01
-1.82488465e+00 8.40802312e-01 1.42310604e-01 -7.44550228e-01
-3.67402405e-01 8.14344347e-01 6.63803220e-01 -4.88397539e-01
8.09723139e-02 1.27331182e-01 -6.44156992e-01 5.72379351e-01
5.72002947e-01 3.31620336e-01 -2.20312048e-02 -9.03023899e-01
-3.10048431e-01 1.02097499e+00 -6.15049362e-01 2.78548151e-01
1.09930348e+00 -4.42081243e-01 -1.00871131e-01 9.39657986e-01
1.77537847e+00 -9.61973295e-02 -1.03778303e+00 -5.20570040e-01
2.97096074e-01 -4.19798255e-01 -6.77302405e-02 -5.67349851e-01
-9.09076631e-01 7.60478437e-01 8.71469200e-01 3.94507587e-01
1.11286139e+00 5.80372848e-02 4.92472202e-01 2.99629599e-01
3.82759362e-01 -1.00445390e+00 7.04680741e-01 3.02494437e-01
4.81683791e-01 -1.81957221e+00 2.18771040e-01 -4.91039246e-01
-6.46064520e-01 1.20266449e+00 3.63512844e-01 3.06928214e-02
7.16793656e-01 -2.39774033e-01 4.38812137e-01 -2.56037176e-01
-6.19713366e-01 -1.78539336e-01 4.15264249e-01 1.99320883e-01
5.62224209e-01 1.59517303e-01 -3.21726263e-01 1.09122843e-01
5.96698046e-01 -4.29475069e-01 2.79407471e-01 1.02573204e+00
-4.69845116e-01 -1.49087095e+00 -2.97895789e-01 2.54960507e-01
-8.78892899e-01 -1.79888792e-02 -8.32785293e-02 6.69253707e-01
2.14505062e-01 1.04875445e+00 -3.26640218e-01 -2.66366333e-01
-6.73903972e-02 3.51127297e-01 7.20892072e-01 -8.59273493e-01
-6.97238088e-01 1.57648399e-01 -1.16455548e-01 -2.75176704e-01
-1.01842356e+00 -5.51646352e-01 -9.60697234e-01 1.81381136e-01
-6.03485107e-01 -4.73606773e-03 4.72311467e-01 1.07234728e+00
5.10646217e-02 4.75741565e-01 1.09712255e+00 -6.15568936e-01
-7.70934105e-01 -9.07715559e-01 -9.02429461e-01 8.31414819e-01
4.02524889e-01 -9.90446270e-01 -3.96468610e-01 -3.03886011e-02] | [9.643478393554688, 4.1897735595703125] |
73b66015-c60a-4a31-ace3-67d23ef3413c | hybrid-y-net-architecture-for-singing-voice | 2303.02599 | null | https://arxiv.org/abs/2303.02599v1 | https://arxiv.org/pdf/2303.02599v1.pdf | Hybrid Y-Net Architecture for Singing Voice Separation | This research paper presents a novel deep learning-based neural network architecture, named Y-Net, for achieving music source separation. The proposed architecture performs end-to-end hybrid source separation by extracting features from both spectrogram and waveform domains. Inspired by the U-Net architecture, Y-Net predicts a spectrogram mask to separate vocal sources from a mixture signal. Our results demonstrate the effectiveness of the proposed architecture for music source separation with fewer parameters. Overall, our work presents a promising approach for improving the accuracy and efficiency of music source separation. | ['Pantaleon Perera', 'Janaka Wijayakulasooriya', 'Udula Ranasinghe', 'Pamudu Ranasinghe', 'Rashen Fernando'] | 2023-03-05 | null | null | null | null | ['music-source-separation'] | ['music'] | [ 1.40293986e-01 -3.77088487e-01 2.25713849e-02 5.03318645e-02
-1.13291872e+00 -5.12307942e-01 1.17503591e-02 -4.43633884e-01
1.78160086e-01 3.63141507e-01 4.09795344e-01 -2.09094696e-02
-4.06843483e-01 -3.11767161e-01 -3.83621693e-01 -5.48287928e-01
-1.52111396e-01 -4.01503623e-01 -5.64259529e-01 4.99588512e-02
1.03473663e-01 3.90959799e-01 -1.37781584e+00 3.70688766e-01
6.99672401e-01 1.13069868e+00 -3.20254564e-02 1.18390334e+00
2.61673518e-03 7.39012420e-01 -9.09441769e-01 -3.28183658e-02
4.12719041e-01 -1.11280060e+00 -1.39016137e-01 -2.12559938e-01
4.36974406e-01 -4.35765713e-01 -4.31375653e-01 1.06356299e+00
1.09739542e+00 4.02204335e-01 7.62400508e-01 -1.21740806e+00
-6.31163776e-01 1.39797318e+00 -3.87227535e-01 2.38843903e-01
1.65639579e-01 -1.77248135e-01 1.14086616e+00 -9.01022434e-01
-1.46606863e-01 1.03134489e+00 1.05902159e+00 2.03546941e-01
-9.29877758e-01 -1.23939753e+00 -3.91245753e-01 2.81654269e-01
-1.29785454e+00 -8.73519599e-01 1.45976365e+00 -2.98961937e-01
1.04471576e+00 2.94140577e-01 3.80345881e-01 1.17066920e+00
-9.71255824e-02 1.11403954e+00 6.59673035e-01 -6.31262720e-01
1.93149477e-01 -3.71337235e-01 1.28135264e-01 2.93208897e-01
-1.97894961e-01 3.26551348e-01 -9.64507341e-01 -4.35921669e-01
9.69483674e-01 -2.59042680e-01 -2.67353237e-01 3.08977932e-01
-1.07360697e+00 4.86591995e-01 2.93819249e-01 6.15380824e-01
-5.29802859e-01 4.97740835e-01 4.87360120e-01 3.22158337e-01
3.11384916e-01 5.87145150e-01 -3.54754359e-01 -4.19354111e-01
-1.58365262e+00 2.65725702e-01 7.49289453e-01 6.86497569e-01
-2.21483037e-01 1.17752790e+00 -2.22562298e-01 1.09709918e+00
2.95567214e-01 6.11401021e-01 9.28735018e-01 -1.16940856e+00
5.25907017e-02 -2.00854883e-01 -3.70501354e-02 -1.00606096e+00
-3.85064125e-01 -1.20114493e+00 -9.59540129e-01 3.61903518e-01
2.03450978e-01 -4.98806596e-01 -6.32889748e-01 1.77875328e+00
-1.30921900e-01 1.03240728e+00 3.95252138e-01 9.99152660e-01
1.14881372e+00 8.06647897e-01 -4.20678705e-01 -1.43065020e-01
8.51518929e-01 -1.32268727e+00 -1.06490791e+00 2.30914593e-01
-3.36119801e-01 -1.03985584e+00 5.98337114e-01 6.92269981e-01
-1.45635331e+00 -1.07994974e+00 -1.17660546e+00 1.43004730e-01
1.65876180e-01 7.31239796e-01 2.47184619e-01 8.59805107e-01
-8.67596924e-01 9.82136190e-01 -6.65221870e-01 2.40284756e-01
1.86388060e-01 2.37117976e-01 1.50024235e-01 8.30450654e-01
-1.10748589e+00 1.42435327e-01 3.56620103e-01 2.45428160e-01
-1.06036091e+00 -6.57843530e-01 -6.20749891e-01 7.79234767e-01
-1.37837812e-01 -5.45153320e-01 1.56920123e+00 -1.11304128e+00
-2.07495093e+00 2.88070855e-03 -2.00707048e-01 -7.49591947e-01
6.59938529e-02 -7.00555861e-01 -9.54483151e-01 1.58260524e-01
-1.24770932e-01 2.05669180e-01 1.35379577e+00 -1.12299693e+00
-4.75479871e-01 4.23451290e-02 -6.62830114e-01 1.25543326e-01
-2.70233870e-01 2.35999033e-01 1.21870480e-01 -1.46545053e+00
1.88160807e-01 -5.66261172e-01 1.04060248e-01 -6.63436174e-01
-5.48781514e-01 -1.67573076e-02 6.41750455e-01 -8.67075741e-01
1.28833818e+00 -2.46287203e+00 -1.34456456e-02 1.18525833e-01
1.99341446e-01 4.61220682e-01 -4.93969828e-01 4.06679302e-01
-4.30575848e-01 -2.94223547e-01 -2.22809508e-01 -4.22619730e-01
3.21163356e-01 -5.37940800e-01 -9.41397190e-01 1.56717032e-01
1.37612307e-02 7.94047892e-01 -6.40646577e-01 1.74382627e-01
1.34237275e-01 7.23420441e-01 -5.83437622e-01 4.00619954e-01
3.88432667e-02 3.64734203e-01 1.80947572e-01 5.67705810e-01
7.01955557e-01 3.31384867e-01 5.97270066e-03 -1.26966938e-01
-4.15671580e-02 5.20590425e-01 -1.49991977e+00 1.92865431e+00
-4.15603697e-01 9.69051123e-01 5.24519622e-01 -6.21929705e-01
1.26866043e+00 9.26304042e-01 6.09063983e-01 -2.08749115e-01
6.35328233e-01 5.16033709e-01 3.43975216e-01 -3.77093226e-01
3.11439246e-01 -2.53605157e-01 2.28175119e-01 5.29809237e-01
6.87963963e-01 -6.19975887e-02 -1.71201274e-01 -4.42068130e-01
8.19790542e-01 2.14833543e-01 2.45269850e-01 1.24390759e-01
4.00703847e-01 -5.82435191e-01 5.53352058e-01 8.55409145e-01
-3.85047555e-01 6.46843851e-01 -7.19733313e-02 4.27747294e-02
-5.88996708e-01 -1.31781149e+00 1.01834476e-01 1.23379660e+00
-2.42983714e-01 -4.64698941e-01 -8.47123802e-01 -1.08028039e-01
-6.96721971e-02 7.64014721e-01 -1.45883307e-01 -2.01609582e-01
-6.72722340e-01 -3.10804754e-01 1.37989295e+00 6.83270335e-01
2.55402803e-01 -1.23792958e+00 -3.28652471e-01 5.76571465e-01
-3.49685162e-01 -7.29918420e-01 -5.94368577e-01 3.44213754e-01
-8.09955359e-01 -7.42661238e-01 -1.00669229e+00 -8.46747518e-01
-3.07782084e-01 2.86510766e-01 6.88747227e-01 -6.70236111e-01
6.63381815e-02 1.38233081e-01 -3.31737071e-01 -8.65073800e-01
-3.62658620e-01 -8.70866403e-02 4.75890011e-01 2.53478885e-01
2.88719863e-01 -1.27375412e+00 -5.14107227e-01 -7.74328336e-02
-6.66936815e-01 -5.11202455e-01 4.96137321e-01 4.76717770e-01
2.79368281e-01 5.88522851e-01 1.19836700e+00 -6.35679141e-02
1.28171027e+00 -5.12385249e-01 -2.62353629e-01 -2.59954780e-01
-2.25140586e-01 -3.71031612e-01 9.47477341e-01 -5.79680026e-01
-1.00420523e+00 -7.25871474e-02 -4.47439313e-01 -9.54742432e-01
-4.07540053e-01 4.10956949e-01 -5.08382767e-02 2.57931501e-01
6.46077871e-01 2.91550159e-01 -3.64545941e-01 -9.45708513e-01
5.06000638e-01 9.82685804e-01 1.03305340e+00 -2.41882533e-01
5.65428495e-01 1.86998069e-01 -4.21364784e-01 -8.52354169e-01
-6.35647178e-01 -5.69282591e-01 -6.46006286e-01 -9.55764353e-02
6.47243381e-01 -9.89848852e-01 -7.66838789e-01 7.04666436e-01
-1.32059348e+00 -6.21718690e-02 -4.01693553e-01 1.08604407e+00
-6.17833793e-01 1.95422508e-02 -7.24807143e-01 -1.21910107e+00
-7.66452253e-01 -8.13009858e-01 9.03269470e-01 2.00518221e-01
-4.04893756e-01 -7.53829777e-01 5.35356283e-01 -2.13588104e-02
5.57647645e-01 -3.51591073e-02 4.68646348e-01 -8.37271810e-01
-1.73826531e-01 -8.32876191e-02 8.49112570e-02 6.97094321e-01
2.75339037e-01 -9.24600214e-02 -1.60327935e+00 -1.11945443e-01
4.61682230e-01 -7.49220476e-02 9.18925166e-01 8.67645264e-01
1.04174387e+00 -2.66187280e-01 2.33870268e-01 9.72565949e-01
1.01230013e+00 6.28286660e-01 2.80243188e-01 -1.11012548e-01
5.97117960e-01 9.46525037e-02 -7.05227479e-02 4.73386675e-01
-2.39098489e-01 3.59149635e-01 1.54492244e-01 -2.46552110e-01
-5.22215068e-01 -2.02314004e-01 5.17712057e-01 1.36101413e+00
7.92102665e-02 -2.20406562e-01 -4.92563874e-01 6.01540923e-01
-1.66208279e+00 -1.24427104e+00 -2.76812259e-02 1.72809529e+00
6.33024275e-01 -1.85199112e-01 4.11516070e-01 8.42642784e-01
6.54736519e-01 2.65401363e-01 -6.50572956e-01 -2.91835248e-01
-5.71337529e-02 6.12619281e-01 -8.17490667e-02 3.74653041e-01
-1.17270374e+00 5.74833333e-01 7.69073248e+00 1.05425727e+00
-1.38207161e+00 2.27488488e-01 -1.78885937e-01 -4.89940256e-01
4.50043380e-02 -5.91086447e-01 -2.31238499e-01 2.61398822e-01
1.21230066e+00 -1.59869254e-01 7.03587711e-01 8.40953410e-01
2.62371153e-01 6.29402757e-01 -9.79356647e-01 1.40857708e+00
3.19562405e-01 -1.15421128e+00 -1.86653033e-01 -3.74951750e-01
5.16523063e-01 6.43727779e-02 2.75663644e-01 3.72982323e-01
1.30321860e-01 -1.11227298e+00 1.05590975e+00 3.94649446e-01
7.12125838e-01 -1.04177225e+00 4.43649590e-01 3.66909206e-01
-1.30024016e+00 -3.41471016e-01 -1.66222975e-01 -2.87058085e-01
2.22931966e-01 5.80501497e-01 -9.63570058e-01 6.33226097e-01
5.38831055e-01 7.68806756e-01 1.03840046e-01 1.30553222e+00
8.61692009e-04 1.28103960e+00 -3.11318189e-02 4.14406031e-01
1.13267332e-01 -1.24952361e-01 1.05757558e+00 1.65507162e+00
7.52117515e-01 -3.41134332e-02 -2.93423068e-02 1.08666623e+00
-2.44755894e-01 -2.31655352e-02 -4.67531979e-01 -2.75355190e-01
5.54403722e-01 1.15052426e+00 -5.47036052e-01 -1.15300998e-01
6.95938319e-02 7.41993368e-01 -2.25372553e-01 5.80425858e-01
-8.23878646e-01 -9.64158893e-01 7.26673305e-01 -4.42810565e-01
4.58289653e-01 -3.60869318e-01 -6.05231047e-01 -9.85989153e-01
-2.97990471e-01 -1.04202306e+00 9.24325213e-02 -9.23487067e-01
-1.26568794e+00 9.00944173e-01 -4.50314224e-01 -1.61323667e+00
-4.86853361e-01 -5.22229075e-01 -9.94003594e-01 1.07646537e+00
-1.34623289e+00 -8.14658523e-01 1.25408724e-01 6.35560453e-01
5.55100501e-01 -8.29084694e-01 1.06494105e+00 4.30207372e-01
-4.05126959e-01 6.16751134e-01 5.86259604e-01 3.55462462e-01
7.05516517e-01 -1.21559131e+00 5.33203959e-01 9.68462944e-01
8.70739281e-01 6.59821510e-01 5.56933105e-01 -3.30029905e-01
-9.98960555e-01 -1.08527851e+00 5.56080639e-01 -1.22215882e-01
4.92175639e-01 -2.07725421e-01 -7.52739549e-01 3.87418389e-01
7.38444149e-01 -4.53766763e-01 1.36259890e+00 1.62374631e-01
-4.68874067e-01 -2.11655289e-01 -8.57107639e-01 3.63358259e-01
7.04018652e-01 -8.18116665e-01 -9.65928555e-01 -3.00841331e-01
7.75133431e-01 -2.96362728e-01 -5.65745354e-01 2.95705289e-01
9.06477392e-01 -9.92985368e-01 1.08951807e+00 -4.77678210e-01
5.25853217e-01 -2.65837342e-01 -2.94102579e-01 -1.76589990e+00
-7.50657201e-01 -1.07584572e+00 -6.39142036e-01 1.34763706e+00
1.83974758e-01 -1.38375074e-01 3.58872652e-01 -3.75346959e-01
-3.05278957e-01 -2.26462692e-01 -8.17347646e-01 -9.94226217e-01
-3.89991850e-02 -9.51208055e-01 8.35727572e-01 9.82773244e-01
-8.76939595e-02 3.84618700e-01 -7.03380942e-01 2.58534789e-01
5.79237223e-01 3.59909326e-01 5.73205054e-01 -1.11396646e+00
-7.94843853e-01 -9.90268767e-01 6.60466496e-03 -8.40872288e-01
3.88113409e-01 -1.05334151e+00 1.44571155e-01 -1.33771539e+00
-4.51465011e-01 1.80587143e-01 -1.07377434e+00 6.42460436e-02
-2.70497482e-02 5.11491776e-01 5.25706351e-01 1.42467067e-01
-1.32148042e-01 7.75922477e-01 8.32850397e-01 -1.69103429e-01
-6.73611045e-01 4.08436745e-01 -6.46100104e-01 1.09714186e+00
8.24150264e-01 -5.26930034e-01 -4.66245621e-01 -5.04271626e-01
-3.48284721e-01 4.43317503e-01 2.97454298e-01 -1.41193616e+00
1.52404204e-01 2.46122330e-01 4.53127861e-01 -8.15732241e-01
6.44384861e-01 -8.24565053e-01 -1.05304699e-02 2.92545915e-01
-6.24571979e-01 -4.62475270e-01 4.43710029e-01 4.01810199e-01
-5.88668287e-01 -1.43489718e-01 5.80574811e-01 2.02461928e-01
-2.39249058e-02 -9.12146196e-02 -3.42275530e-01 -2.44054809e-01
3.93175185e-01 6.72123432e-02 2.54355581e-03 -6.65278375e-01
-9.83537197e-01 -5.19955754e-01 -6.42467737e-01 4.96498019e-01
7.72368371e-01 -1.68980217e+00 -1.00750148e+00 5.48312783e-01
-4.44491476e-01 -6.40477180e-01 3.12643588e-01 5.89339137e-01
-1.26876593e-01 4.57788825e-01 -4.60822910e-01 -2.19482660e-01
-1.18964887e+00 2.90327609e-01 4.95756418e-01 1.55852973e-01
-6.09804034e-01 1.27397227e+00 5.54253869e-02 -4.37545240e-01
6.94953084e-01 -4.04878676e-01 -3.49409461e-01 4.86951359e-02
8.01658034e-01 9.07844484e-01 -4.09925431e-02 -5.11685133e-01
-2.16127172e-01 2.84698725e-01 5.59182405e-01 -5.86695433e-01
1.39656484e+00 3.44347470e-02 -1.60331801e-01 7.45357752e-01
1.05523705e+00 6.70155823e-01 -9.55837011e-01 -2.40045950e-01
-2.56672055e-01 -2.71035194e-01 2.76399583e-01 -1.12243521e+00
-1.02274644e+00 1.07485044e+00 6.33109272e-01 4.77674007e-01
1.51279974e+00 -5.56765020e-01 1.24062943e+00 3.48907888e-01
-2.37368673e-01 -8.57274354e-01 -3.77487112e-03 7.18083978e-01
1.09012842e+00 -8.08285594e-01 -6.29336238e-01 2.70505905e-01
-2.72515267e-01 1.25329220e+00 2.57505804e-01 -6.53379142e-01
8.52612257e-01 5.67847371e-01 4.83226329e-01 1.50524214e-01
-3.44051152e-01 -1.76119611e-01 7.85665035e-01 5.46847761e-01
7.72777915e-01 1.30123332e-01 1.86765626e-01 1.38451195e+00
-8.29722226e-01 1.46080911e-01 3.24899822e-01 3.06206793e-01
-4.90780950e-01 -9.87725616e-01 -1.07662070e+00 2.09962472e-01
-8.78578961e-01 -3.03348631e-01 -7.28322744e-01 6.92223161e-02
1.64474711e-01 1.32218778e+00 -1.32372215e-01 -5.84409356e-01
2.46061787e-01 5.24162650e-01 3.92507583e-01 -3.50351065e-01
-8.87896299e-01 9.94140506e-01 -2.57602513e-01 -2.53927320e-01
-4.50865954e-01 -2.02886701e-01 -1.17139328e+00 9.40567032e-02
-2.47918099e-01 2.84913540e-01 6.87517166e-01 6.66750014e-01
5.65147042e-01 1.40069246e+00 7.56827354e-01 -1.17700040e+00
-5.74510932e-01 -1.30996776e+00 -1.00009787e+00 2.67986790e-04
9.97121811e-01 -1.83667153e-01 -1.59737334e-01 2.22466961e-01] | [15.525400161743164, 5.5669755935668945] |
f9fc5753-7876-4ff7-b512-ad7797300ef8 | monte-carlo-siamese-policy-on-actor-for | 2004.03879 | null | https://arxiv.org/abs/2004.03879v1 | https://arxiv.org/pdf/2004.03879v1.pdf | Monte-Carlo Siamese Policy on Actor for Satellite Image Super Resolution | In the past few years supervised and adversarial learning have been widely adopted in various complex computer vision tasks. It seems natural to wonder whether another branch of artificial intelligence, commonly known as Reinforcement Learning (RL) can benefit such complex vision tasks. In this study, we explore the plausible usage of RL in super resolution of remote sensing imagery. Guided by recent advances in super resolution, we propose a theoretical framework that leverages the benefits of supervised and reinforcement learning. We argue that a straightforward implementation of RL is not adequate to address ill-posed super resolution as the action variables are not fully known. To tackle this issue, we propose to parameterize action variables by matrices, and train our policy network using Monte-Carlo sampling. We study the implications of parametric action space in a model-free environment from theoretical and empirical perspective. Furthermore, we analyze the quantitative and qualitative results on both remote sensing and non-remote sensing datasets. Based on our experiments, we report considerable improvement over state-of-the-art methods by encapsulating supervised models in a reinforcement learning framework. | ['Saumyaa Shah', 'S Manthira Moorthi', 'Debajyoti Dhar', 'Litu Rout'] | 2020-04-08 | null | null | null | null | ['satellite-image-super-resolution'] | ['computer-vision'] | [ 6.81355596e-01 3.79519731e-01 -1.08542189e-01 -1.25293300e-01
-8.70912492e-01 -4.52218562e-01 9.07342911e-01 -4.15646076e-01
-4.64141905e-01 1.16909420e+00 3.02750230e-01 -2.45850235e-01
-5.08006275e-01 -9.92789567e-01 -5.94977558e-01 -8.37308288e-01
-1.38165697e-01 1.04352303e-01 -1.27112061e-01 -4.42148596e-01
1.12745136e-01 6.58951163e-01 -1.44302666e+00 1.81713235e-02
1.13558114e+00 5.65621257e-01 2.30221912e-01 4.74931151e-01
3.53738308e-01 1.00340652e+00 -2.91784823e-01 -8.72356221e-02
6.08166337e-01 -4.54458863e-01 -8.37882459e-01 1.39056787e-01
2.23785654e-01 -5.24509430e-01 -2.14964405e-01 1.20160162e+00
5.69343865e-01 2.44795978e-01 6.09385073e-01 -7.74374485e-01
-6.10867798e-01 7.26609886e-01 -7.11843431e-01 3.22834879e-01
-9.35222115e-03 5.07663906e-01 1.09203362e+00 -4.79530305e-01
7.65818596e-01 1.33239317e+00 6.76706135e-01 5.39450765e-01
-1.38130796e+00 -6.62747383e-01 2.99839407e-01 1.52556241e-01
-1.06951237e+00 -3.02548587e-01 7.71223485e-01 -5.06682694e-01
5.12284100e-01 1.18841112e-01 5.71770966e-01 1.19379342e+00
1.38472384e-02 5.09690762e-01 1.82017303e+00 -5.27593851e-01
5.18140495e-01 -3.87981869e-02 -4.10318017e-01 3.83361042e-01
5.88993244e-02 9.26245272e-01 -2.69338280e-01 -1.89518258e-01
1.18273497e+00 -2.73465604e-01 -1.45420671e-01 -2.91454703e-01
-1.05879915e+00 1.36522520e+00 7.02254772e-01 -9.85538587e-03
-6.28900707e-01 1.50752440e-01 2.32150611e-02 1.54624194e-01
7.08626390e-01 7.72803843e-01 -2.15405464e-01 3.67330998e-01
-1.00449026e+00 3.29968631e-01 4.74468201e-01 2.10247904e-01
7.80260384e-01 3.74900073e-01 1.27143845e-01 6.82010829e-01
2.65167415e-01 5.79669118e-01 -3.00350729e-02 -1.46460927e+00
3.05213258e-02 5.61914481e-02 4.53352183e-01 -6.56497240e-01
-4.14269567e-01 -5.28108120e-01 -1.10989654e+00 8.33579183e-01
3.02630305e-01 -5.45689404e-01 -6.75341964e-01 1.79501891e+00
4.61323738e-01 4.79917228e-01 3.53678703e-01 1.10923481e+00
3.48505557e-01 5.87667763e-01 2.19345838e-01 -5.20740569e-01
8.80806983e-01 -7.08274186e-01 -5.72339177e-01 -3.78369778e-01
4.58767451e-03 -2.25059614e-01 9.64742541e-01 3.99322361e-01
-8.66245985e-01 -4.97814059e-01 -8.67132902e-01 3.84588331e-01
-1.33952945e-01 -1.77547291e-01 7.87144005e-01 4.71167833e-01
-1.00143385e+00 8.69877875e-01 -9.83184695e-01 -4.13260341e-01
6.04808867e-01 9.84993856e-03 3.99902510e-03 1.55279189e-01
-1.42771733e+00 1.03299630e+00 1.77651867e-01 6.39706254e-02
-1.08752191e+00 -6.35302901e-01 -6.74839556e-01 -2.33403757e-01
7.00884581e-01 -8.02614808e-01 1.08718872e+00 -1.15657580e+00
-1.62003362e+00 5.53615510e-01 2.77034640e-01 -8.03636134e-01
7.80930221e-01 -2.78024524e-01 -1.43516645e-01 3.79930973e-01
1.36505321e-01 6.49816632e-01 1.17233491e+00 -1.56976545e+00
-5.45350969e-01 -4.43375558e-01 5.10368586e-01 3.28240484e-01
2.44636253e-01 3.07618491e-02 5.32294214e-01 -6.44295037e-01
-7.05289245e-02 -8.98910046e-01 -8.17955971e-01 8.13235268e-02
-3.79695669e-02 1.85114473e-01 3.75491858e-01 -5.45274794e-01
7.35925019e-01 -1.83781374e+00 3.51694435e-01 -6.58682808e-02
1.68651029e-01 2.90616244e-01 -9.32575837e-02 3.81533444e-01
1.06100604e-01 3.39164317e-01 -5.21248639e-01 1.09210745e-01
-1.90203935e-01 4.77918625e-01 -7.46500671e-01 5.07947624e-01
4.22094792e-01 9.40444291e-01 -1.09667945e+00 -3.25400561e-01
3.27142924e-01 6.56839907e-01 -4.32551056e-01 1.16146885e-01
-4.29376632e-01 1.10318327e+00 -9.17848945e-01 5.12770355e-01
6.13488495e-01 -3.15026015e-01 3.17444235e-01 3.73289697e-02
-2.08095640e-01 -5.49816974e-02 -1.19886017e+00 1.44627917e+00
-3.01034719e-01 3.48945796e-01 2.66745239e-01 -1.30667293e+00
6.51416719e-01 6.30593598e-02 4.46930796e-01 -7.74735928e-01
-2.82192767e-01 -1.07488573e-01 -1.98658854e-01 -6.39169931e-01
2.99379379e-01 -6.07232332e-01 2.26635173e-01 4.83337909e-01
-3.21425200e-01 -3.77715379e-01 -2.65511543e-01 -1.05792254e-01
9.66806948e-01 5.89708626e-01 5.67876339e-01 -1.51560366e-01
4.11402285e-01 2.30390146e-01 4.75962728e-01 1.14318478e+00
-2.56350577e-01 2.94260502e-01 2.14928210e-01 -4.08010870e-01
-1.07688630e+00 -1.08434141e+00 -1.87375918e-01 1.06758225e+00
-7.17196018e-02 1.47329763e-01 -7.48778999e-01 -4.01141524e-01
-7.69366249e-02 5.73910117e-01 -7.71068156e-01 8.29944164e-02
-5.00053644e-01 -1.04290044e+00 4.64932948e-01 3.43315929e-01
7.15211689e-01 -1.29043853e+00 -1.03057480e+00 2.51386166e-01
-5.95614780e-03 -1.19700158e+00 5.05008459e-01 -5.84834255e-02
-9.58549917e-01 -9.39675450e-01 -5.29416740e-01 -6.38984293e-02
1.71942487e-01 3.97111177e-01 1.14687288e+00 -1.96264952e-01
-2.32639432e-01 5.67983627e-01 -5.47370791e-01 -2.51944602e-01
-6.37975216e-01 3.35266478e-02 5.49824014e-02 -3.25958394e-02
-1.87856600e-01 -9.18096483e-01 -6.65430903e-01 4.80163470e-02
-1.07808936e+00 4.49483208e-02 7.34932423e-01 9.73521888e-01
6.02977037e-01 1.04424253e-01 7.72775412e-01 -1.02202308e+00
3.70596588e-01 -5.95411301e-01 -7.38379061e-01 1.69987783e-01
-6.17257118e-01 2.33089849e-01 5.94130635e-01 -4.11934853e-01
-1.29950523e+00 3.57558168e-02 2.69501079e-02 -2.18821362e-01
-3.51322979e-01 6.67183220e-01 1.76061898e-01 -2.07919210e-01
1.00753319e+00 1.55045956e-01 1.55676872e-01 -2.71929651e-01
7.75220513e-01 5.28872311e-01 4.06164378e-01 -5.83090961e-01
1.25494373e+00 1.07892156e+00 1.57632589e-01 -1.05111599e+00
-1.35296166e+00 4.94608320e-02 -4.94445592e-01 -2.67828763e-01
1.02512908e+00 -1.06757927e+00 -3.99899662e-01 2.40592688e-01
-7.34693527e-01 -6.95207536e-01 -5.66198409e-01 3.87797266e-01
-9.64929044e-01 2.77193338e-01 -2.39015326e-01 -1.13769996e+00
-1.86126605e-01 -9.55664515e-01 9.67163026e-01 3.61900449e-01
2.00373843e-01 -9.61701810e-01 1.96275443e-01 5.45523286e-01
6.57495677e-01 6.75478756e-01 5.44560373e-01 -7.47972503e-02
-7.83985794e-01 5.79393744e-01 -2.31768087e-01 2.87295222e-01
6.48939982e-02 -1.09786458e-01 -1.10465991e+00 -3.21696162e-01
1.80781856e-01 -6.36378586e-01 1.05438447e+00 7.12212145e-01
1.01446187e+00 -3.89767110e-01 9.50232148e-02 7.86416471e-01
1.72124791e+00 -2.14290559e-01 7.04262376e-01 6.82856798e-01
4.59738076e-01 8.48529398e-01 8.56722116e-01 6.08629525e-01
3.72085810e-01 4.21036184e-01 7.66898215e-01 -1.84777051e-01
1.23928107e-01 -2.08304510e-01 1.31483093e-01 -4.14840691e-02
-5.46185851e-01 2.52050519e-01 -8.18441272e-01 2.42583454e-01
-1.86676598e+00 -1.35926580e+00 3.98081005e-01 1.91876304e+00
9.13517177e-01 -2.79141143e-02 1.07973106e-01 -1.42657802e-01
4.85559493e-01 6.00407779e-01 -7.86013722e-01 1.24608949e-01
-1.98346287e-01 3.02374482e-01 6.30126536e-01 6.38093412e-01
-1.22126186e+00 1.16647661e+00 6.82470703e+00 6.19884074e-01
-1.21392226e+00 1.31468967e-01 6.71534896e-01 5.51838465e-02
-3.34432125e-01 1.63316593e-01 -5.07490039e-01 2.95696594e-02
9.01461840e-01 9.75538045e-02 9.11398113e-01 4.81648296e-01
7.34359145e-01 -2.86892444e-01 -7.26307452e-01 5.09021282e-01
-2.16741368e-01 -1.43339360e+00 -3.13683972e-02 1.53272703e-01
9.14311886e-01 3.04966360e-01 2.26497993e-01 2.84876436e-01
7.17682183e-01 -1.44240415e+00 3.58618498e-01 6.71182752e-01
7.46896446e-01 -5.91248512e-01 3.67555439e-01 6.46592379e-01
-7.90432096e-01 -2.52065659e-01 -3.80878389e-01 -3.26903135e-01
-6.22376837e-02 2.35383317e-01 -4.52140063e-01 6.57332599e-01
6.77600682e-01 8.15810800e-01 -3.68665308e-01 4.71891612e-01
-3.88403028e-01 8.63816917e-01 -2.15355396e-01 4.89675403e-01
3.88634861e-01 -5.34354866e-01 6.50221944e-01 7.47824192e-01
-7.84633309e-02 4.65803266e-01 2.20137984e-01 1.22816753e+00
3.64771068e-01 -2.82426476e-01 -8.02679420e-01 5.93070276e-02
4.32985306e-01 1.23384964e+00 -5.25266826e-01 6.45167101e-03
-4.14544225e-01 5.24848998e-01 3.93720418e-01 6.88320398e-01
-8.10505271e-01 3.81813616e-01 5.88179171e-01 -2.03707758e-02
4.01586920e-01 -2.45401233e-01 -3.08450967e-01 -1.25730765e+00
-1.99721232e-01 -1.19039941e+00 1.78035259e-01 -8.65988851e-01
-1.41990411e+00 3.48740757e-01 2.32227102e-01 -1.19194341e+00
-4.90725189e-01 -2.95960575e-01 -4.36591983e-01 6.96488917e-01
-2.21019864e+00 -1.30321467e+00 -2.88911462e-01 4.50466990e-01
4.08419311e-01 -1.10425919e-01 7.50923276e-01 -4.35148627e-01
-3.06160033e-01 -3.64955189e-03 3.30442071e-01 -1.81954931e-02
3.53434533e-01 -1.22285855e+00 3.23514432e-01 9.64339435e-01
8.83959010e-02 2.91512519e-01 9.90028977e-01 -5.28652966e-01
-1.25367308e+00 -1.13395238e+00 1.13013588e-01 -2.35393688e-01
8.70666921e-01 1.22924589e-01 -9.77232754e-01 6.19509101e-01
1.25909567e-01 3.63491148e-01 3.55122566e-01 -2.59313017e-01
-3.08412999e-01 -2.62821745e-02 -1.40254080e+00 6.10091925e-01
1.03000414e+00 -4.80753988e-01 -6.13361597e-01 2.47304291e-01
6.94670618e-01 -9.27352458e-02 -9.67875361e-01 5.43342769e-01
3.30244839e-01 -1.16060221e+00 1.28764284e+00 -6.69787049e-01
7.44997323e-01 -2.08288759e-01 -4.22287554e-01 -1.44456446e+00
-3.72530699e-01 -6.28434300e-01 -1.09721072e-01 7.79249489e-01
9.61703807e-02 -7.51873076e-01 6.52354956e-01 2.76483417e-01
4.18753117e-01 -5.77332139e-01 -9.55709398e-01 -6.78285182e-01
6.46254063e-01 -1.61631674e-01 2.96568006e-01 1.00320315e+00
-5.45441747e-01 2.84005642e-01 -5.10109723e-01 5.68500042e-01
1.17795026e+00 1.81507632e-01 6.10631406e-01 -1.28483176e+00
-6.47712290e-01 -2.32769459e-01 8.85403305e-02 -6.44452870e-01
4.51930612e-01 -4.77584541e-01 -9.89731699e-02 -1.37588954e+00
2.26321414e-01 -4.32875454e-01 -2.30660886e-01 3.39838117e-01
-2.55119026e-01 2.10911021e-01 2.19858572e-01 3.59252721e-01
-4.00501281e-01 6.98339045e-01 1.34094834e+00 -8.49137828e-02
-8.29780251e-02 -5.91987073e-02 -8.38752508e-01 7.97860742e-01
9.66693640e-01 -2.43890360e-01 -3.89390767e-01 -3.20667565e-01
3.58094782e-01 3.53577375e-01 7.54309654e-01 -7.14948058e-01
-2.15657920e-01 -8.06927145e-01 1.77253067e-01 -9.81116742e-02
2.72409141e-01 -6.97045267e-01 -9.33944341e-03 3.56276751e-01
-5.90466499e-01 -5.84998846e-01 -8.74760468e-03 7.31788039e-01
8.19403380e-02 -1.66332617e-01 1.13184214e+00 -5.68189025e-01
-6.60542428e-01 1.69895619e-01 -3.11337560e-01 2.09505886e-01
9.24445987e-01 5.29599655e-03 -3.19494814e-01 -5.43391943e-01
-7.83692956e-01 1.78715467e-01 4.77362990e-01 3.43219340e-02
4.44283485e-01 -9.09733832e-01 -1.12813556e+00 1.95986778e-02
-1.40473962e-01 3.82830645e-03 2.86565691e-01 5.66522777e-01
-2.88982809e-01 -4.85361591e-02 -3.09774995e-01 -4.23968643e-01
-7.95007050e-01 5.51998436e-01 6.58659041e-01 -4.70472336e-01
-7.75364459e-01 3.05058241e-01 9.11204815e-02 -5.77748477e-01
-1.00464739e-01 2.85267539e-04 -4.07866478e-01 -2.30378397e-02
5.40781617e-01 4.60894555e-01 -5.22134066e-01 -4.78220910e-01
-5.83976656e-02 5.98981678e-01 1.65769458e-01 -5.66319227e-01
1.73286867e+00 -4.24021035e-01 1.54313102e-01 2.82680392e-01
3.62632930e-01 -1.11295499e-01 -1.97010314e+00 -4.29262489e-01
-1.21526003e-01 -3.19339901e-01 3.76395643e-01 -8.98602426e-01
-1.01010072e+00 7.50500321e-01 6.03082836e-01 3.74691308e-01
1.24441910e+00 -2.94573624e-02 1.93521515e-01 3.89785916e-01
5.44900715e-01 -9.77186143e-01 -2.72501819e-02 4.35148507e-01
9.38487709e-01 -1.63893306e+00 2.77495176e-01 -2.15831026e-01
-8.25678051e-01 7.59601176e-01 3.22666138e-01 -5.22829354e-01
5.93949616e-01 1.27823219e-01 1.16536751e-01 -3.47965979e-03
-6.25467420e-01 -5.72437048e-01 -7.55032524e-02 8.96294296e-01
3.20337638e-02 1.47078648e-01 -6.37899637e-02 1.81214958e-01
-6.60153404e-02 1.28267229e-01 6.83395445e-01 9.10039544e-01
-6.10799134e-01 -9.65098441e-01 -5.15954792e-01 2.70207852e-01
-5.59243858e-01 -8.58466849e-02 -1.50343716e-01 8.00490141e-01
-9.77468863e-02 1.02337003e+00 -2.64248848e-01 -6.84480742e-02
-1.69140488e-01 -3.12683463e-01 5.06388664e-01 -5.57971239e-01
-4.05322194e-01 8.25182348e-02 -1.17087305e-01 -5.55309057e-01
-1.12871397e+00 -7.84057200e-01 -1.02131009e+00 -5.69809563e-02
-1.40223727e-01 -2.04946041e-01 4.06247705e-01 1.19975960e+00
1.58412740e-01 4.20490056e-01 8.32362413e-01 -1.14033604e+00
-1.30811453e+00 -9.42586303e-01 -5.66153049e-01 7.76628554e-02
6.15965068e-01 -7.49383688e-01 -6.01367414e-01 -7.83797652e-02] | [10.22626781463623, -1.7780508995056152] |
60d8c679-e9c7-434c-b9fb-a8ba1d304904 | self-supervised-log-parsing | 2003.07905 | null | https://arxiv.org/abs/2003.07905v1 | https://arxiv.org/pdf/2003.07905v1.pdf | Self-Supervised Log Parsing | Logs are extensively used during the development and maintenance of software systems. They collect runtime events and allow tracking of code execution, which enables a variety of critical tasks such as troubleshooting and fault detection. However, large-scale software systems generate massive volumes of semi-structured log records, posing a major challenge for automated analysis. Parsing semi-structured records with free-form text log messages into structured templates is the first and crucial step that enables further analysis. Existing approaches rely on log-specific heuristics or manual rule extraction. These are often specialized in parsing certain log types, and thus, limit performance scores and generalization. We propose a novel parsing technique called NuLog that utilizes a self-supervised learning model and formulates the parsing task as masked language modeling (MLM). In the process of parsing, the model extracts summarizations from the logs in the form of a vector embedding. This allows the coupling of the MLM as pre-training with a downstream anomaly detection task. We evaluate the parsing performance of NuLog on 10 real-world log datasets and compare the results with 12 parsing techniques. The results show that NuLog outperforms existing methods in parsing accuracy with an average of 99% and achieves the lowest edit distance to the ground truth templates. Additionally, two case studies are conducted to demonstrate the ability of the approach for log-based anomaly detection in both supervised and unsupervised scenario. The results show that NuLog can be successfully used to support troubleshooting tasks. The implementation is available at https://github.com/nulog/nulog. | ['Sasho Nedelkoski', 'Jorge Cardoso', 'Jasmin Bogatinovski', 'Odej Kao', 'Alexander Acker'] | 2020-03-17 | null | null | null | null | ['log-parsing'] | ['computer-code'] | [ 7.65588656e-02 -8.52448121e-02 -1.85101137e-01 -3.59473377e-01
-6.53636694e-01 -4.23538834e-01 3.65515441e-01 8.85461986e-01
-1.03041045e-01 2.35475674e-01 -3.82125899e-02 -6.77990258e-01
4.38873321e-02 -7.44018674e-01 -5.44879735e-01 -3.22839022e-01
-3.97092462e-01 3.29266518e-01 5.24143457e-01 1.13723397e-01
4.81810749e-01 3.28230113e-01 -1.55388212e+00 3.09067339e-01
7.77206182e-01 9.87925231e-01 5.19503094e-02 7.79414892e-01
-4.92534220e-01 7.17376292e-01 -7.28840947e-01 -3.09124291e-02
3.06797117e-01 -2.87524730e-01 -6.80345476e-01 1.43281966e-01
2.03315113e-02 -2.85456836e-01 -6.01838492e-02 9.55920160e-01
1.34784609e-01 -8.62125158e-02 2.56322891e-01 -1.35224688e+00
-1.01312652e-01 6.94198310e-01 -6.59495473e-01 6.02398753e-01
5.79145610e-01 -1.02540240e-01 1.00088930e+00 -6.65582240e-01
2.74236590e-01 7.88230062e-01 5.37130177e-01 1.91814601e-01
-1.04589880e+00 -4.26250815e-01 -8.03903267e-02 1.91353664e-01
-1.00871456e+00 -3.18767816e-01 7.73085952e-01 -6.42763197e-01
1.38224876e+00 1.30891353e-01 6.02808669e-02 6.81093216e-01
5.82515955e-01 5.71392179e-01 6.58514917e-01 -5.69626868e-01
4.31837022e-01 1.81453794e-01 7.20732093e-01 7.83588827e-01
4.19841558e-01 -2.04813212e-01 -6.64356709e-01 -6.87443018e-01
1.82386965e-01 1.96178377e-01 8.48062057e-03 -7.73805454e-02
-8.10944855e-01 6.25855982e-01 -3.12907994e-01 3.19478244e-01
-4.22161937e-01 -1.67346358e-01 8.42198014e-01 6.44703925e-01
3.43619078e-01 2.71145463e-01 -4.59057212e-01 -4.55114126e-01
-1.06483817e+00 -1.50553852e-01 1.07504928e+00 8.08243632e-01
7.78130710e-01 2.36429989e-01 8.55199397e-02 5.88985801e-01
5.53426921e-01 4.59161140e-02 7.51077116e-01 -3.54840368e-01
5.34317970e-01 1.11864328e+00 -1.96960971e-01 -8.38579834e-01
-3.10259312e-01 -3.01520735e-01 -5.55194139e-01 1.70928746e-01
1.05181523e-01 1.60154104e-01 -6.53663993e-01 1.28048587e+00
3.80724639e-01 2.47052893e-01 -2.04201534e-01 2.15848461e-01
1.31084710e-01 6.37851059e-01 -3.83431986e-02 -4.16573673e-01
1.22046173e+00 -6.71262085e-01 -6.95966959e-01 -3.26596081e-01
8.09589624e-01 -8.17402184e-01 1.17272532e+00 5.42221367e-01
-6.17026687e-01 -2.52773345e-01 -1.29573202e+00 5.56743801e-01
-2.63874143e-01 1.68116808e-01 3.94192159e-01 5.92408836e-01
-8.03394258e-01 6.21008396e-01 -1.56953478e+00 -3.72538596e-01
2.58164167e-01 3.13839376e-01 -4.08608586e-01 2.98237324e-01
-5.74098527e-01 3.60815644e-01 6.21685922e-01 -2.72359629e-03
-8.12764883e-01 -3.81237209e-01 -9.14150953e-01 8.91821831e-02
5.08920252e-01 1.83612302e-01 1.26551926e+00 -4.06928778e-01
-1.11711490e+00 6.50324702e-01 -2.94265777e-01 -6.22997522e-01
2.69113272e-01 -5.76427937e-01 -5.79565465e-01 -1.77616760e-01
1.67048857e-01 -5.79261720e-01 8.53412747e-01 -7.60244191e-01
-5.96159339e-01 -3.04683506e-01 -2.88234532e-01 -4.96611059e-01
-5.89030206e-01 3.69254738e-01 -3.58715177e-01 -4.18237358e-01
1.04747966e-01 -4.94371325e-01 9.88850296e-02 -7.16718256e-01
-6.32712781e-01 -2.59323597e-01 1.03395796e+00 -8.99585247e-01
1.97266698e+00 -2.24667501e+00 -2.43490010e-01 3.00615847e-01
1.55296683e-01 2.60835767e-01 3.23421746e-01 8.27840567e-01
-1.80962309e-01 6.80325329e-02 -5.73069930e-01 -5.96718371e-01
-4.67194244e-02 2.35503122e-01 -4.70235825e-01 5.01745105e-01
3.36860955e-01 5.77602863e-01 -6.76651835e-01 -5.21802723e-01
1.74092799e-01 -1.54168874e-01 -4.80524033e-01 7.12607741e-01
-2.29010716e-01 1.26288071e-01 -3.77636552e-01 9.27894235e-01
2.10788324e-01 -1.99755490e-01 7.12669343e-02 2.02907398e-01
-1.49302796e-01 3.70122224e-01 -1.14801621e+00 1.36437106e+00
-4.61572081e-01 5.79902709e-01 -2.49575660e-01 -1.11901200e+00
8.72981727e-01 2.70778149e-01 2.74585783e-01 -3.01675469e-01
8.43242928e-02 1.44136772e-01 -1.84909850e-02 -6.60355747e-01
2.78939545e-01 3.64776671e-01 -2.63952881e-01 9.21848118e-01
1.55845322e-02 4.85101521e-01 3.13213170e-01 3.16665947e-01
1.77362978e+00 6.62954524e-02 7.40301192e-01 1.34615406e-01
6.62226260e-01 6.85495064e-02 7.69201159e-01 5.52660167e-01
-1.02373853e-01 2.99274266e-01 8.47723782e-01 -2.27711752e-01
-8.11428905e-01 -1.02316797e+00 7.43755177e-02 1.00815678e+00
-3.35680008e-01 -7.13051438e-01 -6.49248898e-01 -1.09614539e+00
-1.34645244e-02 8.47318530e-01 -4.21283364e-01 -2.72473991e-01
-6.75476551e-01 -9.96581674e-01 4.53554064e-01 6.61276639e-01
8.57530087e-02 -1.34359717e+00 -8.58114302e-01 4.31727380e-01
2.03171089e-01 -1.08680439e+00 -3.11736614e-01 4.40585911e-01
-1.10315812e+00 -1.36202729e+00 4.65068698e-01 -5.00335574e-01
7.54174948e-01 -1.80205226e-01 1.04118717e+00 3.04268539e-01
-6.71569943e-01 2.45505899e-01 -5.88436186e-01 -2.44086087e-01
-9.62505817e-01 -3.54586504e-02 1.89510137e-01 1.10009558e-01
7.26728261e-01 -8.80450666e-01 -1.41287863e-01 1.39530137e-01
-1.13246548e+00 -5.10426998e-01 8.01772714e-01 8.17801535e-01
5.81592917e-01 1.99037641e-01 6.81518555e-01 -1.31163514e+00
7.99259663e-01 -8.71777058e-01 -9.99903738e-01 1.42674476e-01
-1.00505054e+00 2.82386720e-01 1.00684893e+00 -3.77698660e-01
-8.42475057e-01 5.11878505e-02 -2.10117534e-01 -3.29056859e-01
-4.40743029e-01 6.44710302e-01 -1.84850022e-01 3.00535113e-01
6.06005311e-01 3.49957645e-01 -1.70772359e-01 -7.86880553e-01
-2.39711791e-01 9.78855729e-01 5.46748996e-01 -4.54945445e-01
9.50450420e-01 2.21317559e-01 -3.20416212e-01 -8.17048848e-01
-5.01004279e-01 -8.09664190e-01 -5.26324689e-01 1.05385125e-01
5.11999428e-01 -3.53880435e-01 -3.40597659e-01 4.53803480e-01
-9.66594458e-01 -1.49630368e-01 -1.00981936e-01 1.93444073e-01
-3.05198342e-01 7.12762535e-01 -7.85911143e-01 -9.37863529e-01
-5.79603255e-01 -8.99251401e-01 9.58915293e-01 9.74077880e-02
-4.21482593e-01 -9.07003939e-01 1.88518882e-01 1.51586384e-01
3.29594702e-01 2.66615033e-01 1.22198915e+00 -1.44991302e+00
-6.54773533e-01 -8.13372910e-01 1.55172378e-01 5.63466132e-01
3.02487135e-01 5.99999279e-02 -9.72122848e-01 -2.96847403e-01
2.43814677e-01 -2.81809624e-02 3.27586085e-01 -2.05566436e-01
1.00575280e+00 -2.45882213e-01 -1.74536526e-01 4.22227383e-01
1.46276391e+00 3.60816032e-01 3.75597507e-01 4.54620659e-01
5.92356920e-01 2.21400395e-01 8.34971964e-01 7.55217850e-01
-2.06755623e-02 3.26207489e-01 4.57753688e-01 5.94887316e-01
3.62948954e-01 -2.86004305e-01 7.44907856e-01 1.32103169e+00
4.70982015e-01 -1.26543730e-01 -1.26286530e+00 7.06617415e-01
-1.82661176e+00 -6.52478516e-01 -1.32121399e-01 2.50890851e+00
6.92101359e-01 5.41850030e-01 -7.99575448e-03 5.89240313e-01
4.71791357e-01 -9.18553174e-02 -5.07941425e-01 -5.31936586e-01
4.37916785e-01 3.50993961e-01 2.40877301e-01 3.74016583e-01
-1.04297268e+00 6.74378872e-01 5.03822327e+00 5.23285627e-01
-9.25255716e-01 2.93940723e-01 1.07013777e-01 2.35248521e-01
1.81562603e-01 2.15573058e-01 -9.16464627e-01 6.19313836e-01
1.47211885e+00 -3.53340507e-01 1.48180708e-01 1.10698164e+00
3.05159658e-01 -3.01250905e-01 -1.37264514e+00 6.94923699e-01
-7.31000900e-02 -9.93708193e-01 -2.44646803e-01 1.07842065e-01
2.18669817e-01 9.15626585e-02 -5.73756814e-01 2.35876545e-01
5.02860025e-02 -8.19077671e-01 2.42710143e-01 1.27353162e-01
4.10715163e-01 -5.24846256e-01 1.07012379e+00 5.17789543e-01
-1.22685599e+00 -1.57167524e-01 -2.49277130e-02 -5.46257980e-02
2.70320594e-01 8.04166973e-01 -1.09614038e+00 6.05847120e-01
6.90155506e-01 5.84785461e-01 -6.53786242e-01 1.02988017e+00
-1.44272879e-01 1.20299184e+00 -4.51539993e-01 1.90968946e-01
-2.50522017e-01 -3.29948291e-02 5.79876423e-01 1.18528724e+00
1.24767281e-01 -5.80549479e-01 3.02597433e-01 7.44586766e-01
9.51208919e-02 2.82435775e-01 -4.98966396e-01 -4.01321083e-01
7.42183685e-01 1.08273292e+00 -8.52688789e-01 -1.19515926e-01
-7.06710219e-01 8.73717785e-01 2.47221917e-01 2.49975938e-02
-7.62767792e-01 -6.42870426e-01 5.69486499e-01 2.83948034e-01
1.00686140e-01 -3.00647050e-01 -1.64940029e-01 -1.10135233e+00
5.98072767e-01 -9.16875482e-01 4.96035397e-01 -1.63701419e-02
-1.09425998e+00 7.47954190e-01 -7.18229078e-03 -1.31143081e+00
-6.02074683e-01 -5.18892646e-01 -1.13061941e+00 5.25191784e-01
-1.08148408e+00 -5.17153144e-01 -3.69817883e-01 2.75566190e-01
7.26217985e-01 -2.83611745e-01 8.09391022e-01 4.11157966e-01
-1.01620460e+00 7.05075920e-01 -3.30302417e-02 2.34927103e-01
6.17863715e-01 -1.37134230e+00 6.29847109e-01 1.45831621e+00
1.50010929e-01 7.01235533e-01 7.44340956e-01 -8.28145146e-01
-1.49079132e+00 -1.20540214e+00 7.26389945e-01 -3.88936728e-01
1.01970589e+00 -4.88811761e-01 -1.38326705e+00 8.78851175e-01
-6.47595450e-02 2.01537902e-03 8.91131341e-01 -1.49444774e-01
-1.82395414e-01 -1.98291205e-02 -9.79382813e-01 3.49597558e-02
6.13573968e-01 -5.27973652e-01 -7.00364769e-01 2.41974264e-01
4.80702609e-01 -2.57864565e-01 -8.65868926e-01 1.04064189e-01
1.54080629e-01 -1.08950830e+00 3.99999380e-01 -5.97711802e-01
3.37767124e-01 -3.77139598e-01 -1.47312447e-01 -8.15950215e-01
2.77821660e-01 -7.48076499e-01 -6.98480964e-01 1.46835434e+00
3.86351973e-01 -8.12147200e-01 7.75250316e-01 3.66135061e-01
-1.18346088e-01 -7.53308117e-01 -6.72836423e-01 -7.56567061e-01
-6.69301927e-01 -7.31700122e-01 2.70047903e-01 8.25232148e-01
1.43169001e-01 1.46453246e-01 -1.57183275e-01 5.82349718e-01
5.93290150e-01 4.11381796e-02 7.52440333e-01 -1.32179487e+00
-7.05597579e-01 -3.46412137e-02 -7.68021107e-01 -5.68028927e-01
1.68633148e-01 -7.47682810e-01 1.08060941e-01 -1.11609364e+00
1.02161039e-02 -3.15907121e-01 -3.52084875e-01 6.96710348e-01
-1.95177361e-01 -3.03957343e-01 -3.14076751e-01 2.99627393e-01
-4.74181652e-01 2.26644933e-01 -9.92237926e-02 2.75873303e-01
-5.44332564e-01 3.58480006e-01 -3.01378936e-01 6.95390701e-01
1.04800439e+00 -7.81278193e-01 -4.47523862e-01 -5.40867373e-02
2.52938747e-01 -1.18757151e-02 4.47482206e-02 -1.04500532e+00
1.84791043e-01 1.24526717e-01 -1.25865340e-01 -5.42688310e-01
-2.53166348e-01 -6.87396765e-01 -8.52999985e-02 3.33564878e-01
6.86207414e-02 5.92643499e-01 4.11695361e-01 8.33765566e-01
-5.39182186e-01 -4.52515364e-01 4.58899826e-01 1.46697879e-01
-7.83517957e-01 1.00601904e-01 -4.11059201e-01 1.42756090e-01
1.18205690e+00 -1.56351060e-01 -1.64190575e-01 -7.55973533e-02
-6.14294827e-01 2.91705161e-01 5.55777669e-01 5.10102212e-01
6.20425463e-01 -8.23290527e-01 -3.34121823e-01 6.20806515e-01
2.95519471e-01 6.43041506e-02 -1.94501176e-01 1.04375970e+00
-6.34754658e-01 1.82609797e-01 -8.79354868e-03 -6.78851068e-01
-1.47551608e+00 4.95346785e-01 4.69142459e-02 -5.26832402e-01
-7.85122037e-01 4.39293236e-01 -2.11668536e-01 -1.88583180e-01
2.83358157e-01 -3.05526108e-01 -1.10888183e-02 -1.67588413e-01
6.09142184e-01 4.16231066e-01 5.32914281e-01 -1.51114002e-01
-5.19335687e-01 1.65086046e-01 -4.04052198e-01 1.72970504e-01
1.46142781e+00 9.24133584e-02 -4.95189458e-01 7.91065276e-01
1.01708925e+00 3.04047436e-01 -7.52288163e-01 -2.84667403e-01
8.52776110e-01 -3.22157353e-01 -1.51418403e-01 -4.23800975e-01
-7.65646815e-01 8.30479681e-01 4.36264664e-01 3.97242725e-01
1.32217920e+00 4.45361696e-02 7.31219530e-01 4.11805153e-01
3.88529271e-01 -8.90958488e-01 -6.41412213e-02 3.18747073e-01
3.12949300e-01 -1.22709858e+00 -1.69752374e-01 -3.50580066e-01
-2.40749434e-01 1.06931412e+00 7.40751147e-01 -9.35225934e-02
6.87935770e-01 6.75513685e-01 9.23046768e-02 -1.10536933e-01
-9.41412389e-01 3.07199121e-01 5.55489250e-02 2.76631802e-01
5.66906154e-01 -1.45689428e-01 -2.55854666e-01 8.01022887e-01
1.62660982e-02 -1.83143929e-01 7.03405261e-01 1.61758792e+00
-5.46459615e-01 -1.44838655e+00 -2.53195316e-01 8.57129157e-01
-7.24216104e-01 -2.19204184e-02 -3.37951571e-01 6.11720741e-01
-2.40321174e-01 9.46498454e-01 -9.15686786e-02 -6.25057220e-01
2.87031680e-01 7.18098700e-01 -1.80605814e-01 -9.83588636e-01
-5.56627631e-01 -1.26827762e-01 -5.63199399e-04 -9.45976019e-01
2.20210627e-01 -6.68045700e-01 -1.36614370e+00 -3.25055942e-02
-4.18462992e-01 2.99158633e-01 5.48751354e-01 9.19497728e-01
5.31420112e-01 7.34158337e-01 6.66426599e-01 -3.19408596e-01
-7.77401865e-01 -1.01765418e+00 -2.72139311e-01 5.28957605e-01
3.63067448e-01 -4.94114369e-01 -6.21534169e-01 2.03318283e-01] | [7.4693803787231445, 2.7137584686279297] |
b583a500-6bc7-42a6-ac05-7551315e769e | duck-rumour-detection-on-social-media-by-1 | null | null | https://aclanthology.org/2022.naacl-main.364 | https://aclanthology.org/2022.naacl-main.364.pdf | DUCK: Rumour Detection on Social Media by Modelling User and Comment Propagation Networks | Social media rumours, a form of misinformation, can mislead the public and cause significant economic and social disruption. Motivated by the observation that the user network — which captures \textit{who} engage with a story — and the comment network — which captures \textit{how} they react to it — provide complementary signals for rumour detection, in this paper, we propose DUCK (rumour \underline{d}etection with \underline{u}ser and \underline{c}omment networ\underline{k}s) for rumour detection on social media. We study how to leverage transformers and graph attention networks to jointly model the contents and structure of social media conversations, as well as the network of users who engaged in these conversations. Over four widely used benchmark rumour datasets in English and Chinese, we show that DUCK produces superior performance for detecting rumours, creating a new state-of-the-art. Source code for DUCK is available at: https://github.com/l tian678/DUCK-code. | ['Jey Han Lau', 'Xiuzhen Zhang', 'Lin Tian'] | null | null | null | null | naacl-2022-7 | ['rumour-detection'] | ['natural-language-processing'] | [-3.06876779e-01 4.21661586e-01 -4.75719839e-01 6.54444769e-02
-2.07414657e-01 -5.75244486e-01 8.81906450e-01 5.05257845e-01
2.50317603e-01 6.91074967e-01 7.42010057e-01 -3.55233669e-01
5.36614992e-02 -8.38047981e-01 -4.53312784e-01 -1.99808568e-01
-3.51831138e-01 3.78934294e-01 1.57361820e-01 -5.16626954e-01
4.48506385e-01 1.29174262e-01 -8.42757642e-01 3.58847737e-01
6.18861973e-01 8.06434751e-01 -2.19688430e-01 7.14926362e-01
-4.87617013e-04 1.88050175e+00 -7.43408918e-01 -6.10472083e-01
-3.02246064e-01 -8.23239923e-01 -1.03112376e+00 2.31852874e-01
1.16557442e-01 -5.78492165e-01 -1.00811625e+00 1.15807688e+00
1.44751698e-01 9.83901024e-02 8.21399033e-01 -1.37639880e+00
-1.08243477e+00 1.12329638e+00 -9.23423529e-01 8.67062390e-01
4.85140085e-01 -1.42879203e-01 1.30613792e+00 -7.89213598e-01
7.25929737e-01 1.21346343e+00 7.63449907e-01 2.21145466e-01
-8.52446020e-01 -8.66806507e-01 -3.63287963e-02 1.57213241e-01
-1.11016560e+00 -4.99761492e-01 1.01097775e+00 -5.13948321e-01
4.53983992e-01 5.80801547e-01 4.89752948e-01 1.37271750e+00
1.54866785e-01 1.07814837e+00 9.32462156e-01 2.10907087e-01
-3.07236075e-01 1.78887799e-01 4.78954941e-01 8.52152467e-01
9.32106450e-02 -1.76455945e-01 -7.30411589e-01 -7.28603721e-01
6.99860752e-01 3.40136796e-01 -4.39744234e-01 5.27164936e-01
-1.08188963e+00 1.23632836e+00 7.09341109e-01 1.10583045e-01
-5.85115135e-01 1.04151785e-01 5.37297964e-01 5.92147946e-01
1.08751440e+00 5.54377317e-01 3.47176611e-01 -4.72249016e-02
-9.09885466e-01 5.39409742e-02 1.16641724e+00 8.09401870e-01
5.43577492e-01 -1.05633333e-01 -3.82285938e-02 1.17413831e+00
1.16456695e-01 3.63763392e-01 9.18759704e-02 -7.71001041e-01
4.45939124e-01 5.61859667e-01 2.68047124e-01 -1.70060182e+00
-7.01396823e-01 -4.60534096e-01 -1.14846873e+00 -5.19335389e-01
4.02072251e-01 -3.99634004e-01 -1.60894409e-01 1.24972260e+00
2.36134529e-01 3.81001115e-01 -3.86431962e-01 9.24907684e-01
1.08180559e+00 7.79201984e-01 -5.46348155e-01 -2.37128109e-01
1.34844363e+00 -1.18674433e+00 -8.06895018e-01 -1.21191315e-01
3.64611626e-01 -8.43095541e-01 5.65363348e-01 2.53350019e-01
-1.22679603e+00 1.70666337e-01 -8.18544328e-01 4.51929197e-02
-1.07091412e-01 -5.11677444e-01 3.77737910e-01 3.28479737e-01
-9.12301004e-01 7.14400709e-01 -1.18464783e-01 -2.78025210e-01
6.16107523e-01 -2.35119134e-01 1.64818347e-01 1.50215137e-03
-1.61179411e+00 5.46748638e-01 -1.12662360e-01 1.45089598e-02
-1.25972831e+00 -1.93215817e-01 -5.96866369e-01 -1.55258283e-01
6.53267622e-01 -2.31701940e-01 1.19511878e+00 -7.66285419e-01
-1.09400952e+00 8.23981762e-01 1.05976909e-01 -5.32819748e-01
6.55811429e-01 -1.85167477e-01 -5.58867037e-01 3.02302450e-01
1.15708128e-01 -1.72331572e-01 1.07780957e+00 -1.08494985e+00
-3.80710393e-01 -2.59847045e-01 1.31154329e-01 -6.16611652e-02
-3.88588428e-01 6.61080837e-01 3.21012666e-03 -7.32641518e-01
1.27425464e-02 -8.72602344e-01 3.06995571e-01 -5.29944479e-01
-1.12434137e+00 -2.40338415e-01 8.26581359e-01 -9.14424539e-01
1.47280872e+00 -1.81693590e+00 -1.25480428e-01 1.74422324e-01
1.08889496e+00 1.19615935e-01 2.99357116e-01 9.29878533e-01
3.97374988e-01 3.03390890e-01 8.65137875e-02 -3.52890074e-01
-7.80276060e-02 -3.62117216e-02 -4.88408774e-01 1.00753677e+00
-1.88515753e-01 7.42840588e-01 -1.38750792e+00 -1.11015975e-01
-3.56542677e-01 2.64781296e-01 -3.52482438e-01 3.60453516e-01
-1.62038133e-01 3.92957330e-01 -6.07405007e-01 3.70092332e-01
4.66458380e-01 -9.18225646e-01 -9.73586738e-03 5.38425088e-01
1.11571863e-01 7.67547846e-01 -4.16277230e-01 4.85293120e-01
-9.13586617e-02 9.13007498e-01 5.61277628e-01 -5.44193745e-01
1.16997600e+00 4.06628728e-01 3.05454046e-01 -1.90704629e-01
4.60958660e-01 4.24027517e-02 -3.29253405e-01 -3.95697445e-01
8.15896928e-01 1.56273648e-01 -1.72658831e-01 1.13161254e+00
-3.27980220e-01 -1.86624359e-02 -2.15815187e-01 8.05344224e-01
1.24481976e+00 -8.51519585e-01 6.12708390e-01 -2.94828802e-01
1.74006730e-01 -8.11968073e-02 1.22467265e-01 1.04190242e+00
-3.72670203e-01 3.40776414e-01 1.34232199e+00 -1.84756771e-01
-9.76801455e-01 -6.67754889e-01 2.12512255e-01 1.47147262e+00
2.39811361e-01 -3.46151739e-01 -6.77964926e-01 -7.82169521e-01
1.82059288e-01 8.00475657e-01 -8.36452425e-01 1.20714583e-01
-3.56273800e-01 -9.21031833e-01 8.96410942e-01 -2.61582974e-02
5.04743874e-01 -1.23072517e+00 1.87084988e-01 2.42833361e-01
-6.24457598e-01 -1.01471281e+00 -1.07493210e+00 -3.98601443e-01
-3.56613904e-01 -1.24615037e+00 -6.81322455e-01 -4.14076239e-01
3.34499896e-01 5.85953355e-01 1.34551680e+00 7.39946902e-01
2.35659659e-01 3.74124870e-02 -5.76493800e-01 -3.61391753e-01
-9.39034164e-01 -1.39080346e-01 -6.43592551e-02 2.55966306e-01
7.45071918e-02 -7.82361686e-01 -6.11309052e-01 6.77047670e-01
-7.20134795e-01 5.77911958e-02 -1.17016003e-01 6.07377112e-01
-2.99617857e-01 -8.67741480e-02 9.51934457e-01 -1.30689108e+00
1.00500512e+00 -1.48570716e+00 -1.04954004e-01 -4.57469881e-01
-4.87787098e-01 -6.59717679e-01 6.06702507e-01 -1.22690693e-01
-6.02244258e-01 -1.11536682e+00 2.28119977e-02 -2.25122914e-01
1.32296667e-01 6.35779083e-01 6.13091946e-01 3.59669179e-01
8.03902745e-01 2.20089346e-01 5.35034761e-02 -5.19149959e-01
1.79551065e-01 1.09590840e+00 3.49032134e-01 6.18054084e-02
4.29917395e-01 7.67308235e-01 -4.19949383e-01 -1.19098246e+00
-1.19949877e+00 -8.26469123e-01 -1.01936325e-01 -5.63077211e-01
4.82707113e-01 -6.84045255e-01 -8.03595841e-01 8.58592451e-01
-1.38232660e+00 -3.37292045e-01 3.77825826e-01 8.06507394e-02
-3.22004706e-01 5.17531633e-01 -1.51608360e+00 -9.41435874e-01
-5.49577713e-01 -4.75758135e-01 2.72626907e-01 -1.36392370e-01
-5.23450077e-01 -1.15916097e+00 -1.99348018e-01 7.69535780e-01
4.20693934e-01 2.92572111e-01 4.82015878e-01 -1.12064493e+00
-4.47759330e-01 -2.64707476e-01 -8.77744675e-01 2.30735391e-01
1.51538745e-01 -3.10953736e-01 -8.63281429e-01 -2.79576063e-01
5.83310053e-02 -3.10749561e-01 9.84301090e-01 8.75386223e-03
7.83644915e-01 -1.11210608e+00 -2.02191696e-01 1.53748780e-01
6.87446952e-01 -4.60820615e-01 5.21493852e-01 -3.09464522e-02
8.54626656e-01 5.43669581e-01 8.28389227e-02 9.46205258e-01
6.91515386e-01 2.81537205e-01 8.67957473e-01 1.06399380e-01
-1.60175860e-01 -5.48191845e-01 6.87016428e-01 1.03334033e+00
-1.92681894e-01 -7.61825621e-01 -8.60564947e-01 7.65872538e-01
-1.70755255e+00 -1.20280051e+00 -6.23815119e-01 1.73577476e+00
7.38979220e-01 1.19322181e-01 6.48062110e-01 -4.28190716e-02
1.10332787e+00 9.25018907e-01 -2.40917653e-01 -3.87447208e-01
-6.32802844e-02 -7.18335807e-01 5.93983471e-01 8.17354798e-01
-7.28649795e-01 6.24313712e-01 5.17067575e+00 5.61040342e-01
-8.97817314e-01 4.75740433e-01 6.60291255e-01 -1.60348549e-01
-3.43831629e-01 -2.70580262e-01 -5.56304991e-01 6.75843060e-01
1.02090311e+00 -1.78753555e-01 9.01370525e-01 4.55479145e-01
5.56028426e-01 4.01924849e-02 -6.34451747e-01 4.28706884e-01
4.13171023e-01 -1.52084148e+00 -1.10230856e-01 2.24916130e-01
9.67638195e-01 6.16248965e-01 -4.59785163e-02 3.38865048e-03
7.38008738e-01 -1.00528407e+00 6.99046969e-01 4.06816304e-01
4.66649443e-01 -5.11809647e-01 7.12587595e-01 7.32477605e-01
-7.53450155e-01 2.03149039e-02 -2.62468487e-01 -3.68259341e-01
2.71634161e-01 8.90152037e-01 -1.30118930e+00 3.80691558e-01
4.67683524e-01 1.08063054e+00 -1.31090567e-01 7.17473567e-01
-4.83415753e-01 1.25043213e+00 7.87344053e-02 -4.33195293e-01
5.31658113e-01 -7.56102055e-02 1.33443499e+00 1.42413414e+00
1.73734520e-02 1.78438216e-01 2.76071578e-01 1.03592372e+00
-7.61846304e-01 -4.43594866e-02 -6.26456857e-01 -3.23784739e-01
5.91032684e-01 8.74712169e-01 -6.64841592e-01 -4.53267872e-01
-2.72215992e-01 9.86701906e-01 4.65838224e-01 3.67874146e-01
-8.37856531e-01 -2.40148455e-01 5.71727514e-01 6.64381325e-01
9.06213671e-02 2.28251908e-02 -2.06384379e-02 -9.85070527e-01
-3.93708020e-01 -8.58728826e-01 5.62049389e-01 -6.68676913e-01
-1.61649024e+00 5.36355197e-01 -3.62163633e-01 -7.41535485e-01
4.26532961e-02 6.64541274e-02 -9.77995753e-01 6.71376526e-01
-1.52229249e+00 -7.97470093e-01 -3.30064058e-01 4.98592436e-01
4.11271393e-01 -6.88970685e-02 3.11987400e-01 1.22232936e-01
-4.31179553e-01 2.97954351e-01 3.33630919e-01 5.91865838e-01
4.31419581e-01 -1.08080530e+00 5.64492226e-01 3.01853418e-01
-9.69577432e-02 3.53663087e-01 9.39961255e-01 -9.34473872e-01
-1.06100965e+00 -1.29632699e+00 1.10996938e+00 -4.99655932e-01
1.39121616e+00 -2.27267876e-01 -1.10913360e+00 9.23694968e-01
3.84469718e-01 -1.62014917e-01 4.36066061e-01 1.77387968e-01
-4.81410205e-01 5.78067303e-01 -1.05085433e+00 5.03402591e-01
1.03835404e+00 -6.82585359e-01 -5.03190458e-01 8.40202093e-01
7.07222760e-01 -1.90404877e-01 -7.65045941e-01 -3.31442386e-01
1.97306335e-01 -1.39054739e+00 6.28187001e-01 -6.63148284e-01
9.21849728e-01 2.42199391e-01 2.67871171e-01 -1.51949894e+00
-4.41205233e-01 -1.28583300e+00 -5.42176187e-01 1.05820394e+00
4.59735185e-01 -1.00513756e+00 6.24814987e-01 -1.42356366e-01
1.12854093e-02 -5.37694156e-01 -6.80048585e-01 -5.44514239e-01
2.98548751e-02 -1.32911548e-01 6.15543008e-01 1.45315647e+00
5.52607000e-01 4.51376766e-01 -1.09857059e+00 1.41557544e-01
6.60694122e-01 -9.93992668e-03 6.14679992e-01 -1.17585564e+00
-1.11751258e-01 -5.96390247e-01 4.60572168e-02 -1.34268725e+00
3.88065577e-02 -1.07405972e+00 -1.67663798e-01 -1.41259813e+00
5.56386709e-01 -1.96062684e-01 -2.51419872e-01 1.21863477e-01
7.49184471e-03 3.75626713e-01 1.12376183e-01 7.01061606e-01
-8.01476121e-01 3.54180068e-01 1.46717238e+00 -9.09710377e-02
-1.37964651e-01 2.85315633e-01 -7.93320835e-01 9.02792454e-01
7.27873981e-01 -7.37011492e-01 1.51755800e-02 -5.36831189e-03
6.83884084e-01 7.57624507e-01 6.17862463e-01 -2.17344165e-01
1.65449381e-01 -7.06924275e-02 -3.27037752e-01 -5.53293407e-01
4.19307262e-01 -5.56591488e-02 -3.06611598e-01 2.60701537e-01
-5.48081398e-01 1.59762591e-01 -2.84824967e-01 1.01719141e+00
-1.72023904e-02 -1.62331983e-01 7.77593911e-01 -2.61681825e-01
-8.88698772e-02 3.05043042e-01 -7.25532115e-01 4.54959840e-01
5.00406563e-01 3.80764544e-01 -1.04109955e+00 -1.42052770e+00
-6.66308522e-01 3.62143964e-01 4.52346981e-01 4.67084199e-01
5.15084565e-01 -9.28425491e-01 -1.36790419e+00 -2.77239740e-01
-1.73112497e-01 -5.65179527e-01 2.27607250e-01 1.41791344e+00
-1.76171005e-01 1.90613687e-01 7.12066516e-02 4.60235849e-02
-1.19286788e+00 5.00095546e-01 2.82523483e-01 -2.60091841e-01
-6.92617118e-01 9.64403987e-01 5.86397387e-02 -2.59819776e-01
-1.52565390e-02 2.26678535e-01 -3.74072820e-01 4.25509065e-01
7.49404848e-01 1.04688168e+00 -3.46328795e-01 -9.37123239e-01
-1.25355557e-01 -6.04826093e-01 -2.76831001e-01 3.47880989e-01
1.31960869e+00 -6.21609211e-01 -7.37687767e-01 5.98209381e-01
1.22742677e+00 3.01678181e-01 -7.23893344e-01 -6.51540339e-01
-6.77348375e-02 -4.83115464e-01 3.45336109e-01 -5.34445286e-01
-9.28156972e-01 4.14810121e-01 -7.54210472e-01 1.18489575e+00
2.83264756e-01 4.06262130e-01 1.23705387e+00 1.28567085e-01
2.72474408e-01 -7.86746383e-01 3.33937198e-01 8.89678061e-01
1.20395923e+00 -1.05424082e+00 1.21938154e-01 -7.29920387e-01
-7.95478642e-01 7.56083608e-01 1.55633716e-02 -3.55563998e-01
8.29287171e-01 -2.72552788e-01 -1.43923566e-01 -8.55085373e-01
-7.40836203e-01 2.26627633e-01 3.95508371e-02 -1.33713409e-02
4.11589116e-01 4.26918268e-01 -1.16916753e-01 5.12509704e-01
-3.36335719e-01 -2.19691485e-01 1.20103812e+00 4.63646501e-01
-7.11009443e-01 -2.39780381e-01 -5.90854704e-01 9.56519902e-01
-8.75468850e-01 -2.16508955e-01 -8.90645027e-01 4.78358448e-01
-5.09002209e-01 1.47185957e+00 -2.49197617e-01 -4.60805565e-01
-2.40384564e-01 -1.81928173e-01 -1.25127167e-01 -5.92152536e-01
-7.87219048e-01 -1.61074728e-01 8.42243016e-01 -3.31785768e-01
-2.08179820e-02 -7.90317476e-01 -9.89765286e-01 -1.18753195e+00
-4.66948986e-01 3.84035915e-01 6.13334849e-02 8.09765816e-01
2.10747853e-01 3.32900316e-01 1.09748077e+00 -5.62744558e-01
-6.64412916e-01 -1.34519887e+00 -1.22367275e+00 4.62561190e-01
9.32074070e-01 -4.73069608e-01 -8.38636160e-01 -4.46258515e-01] | [8.18432331085205, 10.142637252807617] |
4538a3e4-dfd3-421c-9f79-a9154dd798b1 | dense-relational-captioning-triple-stream | 1903.05942 | null | https://arxiv.org/abs/1903.05942v4 | https://arxiv.org/pdf/1903.05942v4.pdf | Dense Relational Captioning: Triple-Stream Networks for Relationship-Based Captioning | Our goal in this work is to train an image captioning model that generates more dense and informative captions. We introduce "relational captioning," a novel image captioning task which aims to generate multiple captions with respect to relational information between objects in an image. Relational captioning is a framework that is advantageous in both diversity and amount of information, leading to image understanding based on relationships. Part-of speech (POS, i.e. subject-object-predicate categories) tags can be assigned to every English word. We leverage the POS as a prior to guide the correct sequence of words in a caption. To this end, we propose a multi-task triple-stream network (MTTSNet) which consists of three recurrent units for the respective POS and jointly performs POS prediction and captioning. We demonstrate more diverse and richer representations generated by the proposed model against several baselines and competing methods. | ['Tae-Hyun Oh', 'Jinsoo Choi', 'Dong-Jin Kim', 'In So Kweon'] | 2019-03-14 | dense-relational-captioning-triple-stream-1 | http://openaccess.thecvf.com/content_CVPR_2019/html/Kim_Dense_Relational_Captioning_Triple-Stream_Networks_for_Relationship-Based_Captioning_CVPR_2019_paper.html | http://openaccess.thecvf.com/content_CVPR_2019/papers/Kim_Dense_Relational_Captioning_Triple-Stream_Networks_for_Relationship-Based_Captioning_CVPR_2019_paper.pdf | cvpr-2019-6 | ['relational-captioning', 'relational-captioning'] | ['computer-vision', 'natural-language-processing'] | [ 8.29858243e-01 5.36795020e-01 -2.90478915e-01 -6.24933779e-01
-1.15265584e+00 -3.13103646e-01 9.36920881e-01 -8.46873820e-02
-8.45136940e-02 6.76443696e-01 8.35020363e-01 -2.15781897e-01
5.53787529e-01 -5.32558262e-01 -1.26324451e+00 -4.46744740e-01
4.37354803e-01 5.56446731e-01 1.23775594e-01 -2.12376803e-01
1.35612503e-01 3.09010088e-01 -1.42199481e+00 9.40135777e-01
6.24785423e-01 8.37234318e-01 7.56626308e-01 5.41467190e-01
-1.42434284e-01 1.05597663e+00 -5.20533919e-01 -7.99828947e-01
-1.40700415e-01 -6.45509481e-01 -8.26701462e-01 2.25745693e-01
4.03270453e-01 -1.86436057e-01 -4.25376803e-01 9.52576637e-01
3.61795962e-01 -1.03088632e-01 8.62527013e-01 -1.34861326e+00
-1.26723003e+00 8.72722924e-01 -7.64595509e-01 5.19790165e-02
4.91751254e-01 1.38056546e-01 1.38857424e+00 -8.81902099e-01
7.33301878e-01 1.38121390e+00 1.72905609e-01 8.56675327e-01
-1.11291313e+00 -3.95400465e-01 2.02384308e-01 2.04953238e-01
-1.12996387e+00 -2.69654363e-01 6.74086094e-01 -4.79202509e-01
6.03329957e-01 2.27468442e-02 2.52924502e-01 1.68119597e+00
1.73210263e-01 1.08411956e+00 7.31275797e-01 -3.91939431e-01
-2.24945396e-02 3.37178111e-02 -1.26458600e-01 3.91317427e-01
1.07628621e-01 -2.29491308e-01 -5.25036395e-01 1.50541142e-01
8.15404296e-01 -2.10862570e-02 -2.27307305e-01 -2.47134984e-01
-1.64252174e+00 7.63765454e-01 7.83611596e-01 1.37144014e-01
-7.87820041e-01 5.82924545e-01 3.64470541e-01 -4.81534868e-01
2.30263874e-01 5.15527725e-01 -1.72073364e-01 1.58126727e-01
-6.18042946e-01 1.55201962e-03 3.46064061e-01 1.28117633e+00
5.93821704e-01 -1.93090975e-01 -1.01038349e+00 8.68984997e-01
5.39073408e-01 7.49010265e-01 3.24385822e-01 -7.97228575e-01
7.15233743e-01 2.41773471e-01 2.63244718e-01 -8.45087171e-01
2.09692582e-01 -2.62917459e-01 -8.33806574e-01 -5.30700624e-01
-3.32621396e-01 9.72050205e-02 -1.35131967e+00 1.93969095e+00
-2.74331495e-02 4.86707389e-01 5.65901458e-01 9.37025547e-01
1.23898244e+00 1.16305232e+00 6.58869028e-01 6.07987866e-02
1.80377162e+00 -1.35649097e+00 -7.72966266e-01 -5.89323699e-01
2.26518497e-01 -7.19795942e-01 9.72193897e-01 -2.54209906e-01
-9.61173117e-01 -6.80755675e-01 -6.63048208e-01 -1.49593264e-01
-1.61661595e-01 1.42319024e-01 4.43208873e-01 -1.19389705e-01
-1.14480007e+00 7.41682127e-02 -5.44899046e-01 -2.43548900e-01
6.74755394e-01 6.20263703e-02 -3.16629708e-01 -1.42042339e-01
-1.18997335e+00 8.35975468e-01 4.53999937e-01 1.31287530e-01
-1.22434962e+00 -5.01899242e-01 -1.24653780e+00 1.08123980e-01
1.17545377e-03 -1.12442887e+00 1.44113767e+00 -1.30531180e+00
-1.11889637e+00 1.06894231e+00 -3.96825165e-01 -7.28080511e-01
8.75166059e-03 -1.20683506e-01 -2.73322672e-01 5.79125583e-01
5.43436944e-01 1.64481294e+00 8.49492848e-01 -1.74183023e+00
-4.71729368e-01 8.46668556e-02 -1.21488936e-01 5.67016304e-01
1.61276847e-01 3.09020728e-01 -6.81875587e-01 -6.95920587e-01
-3.97414207e-01 -8.46566379e-01 -3.66247773e-01 -2.74685115e-01
-7.93552458e-01 -3.05006564e-01 3.96200716e-01 -7.06210017e-01
6.08098507e-01 -2.12402606e+00 1.05287172e-01 -2.77440637e-01
-1.27394065e-01 3.46503168e-01 -6.59394145e-01 4.56508368e-01
-2.94209838e-01 2.93597251e-01 -2.48032406e-01 -5.66497922e-01
2.05673533e-03 4.86106426e-01 -8.52596879e-01 -1.63728565e-01
7.58702517e-01 1.41050076e+00 -1.12869823e+00 -8.49309325e-01
1.19489469e-01 7.74930298e-01 -1.92994803e-01 5.77286005e-01
-6.62991464e-01 4.69196111e-01 -5.91575861e-01 4.93885487e-01
4.84592199e-01 -7.02900946e-01 -1.47440866e-01 -4.69103694e-01
2.15253606e-01 3.56315643e-01 -4.05992538e-01 1.68574476e+00
-4.82420117e-01 5.04876316e-01 -5.60332775e-01 -7.64074862e-01
1.03856993e+00 5.16253769e-01 3.28309417e-01 -6.66638494e-01
-1.03103798e-02 6.12106621e-02 -4.21223968e-01 -5.78134894e-01
6.62141323e-01 -1.09118940e-02 -3.36485982e-01 2.05476165e-01
2.21256822e-01 -1.28750756e-01 1.51636884e-01 3.59455377e-01
7.84470379e-01 5.42739153e-01 2.75198579e-01 1.42314002e-01
4.66758847e-01 -3.82181555e-02 3.20097983e-01 6.99725568e-01
-1.73583534e-02 1.16949618e+00 5.37280142e-01 -2.14222237e-01
-1.30021930e+00 -1.18091476e+00 5.93859181e-02 9.20056939e-01
3.16378832e-01 1.61888432e-02 -5.50401390e-01 -6.68471456e-01
-4.10173595e-01 9.06231999e-01 -7.65103400e-01 -8.03987235e-02
-5.60959756e-01 -3.36752236e-01 3.38163912e-01 6.47935390e-01
4.45424438e-01 -1.79560518e+00 -3.24541152e-01 -5.66227175e-03
-6.20686233e-01 -1.67008448e+00 -7.60914207e-01 1.83100831e-02
-4.36899453e-01 -6.32331789e-01 -1.06016350e+00 -1.33611143e+00
8.88857961e-01 2.40634128e-01 1.52467215e+00 -3.46397728e-01
2.21529067e-01 3.21168184e-01 -7.84017384e-01 -4.98562247e-01
-5.85129559e-01 1.44941621e-02 -5.30527830e-01 2.63100028e-01
2.47274905e-01 -2.27830142e-01 -6.87107444e-01 1.21465482e-01
-1.07531977e+00 7.46687293e-01 1.12723005e+00 7.42440343e-01
1.00671053e+00 -8.79395306e-01 6.04636908e-01 -7.43558288e-01
3.86707753e-01 -6.55239880e-01 -3.21432889e-01 5.00296474e-01
1.30144566e-01 2.42653936e-01 4.94537741e-01 -3.32192540e-01
-1.13688660e+00 4.37115878e-01 -2.58784324e-01 -4.79810685e-01
-2.51061678e-01 3.27939630e-01 -2.01537505e-01 3.90343606e-01
4.53593493e-01 7.22987115e-01 -5.75072356e-02 -1.31074607e-01
8.46244752e-01 7.66056478e-01 9.81298029e-01 -5.27204573e-01
5.90093493e-01 2.99697787e-01 -3.69627364e-02 -5.63342869e-01
-1.39493477e+00 -4.87702131e-01 -4.99333113e-01 -6.47699237e-02
1.37883914e+00 -1.28766954e+00 -4.52471763e-01 9.21393409e-02
-1.78124201e+00 -9.57392454e-02 -3.66598338e-01 2.58188903e-01
-9.20883417e-01 1.31946981e-01 -4.59854782e-01 -5.85837960e-01
-6.04571998e-01 -1.29050684e+00 1.74143302e+00 3.96251708e-01
-2.54976824e-02 -7.77362466e-01 -7.55909607e-02 6.58784807e-01
-4.36988324e-02 3.94402117e-01 5.26481688e-01 -8.60659361e-01
-7.87164092e-01 2.34077454e-01 -6.31197631e-01 4.10743564e-01
-2.26190947e-02 -2.66884774e-01 -7.83849776e-01 1.00506745e-01
-3.42186749e-01 -5.18793225e-01 9.16196704e-01 3.37505668e-01
1.24535024e+00 -5.77076077e-01 -5.04770219e-01 3.14911574e-01
1.42951918e+00 3.17416996e-01 1.12341559e+00 2.02645808e-01
9.43728149e-01 7.26969361e-01 6.38969958e-01 1.08846195e-01
5.56186020e-01 6.89330876e-01 6.70270443e-01 -3.28767091e-01
-5.25697649e-01 -7.85773337e-01 4.22307551e-01 5.64822257e-01
4.58192050e-01 -7.31907547e-01 -8.99431467e-01 9.19286788e-01
-1.88574028e+00 -1.06141031e+00 -1.11910827e-01 1.60755396e+00
1.12419927e+00 -1.67418957e-01 -2.38276869e-01 -7.50263810e-01
1.20164359e+00 2.03897491e-01 -4.29561287e-01 -3.12763125e-01
-1.58606723e-01 1.00023754e-01 3.85995239e-01 3.37622881e-01
-1.03684413e+00 1.18675125e+00 5.89589834e+00 5.17552018e-01
-9.46682990e-01 -6.69708624e-02 1.00608134e+00 4.39836770e-01
-6.04012609e-01 -1.96056440e-01 -9.03021872e-01 4.99965787e-01
9.80700850e-01 2.95574870e-02 -4.70760427e-02 8.76912475e-01
1.40625536e-01 1.17183395e-01 -1.12157404e+00 1.16893101e+00
6.41246259e-01 -1.51673114e+00 7.57448614e-01 -1.31377831e-01
8.62216175e-01 1.62302941e-01 2.57613540e-01 1.11142024e-01
4.63369578e-01 -1.10652554e+00 8.04610252e-01 6.26653612e-01
7.95699894e-01 -4.69146878e-01 9.41317916e-01 -2.66109426e-02
-1.02759302e+00 3.00115347e-01 -2.27946192e-01 4.88295734e-01
6.08145952e-01 3.71495426e-01 -1.17941868e+00 5.63945234e-01
4.12440777e-01 9.37791049e-01 -4.65831161e-01 1.05584073e+00
-7.55779564e-01 3.60635310e-01 1.06700994e-01 -9.18053463e-02
4.12939399e-01 -3.55446897e-02 2.94838876e-01 1.24957752e+00
4.25447792e-01 9.22004282e-02 1.15039140e-01 1.09256291e+00
-4.31822836e-01 -2.61617582e-02 -7.32743084e-01 -1.99560821e-01
4.28321689e-01 1.28771961e+00 -7.05498815e-01 -5.01638412e-01
-4.54744399e-01 1.29898965e+00 1.17490634e-01 3.59507740e-01
-9.63977158e-01 -1.10600561e-01 5.30848384e-01 -1.56861101e-03
7.35573709e-01 1.65018216e-01 -5.63870892e-02 -1.09678459e+00
-3.73483673e-02 -6.94263339e-01 2.25724071e-01 -1.54368126e+00
-1.48264837e+00 1.02236998e+00 7.70350173e-02 -1.20683992e+00
-4.30666476e-01 -5.82699478e-01 -3.64483923e-01 9.99929190e-01
-1.84034300e+00 -1.72031736e+00 -2.06927329e-01 3.78518462e-01
7.27867544e-01 6.22036271e-02 6.87250376e-01 7.27387443e-02
-3.29405725e-01 3.42014700e-01 -3.84552807e-01 4.23805147e-01
6.56764805e-01 -1.13287294e+00 7.19918966e-01 9.32558417e-01
4.35983449e-01 4.31887180e-01 8.79266262e-01 -8.22510421e-01
-9.79377508e-01 -1.58207059e+00 1.34203911e+00 -6.20490015e-01
4.60878342e-01 -4.77836698e-01 -6.40590310e-01 8.67631435e-01
5.99997282e-01 1.66467518e-01 5.79360723e-01 -3.67180794e-01
-4.39067245e-01 -1.87118612e-02 -8.52160156e-01 7.90626526e-01
9.36583340e-01 -5.83926797e-01 -8.25626016e-01 5.92386186e-01
1.37465417e+00 -2.94674575e-01 -6.11641765e-01 2.39873469e-01
1.10745750e-01 -5.64809084e-01 1.14898765e+00 -4.13132042e-01
1.03061128e+00 -4.80715543e-01 -1.71894640e-01 -1.18021464e+00
-1.86204523e-01 -5.79297900e-01 1.33096099e-01 1.40438366e+00
6.64357543e-01 -1.61829609e-02 6.28625035e-01 2.01792568e-01
-5.58320940e-01 -5.90780079e-01 -5.08477449e-01 -5.29520690e-01
-2.70341963e-01 -3.23579013e-01 5.79878867e-01 4.81985420e-01
-2.48222709e-01 9.76317465e-01 -6.29229009e-01 2.79735774e-01
4.20576811e-01 1.63825244e-01 4.77978677e-01 -7.62996972e-01
-7.61587620e-02 -1.18570991e-01 -2.96947926e-01 -1.37156689e+00
3.85674655e-01 -9.79985476e-01 5.74541211e-01 -2.08661819e+00
5.58225155e-01 -1.44344792e-01 -1.66228235e-01 7.25304604e-01
-3.50231618e-01 5.89372635e-01 2.01954216e-01 3.58057916e-01
-1.03153527e+00 8.75488937e-01 1.59369552e+00 -2.56792903e-01
1.08802713e-01 -3.04771841e-01 -1.02909863e+00 2.78316170e-01
5.12207687e-01 -4.63826180e-01 -5.52414119e-01 -7.12567270e-01
2.67012507e-01 3.05235535e-01 5.22789955e-01 -7.39749014e-01
8.16839412e-02 -1.61713898e-01 2.19920129e-01 -6.89790606e-01
4.91977483e-01 -7.02507496e-01 5.42433895e-02 1.66812614e-01
-7.54862905e-01 1.08899973e-01 -6.66419789e-02 6.35631204e-01
-4.80485767e-01 -2.28565842e-01 6.90213799e-01 -1.85936213e-01
-9.15867388e-01 5.31793356e-01 3.23664881e-02 -2.30995174e-02
1.26730871e+00 -3.20528224e-02 -4.01702493e-01 -4.76244807e-01
-5.51415563e-01 2.45807454e-01 2.62911171e-01 7.93140292e-01
9.43838358e-01 -1.67659366e+00 -9.81116116e-01 -1.14860937e-01
7.05171108e-01 1.43165678e-01 6.96568713e-02 3.03598434e-01
-4.85504001e-01 6.54576898e-01 -3.28690648e-01 -7.16574430e-01
-9.65793192e-01 8.39127064e-01 8.07194114e-02 -1.94127098e-01
-5.05421758e-01 9.53850865e-01 4.37915772e-01 -2.08727904e-02
9.95973572e-02 -1.65338963e-01 -6.62460268e-01 5.18487650e-04
6.16859436e-01 -5.85676968e-01 -4.00646210e-01 -1.07960057e+00
-2.02083662e-01 3.53751421e-01 -2.06616595e-01 -2.18011782e-01
1.30432045e+00 -1.91007152e-01 -1.59511700e-01 2.17092007e-01
1.21273518e+00 -5.16617596e-01 -1.44561064e+00 -1.31112933e-01
-6.32421821e-02 -1.90923065e-01 -3.13310087e-01 -8.03849816e-01
-7.69618154e-01 7.71091521e-01 2.58106023e-01 -1.65484652e-01
1.00865853e+00 6.11140311e-01 9.98937845e-01 1.29907727e-01
1.27272770e-01 -5.85192859e-01 6.43720627e-01 4.64004189e-01
1.10152102e+00 -1.32412064e+00 -5.60895264e-01 -6.53195262e-01
-1.31785285e+00 7.58043766e-01 7.59062409e-01 1.26675824e-02
8.75171274e-02 -8.79543498e-02 -3.48574556e-02 -1.08578332e-01
-8.01918566e-01 -5.24670541e-01 3.51917386e-01 9.93887484e-01
3.71413469e-01 -6.21665455e-02 -6.55976683e-02 4.03010309e-01
-7.10888728e-02 -1.01419119e-03 5.38277388e-01 4.79657263e-01
-2.66681075e-01 -1.08820748e+00 -1.94372356e-01 2.19221339e-01
-4.42097157e-01 -6.11511230e-01 -2.50501841e-01 2.65937209e-01
1.13765061e-01 8.63940835e-01 2.80710757e-01 -1.47500694e-01
1.91298097e-01 -1.97560638e-01 2.38402233e-01 -1.07096553e+00
-2.48366565e-01 -2.05419719e-01 1.19149506e-01 -3.62388611e-01
-6.74125612e-01 -6.00231111e-01 -1.20937693e+00 5.39688110e-01
1.91012189e-01 2.81906277e-01 6.33361995e-01 8.40959251e-01
5.67656159e-01 6.43448174e-01 5.10750711e-01 -7.47812271e-01
-3.62552106e-01 -9.23688054e-01 -7.44126290e-02 7.17195511e-01
2.30012313e-01 -3.35263073e-01 6.53513372e-02 6.12777770e-01] | [10.8507080078125, 0.9968134164810181] |
fe038ce2-24de-45ae-bb11-b57465f9f228 | imagen-video-high-definition-video-generation | 2210.02303 | null | https://arxiv.org/abs/2210.02303v1 | https://arxiv.org/pdf/2210.02303v1.pdf | Imagen Video: High Definition Video Generation with Diffusion Models | We present Imagen Video, a text-conditional video generation system based on a cascade of video diffusion models. Given a text prompt, Imagen Video generates high definition videos using a base video generation model and a sequence of interleaved spatial and temporal video super-resolution models. We describe how we scale up the system as a high definition text-to-video model including design decisions such as the choice of fully-convolutional temporal and spatial super-resolution models at certain resolutions, and the choice of the v-parameterization of diffusion models. In addition, we confirm and transfer findings from previous work on diffusion-based image generation to the video generation setting. Finally, we apply progressive distillation to our video models with classifier-free guidance for fast, high quality sampling. We find Imagen Video not only capable of generating videos of high fidelity, but also having a high degree of controllability and world knowledge, including the ability to generate diverse videos and text animations in various artistic styles and with 3D object understanding. See https://imagen.research.google/video/ for samples. | ['Tim Salimans', 'David J. Fleet', 'Mohammad Norouzi', 'Ben Poole', 'Diederik P. Kingma', 'Alexey Gritsenko', 'Ruiqi Gao', 'Jay Whang', 'Chitwan Saharia', 'William Chan', 'Jonathan Ho'] | 2022-10-05 | null | null | null | null | ['video-super-resolution', 'video-generation'] | ['computer-vision', 'computer-vision'] | [ 3.62321913e-01 -1.82374746e-01 -7.71213248e-02 -3.45140658e-02
-7.12680936e-01 -7.32154429e-01 1.09586585e+00 -6.88441813e-01
9.15770605e-03 7.16600358e-01 5.86824477e-01 -1.39994755e-01
9.12651718e-02 -7.90564656e-01 -9.31633592e-01 -3.84653896e-01
1.01036951e-01 3.18716288e-01 2.36018017e-01 -2.01169908e-01
2.56789505e-01 3.57298106e-01 -1.68555117e+00 7.12160528e-01
5.96305192e-01 7.43323445e-01 3.55325788e-01 1.48027349e+00
1.86574638e-01 1.25578356e+00 -5.08835673e-01 -1.19884133e-01
3.84324074e-01 -7.37824857e-01 -5.88325083e-01 3.97119999e-01
6.45402908e-01 -7.93806195e-01 -6.82590306e-01 5.98851860e-01
4.43239421e-01 2.09414259e-01 6.41955674e-01 -1.26223826e+00
-1.15814531e+00 6.63660169e-01 -3.58172238e-01 3.98482949e-01
7.66724288e-01 7.76421189e-01 7.07178354e-01 -8.36139739e-01
1.32041883e+00 1.40729845e+00 2.59981871e-01 9.59341228e-01
-1.33984613e+00 -7.47111499e-01 4.49751839e-02 -2.89492588e-02
-1.10141814e+00 -7.67174363e-01 3.40882510e-01 -7.82164872e-01
8.33789647e-01 6.27612397e-02 1.01709831e+00 1.80307007e+00
2.37408251e-01 7.40801871e-01 8.12906146e-01 -1.42234311e-01
2.52971739e-01 -2.09631175e-01 -7.39496589e-01 7.27671146e-01
-4.48315777e-02 3.07214499e-01 -8.62108827e-01 2.61170883e-02
1.60681820e+00 -1.66723400e-01 -3.67436111e-01 -1.01713315e-01
-1.42308378e+00 8.04100633e-01 1.71833690e-02 1.06542952e-01
-4.72698778e-01 6.14686608e-01 5.47407120e-02 2.40458265e-01
4.55455691e-01 5.12237906e-01 -4.15808782e-02 -4.28390443e-01
-1.22970557e+00 5.56857228e-01 5.95961690e-01 1.32216394e+00
3.73946577e-01 6.47679031e-01 -4.71648961e-01 3.19592893e-01
-1.25770301e-01 6.12078071e-01 3.95689279e-01 -1.73013151e+00
3.60446900e-01 -5.28855547e-02 2.88767159e-01 -6.41153157e-01
1.36628315e-01 7.05246478e-02 -5.50873458e-01 3.00379097e-01
1.82492137e-01 -5.33936083e-01 -1.08908546e+00 1.52899289e+00
7.88010135e-02 5.25088847e-01 -8.37994218e-02 9.43407118e-01
5.33153832e-01 9.44382310e-01 1.96092986e-02 -1.89913079e-01
9.65904534e-01 -8.65846872e-01 -6.06185555e-01 1.71361223e-01
2.80887425e-01 -7.65881479e-01 9.94124353e-01 4.28451061e-01
-1.55469406e+00 -6.77351773e-01 -8.06612670e-01 -1.37719229e-01
7.21493214e-02 -1.43361881e-01 4.39521849e-01 3.75473768e-01
-1.60397029e+00 7.09584594e-01 -7.13972092e-01 -3.53806078e-01
4.36194301e-01 1.52422190e-01 -3.34030002e-01 -2.42117479e-01
-1.14314938e+00 4.64072675e-01 2.12981030e-01 -3.88171792e-01
-1.48413742e+00 -8.26833069e-01 -6.76229119e-01 -2.29035154e-01
9.97568443e-02 -1.14406848e+00 1.39951742e+00 -1.52073848e+00
-1.68391061e+00 4.00707334e-01 -5.98915406e-02 -4.59412456e-01
8.20051670e-01 -9.85584185e-02 -3.12371343e-01 6.73669815e-01
1.18084393e-01 1.27877653e+00 1.44363475e+00 -1.06453288e+00
-8.16150188e-01 2.14507401e-01 3.01612794e-01 4.52112645e-01
-1.99378282e-01 -8.57098699e-02 -4.63805079e-01 -7.77485490e-01
-6.92610264e-01 -8.01902533e-01 -2.24143729e-01 2.46481255e-01
-1.30737886e-01 2.24679887e-01 1.05628252e+00 -6.61741257e-01
9.73076642e-01 -2.03990841e+00 4.44626182e-01 -2.06258565e-01
4.50917006e-01 -1.63380489e-01 -4.52233762e-01 3.34969103e-01
2.18709633e-02 4.90681827e-01 1.65092304e-01 -1.11386463e-01
-3.44485670e-01 -1.62875906e-01 -1.92432389e-01 9.79343355e-02
4.37099934e-01 9.35737967e-01 -1.10623658e+00 -5.51669955e-01
4.46445376e-01 8.30609083e-01 -9.76186454e-01 1.88962504e-01
-7.70828068e-01 6.77307427e-01 -6.17380917e-01 4.15914088e-01
2.56399572e-01 -4.92888600e-01 1.05341390e-01 -8.81887078e-02
-2.09596425e-01 -1.44775659e-01 -9.10824656e-01 1.72282720e+00
-5.49800813e-01 1.01328778e+00 2.04944283e-01 -1.66152298e-01
5.38637817e-01 4.10124660e-01 6.98427320e-01 -5.53818345e-01
-1.02936156e-01 -2.19218992e-02 -3.69870126e-01 -6.82211637e-01
1.07606125e+00 -4.58421744e-02 2.84766674e-01 5.60731530e-01
1.35169029e-01 -4.16852176e-01 4.63244766e-01 6.92247987e-01
1.08445108e+00 6.35465741e-01 -3.62313390e-01 -1.10427670e-01
-1.27843365e-01 -1.50576020e-02 -9.64844003e-02 7.52857327e-01
1.50918409e-01 9.62777793e-01 4.30166364e-01 -1.70663118e-01
-1.78844106e+00 -1.09276640e+00 2.65974730e-01 9.37755644e-01
-7.93947652e-02 -4.16147590e-01 -9.08522546e-01 -1.45210192e-01
-1.70558646e-01 7.36839890e-01 -6.94097340e-01 4.53328378e-02
-5.95464051e-01 -3.33585232e-01 5.08452892e-01 4.14725959e-01
3.41930658e-01 -1.12277186e+00 -5.17344952e-01 1.66353032e-01
-3.38017076e-01 -1.26206756e+00 -8.34279537e-01 -4.05178457e-01
-7.95979679e-01 -8.50813866e-01 -1.03469288e+00 -7.55863607e-01
4.88347441e-01 3.42000425e-01 1.10795867e+00 -3.22950520e-02
-2.56708354e-01 7.86045253e-01 -4.31934386e-01 2.92271584e-01
-9.65949774e-01 -2.24682838e-02 3.62703614e-02 8.29078704e-02
-3.42907906e-01 -4.93190169e-01 -8.18910599e-01 5.77550270e-02
-1.26499784e+00 6.48345828e-01 2.23909512e-01 5.01544178e-01
4.11295235e-01 -1.22365706e-01 2.63100237e-01 -6.64953232e-01
7.45709181e-01 -7.20499158e-01 -4.68111008e-01 -3.43031138e-02
-3.06741357e-01 -1.84970051e-02 7.40379393e-01 -7.71350205e-01
-1.23151803e+00 -5.64463297e-03 1.31897166e-01 -9.79455352e-01
-2.45293919e-02 -1.86828349e-03 1.91295788e-01 1.73094034e-01
8.81735981e-01 3.01377088e-01 1.21727288e-02 1.05123341e-01
9.24319267e-01 4.37070847e-01 3.10929626e-01 -6.59442425e-01
6.57566488e-01 5.31925082e-01 -3.27708095e-01 -9.99601483e-01
-1.54902577e-01 3.47785771e-01 -4.47619349e-01 -5.58322251e-01
1.25089586e+00 -1.18825114e+00 -4.65256274e-01 5.18454134e-01
-1.02566016e+00 -1.02006721e+00 -4.97875005e-01 2.59401709e-01
-9.55520689e-01 7.67062381e-02 -1.01756287e+00 -5.71077406e-01
-7.77821243e-02 -1.28541803e+00 1.21006036e+00 1.98795721e-01
-3.05896729e-01 -1.01202726e+00 -1.86631262e-01 2.72978961e-01
6.92981362e-01 3.20263684e-01 5.63753545e-01 9.85439867e-02
-1.21647215e+00 2.55654871e-01 -1.54611319e-01 1.36860594e-01
6.17358014e-02 5.73458254e-01 -5.14221251e-01 -3.85263115e-01
-4.42905009e-01 -4.43540931e-01 7.34331965e-01 7.18704581e-01
8.45546186e-01 -5.26479006e-01 -2.74375200e-01 8.28065157e-01
1.33677495e+00 2.44391158e-01 6.47626698e-01 8.50584581e-02
8.86876643e-01 2.39803180e-01 2.55997777e-01 4.79203999e-01
2.63958633e-01 6.14656031e-01 2.13647217e-01 5.75701073e-02
-4.73188400e-01 -6.70662284e-01 7.28631437e-01 5.46150565e-01
-2.41251528e-01 -6.83268070e-01 -5.53187311e-01 5.21284819e-01
-1.62169349e+00 -1.49855471e+00 7.38533884e-02 1.90024436e+00
7.12955058e-01 6.12717606e-02 3.31544667e-01 -3.90596867e-01
9.49096501e-01 2.93964654e-01 -6.86754405e-01 -2.18561694e-01
-1.90443352e-01 4.25182134e-02 4.25611854e-01 7.80195475e-01
-8.33245456e-01 1.31216550e+00 7.13207531e+00 9.26503360e-01
-1.16042614e+00 -2.56354604e-02 8.35981607e-01 -7.12169051e-01
-8.01803768e-01 -8.43043551e-02 -6.70929015e-01 5.65338552e-01
1.19726205e+00 -4.45751220e-01 8.82078290e-01 4.72527534e-01
6.54218078e-01 1.52607590e-01 -1.00020933e+00 8.65972579e-01
1.34584725e-01 -1.99935830e+00 7.57282674e-01 2.64988653e-02
1.01867259e+00 -9.41712782e-02 2.66605526e-01 2.82399692e-02
7.86789715e-01 -1.13277745e+00 1.19451475e+00 5.77469826e-01
1.46219301e+00 -3.98298889e-01 -2.20172122e-01 -5.93650155e-02
-1.10073531e+00 -2.44718835e-01 -9.46255960e-03 1.34561956e-01
4.52143431e-01 2.19122350e-01 -4.64396387e-01 1.29238814e-01
6.29077971e-01 9.55817819e-01 -3.03437769e-01 4.71841961e-01
-1.95318945e-02 4.81069267e-01 -1.10345729e-01 6.99609071e-02
1.00989804e-01 -8.52699131e-02 6.55225813e-01 1.29741013e+00
7.49593377e-01 2.64389873e-01 9.61230323e-02 9.78312373e-01
-1.32449090e-01 -2.08240539e-01 -8.43285859e-01 -2.62940258e-01
4.67097402e-01 9.97484386e-01 -5.80944777e-01 -5.50901473e-01
-1.68953314e-01 1.20710671e+00 -6.57980964e-02 7.02524424e-01
-9.57531035e-01 1.62354395e-01 6.08876586e-01 5.50359190e-01
5.01712799e-01 -5.14130533e-01 1.59085140e-01 -1.39561629e+00
-2.97987133e-01 -1.04753327e+00 -6.24863617e-02 -1.47945535e+00
-9.15563941e-01 8.06223631e-01 1.31743178e-01 -1.02927458e+00
-6.68653250e-01 -3.25546265e-01 -3.56479526e-01 6.24927819e-01
-1.05830204e+00 -1.10098743e+00 -2.83495486e-01 8.05825353e-01
1.15522301e+00 -1.99452177e-01 3.08613181e-01 1.22717395e-01
-1.69012547e-01 1.15609214e-01 -7.59016201e-02 2.98999771e-02
6.56052232e-01 -8.66004348e-01 7.37903655e-01 9.04659927e-01
-9.07345042e-02 2.00509682e-01 8.31941843e-01 -8.88502598e-01
-1.57212031e+00 -1.24867761e+00 2.32256576e-01 -6.49817765e-01
6.96795881e-01 -4.05305982e-01 -3.68107021e-01 7.91182578e-01
5.03856361e-01 -3.49794924e-01 1.84265405e-01 -6.23287499e-01
-1.16560489e-01 2.86613852e-01 -9.69983041e-01 1.10049844e+00
1.38535047e+00 -5.35999000e-01 9.68513712e-02 3.13519567e-01
1.10788035e+00 -5.80670416e-01 -8.77784908e-01 -6.09998479e-02
7.02753723e-01 -1.01781380e+00 8.10991406e-01 -4.39123005e-01
1.18555713e+00 -1.49335444e-01 -1.41329467e-01 -1.14607608e+00
-6.01947904e-01 -1.03438139e+00 -2.03085020e-01 1.01255071e+00
4.36301619e-01 -7.91686699e-02 8.00284624e-01 6.02134883e-01
9.38368589e-02 -3.33461046e-01 -4.92127150e-01 -4.94669199e-01
1.03515953e-01 -3.73134017e-01 3.85368645e-01 8.50296080e-01
-3.10289472e-01 4.09530908e-01 -6.27813160e-01 -2.29938254e-01
4.80966330e-01 -1.63871765e-01 7.29680836e-01 -6.65053189e-01
-4.58998382e-01 -5.58139622e-01 -2.14164536e-02 -1.28262830e+00
-1.00915723e-01 -6.51572227e-01 -1.71756014e-01 -1.53633642e+00
1.95102125e-01 -1.26999319e-01 2.59916067e-01 3.00642173e-03
2.91850530e-02 2.68833637e-01 4.85402167e-01 3.80788445e-01
-6.29299343e-01 3.40322137e-01 1.73712265e+00 -1.43733937e-02
-3.42038482e-01 -4.61648256e-01 -6.48493111e-01 2.81975687e-01
8.00479233e-01 -1.29553571e-01 -8.17715526e-01 -7.62685359e-01
2.76735932e-01 6.46670878e-01 3.47583622e-01 -9.09251153e-01
-2.31831577e-02 -5.29784083e-01 6.98138297e-01 3.73894796e-02
5.03306270e-01 -3.43993813e-01 5.03123701e-01 1.50554150e-01
-7.81851828e-01 3.59297693e-01 1.23775214e-01 5.81472874e-01
2.62298137e-01 2.52021581e-01 8.04199100e-01 -4.85705912e-01
-6.82402551e-01 7.03224719e-01 -8.27932358e-01 2.04805836e-01
9.92044270e-01 -4.10169482e-01 -5.65606833e-01 -1.01225638e+00
-6.66839182e-01 1.52378067e-01 1.02272737e+00 9.60538447e-01
8.80129099e-01 -1.20194244e+00 -9.63769555e-01 2.36180797e-01
-2.31292441e-01 -2.22857699e-01 4.44645137e-01 2.10578680e-01
-8.32784653e-01 1.40163019e-01 -3.65220577e-01 -5.62404454e-01
-8.55460882e-01 6.23449862e-01 2.90710360e-01 8.94734710e-02
-6.91632986e-01 6.20805621e-01 2.97934353e-01 2.52102613e-01
-1.13622382e-01 -1.51848212e-01 1.05403759e-01 -2.74767190e-01
6.74718022e-01 1.92421675e-01 -6.09630346e-01 -5.88339031e-01
1.22989148e-01 4.18711275e-01 7.21545890e-02 -6.48870349e-01
1.08214092e+00 -2.88110584e-01 3.80051941e-01 1.96621105e-01
9.14571941e-01 -6.77084103e-02 -2.09151983e+00 3.06970417e-01
-6.53169096e-01 -4.10652578e-01 1.98820475e-02 -6.37970507e-01
-1.16854560e+00 6.94983840e-01 3.16261828e-01 1.50354967e-01
9.04003084e-01 -1.82042912e-01 8.12984407e-01 -2.10419446e-01
3.07580233e-01 -1.01723075e+00 5.52322745e-01 3.20440501e-01
9.20822501e-01 -8.00984144e-01 -1.03910238e-01 3.07577336e-03
-8.56077671e-01 1.08464992e+00 6.27250433e-01 -2.11532384e-01
4.52509820e-01 5.83903074e-01 -3.97989713e-02 2.17790436e-02
-1.42131710e+00 9.29581523e-02 -1.55596435e-01 8.84784758e-01
4.20089871e-01 -1.15367115e-01 5.20221107e-02 -1.25059649e-01
-2.80738801e-01 4.58936810e-01 9.91698325e-01 5.84734857e-01
-3.40133518e-01 -8.19988608e-01 -2.39144489e-01 3.88287693e-01
-3.88843328e-01 -3.02302033e-01 -5.38736060e-02 6.67000711e-01
-9.26730875e-03 9.21088338e-01 3.32888454e-01 -6.24132335e-01
-1.13623336e-01 -1.78865075e-01 6.74637616e-01 -4.42928940e-01
-3.79644424e-01 1.23001598e-01 1.13761857e-01 -5.75438678e-01
-4.26268518e-01 -7.74187982e-01 -9.84597862e-01 -6.93076551e-01
-2.51716077e-02 -4.39808108e-02 4.97869879e-01 5.40574491e-01
7.23123550e-01 5.32521486e-01 5.05547702e-01 -1.32996070e+00
-6.29356189e-04 -6.75030649e-01 -3.71631593e-01 4.11895752e-01
4.74284053e-01 -1.90964356e-01 -3.14623624e-01 7.35787749e-01] | [10.916481971740723, -0.6400829553604126] |
e7d410a5-16cf-48b2-b98e-6c571e27baf9 | a-theory-of-human-like-few-shot-learning | 2301.01047 | null | https://arxiv.org/abs/2301.01047v1 | https://arxiv.org/pdf/2301.01047v1.pdf | A Theory of Human-Like Few-Shot Learning | We aim to bridge the gap between our common-sense few-sample human learning and large-data machine learning. We derive a theory of human-like few-shot learning from von-Neuman-Landauer's principle. modelling human learning is difficult as how people learn varies from one to another. Under commonly accepted definitions, we prove that all human or animal few-shot learning, and major models including Free Energy Principle and Bayesian Program Learning that model such learning, approximate our theory, under Church-Turing thesis. We find that deep generative model like variational autoencoder (VAE) can be used to approximate our theory and perform significantly better than baseline models including deep neural networks, for image recognition, low resource language processing, and character recognition. | ['Ming Li', 'Dongbo Bu', 'Rui Wang', 'Zhiying Jiang'] | 2023-01-03 | null | null | null | null | ['common-sense-reasoning'] | ['reasoning'] | [-4.13765386e-03 2.47580320e-01 2.68909894e-02 -2.96923429e-01
-4.64732170e-01 -8.05878788e-02 1.02208912e+00 -1.06672712e-01
-6.39573455e-01 6.83368564e-01 3.23284686e-01 -5.80904931e-02
9.12037771e-03 -1.23276663e+00 -7.75205612e-01 -5.44529855e-01
2.24983364e-01 8.15858960e-01 2.44017392e-01 -3.63734514e-01
2.99093187e-01 -1.13580644e-01 -1.38407004e+00 5.27783819e-02
5.97295761e-01 4.14857805e-01 1.65657595e-01 1.08281159e+00
-8.36205333e-02 1.35152638e+00 -4.73589152e-02 -4.63092744e-01
-9.35005322e-02 -7.42193818e-01 -1.00726116e+00 -3.04083228e-01
2.83222701e-02 -3.35630178e-01 -8.27065706e-01 1.27018285e+00
3.54136169e-01 7.22348571e-01 1.36690104e+00 -9.96837497e-01
-1.17848480e+00 7.48330772e-01 -1.71817198e-01 2.36924186e-01
2.49548927e-02 4.74180251e-01 9.65896487e-01 -9.26703334e-01
7.43890166e-01 1.43828905e+00 8.54173124e-01 1.15024602e+00
-1.19701469e+00 4.35824320e-02 -6.46482706e-01 3.98863792e-01
-1.29588056e+00 -3.20193291e-01 5.34599185e-01 -6.71586990e-01
1.25494766e+00 -1.85993508e-01 9.36740637e-01 1.49071360e+00
5.57812333e-01 8.82372677e-01 8.37772608e-01 -7.85431087e-01
7.62582779e-01 -8.17500148e-03 3.54992032e-01 1.09827328e+00
2.62502700e-01 3.40397418e-01 -6.81000352e-01 -2.88784087e-01
7.65748978e-01 4.81644630e-01 -4.48206032e-04 -3.85750145e-01
-9.23033476e-01 1.22430217e+00 1.07391566e-01 3.37899089e-01
-4.27478254e-01 7.81166375e-01 4.55067128e-01 4.65101637e-02
2.02417448e-01 9.25961584e-02 -2.66923636e-01 -2.54887432e-01
-1.06596458e+00 1.90861851e-01 1.11163342e+00 9.03155029e-01
1.02438986e+00 4.06617612e-01 7.12420791e-02 6.01136446e-01
6.64778709e-01 3.22228611e-01 8.38783979e-01 -1.41960049e+00
-5.50950229e-01 -8.84326845e-02 -8.93860608e-02 -4.73577440e-01
-2.17837393e-02 -4.47610728e-02 -1.16389823e+00 1.28658395e-02
1.70256436e-01 -9.62902904e-02 -8.12987328e-01 1.65695953e+00
-4.85093556e-02 4.09583688e-01 2.73732632e-01 6.48632705e-01
7.76318073e-01 8.07549417e-01 2.14069411e-01 -4.63496625e-01
1.31089365e+00 -7.79138267e-01 -5.99474847e-01 -2.90062606e-01
6.20081306e-01 -1.87364630e-02 1.09870851e+00 4.08309877e-01
-1.08667672e+00 -5.82355559e-01 -1.01452255e+00 -4.35339451e-01
-2.66943634e-01 -9.02436793e-01 9.00818050e-01 9.31940854e-01
-1.01045346e+00 1.01751423e+00 -1.20447457e+00 -6.15888774e-01
6.52620256e-01 -1.14769153e-01 1.80701085e-03 -1.95165768e-01
-1.29846299e+00 8.91547740e-01 5.13816416e-01 -4.24907625e-01
-1.60830665e+00 -5.30437112e-01 -9.32063699e-01 1.06072001e-01
2.29931891e-01 -1.23665345e+00 1.53786278e+00 -8.52226555e-01
-1.60638273e+00 1.06623626e+00 -3.30520242e-01 -8.34136069e-01
7.96256214e-02 -1.04160875e-01 -7.37373680e-02 9.32547599e-02
-4.14978921e-01 5.54035068e-01 8.04196417e-01 -8.30786765e-01
5.64844832e-02 -3.70398700e-01 -3.09624821e-01 -3.14580113e-01
-2.04730242e-01 -1.72372013e-01 1.75481215e-01 -1.40847489e-01
-3.72023582e-01 -7.43054092e-01 -3.02139670e-01 1.01703610e-02
1.19535692e-01 -7.94309497e-01 3.71937752e-01 -2.04302281e-01
7.68225491e-01 -1.90450025e+00 2.32937008e-01 -2.17587009e-01
4.25759256e-01 -2.07858514e-02 5.35662100e-02 5.86879671e-01
4.27874565e-01 1.71991691e-01 -2.44192839e-01 -2.43914068e-01
4.46582437e-01 7.49926329e-01 -3.44799727e-01 6.35749698e-01
-1.22083217e-01 1.29893041e+00 -1.19314516e+00 -7.86780894e-01
4.27567512e-01 6.00733340e-01 -7.31061161e-01 1.29187450e-01
-4.81887132e-01 -7.29074255e-02 -2.01207906e-01 2.99016625e-01
1.93095818e-01 -5.61825156e-01 8.78055245e-02 2.89082378e-01
2.45154440e-01 -3.13558173e-03 -8.93700659e-01 1.94118547e+00
-1.89395368e-01 9.16737735e-01 -3.91303867e-01 -1.19159567e+00
6.18978739e-01 4.42878187e-01 -7.47244107e-04 -2.53716469e-01
2.75009364e-01 -8.10490996e-02 8.48411694e-02 -6.04451835e-01
3.89400184e-01 -7.24542916e-01 -1.28308207e-01 5.47983050e-01
7.31098831e-01 -3.55345726e-01 -5.91709688e-02 6.19976163e-01
1.07440221e+00 2.85869002e-01 7.91338682e-01 -4.99182552e-01
-1.28784562e-02 -1.04081415e-01 5.55899560e-01 1.35511363e+00
-7.08286703e-01 5.14367759e-01 2.54121572e-01 -7.27096438e-01
-1.49451208e+00 -1.32210624e+00 1.06398299e-01 1.59248793e+00
-2.70807892e-01 -5.04266679e-01 -9.91342127e-01 1.60895824e-01
-4.22421545e-01 1.30713511e+00 -8.41820240e-01 -3.51788074e-01
-9.54954848e-02 -7.06447899e-01 8.04953992e-01 3.89532059e-01
3.98976803e-01 -1.16037655e+00 -1.00009990e+00 2.09559336e-01
1.79462835e-01 -6.66251481e-01 -8.07536691e-02 2.62674004e-01
-8.47219110e-01 -7.56456971e-01 -5.80344617e-01 -5.61995983e-01
2.60066390e-02 -2.25855514e-01 1.33057630e+00 -1.97998602e-02
-6.80419743e-01 7.90748298e-01 -1.14627615e-01 -5.53442597e-01
-7.73414016e-01 -3.76676202e-01 5.19716144e-01 -5.72494984e-01
1.10044587e+00 -7.32256532e-01 -5.06114185e-01 -3.13934714e-01
-8.35208058e-01 -1.89551383e-01 2.51217037e-01 1.11759841e+00
3.11790854e-01 -5.86805232e-02 1.08124735e-02 -8.86684299e-01
4.37307298e-01 -6.04754031e-01 -1.31722182e-01 3.12340766e-01
-6.03931606e-01 4.28561121e-01 5.47237337e-01 -3.95903707e-01
-1.16048884e+00 -3.79623920e-01 -6.50738925e-02 -5.26331723e-01
-4.03894931e-01 4.12558228e-01 2.14583009e-01 2.77250230e-01
1.30938566e+00 7.36916959e-01 -4.09819447e-02 -5.41327447e-02
6.77541137e-01 6.80061281e-01 5.22533298e-01 -7.04658031e-01
5.62468171e-01 5.05440652e-01 2.79988311e-02 -1.31187892e+00
-1.02526700e+00 -1.24922588e-01 -6.99293315e-01 -8.80973265e-02
1.14654458e+00 -9.13608968e-01 -9.41141605e-01 3.79459620e-01
-1.32042527e+00 -4.83775735e-01 -5.86561739e-01 4.15923178e-01
-9.72259939e-01 4.60080832e-01 -1.03786671e+00 -1.29073620e+00
-4.47678447e-01 -4.94332343e-01 7.31021166e-01 3.91785562e-01
-3.32333893e-01 -1.35326397e+00 7.56033003e-01 9.80210975e-02
3.66374701e-01 -7.60549540e-03 8.46153438e-01 -7.75584102e-01
-5.34588993e-01 1.58549801e-01 4.66081411e-01 3.24024677e-01
-5.74510157e-01 1.72626972e-01 -1.02681124e+00 -1.01335019e-01
2.43659869e-01 -8.01375568e-01 1.26905966e+00 5.65564752e-01
8.94505084e-01 -3.46552908e-01 -1.59252733e-02 4.71742898e-01
1.78112280e+00 -2.94597268e-01 9.62302446e-01 -1.38485119e-01
6.28546655e-01 3.91059041e-01 -2.81108499e-01 6.83408260e-01
3.10450673e-01 -2.30431616e-01 -8.85700528e-03 9.29476559e-01
2.64135927e-01 -4.96115834e-01 5.29112756e-01 1.29840136e+00
-5.35429418e-01 -7.94583634e-02 -1.13151360e+00 5.61522365e-01
-1.94923139e+00 -1.87983143e+00 8.83241296e-02 1.81350982e+00
1.00781012e+00 2.58641541e-01 1.42803594e-01 -3.26220334e-01
4.71105635e-01 1.64896980e-01 -6.51404917e-01 -4.96567667e-01
-4.81057838e-02 5.74266672e-01 2.06039965e-01 5.06199181e-01
-5.50414562e-01 1.09829390e+00 7.75216866e+00 9.57476854e-01
-3.41380477e-01 7.13602364e-01 2.92356610e-01 -2.39693090e-01
-4.41153944e-01 1.71578139e-01 -5.11066198e-01 1.85396194e-01
1.68332660e+00 -2.93349802e-01 8.08547795e-01 1.03955901e+00
-2.40565851e-01 5.84120937e-02 -1.47153258e+00 1.12673283e+00
1.39579803e-01 -1.65296018e+00 -2.91928370e-02 7.85944164e-02
8.83640409e-01 4.21087295e-01 -2.04262823e-01 8.25361609e-01
1.13628709e+00 -1.33183408e+00 4.80193704e-01 1.09502256e+00
2.64914870e-01 -7.40249336e-01 3.35814029e-01 1.06690657e+00
-6.91253364e-01 5.65003231e-02 -1.09169686e+00 -5.71954131e-01
-1.45588979e-01 4.21866715e-01 -1.75504699e-01 -2.76527852e-01
3.66527170e-01 5.49652219e-01 -8.75474140e-02 4.04170841e-01
-7.70897567e-02 7.33911633e-01 -1.50906697e-01 -7.52888560e-01
1.45968676e-01 -1.97657570e-01 4.20504689e-01 1.23009241e+00
5.87708801e-02 5.23593068e-01 -5.56050651e-02 1.55963337e+00
-1.03195176e-01 -2.29514152e-01 -7.62229800e-01 -4.66197908e-01
5.59862316e-01 8.91536653e-01 -4.37530726e-01 -8.07273507e-01
-5.37277699e-01 1.09594190e+00 3.16049367e-01 2.74470687e-01
-5.44213772e-01 -3.91705811e-01 5.57406306e-01 -2.09106639e-01
3.95989835e-01 -3.45129609e-01 -6.36774004e-02 -1.53526366e+00
-8.11316073e-01 -6.60749078e-01 2.18807027e-01 -8.11685324e-01
-1.52044523e+00 -8.28212965e-03 -1.17155902e-01 -5.19139349e-01
-6.79979026e-01 -8.17437172e-01 -8.78972054e-01 4.25497264e-01
-7.49769032e-01 -9.61596012e-01 1.16903119e-01 6.96852505e-01
6.36818171e-01 -4.41395551e-01 1.26758838e+00 -4.40115243e-01
-3.28768194e-01 1.78074732e-01 3.75223875e-01 4.41806436e-01
9.97010060e-03 -1.41373825e+00 4.39415216e-01 6.16604209e-01
6.28871202e-01 1.08659387e+00 1.16974962e+00 -4.74772871e-01
-1.75221157e+00 -6.06445193e-01 4.99171406e-01 -6.90660954e-01
7.46198893e-01 -4.34901148e-01 -9.75557804e-01 8.17878008e-01
5.61767876e-01 2.49992371e-01 9.64382291e-01 2.36811996e-01
-6.83193147e-01 4.64792430e-01 -1.06419051e+00 5.72925270e-01
8.96604240e-01 -1.14767694e+00 -1.37422323e+00 3.35752755e-01
5.99890471e-01 2.97073454e-01 -9.19519424e-01 -3.46863657e-01
6.31841898e-01 -1.08459198e+00 9.03842688e-01 -1.00896621e+00
7.40658164e-01 2.51138180e-01 -5.46629727e-01 -1.16470909e+00
-5.96252263e-01 -9.13739979e-01 -7.23112226e-01 5.28065264e-01
-2.56320268e-01 -2.16541544e-01 7.61377513e-01 8.13319802e-01
1.83876380e-01 -2.84712166e-01 -8.75482440e-01 -8.27710032e-01
7.59497821e-01 -6.15019143e-01 -7.52052739e-02 8.13425362e-01
3.00545454e-01 7.24033713e-01 -6.38992488e-01 -3.31845313e-01
1.44035661e+00 -4.50780094e-01 5.41553080e-01 -1.65165234e+00
-9.32173312e-01 -2.88329244e-01 -4.88717914e-01 -8.35294366e-01
4.54360276e-01 -9.62606132e-01 2.24992126e-01 -1.36915410e+00
9.62095201e-01 7.37316310e-01 -2.27583796e-01 7.11303800e-02
1.38986096e-01 8.06230009e-02 -8.62975232e-03 2.44598195e-01
-9.38757837e-01 7.78069317e-01 6.87495947e-01 -6.75374344e-02
2.64300972e-01 -6.41814947e-01 -5.35674870e-01 1.13893938e+00
4.23614591e-01 -7.28804231e-01 -1.21750697e-01 -3.21813859e-02
7.30314136e-01 -4.18992676e-02 7.78270304e-01 -8.91355038e-01
6.15305543e-01 -4.78025794e-01 4.82175022e-01 -1.33210957e-01
1.58819899e-01 -2.82629788e-01 -1.69540107e-01 9.26215649e-01
-5.42902589e-01 -6.42942488e-01 -4.84510899e-01 9.55220222e-01
3.79550129e-01 -7.80801654e-01 1.12446654e+00 -9.81160641e-01
-9.75102425e-01 4.43260401e-01 -7.67716050e-01 6.57399535e-01
6.30590856e-01 6.48428351e-02 -3.25225472e-01 -4.18027699e-01
-6.92446947e-01 -1.95460737e-01 5.33259630e-01 -1.10052206e-01
7.60872364e-01 -1.19908226e+00 -7.83443630e-01 -1.30446613e-01
-4.24407423e-02 -4.10034955e-01 3.99175912e-01 4.89299417e-01
-4.97605264e-01 2.05138743e-01 -1.67468354e-01 -5.10626256e-01
-6.30433917e-01 7.38416970e-01 4.43103433e-01 1.86743438e-01
-6.70006692e-01 9.35168386e-01 -2.19975501e-01 -3.86449158e-01
1.11535706e-01 6.22871928e-02 3.58115464e-01 -2.76569724e-01
8.02344561e-01 6.07407928e-01 -8.95342648e-01 -2.76059896e-01
-1.93289071e-01 2.26557940e-01 7.42278481e-03 -4.21339810e-01
1.54084945e+00 -1.19354866e-01 -1.85357273e-01 1.42678392e+00
1.26515710e+00 -6.43541634e-01 -1.13200223e+00 -4.93399292e-01
-2.12811410e-01 1.96870249e-02 3.03695768e-01 -7.28437155e-02
-1.93772186e-02 1.53589702e+00 5.63192487e-01 2.68841088e-01
2.34163210e-01 2.38064066e-01 8.26500893e-01 1.20896649e+00
6.96979463e-01 -1.53790760e+00 3.95041823e-01 7.51960039e-01
1.87866047e-01 -1.49867713e+00 -6.23885132e-02 5.95322907e-01
-5.86214423e-01 1.29176307e+00 2.83882201e-01 -7.18483150e-01
9.14307117e-01 1.37567371e-01 -7.02210486e-01 -3.33651930e-01
-1.34368229e+00 -7.68922642e-02 1.01270273e-01 7.41092622e-01
4.85801309e-01 1.66556060e-01 2.03278258e-01 6.38938785e-01
-1.86102778e-01 5.33026516e-01 7.25311816e-01 7.40155280e-01
-1.21880889e+00 -4.59548652e-01 -1.52373379e-02 2.73381978e-01
-3.47207069e-01 -3.16752791e-01 -1.29034281e-01 4.93333489e-01
1.20185986e-02 5.98390043e-01 2.67065674e-01 -1.26097992e-01
-5.49081564e-01 6.25254035e-01 1.09752691e+00 -8.30450654e-01
-3.22980992e-02 -1.59038082e-01 -4.68752861e-01 -4.21831489e-01
-6.78182006e-01 -5.73850632e-01 -1.36441731e+00 -9.39830661e-01
-8.53606611e-02 8.57997090e-02 2.29873762e-01 1.43862092e+00
-1.79136083e-01 1.84708282e-01 4.48709540e-02 -6.88600898e-01
-1.13985062e+00 -1.19000363e+00 -1.02406001e+00 3.01078796e-01
2.88561165e-01 -2.70132869e-01 -6.54914379e-01 3.37139487e-01] | [5.982165813446045, 4.667880058288574] |
5a2800dd-ee09-4960-8c38-0663a4946ddd | examining-the-presence-of-gender-bias-in | 1902.00496 | null | http://arxiv.org/abs/1902.00496v1 | http://arxiv.org/pdf/1902.00496v1.pdf | Examining the Presence of Gender Bias in Customer Reviews Using Word Embedding | Humans have entered the age of algorithms. Each minute, algorithms shape
countless preferences from suggesting a product to a potential life partner. In
the marketplace algorithms are trained to learn consumer preferences from
customer reviews because user-generated reviews are considered the voice of
customers and a valuable source of information to firms. Insights mined from
reviews play an indispensable role in several business activities ranging from
product recommendation, targeted advertising, promotions, segmentation etc. In
this research, we question whether reviews might hold stereotypic gender bias
that algorithms learn and propagate Utilizing data from millions of
observations and a word embedding approach, GloVe, we show that algorithms
designed to learn from human language output also learn gender bias. We also
examine why such biases occur: whether the bias is caused because of a negative
bias against females or a positive bias for males. We examine the impact of
gender bias in reviews on choice and conclude with policy implications for
female consumers, especially when they are unaware of the bias, and the ethical
implications for firms. | ['S. Rathee', 'H. Mishra', 'A. Mishra'] | 2019-02-01 | null | null | null | null | ['product-recommendation'] | ['miscellaneous'] | [ 7.63724372e-02 5.70355296e-01 -1.02163279e+00 -9.19676661e-01
1.01016209e-01 -5.19968212e-01 8.14803302e-01 4.82723296e-01
-6.81002021e-01 6.88218415e-01 6.71756387e-01 -7.48699486e-01
5.00948429e-02 -8.86508584e-01 -6.57454252e-01 -3.50804150e-01
3.55975181e-01 7.11936593e-01 -5.81741154e-01 -4.95317310e-01
7.74454951e-01 2.49885265e-02 -1.34374022e+00 4.54855859e-02
6.10046864e-01 7.91698873e-01 -3.20678294e-01 3.47431451e-01
-3.24816048e-01 3.59819263e-01 -3.34209979e-01 -1.22492480e+00
4.41297263e-01 -1.96430549e-01 -3.86487901e-01 9.29327086e-02
3.87264431e-01 -1.79236203e-01 2.26306751e-01 1.28285575e+00
4.13565636e-01 -1.97880074e-01 8.10792685e-01 -1.38682759e+00
-1.21881914e+00 1.13419676e+00 -8.14534068e-01 1.70509070e-01
1.21398427e-01 5.83839752e-02 1.43087327e+00 -7.48827755e-01
7.78709650e-01 1.64752245e+00 3.36413950e-01 8.19517374e-01
-1.06346560e+00 -9.21396017e-01 4.00245398e-01 -2.62120485e-01
-4.73722339e-01 -4.08640593e-01 8.41035903e-01 -5.28038561e-01
4.03460264e-01 3.46950203e-01 8.90698314e-01 1.62114680e+00
6.41606688e-01 5.81215560e-01 1.19804430e+00 -2.42313117e-01
2.65854299e-01 1.07740259e+00 4.51880217e-01 1.85781419e-01
1.18659699e+00 2.46821508e-01 -8.63929093e-01 -3.87549758e-01
3.15914422e-01 7.66354352e-02 2.89872408e-01 -2.13241875e-01
-8.39252889e-01 1.56197011e+00 1.57257274e-01 -9.38474834e-02
-4.38555986e-01 1.78742886e-01 2.93991387e-01 4.54845637e-01
7.14977324e-01 7.09777236e-01 -8.72582197e-01 -2.51564443e-01
-4.39684272e-01 2.69835860e-01 1.00078750e+00 6.45307481e-01
5.60176015e-01 -3.52352075e-02 3.77311826e-01 4.91211653e-01
8.24229717e-01 7.32828200e-01 8.68712902e-01 -4.99794036e-01
2.53531575e-01 5.95723629e-01 1.46454886e-01 -1.52715611e+00
-2.95392442e-02 -4.56269354e-01 -3.27654481e-01 -1.11210361e-01
3.89274716e-01 -4.43473399e-01 -6.67058051e-01 1.55682504e+00
1.59249783e-01 -7.66786695e-01 -1.68790013e-01 7.66203642e-01
4.30259138e-01 3.66332024e-01 2.51206785e-01 -8.70482698e-02
1.41363585e+00 -5.21457553e-01 -6.20034873e-01 -8.84575903e-01
5.88466465e-01 -8.05614591e-01 1.06401742e+00 5.74614704e-01
-7.33502626e-01 -3.42491567e-01 -9.66650307e-01 4.08970304e-02
-6.09813333e-01 -1.92027558e-02 1.14957643e+00 1.13418019e+00
-4.74378735e-01 5.34106672e-01 -3.76782238e-01 -6.76603556e-01
4.37106758e-01 4.99551922e-01 -3.90568413e-02 2.62611777e-01
-1.11256599e+00 9.36437309e-01 -3.28327626e-01 -1.28080204e-01
-2.88799793e-01 -3.96524578e-01 -9.36159134e-01 -1.34930193e-01
1.09213926e-01 -4.37906981e-01 1.25530875e+00 -1.75711071e+00
-9.76748168e-01 1.07284319e+00 -2.86819279e-01 -5.72311878e-01
3.39770794e-01 -3.51770550e-01 -6.44505441e-01 -6.26390100e-01
3.10313672e-01 6.44135594e-01 1.01497138e+00 -1.27313745e+00
-7.78738558e-01 -8.25061917e-01 -6.41405284e-02 -1.24266505e-01
-3.86355102e-01 1.17413409e-01 6.44078016e-01 -4.15229231e-01
-1.66024402e-01 -1.14945734e+00 -3.40348393e-01 -3.13042849e-01
-2.93106377e-01 -4.50059831e-01 4.77638870e-01 -4.38886136e-01
1.02292609e+00 -2.13106728e+00 -5.13834655e-01 4.94436026e-01
2.08329469e-01 -2.92944163e-01 2.09818915e-01 5.58467925e-01
3.44891027e-02 7.95064390e-01 4.12411928e-01 -1.10779129e-01
4.95186269e-01 1.84414029e-01 -6.87485874e-01 4.27976340e-01
1.71008185e-01 7.53065050e-01 -9.30013359e-01 -2.47520864e-01
-3.40469837e-01 1.93076938e-01 -5.94406545e-01 -1.93274513e-01
-6.11134693e-02 -2.46603787e-02 -6.18747175e-01 7.48260200e-01
4.51296270e-01 -2.20634192e-01 4.31761354e-01 -1.35870963e-01
-6.70121145e-03 5.64453781e-01 -7.45718122e-01 8.25675726e-01
-4.21831727e-01 7.30290174e-01 1.60868675e-01 -6.23694897e-01
1.17732739e+00 -1.35128751e-01 -1.46573201e-01 -7.53964663e-01
6.87846839e-01 4.28412050e-01 5.80926418e-01 -3.17185879e-01
1.05932486e+00 -3.69513065e-01 -1.97932705e-01 9.08914208e-01
-2.93486416e-01 -8.43094289e-03 1.84169292e-01 1.54005960e-01
3.81177127e-01 -4.08225238e-01 3.24167669e-01 -4.80710566e-01
3.13475281e-02 3.27897705e-02 6.74337029e-01 4.60526943e-01
-2.16776982e-01 1.19951218e-01 6.55946493e-01 -6.95220351e-01
-8.72713983e-01 -7.82022119e-01 -2.30587348e-02 1.36880267e+00
8.99837688e-02 -2.96028435e-01 -1.59980848e-01 -7.10729599e-01
5.85320592e-01 1.06618738e+00 -9.45532560e-01 -4.47966814e-01
-8.83795470e-02 -8.57788563e-01 -4.88080457e-02 5.11882007e-01
9.01745073e-03 -8.49633813e-01 -7.38030791e-01 -1.17339179e-01
4.25759256e-01 -6.50325894e-01 -7.00680733e-01 2.06366628e-01
-9.26531255e-01 -8.59935284e-01 -3.53210777e-01 -5.05584359e-01
8.81964326e-01 -2.29327232e-01 1.35169566e+00 1.64362080e-02
-1.45569727e-01 4.63016033e-01 -1.54752284e-01 -1.25484884e+00
-4.20511335e-01 2.97935307e-01 4.28697497e-01 2.43166387e-01
1.24294281e+00 -4.55479711e-01 -8.50085258e-01 2.00963274e-01
-5.90602279e-01 -4.57565457e-01 6.87351763e-01 5.61688423e-01
-3.24343681e-01 -3.04529727e-01 8.16265523e-01 -1.75624669e+00
1.16444731e+00 -7.57834971e-01 -2.70662427e-01 -1.46373510e-01
-1.61550629e+00 1.62599608e-01 1.37141168e-01 -4.73672271e-01
-9.90379512e-01 -2.30553836e-01 2.07950085e-01 4.42404836e-01
1.72973931e-01 3.56224090e-01 2.75810044e-02 4.39932287e-01
7.93669105e-01 -3.50033402e-01 2.64272958e-01 -1.93328589e-01
5.15941322e-01 8.07848692e-01 -1.88737690e-01 -5.35807133e-01
7.57733703e-01 4.42096770e-01 -5.28055668e-01 -5.87838233e-01
-6.00040853e-01 -1.66854367e-01 -8.55814293e-02 -1.70906663e-01
7.28877902e-01 -6.70645356e-01 -6.79401100e-01 -2.95877308e-01
-9.79422271e-01 2.31530115e-01 -3.23488921e-01 7.11869061e-01
1.49629619e-02 -3.12968761e-01 -4.65476424e-01 -1.10535252e+00
-3.56549203e-01 -8.93022418e-01 6.26140654e-01 5.84288776e-01
-1.01300704e+00 -1.07161152e+00 -5.32369949e-02 3.69515002e-01
6.27342403e-01 -1.70493454e-01 1.01898193e+00 -1.29099607e+00
-9.12354961e-02 -2.30684653e-01 -2.97068469e-02 2.47559085e-01
3.15751523e-01 3.75552237e-01 -9.48084772e-01 -3.86075750e-02
-2.09141210e-01 -2.72882193e-01 6.14289224e-01 4.02930111e-01
5.95171273e-01 -5.93282044e-01 -4.25720513e-01 1.97837967e-02
1.19327033e+00 3.57217908e-01 2.09007964e-01 3.57692659e-01
2.93033868e-01 1.13868558e+00 6.29223406e-01 6.44527018e-01
4.92127955e-01 -2.98501253e-01 2.95040369e-01 1.41602054e-01
6.30104721e-01 -6.13292813e-01 5.68984210e-01 7.44655311e-01
8.02667588e-02 -1.51749969e-01 -5.69290996e-01 7.42747962e-01
-1.61807120e+00 -6.71755493e-01 9.79523659e-02 2.13015437e+00
8.17293942e-01 6.09241903e-01 1.77937135e-01 -2.85710599e-02
6.21242285e-01 2.44050562e-01 -7.08908021e-01 -1.38547277e+00
1.56719521e-01 1.58885837e-01 1.09361446e+00 4.26917940e-01
-4.81174797e-01 5.77181578e-01 6.58431053e+00 2.70259470e-01
-1.37674749e+00 -1.54476985e-01 1.34559321e+00 -4.75949096e-03
-1.11822712e+00 9.91215855e-02 -9.12499011e-01 4.06049311e-01
9.45452511e-01 -5.59984386e-01 1.71108663e-01 1.12351918e+00
1.57371461e-01 -3.92440818e-02 -1.28160048e+00 8.37352931e-01
1.72004715e-01 -9.19619322e-01 4.47399355e-02 5.99820495e-01
6.54030859e-01 -3.48475605e-01 8.02558184e-01 3.27123106e-01
7.31917262e-01 -1.30542469e+00 7.93654740e-01 3.48037370e-02
2.41402060e-01 -8.30204427e-01 7.16581166e-01 7.49898329e-02
-3.08381289e-01 -3.04392457e-01 -5.17274022e-01 -6.28368497e-01
1.84178755e-01 8.95574272e-01 -1.04366660e+00 -2.56565034e-01
3.65164608e-01 7.95548737e-01 -6.40870929e-01 2.02502817e-01
-1.13604747e-01 6.91989541e-01 4.57738005e-02 -5.72482765e-01
1.40686586e-01 -4.40953970e-01 2.10811660e-01 9.06662941e-01
3.41119617e-01 -3.35205466e-01 -4.49410588e-01 8.31308722e-01
-2.61389285e-01 3.45620811e-01 -9.58346069e-01 -9.41570401e-01
4.24184978e-01 1.48533356e+00 -8.62885714e-01 -4.68169528e-05
-3.51607978e-01 6.04049921e-01 -3.23974431e-01 3.27502996e-01
-4.54892755e-01 -2.28572413e-01 9.37239051e-01 4.25391734e-01
-5.78431785e-02 2.25769430e-01 -7.58458614e-01 -8.37476015e-01
-2.95909137e-01 -1.17264509e+00 3.31334886e-03 -3.60894591e-01
-1.48539019e+00 2.05419838e-01 -4.19541895e-01 -5.84160328e-01
-3.64502639e-01 -9.36018825e-01 -6.67587221e-01 5.44871926e-01
-1.15358162e+00 -7.90373862e-01 3.33883017e-01 -2.13050991e-01
5.04864573e-01 -3.48580033e-01 5.11637926e-01 8.14706758e-02
-6.54429197e-02 5.85699439e-01 -2.54222751e-01 5.83314002e-02
1.09214473e+00 -1.38635194e+00 4.85328108e-01 1.72990918e-01
2.19233543e-01 1.11247611e+00 1.06217682e+00 -7.13550150e-01
-1.44204998e+00 -5.77110112e-01 1.54532051e+00 -7.38621593e-01
5.86442828e-01 -3.35124344e-01 2.23192479e-02 8.14983368e-01
5.05278945e-01 -5.49055398e-01 1.24112833e+00 5.64207137e-01
-3.25048476e-01 -3.10782731e-01 -1.14500475e+00 7.21521258e-01
6.97255313e-01 -4.38541591e-01 -6.90827608e-01 8.05107728e-02
7.23875761e-01 2.59673327e-01 -3.31232071e-01 -7.09559098e-02
1.05074823e+00 -8.56549799e-01 5.53680837e-01 -9.22190070e-01
9.10519958e-01 4.72508609e-01 -2.34461218e-01 -1.49480760e+00
-5.64892963e-02 -4.31042671e-01 4.46972787e-01 1.27958155e+00
1.00861251e+00 -9.53899503e-01 8.92813861e-01 1.28674281e+00
5.12552619e-01 -7.89508104e-01 -3.96872908e-01 -1.31070226e-01
1.95064038e-01 -1.38225108e-01 9.15595770e-01 1.02045524e+00
-4.33658324e-02 7.60010064e-01 -2.83565789e-01 -3.18689138e-01
5.04225552e-01 1.20719761e-01 7.86105275e-01 -1.52142513e+00
-2.38131702e-01 -4.54624832e-01 -2.08576974e-02 -8.67046058e-01
6.70148954e-02 -6.07803226e-01 -2.37263009e-01 -1.13527751e+00
4.75674914e-03 -2.65959829e-01 -2.29798526e-01 -2.82649875e-01
1.64810702e-01 -2.71577220e-02 1.99976921e-01 -2.93619692e-01
-1.70259386e-01 1.37158364e-01 1.21995759e+00 -1.90795839e-01
2.07021832e-02 2.33075649e-01 -1.81776416e+00 8.49415243e-01
7.85611391e-01 -6.19095802e-01 -6.78944707e-01 -2.16394678e-01
1.42982697e+00 -5.81545889e-01 -1.40165882e-02 5.81359724e-03
4.55308100e-03 -3.91630203e-01 4.38976377e-01 -1.75727352e-01
1.23166002e-01 -8.72925401e-01 -1.51134551e-01 4.63643432e-01
-6.35478199e-01 6.93306446e-01 -2.55701333e-01 6.66013122e-01
-4.13662940e-02 -2.46939003e-01 1.71965092e-01 -3.59221816e-01
-2.22911894e-01 8.86835679e-02 -5.85131168e-01 7.66245201e-02
6.09793663e-01 -2.30115354e-01 -3.07027042e-01 -7.30964124e-01
-4.57824022e-01 1.95039764e-01 5.24900973e-01 7.39551723e-01
2.38262445e-01 -1.19258618e+00 -5.16123891e-01 1.63424268e-01
1.33727834e-01 -6.15459859e-01 -5.87516904e-01 4.60191667e-01
-1.32332042e-01 3.71737331e-01 -1.10899463e-01 3.41078714e-02
-1.09046876e+00 2.38590404e-01 -2.10971281e-01 6.47434220e-02
2.52386689e-01 1.11624551e+00 1.04093671e-01 -5.50934851e-01
1.79195516e-02 -7.20335245e-01 -3.88021320e-01 6.58592045e-01
2.75311172e-01 3.70834589e-01 -3.59186232e-01 -5.14101326e-01
-2.88711548e-01 3.26633394e-01 -3.04808348e-01 -3.70446116e-01
1.33021498e+00 -1.58106655e-01 -1.29109874e-01 9.34858739e-01
8.80503297e-01 5.00272155e-01 -6.18360817e-01 1.72995061e-01
1.99903965e-01 -8.02515268e-01 -2.11416125e-01 -8.31147909e-01
-1.13125515e+00 8.80112469e-01 4.51507121e-01 4.11176294e-01
3.83888632e-01 -1.30262032e-01 7.46831715e-01 2.34471664e-01
2.59578764e-01 -1.53784025e+00 -4.50016744e-02 4.77974713e-02
5.28598189e-01 -1.55867159e+00 1.16445400e-01 -1.20552234e-01
-1.04763448e+00 9.13544476e-01 6.22746050e-01 -1.83843568e-01
1.03197491e+00 -2.29051605e-01 4.00823355e-01 -3.06354910e-01
-1.12025058e+00 3.72688293e-01 3.43008190e-02 4.94738877e-01
9.92160141e-01 3.70782256e-01 -1.04970217e+00 1.23813856e+00
-6.98562384e-01 -1.49668261e-01 6.61497533e-01 6.36159658e-01
-4.17776018e-01 -1.48603225e+00 -3.84772755e-02 9.61977661e-01
-7.44618058e-01 -1.39055490e-01 -8.33272636e-01 6.75410509e-01
2.68148661e-01 1.01004481e+00 1.47685915e-01 -5.23154855e-01
-2.20485665e-02 1.68984130e-01 9.74881947e-02 -6.81749225e-01
-7.89691627e-01 -3.25447977e-01 4.58970815e-01 -1.25927821e-01
-2.17852548e-01 -8.40970516e-01 -8.89773905e-01 -5.07270336e-01
-3.75923604e-01 4.45287019e-01 9.20087993e-01 7.35025048e-01
4.96105641e-01 -1.19287170e-01 5.89611590e-01 -2.32304052e-01
-9.54085052e-01 -9.97044921e-01 -8.71419668e-01 4.21095639e-01
1.58871576e-01 -4.78960603e-01 -7.51174986e-01 -4.32761386e-02] | [9.31798267364502, 10.135830879211426] |
47caf140-c3d8-448b-b2b4-72c8c429caf5 | lemma-bootstrapping-high-level-mathematical | 2211.08671 | null | https://arxiv.org/abs/2211.08671v1 | https://arxiv.org/pdf/2211.08671v1.pdf | LEMMA: Bootstrapping High-Level Mathematical Reasoning with Learned Symbolic Abstractions | Humans tame the complexity of mathematical reasoning by developing hierarchies of abstractions. With proper abstractions, solutions to hard problems can be expressed concisely, thus making them more likely to be found. In this paper, we propose Learning Mathematical Abstractions (LEMMA): an algorithm that implements this idea for reinforcement learning agents in mathematical domains. LEMMA augments Expert Iteration with an abstraction step, where solutions found so far are revisited and rewritten in terms of new higher-level actions, which then become available to solve new problems. We evaluate LEMMA on two mathematical reasoning tasks--equation solving and fraction simplification--in a step-by-step fashion. In these two domains, LEMMA improves the ability of an existing agent, both solving more problems and generalizing more effectively to harder problems than those seen during training. | ['Armando Solar-Lezama', 'Noah Goodman', 'Omar Costilla-Reyes', 'Gabriel Poesia', 'Zhening Li'] | 2022-11-16 | null | null | null | null | ['mathematical-reasoning'] | ['natural-language-processing'] | [ 6.37065545e-02 6.08924031e-01 1.25868738e-01 -8.43285471e-02
-3.70979220e-01 -7.06127346e-01 3.77381235e-01 4.96514350e-01
-4.35317516e-01 1.30689180e+00 -5.71033806e-02 -4.78859276e-01
-2.56848335e-01 -1.19709122e+00 -7.38260806e-01 -3.50932419e-01
-3.35372597e-01 8.20005834e-01 1.74119800e-01 -5.63802063e-01
3.30632657e-01 6.79611564e-01 -1.36155844e+00 3.71552378e-01
1.20486379e+00 3.56101245e-01 8.98703411e-02 9.21981156e-01
-2.59350926e-01 1.32960951e+00 -7.21469939e-01 -4.52366143e-01
2.98218817e-01 -4.70806152e-01 -1.02181089e+00 -1.27544239e-01
3.02513093e-01 -4.85745728e-01 -3.51349026e-01 8.63997161e-01
-3.28337014e-01 4.40158725e-01 5.99336982e-01 -1.11039460e+00
-8.42513561e-01 1.01913691e+00 -1.77234024e-01 -7.54759014e-02
5.54892600e-01 4.24162060e-01 1.00824738e+00 -3.30609441e-01
5.22365332e-01 1.54940081e+00 6.98615789e-01 8.71094108e-01
-1.33123243e+00 -5.75543821e-01 2.95421153e-01 7.76442289e-02
-9.31046188e-01 5.88910133e-02 4.16981369e-01 -2.75788575e-01
1.37769735e+00 2.53850639e-01 1.01290917e+00 1.92617863e-01
1.72037631e-01 7.69488633e-01 1.15475762e+00 -4.48047817e-01
4.66939569e-01 2.59741575e-01 1.77678689e-01 1.15489388e+00
6.11646175e-01 2.17465848e-01 -2.84164429e-01 -1.43018365e-01
9.22593951e-01 1.22260384e-01 -8.50457475e-02 -3.32823664e-01
-6.32939756e-01 9.47913885e-01 5.64328849e-01 2.77321965e-01
-7.15386987e-01 4.89414573e-01 1.48486584e-01 7.55566239e-01
-5.06473929e-02 1.34191513e+00 -5.49210846e-01 1.30128965e-01
-6.38924956e-01 8.08625579e-01 1.19278085e+00 6.92116499e-01
9.04361129e-01 2.85769403e-01 1.07308559e-01 3.97116810e-01
1.84791628e-02 4.44989800e-01 2.37656444e-01 -1.48032904e+00
3.18297595e-01 8.70518446e-01 1.61643699e-01 -5.12854755e-01
-3.48271251e-01 -4.89127636e-01 -4.42667663e-01 8.65496695e-01
2.96377718e-01 -3.57543439e-01 -7.54579127e-01 1.96344340e+00
2.82485723e-01 -9.44600403e-02 4.64527398e-01 5.47614157e-01
2.91886032e-01 1.08682930e+00 3.41665804e-01 -2.62691677e-01
1.09481192e+00 -9.54340696e-01 -4.37086016e-01 -7.96035603e-02
9.64719594e-01 -5.52034602e-02 8.09069872e-01 8.65884900e-01
-1.76922834e+00 -4.39499199e-01 -8.92304897e-01 -3.21832359e-01
-5.36469460e-01 -5.70280075e-01 1.13866985e+00 3.94568890e-01
-1.14609671e+00 1.08420646e+00 -4.56280738e-01 8.47895741e-02
5.96684515e-01 4.42906052e-01 -1.26545830e-02 -2.08280191e-01
-1.23897517e+00 1.30380130e+00 4.89664584e-01 -4.43452924e-01
-1.09365332e+00 -1.33915961e+00 -7.89326549e-01 4.29844171e-01
6.33373380e-01 -1.12356412e+00 1.78494465e+00 -9.10271108e-01
-1.55532777e+00 5.03455579e-01 2.72896923e-02 -7.96621561e-01
4.20956433e-01 -2.59980619e-01 1.05887726e-01 1.56371921e-01
-4.05188240e-02 7.99589157e-01 8.58125687e-01 -1.47767043e+00
-9.57856059e-01 -2.08844721e-01 1.14035809e+00 1.37482554e-01
-8.05925429e-02 -4.99967486e-01 1.71366602e-01 -3.52887183e-01
-1.09596871e-01 -8.29338491e-01 -5.38485110e-01 1.54313609e-01
2.61617184e-01 -6.74166560e-01 2.87469000e-01 -6.96825325e-01
1.14913654e+00 -1.83533108e+00 9.25028145e-01 2.65896291e-01
9.29718375e-01 5.06004810e-01 -6.99751154e-02 3.19132835e-01
-1.19536996e-01 2.19876170e-01 -1.96219817e-01 -9.76516679e-02
3.57287198e-01 6.06510282e-01 -4.50035721e-01 -2.85610735e-01
2.28492543e-01 1.01900446e+00 -1.23619986e+00 -3.18377942e-01
1.05253682e-01 -1.15971781e-01 -9.43853557e-01 3.99266392e-01
-8.63926470e-01 -2.22752076e-02 -4.13910240e-01 4.19730097e-01
4.58245933e-01 -1.05828576e-01 9.66045111e-02 3.72614235e-01
3.34036276e-02 3.03483605e-01 -1.17723322e+00 1.16184938e+00
-1.08260345e+00 3.89289439e-01 -1.29926533e-01 -9.84569490e-01
5.24974942e-01 7.27091730e-02 -7.00007752e-02 -6.64581060e-01
-1.46073624e-01 6.11416958e-02 2.31091097e-01 -4.76027727e-01
3.50173622e-01 -3.49329233e-01 -6.28964454e-02 5.87348402e-01
-2.22558677e-01 -9.02060747e-01 8.61746073e-01 4.82971191e-01
1.06467867e+00 6.14612289e-02 4.03776139e-01 -3.86664160e-02
5.74433386e-01 3.53581965e-01 4.05532449e-01 1.20106566e+00
2.75177151e-01 -4.44883168e-01 7.29019582e-01 -8.51752043e-01
-1.22385037e+00 -1.29312313e+00 4.59786177e-01 9.70487535e-01
-3.76437217e-01 -5.73040128e-01 -8.54500413e-01 -5.47164142e-01
4.34867144e-01 1.16003752e+00 -6.00194454e-01 -3.44692528e-01
-8.00928533e-01 2.32839167e-01 3.89058322e-01 6.58436656e-01
5.46997011e-01 -1.31364596e+00 -1.16556215e+00 4.52871531e-01
4.00877297e-01 -6.85014307e-01 1.59417436e-01 3.32766294e-01
-1.27612567e+00 -8.68933439e-01 -6.16802692e-01 -5.89171350e-01
8.13109040e-01 8.60476047e-02 1.13090718e+00 8.85456145e-01
-4.32586581e-01 5.20959198e-01 -1.54132843e-01 -6.26053154e-01
-8.42098951e-01 -1.37174115e-01 -7.67274527e-03 -9.56342399e-01
1.29885182e-01 -6.15755081e-01 6.89093396e-02 -4.66053128e-01
-8.15070093e-01 1.17783353e-01 6.11726582e-01 8.01140249e-01
1.56033814e-01 7.23083019e-01 3.55545461e-01 -9.49956298e-01
1.02720141e+00 -2.81879246e-01 -9.68276978e-01 3.88512999e-01
-4.53735739e-01 6.04877353e-01 1.23753214e+00 -5.81983984e-01
-1.07686615e+00 -2.55220413e-01 4.07169044e-01 -2.79151291e-01
-3.27039771e-02 6.42929077e-01 2.74712652e-01 -1.40932471e-01
9.11481977e-01 2.86405504e-01 -7.55876303e-02 -2.00653840e-02
5.45036197e-01 -2.65979376e-02 4.71659780e-01 -1.38923073e+00
1.24220943e+00 -1.10278860e-01 2.89820284e-01 -7.04253137e-01
-8.59296143e-01 1.34139583e-01 -1.01147786e-01 9.44694281e-02
4.57838923e-01 -5.84764659e-01 -1.11729729e+00 1.15402035e-01
-1.18561494e+00 -8.22559059e-01 -6.70716345e-01 7.58698359e-02
-5.28350472e-01 2.17201665e-01 -6.22839034e-01 -1.07880032e+00
8.77107494e-03 -9.88591373e-01 5.08392334e-01 5.94835043e-01
-6.12119496e-01 -1.06789768e+00 5.54785989e-02 9.75935832e-02
3.99099588e-01 -5.42439036e-02 1.62164950e+00 -3.64472985e-01
-8.96086335e-01 6.75155595e-02 -7.55730122e-02 2.97274351e-01
-2.78449327e-01 -1.90476343e-01 -5.67799568e-01 -4.58437353e-02
-8.24227408e-02 -5.37028134e-01 6.48961067e-01 7.77690932e-02
1.30702126e+00 -5.38033009e-01 -1.60517633e-01 3.88282418e-01
1.07207799e+00 2.67857522e-01 7.19188273e-01 4.49905008e-01
3.08562100e-01 3.52794319e-01 3.20779264e-01 5.03801346e-01
2.54610986e-01 1.19984895e-01 4.87550460e-02 2.41385754e-02
1.10339336e-01 -1.81041688e-01 8.51070285e-02 1.80925712e-01
-5.27176082e-01 1.89790577e-01 -9.02376711e-01 4.59449440e-01
-1.79849982e+00 -1.05387235e+00 2.26696968e-01 1.78281999e+00
1.34240949e+00 3.73821110e-01 2.57568449e-01 1.76767662e-01
1.19281128e-01 -4.61011708e-01 -5.90406537e-01 -1.07763994e+00
4.44920689e-01 9.35040295e-01 6.79406598e-02 1.01536500e+00
-3.64227921e-01 1.19859564e+00 6.43307495e+00 2.81670779e-01
-6.61317825e-01 -3.38933945e-01 1.34397298e-01 1.04207344e-01
-5.63409567e-01 1.67079896e-01 -5.66667378e-01 -8.46830085e-02
8.04285109e-01 -4.33225840e-01 1.21545851e+00 1.04205811e+00
-3.87502879e-01 -2.66328424e-01 -1.64177918e+00 5.57858646e-01
-4.18643087e-01 -1.52760184e+00 5.97880006e-01 -3.38492543e-01
8.77374470e-01 -9.91952479e-01 2.08191395e-01 9.21009302e-01
9.14710939e-01 -1.24438727e+00 4.76386875e-01 4.29875731e-01
5.52874431e-02 -9.66572821e-01 4.05650169e-01 6.18689597e-01
-8.83547783e-01 -5.95717728e-01 -3.93711358e-01 -6.75966322e-01
-2.34735280e-01 9.75395963e-02 -9.23662663e-01 3.35507542e-01
3.51708889e-01 1.47605523e-01 -3.00576389e-01 8.45723152e-01
-7.92079210e-01 1.42137304e-01 -3.05177718e-01 -6.74318016e-01
5.26964545e-01 -1.47645801e-01 2.13303491e-01 9.89108264e-01
4.92857099e-02 1.00193727e+00 2.84912825e-01 1.29910374e+00
3.69238630e-02 -3.63642901e-01 -4.72115457e-01 -3.69667500e-01
5.21962523e-01 1.04405344e+00 -3.45281422e-01 -8.51551235e-01
-2.02164710e-01 6.46123827e-01 7.94715881e-01 3.66791815e-01
-8.22979689e-01 -6.43822789e-01 6.03865325e-01 -9.14581493e-02
1.34473071e-01 -4.13083464e-01 -3.52921069e-01 -9.92519677e-01
-2.39928275e-01 -1.44155431e+00 4.56461370e-01 -1.04795313e+00
-9.44791615e-01 1.82932392e-01 3.77143502e-01 -4.05191422e-01
-4.49805111e-01 -8.44532847e-01 -4.97562855e-01 9.16199386e-01
-1.56140232e+00 -6.38004303e-01 -2.64984906e-01 5.70322156e-01
5.54546475e-01 -2.91586220e-01 9.87947345e-01 -3.65620673e-01
5.57865109e-03 4.67790216e-01 -3.29000741e-01 -2.01342046e-01
-7.84498006e-02 -1.58296454e+00 1.68427378e-01 5.22750020e-01
3.57345156e-02 1.04164755e+00 8.42851222e-01 -5.35238683e-01
-1.41078484e+00 -4.70332265e-01 8.50301147e-01 -3.80115986e-01
8.21655273e-01 -3.04665528e-02 -1.08104813e+00 6.24947190e-01
-3.47896516e-02 -6.16980612e-01 2.61244595e-01 2.81856060e-01
-5.83743453e-01 -1.08181603e-01 -1.27921736e+00 1.09556377e+00
9.79092598e-01 -4.88659114e-01 -1.41987908e+00 4.43396568e-01
8.66495252e-01 -5.40304601e-01 -8.60920012e-01 -1.90301538e-01
2.97205418e-01 -6.69244468e-01 1.11018860e+00 -1.21135628e+00
9.22683835e-01 -2.22044855e-01 7.66369328e-02 -1.73452437e+00
-6.10300303e-01 -1.02582562e+00 -5.19030154e-01 4.30006236e-01
3.39650810e-01 -9.31347191e-01 6.22063518e-01 8.55005085e-01
-6.43159822e-02 -7.53359020e-01 -4.68407542e-01 -7.28021860e-01
6.18848801e-01 3.47270444e-02 8.11061144e-01 8.43279362e-01
5.08781791e-01 2.14011371e-01 1.60411477e-01 3.28393430e-01
8.00159514e-01 2.58233249e-01 9.01884079e-01 -1.34966207e+00
-6.24587238e-01 -5.89273334e-01 -5.27019687e-02 -1.03971100e+00
4.70891684e-01 -8.40503216e-01 -1.63385823e-01 -1.41420984e+00
4.80544828e-02 -4.21349853e-01 -2.04244703e-02 7.53513873e-01
-2.04255968e-01 -3.95920724e-01 4.64253873e-01 -1.95560664e-01
-5.01037419e-01 8.29916149e-02 1.38932312e+00 -1.57121599e-01
-2.60760278e-01 -1.96748048e-01 -8.24794412e-01 9.49414730e-01
9.08485651e-01 -5.37399828e-01 -4.65494186e-01 -2.61896253e-01
4.01876330e-01 3.99697483e-01 4.49449331e-01 -1.23774934e+00
2.62964696e-01 -5.58671951e-01 6.06904328e-01 -5.24287559e-02
1.26132488e-01 -8.25353682e-01 -1.55715719e-01 1.19798064e+00
-7.37012327e-01 3.34964454e-01 5.72686493e-01 7.72014540e-03
2.96730906e-01 -7.51227021e-01 8.72339308e-01 -6.87474787e-01
-6.34244919e-01 5.96247949e-02 -5.93442857e-01 2.88306445e-01
1.06905031e+00 -1.08607650e-01 -2.06212208e-01 -5.02394199e-01
-9.41866577e-01 4.23084229e-01 2.80277520e-01 -2.12788403e-01
5.65499723e-01 -8.21200907e-01 -4.92189884e-01 -4.49884608e-02
-2.84499466e-01 3.06331038e-01 -7.52049536e-02 2.14507908e-01
-7.53582656e-01 3.86109024e-01 -4.60200101e-01 2.02503055e-02
-1.13863540e+00 9.29750621e-01 6.13588154e-01 -6.71764851e-01
-5.73512495e-01 8.91795754e-01 2.74103850e-01 -3.86028349e-01
4.21639770e-01 -8.49904716e-01 1.29125714e-01 -4.42112893e-01
7.72123039e-01 3.62648755e-01 -5.51340759e-01 5.47691584e-01
1.34475306e-02 4.97060895e-01 -6.02244616e-01 -9.34172571e-02
1.60349405e+00 4.03317004e-01 -2.22172096e-01 4.30400036e-02
4.22645241e-01 -2.25774154e-01 -1.14267576e+00 -2.93151885e-01
-1.11080155e-01 -3.46754789e-01 -2.78220087e-01 -1.10881019e+00
-5.13441384e-01 8.70828211e-01 -8.92173797e-02 7.15156674e-01
9.26167727e-01 1.68361384e-02 6.29286408e-01 1.41950500e+00
3.86067122e-01 -1.09687829e+00 5.47882557e-01 7.36528873e-01
9.99186099e-01 -6.09538972e-01 3.69334549e-01 -2.77237713e-01
-4.83322948e-01 1.54344893e+00 8.02012086e-01 -3.71283770e-01
-6.17967397e-02 3.38395029e-01 -3.57631505e-01 -7.91217759e-02
-8.87399971e-01 -5.88435642e-02 -1.64384425e-01 9.07760024e-01
-3.66089679e-02 -6.37678877e-02 -4.00104448e-02 4.44484770e-01
-4.16277409e-01 3.82540017e-01 6.89646840e-01 1.11401165e+00
-8.89661789e-01 -1.04465711e+00 -4.20712024e-01 5.00042588e-02
-3.08147725e-02 -1.48785591e-01 -3.95000666e-01 1.03904498e+00
2.11928546e-01 5.18317580e-01 -1.93355635e-01 1.83707237e-01
3.92357379e-01 1.43409535e-01 1.22895992e+00 -8.09954464e-01
-6.20421290e-01 -7.61713982e-01 2.28915527e-01 -4.18798506e-01
7.31871352e-02 -4.72909153e-01 -2.01128268e+00 -7.40174353e-01
2.35526398e-01 4.76969361e-01 2.51480162e-01 1.11694825e+00
-2.33340375e-02 8.66756439e-01 3.34696293e-01 -6.25955462e-01
-1.26681554e+00 -2.34267220e-01 -2.88644135e-01 3.23587418e-01
4.79539394e-01 -6.26122594e-01 -3.73631537e-01 -2.96365023e-02] | [9.068699836730957, 7.113010883331299] |
34818fba-c19f-4755-8dc2-216799359035 | message-based-web-service-composition | 1401.3470 | null | http://arxiv.org/abs/1401.3470v1 | http://arxiv.org/pdf/1401.3470v1.pdf | Message-Based Web Service Composition, Integrity Constraints, and Planning under Uncertainty: A New Connection | Thanks to recent advances, AI Planning has become the underlying technique
for several applications. Figuring prominently among these is automated Web
Service Composition (WSC) at the "capability" level, where services are
described in terms of preconditions and effects over ontological concepts. A
key issue in addressing WSC as planning is that ontologies are not only formal
vocabularies; they also axiomatize the possible relationships between concepts.
Such axioms correspond to what has been termed "integrity constraints" in the
actions and change literature, and applying a web service is essentially a
belief update operation. The reasoning required for belief update is known to
be harder than reasoning in the ontology itself. The support for belief update
is severely limited in current planning tools.
Our first contribution consists in identifying an interesting special case of
WSC which is both significant and more tractable. The special case, which we
term "forward effects", is characterized by the fact that every ramification of
a web service application involves at least one new constant generated as
output by the web service. We show that, in this setting, the reasoning
required for belief update simplifies to standard reasoning in the ontology
itself. This relates to, and extends, current notions of "message-based" WSC,
where the need for belief update is removed by a strong (often implicit or
informal) assumption of "locality" of the individual messages. We clarify the
computational properties of the forward effects case, and point out a strong
relation to standard notions of planning under uncertainty, suggesting that
effective tools for the latter can be successfully adapted to address the
former.
Furthermore, we identify a significant sub-case, named "strictly forward
effects", where an actual compilation into planning under uncertainty exists.
This enables us to exploit off-the-shelf planning tools to solve message-based
WSC in a general form that involves powerful ontologies, and requires reasoning
about partial matches between concepts. We provide empirical evidence that this
approach may be quite effective, using Conformant-FF as the underlying planner. | ['Piergiorgio Bertoli', 'Jörg Hoffmann', 'Marco Pistore', 'Malte Helmert'] | 2014-01-15 | null | null | null | null | ['service-composition'] | ['miscellaneous'] | [ 2.27484450e-01 8.45668197e-01 -7.67410174e-03 -3.40358615e-01
-4.89200324e-01 -8.27333868e-01 1.14969957e+00 3.49554986e-01
-2.09477156e-01 6.01089656e-01 4.45193797e-01 -3.28313738e-01
-4.94572908e-01 -1.10193729e+00 -6.74481094e-01 -5.49177468e-01
-4.65868294e-01 7.14881897e-01 8.79058540e-01 -8.16352606e-01
-9.44152009e-03 4.39689547e-01 -1.87384605e+00 1.53322563e-01
3.95910442e-01 9.33321118e-01 3.59721482e-02 3.16709131e-01
-6.40995681e-01 6.83744073e-01 -3.19278449e-01 -5.18750966e-01
9.00440812e-02 -2.04266995e-01 -1.39998639e+00 2.48557612e-01
-3.80771518e-01 3.05664465e-02 2.29106858e-01 1.13490164e+00
-8.38472247e-02 5.17395325e-02 1.76835954e-01 -1.83112347e+00
-1.06401548e-01 1.00258732e+00 2.77793556e-01 -3.55858862e-01
1.13785148e+00 1.18169014e-03 1.15988302e+00 -1.64833844e-01
8.39957416e-01 1.29296756e+00 4.52600241e-01 7.01972604e-01
-1.15378666e+00 3.51470225e-02 4.59730327e-01 3.46831828e-01
-1.30368531e+00 -5.52145302e-01 3.81607622e-01 -2.33322904e-01
1.25245094e+00 8.77909005e-01 6.95893705e-01 6.16805851e-01
1.73571140e-01 6.67567790e-01 1.01320374e+00 -8.05401802e-01
7.23015368e-01 2.93314010e-01 1.19152375e-01 5.33790827e-01
3.16531658e-01 6.02667257e-02 -2.82618791e-01 -3.65983218e-01
1.44193575e-01 -2.50959396e-01 -2.62691677e-01 -5.36225677e-01
-9.65182841e-01 7.49671280e-01 -1.61902472e-01 6.52934611e-01
-1.41212896e-01 2.95864433e-01 3.33987445e-01 4.41400170e-01
2.08771396e-02 3.85051161e-01 -6.53446674e-01 -2.97801554e-01
-2.35188782e-01 5.16759455e-01 1.65138233e+00 1.33595514e+00
6.27698839e-01 -1.91666946e-01 6.43745661e-01 2.47702897e-01
6.40420377e-01 3.04842889e-01 3.42993259e-01 -1.35906100e+00
-5.86635396e-02 6.55510187e-01 5.06058931e-01 -6.49203002e-01
-4.81787831e-01 -8.94966871e-02 -6.84957579e-02 3.56615841e-01
2.73379117e-01 1.76278830e-01 -2.82350212e-01 1.88143528e+00
6.20579720e-01 6.67890161e-02 4.01513815e-01 4.79581743e-01
1.97808176e-01 4.56377685e-01 3.62209938e-02 -6.51251793e-01
1.48999977e+00 -3.26981455e-01 -7.66326070e-01 -2.49361377e-02
6.62962258e-01 -2.32370511e-01 9.17228043e-01 4.81960416e-01
-1.13121700e+00 5.67956090e-01 -1.04112983e+00 2.94186682e-01
-6.16847932e-01 -1.15667105e+00 1.10834062e+00 6.01060629e-01
-1.11182296e+00 3.88574421e-01 -6.74055636e-01 -6.83206022e-01
-1.70794502e-01 4.32839692e-01 -2.27623135e-01 -2.25804001e-02
-1.43143702e+00 1.00047195e+00 5.13925731e-01 -2.48713687e-01
-8.64833117e-01 -2.07960039e-01 -9.15625215e-01 2.51803070e-01
1.11884630e+00 -3.96576911e-01 1.71630871e+00 -1.03402925e+00
-1.71751535e+00 6.21938527e-01 7.96244815e-02 -6.39715612e-01
5.63649476e-01 4.56894338e-01 -8.40789497e-01 -6.08308800e-02
6.67694509e-02 2.04969421e-01 3.78021240e-01 -1.48379111e+00
-1.13695443e+00 -3.39492708e-01 1.04303598e+00 1.52286157e-01
1.83427766e-01 4.12688941e-01 -3.35704923e-01 4.89295311e-02
4.32212591e-01 -9.17889595e-01 -4.16849017e-01 -3.21095109e-01
-5.63134775e-02 -5.20113587e-01 2.77276188e-01 8.30397829e-02
1.19127619e+00 -1.66613817e+00 1.38831332e-01 6.04428709e-01
-1.00252889e-01 -4.91652757e-01 8.50205496e-02 9.50971842e-01
1.93628315e-02 3.51758510e-01 -2.98051536e-01 2.18350857e-01
9.04001772e-01 8.48835588e-01 -3.34738106e-01 3.59417289e-01
-1.19800307e-02 4.89488691e-01 -8.82633030e-01 -4.81487036e-01
-2.20128568e-03 -3.44805606e-02 -7.34016180e-01 -3.15161914e-01
-9.91682827e-01 4.62582894e-03 -6.03078604e-01 5.90529621e-01
2.70515352e-01 -1.37447238e-01 7.16778159e-01 3.13036889e-01
-5.60401261e-01 3.23721558e-01 -1.86678433e+00 1.75646865e+00
-3.93843085e-01 -1.94198623e-01 2.58206964e-01 -8.86766016e-01
2.59374440e-01 7.53225267e-01 5.86350024e-01 -3.64807457e-01
6.27188419e-05 3.11034292e-01 -6.03015870e-02 -6.37366712e-01
2.75189757e-01 -5.03959537e-01 -5.35401702e-01 5.79331934e-01
-1.70297131e-01 -3.89483005e-01 5.00148833e-01 2.30310649e-01
1.29435265e+00 3.08151573e-01 7.10689306e-01 -5.29809952e-01
7.96683490e-01 3.60542268e-01 7.60301471e-01 6.47505939e-01
-5.29753715e-02 -3.54030356e-02 7.15835869e-01 -3.96836013e-01
-6.33858263e-01 -6.58718348e-01 -1.38584420e-01 1.04891980e+00
2.32899085e-01 -8.94677520e-01 -7.64151752e-01 -7.06151605e-01
-1.54331967e-01 1.11332965e+00 -3.41923058e-01 1.22841768e-01
-4.26159054e-01 -3.58149350e-01 6.15150511e-01 1.44194245e-01
-5.90105588e-03 -8.45036447e-01 -8.96072865e-01 5.29192090e-01
-2.44819120e-01 -1.17421675e+00 1.58129394e-01 1.75021291e-01
-5.26808500e-01 -1.22962940e+00 3.03021282e-01 -1.62987113e-01
3.19532633e-01 -1.82369798e-01 1.00263655e+00 3.29786390e-01
1.90857053e-01 9.72558796e-01 -5.75717688e-01 -6.79595470e-01
-6.23173773e-01 -3.58032763e-01 1.63564831e-01 -9.17085353e-03
2.91163027e-01 -7.85501897e-01 -1.30861606e-02 3.01751554e-01
-1.45330989e+00 -1.03688547e-02 1.23397432e-01 1.53345227e-01
2.91147590e-01 7.26280868e-01 3.76231670e-01 -8.87154400e-01
4.85940635e-01 -5.95084965e-01 -7.95157015e-01 2.80417114e-01
-8.48139644e-01 3.96030396e-01 4.91846859e-01 -8.16435292e-02
-1.13182509e+00 2.24285182e-02 -2.94826746e-01 6.03788257e-01
-4.39460456e-01 9.10067141e-01 -7.69003212e-01 2.56635696e-02
3.50053132e-01 1.87932849e-01 -4.53253277e-02 -3.24498832e-01
4.14961040e-01 4.71043050e-01 3.14255476e-01 -1.27410078e+00
7.84509182e-01 9.70549703e-01 2.57206649e-01 -4.92855728e-01
-3.32884520e-01 -2.92942077e-01 -2.03658491e-01 -1.80531606e-01
5.50500453e-01 -2.90025234e-01 -1.43985653e+00 -1.50847822e-01
-1.04255152e+00 -3.20234090e-01 -7.71976233e-01 2.08455712e-01
-1.05750859e+00 3.50318730e-01 -3.17427188e-01 -1.13215768e+00
3.42501134e-01 -1.25056303e+00 7.53276587e-01 -2.46329322e-01
-3.04491729e-01 -8.35214019e-01 -1.03898078e-01 -2.86322250e-03
4.07829434e-01 8.36201310e-02 1.22478056e+00 -1.11468661e+00
-5.85241318e-01 -4.90840197e-01 2.77406573e-01 -4.56086770e-02
-1.68737277e-01 -7.19019771e-02 -6.15862191e-01 1.19355403e-01
1.21269755e-01 3.29980463e-01 -2.84039348e-01 -1.90409705e-01
4.05878276e-01 -5.32175601e-01 -2.18374342e-01 -2.69235224e-01
1.80429685e+00 5.02967238e-01 7.64271200e-01 6.39708877e-01
-8.06452408e-02 8.70966375e-01 4.36715156e-01 6.26707375e-01
7.88078606e-01 1.07179689e+00 6.43041849e-01 7.05552280e-01
1.11266881e-01 5.13178520e-02 3.48313123e-01 5.96284330e-01
-5.24098694e-01 6.63801581e-02 -1.06085396e+00 2.46079192e-01
-2.19523263e+00 -1.31231070e+00 -1.06183372e-01 2.14790511e+00
1.00445962e+00 1.59593672e-01 3.93243767e-02 3.14811856e-01
4.00938660e-01 -2.05430329e-01 -9.48017389e-02 -3.66570890e-01
6.43458450e-03 9.41701010e-02 4.45245415e-01 1.07930565e+00
-5.94799101e-01 8.34944069e-01 6.05939341e+00 3.57504278e-01
-4.86257881e-01 5.18831551e-01 -4.14598256e-01 2.18596369e-01
-7.31202781e-01 7.35263348e-01 -7.13988066e-01 3.85446012e-01
1.12697792e+00 -5.12338638e-01 8.33783746e-01 9.23356831e-01
1.83193266e-01 -3.84041846e-01 -1.41048288e+00 1.82995364e-01
-8.09018910e-02 -1.26286876e+00 -1.58954546e-01 1.66207850e-01
3.14465284e-01 -3.15578938e-01 -6.64560974e-01 3.63560587e-01
8.59472871e-01 -4.77132142e-01 1.19846106e+00 4.64935869e-01
3.71170491e-01 -5.50093293e-01 5.38375437e-01 5.12686312e-01
-1.23332572e+00 -4.03597206e-01 4.86659184e-02 -1.49834797e-01
4.61894542e-01 2.60982066e-01 -4.51842576e-01 9.65338528e-01
4.57370728e-01 2.78638152e-04 1.13994576e-01 8.25387537e-01
-1.88322693e-01 3.64347026e-02 -7.25982189e-01 -5.24206720e-02
2.41252467e-01 -2.69295335e-01 8.02676082e-01 1.06800973e+00
2.26478204e-01 3.63957763e-01 2.17260346e-01 5.12705564e-01
3.92646104e-01 2.06565056e-02 -4.62033302e-01 2.00859219e-01
3.56896818e-01 7.50283360e-01 -6.85095131e-01 -5.37908018e-01
-7.03567505e-01 4.66650695e-01 -2.34849930e-01 3.64091337e-01
-7.91782498e-01 1.77795514e-02 6.79662883e-01 1.55985489e-01
-6.40721321e-02 -8.69589970e-02 1.81576103e-01 -1.25114763e+00
9.98002961e-02 -1.21078575e+00 5.78882277e-01 -5.91929495e-01
-8.07817817e-01 3.43255252e-01 5.87604463e-01 -9.56737936e-01
-4.64912653e-01 -4.69882786e-01 -1.51343897e-01 4.21299160e-01
-1.51750517e+00 -1.12704825e+00 2.21889522e-02 7.97943711e-01
4.36509728e-01 3.51552516e-01 1.36350036e+00 1.46496147e-01
-1.20767802e-01 -2.48403475e-01 -5.10783195e-01 -6.86874092e-01
7.84580186e-02 -1.34633827e+00 -1.49117038e-01 1.09597147e+00
-3.58144231e-02 6.86240792e-01 1.24804866e+00 -6.54083073e-01
-1.96936786e+00 -5.68981946e-01 1.29493952e+00 -3.53145331e-01
1.14255428e+00 -1.85736209e-01 -5.27341008e-01 1.23324609e+00
3.58654725e-05 -1.87961593e-01 4.92385626e-01 1.32554814e-01
-3.10240030e-01 -2.27921575e-01 -1.26325500e+00 7.68893838e-01
1.29646289e+00 -4.77017820e-01 -8.61799061e-01 6.10061646e-01
9.15944934e-01 -2.00946018e-01 -9.48212564e-01 2.67431498e-01
5.74289143e-01 -8.77555251e-01 7.22011149e-01 -9.22380030e-01
-1.70050383e-01 -7.51622736e-01 -7.45915234e-01 -9.18297470e-01
-9.97357145e-02 -8.93542886e-01 -1.34258911e-01 1.19210935e+00
4.66752976e-01 -9.63832915e-01 4.61418301e-01 1.48244512e+00
-4.91275221e-01 -3.00566733e-01 -1.05682766e+00 -9.70763683e-01
-2.48487115e-01 -1.21834457e+00 1.27742338e+00 1.09577930e+00
8.44488919e-01 -1.76172584e-01 1.79628924e-01 5.36369979e-01
3.81616116e-01 1.78165231e-02 3.30539227e-01 -1.64049172e+00
-5.79037488e-01 -5.00624239e-01 -4.55593526e-01 -4.86685872e-01
3.06391954e-01 -8.97544324e-01 2.90262759e-01 -1.67482615e+00
-4.20112699e-01 -7.17082739e-01 4.65711318e-02 7.04124629e-01
8.20271909e-01 -4.14948344e-01 2.13759020e-01 2.68854767e-01
-6.57969654e-01 -1.44243473e-03 6.98990047e-01 6.81865588e-02
2.74067819e-02 2.55994707e-01 -7.76995599e-01 1.08159292e+00
5.67377687e-01 -3.91959548e-01 -4.66693610e-01 -1.17151164e-01
1.01982450e+00 3.15008044e-01 1.45619124e-01 -8.71256173e-01
6.31345749e-01 -8.61427784e-01 -9.00480509e-01 1.83426738e-01
3.17831546e-01 -1.63417792e+00 8.40065658e-01 4.35419440e-01
-3.76384258e-01 -1.03962377e-01 -1.37216479e-01 4.49238420e-01
-2.35978384e-02 -9.01259780e-01 2.29507864e-01 -4.35280651e-01
-1.06704223e+00 -1.09008566e-01 -7.39317060e-01 -2.28707224e-01
1.17248499e+00 9.85949114e-02 -1.32143632e-01 -5.13864338e-01
-1.04579377e+00 2.15236731e-02 5.16361296e-01 1.10756315e-01
2.30726182e-01 -1.10697758e+00 -2.01120928e-01 -1.54850394e-01
3.36354971e-01 -1.09538466e-01 -2.08778620e-01 1.02945244e+00
-2.40023240e-01 4.36913908e-01 1.34496346e-01 -1.29689410e-01
-9.63541448e-01 7.25026548e-01 2.95825720e-01 5.61062898e-03
-5.42115450e-01 3.42280328e-01 -2.77914643e-01 -3.17745030e-01
3.48103315e-01 -5.28724432e-01 -5.63940667e-02 -1.23183489e-01
6.67067885e-01 1.37580439e-01 2.29759827e-01 -5.08803487e-01
-5.84173679e-01 2.12330088e-01 3.88494670e-01 -5.56815326e-01
1.43872225e+00 -4.41057265e-01 -5.42846501e-01 4.60540026e-01
3.74245584e-01 1.94334835e-01 -7.39552140e-01 -3.37637484e-01
5.85244000e-01 -1.19258940e-01 -3.59867990e-01 -9.09052014e-01
-4.23279077e-01 1.46624222e-01 1.75548390e-01 1.08070195e+00
1.01399422e+00 3.10076445e-01 2.91178256e-01 5.99241078e-01
1.12574613e+00 -1.24292850e+00 -7.42321908e-01 5.02789319e-01
1.05727565e+00 -7.03396797e-01 -5.48202638e-03 -6.79680943e-01
-3.98989022e-01 1.14375210e+00 1.17623713e-02 2.08954647e-01
3.64602715e-01 8.39211941e-01 -2.99439996e-01 -4.17433172e-01
-9.69320595e-01 -6.40614212e-01 -3.77633035e-01 5.64004302e-01
-7.40887076e-02 1.97177887e-01 -7.16818392e-01 7.97339857e-01
-7.50103593e-02 2.25079075e-01 7.80350029e-01 1.40150177e+00
-6.45816445e-01 -1.56585884e+00 -3.47662359e-01 -1.58200428e-01
-4.64870274e-01 2.51292977e-02 -1.32640287e-01 1.20363975e+00
3.33587915e-01 1.22957456e+00 -1.40118629e-01 -1.40878543e-01
5.42425096e-01 4.62272823e-01 5.60686231e-01 -6.67587757e-01
-4.20794994e-01 -7.78841302e-02 6.88237786e-01 -8.90189648e-01
-7.21027136e-01 -7.54631996e-01 -1.81337440e+00 -3.87073547e-01
-2.58716315e-01 5.42159855e-01 1.03613639e+00 1.43842697e+00
-6.64374083e-02 3.31233412e-01 2.62969643e-01 -6.27786338e-01
-7.95300186e-01 -3.87233824e-01 -5.63669801e-01 3.42358351e-01
-1.56828195e-01 -5.55355430e-01 -5.56210458e-01 2.73657233e-01] | [8.647599220275879, 6.834794044494629] |
4059848f-988c-47d2-b5ac-d12c19265ae3 | cogalex-vi-shared-task-transrelation-a-robust | null | null | https://aclanthology.org/2020.cogalex-1.7 | https://aclanthology.org/2020.cogalex-1.7.pdf | CogALex-VI Shared Task: Transrelation - A Robust Multilingual Language Model for Multilingual Relation Identification | We describe our submission to the CogALex-VI shared task on the identification of multilingual paradigmatic relations building on XLM-RoBERTa (XLM-R), a robustly optimized and multilingual BERT model. In spite of several experiments with data augmentation, data addition and ensemble methods with a Siamese Triple Net, Translrelation, the XLM-R model with a linear classifier adapted to this specific task, performed best in testing and achieved the best results in the final evaluation of the shared task, even for a previously unseen language. | ['Dagmar Gromann', 'Barbara Heinisch', 'Christian Lang', 'Lennart Wachowiak'] | 2020-12-12 | null | null | null | null | ['multilingual-text-classification', 'hypernym-discovery'] | ['miscellaneous', 'natural-language-processing'] | [-4.43313122e-01 3.45290542e-01 -3.98572683e-01 -4.39669251e-01
-7.84414649e-01 -5.17642021e-01 1.06903410e+00 1.36677459e-01
-5.81280053e-01 1.05361676e+00 3.31992149e-01 -7.00046718e-01
-7.41616607e-01 -1.83683604e-01 -6.56576514e-01 -4.01632264e-02
-4.42924976e-01 1.42627573e+00 -6.31056651e-02 -8.44534636e-01
-2.80909181e-01 2.28485510e-01 -7.07492352e-01 5.94139278e-01
1.02307796e+00 4.41153407e-01 -1.55892387e-01 4.74125087e-01
4.03617710e-01 1.11903322e+00 -4.82883811e-01 -9.25513327e-01
1.14410579e-01 -7.75493011e-02 -1.42697966e+00 -7.35187650e-01
6.82152927e-01 4.84342277e-01 -4.56009895e-01 6.27632201e-01
4.35809433e-01 7.17273802e-02 5.16447127e-01 -1.50000751e+00
-7.13153958e-01 1.96194184e+00 -2.56062478e-01 4.02457923e-01
6.85408890e-01 -1.93673998e-01 1.43766153e+00 -8.09766769e-01
1.41712546e+00 1.72392809e+00 8.87322128e-01 -1.87435925e-01
-1.45964444e+00 -6.33835375e-01 1.41414210e-01 5.15319705e-01
-1.44257510e+00 -5.28400719e-01 3.08511645e-01 -1.44976690e-01
1.71312094e+00 2.88960099e-01 4.37123120e-01 1.39580989e+00
1.33352995e-01 6.54619694e-01 1.54644048e+00 -4.82972562e-01
-5.59197605e-01 3.21697891e-01 4.20608729e-01 4.72695559e-01
1.47003144e-01 1.32739320e-01 -8.31231236e-01 -1.42509162e-01
-2.47453824e-02 -1.22160053e+00 -2.22944558e-01 -2.45033647e-03
-1.71241188e+00 3.56708825e-01 5.41254759e-01 9.20738816e-01
-1.65486023e-01 -4.19870824e-01 5.92267215e-01 9.36144412e-01
6.52203023e-01 9.69872236e-01 -1.38398480e+00 -1.23065814e-01
-5.11158705e-01 3.79269838e-01 1.20304596e+00 9.48073685e-01
3.53737444e-01 -2.10238084e-01 -6.11830130e-02 1.12135828e+00
-7.38614099e-03 2.10859001e-01 4.20320362e-01 -9.03967500e-01
1.02479839e+00 7.25606322e-01 -5.16880810e-01 -6.00185156e-01
-1.00429809e+00 -6.47894681e-01 -4.75435019e-01 -2.30052188e-01
6.69981420e-01 -2.40435675e-01 -2.85704285e-01 1.81244719e+00
-9.09778997e-02 -2.99639463e-01 5.02754450e-01 6.11118019e-01
1.00106645e+00 1.76654339e-01 1.82142168e-01 -7.73656368e-02
1.15527463e+00 -1.19343472e+00 -6.49914682e-01 -1.37047261e-01
1.25568485e+00 -7.29179382e-01 7.23335981e-01 5.11850357e-01
-9.10773218e-01 -4.26634401e-01 -1.01899457e+00 -4.89735991e-01
-9.14920747e-01 1.68995429e-02 1.01973748e+00 1.79036096e-01
-1.11248934e+00 5.00282288e-01 -1.16522148e-01 -8.54412138e-01
-7.40570202e-02 3.39954674e-01 -1.13346851e+00 -1.18833072e-01
-1.82359278e+00 1.79237545e+00 9.84014332e-01 5.79387099e-02
-4.16491687e-01 -8.21170390e-01 -7.54425049e-01 -4.02024686e-01
4.69082981e-01 -3.33060235e-01 7.47667551e-01 -3.77733111e-01
-1.17450464e+00 1.43649614e+00 3.63592088e-01 -7.73445189e-01
5.17032266e-01 -3.06375206e-01 -1.07180762e+00 -3.64170104e-01
3.33892286e-01 5.63520074e-01 -5.65877073e-02 -1.03009665e+00
-4.22737122e-01 -2.79643238e-01 6.33583637e-04 1.34050995e-01
3.24609242e-02 6.15315914e-01 4.58609983e-02 -6.38332248e-01
-7.26053342e-02 -8.61435056e-01 -3.03369085e-03 -1.10245407e+00
-6.65677309e-01 -6.62421703e-01 6.03862941e-01 -1.09047353e+00
8.86980832e-01 -1.72017598e+00 4.75450307e-01 3.78591567e-01
-8.88220128e-03 3.31857979e-01 -4.43054855e-01 5.04999459e-01
-6.74600244e-01 1.14621103e-01 2.37945765e-01 -4.19588894e-01
1.04722176e-02 2.17899829e-01 -4.21401225e-02 2.53036410e-01
-6.83757290e-02 1.24485385e+00 -9.13284123e-01 -6.61538064e-01
-1.49370849e-01 5.48865646e-02 -1.70702770e-01 -2.31767267e-01
-2.67862737e-01 6.38824165e-01 2.81539232e-01 4.28679615e-01
2.43465096e-01 -1.15890995e-01 9.22357500e-01 -2.55202264e-01
-1.21865213e-01 6.64608598e-01 -7.67248511e-01 1.94736290e+00
-3.76899779e-01 8.70689034e-01 8.40074271e-02 -7.74304807e-01
8.52823734e-01 4.97571230e-01 4.34179187e-01 -8.26272726e-01
1.33708969e-01 7.05881953e-01 7.05526292e-01 -3.06551427e-01
4.58178967e-01 4.40031856e-01 -2.87386805e-01 4.25179571e-01
6.06458008e-01 1.20992199e-01 6.99332595e-01 7.03058004e-01
1.02132630e+00 3.99099022e-01 4.52616781e-01 -6.39435232e-01
6.89311326e-01 9.84576419e-02 6.44393265e-01 3.70942384e-01
6.62667528e-02 2.13348540e-03 4.79583085e-01 -4.27796364e-01
-7.71431088e-01 -8.14678252e-01 -5.25176823e-01 1.36519742e+00
-3.72419775e-01 -7.73391843e-01 -3.77622526e-03 -9.20379162e-01
9.52224731e-02 1.05237484e+00 -3.20561975e-01 1.34041324e-01
-8.64009142e-01 -1.01829720e+00 1.01505184e+00 5.46699874e-02
1.75035655e-01 -1.14850152e+00 4.71012890e-01 2.43726239e-01
-4.18417692e-01 -1.73911130e+00 -4.78004571e-03 4.41129655e-01
-1.62944376e-01 -1.44199085e+00 5.11495844e-02 -8.26976299e-01
4.01092730e-02 -7.76394367e-01 1.47362733e+00 -2.78871477e-01
-7.17464685e-02 -2.16479242e-01 -2.98094660e-01 -1.24571346e-01
-7.13714898e-01 6.97315335e-01 1.27505109e-01 -4.92357016e-01
4.45142299e-01 -6.16618752e-01 3.61557782e-01 6.19204864e-02
-3.17190811e-02 6.49144500e-02 3.59879971e-01 8.50395262e-01
-9.88863036e-02 -1.39912948e-01 8.61558616e-01 -1.04921257e+00
8.43113899e-01 -5.59392154e-01 -4.13858056e-01 6.67170823e-01
-7.21021354e-01 -1.50665119e-01 4.72846150e-01 -2.04549357e-01
-9.33726966e-01 -6.24868870e-01 3.32398340e-02 1.98388770e-01
9.95796323e-02 7.96659470e-01 -3.38152587e-01 -5.70662096e-02
7.93661952e-01 -6.38432801e-01 -3.41292650e-01 -7.17139006e-01
7.93673575e-01 5.18001199e-01 6.03218138e-01 -9.23470020e-01
4.09807533e-01 -3.18281353e-01 2.05168366e-01 -5.66064537e-01
-7.54514992e-01 1.56057132e-02 -9.32152271e-01 1.05299145e-01
7.14684248e-01 -9.84139204e-01 -7.14565516e-01 4.89033729e-01
-1.23822737e+00 -4.96933192e-01 -5.48732579e-02 6.61718726e-01
-2.91646063e-01 -1.02432467e-01 -7.60027349e-01 -3.70639324e-01
-2.45425880e-01 -9.00930762e-01 4.19829905e-01 -4.59903359e-01
-9.03385520e-01 -1.43350053e+00 3.38405132e-01 7.25572169e-01
1.27208710e-01 2.92006761e-01 1.60382521e+00 -1.37715149e+00
-2.74630561e-02 3.80695285e-03 -2.10613638e-01 2.29571447e-01
-2.34253809e-01 5.35230637e-02 -5.82287014e-01 -8.10439214e-02
-7.64590383e-01 -7.58041263e-01 6.52950406e-01 -2.88579702e-01
3.55079442e-01 -7.02853203e-02 -4.41694349e-01 5.09065807e-01
9.76170897e-01 -8.04503784e-02 1.37911722e-01 5.69104612e-01
7.88057148e-01 7.78094053e-01 4.74499673e-01 -3.14724803e-01
1.00786412e+00 9.61633861e-01 2.19115838e-02 2.10112035e-02
-4.46783155e-01 -2.13918090e-01 2.81107247e-01 1.18504643e+00
-1.18108884e-01 -3.25386643e-01 -1.47714663e+00 6.08961642e-01
-2.10378957e+00 -6.77309275e-01 -6.40739262e-01 1.63712704e+00
1.08470523e+00 8.47061798e-02 1.26342297e-01 1.62042975e-01
3.65313530e-01 -3.89141813e-02 -8.07852596e-02 -5.42111695e-01
-1.14591253e+00 3.50129277e-01 2.20302001e-01 6.67711794e-01
-9.17417943e-01 1.46551025e+00 7.69486856e+00 6.62683964e-01
-6.71831250e-01 5.97632825e-01 2.43240848e-01 -7.35415295e-02
-1.01818621e-01 1.18511163e-01 -8.59608889e-01 2.21043512e-01
1.29413998e+00 -2.17531800e-01 9.18703198e-01 1.43887550e-01
-3.77148181e-01 -1.13115124e-01 -1.43658507e+00 4.37328994e-01
1.01361990e-01 -9.11538303e-01 -1.15488172e-01 -3.19230296e-02
1.02814043e+00 6.59923673e-01 -2.60655254e-01 7.09121108e-01
9.65632617e-01 -1.12275004e+00 5.25642037e-01 4.76870835e-01
4.66642946e-01 -6.88016653e-01 8.09829772e-01 3.63222927e-01
-6.66964650e-01 -1.17993474e-01 1.73754841e-01 6.38202019e-03
1.83655694e-01 6.33486390e-01 -6.06656075e-01 1.27841377e+00
5.78947365e-01 8.69710386e-01 -8.83121312e-01 5.13956606e-01
-5.26098430e-01 4.32728916e-01 -4.15075839e-01 4.05282497e-01
1.65913049e-02 -2.37389337e-02 9.65193152e-01 1.32059038e+00
-2.51956224e-01 -4.06541079e-01 3.02145541e-01 3.77220303e-01
-3.85569334e-01 3.96293461e-01 -5.32350838e-01 -1.46342248e-01
4.34013724e-01 1.50736272e+00 -1.74861580e-01 -2.32257470e-01
-2.51970589e-01 4.05872315e-01 1.13226092e+00 3.79777342e-01
-5.62595785e-01 -2.21727103e-01 1.11965381e-01 -2.94809014e-01
-4.57908541e-01 -1.92560524e-01 -1.45534039e-01 -1.09445500e+00
-3.17521900e-01 -1.20728481e+00 6.53686464e-01 -9.05710042e-01
-1.71800673e+00 9.90432620e-01 1.66923061e-01 -4.67970937e-01
-5.93045235e-01 -7.59591818e-01 -1.26067773e-01 1.05895627e+00
-1.03581965e+00 -1.90876102e+00 3.18443447e-01 6.59627080e-01
-2.93902576e-01 -8.86988938e-01 1.08229148e+00 6.50211573e-01
-5.27665079e-01 7.82881498e-01 -2.10814893e-01 8.35506693e-02
1.47944021e+00 -1.38822699e+00 3.36057276e-01 4.88507628e-01
3.35287243e-01 6.07549965e-01 5.42841673e-01 -7.76210845e-01
-9.22993898e-01 -8.89900506e-01 1.64657784e+00 -6.86221480e-01
1.48572052e+00 -4.47746962e-01 -5.35988867e-01 1.46342695e+00
8.06423008e-01 -4.31764811e-01 5.28869510e-01 9.94295657e-01
-6.35681152e-01 -1.98989287e-01 -9.42287505e-01 5.10960996e-01
1.38694048e+00 -6.02690697e-01 -8.07511508e-01 1.02460814e+00
6.75412595e-01 -3.78726989e-01 -1.69244313e+00 6.52590752e-01
3.03742737e-01 -5.60918629e-01 6.78737283e-01 -1.10595322e+00
3.45764637e-01 1.17272303e-01 -2.68445551e-01 -1.75739002e+00
-3.73360962e-01 -7.34811544e-01 4.70218398e-02 1.51107049e+00
1.04548872e+00 -1.03439426e+00 6.67378381e-02 2.13965699e-02
-1.45934656e-01 -4.23444062e-01 -1.09343028e+00 -9.67609644e-01
3.10837656e-01 -5.08484364e-01 5.51038504e-01 1.62591839e+00
3.69777113e-01 9.46184278e-01 -4.04481798e-01 -1.65565982e-01
5.74773669e-01 -1.48135513e-01 7.52049685e-01 -1.38736534e+00
-2.89537281e-01 -4.77168202e-01 -3.63100946e-01 -8.01105499e-02
1.02836335e+00 -1.75327754e+00 -4.74925160e-01 -1.50249827e+00
-1.05616702e-02 -6.40469551e-01 -3.04768533e-01 6.94506109e-01
-1.22189103e-02 3.42828184e-01 1.67149767e-01 5.84928552e-03
-5.79985499e-01 1.71437591e-01 1.01241910e+00 -9.99437571e-02
-1.03673767e-02 -3.13959539e-01 -8.11413765e-01 6.08409762e-01
5.85470557e-01 -2.61297077e-01 -2.01122761e-01 -7.24954367e-01
7.28977799e-01 -4.96022329e-02 1.77252650e-01 -7.07902908e-01
1.85697943e-01 3.12123537e-01 1.40712380e-01 -7.12060750e-01
4.08836126e-01 -5.78208029e-01 3.82867664e-01 1.32705048e-01
-5.03305018e-01 3.80098462e-01 1.10674277e-01 -2.97307998e-01
-1.50310040e-01 1.73315808e-01 4.49434310e-01 1.22077763e-01
-5.08415401e-01 -1.01480253e-01 -5.21286987e-02 3.18671465e-01
9.16465282e-01 5.39279997e-01 -9.64040339e-01 -2.50204913e-02
-1.19669890e+00 6.87348306e-01 -3.68097611e-03 9.90665019e-01
-2.37953708e-01 -1.35131931e+00 -1.38413024e+00 -6.27750009e-02
1.34151027e-01 -7.01920569e-01 -2.13482007e-01 1.21667659e+00
-2.14077607e-01 9.72978592e-01 -2.45415315e-01 -1.61670610e-01
-1.19621861e+00 4.37200934e-01 4.24904674e-01 -9.88668501e-01
-3.12755495e-01 7.03726649e-01 -7.51078486e-01 -1.12880409e+00
-1.03829190e-01 1.82514429e-01 -4.07126307e-01 2.41709620e-01
-2.61235870e-02 7.69150257e-01 4.05530602e-01 -1.06969285e+00
-5.75750172e-01 1.81725204e-01 -2.76971340e-01 -3.95848244e-01
1.36880529e+00 1.69695616e-01 -1.01265740e+00 8.32946479e-01
1.22488678e+00 2.10546747e-01 1.68583661e-01 -4.64786440e-01
5.65117419e-01 2.30221003e-01 -1.47525519e-01 -1.76075268e+00
-8.04498851e-01 7.60705248e-02 1.75653528e-02 3.15526575e-02
4.19519335e-01 1.44452184e-01 3.14159632e-01 4.71984386e-01
5.51005363e-01 -1.10072184e+00 -5.31283021e-01 1.27178228e+00
1.14509463e+00 -1.18366122e+00 2.23561853e-01 -4.31516737e-01
-5.25101244e-01 7.40243852e-01 9.07652438e-01 5.13930917e-02
7.58926809e-01 2.99325317e-01 2.34888196e-01 -3.25779200e-01
-1.19080436e+00 -7.76918158e-02 5.55843294e-01 6.40960038e-01
6.75847709e-01 3.86298269e-01 -7.26901710e-01 5.15469968e-01
-8.29622686e-01 -4.29005593e-01 -2.01821178e-01 3.58641416e-01
4.92449462e-01 -1.45755267e+00 -9.20510665e-02 3.10754031e-01
-4.25535113e-01 -2.99669504e-01 -8.42840612e-01 1.42094767e+00
4.88236696e-01 1.08973300e+00 1.43479416e-02 -4.56293017e-01
4.63375270e-01 3.45982909e-01 6.70622885e-01 -3.82008046e-01
-1.40074289e+00 -3.43873143e-01 1.15087807e+00 -4.62174743e-01
-4.78804082e-01 -9.44972515e-01 -9.16826844e-01 -4.69859004e-01
-6.06691614e-02 1.90614790e-01 6.19633079e-01 1.17807138e+00
9.61140692e-02 7.72896111e-01 1.74769104e-01 -1.91747621e-01
-8.92964602e-02 -1.61658740e+00 -4.81543750e-01 3.94177258e-01
-3.13133985e-01 -3.65492076e-01 -8.82805362e-02 -5.33704698e-01] | [9.975334167480469, 9.23398208618164] |
dd2cafd0-6e2d-4f4f-9b5b-6408ffa49e2b | mgrr-net-multi-level-graph-relational | 2204.01349 | null | https://arxiv.org/abs/2204.01349v3 | https://arxiv.org/pdf/2204.01349v3.pdf | MGRR-Net: Multi-level Graph Relational Reasoning Network for Facial Action Units Detection | The Facial Action Coding System (FACS) encodes the action units (AUs) in facial images, which has attracted extensive research attention due to its wide use in facial expression analysis. Many methods that perform well on automatic facial action unit (AU) detection primarily focus on modeling various types of AU relations between corresponding local muscle areas, or simply mining global attention-aware facial features, however, neglect the dynamic interactions among local-global features. We argue that encoding AU features just from one perspective may not capture the rich contextual information between regional and global face features, as well as the detailed variability across AUs, because of the diversity in expression and individual characteristics. In this paper, we propose a novel Multi-level Graph Relational Reasoning Network (termed MGRR-Net) for facial AU detection. Each layer of MGRR-Net performs a multi-level (i.e., region-level, pixel-wise and channel-wise level) feature learning. While the region-level feature learning from local face patches features via graph neural network can encode the correlation across different AUs, the pixel-wise and channel-wise feature learning via graph attention network can enhance the discrimination ability of AU features from global face features. The fused features from the three levels lead to improved AU discriminative ability. Extensive experiments on DISFA and BP4D AU datasets show that the proposed approach achieves superior performance than the state-of-the-art methods. | ['Hu Han', 'Xiao Liu', 'Songpei Xu', 'Joemon M. Jose', 'Xuri Ge'] | 2022-04-04 | null | null | null | null | ['relational-reasoning'] | ['natural-language-processing'] | [ 1.86111376e-01 -1.34865552e-01 -3.87254149e-01 -4.82155502e-01
-5.40818155e-01 8.06331038e-02 3.63662302e-01 -2.35367984e-01
-1.44948196e-02 2.25033000e-01 3.02611083e-01 4.21833396e-01
-7.57311657e-02 -8.05430830e-01 -4.46436495e-01 -1.06052053e+00
-2.93668747e-01 -2.43616506e-01 -7.65244290e-02 -3.92735422e-01
-8.29566941e-02 8.07945609e-01 -1.68312979e+00 5.92637599e-01
4.50845897e-01 1.69309568e+00 -2.77002096e-01 2.00329378e-01
-1.21594422e-01 9.99316692e-01 -2.85251826e-01 -1.93828195e-01
1.62224248e-01 -8.59283209e-01 -3.73280078e-01 2.29847282e-01
4.05603677e-01 -3.30656648e-01 -3.48752201e-01 1.24476922e+00
4.57383156e-01 -1.38900220e-01 6.71492696e-01 -1.46591103e+00
-8.08173478e-01 1.21065706e-01 -1.05680716e+00 2.48631492e-01
3.58630240e-01 1.87429085e-01 1.10012722e+00 -8.64563584e-01
5.57614684e-01 1.44243681e+00 2.64645308e-01 5.33800602e-01
-9.61966276e-01 -8.92845631e-01 3.29468369e-01 4.67166692e-01
-1.75424707e+00 -4.31278199e-01 1.07274663e+00 -4.14749801e-01
8.60551178e-01 1.00083135e-01 8.32140744e-01 8.08942020e-01
3.29677790e-01 8.47514391e-01 1.25379741e+00 -1.00726552e-01
-2.01676607e-01 -3.30134153e-01 -2.12661743e-01 1.17009485e+00
-3.09541583e-01 -1.25089556e-01 -7.60836720e-01 -6.89797103e-02
1.05371332e+00 1.73137546e-01 -2.24479333e-01 -2.91075949e-02
-6.48504376e-01 7.95212746e-01 8.09261918e-01 4.27303195e-01
-6.10091984e-01 2.72783458e-01 4.55429435e-01 2.86615998e-01
5.42165697e-01 -2.16156557e-01 -2.99563140e-01 -1.58300195e-02
-6.55037165e-01 -6.80253506e-02 2.02070147e-01 5.26590407e-01
1.19997799e+00 1.13387227e-01 -3.06148529e-01 1.00121474e+00
4.72551763e-01 2.00671956e-01 5.87850988e-01 -7.82186806e-01
1.69861749e-01 1.03325450e+00 -6.73818886e-01 -1.52655160e+00
-3.86267632e-01 -1.15765110e-02 -9.19346392e-01 2.36139670e-01
7.66567215e-02 -1.14314459e-01 -7.16816127e-01 1.98180985e+00
2.89294988e-01 2.06054688e-01 -3.21947575e-01 1.02087533e+00
1.14517367e+00 5.33233345e-01 2.19979912e-01 -5.27225673e-01
1.40185273e+00 -7.10540891e-01 -8.07015359e-01 -1.10310651e-01
5.83279490e-01 -3.18489343e-01 4.97197479e-01 1.63986102e-01
-8.60867202e-01 -6.34711087e-01 -8.34124386e-01 1.14306591e-01
-3.60874027e-01 1.67138278e-01 7.39551365e-01 2.86280096e-01
-9.96530116e-01 2.15627745e-01 -4.56965387e-01 -2.23823622e-01
9.38340366e-01 6.51914656e-01 -8.02415252e-01 -8.35587922e-03
-1.26378787e+00 4.96600866e-01 -1.17688157e-01 3.81745398e-01
-7.10181415e-01 -1.86004207e-01 -9.01907563e-01 1.82794794e-01
4.71738696e-01 -1.34527221e-01 7.11076915e-01 -1.47223949e+00
-1.59384823e+00 7.51009285e-01 -2.16540486e-01 1.29125968e-01
-1.59108005e-02 3.10701311e-01 -4.44145530e-01 4.26731646e-01
-2.03669459e-01 8.15429330e-01 9.24893260e-01 -8.99551749e-01
-3.55011731e-01 -7.74037480e-01 -3.74327786e-02 2.60754257e-01
-4.88999158e-01 3.69451493e-01 -3.51240784e-01 -5.71125746e-01
2.29889020e-01 -5.61547160e-01 1.15916222e-01 4.51845080e-01
1.92597017e-01 -6.36275768e-01 9.04961884e-01 -4.96269345e-01
1.19868171e+00 -2.19968748e+00 2.69614875e-01 3.80232900e-01
2.57297993e-01 1.42028764e-01 -2.43495554e-01 1.84137702e-01
-2.56818384e-01 3.78742144e-02 5.75243905e-02 -1.25003792e-02
-3.51623118e-01 2.57991284e-01 2.25915298e-01 5.44702768e-01
5.99220991e-01 1.09844458e+00 -7.71542192e-01 -8.24040413e-01
1.13323502e-01 5.69374561e-01 -4.86629367e-01 7.41588771e-02
-1.27195939e-01 3.86067986e-01 -7.74713933e-01 1.19357228e+00
4.93394911e-01 -1.44727245e-01 1.58323124e-01 -3.77830476e-01
3.18143815e-01 -3.88205171e-01 -8.71652782e-01 1.72892642e+00
-2.72411853e-01 4.53800201e-01 3.17882121e-01 -1.13706875e+00
1.12606096e+00 4.28067565e-01 9.02307928e-01 -8.88015330e-01
5.47289073e-01 -1.87302027e-02 2.44676813e-01 -6.47451043e-01
-1.15537584e-01 -2.72311419e-01 1.06448606e-01 3.54185581e-01
4.62794662e-01 5.71116626e-01 -2.03381106e-01 -5.88546917e-02
9.68862057e-01 1.72815457e-01 4.90590960e-01 -4.01892364e-02
8.07382405e-01 -6.42279863e-01 8.68616641e-01 -2.41281446e-02
-5.52989542e-01 3.23366046e-01 9.91295159e-01 -3.19760203e-01
-2.28448004e-01 -5.17829955e-01 -2.26838306e-01 1.28799641e+00
1.19091339e-01 -5.98361611e-01 -7.62766659e-01 -8.84478748e-01
1.13323465e-01 -2.47179642e-01 -1.20528197e+00 -3.51590306e-01
-2.47690812e-01 -7.58693337e-01 4.31818992e-01 5.99985659e-01
7.92882740e-01 -1.28812957e+00 -3.49853784e-01 -3.29543976e-03
9.31503251e-02 -9.39474285e-01 -3.48046690e-01 -5.14470488e-02
-5.21505296e-01 -9.87947702e-01 -5.04601836e-01 -6.48026228e-01
7.36247420e-01 1.34026691e-01 5.36841631e-01 2.54601181e-01
-5.72279394e-01 2.35588714e-01 -4.83156592e-01 -1.41052008e-01
1.16744995e-01 -4.42864329e-01 -7.76348412e-02 9.08152699e-01
6.58271074e-01 -5.48932850e-01 -5.81459403e-01 3.07568848e-01
-8.04872513e-01 -2.51311302e-01 7.86463857e-01 9.25723910e-01
7.16983497e-01 1.04008019e-02 6.81115746e-01 -6.95198715e-01
6.27253473e-01 -5.66491604e-01 -5.83460145e-02 2.93501467e-01
-2.26614863e-01 -2.49269709e-01 3.40951264e-01 -3.15667510e-01
-1.00238991e+00 6.68129399e-02 -1.23152077e-01 -7.79505432e-01
-1.09564699e-01 6.20987117e-01 -5.95816076e-01 -5.38012445e-01
4.92886633e-01 1.77204445e-01 4.37156290e-01 -2.87096240e-02
1.93258941e-01 7.34205723e-01 1.34120762e-01 -5.18282771e-01
2.52151489e-01 4.86590654e-01 2.21942231e-01 -7.09671080e-01
-5.49983323e-01 -2.93734938e-01 -7.51776516e-01 -6.30037546e-01
1.18538296e+00 -9.46362138e-01 -7.84189463e-01 6.20177031e-01
-8.28487039e-01 -2.85221059e-02 4.24038619e-02 1.20871298e-01
-5.80564260e-01 1.80882290e-01 -7.31327415e-01 -7.59364665e-01
-1.84589252e-01 -1.32563591e+00 1.33463252e+00 3.19171578e-01
-3.56502831e-02 -7.40280390e-01 -2.55605996e-01 2.20136791e-01
1.11800961e-01 5.48078835e-01 7.62086868e-01 -1.78869337e-01
-4.49878663e-01 -3.62189621e-01 -4.65181142e-01 4.96489048e-01
5.16906381e-01 1.81447491e-01 -1.01432502e+00 1.03886150e-01
-1.36467516e-01 -5.61750889e-01 7.81454146e-01 3.31296712e-01
1.47149062e+00 -2.88346559e-01 -2.07128778e-01 5.63275695e-01
1.21528721e+00 2.88169891e-01 6.70257688e-01 -1.48153588e-01
8.78009975e-01 6.61248386e-01 7.38497615e-01 5.81905186e-01
4.18163389e-02 8.74205351e-01 7.46978223e-01 -1.79797456e-01
-2.13339105e-02 2.45221443e-02 5.39913297e-01 4.03532028e-01
-4.93777275e-01 2.03917906e-01 -4.37633723e-01 1.85643449e-01
-1.98280430e+00 -1.06465113e+00 3.21744740e-01 1.66445470e+00
5.90893984e-01 -2.50864953e-01 2.89403081e-01 -7.92605728e-02
5.18978119e-01 4.37546134e-01 -4.54268306e-01 -6.45524561e-01
-2.51907170e-01 1.44297808e-01 3.43073420e-02 2.59391516e-02
-8.82461548e-01 9.84349132e-01 5.42940950e+00 1.08881867e+00
-1.35708714e+00 2.36810133e-01 8.55685055e-01 -1.23044653e-02
8.94748643e-02 -4.19670820e-01 -4.83233303e-01 3.36238414e-01
5.26851058e-01 9.58067030e-02 3.94552886e-01 8.25417876e-01
5.11519052e-02 5.14416099e-02 -8.80952477e-01 1.24568439e+00
4.63716090e-01 -1.11189723e+00 1.96618333e-01 2.69264132e-01
5.15298724e-01 -2.88214654e-01 3.27585079e-02 1.61564678e-01
-1.29583910e-01 -1.29685712e+00 3.85870636e-01 6.82195723e-01
9.57611203e-01 -8.12700331e-01 8.17445040e-01 7.50822946e-02
-1.54540372e+00 -2.13473991e-01 -3.56264353e-01 -6.29367083e-02
-1.61186501e-01 2.10371651e-02 -3.18818003e-01 5.07133782e-01
7.93405235e-01 1.21735239e+00 -5.09327173e-01 2.47101679e-01
-2.57295400e-01 3.05381477e-01 -1.21607631e-01 -5.66585362e-02
3.03888857e-01 -3.83247495e-01 9.41221043e-02 1.03690028e+00
1.36522785e-01 5.30355453e-01 -2.11089998e-02 7.98035383e-01
-6.62024394e-02 4.19442087e-01 -6.27601266e-01 -2.09768936e-01
-1.17460549e-01 1.59860134e+00 -4.32488829e-01 -6.97604567e-02
-8.56218040e-01 8.46037328e-01 5.46897650e-01 3.15078139e-01
-7.33731389e-01 -2.59831361e-02 1.04986072e+00 1.54144734e-01
2.29832619e-01 1.59278326e-02 6.93296418e-02 -9.06084359e-01
-3.31171490e-02 -8.03084433e-01 5.87397397e-01 -7.83978939e-01
-1.25693381e+00 6.98792219e-01 -1.77511320e-01 -1.26338196e+00
-1.17633417e-01 -8.10187340e-01 -6.87524855e-01 7.91424155e-01
-1.33873188e+00 -1.49588275e+00 -5.60240507e-01 1.01222551e+00
3.16075295e-01 -2.99081057e-01 7.61188269e-01 1.70339257e-01
-7.99945533e-01 8.72489870e-01 -5.17430425e-01 5.20321906e-01
4.20796454e-01 -6.83586240e-01 -6.01487458e-01 4.48478073e-01
1.54606342e-01 5.45645118e-01 -1.60207283e-02 -4.20914382e-01
-1.42191672e+00 -1.02775586e+00 6.01356328e-01 1.14097483e-02
6.48218870e-01 -2.38569677e-01 -9.08705115e-01 6.29668772e-01
-3.58680487e-02 6.94233477e-01 7.09245563e-01 9.03940201e-02
-4.46031839e-01 -5.33156753e-01 -1.10965443e+00 3.63591790e-01
1.17043436e+00 -7.23561466e-01 -1.07913781e-02 1.09721422e-01
1.40750155e-01 -8.68748501e-02 -1.25029922e+00 6.00657463e-01
7.29058564e-01 -1.13418007e+00 5.68674922e-01 -6.14189625e-01
5.96316159e-01 -2.94760555e-01 -2.41660625e-01 -9.86549616e-01
-5.14229834e-01 -2.61341065e-01 -5.51718287e-02 1.30680525e+00
1.35055091e-02 -5.44242322e-01 6.04271293e-01 3.03378969e-01
5.01185320e-02 -1.30972469e+00 -1.20108914e+00 -2.89262533e-01
-2.85450667e-01 -2.97100812e-01 6.82574213e-01 9.13749576e-01
2.59499043e-01 3.36366266e-01 -3.78877074e-01 -1.19894683e-01
7.96094835e-02 2.22805932e-01 4.55086589e-01 -1.06827331e+00
-1.58595577e-01 -8.12571466e-01 -1.20961428e+00 -6.19442284e-01
4.55578685e-01 -9.89332795e-01 -1.59762293e-01 -1.24439240e+00
3.85157466e-01 -2.32476264e-01 -6.47909462e-01 9.20983076e-01
-1.36318132e-01 4.88593876e-01 1.15533672e-01 2.06711907e-02
-5.79910040e-01 7.94504166e-01 1.53203166e+00 -1.40956074e-01
4.40147482e-02 -4.40243155e-01 -6.94622636e-01 5.75371265e-01
4.85734940e-01 -5.71222268e-02 -2.47928947e-01 -7.07893670e-02
-8.83155987e-02 1.30043551e-01 3.74598533e-01 -8.69314969e-01
1.16245605e-01 -3.45906258e-01 6.45120442e-01 -6.10613264e-02
5.79135299e-01 -8.32403839e-01 -1.43195912e-01 1.66377738e-01
-1.31116971e-01 -2.44667292e-01 5.07063232e-02 5.34661651e-01
-7.20523536e-01 4.23478782e-01 8.25350463e-01 -2.72970617e-01
-9.96654332e-01 9.57368433e-01 -1.95897728e-01 -4.88046587e-01
1.24580061e+00 -3.92763555e-01 -1.64168850e-01 -5.38592219e-01
-8.38608801e-01 7.76122585e-02 2.13020667e-01 6.39019370e-01
6.43354654e-01 -1.58850229e+00 -5.96325696e-01 5.93716204e-01
4.43705767e-01 -2.29820117e-01 5.27145028e-01 1.12293100e+00
-2.00255781e-01 1.84657380e-01 -7.13016212e-01 -6.66456819e-01
-1.53129041e+00 3.95952940e-01 5.53212762e-01 3.32175605e-02
-1.77977815e-01 1.23570204e+00 6.50890827e-01 4.34453674e-02
-5.51632792e-02 1.38097510e-01 -6.37887120e-01 3.38562965e-01
4.55821007e-01 5.36194220e-02 -5.39327152e-02 -1.39928579e+00
-4.42629158e-01 1.02976608e+00 9.95573923e-02 4.28873867e-01
1.12405205e+00 3.71869355e-02 -4.56927747e-01 2.67075062e-01
1.48356771e+00 -2.79276282e-01 -1.25685787e+00 -3.31766456e-01
-4.55886632e-01 -5.25547147e-01 2.40718246e-01 -4.51959044e-01
-1.73810542e+00 1.13891625e+00 6.11058056e-01 -1.45069852e-01
1.64377022e+00 2.83996105e-01 4.14352119e-01 -1.82584330e-01
4.11817938e-01 -9.55329895e-01 3.43180895e-01 3.75949964e-03
1.00827920e+00 -1.17474842e+00 -6.45421892e-02 -6.11181915e-01
-7.18101025e-01 1.22921479e+00 1.07086861e+00 -1.30504593e-01
8.38884175e-01 7.19785541e-02 8.76255240e-03 -5.38254023e-01
-7.59768248e-01 -4.02322084e-01 4.72785443e-01 4.80612844e-01
5.04781663e-01 5.16725518e-02 -2.27924332e-01 7.30607092e-01
1.96641743e-01 7.04660416e-02 -2.03933597e-01 7.86718071e-01
-3.74232560e-01 -1.02329743e+00 -5.49844094e-02 7.11662292e-01
-4.81959283e-01 1.50236681e-01 -6.76375568e-01 8.69543731e-01
6.16195023e-01 7.56197572e-01 2.03543708e-01 -7.39942610e-01
5.61787821e-02 5.76806292e-02 6.71124399e-01 -4.72424299e-01
-4.96005446e-01 2.42869943e-01 -1.93774149e-01 -9.90289271e-01
-6.90075159e-01 -7.50770569e-01 -1.47309923e+00 -1.32406995e-01
-9.89630818e-02 -2.17325613e-01 2.97892928e-01 9.00838494e-01
4.17618692e-01 4.50773329e-01 7.95128226e-01 -7.69302726e-01
8.12377557e-02 -1.05904567e+00 -1.01154995e+00 5.28740227e-01
2.30382890e-01 -1.08759022e+00 -2.80888289e-01 -1.28425151e-01] | [13.645766258239746, 1.6398398876190186] |
e7c23473-00f5-4fce-aa8a-c6420dcd1bc4 | globally-gated-deep-linear-networks | 2210.17449 | null | https://arxiv.org/abs/2210.17449v2 | https://arxiv.org/pdf/2210.17449v2.pdf | Globally Gated Deep Linear Networks | Recently proposed Gated Linear Networks present a tractable nonlinear network architecture, and exhibit interesting capabilities such as learning with local error signals and reduced forgetting in sequential learning. In this work, we introduce a novel gating architecture, named Globally Gated Deep Linear Networks (GGDLNs) where gating units are shared among all processing units in each layer, thereby decoupling the architectures of the nonlinear but unlearned gatings and the learned linear processing motifs. We derive exact equations for the generalization properties in these networks in the finite-width thermodynamic limit, defined by $P,N\rightarrow\infty, P/N\sim O(1)$, where P and N are the training sample size and the network width respectively. We find that the statistics of the network predictor can be expressed in terms of kernels that undergo shape renormalization through a data-dependent matrix compared to the GP kernels. Our theory accurately captures the behavior of finite width GGDLNs trained with gradient descent dynamics. We show that kernel shape renormalization gives rise to rich generalization properties w.r.t. network width, depth and L2 regularization amplitude. Interestingly, networks with sufficient gating units behave similarly to standard ReLU networks. Although gatings in the model do not participate in supervised learning, we show the utility of unsupervised learning of the gating parameters. Additionally, our theory allows the evaluation of the network's ability for learning multiple tasks by incorporating task-relevant information into the gating units. In summary, our work is the first exact theoretical solution of learning in a family of nonlinear networks with finite width. The rich and diverse behavior of the GGDLNs suggests that they are helpful analytically tractable models of learning single and multiple tasks, in finite-width nonlinear deep networks. | ['Haim Sompolinsky', 'Qianyi Li'] | 2022-10-31 | null | null | null | null | ['l2-regularization'] | ['methodology'] | [ 6.27025217e-02 9.23568159e-02 -1.58743426e-01 -2.81032354e-01
6.46097288e-02 -4.67130512e-01 6.01151586e-01 5.08804582e-02
-7.04797566e-01 9.93373990e-01 -1.24906778e-01 -3.92616570e-01
-4.62234139e-01 -9.25394535e-01 -8.83691192e-01 -1.56470251e+00
-5.70600510e-01 4.56799686e-01 6.27774537e-01 -5.38016915e-01
9.71683413e-02 7.02421367e-01 -1.29883528e+00 1.05588593e-01
7.37783134e-01 8.92310321e-01 1.96786419e-01 8.75968158e-01
9.44756791e-02 6.89805031e-01 1.87001806e-02 1.07008033e-01
4.34542745e-01 -4.37768340e-01 -5.06257534e-01 -5.53094745e-01
4.68343616e-01 7.76208341e-02 -5.19838989e-01 9.39707339e-01
5.57978451e-01 6.26996338e-01 8.13498139e-01 -6.20057642e-01
-7.51796067e-01 7.33485162e-01 -1.72315761e-01 5.16257763e-01
-5.61219335e-01 1.02065362e-01 1.21867657e+00 -9.09124434e-01
5.73979795e-01 1.10852921e+00 7.85442829e-01 8.86655986e-01
-1.69294262e+00 -5.70867181e-01 3.17539871e-01 -3.09804082e-01
-1.11433530e+00 -3.68165016e-01 4.51376736e-01 -3.50021303e-01
1.43723583e+00 -1.12321526e-01 6.56087995e-01 8.55631053e-01
6.33434474e-01 1.60380527e-01 1.11400592e+00 -8.32321405e-01
3.60388517e-01 6.41560480e-02 7.40707159e-01 1.05381644e+00
6.31355941e-01 3.59424591e-01 -7.25018501e-01 -1.16206683e-01
9.87627745e-01 -5.06784860e-03 -3.72855008e-01 -5.12318432e-01
-7.41579890e-01 9.41272259e-01 5.67148268e-01 5.15111387e-01
1.64249574e-03 5.69725633e-01 4.51057166e-01 5.64249396e-01
5.55900335e-01 3.17663401e-01 -8.23503673e-01 3.33196253e-01
-7.59989917e-01 3.36249441e-01 9.23644423e-01 7.41527319e-01
1.07636845e+00 2.43027464e-01 6.16815463e-02 5.01314759e-01
3.39301944e-01 3.92796218e-01 6.31723523e-01 -6.59860194e-01
3.02638888e-01 3.67323369e-01 -2.66779572e-01 -3.02872866e-01
-8.11152577e-01 -6.09370589e-01 -1.08974504e+00 3.25003356e-01
6.51707649e-01 -2.27477506e-01 -1.00545645e+00 2.14354587e+00
-1.57049865e-01 -2.52312690e-01 5.67472354e-02 5.21185577e-01
5.04736722e-01 5.95922053e-01 3.57379615e-01 -4.18025672e-01
9.96565700e-01 -7.31058478e-01 -3.59765679e-01 -2.64051467e-01
8.54908109e-01 -9.76326838e-02 1.03810596e+00 2.98325747e-01
-1.21388924e+00 -5.06583750e-01 -9.43777561e-01 -1.63034841e-01
-4.72727269e-01 -1.38055518e-01 9.27791476e-01 6.51233912e-01
-1.51909316e+00 1.34666729e+00 -1.06374633e+00 -4.15357649e-01
3.04147780e-01 9.19999719e-01 -2.38249898e-01 1.96589589e-01
-1.28930914e+00 8.76391292e-01 6.98783457e-01 2.26086870e-01
-8.12686265e-01 -6.22449398e-01 -3.83131981e-01 7.69189699e-03
-9.48957279e-02 -8.07807326e-01 8.93487871e-01 -6.48035586e-01
-1.54278851e+00 6.82918847e-01 -1.91604882e-01 -6.79325521e-01
1.84972301e-01 -1.19490482e-01 3.90518010e-02 -1.44466516e-02
-3.85865659e-01 4.06041592e-01 8.71452868e-01 -9.08640742e-01
-6.40018210e-02 -5.08127153e-01 -1.29720956e-01 8.07161182e-02
-3.91946286e-01 -4.18943256e-01 1.76073462e-01 -3.26413184e-01
2.85137653e-01 -9.49474454e-01 -4.13556367e-01 -2.91192949e-01
-4.96096462e-02 -1.29191667e-01 5.54249108e-01 -3.31487030e-01
1.20162117e+00 -1.98688555e+00 3.97094190e-01 5.07351756e-01
4.78248894e-01 6.19840771e-02 -3.98747437e-02 4.32583958e-01
-1.40420616e-01 2.37038091e-01 -1.86903894e-01 -1.62639394e-01
-7.41456002e-02 6.34627700e-01 -4.74234670e-01 4.47007090e-01
2.41921544e-02 1.19068921e+00 -6.12068772e-01 -1.32127935e-02
-2.33846337e-01 5.30681729e-01 -7.57318616e-01 4.24462222e-02
-1.66984990e-01 4.59200114e-01 -3.11350763e-01 1.90797120e-01
5.11729300e-01 -2.58186549e-01 3.36819023e-01 -1.36803672e-01
-2.98806816e-01 2.21466586e-01 -6.58193767e-01 1.28246272e+00
-3.66231203e-01 5.08747637e-01 2.07270429e-01 -1.42761564e+00
7.73032069e-01 2.02698857e-01 -9.76144820e-02 -6.31042659e-01
-3.90630588e-02 5.14008701e-01 4.40482140e-01 -1.40461773e-01
1.95691958e-01 -8.42927873e-01 9.49580744e-02 3.66993368e-01
8.21295142e-01 3.07579730e-02 3.67805034e-01 1.45697489e-01
1.11602199e+00 -1.84504911e-01 2.29561910e-01 -7.94296443e-01
3.27966779e-01 -5.99038243e-01 2.82060057e-01 1.34573495e+00
7.62244407e-03 2.06552565e-01 8.62837493e-01 -4.48332697e-01
-1.33631480e+00 -1.17847621e+00 -3.51110011e-01 1.91268218e+00
-2.59710610e-01 -1.88691542e-01 -6.73434615e-01 -1.67631581e-01
-6.01304695e-02 3.15294087e-01 -8.88007760e-01 -5.59809327e-01
-9.89301741e-01 -1.37886679e+00 6.52144074e-01 4.28227931e-01
4.07969326e-01 -1.18276834e+00 -2.32079908e-01 1.90204099e-01
7.01915264e-01 -7.48887360e-01 -2.11373255e-01 1.18642843e+00
-1.65597951e+00 -7.81713724e-01 -3.50435555e-01 -8.68396103e-01
7.98740149e-01 -2.09072217e-01 9.90244925e-01 9.63257775e-02
-2.35283405e-01 1.86840445e-01 9.14413854e-02 1.78650897e-02
-2.59530336e-01 6.08309686e-01 3.55930716e-01 -3.80443543e-01
8.32750350e-02 -1.11354637e+00 -6.87623501e-01 1.06972791e-01
-9.19780493e-01 -2.41823837e-01 6.71942413e-01 1.13221705e+00
5.02781749e-01 -8.39615911e-02 4.86960351e-01 -1.11515093e+00
8.52259755e-01 -1.70354113e-01 -5.64353526e-01 9.89995077e-02
-7.23974884e-01 7.35642016e-01 9.37314868e-01 -5.97324848e-01
-1.12870884e+00 -2.37752020e-01 -7.09000975e-02 1.97228789e-01
7.66023993e-02 3.60948801e-01 8.62926468e-02 -6.90689325e-01
9.59584713e-01 2.94914842e-01 -1.97077349e-01 -3.47829014e-01
5.35706878e-01 -2.52831548e-01 1.13009967e-01 -9.21742141e-01
6.85659587e-01 4.92079109e-01 3.90188813e-01 -1.03800607e+00
-9.12474036e-01 -1.51778474e-01 -9.31301534e-01 7.06547350e-02
6.78714156e-01 -6.76069081e-01 -8.49356115e-01 4.58956659e-01
-9.95962739e-01 -8.25135052e-01 -7.48540938e-01 4.87859279e-01
-6.21571183e-01 -1.12553455e-01 -1.14502525e+00 -8.57117772e-01
-4.61433232e-01 -6.71555758e-01 3.09690773e-01 3.08499098e-01
1.75517753e-01 -1.71977186e+00 2.65768528e-01 -3.01297009e-01
7.57997870e-01 -1.79767758e-01 1.50203276e+00 -8.79961014e-01
-4.43786740e-01 8.87222290e-02 -1.58952370e-01 6.23477221e-01
-4.01185781e-01 -2.10494578e-01 -9.76418853e-01 -6.11892462e-01
2.29596853e-01 -4.19617057e-01 1.76006937e+00 7.11666167e-01
8.37830186e-01 -2.41782412e-01 -2.42560074e-01 8.72512996e-01
1.56104672e+00 -3.21812518e-02 3.47630292e-01 -9.36420262e-02
7.36044943e-01 2.89364487e-01 -4.28481668e-01 1.51032388e-01
-3.87706965e-01 -6.05642833e-02 2.14466184e-01 2.15637639e-01
8.73976871e-02 -1.32032573e-01 6.34250700e-01 1.33759141e+00
-8.63428414e-01 6.01418428e-02 -8.43910992e-01 6.61645755e-02
-1.92433119e+00 -8.26265335e-01 -8.11566487e-02 2.35986829e+00
8.90735865e-01 6.73647404e-01 -3.16533774e-01 -4.05073106e-01
3.99628878e-01 1.95531413e-01 -7.54271030e-01 -5.82910001e-01
-4.78504449e-01 1.06351030e+00 9.72735882e-01 8.13782632e-01
-7.71198332e-01 1.03279722e+00 6.98080826e+00 8.33937287e-01
-9.73525643e-01 5.51700652e-01 4.14207429e-01 -3.08755010e-01
-2.56353348e-01 2.69526333e-01 -1.28245199e+00 -4.89720367e-02
1.35140300e+00 6.38832077e-02 5.60851693e-01 4.70079362e-01
8.54001790e-02 -6.17518574e-02 -1.00416183e+00 3.94097149e-01
-2.07049638e-01 -1.39137185e+00 2.02431738e-01 1.88344523e-01
8.77496660e-01 4.92959023e-01 2.25385725e-01 5.34983218e-01
4.12918925e-01 -1.11344278e+00 3.12545449e-01 7.45342553e-01
5.91529429e-01 -6.65679395e-01 4.67047691e-01 4.54178303e-01
-1.10087812e+00 -3.54724824e-01 -8.66077900e-01 -4.15020943e-01
-1.08203538e-01 7.22495854e-01 -2.68649280e-01 -3.55100594e-02
3.09954762e-01 4.94295418e-01 -5.76891303e-01 6.79816246e-01
-1.09301470e-01 7.66827643e-01 -4.22236830e-01 -1.46637559e-01
2.88324803e-01 -5.72830975e-01 2.66777456e-01 1.28829741e+00
2.78635740e-01 1.34692267e-01 -7.87376091e-02 8.83264661e-01
-3.52121770e-01 2.32431535e-02 -5.93827784e-01 -2.20928758e-01
-2.89787594e-02 1.20155418e+00 -9.42442894e-01 -5.89141399e-02
-1.87846959e-01 7.06972718e-01 9.85242486e-01 7.60659218e-01
-3.71917188e-01 -3.66973162e-01 5.44448972e-01 3.31193417e-01
4.80294973e-01 -6.65006578e-01 -1.27048060e-01 -1.24464107e+00
-1.64743856e-01 -1.17609113e-01 -1.63917486e-02 -2.20630467e-01
-1.40978694e+00 4.62820649e-01 -1.69992492e-01 -3.69715363e-01
1.74838230e-01 -1.27203000e+00 -7.70113707e-01 9.40139174e-01
-1.36359239e+00 -9.89842713e-01 2.60169595e-01 6.16446555e-01
-2.14839891e-01 -1.98774889e-01 1.00024152e+00 8.11333284e-02
-5.38655281e-01 3.74487162e-01 6.29658580e-01 -6.69002458e-02
4.91919518e-01 -1.36439431e+00 2.79165894e-01 4.09279913e-01
-4.83524501e-02 1.23753846e+00 6.79817796e-01 -5.16884327e-01
-1.12054121e+00 -9.27079678e-01 7.50371099e-01 -1.61261991e-01
1.06483078e+00 -8.83113921e-01 -1.29150796e+00 7.47373581e-01
-1.13702647e-03 5.23507297e-01 7.09892690e-01 2.77021021e-01
-3.56400907e-01 2.96859466e-03 -7.33759284e-01 3.63040835e-01
1.40518534e+00 -7.59582520e-01 -4.35886353e-01 3.88372362e-01
8.31718802e-01 -1.91698730e-01 -6.19312406e-01 1.86207145e-01
5.39627910e-01 -1.11080718e+00 7.83221126e-01 -1.09499109e+00
-1.06488235e-01 3.18378747e-01 9.34552029e-03 -1.18961036e+00
-5.30588627e-01 -6.51758194e-01 -2.95175284e-01 5.49825013e-01
6.79970741e-01 -9.81814504e-01 8.47211361e-01 5.60494542e-01
-5.08994497e-02 -8.56814504e-01 -1.20243585e+00 -6.67775869e-01
7.24240124e-01 -3.88317376e-01 -3.48605543e-01 4.86950219e-01
-1.92679703e-01 4.17755991e-01 -6.10001758e-02 -3.22523937e-02
7.14754939e-01 -2.89504707e-01 -1.13182046e-01 -1.56613040e+00
-4.09034282e-01 -4.21791732e-01 -1.81360051e-01 -1.07733798e+00
3.84004086e-01 -1.25181592e+00 -2.18359411e-01 -1.04178417e+00
2.15427339e-01 -5.50908029e-01 -5.70156991e-01 3.07936251e-01
2.96724111e-01 7.07873777e-02 -1.24358147e-01 3.81265193e-01
-5.50132692e-01 3.76200706e-01 1.18812716e+00 3.77613038e-01
-4.84839767e-01 5.34902290e-02 -4.36663747e-01 9.11458850e-01
9.99368787e-01 -5.61728060e-01 -3.14056605e-01 -6.00184351e-02
8.47047329e-01 -4.00446236e-01 4.82016772e-01 -8.82321239e-01
5.07372737e-01 1.05731614e-01 4.68360066e-01 2.71854326e-02
6.55942708e-02 -3.12443316e-01 -4.83898908e-01 5.75393617e-01
-5.83660066e-01 -1.87981874e-01 1.85149759e-01 6.31237686e-01
1.36273876e-01 -5.19405484e-01 9.72965658e-01 -5.06831110e-01
-3.03801119e-01 4.60501909e-01 -5.28232217e-01 3.16520572e-01
4.88956243e-01 4.10214216e-02 -2.91113943e-01 5.15722036e-02
-1.34125614e+00 -1.28302306e-01 1.34096071e-01 -3.04565936e-01
4.38379735e-01 -1.14657485e+00 -1.89031705e-01 4.78983790e-01
-3.93733710e-01 5.30143827e-02 2.83613920e-01 9.24504578e-01
-5.57957172e-01 4.78005856e-01 -3.04473251e-01 -2.98532069e-01
-7.27936566e-01 1.61126241e-01 6.93759978e-01 -8.03544343e-01
-2.65675366e-01 1.04689324e+00 3.60574007e-01 -7.11939931e-01
-6.06158897e-02 -7.32255459e-01 9.51826125e-02 2.31513754e-01
2.56458312e-01 2.64512956e-01 1.52333856e-01 -2.33381689e-01
9.46743190e-02 6.64967835e-01 -4.24926728e-01 6.16226979e-02
1.41472936e+00 1.44916847e-01 -6.95125341e-01 8.71200383e-01
1.16663146e+00 -2.84014702e-01 -1.30455065e+00 -3.99013489e-01
7.77496248e-02 4.43870753e-01 -8.93695951e-02 -3.90711695e-01
-9.06068146e-01 1.14135146e+00 6.23883188e-01 3.11356306e-01
7.17445612e-01 2.83109695e-01 4.10976052e-01 8.61579835e-01
3.15541923e-01 -1.17547452e+00 1.45570740e-01 1.21590185e+00
6.33775234e-01 -9.09193516e-01 -5.61387055e-02 2.85363030e-02
-4.21801023e-02 1.43300605e+00 5.73551357e-01 -6.25489593e-01
9.31040645e-01 3.25950652e-01 -4.89383250e-01 -1.75385565e-01
-9.97808456e-01 -9.65509117e-02 1.58690706e-01 4.71603692e-01
7.96361983e-01 1.61049236e-02 -3.70163500e-01 5.24300039e-01
-6.62903339e-02 -3.04929644e-01 2.54337668e-01 7.17456043e-01
-1.02845562e+00 -1.18781400e+00 1.82141438e-01 6.31065965e-01
-2.89550394e-01 -4.85371202e-01 -1.17376838e-02 9.09656405e-01
2.47650206e-01 2.66353600e-02 -7.98823833e-02 1.09186908e-02
2.47795105e-01 6.58294737e-01 8.62558126e-01 -7.59625733e-01
-6.84593976e-01 -2.27123484e-01 -3.85387778e-01 -2.50889868e-01
-3.11900884e-01 -1.77062526e-01 -1.63369405e+00 -2.85656095e-01
-4.47719783e-01 2.24360153e-01 2.57853270e-01 1.06623459e+00
7.26119503e-02 5.43460906e-01 -9.54928324e-02 -1.21853197e+00
-8.19816589e-01 -8.63043010e-01 -1.13089430e+00 2.20533311e-01
5.27411640e-01 -5.54981887e-01 -8.18974137e-01 -2.60421872e-01] | [7.989327907562256, 3.4949376583099365] |
dc25f760-49cd-4d43-904b-e9427b7de091 | televit-teleconnection-driven-transformers | 2306.10940 | null | https://arxiv.org/abs/2306.10940v1 | https://arxiv.org/pdf/2306.10940v1.pdf | TeleViT: Teleconnection-driven Transformers Improve Subseasonal to Seasonal Wildfire Forecasting | Wildfires are increasingly exacerbated as a result of climate change, necessitating advanced proactive measures for effective mitigation. It is important to forecast wildfires weeks and months in advance to plan forest fuel management, resource procurement and allocation. To achieve such accurate long-term forecasts at a global scale, it is crucial to employ models that account for the Earth system's inherent spatio-temporal interactions, such as memory effects and teleconnections. We propose a teleconnection-driven vision transformer (TeleViT), capable of treating the Earth as one interconnected system, integrating fine-grained local-scale inputs with global-scale inputs, such as climate indices and coarse-grained global variables. Through comprehensive experimentation, we demonstrate the superiority of TeleViT in accurately predicting global burned area patterns for various forecasting windows, up to four months in advance. The gain is especially pronounced in larger forecasting windows, demonstrating the improved ability of deep learning models that exploit teleconnections to capture Earth system dynamics. Code available at https://github.com/Orion-Ai-Lab/TeleViT. | ['Ioannis Papoutsis', 'Gustau Camps-Valls', 'Dimitrios Michail', 'Spyros Kondylatos', 'Nikolaos Ioannis Bountos', 'Ioannis Prapas'] | 2023-06-19 | null | null | null | null | ['management'] | ['miscellaneous'] | [-2.74922311e-01 -6.91807866e-01 3.01752985e-02 -8.62145126e-02
-8.09678510e-02 -7.57828653e-01 8.90743434e-01 2.22035855e-01
-2.69791722e-01 8.98029268e-01 3.46647382e-01 -8.37301075e-01
-3.42985690e-01 -1.24466693e+00 -3.91165465e-01 -6.43389225e-01
-7.86624372e-01 4.28748310e-01 4.05909419e-02 -7.07759738e-01
-5.87604232e-02 9.00862217e-01 -1.44922447e+00 -2.25252107e-01
8.36637378e-01 7.74255633e-01 3.00174385e-01 1.08391488e+00
9.95388106e-02 5.00548840e-01 -2.17616647e-01 4.11637157e-01
2.68629432e-01 3.40950377e-02 -6.06989920e-01 -3.44493330e-01
5.01463227e-02 -4.55226362e-01 -3.66011679e-01 7.82629669e-01
2.64218658e-01 2.24848226e-01 4.01035726e-01 -8.95912468e-01
-1.25496328e-01 3.05371642e-01 -4.32851434e-01 5.27315736e-01
-3.82724732e-01 6.92343771e-01 8.05694044e-01 -5.85188746e-01
2.17716098e-01 1.15272236e+00 9.17235315e-01 -2.93662578e-01
-1.53343427e+00 -5.39907694e-01 4.32231873e-01 9.50455889e-02
-1.23745358e+00 -3.45401466e-01 5.08684158e-01 -8.09824467e-01
1.16334414e+00 3.88326615e-01 9.51367319e-01 8.24100792e-01
8.21620524e-01 6.80996990e-03 1.30588639e+00 -8.72883946e-02
1.86429709e-01 -5.60094774e-01 2.70003602e-02 2.97229350e-01
6.75086007e-02 8.84946764e-01 -3.04767460e-01 4.92460430e-02
9.67777669e-01 4.74943787e-01 -3.55048209e-01 2.78901905e-01
-1.18596423e+00 6.87323093e-01 1.24170792e+00 2.90149510e-01
-1.10758662e+00 5.28485119e-01 2.81422794e-01 3.11307788e-01
9.50653851e-01 4.29662138e-01 -7.18311012e-01 -6.08535819e-02
-1.26555502e+00 3.29293817e-01 5.26371181e-01 3.85652602e-01
7.34134495e-01 2.82817006e-01 1.84923902e-01 6.04543686e-01
2.58663774e-01 1.29245746e+00 -8.70147645e-02 -9.75595951e-01
-9.89463106e-02 3.58878613e-01 3.13434303e-01 -1.04732585e+00
-6.52214646e-01 -8.08296859e-01 -1.66402912e+00 4.63465959e-01
1.44119948e-01 -3.33196282e-01 -1.05466771e+00 1.72416317e+00
7.26773217e-02 1.87417835e-01 -1.62695974e-01 1.08399010e+00
3.68079096e-02 1.13619113e+00 2.55985707e-01 -1.44531816e-01
1.25094450e+00 -4.87080276e-01 -5.52386642e-01 -5.49057841e-01
2.05974653e-01 -4.85901266e-01 6.17780626e-01 -1.52370036e-01
-4.43433434e-01 -3.37984681e-01 -6.08982027e-01 5.63694298e-01
-8.43798578e-01 -1.41836390e-01 8.63975942e-01 -3.17779541e-01
-1.14122939e+00 7.35178411e-01 -1.16837966e+00 -5.97490907e-01
9.84412357e-02 -1.41339144e-02 1.28487602e-01 2.59129971e-01
-1.59211886e+00 1.00208569e+00 8.13836679e-02 1.03930008e+00
-1.09068513e+00 -8.68052423e-01 -4.41239208e-01 4.76102829e-01
-1.01718850e-01 -7.73719013e-01 1.10613704e+00 -7.12033033e-01
-8.49684656e-01 2.64257312e-01 -1.30551293e-01 -7.28816807e-01
3.81317317e-01 -4.50618006e-02 -4.00156319e-01 -1.15603961e-01
1.49670750e-01 6.72486961e-01 6.92058802e-01 -1.20780718e+00
-6.04015529e-01 -6.45385444e-01 2.52601299e-02 2.16624513e-01
-4.45054956e-02 -8.59886333e-02 2.37449378e-01 -8.14680576e-01
2.34743040e-02 -1.24843216e+00 -6.19154692e-01 7.79065341e-02
-7.55048618e-02 3.93116593e-01 6.72121227e-01 -9.12231088e-01
1.07418430e+00 -1.80847228e+00 5.72539717e-02 4.86918539e-03
3.19970995e-01 4.54948395e-01 -1.20838374e-01 1.02184153e+00
5.24395630e-02 2.72315443e-01 -6.10163391e-01 7.60975853e-02
-1.12344556e-01 5.83446324e-01 -8.18818033e-01 1.95240915e-01
2.27632314e-01 7.11569905e-01 -8.34686816e-01 1.72483489e-01
5.59768319e-01 5.09339511e-01 1.36886731e-01 7.59272277e-02
-6.26322687e-01 7.05222726e-01 -3.81274223e-01 5.17248273e-01
8.77164304e-01 -6.61003739e-02 6.14386089e-02 3.31932366e-01
-7.31544077e-01 1.54422149e-01 -6.75583780e-01 8.89305174e-01
-8.46883118e-01 9.75604355e-01 5.51967561e-01 -5.86726010e-01
8.52513969e-01 3.62943769e-01 3.38386387e-01 -9.57451999e-01
-2.54421115e-01 2.99937099e-01 -7.39979744e-03 -2.32072785e-01
3.08939487e-01 -4.43354487e-01 2.29466483e-01 5.41749895e-01
-4.55957562e-01 -9.21420008e-02 -1.75242618e-01 1.87826920e-02
8.56446207e-01 -1.86884388e-01 -6.56933710e-02 -5.60681403e-01
1.34738475e-01 3.92603636e-01 4.82121706e-01 5.26673853e-01
-2.33977288e-01 2.13248298e-01 2.56090015e-01 -9.30424929e-01
-8.20118427e-01 -7.95059681e-01 -2.62925863e-01 8.90978873e-01
-2.44960085e-01 -6.70788344e-03 2.69142780e-02 -2.15836018e-01
4.61406529e-01 6.74472392e-01 -5.68453252e-01 -8.59717131e-02
-1.97505057e-01 -1.08692408e+00 5.50936460e-01 6.23125076e-01
6.70037687e-01 -1.02607739e+00 -8.38242173e-01 5.14245391e-01
-3.47344905e-01 -5.46288669e-01 1.32718071e-01 4.76194769e-01
-1.21786237e+00 -7.90716410e-01 -7.12684751e-01 -9.60259512e-02
2.67138898e-01 4.92740959e-01 1.20811903e+00 -3.95117477e-02
-2.09775463e-01 6.77847862e-02 -1.24760419e-01 -2.48408124e-01
-9.99542847e-02 2.31722385e-01 1.09828711e-01 -2.97993511e-01
-2.10454259e-02 -1.23044348e+00 -9.77872729e-01 2.30817705e-01
-8.12824905e-01 2.65883088e-01 5.97714424e-01 6.79643512e-01
2.37974182e-01 2.57527351e-01 5.75233281e-01 -3.93668890e-01
3.87950271e-01 -6.63415134e-01 -8.93365145e-01 -7.12542236e-02
-8.77885342e-01 9.64963213e-02 5.60284197e-01 4.30012010e-02
-1.04741502e+00 -1.46547973e-01 6.72722831e-02 -2.66935199e-01
-3.48679721e-01 1.15841484e+00 4.22123671e-01 1.62887663e-01
4.88087952e-01 3.35921556e-01 -1.29300192e-01 -5.62001288e-01
4.27646607e-01 7.21110046e-01 6.17564261e-01 -3.55388433e-01
9.06805992e-01 5.62179685e-01 -1.35189611e-02 -1.11008012e+00
-4.83856261e-01 -4.84504282e-01 -7.76536167e-01 -6.45981610e-01
5.22510290e-01 -1.27147520e+00 -5.32301545e-01 1.00830793e+00
-1.29112327e+00 -8.02278638e-01 7.41803497e-02 4.46432650e-01
-7.34232217e-02 -4.42481905e-01 -3.36355180e-01 -8.83153319e-01
-6.89501822e-01 -4.43496615e-01 7.08738625e-01 1.05022378e-01
4.70347367e-02 -1.36436808e+00 7.48166859e-01 -1.37362614e-01
1.15434110e+00 5.64782798e-01 8.41821313e-01 4.06277478e-01
-7.31349707e-01 -7.50527307e-02 -4.65538025e-01 1.72087848e-01
3.51498485e-01 2.64516592e-01 -9.61648703e-01 -3.07839900e-01
-2.39575744e-01 2.26990566e-01 1.40098989e+00 5.99196553e-01
6.12893224e-01 -6.09924734e-01 -3.13231170e-01 9.98738468e-01
1.64214087e+00 6.67128935e-02 4.72353309e-01 4.07458365e-01
4.66935277e-01 4.89592284e-01 4.03059274e-01 5.34696639e-01
5.82105935e-01 3.05336982e-01 9.50585365e-01 -5.51346958e-01
-6.10442087e-02 3.64008993e-02 1.48231372e-01 5.32468975e-01
-4.06790286e-01 -4.30779830e-02 -1.57426429e+00 1.01376712e+00
-1.93760955e+00 -1.39162409e+00 -3.62744093e-01 2.18528247e+00
4.22963887e-01 -8.68034512e-02 -1.14020079e-01 -3.79588813e-01
5.52062750e-01 8.76111567e-01 -9.02051389e-01 -3.32153171e-01
-1.36616588e-01 1.19602010e-01 9.49868679e-01 7.27686882e-01
-1.13483095e+00 1.09031141e+00 5.90546227e+00 -3.55181992e-02
-1.79662919e+00 1.84030861e-01 5.07574439e-01 5.33280522e-02
-3.46302718e-01 3.67886961e-01 -3.04972470e-01 2.15240940e-01
1.33472431e+00 -9.06057209e-02 8.56407225e-01 2.99495250e-01
1.09770322e+00 -3.29673022e-01 -2.63618439e-01 2.01982602e-01
-8.61695409e-01 -1.44999290e+00 -2.21953332e-01 3.25265050e-01
7.32779205e-01 8.24044824e-01 -1.14745438e-01 2.12419614e-01
6.52108192e-01 -8.83573771e-01 4.43277597e-01 8.90482962e-01
5.46923399e-01 -5.83274722e-01 5.74040353e-01 3.25076908e-01
-1.61430788e+00 -2.66358461e-02 -2.10016429e-01 -6.51915550e-01
8.04338008e-02 1.08706629e+00 -5.51314890e-01 4.97197300e-01
8.51393163e-01 1.09309566e+00 -2.74911851e-01 9.42843318e-01
-2.24253595e-01 1.02092564e+00 -6.97402894e-01 4.97184306e-01
4.85572100e-01 -5.07060647e-01 6.68220222e-01 9.37333345e-01
3.31074297e-01 2.83659071e-01 -4.10441756e-02 8.07132840e-01
1.92509919e-01 -5.38867176e-01 -8.71625423e-01 -4.39713150e-02
5.49644768e-01 1.19039059e+00 -3.64129782e-01 -1.44535214e-01
1.05713822e-01 6.88653469e-01 9.65208113e-02 7.93850780e-01
-6.64759517e-01 -2.41490141e-01 1.32573521e+00 1.61760285e-01
2.49437883e-01 -8.75924885e-01 -2.99499363e-01 -1.08213985e+00
-3.85917947e-02 -6.71964526e-01 8.74559432e-02 -9.50329304e-01
-8.41665030e-01 6.09290242e-01 -2.65354127e-01 -1.17443681e+00
-1.56273633e-01 -2.24270433e-01 -1.19703305e+00 1.19923830e+00
-2.08337355e+00 -1.46832097e+00 -4.73386705e-01 2.94935167e-01
1.46143585e-01 5.05690336e-01 1.06218719e+00 -6.80093318e-02
-6.00392520e-01 -6.36560857e-01 6.38391852e-01 -2.33929947e-01
5.48531651e-01 -1.33280385e+00 7.97444642e-01 9.86105919e-01
-2.76212931e-01 4.11129206e-01 7.50594497e-01 -7.25182235e-01
-1.44028234e+00 -1.63399255e+00 1.04028034e+00 9.32095498e-02
9.67082500e-01 -1.36290103e-01 -1.01581514e+00 8.50736380e-01
1.65896773e-01 1.16735905e-01 1.49610355e-01 4.55737591e-01
-3.68068099e-01 -7.06667781e-01 -7.35514939e-01 4.74638402e-01
6.33266866e-01 -7.34482586e-01 -2.66197294e-01 6.09320164e-01
3.79307985e-01 2.50030220e-01 -1.01819646e+00 4.85645741e-01
7.81096876e-01 -7.63663590e-01 6.77481472e-01 -3.77044469e-01
5.41284740e-01 -3.54394406e-01 -2.37528980e-01 -1.92809641e+00
-8.14112842e-01 -4.02756810e-01 8.82933736e-02 9.96439993e-01
2.65694827e-01 -1.10193515e+00 7.03929141e-02 1.49298370e-01
1.93932325e-01 -2.02709451e-01 -9.41652775e-01 -8.66108418e-01
3.56190145e-01 -3.01430225e-01 6.64721131e-01 9.22390223e-01
-2.69088358e-01 2.54439980e-01 -3.39590251e-01 9.93253112e-01
5.68610430e-01 5.79747677e-01 3.74630898e-01 -1.60716307e+00
1.62598670e-01 -5.64727783e-01 -3.49411145e-02 -6.11648977e-01
-1.35036990e-01 -4.77137864e-01 8.79662260e-02 -1.48323107e+00
-1.72354564e-01 -7.11916566e-01 -4.10922319e-01 8.21146250e-01
-5.52173443e-02 1.10257775e-01 1.63348794e-01 5.13094842e-01
3.79239559e-01 1.14520264e+00 9.76986885e-01 -3.44849348e-01
-7.51705691e-02 1.28063098e-01 -1.27742529e-01 3.99485618e-01
1.05676556e+00 -5.68145394e-01 3.55919311e-03 -9.11002040e-01
2.53504783e-01 3.97804916e-01 8.46996963e-01 -1.31359887e+00
1.19665228e-01 -6.71091735e-01 2.95685053e-01 -7.49618411e-01
1.98313311e-01 -8.35279882e-01 5.24086118e-01 7.11230338e-01
-3.61695774e-02 1.47832155e-01 6.64885819e-01 6.23647213e-01
-2.64397711e-01 6.38524592e-01 6.27501488e-01 -1.06075555e-01
-6.43081486e-01 6.26907468e-01 -9.10553992e-01 -4.30085152e-01
6.24236941e-01 4.98906553e-01 -6.10615492e-01 -4.52705473e-01
-8.27333868e-01 8.45216453e-01 5.85257173e-01 3.20236593e-01
2.25592002e-01 -8.97080541e-01 -9.48660553e-01 3.53049934e-01
2.56102555e-03 -1.93616301e-01 4.83461916e-01 8.12017262e-01
-7.22425818e-01 6.61886334e-01 -3.41895282e-01 -4.74308342e-01
-8.75860989e-01 2.07393855e-01 8.90847147e-01 -2.97510505e-01
-6.31537676e-01 5.14933586e-01 -1.08335838e-01 -7.66369820e-01
-3.55189174e-01 -6.96589887e-01 2.08825007e-01 3.57050300e-01
3.84452134e-01 1.11042693e-01 9.34418738e-02 -5.19203305e-01
-5.20747006e-01 4.29931581e-01 4.48046923e-01 -3.55748177e-01
1.75972342e+00 -3.17173928e-01 -3.14151704e-01 6.59130931e-01
5.69145322e-01 -6.55627429e-01 -1.59014475e+00 -2.56578237e-01
-1.41803175e-01 -4.48324710e-01 9.73031640e-01 -1.18801618e+00
-1.42856789e+00 1.18236721e+00 7.46887207e-01 3.51050287e-01
1.14881945e+00 -5.51482737e-01 8.67399931e-01 3.55450481e-01
3.18650991e-01 -6.05730474e-01 -7.48464882e-01 7.82605767e-01
1.26853979e+00 -1.07849586e+00 -1.76438689e-01 3.39639544e-01
-1.89333141e-01 1.01681364e+00 1.62602127e-01 -9.55281258e-02
9.71192539e-01 1.87637463e-01 3.71194124e-01 -3.44685078e-01
-1.13263226e+00 -5.15675485e-01 -2.11102262e-01 2.89691746e-01
-3.82322632e-03 9.00315344e-01 2.74647564e-01 -1.01631790e-01
7.65121356e-02 1.37712300e-01 3.35885078e-01 8.18502605e-01
-6.35199785e-01 -6.47457302e-01 -6.15898967e-01 4.55286264e-01
2.88878918e-01 -4.24530178e-01 -2.33133137e-01 3.76783729e-01
-3.10946256e-01 8.18717718e-01 2.96067774e-01 -1.76497892e-01
1.36312507e-02 -1.03828749e-02 -1.60867780e-01 -3.28486055e-01
-5.97018778e-01 -8.98441952e-03 -6.39721751e-02 -5.17749965e-01
-1.49421081e-01 -8.02930713e-01 -8.47224534e-01 -1.02921689e+00
-1.82317402e-02 -2.76645236e-02 8.35282803e-01 9.24349308e-01
7.26458490e-01 5.61579227e-01 8.20324957e-01 -1.41701603e+00
-4.47532952e-01 -1.31051636e+00 -7.57228196e-01 -3.64665180e-01
9.94393826e-01 -5.12916446e-01 -5.17835975e-01 -1.85626268e-01] | [9.497491836547852, -1.5511375665664673] |
9efd0593-8b92-47f1-bcb6-d49c64a365f6 | gpt-re-in-context-learning-for-relation | 2305.02105 | null | https://arxiv.org/abs/2305.02105v1 | https://arxiv.org/pdf/2305.02105v1.pdf | GPT-RE: In-context Learning for Relation Extraction using Large Language Models | In spite of the potential for ground-breaking achievements offered by large language models (LLMs) (e.g., GPT-3), they still lag significantly behind fully-supervised baselines (e.g., fine-tuned BERT) in relation extraction (RE). This is due to the two major shortcomings of LLMs in RE: (1) low relevance regarding entity and relation in retrieved demonstrations for in-context learning; and (2) the strong inclination to wrongly classify NULL examples into other pre-defined labels. In this paper, we propose GPT-RE to bridge the gap between LLMs and fully-supervised baselines. GPT-RE successfully addresses the aforementioned issues by (1) incorporating task-specific entity representations in demonstration retrieval; and (2) enriching the demonstrations with gold label-induced reasoning logic. We evaluate GPT-RE on four widely-used RE datasets, and observe that GPT-RE achieves improvements over not only existing GPT-3 baselines, but also fully-supervised baselines. Specifically, GPT-RE achieves SOTA performances on the Semeval and SciERC datasets, and competitive performances on the TACRED and ACE05 datasets. | ['Sadao Kurohashi', 'Jiwei Li', 'Haiyue Song', 'Qianying Liu', 'Zhuoyuan Mao', 'Fei Cheng', 'Zhen Wan'] | 2023-05-03 | null | null | null | null | ['relation-extraction'] | ['natural-language-processing'] | [ 1.03804981e-02 3.59187603e-01 -8.60366225e-01 -2.01456010e-01
-1.10980690e+00 -5.86814284e-01 9.72740650e-01 1.65407866e-01
-5.36234558e-01 9.96301949e-01 3.75929445e-01 -4.30488557e-01
-3.22089583e-01 -7.40653157e-01 -9.16257322e-01 -5.94448596e-02
2.73432918e-02 6.91915751e-01 1.89726666e-01 -2.97168344e-01
-1.93218276e-01 3.74016672e-01 -1.50533724e+00 3.82565141e-01
1.15120113e+00 7.92066395e-01 -1.92786291e-01 4.11615878e-01
2.65688561e-02 1.51019287e+00 -8.06670904e-01 -6.47789121e-01
-3.18979412e-01 -1.86720803e-01 -1.03689420e+00 -4.43104893e-01
7.39922881e-01 -2.20233321e-01 -6.91680551e-01 7.56521642e-01
2.10320607e-01 1.69906914e-01 1.10681868e+00 -1.68421948e+00
-7.71265209e-01 1.06390071e+00 -2.60578334e-01 1.10302478e-01
5.09124041e-01 8.12744256e-03 1.48356783e+00 -9.44768012e-01
9.93981361e-01 1.56160772e+00 5.24077654e-01 4.85648215e-01
-1.22150505e+00 -8.96541417e-01 2.92529404e-01 6.50609553e-01
-1.56173038e+00 -4.46719825e-01 5.17150760e-01 -5.09542227e-01
1.50363815e+00 3.22224647e-01 2.74178982e-01 1.51900458e+00
-1.54016212e-01 1.21009612e+00 1.15529597e+00 -5.74308634e-01
-8.11482370e-02 1.85263902e-01 5.04371464e-01 7.72152007e-01
3.72428089e-01 2.83890992e-01 -6.67691350e-01 3.37309316e-02
4.07777220e-01 -3.78844738e-01 -2.98788518e-01 -3.12243134e-01
-1.45323980e+00 6.16663754e-01 5.29790461e-01 2.47862265e-01
-4.56962474e-02 2.07385957e-01 4.57625896e-01 3.63081723e-01
2.84661114e-01 8.37394655e-01 -7.16367304e-01 -6.42192662e-02
-7.39162683e-01 5.20343840e-01 9.65861976e-01 1.46974945e+00
5.33192515e-01 -2.69362122e-01 -6.87368572e-01 6.89421296e-01
2.18379170e-01 5.50119162e-01 3.66769344e-01 -8.85369062e-01
9.91591156e-01 6.85945690e-01 4.58596796e-02 -7.31565952e-01
-3.94837022e-01 -3.72091144e-01 -4.52640772e-01 -2.32454449e-01
2.76935637e-01 -1.04864366e-01 -7.74200737e-01 1.95654559e+00
-3.83959822e-02 1.84016749e-01 6.06355548e-01 5.90809762e-01
1.68974888e+00 3.89539927e-01 6.25192165e-01 1.17754690e-01
1.34173536e+00 -1.57513094e+00 -8.44102025e-01 -3.97611469e-01
9.80795443e-01 -2.95649827e-01 1.24654520e+00 -1.32175991e-02
-7.07011104e-01 -5.78551829e-01 -1.20961428e+00 -3.20068061e-01
-7.88220823e-01 5.94920516e-01 8.15575600e-01 2.14587659e-01
-7.21507072e-01 5.31721294e-01 -3.00688297e-01 -4.50912952e-01
3.29563141e-01 9.06105638e-02 -3.03973913e-01 -1.49783820e-01
-1.58191347e+00 1.23104739e+00 5.58751762e-01 -2.19736733e-02
-1.12903225e+00 -9.08336461e-01 -1.15186203e+00 8.85092095e-02
1.01444674e+00 -6.69191301e-01 1.38199663e+00 -4.99498881e-02
-1.18929017e+00 8.66648972e-01 -9.99630466e-02 -4.80876178e-01
5.31133890e-01 -7.19687521e-01 -5.07698417e-01 -1.00154892e-01
3.33355188e-01 8.04086566e-01 1.76314428e-01 -1.26649344e+00
-6.67995453e-01 7.24644810e-02 5.31788707e-01 4.38429862e-01
-8.49438533e-02 -1.59917369e-01 -3.18170696e-01 -4.35237348e-01
-4.26685065e-01 -1.03125799e+00 3.95588093e-02 -7.14643002e-01
-9.02788401e-01 -1.05700076e+00 7.01035619e-01 -3.87539834e-01
1.37866306e+00 -1.83474052e+00 2.90533245e-01 -2.40588814e-01
3.68628651e-01 5.29230654e-01 -2.89481461e-01 4.58312333e-01
-8.08162764e-02 2.99672812e-01 2.76417851e-01 -2.44776040e-01
4.41999048e-01 3.22602659e-01 -5.02080977e-01 -7.37982392e-02
4.13032532e-01 1.44734228e+00 -1.27339780e+00 -9.73270714e-01
3.03529918e-01 1.45258635e-01 -1.56686515e-01 3.20783108e-01
-5.10565400e-01 2.50494480e-01 -7.39656091e-01 7.27611542e-01
-1.20735034e-01 -4.74364400e-01 2.20173493e-01 -4.76963907e-01
6.72607645e-02 1.06432247e+00 -7.92225540e-01 1.60574734e+00
-4.81171936e-01 7.28929996e-01 -6.20734334e-01 -8.17775548e-01
6.04571044e-01 3.99878055e-01 1.54463992e-01 -7.75789559e-01
-2.04395235e-01 2.02688947e-01 -1.73638329e-01 -7.66814113e-01
7.07042038e-01 2.55468279e-01 -4.11064148e-01 8.59677196e-02
4.23745692e-01 -2.85131246e-01 5.13996720e-01 5.90299964e-01
1.13613474e+00 7.44503915e-01 3.25401813e-01 -1.17898762e-01
5.42582989e-01 2.15849802e-01 5.55644870e-01 7.17536807e-01
-1.19313046e-01 7.75044635e-02 4.91956443e-01 1.10378750e-01
-4.81671542e-01 -8.24140251e-01 -1.02690391e-01 1.13808846e+00
2.99658805e-01 -1.01919186e+00 -4.19542938e-01 -1.25541413e+00
1.88367292e-01 1.14371514e+00 -4.66286927e-01 -2.88315088e-01
-6.23823583e-01 -3.44902158e-01 1.07196629e+00 6.11367464e-01
7.75422931e-01 -1.20801830e+00 -3.04354459e-01 4.37767245e-03
-5.21205127e-01 -1.65730047e+00 3.24254297e-02 4.11521465e-01
-6.11103237e-01 -1.28503752e+00 -3.42837185e-01 -7.48801112e-01
3.96096110e-01 7.54178613e-02 1.71947265e+00 -4.12662625e-02
7.61088207e-02 5.59076786e-01 -3.97975862e-01 -3.44894588e-01
-2.75753826e-01 3.56186658e-01 -2.62301803e-01 -1.01091206e+00
7.75061667e-01 -2.24966109e-01 -9.25690010e-02 2.96498328e-01
-4.00370151e-01 1.88460171e-01 6.74507141e-01 6.91838562e-01
5.53292155e-01 1.29332498e-01 8.29321325e-01 -1.25268245e+00
6.54116809e-01 -2.74860054e-01 -3.07829082e-01 8.32625389e-01
-9.46212053e-01 1.20354719e-01 2.96001196e-01 -2.88078427e-01
-1.05552256e+00 -4.82007533e-01 1.60615280e-01 -2.57841259e-01
-6.59277067e-02 6.81383550e-01 -3.38726670e-01 2.63714075e-01
6.80551291e-01 -2.67131776e-01 -6.88667178e-01 -3.52405936e-01
7.71509945e-01 5.02498329e-01 5.39648771e-01 -1.12998402e+00
7.93520868e-01 -2.77368575e-01 8.01737010e-02 -4.80458349e-01
-1.46264029e+00 -2.80477583e-01 -4.77242559e-01 1.35512315e-02
8.73325050e-01 -1.24283874e+00 -5.95589280e-01 2.43793782e-02
-1.04741108e+00 -6.93608403e-01 -5.50537825e-01 4.75631416e-01
-4.84028399e-01 1.27816409e-01 -9.20021296e-01 -5.89087844e-01
-2.02661127e-01 -1.12998855e+00 1.11349273e+00 3.57481033e-01
-4.24701929e-01 -8.94242287e-01 4.68489677e-02 6.39370918e-01
1.74202956e-02 2.53524870e-01 1.21808255e+00 -8.70925903e-01
-6.40401959e-01 -4.60811565e-03 -4.10585612e-01 2.36182228e-01
2.73599220e-03 1.44854924e-02 -1.04300284e+00 4.74069118e-02
-6.28958046e-01 -1.07424951e+00 9.02427197e-01 -7.26255104e-02
8.47749829e-01 -2.48648614e-01 -8.37118387e-01 3.38817030e-01
9.98402476e-01 9.13079754e-02 4.33410585e-01 6.55867338e-01
7.87304223e-01 6.42588735e-01 1.18387794e+00 -3.26320648e-01
8.68940890e-01 8.10482681e-01 1.34355247e-01 -8.98738876e-02
-7.03101933e-01 -9.02873278e-01 4.09616560e-01 7.05422521e-01
-2.24068016e-01 -3.25410128e-01 -8.16198945e-01 6.85744643e-01
-2.02237916e+00 -7.03534186e-01 -2.82309949e-01 1.72223949e+00
1.27031016e+00 8.13340098e-02 -1.90493912e-01 -1.45162195e-01
2.87397653e-01 2.61528850e-01 -5.18464923e-01 3.83000039e-02
-4.45616215e-01 2.30722755e-01 1.67126730e-01 3.56730759e-01
-1.36890614e+00 1.38315952e+00 5.85010338e+00 1.11734498e+00
-5.04908741e-01 -1.79950967e-02 2.81251192e-01 7.12912381e-02
-1.51055336e-01 7.94538334e-02 -1.26877952e+00 -7.26646632e-02
8.94493699e-01 -2.85745680e-01 1.80016905e-01 8.31279933e-01
-5.89816630e-01 9.55997184e-02 -1.72618389e+00 8.08980167e-01
1.57693997e-01 -1.09351850e+00 4.41351607e-02 -8.38817880e-02
8.99098158e-01 5.47147207e-02 1.76955145e-02 1.37467325e+00
7.37487972e-01 -1.31486630e+00 7.34158933e-01 1.90349758e-01
9.54493821e-01 -4.46078360e-01 7.77224123e-01 3.41750920e-01
-1.16263986e+00 1.81412473e-01 -1.92877010e-01 1.45400286e-01
-1.46237180e-01 3.76195312e-01 -8.01876247e-01 9.72767174e-01
4.77791369e-01 1.05889165e+00 -8.06835949e-01 6.23009562e-01
-1.27076089e+00 6.94833517e-01 -2.02356249e-01 -1.05423644e-01
2.33092442e-01 1.65298954e-01 5.20519674e-01 1.43090677e+00
-1.21782839e-01 7.84206539e-02 2.62923300e-01 1.10491002e+00
-4.41261411e-01 -1.67907879e-01 -6.62958682e-01 -3.57644528e-01
7.32527792e-01 1.24871635e+00 -4.56213020e-02 -7.38564730e-01
-4.09165204e-01 4.76982296e-01 7.33909726e-01 7.83473015e-01
-9.43043172e-01 -3.40999782e-01 2.44076192e-01 -2.65742570e-01
1.88305855e-01 -4.77608666e-02 1.70699224e-01 -1.23640800e+00
-2.37228237e-02 -9.78896618e-01 5.63879073e-01 -7.88564384e-01
-1.35363591e+00 6.98657930e-01 4.39433068e-01 -9.75570440e-01
-4.85005140e-01 -4.53128755e-01 -1.22893974e-01 5.81496239e-01
-2.17502999e+00 -1.49634516e+00 -2.27548525e-01 4.11576420e-01
5.18791676e-01 -1.64013401e-01 1.03880548e+00 5.02189100e-01
-7.20466137e-01 7.63682425e-01 -4.69861895e-01 2.77741194e-01
9.82962132e-01 -1.53497803e+00 3.08191776e-01 5.56814611e-01
5.92514277e-01 7.40494251e-01 5.58946490e-01 -8.43177319e-01
-1.13036966e+00 -1.23982823e+00 1.46238196e+00 -9.46800947e-01
8.88661206e-01 -2.66438931e-01 -6.50563419e-01 1.36490035e+00
2.31154799e-01 -1.02007985e-01 5.73183954e-01 7.19966948e-01
-8.09497416e-01 2.35181630e-01 -8.53747547e-01 7.64484406e-01
1.28148949e+00 -7.73682237e-01 -1.08760679e+00 6.82079613e-01
9.80493665e-01 -6.44831240e-01 -1.18786693e+00 8.72472346e-01
3.31271231e-01 -5.65401912e-01 1.23480785e+00 -9.60740387e-01
7.12024152e-01 -1.23165071e-01 -2.02331290e-01 -1.26262712e+00
-1.05939046e-01 -5.05293548e-01 -5.43980777e-01 1.68442547e+00
8.40879679e-01 -3.99411082e-01 5.23739994e-01 6.35398805e-01
-1.32368118e-01 -9.82686460e-01 -6.66174889e-01 -9.73031342e-01
1.96329609e-01 -2.93455541e-01 5.63473880e-01 1.00468600e+00
1.11836299e-01 1.09460211e+00 1.47710638e-02 5.25236130e-03
5.09062588e-01 2.90486366e-01 9.73323345e-01 -1.10466421e+00
-2.75126189e-01 -3.41255188e-01 3.05609088e-02 -1.40838397e+00
6.86138749e-01 -1.05497372e+00 7.90924802e-02 -1.89551866e+00
4.63782400e-01 -9.70067382e-01 -4.29181248e-01 8.12113106e-01
-5.02262652e-01 -4.65913787e-02 8.68145451e-02 3.18722993e-01
-1.21835113e+00 6.39870465e-01 1.19401252e+00 -3.41763616e-01
-7.46781528e-02 -2.21857354e-01 -6.28850460e-01 8.30522180e-01
4.56429154e-01 -4.86540645e-01 -5.86776257e-01 -4.53426003e-01
4.10055548e-01 -9.72149707e-03 5.01667619e-01 -8.22085798e-01
2.25222290e-01 -1.17020093e-01 6.59923479e-02 -7.35626876e-01
4.25053805e-01 -6.57780170e-01 -3.13093454e-01 -9.27985981e-02
-8.16451311e-01 -3.30669940e-01 3.60241294e-01 5.71160913e-01
-3.88493568e-01 -3.60178411e-01 -1.40270427e-01 4.81394380e-02
-9.43581700e-01 -1.58806697e-01 -2.07976205e-03 4.23191607e-01
7.04046726e-01 2.91283756e-01 -1.16194487e+00 -2.09782258e-01
-3.29616189e-01 6.03836238e-01 1.98087558e-01 5.43224156e-01
2.22415611e-01 -1.35462463e+00 -6.16463542e-01 -3.78819138e-01
5.54585636e-01 2.89689302e-01 -3.48830223e-02 7.92650640e-01
1.60990819e-01 1.08457005e+00 3.51445884e-01 -3.69783133e-01
-1.37154961e+00 5.12993574e-01 1.55347034e-01 -1.04052794e+00
-7.09582627e-01 9.94947731e-01 1.86596081e-01 -6.02516115e-01
6.48069799e-01 -6.25200570e-01 -5.53718388e-01 -2.09566876e-01
1.86215654e-01 2.14943081e-01 -7.69392634e-03 -4.85923856e-01
-4.04457599e-01 4.86344099e-01 -2.23501667e-01 7.43449405e-02
1.18277240e+00 2.15055987e-01 1.28013030e-01 5.52501619e-01
8.95653069e-01 2.67876923e-01 -8.95331323e-01 -4.58229661e-01
4.84143764e-01 8.00286829e-02 1.88965425e-02 -1.46775913e+00
-7.03095078e-01 6.99511468e-01 -8.65427703e-02 -1.42155513e-01
5.92432082e-01 3.22827101e-01 8.05681765e-01 8.38119745e-01
5.16968846e-01 -8.33427787e-01 1.14639118e-01 5.32095671e-01
8.23644102e-01 -1.27180195e+00 4.72010851e-01 -6.59015357e-01
-7.99869239e-01 6.07084990e-01 1.01449645e+00 1.02405455e-02
5.16410172e-02 1.90028816e-01 -2.50541568e-01 -5.66742897e-01
-1.01465619e+00 -4.78730917e-01 6.17982745e-01 6.12435937e-01
7.68952727e-01 2.17695877e-01 -1.86890170e-01 8.66957903e-01
-2.14608729e-01 1.85833916e-01 -6.62868842e-02 9.75932777e-01
6.92162886e-02 -1.15866530e+00 1.55556872e-01 2.71026611e-01
-2.12397292e-01 -2.39885360e-01 -7.98556983e-01 1.30086410e+00
8.88380036e-02 1.19126809e+00 -4.58679646e-01 -4.44251657e-01
7.43172169e-01 2.00884581e-01 7.38537371e-01 -8.01555574e-01
-7.48066366e-01 -3.59070688e-01 9.67425525e-01 -5.80947876e-01
-6.38060212e-01 -4.57178086e-01 -1.26873028e+00 2.31218606e-01
-4.45635349e-01 3.92375708e-01 3.47345948e-01 1.05191314e+00
2.79865384e-01 1.11770582e+00 -1.59830138e-01 -1.86353654e-01
-6.65375054e-01 -1.28255594e+00 -1.77930132e-01 4.62109298e-01
3.71939093e-02 -1.08581722e+00 -3.62466663e-01 -2.57817656e-01] | [9.598231315612793, 8.59192943572998] |
79f5f1c1-43cc-494c-a9b3-4a65f3612a90 | efficient-neural-network-based-classification | 2305.07639 | null | https://arxiv.org/abs/2305.07639v1 | https://arxiv.org/pdf/2305.07639v1.pdf | Efficient Neural Network based Classification and Outlier Detection for Image Moderation using Compressed Sensing and Group Testing | Popular social media platforms employ neural network based image moderation engines to classify images uploaded on them as having potentially objectionable content. Such moderation engines must answer a large number of queries with heavy computational cost, even though the actual number of images with objectionable content is usually a tiny fraction. Inspired by recent work on Neural Group Testing, we propose an approach which exploits this fact to reduce the overall computational cost of such engines using the technique of Compressed Sensing (CS). We present the quantitative matrix-pooled neural network (QMPNN), which takes as input $n$ images, and a $m \times n$ binary pooling matrix with $m < n$, whose rows indicate $m$ pools of images i.e. selections of $r$ images out of $n$. The QMPNN efficiently outputs the product of this matrix with the unknown sparse binary vector indicating whether each image is objectionable or not, i.e. it outputs the number of objectionable images in each pool. For suitable matrices, this is decoded using CS decoding algorithms to predict which images were objectionable. The computational cost of running the QMPNN and the CS algorithms is significantly lower than the cost of using a neural network with the same number of parameters separately on each image to classify the images, which we demonstrate via extensive experiments. Our technique is inherently resilient to moderate levels of errors in the prediction from the QMPNN. Furthermore, we present pooled deep outlier detection, which brings CS and group testing techniques to deep outlier detection, to provide for the case when the objectionable images do not belong to a set of pre-defined classes. This technique enables efficient automated moderation of off-topic images shared on topical forums dedicated to sharing images of a certain single class, many of which are currently human-moderated. | ['Ajit Rajwade', 'Sanyam Saxena', 'Sabyasachi Ghosh'] | 2023-05-12 | null | null | null | null | ['outlier-detection'] | ['methodology'] | [ 5.01362324e-01 1.95508882e-01 -7.42042214e-02 -1.95822313e-01
-9.15963531e-01 -3.76179606e-01 1.07222989e-01 1.24715209e-01
-5.98636150e-01 3.28651965e-01 -2.93255687e-01 -2.99289227e-01
-2.02926956e-02 -9.74955976e-01 -1.29994595e+00 -6.79665565e-01
-5.48024833e-01 2.56162852e-01 3.51660401e-02 9.87562835e-02
3.56830239e-01 2.62706667e-01 -2.03260612e+00 6.29157960e-01
6.38059974e-01 1.73217189e+00 2.46029962e-02 8.46970677e-01
2.57503420e-01 9.62209940e-01 -6.71999395e-01 -3.88989478e-01
7.61407137e-01 -2.82974333e-01 -3.11677307e-01 2.84667999e-01
9.91385877e-01 -5.80886543e-01 -6.58505619e-01 1.52273452e+00
4.25615788e-01 1.29316524e-01 3.44934523e-01 -1.50048244e+00
-7.53277481e-01 7.39393353e-01 -3.83948267e-01 4.65007961e-01
2.63521314e-01 3.52354497e-01 1.09660411e+00 -1.16389608e+00
6.83116138e-01 8.98484170e-01 5.58254242e-01 4.70942035e-02
-1.13797331e+00 -8.67495596e-01 3.58318281e-03 3.56757849e-01
-1.62557197e+00 -6.98111415e-01 5.73678970e-01 -3.40022564e-01
7.46162236e-01 4.86724377e-01 4.97296453e-01 6.82465792e-01
-4.87038530e-02 7.12557495e-01 9.15096402e-01 -1.77396327e-01
5.71568012e-01 2.78923839e-01 -2.18200281e-01 8.28334928e-01
3.26767683e-01 -2.80929834e-01 -8.24203789e-01 -3.30258161e-01
4.69507039e-01 1.43421695e-01 -3.24520826e-01 -1.32950500e-01
-1.17006660e+00 8.62693667e-01 5.09586573e-01 6.85548484e-02
-5.08548141e-01 2.87065625e-01 2.72101790e-01 6.38447702e-01
5.83363235e-01 2.99186528e-01 -3.81251007e-01 3.08836132e-01
-1.40129161e+00 2.40495652e-01 7.85534382e-01 6.66552424e-01
1.10826659e+00 -7.43615851e-02 -1.64417848e-01 6.60055578e-01
9.69768837e-02 5.62465131e-01 5.15129685e-01 -1.36076820e+00
5.08413076e-01 7.24982917e-01 2.19337344e-02 -1.61794579e+00
7.15927109e-02 -4.08021718e-01 -1.03322566e+00 3.01108092e-01
2.32912660e-01 1.16927162e-01 -7.50849783e-01 1.73319852e+00
1.76675618e-01 2.74361700e-01 -2.83061832e-01 8.36566925e-01
6.04830325e-01 5.80101252e-01 -2.64841020e-01 -9.11257714e-02
1.28791952e+00 -5.64869404e-01 -2.20434636e-01 -1.65237278e-01
5.11533558e-01 -5.75618565e-01 1.12169397e+00 6.34048164e-01
-1.32807314e+00 -4.84796137e-01 -1.09485674e+00 2.68047154e-01
-4.57595050e-01 1.23777084e-01 4.90244478e-01 7.14435399e-01
-1.39689922e+00 7.09589660e-01 -4.36411738e-01 5.63565530e-02
9.05340016e-01 7.34596312e-01 -4.55865920e-01 -1.89354837e-01
-1.01899159e+00 2.62061626e-01 2.11901709e-01 3.25995847e-03
-1.03066540e+00 -5.31609178e-01 -7.34741211e-01 2.45653927e-01
5.67470491e-01 -3.64353001e-01 7.93551087e-01 -1.47952938e+00
-9.28361654e-01 1.03770185e+00 -2.28938952e-01 -7.48040020e-01
5.93460619e-01 1.89469472e-01 -3.77275020e-01 5.53671598e-01
3.99528950e-01 9.21807349e-01 1.37200701e+00 -9.46339130e-01
-6.59213662e-01 -3.99827093e-01 -4.27746549e-02 2.31295060e-02
-6.47559345e-01 2.96942447e-03 -5.85812867e-01 -6.93604648e-01
5.55189848e-01 -9.31631982e-01 -2.76358277e-01 3.03426713e-01
-7.48930573e-01 -1.60998538e-01 6.81775689e-01 -5.80650270e-01
1.11206007e+00 -2.34445643e+00 -6.64867908e-02 7.05886900e-01
5.88626206e-01 -8.41745436e-02 -3.72523606e-01 9.62738544e-02
-8.60708058e-02 3.06377560e-01 -1.94745064e-01 -4.50901687e-01
5.96568361e-02 -1.03244387e-01 -3.91272396e-01 7.27528870e-01
2.99399141e-02 6.39614224e-01 -5.73350668e-01 -3.27327639e-01
-3.52375247e-02 5.27567789e-02 -7.45389640e-01 1.52291372e-01
-3.82698298e-01 4.30825949e-02 -4.40850370e-02 9.10241723e-01
9.40366447e-01 -6.40931427e-01 -7.53699765e-02 -1.04734331e-01
1.55027881e-01 4.80370261e-02 -1.53617442e+00 1.21582651e+00
4.27226461e-02 6.49986327e-01 2.05270231e-01 -1.12217462e+00
3.82573694e-01 7.63557106e-02 6.35601819e-01 -7.52583146e-01
2.81988308e-02 4.22675282e-01 -1.24394141e-01 -5.40501416e-01
6.83126271e-01 2.75603354e-01 -5.60090952e-02 7.27508307e-01
-6.55297330e-03 3.45270991e-01 4.06444967e-01 6.35722339e-01
1.54064882e+00 -6.53410494e-01 -1.54269919e-01 3.98661345e-02
3.97986770e-01 -1.97251901e-01 3.77208173e-01 1.40633118e+00
-3.58709663e-01 7.01317191e-01 7.33324647e-01 -3.61882687e-01
-1.21647930e+00 -8.74286532e-01 -5.18723093e-02 1.20400774e+00
2.40080673e-02 -3.02108318e-01 -7.23057389e-01 -6.84319317e-01
6.04058206e-02 1.33244887e-01 -6.12900972e-01 -3.17201205e-02
-3.55243593e-01 -6.60200298e-01 3.44078511e-01 -1.64537311e-01
6.64109409e-01 -1.35759413e+00 -4.11967963e-01 -7.60663822e-02
-2.69758224e-01 -1.11900902e+00 -3.01170141e-01 5.70209436e-02
-5.46543360e-01 -1.26816189e+00 -5.52028179e-01 -6.27638400e-01
9.13047493e-01 4.24958646e-01 1.03310192e+00 4.49526012e-01
-8.29849541e-02 3.02580118e-01 -1.70198068e-01 -1.29898056e-01
-3.87474537e-01 1.42452391e-02 1.90825775e-01 4.57889378e-01
3.91497523e-01 -5.78340948e-01 -1.07033849e+00 3.41080517e-01
-1.35897040e+00 -2.79892921e-01 5.88508010e-01 4.78263706e-01
7.28148758e-01 4.93889004e-01 3.18066686e-01 -8.04978848e-01
3.64006579e-01 -9.14826274e-01 -5.61349034e-01 -2.52397835e-01
-4.67200398e-01 -3.47379416e-01 5.82828343e-01 -5.60375214e-01
-9.53908041e-02 6.92976965e-03 8.16462114e-02 -6.11528456e-01
4.71792631e-02 5.90302944e-01 -1.29152372e-01 -2.59742349e-01
7.91786909e-01 2.90333033e-01 1.87625378e-01 -7.60314707e-03
1.74939126e-01 6.72981799e-01 5.14525652e-01 1.17491856e-02
7.41877615e-01 7.34445453e-01 -2.91902989e-01 -6.57026768e-01
-6.50097191e-01 -4.02456045e-01 -3.84794101e-02 -2.99741596e-01
4.67243284e-01 -1.17368436e+00 -9.00637507e-01 4.91280079e-01
-8.15093517e-01 -4.05190527e-01 -4.95464385e-01 2.13606089e-01
-4.30660546e-01 4.90748256e-01 -5.99509895e-01 -6.30895138e-01
-2.36696288e-01 -1.32921183e+00 8.13155234e-01 -1.86262384e-01
-1.25476047e-01 -1.74682215e-01 -5.06719708e-01 4.14532691e-01
4.62461293e-01 -6.36302233e-02 6.94628417e-01 -8.22154522e-01
-1.24328220e+00 -6.87783122e-01 -3.84544492e-01 7.19289124e-01
-5.43973923e-01 -2.98899561e-01 -8.40379596e-01 -5.23602664e-01
1.51205078e-01 -3.71239871e-01 1.07632673e+00 5.91854334e-01
1.83879709e+00 -8.31809819e-01 -1.08232930e-01 4.61455584e-01
1.36592960e+00 -1.97469532e-01 8.88552010e-01 3.61993045e-01
4.08594698e-01 4.47451711e-01 2.76610851e-01 6.56467974e-01
1.81330308e-01 3.78665328e-01 9.24912035e-01 7.77435303e-02
1.11357711e-01 7.18340799e-02 6.90490901e-01 4.07504529e-01
3.19220901e-01 -5.16357720e-01 -5.47994435e-01 6.06094539e-01
-1.72634995e+00 -1.03108478e+00 -5.51737845e-02 2.39869237e+00
6.14888251e-01 2.39852548e-01 -5.68514131e-02 2.24522993e-01
7.28313446e-01 3.45084786e-01 -7.98629701e-01 6.30175918e-02
-4.44182694e-01 3.22603732e-01 7.76166618e-01 2.47456580e-01
-1.21724379e+00 5.74438274e-01 5.65232515e+00 1.06540322e+00
-9.63058770e-01 3.36572707e-01 1.18615663e+00 -4.47254658e-01
-3.31911027e-01 -1.38636187e-01 -5.93886614e-01 8.94019544e-01
1.19002616e+00 3.84039462e-01 6.66556776e-01 8.89034867e-01
2.04432696e-01 -4.45261568e-01 -9.90479171e-01 1.13052475e+00
3.78337443e-01 -1.62258601e+00 1.48893222e-01 2.77006269e-01
8.88801455e-01 3.95296127e-01 6.66642189e-01 7.13878721e-02
1.77135140e-01 -1.03387892e+00 8.59025598e-01 3.19808811e-01
9.07226741e-01 -5.46101809e-01 5.99848986e-01 3.52127790e-01
-5.22046387e-01 -4.36599076e-01 -6.23726845e-01 4.41112928e-02
-2.54339516e-01 1.10824049e+00 -3.75684917e-01 -1.62839055e-01
1.12835598e+00 4.03829992e-01 -7.46191740e-01 8.19262683e-01
2.25999221e-01 5.66601694e-01 -6.00869894e-01 1.30266085e-01
7.10885972e-03 8.20269156e-03 7.09576309e-01 7.14515388e-01
7.39634275e-01 -1.73983037e-01 1.25243813e-02 8.24927092e-01
-6.78198099e-01 1.86289594e-01 -6.10879838e-01 5.45629971e-02
5.95118284e-01 1.17571843e+00 -8.45676899e-01 -7.70505011e-01
-3.16563994e-01 1.27873492e+00 2.43379921e-01 2.65827149e-01
-6.61066115e-01 -2.89910525e-01 5.12139201e-01 3.99405152e-01
5.86058021e-01 4.12574783e-02 -1.11248322e-01 -1.33763564e+00
3.56901169e-01 -1.29506207e+00 4.75161314e-01 -7.97564447e-01
-1.27335083e+00 5.42783260e-01 -4.70674664e-01 -1.22824466e+00
1.64273947e-01 -6.00395322e-01 -3.26920390e-01 5.42960167e-01
-1.22121251e+00 -5.96778154e-01 -3.26346010e-01 7.69392252e-01
1.98712990e-01 -4.12482947e-01 4.74621534e-01 5.54207742e-01
-6.09850943e-01 7.40365326e-01 1.96702722e-02 5.21828383e-02
6.10466719e-01 -9.35383081e-01 5.00718467e-02 9.62130725e-01
2.53573149e-01 4.91614640e-01 7.17549503e-01 -5.83444417e-01
-1.27267122e+00 -1.28754985e+00 9.53154385e-01 -3.85443956e-01
6.34734631e-01 -4.21732992e-01 -6.88155711e-01 6.61757946e-01
-5.01493178e-03 3.43518049e-01 7.93055296e-01 -2.49461830e-01
-3.46487731e-01 -9.25215557e-02 -1.17346954e+00 5.52715659e-01
8.55309367e-01 -7.31950402e-01 1.31194025e-01 8.17623734e-01
6.91575646e-01 -2.95801103e-01 -5.59640229e-01 1.96746379e-01
2.95886725e-01 -1.31465709e+00 1.06363130e+00 -2.76618302e-01
7.61434734e-01 -3.49994838e-01 -7.00198650e-01 -6.84890985e-01
-2.38550790e-02 -5.01538515e-01 -3.88646603e-01 6.58149779e-01
4.80026960e-01 -5.54375768e-01 1.29583597e+00 5.52878380e-01
9.82997715e-02 -6.39537096e-01 -1.17478383e+00 -4.22236055e-01
-4.25970018e-01 -8.27546954e-01 4.65630174e-01 8.73791158e-01
-4.52269167e-01 -2.69491494e-01 -3.71360749e-01 4.45040613e-01
6.60869181e-01 -1.77195355e-01 8.03084850e-01 -9.53541696e-01
-4.56866860e-01 -2.27709010e-01 -5.40163040e-01 -9.35536027e-01
4.73299921e-02 -9.63327229e-01 -8.60949308e-02 -7.78221190e-01
3.82447511e-01 -5.91865718e-01 -5.12682140e-01 3.15995723e-01
9.58825797e-02 9.62191641e-01 1.27117649e-01 6.18776858e-01
-1.03295505e+00 8.36389512e-02 6.66614175e-01 -3.21070224e-01
-1.54667392e-01 -1.83977671e-02 -7.59987891e-01 7.74311543e-01
5.91549277e-01 -7.18979776e-01 -4.87889275e-02 -2.46942222e-01
8.95928025e-01 -1.20336466e-01 8.72976959e-01 -1.29606807e+00
3.93295884e-01 3.08748752e-01 5.52164853e-01 -7.03206122e-01
1.41053289e-01 -8.57218981e-01 5.39708920e-02 4.86314893e-01
-4.92980719e-01 1.07780337e-01 -2.38724336e-01 8.64891231e-01
-2.15222672e-01 -2.02281713e-01 5.73659658e-01 -4.22430933e-01
-6.22877657e-01 7.17104554e-01 -5.79675853e-01 -2.05108315e-01
9.21738982e-01 -3.59078556e-01 -1.02451913e-01 -6.96683943e-01
-9.45025265e-01 -3.32644433e-02 3.98585320e-01 -6.67495504e-02
7.66140878e-01 -1.24915040e+00 -6.35164618e-01 5.04463494e-01
9.37329680e-02 -1.30737675e-02 6.18174791e-01 9.01833415e-01
-4.48993742e-01 -9.89852995e-02 1.53728306e-01 -8.14146399e-01
-1.03943384e+00 5.33434629e-01 3.00952971e-01 -1.04045704e-01
-3.73971522e-01 1.07295704e+00 -1.31387427e-01 -3.29971164e-01
3.37188393e-01 -9.16170776e-02 2.16531232e-01 1.47099972e-01
6.24349952e-01 2.55141854e-01 2.37995833e-01 -7.18743920e-01
-1.00097224e-01 3.45902927e-02 -1.08027846e-01 -1.75566435e-01
1.32978809e+00 -1.74355194e-01 -5.84258080e-01 1.87268689e-01
1.51418924e+00 1.31027728e-01 -1.21503854e+00 -3.91759008e-01
-2.91917682e-01 -6.23985231e-01 1.03795491e-01 -4.81437355e-01
-1.33464313e+00 4.65294898e-01 8.57183695e-01 5.26047945e-01
1.21577525e+00 1.42342791e-01 8.79508018e-01 4.63139683e-01
2.52811998e-01 -1.19777203e+00 4.22048002e-01 1.21886559e-01
7.29833841e-01 -1.52719223e+00 4.20810748e-03 -1.60433620e-01
-1.25658602e-01 7.42997825e-01 4.02589917e-01 -4.32221144e-01
6.99479520e-01 -8.67992416e-02 -2.58249551e-01 -4.77435589e-01
-5.43819547e-01 1.64905444e-01 5.57972938e-02 1.12551533e-01
-2.51516908e-01 7.34769329e-02 1.06032804e-01 3.34223509e-01
-1.98617443e-01 -3.62222344e-02 5.15592277e-01 6.22217000e-01
-6.19809270e-01 -6.66128993e-01 -5.14110148e-01 1.07183444e+00
-7.60409296e-01 -3.96786779e-01 -2.79463511e-02 3.12612534e-01
5.06425321e-01 8.77674818e-01 4.82835144e-01 -5.03889859e-01
-2.31520217e-02 -1.52682438e-01 -5.09225624e-03 -4.70900953e-01
-5.28378487e-01 -2.47500464e-01 -2.17757389e-01 -1.03182530e+00
-2.81317234e-01 -6.79811001e-01 -7.58422613e-01 -5.76318443e-01
-1.41056329e-01 -1.46152109e-01 6.55987620e-01 6.75056636e-01
6.11977458e-01 -8.51118118e-02 7.85099924e-01 -9.15162086e-01
-5.59778452e-01 -7.17772722e-01 -6.67952418e-01 6.68179452e-01
4.30642843e-01 -1.13756008e-01 -8.47645104e-01 2.89799478e-02] | [11.49918270111084, 0.8434935808181763] |
850722ec-18b4-4421-bc8f-dc19fe1e57d3 | local-low-rank-approximation-with-superpixel | null | null | https://ieeexplore.ieee.org/document/9861684 | https://ieeexplore.ieee.org/document/9861684 | Local Low-Rank Approximation With Superpixel-Guided Locality Preserving Graph for Hyperspectral Image Classification | Given the detrimental effect of spectral variations in a hyperspectral image (HSI), this article investigates to recover its discriminative representation to improve the classification performance. We propose a new method, namely local low-rank approximation with superpixel-guided locality preserving graph (LLRA-SLPG), which can reduce the spectral variations and preserve the local manifold structure of an HSI. Specifically, the LLRA-SLPG method first clusters pixels of an HSI into several groups (i.e., superpixels). By taking advantage of the local manifold structure, a Laplacian graph is constructed from the superpixels to ensure that a typical pixel should be similar to its neighbors within the same superpixel. The LLRA-SLPG model can increase the compactness of pixels belonging to the same class by reducing spectral variations while promoting local consistency via the Laplacian graph. The objective function of the LLRA-SLPG model can be solved efficiently in an iterative manner. Experimental results on four benchmark datasets validate the superiority of the LLRA-SLPG model over state-of-the-art methods, particularly in cases where only extremely few training pixels are available. | ['and Weijia Zhang', 'Yuheng Jia', 'Yu Zhang', 'Shujun Yang'] | 2022-08-18 | null | null | null | journal-2022-8 | ['superpixels'] | ['computer-vision'] | [ 4.62737203e-01 -9.59105268e-02 -1.23621598e-01 6.65722200e-06
-6.32414639e-01 -3.21860224e-01 1.83883533e-01 -1.39035851e-01
8.28730837e-02 6.48934186e-01 1.01168536e-01 1.55966818e-01
-3.68004292e-01 -8.18887889e-01 -5.13033152e-01 -1.33417511e+00
1.34187117e-01 -3.11418205e-01 3.04234087e-01 2.53646821e-01
2.29915783e-01 7.21693635e-01 -1.36822486e+00 1.53528601e-01
1.43939924e+00 8.65559042e-01 6.85532033e-01 2.18261611e-02
1.46145940e-01 6.75366998e-01 4.88045774e-02 4.16376382e-01
5.45868993e-01 -6.01487041e-01 -6.24250889e-01 7.20600069e-01
6.36120379e-01 -4.95163277e-02 -4.57421541e-01 1.62765396e+00
1.83996320e-01 5.32545209e-01 5.16557157e-01 -9.34325755e-01
-8.26179028e-01 2.94640332e-01 -1.16935349e+00 -3.35126631e-02
-1.58770517e-01 -1.51406089e-02 1.01526606e+00 -1.00279760e+00
4.67229396e-01 1.09393024e+00 3.68629098e-01 -2.10567325e-01
-1.46223533e+00 -4.99036998e-01 3.48676115e-01 3.26297224e-01
-1.95088959e+00 -3.86796221e-02 9.59695756e-01 -3.53091538e-01
1.98620543e-01 4.68978494e-01 4.52992260e-01 5.79912262e-03
-7.22279027e-02 6.22742593e-01 1.31275535e+00 -2.64190763e-01
1.50436014e-01 -1.76489264e-01 2.36505866e-01 9.03836191e-01
3.97766978e-01 -3.04278910e-01 -5.10131657e-01 -1.34248957e-01
7.98722148e-01 2.23976359e-01 -7.95681477e-01 -5.53301275e-01
-1.07686985e+00 6.48559928e-01 9.97849286e-01 4.99749154e-01
-5.82284093e-01 -3.05847198e-01 -2.09378451e-02 -9.84206796e-02
5.97680867e-01 1.18088461e-01 1.06159873e-01 8.43739629e-01
-1.03017604e+00 -3.29113543e-01 2.20754474e-01 6.56796157e-01
1.48200142e+00 -1.33014426e-01 -2.46752739e-01 9.63727891e-01
3.39413941e-01 5.30557573e-01 2.19842941e-01 -9.70646977e-01
4.09169376e-01 9.11376059e-01 -7.86252022e-02 -1.47347605e+00
-1.32381886e-01 -6.83761477e-01 -1.29582334e+00 1.41794384e-02
1.31775975e-01 2.29468241e-01 -6.93772137e-01 1.43573844e+00
4.38954979e-01 6.10853553e-01 -1.86032131e-02 1.08562422e+00
5.90816915e-01 1.08459210e+00 -2.11233087e-02 -6.02722943e-01
9.10232902e-01 -1.04252124e+00 -4.77116674e-01 -3.39434445e-01
6.12127244e-01 -5.53558171e-01 9.20018971e-01 8.26178119e-02
-6.16751313e-01 -5.72032928e-01 -8.87254238e-01 8.17371085e-02
8.97850655e-03 6.34416044e-01 2.97634959e-01 1.64934814e-01
-9.14684892e-01 6.42742872e-01 -6.86129689e-01 -3.14762115e-01
4.10775572e-01 9.35947299e-02 -4.68506843e-01 -5.59433758e-01
-7.97355294e-01 2.50483155e-01 6.58508539e-01 3.59917015e-01
-5.81342578e-01 -7.92978287e-01 -7.06025004e-01 5.75671308e-02
4.61567700e-01 -2.32952088e-01 2.40864784e-01 -1.03275108e+00
-1.07714379e+00 6.60266995e-01 -4.81382102e-01 1.69715285e-01
6.20125569e-02 2.83251584e-01 -3.57441902e-01 5.60611844e-01
3.51036489e-01 2.77285665e-01 8.60664248e-01 -1.58254361e+00
-6.67710423e-01 -5.79462409e-01 -4.18686360e-01 4.33969706e-01
-5.42181551e-01 -4.35745746e-01 -5.83132029e-01 -5.37092030e-01
8.14373016e-01 -1.08513141e+00 -2.42354557e-01 9.43006203e-02
-6.58451498e-01 -9.49896947e-02 1.18967712e+00 -9.33169305e-01
1.31292272e+00 -2.45520210e+00 2.27780566e-01 6.18898451e-01
1.21284798e-01 5.08933365e-01 -3.87658328e-01 3.24045300e-01
-3.26196939e-01 -6.18209951e-02 -6.30009353e-01 1.68481529e-01
-4.05633718e-01 5.31421341e-02 1.39009058e-02 8.15337777e-01
-5.50556108e-02 6.80358708e-01 -1.04443657e+00 -4.89697874e-01
3.33829910e-01 4.72455621e-01 -1.13498539e-01 1.28536031e-01
1.70513675e-01 5.43417335e-01 -7.14355588e-01 6.14276767e-01
1.08510172e+00 -4.24230248e-01 4.19383347e-02 -7.89749146e-01
-4.52748984e-01 -1.46248206e-01 -1.29748273e+00 1.59492671e+00
-6.00622967e-02 3.44105035e-01 3.78438652e-01 -9.88640249e-01
8.62936616e-01 -1.85100228e-01 5.64995468e-01 -3.26405108e-01
-4.19157803e-01 2.89535880e-01 -4.44523484e-01 -2.84389555e-01
1.39344439e-01 -9.43712518e-03 5.27705431e-01 2.02371612e-01
-4.71045196e-01 1.36043578e-01 3.02051932e-01 2.24411011e-01
5.77316821e-01 -4.82312925e-02 3.35783601e-01 -6.17931128e-01
9.46191847e-01 -2.29840904e-01 7.80960739e-01 4.91838843e-01
-1.57834098e-01 5.67551732e-01 8.36565346e-02 -2.39527933e-02
-6.18097842e-01 -8.78499091e-01 -2.51044363e-01 8.44541609e-01
5.36352813e-01 -1.68743789e-01 -6.25335693e-01 -4.95116949e-01
-3.29339728e-02 6.12168491e-01 -3.54882270e-01 -2.39293993e-01
-2.22639605e-01 -1.09583652e+00 -6.11335076e-02 1.17401056e-01
1.08298123e+00 -5.91497540e-01 -1.09044157e-01 3.91716734e-02
-4.60352212e-01 -9.15999472e-01 -6.06032252e-01 -1.34423122e-01
-8.63881528e-01 -9.04651403e-01 -6.67804480e-01 -9.46324050e-01
1.07052732e+00 1.18222022e+00 4.54913348e-01 3.04114651e-02
-1.24544129e-01 7.33316541e-02 -4.89567667e-01 3.46250713e-01
-8.62943679e-02 -1.49260625e-01 -4.20184374e-01 6.68574214e-01
-4.13431413e-02 -5.56270599e-01 -8.11269641e-01 4.24587786e-01
-1.15884447e+00 2.47393787e-01 7.76594639e-01 8.44935775e-01
1.17645180e+00 7.15824783e-01 -6.71594357e-03 -8.55268002e-01
7.19561381e-03 -4.37421471e-01 -6.46668673e-01 5.96787930e-01
-6.44356847e-01 -1.74688801e-01 5.66917598e-01 -8.72976333e-02
-1.10832250e+00 5.62254131e-01 5.64875126e-01 -4.48482841e-01
7.43122622e-02 8.44656587e-01 -5.28324544e-01 -6.57649219e-01
3.65557373e-01 6.78259492e-01 6.27214760e-02 -4.09279168e-01
3.26383919e-01 5.73571622e-01 4.60789263e-01 -6.50374293e-02
1.11596501e+00 7.81930864e-01 4.37836289e-01 -1.35531342e+00
-1.02469742e+00 -8.33117604e-01 -8.28929007e-01 -2.28995949e-01
7.39353180e-01 -1.12138987e+00 -2.01689288e-01 5.24512112e-01
-5.70840240e-01 -2.02246889e-01 -1.42799497e-01 4.96478438e-01
-1.68728501e-01 1.05991054e+00 -4.06252563e-01 -6.29078150e-01
-1.74759820e-01 -8.68376136e-01 9.33095694e-01 4.78610218e-01
5.29138625e-01 -9.53969240e-01 -2.47414291e-01 4.08782899e-01
4.87427860e-02 3.67263615e-01 9.24412429e-01 -1.21459179e-02
-8.39611471e-01 -2.30339579e-02 -5.96846819e-01 5.29222906e-01
5.37787080e-01 -5.17037213e-02 -8.47363591e-01 -6.00330591e-01
-3.14190350e-02 7.81847835e-02 1.05659854e+00 4.84128684e-01
1.45117390e+00 -4.13310140e-01 -3.38334948e-01 7.90498912e-01
1.74816287e+00 -1.95348948e-01 6.14171267e-01 2.19034523e-01
1.22582746e+00 6.66364253e-01 8.24947774e-01 3.59511763e-01
1.36006385e-01 3.71162534e-01 5.81490099e-01 -4.17698115e-01
-2.43581638e-01 -2.22133070e-01 3.47317398e-01 8.59021962e-01
-8.87756050e-02 4.55043558e-03 -5.62213361e-01 5.00528097e-01
-1.95333385e+00 -8.78321469e-01 -6.91132963e-01 2.37381673e+00
5.57889521e-01 -7.07596481e-01 -3.24648291e-01 7.30405152e-02
1.07439506e+00 6.50897324e-01 -6.53029323e-01 4.78533566e-01
-5.21619320e-01 -2.94080466e-01 7.26765454e-01 4.51962233e-01
-1.22908962e+00 9.79941428e-01 4.56381941e+00 1.10228300e+00
-1.00162029e+00 -1.01119816e-01 7.58106351e-01 3.72711003e-01
-1.98610112e-01 1.83401451e-01 -5.15865386e-01 3.78589988e-01
3.31419021e-01 -2.45800331e-01 7.25688696e-01 4.38732862e-01
5.85727453e-01 -2.80384481e-01 -3.60539079e-01 1.07069016e+00
2.45548218e-01 -1.13250756e+00 3.16843599e-01 8.87090191e-02
1.26293683e+00 -5.41452877e-02 5.22513837e-02 -4.09424335e-01
-1.15087479e-01 -6.28475606e-01 2.21976280e-01 6.13207042e-01
7.94600129e-01 -9.01202440e-01 4.46727931e-01 4.09725308e-01
-1.66584039e+00 -5.88816293e-02 -8.81970346e-01 4.74684626e-01
-1.78938866e-01 1.13065827e+00 -2.53664225e-01 1.07458162e+00
6.60649121e-01 1.29260564e+00 -7.26690471e-01 1.06197536e+00
-4.97054011e-01 6.40306115e-01 -1.10791132e-01 5.48430085e-01
5.33636689e-01 -1.22092354e+00 7.14566469e-01 8.96802783e-01
4.79462206e-01 4.25278306e-01 5.87051809e-01 1.04618430e+00
5.37851006e-02 4.88245726e-01 -4.34435666e-01 -4.59249914e-02
3.31660956e-01 1.56851995e+00 -6.66331589e-01 -4.72555161e-02
-3.96961600e-01 1.15609598e+00 3.10186148e-01 6.71187699e-01
-3.18974495e-01 -3.16395491e-01 4.01598573e-01 1.13785975e-01
4.16109741e-01 -1.98230311e-01 -8.52337480e-02 -1.04386163e+00
1.32407725e-01 -6.91359401e-01 5.05025148e-01 -1.02747953e+00
-1.23890769e+00 2.81547815e-01 -3.40766579e-01 -1.49070191e+00
5.28184772e-01 -2.57328004e-01 -6.18831635e-01 9.63335097e-01
-1.93354809e+00 -1.40866947e+00 -8.31448197e-01 7.89481044e-01
2.00053424e-01 2.33586296e-01 5.33234179e-01 3.99892917e-03
-8.74145925e-01 3.94292623e-02 6.43495083e-01 -9.92636830e-02
5.44490159e-01 -9.59067583e-01 -5.34293950e-01 1.32729471e+00
-3.29115316e-02 4.73939866e-01 2.61888862e-01 -7.18982637e-01
-1.32439601e+00 -1.78248882e+00 5.93993485e-01 5.09297132e-01
7.87443340e-01 2.67791659e-01 -1.33349526e+00 3.91813397e-01
-7.96730444e-02 2.51074374e-01 7.38533854e-01 -5.27526796e-01
-2.35099271e-01 -4.87715662e-01 -9.71873581e-01 5.21869600e-01
8.68671000e-01 -7.41489410e-01 -4.99403104e-02 8.67572665e-01
3.17964882e-01 7.44962618e-02 -8.98308158e-01 3.94841164e-01
-8.49640220e-02 -9.17368293e-01 8.73246610e-01 9.40184519e-02
2.03405082e-01 -9.67659771e-01 -1.51989251e-01 -1.54933882e+00
-9.02432024e-01 -4.16016847e-01 3.00325006e-01 1.31666958e+00
1.14437073e-01 -6.46858335e-01 5.04134834e-01 4.01135534e-01
-2.02867821e-01 -2.03718707e-01 -8.90686512e-01 -9.53063309e-01
-3.34419489e-01 2.43018582e-01 2.90604562e-01 1.14091575e+00
-2.42698207e-01 5.04132360e-02 -3.73762518e-01 9.92730856e-01
1.08411753e+00 5.88125050e-01 3.71857941e-01 -1.21100664e+00
-6.90641478e-02 -2.43436709e-01 -4.45022196e-01 -1.01356399e+00
3.31350505e-01 -1.26870990e+00 1.77946150e-01 -1.72750974e+00
6.61392272e-01 -1.68646336e-01 -4.67892438e-01 5.83217084e-01
-6.37080908e-01 4.33359355e-01 7.14950413e-02 6.75420821e-01
-4.57273543e-01 7.28302658e-01 1.52718937e+00 -3.22056413e-01
-2.92982399e-01 -1.95848957e-01 -6.97697878e-01 5.28327048e-01
6.49224818e-01 -2.16474339e-01 -4.39821035e-01 -1.94456592e-01
-2.26715147e-01 -3.39584887e-01 4.42466527e-01 -1.03484046e+00
2.38227487e-01 -3.86723965e-01 3.54888171e-01 -5.57943940e-01
-5.75834550e-02 -8.50139678e-01 3.36315781e-01 5.27893245e-01
-1.43527493e-01 -7.23223269e-01 -1.24384984e-02 8.85174870e-01
-3.71817559e-01 -1.60930499e-01 1.10577822e+00 8.21138993e-02
-1.01538885e+00 5.66799700e-01 -1.43587261e-01 -3.68592083e-01
1.24731290e+00 -1.26898006e-01 -3.81706178e-01 -1.04885198e-01
-4.11777228e-01 2.45084420e-01 6.55840993e-01 -6.56462014e-02
6.62191510e-01 -1.43292320e+00 -8.33258152e-01 2.45932713e-01
4.27463800e-01 1.11655459e-01 5.27788401e-01 1.07652736e+00
-5.68801105e-01 3.04988652e-01 -1.55778289e-01 -6.93281293e-01
-1.56064820e+00 4.54126477e-01 3.20890456e-01 -1.40426472e-01
-7.59822786e-01 7.56125212e-01 6.71123445e-01 -1.82410911e-01
-2.42747307e-01 -6.36308640e-02 -2.44049400e-01 -9.47471783e-02
5.39224386e-01 6.40617490e-01 -1.34537533e-01 -1.09485829e+00
-3.37707490e-01 8.45311642e-01 1.84539109e-01 3.51620167e-01
1.43639374e+00 -3.91867220e-01 -8.37426424e-01 2.78237164e-01
1.26411021e+00 5.19197620e-02 -1.53213680e+00 -4.41516221e-01
-3.04386407e-01 -9.66904223e-01 7.14760900e-01 -3.47177178e-01
-1.49338436e+00 5.68715394e-01 5.97189963e-01 8.88609588e-02
1.55869246e+00 -1.74096718e-01 4.02120650e-01 2.36151859e-01
3.35771054e-01 -1.02528358e+00 9.04283766e-03 1.79785863e-01
8.55183959e-01 -9.97185290e-01 4.47957218e-01 -1.05796635e+00
-6.94200993e-01 9.70792592e-01 2.74150014e-01 -1.46657854e-01
5.78960538e-01 -5.69476187e-01 -2.33798787e-01 -9.95827392e-02
4.38644774e-02 -3.33872885e-01 6.60378277e-01 4.99149591e-01
1.33922383e-01 2.52853811e-01 -2.92711943e-01 -8.64507928e-02
4.48957443e-01 -3.86499286e-01 3.10335159e-01 6.58816874e-01
-6.91094279e-01 -7.43601561e-01 -4.62284714e-01 5.53013027e-01
1.38357580e-01 -2.39823349e-02 -5.32691777e-01 4.05715168e-01
-2.63230246e-03 1.13883090e+00 -2.72582173e-01 -3.37953031e-01
8.75058472e-02 -1.74415559e-01 5.45742549e-02 -7.11399317e-01
-7.07619777e-03 4.71175373e-01 -4.16268975e-01 -7.71545172e-01
-7.24917889e-01 -8.22961807e-01 -1.32180572e+00 -4.77319323e-02
-4.67371196e-01 1.54924139e-01 2.01718211e-01 7.62073100e-01
4.00253594e-01 2.47458085e-01 1.22981346e+00 -7.38096297e-01
-3.57077092e-01 -6.49211466e-01 -1.53238499e+00 4.51813012e-01
1.69261977e-01 -2.91675001e-01 -7.23961651e-01 5.45504764e-02] | [10.142478942871094, -1.8961827754974365] |
0d51b08a-4528-410f-af47-6ad3885584af | leaf-only-sam-a-segment-anything-pipeline-for | 2305.09418 | null | https://arxiv.org/abs/2305.09418v2 | https://arxiv.org/pdf/2305.09418v2.pdf | Leaf Only SAM: A Segment Anything Pipeline for Zero-Shot Automated Leaf Segmentation | Segment Anything Model (SAM) is a new foundation model that can be used as a zero-shot object segmentation method with the use of either guide prompts such as bounding boxes, polygons, or points. Alternatively, additional post processing steps can be used to identify objects of interest after segmenting everything in an image. Here we present a method using segment anything together with a series of post processing steps to segment potato leaves, called Leaf Only SAM. The advantage of this proposed method is that it does not require any training data to produce its results so has many applications across the field of plant phenotyping where there is limited high quality annotated data available. We compare the performance of Leaf Only SAM to a Mask R-CNN model which has been fine-tuned on our small novel potato leaf dataset. On the evaluation dataset, Leaf Only SAM finds an average recall of 63.2 and an average precision of 60.3, compared to recall of 78.7 and precision of 74.7 for Mask R-CNN. Leaf Only SAM does not perform better than the fine-tuned Mask R-CNN model on our data, but the SAM based model does not require any extra training or annotation of our new dataset. This shows there is potential to use SAM as a zero-shot classifier with the addition of post processing steps. | ['Avril Britten', 'Fraser MacFarlane', 'Dominic Williams'] | 2023-05-16 | null | null | null | null | ['plant-phenotyping'] | ['computer-vision'] | [ 3.18999797e-01 3.52218717e-01 -1.81875840e-01 -3.92673314e-01
-5.83751142e-01 -9.90671873e-01 2.36750215e-01 5.41795671e-01
-1.69553742e-01 2.79459506e-01 -6.43661499e-01 -6.53517365e-01
-1.61964938e-01 -1.03990769e+00 -5.95533967e-01 -5.01761854e-01
3.00114304e-01 6.73367739e-01 8.15439880e-01 -7.59903267e-02
-1.93234701e-02 9.30442452e-01 -1.37758589e+00 3.70557189e-01
4.80368853e-01 1.05312979e+00 5.60046971e-01 9.03782666e-01
-3.61943036e-01 2.32323155e-01 -8.15461934e-01 9.05921683e-02
4.56416965e-01 -5.77706564e-03 -6.20181799e-01 1.87847033e-01
5.56386888e-01 -3.43782634e-01 5.18690705e-01 5.54588199e-01
5.10146320e-01 -2.06052870e-01 4.06772763e-01 -1.14631820e+00
-5.55636942e-01 8.21055949e-01 -8.16959977e-01 -4.07447144e-02
-5.78810051e-02 3.56587589e-01 6.85446918e-01 -5.81833184e-01
6.34703755e-01 9.46411073e-01 1.22344851e+00 6.57778308e-02
-1.65794837e+00 -3.20536494e-01 -9.03097391e-02 -4.06890988e-01
-1.36981773e+00 4.50781696e-02 2.78794497e-01 -5.44907808e-01
9.60531116e-01 3.44730467e-01 5.77768445e-01 4.49448764e-01
-9.85281765e-02 5.78033686e-01 8.62710834e-01 -5.38633287e-01
2.00589448e-01 1.51441246e-01 5.01774728e-01 8.44500244e-01
3.73547703e-01 4.20041941e-02 3.37511063e-01 2.01180335e-02
6.76386356e-01 -5.99751845e-02 3.58298309e-02 -2.70859540e-01
-9.56111550e-01 8.98853600e-01 6.50514901e-01 5.06915987e-01
-4.86758828e-01 -1.35043055e-01 4.93617207e-01 -1.54531062e-01
4.44444597e-01 7.53893554e-01 -9.18993890e-01 2.65527815e-01
-1.62672007e+00 3.42414081e-02 9.18893695e-01 1.14084494e+00
6.73593640e-01 -3.98031883e-02 -6.62991047e-01 7.44044542e-01
1.06943555e-01 2.13890553e-01 -1.88821200e-02 -9.30240989e-01
-1.97128773e-01 9.89521861e-01 1.32245615e-01 -7.68325746e-01
-7.44775951e-01 -1.11820176e-01 -4.87394750e-01 4.91105676e-01
7.07576752e-01 -1.78691879e-01 -1.60671306e+00 1.20021427e+00
1.75767764e-01 -2.72181511e-01 -2.85145968e-01 4.01505083e-01
1.34749806e+00 7.26340830e-01 4.02698219e-01 1.74193054e-01
1.57833862e+00 -6.67288125e-01 -4.29872274e-01 -5.84717095e-02
7.13942528e-01 -8.89637232e-01 1.04530513e+00 3.43758643e-01
-8.32616389e-01 -7.88911104e-01 -1.27423608e+00 -3.18765976e-02
-1.08172274e+00 6.45698547e-01 8.40270758e-01 9.03939664e-01
-9.73969638e-01 8.21601510e-01 -7.32239485e-01 -1.03578711e+00
8.82016838e-01 5.11726141e-01 -3.38755667e-01 -9.06268805e-02
-3.18008870e-01 7.99362242e-01 9.47432935e-01 2.37842258e-02
-7.78259397e-01 -8.16897511e-01 -7.23772347e-01 3.36876333e-01
7.00213373e-01 -7.61759933e-03 1.23729253e+00 -7.42147684e-01
-1.35516346e+00 1.13836575e+00 1.33510023e-01 -4.07396704e-01
1.34506673e-01 1.20096572e-01 1.03688136e-01 7.53561184e-02
1.51897758e-01 1.30633867e+00 3.21739793e-01 -1.27161658e+00
-6.86387718e-01 -2.65566438e-01 1.26352414e-01 -6.13645732e-01
1.60640761e-01 7.49584585e-02 -3.11372131e-01 -2.76486367e-01
1.94315657e-01 -9.39874589e-01 -3.13421041e-01 3.86788994e-01
-3.62166226e-01 -1.31415814e-01 1.29302359e+00 -8.07286501e-01
8.51710558e-01 -1.89338768e+00 -5.47485590e-01 1.01098344e-02
3.64341661e-02 9.50497329e-01 -3.79908949e-01 1.87154293e-01
-2.58163929e-01 5.28171301e-01 -5.64810038e-01 2.59833753e-01
-2.36941129e-01 2.52003521e-01 9.09219086e-02 1.58405840e-01
7.16644526e-01 1.09140456e+00 -6.83054090e-01 -4.42781121e-01
4.92435902e-01 4.01330292e-01 -9.15592238e-02 1.00318678e-01
-6.37172341e-01 7.46658295e-02 4.12633084e-02 1.10038996e+00
1.04728174e+00 -1.57481119e-01 -6.12461604e-02 -2.90079951e-01
-6.00742877e-01 -2.32247576e-01 -9.68226612e-01 1.36413705e+00
-1.43106421e-02 7.60885000e-01 3.72629985e-02 -1.04218316e+00
1.19694436e+00 3.95412445e-01 4.50830668e-01 -1.49282515e-01
2.49397457e-01 7.87972584e-02 1.56535968e-01 -2.53420174e-01
3.79604876e-01 5.29811621e-01 2.52079338e-01 1.20287046e-01
2.66969681e-01 -4.93033707e-01 5.62598825e-01 -4.53419015e-02
1.32503581e+00 6.34111941e-01 5.50904691e-01 -4.96541947e-01
1.22351393e-01 7.21573055e-01 5.81620693e-01 8.77875447e-01
-2.72406042e-01 8.44393313e-01 6.06896758e-01 -4.89737749e-01
-1.10202360e+00 -6.21029496e-01 -2.25782037e-01 1.09946334e+00
-3.55533421e-01 -3.16798031e-01 -8.50549042e-01 -8.36074293e-01
-3.25420648e-02 6.97417080e-01 -6.60272717e-01 2.06479058e-01
-3.07321399e-01 -8.81520629e-01 7.35194504e-01 7.57633865e-01
6.39612138e-01 -1.41647196e+00 -1.19637370e+00 3.05522829e-01
3.12045693e-01 -1.11733079e+00 1.13926001e-01 1.03406227e+00
-8.14815283e-01 -1.05270994e+00 -6.90805137e-01 -7.98762321e-01
6.99910641e-01 3.28628302e-01 1.19766653e+00 1.68211274e-02
-7.28762329e-01 -1.60938296e-02 -7.00069427e-01 -8.95365655e-01
-2.72809893e-01 4.84107882e-01 -9.45961058e-01 -7.68567085e-01
6.98086381e-01 -1.00702092e-01 -2.45249018e-01 2.79257149e-01
-9.50743318e-01 -5.27180061e-02 5.37820399e-01 6.60367489e-01
6.03256404e-01 -1.68273710e-02 4.47997630e-01 -1.07260859e+00
1.90825462e-01 -3.16383034e-01 -9.50850248e-01 4.94990408e-01
-2.36365229e-01 -3.94643724e-01 2.56455779e-01 -5.04000247e-01
-7.15426803e-01 6.40740931e-01 -5.88578358e-03 -1.88274279e-01
-6.46863937e-01 4.98776883e-01 -3.23418170e-01 -1.02453411e-01
7.55770385e-01 -5.74007511e-01 -5.98028004e-02 -5.12439251e-01
4.66774404e-01 6.72297537e-01 5.04097164e-01 -1.93530977e-01
6.39769733e-01 4.69874978e-01 1.31408349e-01 -8.58037114e-01
-9.58210409e-01 -6.32324338e-01 -1.26068985e+00 -5.25338799e-02
1.07164586e+00 -3.44940364e-01 -6.62370205e-01 2.94292241e-01
-1.23604643e+00 -7.20175922e-01 -6.62278056e-01 -3.34539674e-02
-3.47055048e-01 1.52078211e-01 -2.84403205e-01 -5.66489041e-01
-4.17617381e-01 -8.95321667e-01 1.16274738e+00 6.18931174e-01
-2.81278580e-01 -6.12777412e-01 -4.98671740e-01 -5.63291721e-02
2.41180658e-01 6.42778397e-01 8.30229819e-01 -8.66026700e-01
-4.03200120e-01 -6.24635398e-01 -8.42801750e-01 2.59079009e-01
3.46999437e-01 5.60413122e-01 -1.09599459e+00 -7.27811009e-02
-4.51966703e-01 -1.53944433e-01 7.75037885e-01 7.03223944e-01
1.30315149e+00 2.16352940e-01 -5.54364920e-01 4.87803102e-01
1.64276564e+00 6.87307537e-01 6.33933067e-01 2.35434428e-01
6.97792411e-01 5.58899105e-01 8.88252974e-01 1.97796211e-01
-8.24917927e-02 5.09079099e-01 6.08551562e-01 -4.54715252e-01
-7.66993091e-02 4.22142521e-02 -1.73899040e-01 -1.00405157e-01
8.68718550e-02 -4.56849009e-01 -1.16006422e+00 6.81890309e-01
-1.79447258e+00 -8.44341815e-01 -3.88934195e-01 2.11348248e+00
5.40393710e-01 1.47443026e-01 1.05731554e-01 3.14799249e-01
6.44433796e-01 -1.52091354e-01 -4.33722466e-01 -8.51137578e-01
1.93725750e-02 7.38635719e-01 9.78848875e-01 2.02449396e-01
-1.45078027e+00 1.41599607e+00 6.54021120e+00 6.00531995e-01
-1.05661118e+00 -7.84186870e-02 6.17070496e-01 4.69437212e-01
5.42925954e-01 3.43576699e-01 -9.98651266e-01 6.37338981e-02
9.37629402e-01 4.57450479e-01 1.13613136e-01 1.06150556e+00
1.74966738e-01 -6.44819736e-01 -9.96479094e-01 3.47178131e-01
2.19122618e-02 -1.11692381e+00 -4.02040660e-01 -1.04049832e-01
3.69051725e-01 -2.00217709e-01 -5.51469386e-01 3.54309529e-01
6.56394184e-01 -1.19433892e+00 4.60059315e-01 2.47297183e-01
8.46461654e-01 -3.88739526e-01 8.27631176e-01 4.36773956e-01
-1.28137183e+00 3.62160951e-02 -6.70550168e-01 2.45021686e-01
-1.31423101e-01 6.25247717e-01 -1.43304217e+00 4.49325323e-01
8.06008518e-01 3.59284759e-01 -1.11657429e+00 1.25550258e+00
-2.13047154e-02 7.26702273e-01 -7.43952096e-01 1.44777104e-01
3.20167094e-01 2.28261411e-01 2.18229592e-01 1.58447707e+00
3.25668871e-01 -9.54208821e-02 5.30918658e-01 1.09538555e+00
3.53874743e-01 1.47659788e-02 -6.00053072e-01 -2.97416240e-01
3.76923829e-01 1.72801995e+00 -1.60642552e+00 -4.15478557e-01
-1.57091215e-01 8.89620781e-01 -9.65723842e-02 -1.87081471e-01
-6.17820859e-01 -6.37959301e-01 -1.62590623e-01 2.98875749e-01
8.70174587e-01 -1.43204719e-01 -5.46357572e-01 -1.78396553e-01
-3.89382482e-01 -6.97366416e-01 2.52179444e-01 -1.15599179e+00
-1.08307397e+00 3.43734443e-01 9.92096141e-02 -8.10478926e-01
8.39722529e-02 -9.93756652e-01 -7.27106035e-01 9.10366654e-01
-1.00516427e+00 -1.74125576e+00 -8.83790553e-01 -1.46870330e-01
6.91992581e-01 6.31964728e-02 1.15330100e+00 -9.39865038e-02
-5.79325199e-01 2.61505753e-01 -2.21957892e-01 2.86112487e-01
4.50297624e-01 -1.37568998e+00 3.19287032e-01 6.97044075e-01
-7.87316263e-02 2.67269731e-01 4.06951100e-01 -8.04709136e-01
-8.36463928e-01 -1.35175109e+00 6.72086179e-01 -3.99318665e-01
1.64833009e-01 -1.49855763e-01 -8.79852533e-01 8.22614014e-01
3.01245421e-01 1.83608949e-01 5.63816428e-01 -5.82965575e-02
-2.40894824e-01 4.79989462e-02 -1.68857348e+00 2.02475429e-01
6.09147906e-01 8.55210125e-02 -1.26862153e-01 2.96667933e-01
8.25236917e-01 -2.87551612e-01 -9.24268961e-01 8.26594651e-01
5.74271142e-01 -7.18603373e-01 7.49599040e-01 -3.12353075e-01
1.81895211e-01 -4.48346615e-01 -2.10281804e-01 -1.13555074e+00
-7.02188015e-01 -3.05705935e-01 2.32904017e-01 1.78047788e+00
5.55937469e-01 -3.15400153e-01 7.17134833e-01 3.05439442e-01
-1.44840211e-01 -6.26177311e-01 -2.00165555e-01 -7.74665117e-01
2.59397086e-03 -2.12896645e-01 7.26442456e-01 7.96239376e-01
-6.24219239e-01 -5.92929497e-02 3.83204788e-01 1.68144971e-01
8.76812488e-02 3.40860337e-02 7.37520039e-01 -1.67261481e+00
4.45958525e-02 -3.77073050e-01 -2.95523167e-01 -4.69614536e-01
-2.17303157e-01 -9.84906554e-01 3.58154595e-01 -1.96013737e+00
9.85009596e-02 -5.85662186e-01 2.17910111e-02 1.06700206e+00
3.77851166e-02 2.43268490e-01 3.24481100e-01 -1.26268029e-01
-5.05905524e-02 -3.84373099e-01 8.16758394e-01 -1.75664827e-01
-5.63706696e-01 5.12538943e-03 -7.08757877e-01 8.80292654e-01
1.33346951e+00 -5.26681781e-01 -1.18132509e-01 -7.77594969e-02
-8.88798386e-02 -3.51763904e-01 5.40898919e-01 -1.15143096e+00
-2.82936215e-01 -2.35715462e-03 7.61589468e-01 -1.08946455e+00
1.51212901e-01 -9.48584855e-01 4.79260087e-02 4.32599932e-01
-4.48297523e-02 -4.18013483e-02 7.61777043e-01 -4.99958619e-02
4.73295718e-01 -8.19609404e-01 1.14920974e+00 -5.62258542e-01
-7.82829404e-01 9.58634317e-02 -4.76199895e-01 -5.39546907e-01
1.31674969e+00 -6.99333429e-01 -3.81559044e-01 2.73039490e-01
-8.35639715e-01 1.03289366e-01 2.93432832e-01 4.24197823e-01
8.96485448e-02 -7.94189513e-01 -5.39276481e-01 1.02709472e-01
1.65221423e-01 2.65184522e-01 -3.38114560e-01 4.72100526e-01
-1.03360152e+00 6.95857406e-01 -5.30966341e-01 -8.65795612e-01
-1.44009936e+00 7.22057343e-01 1.02201305e-01 -3.35563242e-01
-6.57344460e-01 6.31349146e-01 -1.96724832e-01 -6.39712811e-01
2.19426066e-01 -7.36885250e-01 -4.62849110e-01 3.71472389e-01
2.42328569e-01 3.01368177e-01 1.35069937e-01 -3.12142938e-01
-1.82719588e-01 2.71589756e-01 2.62069311e-02 3.85971591e-02
1.51006019e+00 5.80349267e-01 -5.89932390e-02 6.26048803e-01
5.01463354e-01 -3.38524282e-01 -1.06481361e+00 2.86868185e-01
5.22149801e-01 -1.99589714e-01 6.15750141e-02 -1.26613343e+00
-1.04691517e+00 8.02040696e-01 1.05714619e+00 5.41591644e-01
1.08869064e+00 4.87522557e-02 4.06084090e-01 2.45818913e-01
9.17939618e-02 -9.16673183e-01 -3.62966985e-01 3.80172402e-01
8.37937355e-01 -1.40384960e+00 -3.22844088e-02 -6.98332906e-01
-2.61507243e-01 1.00780582e+00 9.02063131e-01 -1.76936463e-01
8.12500119e-01 7.13716507e-01 -2.71691871e-03 -3.46745849e-01
-4.45962608e-01 -7.48657048e-01 -4.04697988e-04 1.18965507e+00
6.91613972e-01 3.11869442e-01 -4.01121527e-01 4.64497060e-01
5.32965958e-02 2.97538668e-01 3.59587520e-01 1.22289586e+00
-6.81187928e-01 -1.05420697e+00 -4.85154867e-01 7.08009183e-01
-4.11666304e-01 2.04088576e-02 -7.60732889e-01 9.39463794e-01
6.29014075e-01 9.42395985e-01 1.93767965e-01 -4.96227890e-02
3.57509822e-01 2.56296426e-01 4.63506132e-01 -1.13260019e+00
-1.00850368e+00 -1.54903214e-02 -1.25310674e-01 -3.73703510e-01
-2.53914922e-01 -5.89987934e-01 -1.00799465e+00 -1.80244520e-01
-7.23445237e-01 -4.69516754e-01 9.19354618e-01 6.62301362e-01
3.51023495e-01 7.77686000e-01 4.61481065e-02 -1.04359758e+00
2.24695429e-02 -1.21564925e+00 -3.03884178e-01 -1.78522512e-01
-1.21876627e-01 -3.90517026e-01 2.01822758e-01 3.24335724e-01] | [9.112325668334961, -1.5309500694274902] |
ff20f41d-ed57-4101-b34e-8b07b6e859a2 | unsupervised-long-term-person-re | 2202.03087 | null | https://arxiv.org/abs/2202.03087v2 | https://arxiv.org/pdf/2202.03087v2.pdf | Unsupervised Long-Term Person Re-Identification with Clothes Change | We investigate unsupervised person re-identification (Re-ID) with clothes change, a new challenging problem with more practical usability and scalability to real-world deployment. Most existing re-id methods artificially assume the clothes of every single person to be stationary across space and time. This condition is mostly valid for short-term re-id scenarios since an average person would often change the clothes even within a single day. To alleviate this assumption, several recent works have introduced the clothes change facet to re-id, with a focus on supervised learning person identity discriminative representation with invariance to clothes changes. Taking a step further towards this long-term re-id direction, we further eliminate the requirement of person identity labels, as they are significantly more expensive and more tedious to annotate in comparison to short-term person re-id datasets. Compared to conventional unsupervised short-term re-id, this new problem is drastically more challenging as different people may have similar clothes whilst the same person can wear multiple suites of clothes over different locations and times with very distinct appearance. To overcome such obstacles, we introduce a novel Curriculum Person Clustering (CPC) method that can adaptively regulate the unsupervised clustering criterion according to the clustering confidence. Experiments on three long-term person re-id datasets show that our CPC outperforms SOTA unsupervised re-id methods and even closely matches the supervised re-id models. | ['Jun Guo', 'Xiatian Zhu', 'Peng Xu', 'Mingkun Li'] | 2022-02-07 | null | null | null | null | ['unsupervised-long-term-person-re', 'unsupervised-person-re-identification'] | ['computer-vision', 'computer-vision'] | [ 4.02373858e-02 -3.40263575e-01 2.77910918e-01 -4.71006393e-01
6.39062598e-02 -8.56271803e-01 6.29165053e-01 7.09181204e-02
-5.35878897e-01 4.25832182e-01 -3.38180810e-02 4.13259357e-01
-1.71980157e-01 -5.31389654e-01 -4.70351160e-01 -6.85252547e-01
1.45189658e-01 8.05408120e-01 2.68002562e-02 -9.79948565e-02
-7.77325854e-02 5.16749561e-01 -1.86512268e+00 -2.94803768e-01
6.39399111e-01 3.49117637e-01 4.89699543e-02 6.75877810e-01
2.53690779e-01 1.24573313e-01 -7.49900579e-01 -7.20720887e-01
6.14856541e-01 -4.53234196e-01 -7.36574411e-01 5.12785256e-01
8.71887624e-01 -1.44400358e-01 -1.56065270e-01 1.28500903e+00
6.99632525e-01 4.87289280e-01 6.12885177e-01 -1.37319589e+00
-9.13506389e-01 1.93705887e-01 -6.07766807e-01 -9.38953310e-02
6.70777440e-01 -1.54979303e-01 4.69013929e-01 -5.30375957e-01
3.57018441e-01 1.27594590e+00 1.05829322e+00 1.07569969e+00
-1.42133439e+00 -6.41933799e-01 5.31331956e-01 1.76450133e-01
-1.79365754e+00 -4.26869333e-01 7.35040843e-01 -4.16360319e-01
2.89001584e-01 4.27681535e-01 7.55757451e-01 1.28582811e+00
-3.94776404e-01 4.76021677e-01 1.26482475e+00 -4.54524994e-01
5.49811684e-02 5.27507484e-01 3.37063849e-01 4.94707704e-01
4.47319388e-01 -3.64465892e-01 -1.72483519e-01 -7.42919371e-02
6.03891611e-01 5.41914761e-01 -1.24208972e-01 -3.94842356e-01
-1.30184782e+00 3.66725206e-01 -1.28600579e-02 2.93052644e-01
-6.72658086e-02 -9.97699276e-02 3.62428665e-01 4.56373036e-01
2.40061164e-01 2.84126788e-01 -1.58032715e-01 1.07284030e-02
-8.25320065e-01 4.28291619e-01 7.92382777e-01 1.15118682e+00
7.48109162e-01 -3.09837401e-01 -9.97756645e-02 1.17359042e+00
1.12270024e-02 7.20068634e-01 6.61908209e-01 -6.61624789e-01
-1.04777962e-01 5.79207361e-01 4.77822959e-01 -1.16622245e+00
-3.36140543e-01 -4.38592821e-01 -1.05912793e+00 -2.93870717e-02
6.90969944e-01 3.05523019e-04 -8.41854811e-01 1.75537813e+00
4.64265555e-01 1.99037105e-01 -2.26593286e-01 8.92467976e-01
5.42497993e-01 7.84385130e-02 1.29087433e-01 -2.56379545e-01
1.48045218e+00 -9.97494996e-01 -7.38160431e-01 -2.27115471e-02
1.23432226e-01 -6.14217520e-01 1.03222251e+00 3.73358786e-01
-8.30525935e-01 -1.06987691e+00 -9.64900434e-01 3.64857793e-01
-7.28884876e-01 1.65598109e-01 4.26717579e-01 1.18243253e+00
-1.00880098e+00 5.62281489e-01 -3.17362845e-01 -1.03649557e+00
1.67042110e-02 5.56720614e-01 -4.26567167e-01 -3.92530747e-02
-1.02971280e+00 5.69484830e-01 7.37522542e-02 1.42896309e-01
-6.11309707e-01 -4.91143525e-01 -6.72874093e-01 -3.10407609e-01
3.31863075e-01 -6.57649338e-01 9.14023578e-01 -1.25532758e+00
-1.52292454e+00 1.33593607e+00 -2.94672191e-01 -7.59564489e-02
7.76493251e-01 -1.27709955e-01 -8.55992258e-01 -2.42758185e-01
1.93604410e-01 3.03664446e-01 1.10835612e+00 -1.66859066e+00
-5.43409526e-01 -6.42283380e-01 2.89768958e-03 2.57726520e-01
-6.88126862e-01 1.16820440e-01 -8.51871371e-01 -8.68697286e-01
-3.35943289e-02 -1.54748559e+00 -5.13858721e-02 -2.28405148e-01
-2.93538988e-01 -5.10202110e-01 8.19004893e-01 -5.27435184e-01
1.04338682e+00 -2.01572108e+00 2.24343911e-01 3.06770772e-01
1.45427987e-01 2.18426034e-01 1.07849084e-01 1.26107052e-01
-1.98954359e-01 -1.51472008e-02 8.01078975e-02 -8.60830724e-01
1.38059542e-01 2.93974459e-01 1.50487483e-01 7.41819680e-01
-5.38873434e-01 6.05396092e-01 -1.07019579e+00 -4.03204888e-01
3.20748478e-01 2.17273399e-01 -1.81253161e-03 2.90818781e-01
4.11889642e-01 7.03507185e-01 2.01054916e-01 7.49456644e-01
7.14283347e-01 -7.21236989e-02 3.34236026e-01 -2.72278309e-01
6.87897429e-02 -6.13221109e-01 -1.46346784e+00 1.41407430e+00
-2.22286806e-01 2.47905985e-01 9.30060148e-02 -9.97608125e-01
7.99334824e-01 2.34217733e-01 7.89626181e-01 -3.52464944e-01
6.81896554e-03 -1.78682074e-01 -3.58516395e-01 -3.51942182e-01
5.45156062e-01 -1.03302680e-01 -3.90735328e-01 4.68838692e-01
-2.18607426e-01 5.37396371e-01 1.68174356e-01 -5.80005907e-02
6.69422030e-01 9.98710617e-02 1.95304856e-01 -3.96195203e-01
7.13317633e-01 -2.56610185e-01 6.59307420e-01 1.01936126e+00
-6.80988669e-01 6.58408225e-01 -5.13656259e-01 -5.48862576e-01
-1.02987158e+00 -1.06940401e+00 5.89751266e-02 1.31659997e+00
7.30547547e-01 -1.53907090e-01 -1.04093778e+00 -8.81588459e-01
1.34008408e-01 1.46871209e-01 -7.89061427e-01 8.11812207e-02
-4.74884361e-01 -9.04140115e-01 6.43048167e-01 2.89351225e-01
8.28957081e-01 -6.36745036e-01 -7.86276832e-02 9.52222869e-02
-3.75614494e-01 -1.16638291e+00 -9.34953988e-01 -3.09089273e-01
-2.05643922e-01 -9.26705182e-01 -1.33839440e+00 -9.33312595e-01
1.08254337e+00 6.10183895e-01 9.27265525e-01 1.39477804e-01
-4.28396791e-01 1.21768332e+00 -4.78768498e-01 -3.22773099e-01
-2.21542194e-01 -7.84648657e-02 8.54815781e-01 6.67031825e-01
7.79584885e-01 -5.04133999e-01 -7.77847171e-01 8.06264639e-01
-4.58036095e-01 -2.09780753e-01 3.19067240e-02 5.71162939e-01
4.64868158e-01 5.83017945e-01 5.25039911e-01 -8.22511435e-01
5.36400378e-01 -3.49406868e-01 -1.86821908e-01 6.04401886e-01
-8.22419405e-01 -4.48840141e-01 6.78746939e-01 -1.01142633e+00
-9.44502234e-01 2.38545299e-01 3.02163273e-01 -4.65767473e-01
-5.18370628e-01 -3.71279895e-01 -4.12662745e-01 -1.95222542e-01
4.07695413e-01 3.29837322e-01 -1.52624413e-01 -6.05399370e-01
1.68630376e-01 8.10135961e-01 8.76483560e-01 -6.39241934e-01
1.26412868e+00 7.91822433e-01 -3.31357539e-01 -7.77113020e-01
-7.50001192e-01 -1.02437663e+00 -1.24494386e+00 -5.85877180e-01
1.06702542e+00 -9.93219376e-01 -1.03955245e+00 8.13417673e-01
-8.02115798e-01 -2.68424809e-01 -2.45053500e-01 1.49765223e-01
-1.31011188e-01 8.43031287e-01 -2.98225969e-01 -8.94809425e-01
-2.61592031e-01 -6.07332528e-01 8.59570980e-01 4.26806867e-01
-4.79849368e-01 -1.22108662e+00 9.83568467e-03 4.56112325e-01
1.35057941e-01 3.36826414e-01 2.49375612e-01 -3.86372745e-01
1.12189129e-01 -3.32601577e-01 -6.48659244e-02 2.92023420e-01
7.25476623e-01 -4.77444649e-01 -8.85178149e-01 -8.21068823e-01
-2.38843828e-01 3.07800621e-02 4.80727345e-01 -7.33250529e-02
1.21395922e+00 -2.52465874e-01 -4.09932166e-01 5.31369507e-01
1.39454556e+00 1.05868736e-02 2.75771171e-01 3.59639347e-01
1.14952183e+00 8.84382665e-01 5.34220934e-01 5.76829672e-01
7.11141944e-01 9.91479933e-01 -1.52217552e-01 -3.44296485e-01
-3.18056792e-01 -1.41927227e-01 4.28198576e-01 7.60973454e-01
-7.11784661e-01 -1.49683207e-01 -5.72554767e-01 7.19338655e-01
-1.97778070e+00 -1.17002571e+00 -7.22763911e-02 2.63066745e+00
5.70202291e-01 -3.72605801e-01 7.31760561e-01 1.38058677e-01
1.16722417e+00 -2.72157192e-01 -6.12472773e-01 -3.48674022e-02
-5.08582145e-02 -2.77584761e-01 8.16512465e-01 2.70064443e-01
-1.27336991e+00 7.49252319e-01 6.31338215e+00 5.43663621e-01
-7.33034909e-01 3.20628434e-01 2.36307889e-01 9.01622176e-02
1.70084864e-01 -3.37629795e-01 -1.01761305e+00 7.73685813e-01
6.45414293e-01 7.88912624e-02 7.61208057e-01 7.31432140e-01
1.00237854e-01 1.25562772e-01 -1.17068446e+00 1.68054903e+00
5.36305368e-01 -4.64970440e-01 -1.52077898e-01 8.53116736e-02
8.64767253e-01 -5.45074821e-01 -8.17980394e-02 2.34685495e-01
4.81657445e-01 -8.25833976e-01 6.72683954e-01 6.41773522e-01
9.81235862e-01 -7.03182399e-01 6.00200891e-01 1.60846293e-01
-1.55265927e+00 -1.53403088e-01 -4.33588207e-01 6.35288432e-02
6.59455881e-02 2.61675566e-01 -4.58574355e-01 5.22040844e-01
1.20789051e+00 6.64168775e-01 -9.30143118e-01 6.05489671e-01
1.98436603e-01 1.45540848e-01 -1.47046775e-01 2.95125544e-01
-3.45583379e-01 -2.06836447e-01 4.67794180e-01 1.44809544e+00
2.55489022e-01 -7.87233189e-02 5.96797407e-01 4.34468687e-01
1.52116120e-01 -1.23006545e-01 -3.81120801e-01 3.74120414e-01
3.36887687e-01 1.26899028e+00 -8.19904089e-01 -5.32715559e-01
-4.19134796e-01 2.03060246e+00 1.48056895e-01 5.09916306e-01
-8.95250678e-01 -7.89270774e-02 6.73804820e-01 4.85162169e-01
9.98552814e-02 -3.58459920e-01 2.61815041e-01 -1.18933070e+00
2.71883253e-02 -8.20562363e-01 6.30725443e-01 -3.39858890e-01
-1.94699514e+00 3.74634802e-01 1.58878982e-01 -1.35648119e+00
-3.58623266e-02 -4.03354704e-01 -1.99033752e-01 6.10589147e-01
-1.23719668e+00 -1.62096274e+00 -8.63812208e-01 1.12849545e+00
5.39162874e-01 -3.29220176e-01 1.06131566e+00 6.04575038e-01
-7.42741466e-01 1.10501432e+00 1.48343235e-01 2.48407856e-01
1.19016111e+00 -1.50131178e+00 2.33501121e-01 8.23696256e-01
-4.69809137e-02 1.07227707e+00 7.51080096e-01 -8.04001570e-01
-1.34923232e+00 -1.34803581e+00 6.25459969e-01 -7.88947344e-01
1.93703502e-01 -7.50882566e-01 -6.49437368e-01 7.25511193e-01
-1.56616881e-01 4.05101255e-02 9.40300584e-01 3.49971950e-01
-3.78175139e-01 -4.08062488e-01 -1.31248689e+00 5.77276230e-01
1.53510284e+00 -5.61734140e-01 -2.96654195e-01 4.73106682e-01
1.50076494e-01 9.24595539e-03 -1.13650358e+00 2.77613997e-01
7.65808225e-01 -7.58313060e-01 1.25493670e+00 -9.86004770e-02
-6.03092611e-01 -7.02013493e-01 9.59153399e-02 -1.02166831e+00
-7.11765409e-01 -8.83591950e-01 3.71787921e-02 1.66349006e+00
-3.50710124e-01 -6.11446857e-01 8.18712711e-01 9.25605714e-01
4.62491870e-01 7.09872097e-02 -5.56318820e-01 -1.31264973e+00
-3.07250619e-01 -1.42067913e-02 7.64808357e-01 1.33457983e+00
-2.36936450e-01 4.10002954e-02 -9.68951523e-01 4.54635680e-01
1.03083897e+00 1.44003108e-01 1.23157823e+00 -1.55199063e+00
-4.76083368e-01 -1.57299921e-01 -4.64452565e-01 -1.07186472e+00
1.88273266e-01 -7.64267981e-01 7.54656494e-02 -1.23412728e+00
7.43228793e-01 -7.48443127e-01 -3.35469574e-01 4.89117771e-01
-3.40054810e-01 6.01694643e-01 1.11511521e-01 6.25159621e-01
-9.16183293e-01 1.53265983e-01 8.91627908e-01 -3.28600943e-01
-2.75306404e-01 2.33445272e-01 -6.66804910e-01 7.07437634e-01
6.48166537e-01 -3.17569226e-01 -4.34939831e-01 -2.18188167e-01
-1.72431707e-01 -6.19206250e-01 6.03596866e-01 -1.25240707e+00
4.30820733e-01 -1.02353856e-01 5.54483473e-01 -2.83138394e-01
2.54168302e-01 -1.13601911e+00 5.04107893e-01 1.60650864e-01
4.07081954e-02 1.46077529e-01 -1.37379244e-01 7.82618582e-01
1.84018552e-01 -1.93267018e-01 8.87896299e-01 -4.94111806e-01
-9.50048566e-01 5.07892668e-01 -3.35238338e-01 -4.27764744e-01
1.25804174e+00 -7.08519936e-01 1.93341330e-01 -3.51323158e-01
-1.07956862e+00 1.01101771e-01 1.04020596e+00 6.29413009e-01
3.46013576e-01 -1.32118058e+00 -7.29758024e-01 1.57814667e-01
3.99185598e-01 -3.36798966e-01 3.25859904e-01 4.44844335e-01
1.90574359e-02 -1.19584367e-01 -1.44106135e-01 -6.45925939e-01
-1.86699080e+00 9.12881136e-01 3.74872148e-01 -8.24188441e-02
-8.07627738e-01 6.62928462e-01 1.87445983e-01 -7.06143796e-01
4.05523628e-01 5.15812576e-01 -3.48402351e-01 1.67810861e-02
6.21810555e-01 5.52435517e-01 -2.95468539e-01 -1.12056768e+00
-3.74700844e-01 1.05042911e+00 -1.67531505e-01 1.30583093e-01
8.78887117e-01 -7.66343057e-01 -1.39206678e-01 5.94357848e-01
8.69783163e-01 2.65585750e-01 -1.16756368e+00 -3.83909762e-01
-4.94754687e-02 -5.05262434e-01 -6.27560377e-01 -5.49046397e-01
-8.10355604e-01 2.11946711e-01 1.12045705e+00 2.97074646e-01
1.00415432e+00 -6.86872900e-02 9.35030520e-01 3.45893741e-01
6.14098608e-01 -1.60893619e+00 1.89464435e-01 1.63530380e-01
5.58184981e-01 -1.40517676e+00 2.47268498e-01 -3.69144917e-01
-5.59783161e-01 7.39732385e-01 6.18588865e-01 5.26762195e-02
5.44325769e-01 -1.56138331e-01 1.31283641e-01 -1.01649292e-01
2.88274944e-01 -3.78846556e-01 2.95234799e-01 1.00197804e+00
-6.33817911e-02 3.35350871e-01 1.46045789e-01 5.78374743e-01
-1.74957931e-01 -4.36057806e-01 1.57865971e-01 6.21867299e-01
1.12312213e-01 -1.27604687e+00 -7.14047968e-01 -8.87338631e-03
-2.90718406e-01 4.54282492e-01 -5.10231555e-01 5.96574128e-01
7.95412600e-01 1.19077969e+00 4.03221957e-02 -4.77945685e-01
3.97086471e-01 2.25025509e-02 6.99154735e-01 -6.26924336e-01
-6.45029128e-01 -1.45682722e-01 -1.41579017e-01 -2.19251737e-01
-9.32408571e-01 -8.24868262e-01 -8.77700984e-01 -7.20828831e-01
-1.82600260e-01 -1.00098625e-01 4.14697945e-01 7.78859437e-01
1.30849883e-01 3.12544554e-01 7.86319137e-01 -8.65983605e-01
-1.73622578e-01 -8.07697475e-01 -6.97736382e-01 1.34140456e+00
1.99191630e-01 -6.93765998e-01 -3.32225502e-01 5.75246394e-01] | [14.707045555114746, 1.0508508682250977] |
a9dfa86c-eff0-4e08-a101-b8ee30ed0bd3 | a-review-for-tone-mapping-operators-on-wide | 2101.03003 | null | https://arxiv.org/abs/2101.03003v1 | https://arxiv.org/pdf/2101.03003v1.pdf | A review for Tone-mapping Operators on Wide Dynamic Range Image | The dynamic range of our normal life can exceeds 120 dB, however, the smart-phone cameras and the conventional digital cameras can only capture a dynamic range of 90 dB, which sometimes leads to loss of details for the recorded image. Now, some professional hardware applications and image fusion algorithms have been devised to take wide dynamic range (WDR), but unfortunately existing devices cannot display WDR image. Tone mapping (TM) thus becomes an essential step for exhibiting WDR image on our ordinary screens, which convert the WDR image into low dynamic range (LDR) image. More and more researchers are focusing on this topic, and give their efforts to design an excellent tone mapping operator (TMO), showing detailed images as the same as the perception that human eyes could receive. Therefore, it is important for us to know the history, development, and trend of TM before proposing a practicable TMO. In this paper, we present a comprehensive study of the most well-known TMOs, which divides TMOs into traditional and machine learning-based category. | ['Ziyi Liu'] | 2021-01-08 | null | null | null | null | ['tone-mapping'] | ['computer-vision'] | [ 4.64922667e-01 -5.99491656e-01 -3.31522077e-02 -1.74090415e-01
-1.87743083e-01 -2.18046397e-01 8.47717747e-02 -6.28515661e-01
-2.07204610e-01 4.89564329e-01 -1.28308877e-01 -4.78869110e-01
1.26968279e-01 -8.45471621e-01 -1.55818075e-01 -7.81658649e-01
3.79836142e-01 -4.18772310e-01 4.77755547e-01 -3.87515336e-01
2.64196903e-01 3.39095503e-01 -1.64684606e+00 1.89371124e-01
1.01788712e+00 1.17145061e+00 4.82305259e-01 4.34023976e-01
-7.68735781e-02 5.80688000e-01 -7.90275693e-01 -4.64333087e-01
3.01439971e-01 -4.76806879e-01 -1.49563864e-01 1.52465463e-01
4.64864485e-02 -5.46038568e-01 -5.34062982e-01 1.46852982e+00
4.35659349e-01 -2.27103740e-01 3.52581739e-01 -1.08981907e+00
-1.02698362e+00 4.33403581e-01 -9.84469116e-01 2.26710320e-01
5.00318646e-01 1.85303882e-01 3.82239550e-01 -5.36385953e-01
1.26835927e-01 7.94469893e-01 2.97297508e-01 3.94710690e-01
-1.00107205e+00 -9.51181889e-01 -4.42093670e-01 4.57181722e-01
-1.48499668e+00 -2.61161089e-01 9.59585726e-01 1.43749733e-02
6.03466153e-01 4.66547519e-01 8.22567225e-01 6.98807061e-01
4.28866625e-01 3.41344833e-01 1.79928315e+00 -5.65487921e-01
-2.41066635e-01 4.72695917e-01 -1.55822217e-01 2.81734526e-01
2.50991255e-01 2.72020400e-01 -3.83348137e-01 3.40940982e-01
9.73517120e-01 1.10241361e-01 -7.51810193e-01 2.20716178e-01
-1.08016002e+00 1.48221746e-01 3.30940664e-01 6.39714360e-01
7.64236376e-02 -3.06297928e-01 -1.26601487e-01 5.73951483e-01
5.14251776e-02 3.65906835e-01 5.59918359e-02 -2.38415152e-01
-6.01225793e-01 -6.23147249e-01 2.93572515e-01 8.38625371e-01
6.14671946e-01 2.54272837e-02 3.58379006e-01 1.07041538e+00
1.28615960e-01 7.69196808e-01 5.24027765e-01 -9.50055063e-01
2.69702196e-01 4.11194682e-01 9.54365060e-02 -9.75429296e-01
-3.37815136e-02 -1.34167358e-01 -9.93170500e-01 4.15998638e-01
1.06208280e-01 -3.79482359e-02 -7.20972538e-01 1.06189585e+00
-6.98163137e-02 -1.13498844e-01 5.82669936e-02 1.18238938e+00
8.36119592e-01 1.11286461e+00 -4.21336532e-01 -6.42660439e-01
1.43363786e+00 -4.13054049e-01 -1.20578599e+00 -1.07771099e-01
-5.74632920e-02 -1.30928445e+00 1.52280521e+00 1.09800649e+00
-1.00519800e+00 -9.00760949e-01 -1.36531758e+00 2.72093173e-02
3.15561406e-02 3.04801017e-01 5.66397905e-01 9.63122666e-01
-8.59455109e-01 1.81502879e-01 -4.03472692e-01 -2.15230823e-01
-8.76069069e-02 -8.14167410e-02 -1.60647482e-01 -1.88208863e-01
-1.22343075e+00 9.67627645e-01 7.83085525e-02 3.76231641e-01
-2.22606838e-01 -4.16133463e-01 -1.94910392e-01 -1.65343553e-01
4.40140307e-01 -2.73287952e-01 1.10044968e+00 -8.67368639e-01
-1.78629482e+00 1.00112891e+00 3.73875648e-02 1.84611939e-02
2.80719966e-01 -1.13021374e-01 -1.29469776e+00 3.54570121e-01
-3.50015849e-01 4.02884409e-02 8.44557285e-01 -1.26656783e+00
-6.36025250e-01 -9.50523093e-02 7.32825324e-03 1.91689014e-01
-3.59811276e-01 2.92887509e-01 -5.01959503e-01 -4.85672772e-01
3.70906144e-01 -6.25595093e-01 2.20235646e-01 7.57211819e-02
-3.75924170e-01 9.35327113e-02 1.09139800e+00 -5.03056645e-01
1.61540389e+00 -2.42619824e+00 -4.56926703e-01 9.92076239e-04
1.36731014e-01 5.33107638e-01 3.19732606e-01 3.08423787e-01
-7.03454688e-02 -1.83861271e-01 -6.88376650e-02 5.20767689e-01
-3.19385111e-01 -2.10602939e-01 -6.24521971e-01 2.51998007e-01
-3.51797432e-01 4.21238363e-01 -3.08077455e-01 -5.70809186e-01
4.20441985e-01 3.45385313e-01 1.96999684e-02 2.57898718e-01
2.60734230e-01 3.67764801e-01 -3.22347373e-01 8.89942050e-01
9.36487556e-01 -1.64380893e-01 -1.14099391e-01 -8.08282375e-01
-4.87692088e-01 -3.04669768e-01 -1.16740000e+00 1.04809642e+00
-4.02632952e-01 9.14104402e-01 -8.26389641e-02 -5.17027259e-01
1.42438722e+00 1.61455527e-01 4.25455242e-01 -1.30284345e+00
1.97624087e-01 4.78207856e-01 -1.30120650e-01 -9.84742641e-01
5.82675576e-01 -4.37988102e-01 6.88072518e-02 3.53829265e-01
-5.91662347e-01 -1.68142319e-01 -1.94763467e-01 3.52527648e-02
5.64055562e-01 -2.99071580e-01 3.07129502e-01 5.83563745e-02
5.25604725e-01 -1.59392133e-01 3.84657353e-01 2.71744102e-01
-1.85060009e-01 6.90927386e-01 2.15393245e-01 -3.35360706e-01
-9.72694159e-01 -1.24066877e+00 -3.80719751e-01 5.59899449e-01
9.63635385e-01 7.70779923e-02 -5.70464253e-01 1.52681604e-01
-4.25313771e-01 5.51824212e-01 9.71028358e-02 -3.25097322e-01
-5.72869420e-01 -7.21691370e-01 3.92605811e-01 2.01023072e-01
1.26997578e+00 -8.28495681e-01 -6.11880362e-01 -1.92008857e-02
-3.41574490e-01 -1.08120787e+00 -4.82549340e-01 -2.98033953e-01
-6.44685328e-01 -9.97301161e-01 -7.44268179e-01 -8.13288629e-01
3.99570882e-01 7.58278906e-01 7.24244297e-01 1.04806930e-01
-4.58857507e-01 2.05351189e-01 -4.84820187e-01 -2.92300284e-01
-2.33749524e-01 -4.38796580e-01 -3.54330689e-02 2.37741590e-01
5.73868930e-01 -5.80559790e-01 -8.03803980e-01 7.53326714e-01
-9.85628247e-01 2.67062217e-01 9.73565817e-01 2.76632637e-01
5.75142980e-01 5.04806817e-01 4.07640129e-01 -5.59839368e-01
6.27370358e-01 1.29184231e-01 -8.00024092e-01 4.09727067e-01
-7.14377642e-01 -5.19678891e-01 5.80192387e-01 -6.22750998e-01
-1.45973825e+00 -4.42623198e-01 -1.19847476e-01 -2.71614432e-01
-2.38826517e-02 1.27639264e-01 -3.39465022e-01 -2.38841191e-01
7.72167683e-01 4.67393607e-01 9.40900221e-02 -4.42868292e-01
4.43623029e-02 1.46193671e+00 8.67319465e-01 -6.53694049e-02
8.10287416e-01 4.71745163e-01 -8.16538706e-02 -9.34147179e-01
-4.96498644e-01 -2.51452863e-01 5.13901263e-02 -6.39379144e-01
7.40542114e-01 -8.89409244e-01 -8.92239988e-01 6.87732697e-01
-6.53284132e-01 -1.46960437e-01 1.50557622e-01 8.42797756e-01
-2.35101804e-01 3.92887473e-01 -6.58993125e-01 -6.04294360e-01
-1.08231239e-01 -9.93934214e-01 6.57583594e-01 8.45769465e-01
3.32000792e-01 -3.99625599e-01 -8.16267878e-02 4.85622793e-01
7.56552398e-01 -1.14266694e-01 6.55312657e-01 5.08130729e-01
-8.85920048e-01 -1.32502913e-01 -5.43862224e-01 3.50667417e-01
3.62413555e-01 2.17941880e-01 -1.16422391e+00 -6.15499355e-02
4.25784171e-01 -1.95891690e-02 7.70560324e-01 3.93383175e-01
1.34771299e+00 1.75139010e-01 -3.04437429e-01 7.11267710e-01
1.78725946e+00 9.15629089e-01 1.26863861e+00 4.05886620e-01
2.85835057e-01 2.20902622e-01 1.00324082e+00 5.45629188e-02
-2.43467093e-02 7.72329748e-01 2.26878807e-01 -5.17486691e-01
-2.12382525e-01 -2.06575885e-01 3.23046803e-01 8.02529812e-01
-2.70079345e-01 -3.25691462e-01 -6.28532112e-01 -7.80248791e-02
-1.05707026e+00 -9.79317546e-01 -2.73145318e-01 2.10031104e+00
8.53452444e-01 1.57557771e-01 -2.18598366e-01 2.71475971e-01
1.06199622e+00 6.92955106e-02 -4.50166255e-01 -3.11850429e-01
-3.45836133e-01 4.37861448e-03 4.11898166e-01 1.91680774e-01
-7.01089382e-01 4.16741699e-01 6.63724995e+00 1.07451069e+00
-1.77389228e+00 -1.93339005e-01 5.72228968e-01 1.94741599e-02
-3.72118622e-01 -1.69606566e-01 -4.98642385e-01 6.48192346e-01
4.02282208e-01 -2.16321588e-01 5.83749294e-01 5.95630586e-01
3.28370839e-01 -5.56355596e-01 -4.80899543e-01 1.61468971e+00
1.67671014e-02 -8.27227116e-01 -1.59469426e-01 1.10297613e-01
5.62733114e-01 -6.31795526e-01 5.33962548e-01 -6.39844760e-02
-3.65025759e-01 -8.48783553e-01 2.38324478e-01 7.16975868e-01
1.36328435e+00 -6.34446204e-01 5.63760698e-01 1.12910643e-01
-1.08585095e+00 -6.20834492e-02 -7.54536808e-01 7.54784569e-02
1.69190109e-01 9.07126844e-01 -1.31698206e-01 3.45938146e-01
9.97406006e-01 4.04551834e-01 -4.83594835e-01 1.08065593e+00
-4.11714502e-02 4.73321497e-01 -2.22063765e-01 6.36024848e-02
-3.12766045e-01 -4.68363166e-01 4.59105939e-01 6.53062880e-01
7.40244269e-01 5.37852466e-01 -1.53521404e-01 6.90230846e-01
6.38929009e-03 -7.78730512e-02 -4.77934510e-01 7.29791000e-02
6.15530372e-01 1.28356063e+00 -7.41940558e-01 -1.29106283e-01
-8.12820733e-01 8.19674015e-01 -6.19931102e-01 3.33317220e-01
-1.02817738e+00 -9.77274060e-01 3.47023964e-01 3.39964628e-01
-9.18950662e-02 -1.48335531e-01 -3.20205152e-01 -1.16437733e+00
1.53519705e-01 -8.64310682e-01 1.66083083e-01 -1.40815330e+00
-1.02678573e+00 4.89880532e-01 -1.52379215e-01 -1.90884078e+00
4.73211020e-01 -5.18606305e-01 -7.27669537e-01 8.10261309e-01
-1.56246781e+00 -7.47786820e-01 -6.66613042e-01 6.28901422e-01
2.71611840e-01 -1.23198882e-01 2.44587526e-01 5.60027242e-01
-4.38292265e-01 4.08097655e-01 7.18991980e-02 -1.08499803e-01
1.13412797e+00 -7.37536907e-01 -2.43775904e-01 8.16977382e-01
-1.36952803e-01 5.63927948e-01 7.44087100e-01 -4.21904892e-01
-1.35111833e+00 -5.87085843e-01 4.57342088e-01 9.86335874e-02
2.53651470e-01 5.98082617e-02 -9.72420096e-01 3.62947613e-01
3.10554445e-01 -2.04095528e-01 6.07572913e-01 -4.17097360e-01
-1.39492258e-01 -9.16379869e-01 -1.07497501e+00 6.69466674e-01
9.40993845e-01 -6.40882254e-01 -4.58782971e-01 -1.23177774e-01
6.88044965e-01 -3.97313625e-01 -8.01328242e-01 4.37847853e-01
7.28321075e-01 -1.58797884e+00 8.69247496e-01 5.91827393e-01
2.61322409e-01 -7.52994657e-01 -2.95192804e-02 -9.38176215e-01
-4.24094982e-02 -6.90819204e-01 5.30988336e-01 1.35247207e+00
2.17412010e-01 -8.68796468e-01 4.06410456e-01 4.00663227e-01
-1.34614870e-01 -4.07845378e-01 -5.24530470e-01 -7.77782440e-01
-6.11810327e-01 -3.66945833e-01 5.45305908e-01 8.24478149e-01
3.64182033e-02 9.11301896e-02 -6.67508066e-01 2.46985182e-01
6.07407451e-01 4.62298781e-01 5.21433353e-01 -7.57907987e-01
-2.94534624e-01 -2.93335319e-01 -2.87502080e-01 -1.09148359e+00
-7.15845585e-01 -1.95420474e-01 -2.95640022e-01 -1.29839182e+00
2.48419955e-01 -3.24287206e-01 -1.54723063e-01 2.06874264e-03
-5.40244989e-02 6.50473833e-01 1.11469075e-01 5.19315660e-01
-2.59810507e-01 2.99542189e-01 1.69370973e+00 4.05431539e-02
-2.14663908e-01 8.83464888e-02 -8.94546628e-01 6.43314242e-01
8.05565715e-01 1.68472771e-02 -4.56310958e-01 -4.06781316e-01
3.23494017e-01 3.43409151e-01 2.52630860e-01 -1.08347631e+00
4.53660816e-01 -3.96845281e-01 5.97274482e-01 -6.13056719e-01
2.23930225e-01 -9.92110848e-01 5.75266361e-01 3.91874105e-01
-4.82089631e-03 -2.10549995e-01 -2.15755299e-01 1.68770999e-01
-5.32321036e-01 -1.30771399e-02 1.15650630e+00 -1.55229181e-01
-9.56701279e-01 -1.89147727e-03 -3.70133847e-01 -3.17976564e-01
1.14655185e+00 -7.44002163e-01 -7.25483477e-01 -4.36504096e-01
-3.54370296e-01 -2.29874417e-01 6.09389007e-01 1.34449437e-01
8.11936259e-01 -1.33290577e+00 -1.97481230e-01 5.01273036e-01
-5.04432768e-02 -4.76140529e-01 8.03432941e-01 7.41280019e-01
-8.32231462e-01 1.26748934e-01 -5.51794529e-01 -4.80593652e-01
-1.47027898e+00 6.41007185e-01 3.94616663e-01 3.78874749e-01
-8.07386458e-01 3.17604959e-01 9.05846879e-02 4.14727658e-01
1.32333685e-03 -6.76936656e-02 -4.30821091e-01 -2.50984341e-01
7.92038679e-01 5.80207407e-01 -1.05123110e-01 -2.56325603e-01
-1.72969941e-02 1.15497911e+00 2.20262781e-01 -5.66951782e-02
8.50575626e-01 -6.65960252e-01 -1.49305940e-01 3.90502095e-01
1.11427689e+00 2.54302949e-01 -9.14775074e-01 -2.13791907e-01
-6.57599628e-01 -1.01298892e+00 2.90339701e-02 -7.34045684e-01
-1.14929020e+00 9.72625792e-01 9.28536892e-01 6.15315318e-01
2.04272580e+00 6.68988153e-02 9.68789399e-01 2.58548319e-01
5.87814987e-01 -1.23869407e+00 3.92502189e-01 -1.77423283e-01
6.26684248e-01 -8.67303848e-01 -1.40787577e-02 -6.94200516e-01
-7.66280234e-01 1.28421104e+00 7.65518308e-01 1.14480577e-01
4.63844806e-01 4.13894802e-01 4.84170198e-01 9.79437977e-02
-5.51786542e-01 -2.11630151e-01 -1.83797106e-02 7.88031399e-01
2.06813812e-01 -1.84389561e-01 -3.15529019e-01 2.97713667e-01
-3.66610318e-01 1.45361811e-01 8.38491857e-01 5.62993288e-01
-9.27862406e-01 -8.94201338e-01 -6.53245568e-01 3.95814955e-01
-6.11541688e-01 2.69725323e-01 -4.03340273e-02 7.03174055e-01
1.25009701e-01 1.10907495e+00 8.30953717e-02 -6.50518060e-01
5.28709948e-01 -3.94769162e-01 5.11661947e-01 5.83145814e-03
1.60598665e-01 3.85599136e-01 -1.55308709e-01 -4.93472248e-01
-4.97389793e-01 -1.30896688e-01 -9.60697174e-01 -5.57444513e-01
-3.78742367e-01 -1.24093585e-01 4.28152591e-01 4.60354209e-01
-1.19660370e-01 3.31532419e-01 9.19475853e-01 -1.40873253e-01
-1.53268710e-01 -6.95460856e-01 -1.17758954e+00 2.41822168e-01
3.14613521e-01 -4.16712642e-01 -3.27564746e-01 7.84868300e-02] | [10.858039855957031, -2.440796136856079] |
b7b292e8-f46a-40f9-9d05-045d138a214b | mapping-quantum-circuits-to-ibm-qx | 1907.02026 | null | https://arxiv.org/abs/1907.02026v1 | https://arxiv.org/pdf/1907.02026v1.pdf | Mapping Quantum Circuits to IBM QX Architectures Using the Minimal Number of SWAP and H Operations | The recent progress in the physical realization of quantum computers (the first publicly available ones--IBM's QX architectures--have been launched in 2017) has motivated research on automatic methods that aid users in running quantum circuits on them. Here, certain physical constraints given by the architectures which restrict the allowed interactions of the involved qubits have to be satisfied. Thus far, this has been addressed by inserting SWAP and H operations. However, it remains unknown whether existing methods add a minimum number of SWAP and H operations or, if not, how far they are away from that minimum--an NP-complete problem. In this work, we address this by formulating the mapping task as a symbolic optimization problem that is solved using reasoning engines like Boolean satisfiability solvers. By this, we do not only provide a method that maps quantum circuits to IBM's QX architectures with a minimal number of SWAP and H operations, but also show by experimental evaluation that the number of operations added by IBM's heuristic solution exceeds the lower bound by more than 100% on average. An implementation of the proposed methodology is publicly available at http://iic.jku.at/eda/research/ibm_qx_mapping. | ['Alwin Zulehner', 'Lukas Burgholzer', 'Robert Wille'] | 2019-07-03 | null | null | null | null | ['quantum-circuit-mapping'] | ['methodology'] | [ 2.39628389e-01 2.58148819e-01 8.56880099e-02 -3.30266178e-01
-6.46285713e-01 -7.85453379e-01 2.95283586e-01 1.00131594e-01
-2.19035074e-01 9.95427489e-01 -4.18642730e-01 -9.22024667e-01
-2.67521888e-01 -9.83462214e-01 -8.15124512e-01 -6.24490917e-01
-3.29507678e-03 6.71322763e-01 8.00644010e-02 -6.09427392e-01
7.27625847e-01 4.15702432e-01 -1.39597857e+00 3.07780653e-02
7.92771637e-01 8.25292706e-01 -1.10652603e-01 6.52868867e-01
1.64363340e-01 5.03239810e-01 -2.91635692e-01 -1.66081950e-01
5.41583180e-01 -7.60778606e-01 -1.15374017e+00 -5.12552857e-01
3.82328592e-02 -3.81690413e-01 -4.00847465e-01 1.60060358e+00
9.30073187e-02 -5.29104806e-02 2.50425816e-01 -1.50091124e+00
-4.86806214e-01 9.50868487e-01 6.32564053e-02 5.76501563e-02
3.97020042e-01 4.30040896e-01 1.37617445e+00 -7.92612433e-02
8.50603938e-01 9.36176717e-01 7.01876581e-02 5.49544394e-01
-1.36599684e+00 -6.02259338e-01 -5.52698612e-01 5.55156231e-01
-1.74744022e+00 -2.85807610e-01 6.56936824e-01 -1.85153127e-01
1.55207658e+00 5.17130375e-01 5.92005670e-01 4.82393861e-01
4.39026326e-01 4.14831638e-02 1.49267602e+00 -7.89364815e-01
6.54204965e-01 2.00057268e-01 4.71312314e-01 6.98822141e-01
5.29077172e-01 2.95060664e-01 -2.94089764e-01 -5.30954413e-02
5.76252699e-01 -6.01474404e-01 -1.07412692e-02 -1.76530734e-01
-1.10055578e+00 7.29254127e-01 3.63978595e-01 5.79193830e-01
-1.44728914e-01 5.43098986e-01 2.88420469e-01 5.65202236e-01
-3.48682553e-01 9.45183277e-01 -3.30383420e-01 -2.30460882e-01
-6.44283950e-01 3.08486074e-01 1.08592772e+00 1.13563156e+00
1.06351650e+00 -4.60347265e-01 2.32881889e-01 -4.00551856e-01
-6.57741725e-03 5.05983293e-01 -2.47547045e-01 -1.15209031e+00
3.28244269e-01 4.45009202e-01 3.73495638e-01 -6.37830615e-01
-3.16955864e-01 1.57355387e-02 -6.39811158e-01 1.82468425e-02
3.86720181e-01 -1.10899292e-01 -4.85112548e-01 1.67511129e+00
1.53271466e-01 6.94557652e-02 4.89502475e-02 1.05045760e+00
1.96450323e-01 8.21956456e-01 -3.35634887e-01 -3.21228594e-01
1.38690066e+00 -6.28279626e-01 -9.57365513e-01 8.49631801e-02
9.60650146e-01 -6.60147846e-01 5.78725636e-01 6.62404001e-01
-1.12389743e+00 -2.99954891e-01 -1.44127381e+00 -8.93779173e-02
-3.40338767e-01 -2.42001817e-01 1.12877274e+00 1.09685278e+00
-1.16834271e+00 8.18742871e-01 -8.12550604e-01 -2.57086664e-01
-2.41114095e-01 9.78693604e-01 -2.94581622e-01 -1.19879693e-01
-1.46555436e+00 1.18385017e+00 6.92331493e-01 3.02707016e-01
-5.61068714e-01 -1.36627287e-01 -3.85192066e-01 2.32610300e-01
6.92227781e-01 -6.55821741e-01 1.15326297e+00 -5.59275985e-01
-1.89389920e+00 5.53004205e-01 -1.24558788e-02 -5.74362576e-01
1.02529988e-01 4.12847489e-01 -3.81102264e-01 1.69331312e-01
-1.87808707e-01 4.47826058e-01 2.39126295e-01 -7.27607250e-01
-2.92703390e-01 -3.52319270e-01 8.54420662e-01 -1.90323591e-01
1.25888363e-02 5.78300208e-02 -2.34424457e-01 4.15622115e-01
3.19808483e-01 -1.42817736e+00 -4.85741943e-01 -4.57290500e-01
-8.02928925e-01 -2.62409836e-01 1.86634641e-02 -1.62870005e-01
1.06795454e+00 -1.96437323e+00 5.24460852e-01 5.95200121e-01
-4.77938801e-02 3.45553607e-02 6.22951658e-03 7.90318429e-01
8.39379504e-02 2.03930646e-01 -1.83091443e-02 2.20774785e-01
5.94501734e-01 3.72534543e-01 -3.58445823e-01 5.21080434e-01
1.24849141e-01 7.92690098e-01 -8.48086536e-01 -2.82335490e-01
3.88445854e-01 3.76986004e-02 -9.18852568e-01 -6.62165284e-02
-4.07956421e-01 6.01141810e-01 -5.44359684e-01 4.07123178e-01
9.02003288e-01 -1.76270291e-01 6.07776761e-01 -2.07321450e-01
-4.85635459e-01 6.68444872e-01 -1.47013450e+00 1.73252833e+00
-6.24554232e-02 3.68558168e-01 1.24095427e-02 -1.06284428e+00
6.94671988e-01 8.36845487e-02 1.93244636e-01 -9.75039899e-01
4.73932266e-01 5.78337729e-01 4.88146335e-01 -3.62415045e-01
6.24630690e-01 -2.54036099e-01 -2.87368894e-01 5.18595278e-01
-6.20579626e-03 -5.65489292e-01 5.04914880e-01 3.99530679e-02
1.36965609e+00 -2.97435466e-03 2.29075402e-01 -4.54349726e-01
7.51730442e-01 3.45317096e-01 5.50369799e-01 9.73499417e-01
-2.89519578e-01 1.96501352e-02 8.46842825e-01 -1.96818799e-01
-1.23545504e+00 -6.68713450e-01 -2.82959700e-01 4.16600883e-01
5.66294491e-01 -7.96892881e-01 -9.64857101e-01 1.34094013e-03
-2.55829006e-01 1.00258470e+00 -2.35985935e-01 -1.93652421e-01
-5.91714561e-01 -5.73656797e-01 4.24931556e-01 1.71600692e-02
2.91430295e-01 -8.45639825e-01 -5.26401043e-01 1.75998241e-01
-8.79001897e-03 -1.18284273e+00 1.60097733e-01 6.07427716e-01
-7.33780324e-01 -9.68011081e-01 2.28242889e-01 -2.11371198e-01
7.81092286e-01 -1.70942172e-01 7.25103021e-01 1.49089187e-01
-4.39308375e-01 -2.29803741e-01 -3.12585711e-01 1.36658708e-02
-5.07522762e-01 1.26566127e-01 1.24642298e-01 -4.35389847e-01
4.08006519e-01 -4.65038508e-01 -4.09396231e-01 4.80797738e-02
-8.28441560e-01 6.06896989e-02 5.28537214e-01 6.03318989e-01
4.99851733e-01 4.63445514e-01 -4.77046124e-05 -8.25093567e-01
1.94551751e-01 -8.42730403e-02 -1.22369623e+00 2.75148243e-01
-5.78462541e-01 6.27942801e-01 7.77488112e-01 1.51817352e-01
-4.43711609e-01 1.46690398e-01 -1.96874470e-01 -1.06216468e-01
-1.13214597e-01 3.93363863e-01 -2.53744781e-01 -3.25187594e-01
3.86006325e-01 5.29547855e-02 -4.62181360e-01 5.57630062e-02
5.39739132e-01 5.99770367e-01 3.01082432e-01 -6.89911723e-01
9.58945811e-01 3.27059448e-01 6.91068590e-01 -4.95452076e-01
-5.78760743e-01 -1.73361599e-01 -6.77121818e-01 7.41351917e-02
9.29390848e-01 -3.53650570e-01 -1.18727434e+00 -1.12997387e-02
-1.19416237e+00 -9.31724906e-02 -9.75247249e-02 5.17648637e-01
-6.73468947e-01 3.67587984e-01 -5.50714672e-01 -1.07417011e+00
-5.79209961e-02 -1.58381319e+00 7.32052326e-01 1.50898397e-01
-1.30402207e-01 -4.59050894e-01 5.21272607e-02 4.98019934e-01
3.34637940e-01 -1.39259160e-01 1.09649289e+00 -4.86825734e-01
-1.28780830e+00 -1.18567280e-01 -1.20182753e-01 2.07443506e-01
-3.25054318e-01 -9.65579413e-03 -6.95162714e-01 -3.60915214e-01
-1.07874200e-01 -1.23978056e-01 4.01457220e-01 1.19722724e-01
8.77056360e-01 -5.05942218e-02 -2.23111510e-01 4.01200205e-01
1.66059470e+00 2.76859492e-01 8.63407433e-01 2.78050840e-01
1.48681536e-01 3.59507710e-01 6.75865531e-01 3.15267324e-01
1.27731031e-02 9.05518174e-01 6.72081828e-01 5.50731063e-01
3.80344003e-01 -2.26228517e-02 1.44563153e-01 1.11745167e+00
-2.84266442e-01 2.45089224e-03 -8.83062482e-01 8.55090097e-02
-1.73359692e+00 -8.21114600e-01 -4.26574022e-01 2.20311713e+00
6.56527102e-01 4.77761626e-01 -3.65808666e-01 2.96230376e-01
6.32351875e-01 -3.84449512e-01 -3.15546960e-01 -8.96226883e-01
1.38228610e-01 7.63824880e-01 7.66519070e-01 7.62204170e-01
-7.43826270e-01 9.79160607e-01 5.96709633e+00 7.14716434e-01
-1.07035756e+00 1.37413546e-01 1.31830007e-01 7.09601343e-02
-3.62692535e-01 9.37955916e-01 -7.97218978e-01 2.05727234e-01
1.37173831e+00 -1.70304760e-01 9.65984285e-01 5.17563045e-01
1.16962112e-01 -2.15424821e-01 -1.34569204e+00 8.59535873e-01
-3.37903500e-01 -1.34097183e+00 -3.26905817e-01 2.94571847e-01
6.21064901e-01 -1.98753938e-01 -1.71993390e-01 4.34907943e-01
-7.21341581e-04 -1.01769817e+00 8.98012578e-01 3.29078734e-01
6.70933187e-01 -8.14516783e-01 8.61409187e-01 3.56339484e-01
-7.66745627e-01 -1.35770574e-01 -5.42644024e-01 -5.98817229e-01
9.86524448e-02 5.23308516e-01 -7.24326730e-01 8.48144531e-01
2.29833767e-01 -9.33144391e-02 -2.33638197e-01 8.59381437e-01
-5.44866502e-01 4.42528248e-01 -4.76142228e-01 -4.55811650e-01
3.23849261e-01 -7.79894114e-01 3.64152640e-01 7.21971273e-01
4.46393520e-01 5.54857850e-01 -1.59465000e-01 1.36430907e+00
4.52378429e-02 -1.89745948e-01 -5.15316129e-01 -4.12976593e-01
5.91802180e-01 1.12527335e+00 -8.39104116e-01 -1.18083112e-01
-1.85565874e-01 8.87835979e-01 1.42713502e-01 5.62534779e-02
-1.08003366e+00 -6.38762474e-01 4.40371752e-01 -1.05280504e-01
1.58648387e-01 -4.31286752e-01 -2.30468780e-01 -1.12410057e+00
-1.01713929e-02 -7.38100111e-01 -2.38688156e-01 -7.97401845e-01
-8.58396232e-01 4.54895943e-01 -6.61248565e-02 -8.17190468e-01
-2.54820496e-01 -9.01450634e-01 -3.81725490e-01 8.91694188e-01
-1.33453083e+00 -6.02497339e-01 3.14412087e-01 1.54934525e-01
-3.94766510e-01 2.15439335e-01 1.11370206e+00 3.25973332e-01
-7.48966694e-01 3.11819613e-01 2.20346838e-01 -2.13987902e-01
3.14000905e-01 -1.13249135e+00 1.40965402e-01 8.13503861e-01
-9.87855345e-02 8.88842762e-01 1.21553028e+00 -3.85368973e-01
-2.42681360e+00 -3.97303849e-01 1.17120624e+00 -3.22307378e-01
9.09560621e-01 -4.93129790e-01 -5.02518296e-01 7.46599555e-01
3.12025070e-01 -1.64809301e-01 3.66824210e-01 2.37001032e-01
-1.91644683e-01 -1.46279007e-01 -1.05263078e+00 4.53853846e-01
9.12674487e-01 -8.53778541e-01 -5.34522533e-01 6.95903182e-01
5.00439286e-01 -2.82451302e-01 -7.74776936e-01 1.28176108e-01
2.15481162e-01 -1.07275367e+00 6.85119867e-01 -6.15949512e-01
2.86442965e-01 -6.71118975e-01 -3.26569766e-01 -7.93527544e-01
-3.41320178e-03 -9.83336329e-01 6.61621690e-02 8.29168797e-01
3.22245389e-01 -7.53934205e-01 6.43730342e-01 8.47192824e-01
-1.58639014e-01 -4.83585864e-01 -1.39828646e+00 -8.79178107e-01
1.83928445e-01 -5.00148773e-01 6.78935647e-01 9.49122787e-01
5.28185487e-01 3.36519212e-01 -1.37754396e-01 4.89952981e-01
5.68994761e-01 4.77461308e-01 6.87132895e-01 -8.06918085e-01
-5.80480456e-01 -3.91505748e-01 -6.87460124e-01 -7.34546423e-01
2.61049986e-01 -1.02938223e+00 -1.17100015e-01 -1.25383592e+00
2.35547408e-01 -4.93855149e-01 -3.05513114e-01 4.65581477e-01
4.39219832e-01 -4.09965292e-02 2.27757618e-01 3.37322243e-02
-7.04601645e-01 3.12556982e-01 1.14188302e+00 -4.62250896e-02
3.79921421e-02 -3.40647250e-01 -3.97474527e-01 2.27981597e-01
6.79099262e-01 -5.76147676e-01 -5.89434840e-02 -2.29098260e-01
8.94546926e-01 4.68916386e-01 1.67898476e-01 -1.13685107e+00
6.12120211e-01 -2.23598227e-01 -5.76340914e-01 -4.60307509e-01
4.33857501e-01 -7.91827917e-01 5.71004570e-01 8.32548440e-01
-6.27146289e-02 -5.93481362e-02 2.42139660e-02 1.05100989e-01
-7.09096268e-02 -8.50365877e-01 4.95193243e-01 -1.70077626e-02
-5.44804811e-01 -2.99555242e-01 -3.88333589e-01 -4.53779548e-01
1.16069508e+00 1.06357500e-01 -4.89558816e-01 1.45805944e-02
-7.27671325e-01 1.85017809e-01 4.67996746e-01 -1.13993838e-01
5.55941649e-02 -1.05930793e+00 -3.08504879e-01 1.95669681e-01
-3.95851247e-02 -4.55229074e-01 2.84453332e-01 1.05232358e+00
-9.83423471e-01 1.11166739e+00 -5.26893020e-01 -3.10851187e-01
-9.00541186e-01 6.72541022e-01 3.07548016e-01 -2.91250587e-01
-2.92353183e-01 5.36559224e-01 -3.99690062e-01 -4.64596868e-01
-2.75673598e-01 -6.06567979e-01 4.22930986e-01 -5.36721110e-01
2.79802531e-01 2.64833182e-01 2.64061421e-01 -4.52193052e-01
-4.83135879e-01 3.99784535e-01 1.39166996e-01 -2.60916173e-01
1.24825215e+00 1.29189864e-01 -7.51704156e-01 2.24401981e-01
8.86800349e-01 -1.06467545e-01 -2.65211731e-01 1.19086742e-01
1.96730811e-02 -2.03014433e-01 1.44260064e-01 -5.59737742e-01
-5.98148227e-01 9.80467737e-01 4.57280129e-01 5.50471723e-01
1.13266647e+00 -1.40174627e-01 6.40216887e-01 1.12775970e+00
1.06579959e+00 -1.19387043e+00 -6.35315776e-01 7.18557119e-01
2.75667757e-01 -8.56496990e-01 6.69052303e-02 -6.68629706e-01
5.63524365e-02 1.18619847e+00 3.45195532e-01 -2.12428510e-01
3.11961293e-01 2.83901691e-01 -4.56768572e-01 -2.27699712e-01
-7.22740054e-01 -2.74673462e-01 -1.97568879e-01 -2.89553832e-02
1.92357317e-01 4.67062801e-01 -6.70889139e-01 4.46097493e-01
-4.87764508e-01 1.97986946e-01 9.28964913e-01 1.09445930e+00
-4.45924878e-01 -1.63612580e+00 -5.75217068e-01 8.15369189e-02
-7.44009763e-02 -1.86190113e-01 -4.32711363e-01 7.83680856e-01
3.18005085e-01 1.02230918e+00 -1.62958503e-01 -4.44135338e-01
2.78722882e-01 1.00213736e-01 1.09094405e+00 -5.71142256e-01
-5.02677381e-01 -2.48187453e-01 2.45905325e-01 -6.37329698e-01
-2.66663313e-01 -5.18263400e-01 -1.73809516e+00 -7.33352184e-01
-6.75837457e-01 5.00880718e-01 9.04536307e-01 1.11130548e+00
2.29789689e-01 5.25078535e-01 4.28511113e-01 -7.24123240e-01
-1.12969112e+00 -6.00214064e-01 -7.85271585e-01 -1.30441962e-02
-1.47429258e-01 -3.78669232e-01 -4.84674156e-01 -4.08071816e-01] | [5.640480041503906, 4.907327175140381] |
4fa3cee0-73db-4369-a139-c52e1d0a4949 | fine-grained-adversarial-semi-supervised | 2110.05848 | null | https://arxiv.org/abs/2110.05848v1 | https://arxiv.org/pdf/2110.05848v1.pdf | Fine-Grained Adversarial Semi-supervised Learning | In this paper we exploit Semi-Supervised Learning (SSL) to increase the amount of training data to improve the performance of Fine-Grained Visual Categorization (FGVC). This problem has not been investigated in the past in spite of prohibitive annotation costs that FGVC requires. Our approach leverages unlabeled data with an adversarial optimization strategy in which the internal features representation is obtained with a second-order pooling model. This combination allows to back-propagate the information of the parts, represented by second-order pooling, onto unlabeled data in an adversarial training setting. We demonstrate the effectiveness of the combined use by conducting experiments on six state-of-the-art fine-grained datasets, which include Aircrafts, Stanford Cars, CUB-200-2011, Oxford Flowers, Stanford Dogs, and the recent Semi-Supervised iNaturalist-Aves. Experimental results clearly show that our proposed method has better performance than the only previous approach that examined this problem; it also obtained higher classification accuracy with respect to the supervised learning methods with which we compared. | ['Alberto del Bimbo', 'Francesco Turchini', 'Federico Pernici', 'Daniele Mugnai'] | 2021-10-12 | null | null | null | null | ['fine-grained-visual-categorization'] | ['computer-vision'] | [ 1.25166580e-01 1.78078830e-01 -8.60646218e-02 -5.80983639e-01
-5.75156868e-01 -1.01445293e+00 8.98845911e-01 -8.46537650e-02
-5.49864888e-01 9.82464731e-01 -1.05074167e-01 2.60757543e-02
-1.61854140e-02 -7.19956875e-01 -8.63200068e-01 -6.59607232e-01
7.37663656e-02 2.64873683e-01 4.04752672e-01 3.39372287e-04
9.95552540e-02 7.28489280e-01 -1.70492101e+00 5.89215338e-01
7.73847342e-01 1.49948823e+00 -1.54686853e-01 4.37384993e-01
-1.37286246e-01 1.06700695e+00 -5.65825403e-01 -4.45044249e-01
5.32321811e-01 -6.11321293e-02 -1.08135748e+00 3.05762708e-01
9.56766605e-01 -1.94347411e-01 -1.48747265e-01 1.13109398e+00
1.44482076e-01 5.56753017e-02 9.06839848e-01 -1.27385187e+00
-8.94412279e-01 3.90185028e-01 -5.72884142e-01 2.02505812e-01
-2.00410053e-01 1.78154916e-01 8.13882530e-01 -7.84362078e-01
7.88665533e-01 1.32653666e+00 6.51585877e-01 5.15174747e-01
-1.24022090e+00 -7.20981300e-01 2.04014957e-01 2.09101886e-02
-1.44704771e+00 -2.64456600e-01 4.74264354e-01 -6.25383258e-01
7.99219906e-01 9.62118618e-03 1.09293364e-01 8.75288486e-01
-1.32418528e-01 7.27770627e-01 1.67143011e+00 -6.19435906e-01
2.77181745e-01 3.23333472e-01 5.02762675e-01 1.14528453e+00
1.07864872e-01 4.46382672e-01 -1.92653522e-01 -5.85191548e-02
5.49784303e-01 -1.99000850e-01 -1.26072783e-02 -4.62769449e-01
-1.02043462e+00 8.68721902e-01 8.98149133e-01 4.90474522e-01
-1.93043649e-01 -1.41383037e-02 3.68026167e-01 3.99495810e-01
7.82784164e-01 7.21169353e-01 -5.77896237e-01 4.11400437e-01
-1.06817770e+00 -1.59960128e-02 6.85394824e-01 7.57467687e-01
7.95387685e-01 1.30659536e-01 -4.66402501e-01 6.72887504e-01
2.79685825e-01 3.30986649e-01 5.93413115e-01 -8.60407531e-01
3.63049448e-01 8.38125885e-01 7.91604519e-02 -5.89041471e-01
-1.77424133e-01 -3.53703469e-01 -6.97906137e-01 7.48492122e-01
6.45689487e-01 -8.83427933e-02 -1.37372279e+00 1.58835292e+00
2.34392226e-01 1.56886920e-01 9.00248811e-02 9.15895700e-01
9.73274946e-01 2.97639281e-01 4.81770873e-01 2.58291453e-01
1.12370002e+00 -1.38107634e+00 -4.20474023e-01 -1.34344578e-01
3.53969991e-01 -6.36841059e-01 9.40896153e-01 3.88017386e-01
-6.04083121e-01 -9.95673299e-01 -1.38309371e+00 7.89796710e-02
-1.05899036e+00 5.37231922e-01 8.77900720e-01 7.46417105e-01
-8.61534595e-01 9.03605580e-01 -7.00687706e-01 -3.35365802e-01
8.80009711e-01 2.74514377e-01 -7.51983762e-01 8.13660100e-02
-1.03040552e+00 8.00820053e-01 6.52576208e-01 -1.22371532e-01
-1.05030894e+00 -6.92580462e-01 -7.30513692e-01 1.40423581e-01
3.74592036e-01 -4.15002793e-01 9.59306777e-01 -1.21340835e+00
-1.27904415e+00 1.19412148e+00 3.01416188e-01 -6.97114050e-01
5.58730543e-01 -1.34159490e-01 -2.31799528e-01 2.47876257e-01
4.54320274e-02 1.11303353e+00 1.14135337e+00 -1.25042498e+00
-5.69244802e-01 -5.18648565e-01 3.21029872e-01 -1.33897260e-01
-3.50244313e-01 -2.17185721e-01 -1.12125508e-01 -8.22721422e-01
-2.76886672e-01 -9.23366904e-01 -9.07842666e-02 2.01665744e-01
-4.62160051e-01 -2.49496549e-01 9.30620551e-01 -4.28161055e-01
9.31847990e-01 -2.36827254e+00 1.04003973e-01 4.09513004e-02
9.79956612e-02 7.39740670e-01 -8.26964006e-02 2.70357847e-01
-2.10057512e-01 2.99189031e-01 -2.56251007e-01 -5.06328702e-01
1.13992564e-01 1.50351629e-01 -4.73907977e-01 5.22450328e-01
4.65156972e-01 1.09783351e+00 -6.55755699e-01 -6.92268789e-01
5.01927495e-01 1.67103574e-01 -2.50248462e-01 3.44079763e-01
-3.24416220e-01 8.95603076e-02 -4.72701848e-01 8.68440628e-01
7.80561805e-01 -3.21442157e-01 -2.25887135e-01 -3.01074833e-01
-8.46692845e-02 -5.01720011e-01 -8.56708109e-01 1.61975253e+00
-2.66203642e-01 2.92452604e-01 -1.23431101e-01 -9.11076665e-01
8.09466302e-01 4.68189530e-02 -1.17828690e-01 -4.34933662e-01
2.83031344e-01 2.85776239e-03 -2.40066215e-01 -3.09135020e-01
3.21057558e-01 -7.95650482e-02 -3.29216033e-01 1.47404686e-01
7.03050375e-01 -1.20907553e-01 3.57600093e-01 1.72666475e-01
7.51617730e-01 4.53554511e-01 2.78321683e-01 -4.32139665e-01
6.37031794e-01 2.33223140e-01 1.84434503e-01 8.24374914e-01
-5.39667010e-01 5.87842822e-01 3.28992277e-01 -3.28869849e-01
-7.47148991e-01 -1.03593123e+00 -1.81386054e-01 1.08329487e+00
-8.43871757e-02 -1.32117532e-02 -1.01564574e+00 -1.41175377e+00
3.06611627e-01 5.60048044e-01 -1.13142526e+00 -1.97921157e-01
-7.61311948e-02 -5.29478192e-01 9.57945645e-01 7.38249600e-01
9.42687929e-01 -1.19717729e+00 -4.44713652e-01 -1.71811715e-01
1.85746670e-01 -1.12150145e+00 -2.46777922e-01 5.58357358e-01
-7.01683283e-01 -1.19840634e+00 -6.51551783e-01 -7.97189057e-01
6.82357788e-01 -1.19614154e-01 9.71924901e-01 -1.89179868e-01
-7.12229967e-01 1.82944268e-01 -4.13858205e-01 -3.62147748e-01
-3.22330534e-01 1.00310773e-01 -1.87754318e-01 2.66550750e-01
4.44336325e-01 -6.78905770e-02 -2.04009593e-01 2.35966399e-01
-9.55821574e-01 -4.18793082e-01 7.04905033e-01 1.07402110e+00
6.68827474e-01 1.52708307e-01 6.34268641e-01 -1.21909928e+00
2.35390171e-01 -3.04573059e-01 -8.25263619e-01 4.86103088e-01
-3.45785350e-01 2.30547115e-01 8.94189298e-01 -3.36306989e-01
-1.36935210e+00 2.96823382e-01 2.70194281e-02 -7.78713107e-01
-6.68391764e-01 1.53610438e-01 -4.41957675e-02 -4.98124301e-01
8.36599290e-01 -2.16365635e-01 -2.54818201e-01 -6.26278758e-01
6.29045248e-01 7.85265207e-01 5.98607123e-01 -3.27881068e-01
8.21389079e-01 6.55018628e-01 5.37325703e-02 -6.04475677e-01
-1.15646267e+00 -4.84398037e-01 -1.11542225e+00 1.25401774e-02
1.12611198e+00 -8.64643633e-01 -4.24986213e-01 5.77870488e-01
-7.69835234e-01 -2.22103819e-01 -5.57218969e-01 2.54585236e-01
-5.69188774e-01 3.48729968e-01 -6.19325936e-01 -5.63376844e-01
-3.02474916e-01 -9.20577705e-01 1.13226175e+00 2.23865509e-01
6.39396757e-02 -9.45860982e-01 -1.63892284e-01 4.27255869e-01
4.29245800e-01 4.07928795e-01 8.90835822e-01 -7.62019336e-01
-4.74478036e-01 -1.37055457e-01 -5.52251339e-01 7.72989750e-01
1.57421663e-01 -4.82437611e-02 -1.40235293e+00 -2.48578444e-01
-4.86953706e-01 -9.53945696e-01 1.27710330e+00 4.23113331e-02
1.36066997e+00 7.09692463e-02 -3.57875258e-01 5.95315874e-01
1.51041400e+00 -1.23702340e-01 4.72268641e-01 2.80445695e-01
5.98077714e-01 6.49489343e-01 1.04967570e+00 1.14785627e-01
-2.07321033e-01 5.19521415e-01 6.40623152e-01 -3.58297378e-01
-6.81886196e-01 -3.29121619e-01 -6.97922111e-02 -8.92289355e-03
-1.11143254e-02 -2.73793906e-01 -4.24986005e-01 4.20964479e-01
-1.70852411e+00 -8.42325270e-01 2.62725383e-01 1.87926149e+00
6.08797133e-01 1.52248636e-01 -1.62502289e-01 1.89790726e-01
5.40612280e-01 1.29567191e-01 -5.23074806e-01 -4.77307081e-01
-1.19084254e-01 4.61550951e-01 7.17860460e-01 1.91563949e-01
-1.90644944e+00 1.35238397e+00 6.24681282e+00 1.08375502e+00
-1.12097955e+00 2.23952651e-01 7.00973094e-01 2.52884835e-01
3.71457130e-01 -5.00410795e-01 -7.92261243e-01 2.46798873e-01
6.99839890e-01 4.07644391e-01 4.05241519e-01 1.09740126e+00
-4.08398747e-01 6.56437278e-02 -9.40936148e-01 5.21999359e-01
1.75384730e-01 -1.14103758e+00 -2.74379514e-02 -2.89630234e-01
9.37705159e-01 2.45146845e-02 2.00843915e-01 5.06330192e-01
4.72130388e-01 -1.05282760e+00 7.66970038e-01 4.34894145e-01
1.01675129e+00 -5.67344546e-01 8.62608254e-01 2.88791209e-01
-9.81429040e-01 -1.07260227e-01 -5.62114835e-01 1.28608808e-01
-1.92819998e-01 3.58059078e-01 -8.73835385e-01 7.13456035e-01
8.82553816e-01 4.36258048e-01 -1.02639437e+00 8.28483105e-01
-4.19762343e-01 6.97042704e-01 -1.56284899e-01 1.85367674e-01
5.98375261e-01 1.73118204e-01 1.49297610e-01 1.29481685e+00
-1.38757363e-01 -2.53273576e-01 2.96783179e-01 8.14260423e-01
-2.54571795e-01 -8.68526846e-03 -6.19970083e-01 -3.61452341e-01
9.65513885e-02 1.51482248e+00 -9.61680353e-01 -4.41268563e-01
-3.73201936e-01 1.09253204e+00 7.34597147e-01 2.47892305e-01
-9.09719765e-01 -5.45158625e-01 3.32618684e-01 -5.74085712e-02
7.54707873e-01 1.83708638e-01 -1.35866359e-01 -1.14225745e+00
-2.19122112e-01 -6.18796289e-01 5.71078718e-01 -7.10031271e-01
-1.79850602e+00 1.04616952e+00 4.24388796e-02 -1.16041744e+00
-2.30715007e-01 -8.60547960e-01 -2.20151618e-01 8.00989926e-01
-1.68601394e+00 -1.77315235e+00 -3.71415734e-01 6.97635770e-01
4.99584705e-01 -3.35511774e-01 1.03460240e+00 1.36647582e-01
-3.35446894e-01 7.07034528e-01 -1.36306867e-01 9.31964591e-02
9.08388376e-01 -1.49933374e+00 1.64001018e-01 8.77115011e-01
1.31706178e-01 4.17944163e-01 3.01396340e-01 -6.25156820e-01
-7.31493890e-01 -1.56403971e+00 6.50064945e-01 -4.08864051e-01
7.34724045e-01 -3.90484214e-01 -7.22041070e-01 7.18622983e-01
3.05761129e-01 7.34425783e-01 6.39747322e-01 -3.34040761e-01
-6.56270862e-01 5.46455383e-02 -1.68167424e+00 1.37842834e-01
9.06724513e-01 -5.49090624e-01 -9.28771138e-01 1.78773746e-01
7.63211429e-01 -1.48318544e-01 -7.46032059e-01 5.50262094e-01
3.63709688e-01 -8.54642749e-01 9.50976908e-01 -9.73955214e-01
3.45632732e-01 -5.01782596e-01 -3.29009503e-01 -1.42551398e+00
-4.88823891e-01 1.73576251e-02 1.01188958e-01 1.41183650e+00
3.45994741e-01 -7.18621016e-01 7.37568915e-01 2.88832426e-01
-1.65900975e-01 -4.64450330e-01 -6.58146203e-01 -8.15412283e-01
1.10813618e-01 1.71740532e-01 4.62054133e-01 8.69633973e-01
-3.86212677e-01 1.18769556e-01 -2.67333716e-01 1.45293415e-01
7.64281213e-01 4.84920651e-01 5.55782139e-01 -1.30816364e+00
-4.19921875e-01 -6.23497404e-02 -5.96583664e-01 -5.78948081e-01
5.77358186e-01 -1.00892496e+00 9.15011577e-03 -1.19467306e+00
1.14534080e-01 -5.30400932e-01 -4.71695870e-01 8.56542885e-01
-1.45248100e-01 9.13759589e-01 3.06665838e-01 2.62159090e-02
-6.78405821e-01 3.37620884e-01 1.23077238e+00 -3.46364826e-01
1.83846340e-01 -1.61466733e-01 -5.91119766e-01 8.75732839e-01
6.77687585e-01 -3.72104645e-01 -3.29412729e-01 -3.92948203e-02
-6.72438323e-01 -4.28935260e-01 5.50307393e-01 -1.00776005e+00
-8.45755413e-02 1.55346125e-01 8.23309064e-01 -3.94621849e-01
3.39368194e-01 -9.25931156e-01 -9.79284570e-02 3.34970504e-01
-5.14672995e-01 -5.06210744e-01 4.91501361e-01 5.42829394e-01
-4.45052594e-01 -4.15118963e-01 1.03386867e+00 -1.47934839e-01
-1.14253783e+00 2.71086425e-01 -1.22065611e-01 -1.29421130e-01
1.39077663e+00 4.89793718e-02 -6.26817703e-01 2.18159571e-01
-9.81367767e-01 -6.81767836e-02 4.40830588e-01 5.30884981e-01
1.14955015e-01 -1.18072605e+00 -4.51974362e-01 3.56846005e-01
4.23478812e-01 -4.29338425e-01 3.12880069e-01 3.14946026e-01
-4.05353993e-01 6.42211080e-01 -6.96274221e-01 -3.83434355e-01
-1.27160716e+00 1.13640773e+00 1.66878030e-01 -5.54301381e-01
-1.55941471e-01 9.73763227e-01 4.15748596e-01 -5.63646376e-01
3.46721858e-01 -1.78905845e-01 -4.92130965e-01 7.80558735e-02
4.22679335e-01 2.67919749e-01 3.16890657e-01 -5.89709461e-01
-3.60045224e-01 4.17396128e-01 -1.47658467e-01 2.30659470e-01
1.08152235e+00 -1.06229922e-02 1.05601542e-01 2.03341201e-01
1.29799271e+00 -1.80335656e-01 -1.29409587e+00 -7.18022883e-02
-1.22048676e-01 -5.66378057e-01 2.77411908e-01 -1.23512578e+00
-1.11259711e+00 9.46871459e-01 8.17447245e-01 2.62824357e-01
1.08641827e+00 7.64837191e-02 2.07566917e-01 2.75685161e-01
5.57984948e-01 -8.04885924e-01 -4.32187095e-02 1.95508778e-01
6.94457591e-01 -1.45123684e+00 -2.11989328e-01 -7.88411021e-01
-6.55759037e-01 9.78480637e-01 8.00197303e-01 -4.36060518e-01
7.14041173e-01 2.90016562e-01 9.55031514e-02 -8.04397166e-02
-5.38362384e-01 -5.51483512e-01 4.61438864e-01 5.65438092e-01
2.26226032e-01 1.43188104e-01 -7.06754103e-02 6.11238480e-01
1.23015970e-01 2.93130636e-01 1.25796288e-01 9.43680942e-01
-3.00358027e-01 -1.00922537e+00 -2.75357634e-01 4.79249239e-01
-7.28061914e-01 -2.44172346e-02 -6.55155122e-01 8.78218532e-01
4.50639725e-01 9.59144711e-01 -5.68231978e-02 -2.37391695e-01
3.46061975e-01 2.02638745e-01 6.38845682e-01 -7.17524230e-01
-8.90544593e-01 -2.35499948e-01 1.59474075e-01 -7.46901333e-01
-6.38651311e-01 -3.06308389e-01 -1.00578821e+00 8.13454911e-02
-4.02929336e-01 1.46865889e-01 5.64552665e-01 8.82278800e-01
3.25933248e-01 7.44740665e-01 4.51095194e-01 -9.19413388e-01
-7.65411079e-01 -1.11700118e+00 -7.59587705e-01 6.27414763e-01
2.37500131e-01 -9.82476950e-01 -4.61403489e-01 2.62216002e-01] | [9.618657112121582, 2.0863378047943115] |
3a1eb399-4ca5-487e-a674-afc66af0e18a | semi-supervised-semantic-segmentation-with-9 | 2304.11539 | null | https://arxiv.org/abs/2304.11539v1 | https://arxiv.org/pdf/2304.11539v1.pdf | Semi-Supervised Semantic Segmentation With Region Relevance | Semi-supervised semantic segmentation aims to learn from a small amount of labeled data and plenty of unlabeled ones for the segmentation task. The most common approach is to generate pseudo-labels for unlabeled images to augment the training data. However, the noisy pseudo-labels will lead to cumulative classification errors and aggravate the local inconsistency in prediction. This paper proposes a Region Relevance Network (RRN) to alleviate the problem mentioned above. Specifically, we first introduce a local pseudo-label filtering module that leverages discriminator networks to assess the accuracy of the pseudo-label at the region level. A local selection loss is proposed to mitigate the negative impact of wrong pseudo-labels in consistency regularization training. In addition, we propose a dynamic region-loss correction module, which takes the merit of network diversity to further rate the reliability of pseudo-labels and correct the convergence direction of the segmentation network with a dynamic region loss. Extensive experiments are conducted on PASCAL VOC 2012 and Cityscapes datasets with varying amounts of labeled data, demonstrating that our proposed approach achieves state-of-the-art performance compared to current counterparts. | ['Yazhou Yao', 'Qiong Wang', 'Tao Chen', 'Rui Chen'] | 2023-04-23 | null | null | null | null | ['semi-supervised-semantic-segmentation', 'pseudo-label'] | ['computer-vision', 'miscellaneous'] | [ 3.84021550e-01 2.66849458e-01 -3.74505758e-01 -9.05006230e-01
-1.04643512e+00 -6.05302095e-01 3.09643596e-01 -3.28442529e-02
-5.43079674e-01 8.26357961e-01 -3.01476657e-01 -8.15855414e-02
2.35598207e-01 -4.33460027e-01 -7.38147855e-01 -7.66414165e-01
4.96601760e-01 2.57102311e-01 4.19191211e-01 2.29019061e-01
1.47684619e-01 1.67532757e-01 -1.36692882e+00 1.39201522e-01
1.49306202e+00 1.21920931e+00 2.82788873e-01 9.18809697e-02
-8.72561634e-02 7.68276095e-01 -5.06510496e-01 -2.98282266e-01
5.65835059e-01 -3.09275866e-01 -7.06272125e-01 5.65471947e-01
6.19618893e-01 -1.38810232e-01 1.96752146e-01 1.35421038e+00
2.77217209e-01 1.98279381e-01 6.25295937e-01 -1.12637699e+00
-3.93505692e-01 5.26008546e-01 -7.84285903e-01 -1.33472309e-01
-2.86193937e-01 1.52711421e-01 9.90099669e-01 -8.96864712e-01
5.57473421e-01 1.00720632e+00 7.97924876e-01 6.05975211e-01
-1.16264844e+00 -7.60040104e-01 4.52316254e-01 -2.14831159e-01
-1.38629556e+00 -1.69843644e-01 7.98827529e-01 -3.27412277e-01
1.00197166e-01 -4.79253903e-02 3.47798526e-01 9.12613392e-01
-1.37675405e-01 7.23136306e-01 1.19387639e+00 -3.54427636e-01
3.80113512e-01 3.51108342e-01 2.00666189e-01 6.80966377e-01
1.58972934e-01 7.36777261e-02 -2.53702551e-01 2.25074232e-01
7.20052481e-01 -9.32728946e-02 -1.66403815e-01 -6.02014422e-01
-7.65531123e-01 6.71994627e-01 7.12927699e-01 2.91934814e-02
-1.56135291e-01 -1.51215181e-01 3.01896036e-01 -3.09255682e-02
8.40808153e-01 2.97826618e-01 -6.23955488e-01 3.98770630e-01
-1.12396324e+00 -4.82367612e-02 4.34915781e-01 9.06292617e-01
9.06076252e-01 1.43188452e-02 -3.61449808e-01 1.16740751e+00
4.23267215e-01 1.64507210e-01 2.99816549e-01 -9.85130906e-01
6.62071168e-01 8.46145332e-01 1.03825174e-01 -7.24904597e-01
-2.53136814e-01 -8.73897493e-01 -5.91075420e-01 2.08822548e-01
6.19229853e-01 -1.30128577e-01 -1.29644191e+00 1.78524601e+00
6.37572169e-01 4.47082311e-01 -1.64568543e-01 1.31069374e+00
6.47132576e-01 3.05876195e-01 3.75122219e-01 4.70612291e-03
7.88428128e-01 -1.39637423e+00 -4.52788144e-01 -5.15273273e-01
6.62877738e-01 -4.97251600e-01 1.15856302e+00 3.59386690e-02
-7.79583812e-01 -7.33400583e-01 -1.05374539e+00 6.75787851e-02
-1.39141187e-01 4.87571687e-01 3.24582458e-01 5.86750686e-01
-7.43546009e-01 6.10979140e-01 -6.79638207e-01 4.47158143e-02
6.41840577e-01 1.75089210e-01 -3.59659828e-02 -8.41815323e-02
-1.02935410e+00 5.19648790e-01 6.28032625e-01 3.20478380e-01
-6.53754771e-01 -7.86873937e-01 -8.07452798e-01 -2.01665252e-01
6.17648244e-01 -1.03543080e-01 1.04142010e+00 -1.49444664e+00
-1.24614370e+00 9.80745316e-01 9.59260240e-02 -3.78303617e-01
7.84578323e-01 -5.70664965e-02 -2.36608043e-01 1.02045760e-01
4.96363938e-01 1.07908297e+00 7.74974704e-01 -1.63107467e+00
-7.78047144e-01 -4.25124198e-01 -1.14009872e-01 4.16099727e-01
-2.50991248e-02 -4.93969500e-01 -6.70862794e-01 -8.27708542e-01
6.09125733e-01 -1.06150079e+00 -3.95402223e-01 1.58281007e-03
-5.79724133e-01 -6.64886683e-02 7.15028226e-01 -5.06238222e-01
9.48429346e-01 -2.15426373e+00 -2.63517350e-01 3.20792913e-01
1.72812380e-02 3.37147355e-01 -2.19800040e-01 -4.45625395e-01
3.31185455e-03 7.91570246e-02 -6.70947790e-01 -6.33181155e-01
-4.12135899e-01 2.40509033e-01 -8.64232183e-02 5.84413052e-01
4.99944806e-01 7.75204122e-01 -8.72239351e-01 -6.63125396e-01
2.17078313e-01 4.41077620e-01 -3.60451400e-01 8.77854005e-02
-4.06563640e-01 1.01162064e+00 -5.79991162e-01 8.49978149e-01
9.01030660e-01 -4.48654205e-01 4.89246733e-02 -9.69682634e-02
5.11189401e-02 2.11048320e-01 -1.26746500e+00 1.68651283e+00
-2.70464331e-01 1.00767024e-01 3.12646955e-01 -1.03736067e+00
9.55309808e-01 5.55864759e-02 3.96169573e-01 -6.77807033e-01
1.86883643e-01 3.96407813e-01 -3.63176495e-01 -1.03639714e-01
5.76381147e-01 2.01138817e-02 5.51056266e-02 3.20878655e-01
-4.04190570e-02 6.57754717e-03 3.20241712e-02 -7.40697756e-02
5.92473388e-01 5.59928536e-01 -1.13953859e-01 -3.46334487e-01
5.46302021e-01 2.10470960e-01 1.06460452e+00 6.45837784e-01
-6.46272123e-01 8.91560018e-01 2.41740718e-01 -3.29651415e-01
-9.42428768e-01 -8.80145252e-01 -3.38391393e-01 1.20226693e+00
5.01231909e-01 2.93871880e-01 -9.33844388e-01 -1.31037796e+00
-3.80312987e-02 6.74626052e-01 -5.57488322e-01 -1.02593683e-01
-3.88917983e-01 -8.12890768e-01 5.08624077e-01 6.49252653e-01
7.65328348e-01 -9.56959665e-01 -3.70595813e-01 5.11050001e-02
-2.88584650e-01 -1.18209958e+00 -6.18958831e-01 4.37974721e-01
-1.03359962e+00 -1.02918506e+00 -7.01619446e-01 -1.01540732e+00
1.16034508e+00 2.78850168e-01 1.00131607e+00 1.57699827e-02
-1.28129810e-01 -2.42050916e-01 -4.06349629e-01 1.63103640e-02
-4.29063380e-01 1.83918506e-01 -2.92967200e-01 1.62137792e-01
1.85496554e-01 -1.37484968e-01 -6.71144903e-01 6.72030747e-01
-8.79749894e-01 -2.63831671e-02 4.21795100e-01 1.00660729e+00
1.15047693e+00 3.14968522e-03 9.23075616e-01 -1.36088777e+00
1.04793891e-01 -4.28255081e-01 -7.91533470e-01 3.73939812e-01
-9.45265949e-01 8.20356235e-03 7.74390757e-01 -4.51630950e-01
-1.40340006e+00 4.39781696e-01 2.57400125e-02 -4.47671056e-01
-2.12323248e-01 2.59219319e-01 -1.84259027e-01 -1.45870566e-01
5.95756173e-01 -9.63584557e-02 -1.57722577e-01 -3.61711591e-01
5.23298979e-01 5.46742976e-01 5.50528407e-01 -4.73284900e-01
5.20198882e-01 5.30631781e-01 -1.50078565e-01 -1.95808738e-01
-1.42660904e+00 -7.33710408e-01 -7.01399326e-01 -2.95321524e-01
7.12792575e-01 -1.17509294e+00 -4.19892557e-02 4.91826683e-01
-6.70684159e-01 -4.91666943e-01 -4.17654544e-01 2.34280497e-01
-2.74955541e-01 3.28145087e-01 -5.07481039e-01 -6.99804425e-01
-2.17889741e-01 -1.20504642e+00 1.13477635e+00 5.56392729e-01
2.17989400e-01 -9.19232905e-01 -2.47210935e-01 6.91511095e-01
8.21035579e-02 2.44289204e-01 4.09260869e-01 -6.95588768e-01
-5.46751142e-01 -1.56779587e-01 -5.81764162e-01 7.02614725e-01
4.89013419e-02 -2.56246239e-01 -1.16139328e+00 -2.29052499e-01
4.03237604e-02 -6.43757045e-01 9.96492922e-01 5.22681773e-01
1.25997341e+00 -4.88477349e-02 -1.85428515e-01 6.31929934e-01
1.47388971e+00 3.44663002e-02 3.02361667e-01 2.39449635e-01
9.80116367e-01 7.92510033e-01 1.11589181e+00 2.65477747e-01
3.93113583e-01 4.43800718e-01 4.73779976e-01 -2.02635199e-01
-2.83821642e-01 -2.71945268e-01 -1.81303009e-01 5.73966861e-01
3.63500446e-01 -2.01009929e-01 -6.67833924e-01 4.96279299e-01
-1.83958554e+00 -3.37745398e-01 -1.53096959e-01 2.29611564e+00
9.60883498e-01 2.87248850e-01 -1.38408646e-01 -1.01823181e-01
1.08539200e+00 1.14319980e-01 -9.30783570e-01 7.04632252e-02
-2.26839453e-01 -1.12229027e-01 8.22247207e-01 4.18696970e-01
-1.41464329e+00 1.20895123e+00 5.56076908e+00 9.37173963e-01
-1.25886774e+00 2.11919725e-01 1.24098933e+00 1.95481196e-01
-1.85880095e-01 -6.96425736e-02 -7.98347771e-01 5.01349151e-01
5.03500342e-01 5.43679714e-01 1.97211534e-01 1.05579805e+00
2.88880408e-01 -3.56584340e-01 -7.39397705e-01 6.21989787e-01
-5.58432341e-02 -8.77733231e-01 -2.08487064e-01 -2.35082313e-01
1.27283013e+00 1.35591045e-01 7.65933767e-02 1.45887181e-01
2.33642131e-01 -7.54848838e-01 8.90732288e-01 2.56130338e-01
8.77461195e-01 -7.06646919e-01 7.49748409e-01 3.04225057e-01
-1.19765711e+00 -1.36524469e-01 -2.92987257e-01 3.23862076e-01
1.02775864e-01 7.88437665e-01 -8.52715194e-01 5.19331694e-01
6.22950077e-01 7.00822234e-01 -6.90425098e-01 9.43793833e-01
-5.04955471e-01 7.16879189e-01 -2.41474107e-01 3.49686056e-01
3.97921145e-01 -3.56297195e-01 2.25622624e-01 9.29397404e-01
-1.16779141e-01 -1.81685135e-01 4.07220423e-01 1.06608260e+00
-2.58352786e-01 1.67260647e-01 -3.87675054e-02 3.37955505e-01
6.40727401e-01 1.25409555e+00 -1.11039853e+00 -2.34363392e-01
-1.63586482e-01 9.47483361e-01 4.39797759e-01 4.71630573e-01
-9.19240594e-01 2.71021407e-02 2.39159260e-03 -1.67567469e-02
1.43756092e-01 1.86872050e-01 -8.60124290e-01 -1.06972241e+00
2.25589871e-01 -5.29457927e-01 2.76263893e-01 -5.18003285e-01
-1.27071154e+00 4.52779472e-01 -3.18533778e-01 -1.25795424e+00
1.61341447e-02 -1.47257447e-01 -4.25795287e-01 8.02855015e-01
-1.72752368e+00 -1.16676259e+00 -3.88715267e-01 1.46013588e-01
7.72944212e-01 1.64682657e-01 2.81432182e-01 4.28469360e-01
-7.83675373e-01 7.77408779e-01 1.56682134e-01 8.39902908e-02
9.04982984e-01 -1.19822228e+00 2.49316335e-01 9.41778898e-01
-9.99096483e-02 2.97802567e-01 3.88933450e-01 -7.97394991e-01
-4.54193860e-01 -1.66488469e+00 6.21581078e-01 -1.46972388e-01
2.52119601e-01 -1.40828878e-01 -1.02383900e+00 4.95274305e-01
-4.15624052e-01 4.77521509e-01 3.62400740e-01 -2.48316213e-01
-3.83345157e-01 -1.83720529e-01 -1.50843179e+00 4.37651157e-01
9.00596559e-01 -3.25854003e-01 -2.12507099e-01 2.37735286e-01
8.46821308e-01 -5.13633370e-01 -5.34468412e-01 8.10127079e-01
3.96206915e-01 -9.57778573e-01 7.50299931e-01 -2.62690466e-02
3.55683595e-01 -5.14022768e-01 9.71636400e-02 -1.15091825e+00
-6.02728724e-02 4.67213010e-03 3.47742707e-01 1.45772922e+00
6.35984063e-01 -5.13365865e-01 1.20583916e+00 7.67400384e-01
-2.75139630e-01 -8.58361959e-01 -9.51978683e-01 -6.75221741e-01
-1.05428532e-01 -2.81236559e-01 2.38410428e-01 1.03092337e+00
-5.36271155e-01 2.44106770e-01 -2.36808062e-01 1.90719515e-01
8.17526102e-01 6.51747659e-02 3.66237938e-01 -1.15387118e+00
-8.25504586e-02 -2.44307294e-01 1.77664738e-02 -1.06801605e+00
3.82426143e-01 -9.43405628e-01 5.87195098e-01 -1.24380648e+00
1.02808446e-01 -1.06452060e+00 -4.36131954e-01 5.08177757e-01
-4.45260108e-01 5.80544889e-01 -7.10992664e-02 4.82154936e-01
-9.81938004e-01 6.19133770e-01 1.51615989e+00 -6.26118034e-02
-4.33933079e-01 1.56780034e-01 -5.42102456e-01 7.67172098e-01
8.95335555e-01 -6.61711156e-01 -6.07774794e-01 -3.37213337e-01
-1.83727905e-01 -1.71422899e-01 3.41039121e-01 -9.14686620e-01
7.10473806e-02 -1.53665677e-01 3.22054654e-01 -6.37664497e-01
-4.49529238e-04 -8.44072998e-01 -2.61915952e-01 1.21585228e-01
-6.18107438e-01 -3.88915449e-01 -2.84135938e-02 7.74967432e-01
-2.19123840e-01 -3.44786227e-01 1.18585265e+00 -9.28654820e-02
-7.66625881e-01 2.15000942e-01 1.98167741e-01 3.64223778e-01
1.01825655e+00 -3.63547802e-01 -1.02465160e-01 -8.75825435e-03
-6.19669080e-01 5.73658764e-01 6.75547957e-01 3.47503453e-01
3.43280971e-01 -1.05711043e+00 -4.03668612e-01 1.94793597e-01
2.82995969e-01 5.42298853e-01 2.48008087e-01 5.71655452e-01
-4.33186024e-01 1.26009464e-01 9.27162021e-02 -7.82484829e-01
-8.78255069e-01 2.51490742e-01 3.44045907e-01 -3.51166159e-01
-5.31711280e-01 1.14103913e+00 1.72124937e-01 -8.26099098e-01
5.59626460e-01 -2.29296505e-01 -9.97611061e-02 -7.32658654e-02
1.35773823e-01 2.87500113e-01 2.64416207e-02 -7.34410524e-01
-3.05628419e-01 4.60962653e-01 -1.28755212e-01 1.45657305e-02
1.00739777e+00 -5.43430448e-01 9.70022604e-02 4.72091973e-01
9.65355635e-01 -3.26114297e-01 -1.97337997e+00 -2.40250692e-01
3.35501470e-02 -2.91965276e-01 6.48330674e-02 -1.09407520e+00
-1.40623248e+00 5.59368968e-01 7.59327650e-01 -2.64892966e-01
1.06939709e+00 -1.07802376e-01 7.57021189e-01 2.13629622e-02
3.81051570e-01 -1.55205429e+00 4.17868942e-02 3.10333937e-01
2.61265844e-01 -1.64634466e+00 -2.30130583e-01 -7.68066943e-01
-9.63747144e-01 6.53154552e-01 8.44372690e-01 -1.71211451e-01
4.67256069e-01 -3.97978202e-02 4.39770550e-01 1.43718004e-01
-2.45195284e-01 -7.82324895e-02 2.96975225e-01 3.08841228e-01
2.85112709e-01 6.17370009e-02 -4.64075416e-01 5.41955471e-01
2.65262753e-01 -2.04920443e-03 1.54389068e-01 8.46108079e-01
-4.15214509e-01 -9.49288487e-01 -2.88759083e-01 3.43358010e-01
-5.99461019e-01 -1.04104551e-02 -2.19517544e-01 4.44477946e-01
3.99657786e-01 1.04801106e+00 -4.79169898e-02 -2.60569930e-01
1.29837440e-02 1.29015252e-01 7.04382136e-02 -6.35378540e-01
-5.94015896e-01 2.85759807e-01 -1.61856398e-01 -5.44995368e-01
-6.75717711e-01 -3.49650472e-01 -1.48432565e+00 3.14366072e-01
-6.92221045e-01 8.06337297e-02 6.93634331e-01 8.72553170e-01
2.59524763e-01 5.69897771e-01 7.28916228e-01 -6.39552951e-01
-7.52184987e-01 -9.19997811e-01 -7.57835984e-01 6.55740738e-01
1.95211947e-01 -6.53512716e-01 -4.93666053e-01 1.14774972e-01] | [9.55556869506836, 1.1112116575241089] |
bb31ff74-7e3d-4f55-a42e-d6b2659ba843 | brain-morphometry-estimation-from-hours-to | null | null | https://doi.org/10.3389/fneur.2020.00244 | https://www.frontiersin.org/articles/10.3389/fneur.2020.00244/pdf | Brain Morphometry Estimation: From Hours to Seconds Using Deep Learning | Motivation: Brain morphometry from magnetic resonance imaging (MRI) is a promising neuroimaging biomarker for the non-invasive diagnosis and monitoring of neurodegenerative and neurological disorders. Current tools for brain morphometry often come with a high computational burden, making them hard to use in clinical routine, where time is often an issue. We propose a deep learning-based approach to predict the volumes of anatomically delineated subcortical regions of interest (ROI), and mean thicknesses and curvatures of cortical parcellations directly from T1-weighted MRI. Advantages are the timely availability of results while maintaining a clinically relevant accuracy.
Materials and Methods: An anonymized dataset of 574 subjects (443 healthy controls and 131 patients with epilepsy) was used for the supervised training of a convolutional neural network (CNN). A silver-standard ground truth was generated with FreeSurfer 6.0.
Results: The CNN predicts a total of 165 morphometric measures directly from raw MR images. Analysis of the results using intraclass correlation coefficients showed, in general, good correlation with FreeSurfer generated ground truth data, with some of the regions nearly reaching human inter-rater performance (ICC > 0.75). Cortical thicknesses predicted by the CNN showed cross-sectional annual age-related gray matter atrophy rates both globally (thickness change of −0.004 mm/year) and regionally in agreement with the literature. A statistical test to dichotomize patients with epilepsy from healthy controls revealed similar effect sizes for structures affecting all subtypes as reported in a large-scale epilepsy study.
Conclusions: We demonstrate the general feasibility of using deep learning to estimate human brain morphometry directly from T1-weighted MRI within seconds. A comparison of the results to other publications shows accuracies of comparable magnitudes for the subcortical volumes and cortical thicknesses. | ['Christian Rummel', 'Mauricio Reyes', 'Roland Wiest', 'Yannick Suter', 'Michael Rebsamen'] | 2020-04-08 | null | null | null | null | ['brain-morphometry'] | ['medical'] | [-1.63162693e-01 1.51275799e-01 9.07504484e-02 -6.43821120e-01
-7.71241069e-01 -8.58465880e-02 1.60662025e-01 1.50960237e-01
-7.73051083e-01 9.37083066e-01 2.77676344e-01 -1.97481155e-01
-5.28251082e-02 -7.11783409e-01 -5.94745576e-01 -6.31829023e-01
-8.45983624e-01 9.23097908e-01 -4.05930588e-03 1.35629028e-01
2.19906643e-01 4.27861750e-01 -7.99159646e-01 -4.69909348e-02
9.62796926e-01 1.01012409e+00 3.32559615e-01 3.08623403e-01
3.11162304e-02 1.51437789e-01 -2.83267945e-01 -1.90885440e-01
3.56041074e-01 -1.11190692e-01 -8.54860544e-01 -4.37106304e-02
7.43109941e-01 -6.43213511e-01 -8.26802999e-02 9.29983735e-01
6.74377978e-01 -2.58834988e-01 6.81214392e-01 -4.28805202e-01
-7.86803961e-01 6.58631027e-01 -6.51521027e-01 6.34168506e-01
-1.66825771e-01 4.32552665e-01 4.68580842e-01 -6.63173854e-01
7.54573405e-01 7.36543357e-01 9.73937809e-01 6.37084424e-01
-1.33075500e+00 -9.18660879e-01 -3.43298495e-01 2.39188761e-01
-1.34534538e+00 -4.75569189e-01 7.43789375e-02 -1.02585459e+00
1.08040833e+00 7.23025238e-04 1.27070737e+00 4.70076829e-01
8.87091577e-01 -5.79340719e-02 1.51626873e+00 1.40634447e-01
3.03084403e-01 -4.95795131e-01 2.76623201e-02 7.45201170e-01
4.97764230e-01 -2.59176567e-02 3.12453769e-02 -2.10435957e-01
9.35230315e-01 -2.46439368e-01 -2.90572971e-01 -1.83070973e-01
-1.43247807e+00 7.15585351e-01 6.50069892e-01 5.98376334e-01
-4.64307725e-01 1.02756418e-01 6.28220916e-01 2.19582304e-01
9.17018712e-01 1.48896053e-01 -5.26263893e-01 1.66683987e-01
-1.51498783e+00 2.37798616e-01 7.20121041e-02 2.45972827e-01
2.39016056e-01 2.82215208e-01 2.89163232e-01 9.03463304e-01
3.39376450e-01 5.39129853e-01 1.12726426e+00 -7.59168029e-01
3.98501247e-01 5.91875613e-01 -4.17482972e-01 -7.67367542e-01
-1.10619116e+00 -3.67311478e-01 -9.94612336e-01 5.14444530e-01
6.18059158e-01 -1.81975976e-01 -1.01590455e+00 1.60796165e+00
4.02266596e-04 -5.15202165e-01 -6.44662559e-01 1.24136102e+00
5.18364012e-01 -2.93795228e-01 3.35164636e-01 -1.84740111e-01
1.30466509e+00 -3.70685071e-01 -1.93010166e-01 -3.56155962e-01
8.21263731e-01 -1.54516682e-01 8.40278566e-01 6.66028485e-02
-1.24659073e+00 6.13726676e-02 -7.42374182e-01 3.89649123e-02
-2.20460594e-01 9.87655818e-02 5.51756084e-01 7.07720757e-01
-1.59871507e+00 8.90932381e-01 -1.27634335e+00 -3.58752757e-01
9.31506634e-01 6.74505591e-01 -8.97025168e-01 2.64204502e-01
-7.85605252e-01 1.28596854e+00 1.45455316e-01 4.40178700e-02
-6.48981273e-01 -1.16002810e+00 -4.41917956e-01 -3.19312841e-01
-5.15711308e-01 -7.61059165e-01 7.80512810e-01 -8.14526260e-01
-9.92315352e-01 1.42089653e+00 1.87240867e-03 -7.41398811e-01
6.96187139e-01 1.40844271e-01 -2.51167327e-01 3.58028829e-01
3.51369202e-01 6.09125614e-01 4.29220289e-01 -5.74047148e-01
8.54932219e-02 -1.21674931e+00 -6.46813750e-01 -6.23070672e-02
9.18553174e-02 2.76111603e-01 5.27384162e-01 -4.23964769e-01
5.42735517e-01 -6.60989165e-01 -4.11632955e-01 2.94810683e-01
-3.27308685e-01 2.45612487e-01 2.97254920e-01 -1.55371368e+00
5.56308389e-01 -1.45117378e+00 -7.37276301e-02 2.06570044e-01
9.64762151e-01 -1.23821311e-01 1.70913905e-01 -4.00780767e-01
-4.79879916e-01 4.37414259e-01 -7.29505122e-01 -1.44611478e-01
-2.72323161e-01 -4.87041682e-01 4.10054296e-01 1.29259443e+00
-2.03617573e-01 1.17308700e+00 -5.85838914e-01 -3.96004975e-01
6.84740171e-02 5.55468619e-01 -3.62649232e-01 -1.18977889e-01
4.31853324e-01 7.34888971e-01 -3.08812223e-03 5.73134243e-01
5.99836171e-01 -3.22680399e-02 1.29906908e-01 -9.97274891e-02
-9.36966911e-02 1.32016346e-01 -2.35645562e-01 1.59665430e+00
-2.47454762e-01 6.46822631e-01 4.51557249e-01 -1.00190389e+00
1.06203389e+00 2.83955127e-01 7.45952785e-01 -8.67594957e-01
4.78530377e-01 5.45265496e-01 6.80048287e-01 -2.47292876e-01
-4.24573243e-01 -7.62467980e-01 4.93427128e-01 5.89007676e-01
9.37129781e-02 -2.44782478e-01 -7.65410252e-03 -2.21925184e-01
1.18654454e+00 -1.87471479e-01 3.25481236e-01 -7.71295905e-01
3.22412997e-01 -1.07302196e-01 3.64698827e-01 -2.39574648e-02
-6.28713429e-01 7.03538060e-01 3.62669885e-01 -7.98526883e-01
-1.36911941e+00 -1.27112854e+00 -6.38622761e-01 3.05392712e-01
-7.59850264e-01 -8.38533393e-04 -1.02626467e+00 -3.74306947e-01
1.05259992e-01 2.08035484e-01 -9.05628085e-01 6.56100586e-02
-5.81743240e-01 -1.19015372e+00 4.73789215e-01 5.16438186e-01
3.96211386e-01 -8.86564910e-01 -6.59768879e-01 1.60816774e-01
3.70246288e-03 -7.93434560e-01 -3.70955229e-01 -6.94374219e-02
-1.56102836e+00 -1.22957087e+00 -1.27509189e+00 -6.99359357e-01
8.62654924e-01 -4.43455637e-01 9.79625463e-01 1.28702238e-01
-7.36484647e-01 -3.54255084e-03 2.41279542e-01 -1.70028850e-01
-1.24865606e-01 3.42746563e-02 5.32283425e-01 -5.33013701e-01
3.88753027e-01 -1.05856931e+00 -1.11829555e+00 9.25012007e-02
-4.39835161e-01 -1.01641908e-01 7.10816622e-01 3.55867982e-01
8.08684170e-01 -5.78270257e-01 7.18206704e-01 -4.54522938e-01
6.19183898e-01 -5.63762128e-01 -3.43431056e-01 -2.17301011e-01
-9.68377590e-01 -4.40856516e-01 4.22276706e-01 -2.30700418e-01
-3.14575225e-01 -3.60121787e-01 -4.84099761e-02 -1.03899106e-01
-2.66906619e-01 3.71714473e-01 2.40350112e-01 -3.10148984e-01
6.30201995e-01 1.17080905e-01 5.59440672e-01 -3.73773694e-01
-5.16761206e-02 4.62922424e-01 4.33735102e-01 -2.41242737e-01
3.12472433e-01 6.50753081e-01 1.11862011e-01 -6.93720162e-01
-1.37001187e-01 -9.87274796e-02 -1.09840286e+00 -2.67303020e-01
1.04478526e+00 -6.02966726e-01 -3.96646559e-01 3.92797798e-01
-7.43199229e-01 -8.52307439e-01 -6.20508045e-02 8.40475500e-01
-6.75520539e-01 2.20948920e-01 -6.22398019e-01 -2.37215891e-01
-1.28101182e+00 -1.38261557e+00 7.91503072e-01 -2.43999422e-01
-5.01018643e-01 -1.20529783e+00 2.43945450e-01 4.00023222e-01
8.44495416e-01 5.11604488e-01 1.18203318e+00 -6.71164274e-01
4.82542403e-02 -2.87346005e-01 -4.76000786e-01 1.21822499e-01
1.87851906e-01 -4.45425570e-01 -4.49131191e-01 -1.93467736e-01
6.47578016e-02 -1.50304660e-01 6.14654481e-01 9.43884373e-01
1.09059942e+00 -2.01508179e-01 -8.77272487e-02 5.58641374e-01
1.50338936e+00 1.22677453e-01 5.77149034e-01 5.95927358e-01
5.45591533e-01 6.77335262e-01 -4.36905533e-01 1.03350155e-01
4.35096264e-01 4.36408043e-01 2.55890459e-01 -7.05179758e-04
-3.06755006e-01 5.28108537e-01 2.17695311e-01 8.87067676e-01
-4.76626128e-01 8.82690430e-01 -1.41385174e+00 8.79955947e-01
-1.00685453e+00 -7.38911986e-01 -5.20291865e-01 2.37680006e+00
8.45526397e-01 -5.04526123e-02 4.28423226e-01 -3.07337910e-01
8.22399080e-01 -9.16287974e-02 -7.10917771e-01 -3.40090871e-01
-3.16250138e-02 4.46739852e-01 8.17066014e-01 3.42876762e-01
-5.99345922e-01 5.33798456e-01 7.17053938e+00 -3.91418627e-03
-1.43685400e+00 5.09649336e-01 8.76294672e-01 -5.79160750e-01
6.60971776e-02 -3.68949831e-01 -4.49675359e-02 4.71727878e-01
1.44369793e+00 -4.05728996e-01 6.64747596e-01 4.44460332e-01
5.86681664e-01 -1.81281030e-01 -7.35769987e-01 7.66196609e-01
-7.28560090e-02 -1.13026512e+00 -3.11007023e-01 3.47064227e-01
4.67868060e-01 8.17089736e-01 -3.80513933e-03 -1.00403473e-01
-1.45428121e-01 -1.33223116e+00 6.82821274e-01 8.49990010e-01
1.40832031e+00 -4.47799802e-01 7.35414445e-01 -1.12332903e-01
-8.92170072e-01 2.23990709e-01 -4.10670877e-01 -7.79015496e-02
3.08914572e-01 1.02100921e+00 -1.02274585e+00 -1.33402720e-01
8.17756057e-01 3.65259916e-01 -5.52125573e-01 1.30213439e+00
1.53983429e-01 5.60268939e-01 -1.84764042e-01 4.62966710e-01
-4.62379940e-02 -4.26297069e-01 4.34490323e-01 9.74983335e-01
4.22426820e-01 4.11497913e-02 -4.04324502e-01 1.35495567e+00
-6.66978881e-02 5.21124780e-01 -3.90858740e-01 -1.17875293e-01
-6.73536351e-03 1.36919272e+00 -1.09160411e+00 -1.48500115e-01
-4.56468642e-01 4.30312544e-01 3.77528042e-01 1.06347818e-02
-3.69773805e-01 -2.46848956e-01 5.90482295e-01 7.16891408e-01
-2.79621601e-01 -5.11067092e-01 -9.26312864e-01 -9.03377712e-01
8.17627758e-02 -5.59417188e-01 -9.48979780e-02 -6.45362437e-01
-1.23682106e+00 6.77769423e-01 -2.15313017e-01 -6.07732475e-01
3.28505188e-02 -6.19486451e-01 -8.03883493e-01 1.31424212e+00
-1.10824120e+00 -7.79938996e-01 -9.58181396e-02 3.99398416e-01
8.97740945e-02 -9.77351069e-02 9.51702774e-01 1.53000414e-01
-4.46210384e-01 1.90841570e-01 -2.49353498e-02 3.73579741e-01
6.10995591e-01 -1.40887165e+00 3.96001011e-01 4.57859129e-01
-6.08271122e-01 1.05669141e+00 3.55189115e-01 -1.10881054e+00
-9.27328885e-01 -1.14261854e+00 9.01472867e-01 -4.08612713e-02
1.01493633e+00 -1.42547768e-02 -9.42349315e-01 7.59105563e-01
-9.76244360e-02 4.18899655e-02 8.03257525e-01 -2.25070007e-02
-1.93987474e-01 7.48985633e-02 -1.57598865e+00 3.01746666e-01
8.03833663e-01 -4.21653628e-01 -7.85341918e-01 4.69382167e-01
1.60654217e-01 -2.41250515e-01 -1.39266801e+00 2.85302043e-01
9.76101518e-01 -8.92321885e-01 8.74293923e-01 -5.02781451e-01
4.80570585e-01 2.54672557e-01 1.40048355e-01 -1.40609729e+00
-2.86866605e-01 9.48716998e-02 1.30917758e-01 5.40358841e-01
3.35518032e-01 -7.53456891e-01 9.19836521e-01 1.10849845e+00
-2.73843080e-01 -1.05384541e+00 -1.25190353e+00 -8.23327005e-01
9.95386839e-01 -2.84707189e-01 4.76514161e-01 6.86959624e-01
2.64834255e-01 -3.02309841e-01 4.37588573e-01 -1.44658744e-01
8.65568459e-01 -3.20319563e-01 -4.79852036e-02 -1.52208877e+00
3.96567017e-01 -7.94896543e-01 -9.13712263e-01 1.82546243e-01
4.16170627e-01 -1.57035506e+00 -2.74957895e-01 -1.95934820e+00
4.76289809e-01 -3.78579915e-01 -9.75909829e-02 5.27480364e-01
3.74484003e-01 5.11666059e-01 -1.49255633e-01 3.25567126e-01
1.95515394e-01 3.38518620e-01 1.32978642e+00 -1.50028944e-01
-2.02825397e-01 -4.40337867e-01 -4.95427370e-01 8.93455327e-01
1.28244829e+00 -3.23024333e-01 -8.69752988e-02 -4.89827603e-01
-1.91971719e-01 -7.72581697e-02 4.82093573e-01 -1.34400594e+00
-2.07256228e-01 1.01188108e-01 8.72968137e-01 -9.51189026e-02
-2.41008159e-02 -4.91223335e-01 8.43154564e-02 7.09730625e-01
-8.56439397e-03 2.80914187e-01 1.78409159e-01 -5.96820973e-02
4.16031510e-01 8.55686516e-02 1.14302409e+00 -4.59976822e-01
-4.34276275e-02 7.56102562e-01 -6.01331294e-01 2.56307386e-02
6.25409663e-01 -3.26976895e-01 -3.49742651e-01 -1.86982557e-01
-1.12777305e+00 -3.15302372e-01 5.22523224e-01 -5.19011617e-02
6.46182537e-01 -1.20737863e+00 -8.72686803e-01 -1.40994266e-01
-3.08486819e-01 -3.51363569e-01 4.38804388e-01 1.85547900e+00
-8.21031749e-01 6.04478598e-01 -9.00519729e-01 -3.96511912e-01
-8.96061361e-01 9.88046005e-02 9.15606558e-01 -1.95232764e-01
-1.06841481e+00 6.52150810e-01 3.07353716e-02 -3.78364801e-01
-2.55059451e-01 -5.47484815e-01 -3.11993193e-02 -2.04509739e-02
8.00544620e-01 5.24108231e-01 4.02232260e-01 -1.07250237e+00
-4.66529787e-01 5.13675094e-01 -5.22933677e-02 -1.41947642e-01
1.73605835e+00 -2.74351001e-01 -7.72369444e-01 3.38184267e-01
1.27661777e+00 -3.06256682e-01 -8.14864218e-01 2.57251620e-01
-5.75537756e-02 -3.12310029e-02 5.23039818e-01 -1.01142621e+00
-1.64816272e+00 8.91487837e-01 1.25072718e+00 -1.88322991e-01
7.26327717e-01 7.22788423e-02 9.61788833e-01 -2.03002691e-01
8.11186612e-01 -8.18412244e-01 -6.54291749e-01 1.68990999e-01
1.17456305e+00 -7.79154658e-01 9.15315673e-02 1.74503073e-01
-2.24850789e-01 1.14596212e+00 5.38186312e-01 -5.08445442e-01
7.53232002e-01 2.29405567e-01 3.36260535e-02 -6.81423664e-01
-1.04584806e-01 1.83598623e-01 3.22220474e-01 7.88123488e-01
7.79358149e-01 3.90906274e-01 -8.95531774e-01 9.40595984e-01
-6.71654463e-01 9.33005884e-02 4.25994158e-01 4.58496869e-01
-6.06984198e-01 -7.98250437e-01 -3.45244408e-01 1.38286543e+00
-7.85550356e-01 -2.13531956e-01 -4.82297689e-01 7.36304998e-01
1.81069255e-01 3.68397921e-01 2.47221589e-01 -5.80659173e-02
3.62630002e-04 2.98037827e-01 6.29483640e-01 -6.12501264e-01
-5.01623452e-01 -2.66456634e-01 -1.23096816e-01 -7.52378583e-01
-2.94229388e-01 -9.99155641e-01 -1.53746986e+00 -2.70999163e-01
6.74724728e-02 -4.40900236e-01 8.78780186e-01 1.10592401e+00
2.07337171e-01 3.86772037e-01 -1.06827067e-02 -1.06818771e+00
2.98853405e-02 -1.31423283e+00 -1.07632577e+00 1.75330535e-01
1.96622610e-01 -8.51264656e-01 -4.27793801e-01 2.59632841e-02] | [14.10041332244873, -2.1630358695983887] |
c6f7c219-85eb-49eb-94ef-93a9f067ca4d | reflection-separation-via-multi-bounce | null | null | https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/2055_ECCV_2020_paper.php | https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123580766.pdf | Reflection Separation via Multi-bounce Polarization State Tracing | Reflection removal from photographs is an important task in computational photography, but also for computer vision tasks that involve imaging through windows and similar settings. Traditionally, the problem is approached as a single reflection removal problem under very controlled scenarios. In this paper we aim to generalize the reflection removal to real-world scenarios with more complicated light interactions. To this end, we propose a simple yet efficient learning framework for supervised image reflection separation with a polarization simulation engine for synthetic polarized data generation and loss function design. Instead of a conventional image sensor, we use a polarization sensor that instantaneously captures four linearly polarized photos of the scene in the same image. Through a combination of a new polarization-guided image formation model and a novel supervised learning framework for the interpretation of a ray-tracing polarized image formation model, a general method is obtained to tackle general image reflection removal problems. We demonstrate our method with extensive experiments on both real and synthetic data and demonstrate the unprecedented quality of image reconstructions. | ['Simeng Qiu', 'Rui Li', 'Wolfgang Heidrich', 'Guangming Zang'] | null | null | null | null | eccv-2020-8 | ['reflection-removal'] | ['computer-vision'] | [ 1.16804421e+00 5.13389856e-02 4.35320318e-01 -2.89729089e-01
-6.99104309e-01 -3.24503541e-01 4.95390505e-01 -5.17236412e-01
-2.15662658e-01 5.55742502e-01 -1.22773640e-01 -3.54588926e-01
-1.88344046e-01 -6.59108102e-01 -8.25269163e-01 -1.30550110e+00
5.43328226e-01 3.46719146e-01 1.86892543e-02 -1.64499998e-01
2.84228623e-01 6.40286922e-01 -1.71491313e+00 1.05612971e-01
6.10809028e-01 7.00217426e-01 3.36896449e-01 6.63370013e-01
1.24023348e-01 6.58747077e-01 -1.52203768e-01 -1.35128617e-01
5.16347528e-01 -4.17443275e-01 -6.05388880e-01 7.34168589e-01
6.20342433e-01 -3.44356805e-01 -1.00385107e-01 1.01337039e+00
3.51315320e-01 1.94121432e-02 5.74114084e-01 -9.68555987e-01
-3.89341079e-02 -6.41977489e-02 -5.15964687e-01 -3.34219694e-01
3.75578523e-01 1.30836606e-01 5.93251228e-01 -6.69212580e-01
4.61373121e-01 6.97963834e-01 5.64846337e-01 5.04735827e-01
-1.37554574e+00 -1.22860171e-01 -2.59825915e-01 -1.69360891e-01
-9.01230931e-01 -6.30907178e-01 1.17654777e+00 -5.43765604e-01
4.72282916e-01 2.95556098e-01 4.28314090e-01 8.13219547e-01
-6.18772097e-02 3.83679807e-01 1.56849957e+00 -8.95747721e-01
7.00331032e-02 4.49722290e-01 1.63977936e-01 7.72272170e-01
2.35327005e-01 1.83027372e-01 -3.68075192e-01 6.77175224e-02
5.30042827e-01 2.12814718e-01 -6.38887465e-01 -7.12162375e-01
-8.70036185e-01 3.06567192e-01 1.56864703e-01 -2.03464776e-01
-2.73406148e-01 -2.26717353e-01 -3.19946140e-01 1.64833784e-01
5.53480148e-01 5.84955037e-01 -1.42399380e-02 5.19125104e-01
-5.79165995e-01 9.34657454e-02 9.24121797e-01 6.69404745e-01
9.98558819e-01 -8.35449398e-02 8.08367953e-02 9.11246359e-01
2.34047562e-01 9.01372433e-01 -2.72279650e-01 -1.18446004e+00
2.47262269e-01 3.70133221e-01 4.93880272e-01 -6.71063662e-01
-1.93979517e-01 -3.28437656e-01 -6.54355645e-01 8.09029341e-01
4.48899090e-01 -1.92567229e-01 -8.83196473e-01 1.52819026e+00
3.70355517e-01 8.82796571e-02 5.15398502e-01 9.51475143e-01
3.68105829e-01 7.21007288e-01 -5.78452528e-01 -6.49495602e-01
1.14088011e+00 -7.69507170e-01 -3.18908513e-01 -1.53361857e-01
1.38951287e-01 -8.23766649e-01 8.60537887e-01 9.28587615e-01
-1.27840900e+00 -4.31428850e-02 -9.86164987e-01 -7.36311898e-02
1.61777526e-01 3.74931604e-01 4.30966854e-01 5.90724885e-01
-6.79262280e-01 4.69474256e-01 -4.03016865e-01 -3.92284840e-01
1.79970548e-01 5.97747825e-02 -2.45583057e-01 -4.44263965e-01
-3.11900020e-01 5.20430207e-01 -2.23811984e-01 1.42163247e-01
-5.63724816e-01 -5.80675006e-01 -5.19154251e-01 -2.14443520e-01
4.92955774e-01 -8.97123694e-01 9.44694936e-01 -9.52267051e-01
-2.00915360e+00 1.18326485e+00 -2.57547379e-01 -1.29999235e-01
3.06154042e-01 -2.32418105e-01 -9.27466974e-02 3.26047868e-01
-2.38875002e-01 -3.20200503e-01 1.17438579e+00 -1.99002504e+00
-3.24445754e-01 -4.03254002e-01 1.72896031e-02 3.43243718e-01
2.92272791e-02 -1.19583972e-01 -3.77303123e-01 6.97758794e-02
3.54763925e-01 -9.17681038e-01 -3.45374674e-01 -3.09782159e-02
-5.79902053e-01 6.26526952e-01 8.67098629e-01 -2.86196947e-01
2.64851272e-01 -2.01547432e+00 1.90281585e-01 1.28453240e-01
9.93896648e-02 5.53191960e-01 -2.30790049e-01 5.32048047e-01
-1.90062433e-01 -4.57335413e-01 -6.67540252e-01 -6.58283532e-01
-5.22248268e-01 1.25286326e-01 -5.57889581e-01 6.36222005e-01
5.86682484e-02 3.72074962e-01 -6.18801892e-01 -5.87916821e-02
4.70661044e-01 6.49200320e-01 -6.82015777e-01 4.09380406e-01
-2.01260597e-01 8.24797809e-01 -3.19931000e-01 4.25070018e-01
1.09871721e+00 -9.73346531e-02 3.40162188e-01 -4.36103433e-01
-5.72581828e-01 -8.11127946e-02 -1.08395636e+00 1.31509149e+00
-8.40027928e-01 4.44249719e-01 5.13330996e-01 -1.32946575e+00
9.07100260e-01 -7.73586892e-03 6.34514451e-01 -9.01785731e-01
-2.28974968e-02 2.27647170e-01 -5.40121198e-01 -8.14016283e-01
4.41662341e-01 -4.91541505e-01 1.86706275e-01 6.50681794e-01
-1.78239688e-01 -5.24441540e-01 -1.63638845e-01 -1.82385787e-01
9.51852381e-01 2.70182669e-01 6.93598986e-02 -1.46509102e-02
6.85649455e-01 4.87066656e-02 3.75304163e-01 6.89511120e-01
5.43919563e-01 9.10942793e-01 1.53109327e-01 -3.66593748e-01
-1.13931727e+00 -1.21473801e+00 -1.16595350e-01 4.27814603e-01
4.43491280e-01 4.02657650e-02 -5.86798608e-01 -1.33304313e-01
-2.95504630e-01 5.79713464e-01 1.31465092e-01 9.09589604e-02
-7.05793142e-01 -1.08502233e+00 9.29230731e-03 -2.29970828e-01
5.63076913e-01 -8.39539528e-01 -7.24932015e-01 -2.14078173e-01
-2.39396811e-01 -1.40530777e+00 4.17804927e-01 -2.78077964e-02
-5.98944724e-01 -1.46107590e+00 -6.31593823e-01 -6.20357692e-01
9.02105570e-01 8.31328928e-01 9.61084127e-01 -1.53354779e-01
-7.31366396e-01 9.55942392e-01 -2.18550056e-01 -3.45824420e-01
-4.74120945e-01 -6.48552358e-01 -1.30636334e-01 5.96714497e-01
-1.23676233e-01 -7.69295633e-01 -6.94456458e-01 2.89016992e-01
-1.09382904e+00 4.73639429e-01 7.06602931e-01 8.01812708e-01
6.09986007e-01 2.60732502e-01 -2.25512221e-01 -9.52374697e-01
3.50638151e-01 -2.26143271e-01 -1.08501744e+00 2.63142258e-01
-4.99370426e-01 1.44907553e-02 7.45023251e-01 -3.27270068e-02
-1.78655350e+00 2.30607256e-01 1.68161094e-02 -1.38525009e-01
-3.57891619e-01 2.53040763e-03 -3.01164150e-01 -4.56704497e-01
5.31884789e-01 4.63491142e-01 2.62762159e-01 -5.92665434e-01
1.75821260e-01 5.44114113e-01 6.14449322e-01 -4.73003179e-01
9.17768180e-01 1.23093510e+00 6.03933096e-01 -1.62477076e+00
-8.51818860e-01 -6.39058769e-01 -3.27139229e-01 -2.76227266e-01
6.94540262e-01 -7.94159591e-01 -9.47154701e-01 7.92108119e-01
-1.09958875e+00 -4.92732018e-01 -3.42948467e-01 5.48328578e-01
-7.44277656e-01 5.99891186e-01 -2.27348164e-01 -9.67141151e-01
-1.24244317e-01 -1.07887852e+00 1.31460893e+00 2.88636893e-01
6.67786896e-01 -9.68790591e-01 4.75009650e-01 7.81073630e-01
1.33926123e-01 8.27831477e-02 7.64575779e-01 4.51708555e-01
-1.19693959e+00 1.27755016e-01 -3.67052764e-01 8.50303233e-01
2.86491625e-02 -1.01206325e-01 -1.26664364e+00 -7.90483505e-02
6.11699522e-01 -2.49895707e-01 1.13397050e+00 3.32018197e-01
1.03566670e+00 -2.01667964e-01 -3.74966562e-01 1.19330931e+00
1.97527885e+00 -1.56952795e-02 9.25246477e-01 1.48331821e-01
7.85582542e-01 9.21269357e-01 4.76114661e-01 3.31196100e-01
-6.79538585e-03 9.25380409e-01 5.68606198e-01 -4.28978920e-01
-2.53426671e-01 1.20523967e-01 2.52437770e-01 4.32608604e-01
-4.46269155e-01 -5.24411321e-01 -7.25054741e-01 2.52360821e-01
-1.65326083e+00 -9.39799488e-01 -4.60311741e-01 2.55725050e+00
4.24013048e-01 -4.17363346e-01 -3.56558263e-01 1.42448604e-01
3.96521807e-01 1.09369859e-01 -1.68779328e-01 -1.26954123e-01
-3.00525904e-01 2.50680119e-01 5.76564372e-01 9.87297475e-01
-8.52133811e-01 6.24964178e-01 6.13554955e+00 2.08933756e-01
-1.45179105e+00 -2.05614775e-01 1.62687406e-01 1.77673176e-01
-5.69148958e-01 1.98794708e-01 -6.71413541e-01 1.63537115e-01
1.82705283e-01 3.73048455e-01 4.86263901e-01 2.77864248e-01
3.80715191e-01 -5.69889665e-01 -9.68562663e-01 1.27025616e+00
2.84834862e-01 -1.03327310e+00 9.74597335e-02 5.92777389e-04
6.46181881e-01 -1.49109349e-01 2.72353351e-01 -6.59026623e-01
-1.59164309e-03 -6.72902584e-01 2.90677473e-02 9.49173152e-01
6.92557275e-01 -2.13792726e-01 2.85357889e-02 5.08457363e-01
-4.57722694e-01 -1.20755769e-01 -1.97605297e-01 -1.53029680e-01
4.62546676e-01 1.00974751e+00 -6.44531310e-01 7.36040950e-01
4.47110534e-01 7.84869015e-01 2.04433203e-02 8.99314463e-01
-3.09272170e-01 4.23465848e-01 -2.71471679e-01 4.79496270e-01
-1.22425571e-01 -9.09509718e-01 9.92771089e-01 9.35335219e-01
3.79230618e-01 2.73996294e-01 -1.29590109e-01 9.27524507e-01
2.50198811e-01 -2.80271083e-01 -8.83015752e-01 4.95220453e-01
-3.55540752e-01 1.55611968e+00 -3.77554417e-01 1.06943071e-01
-4.65388477e-01 1.04265285e+00 8.60728025e-02 8.62325788e-01
-4.58687305e-01 7.44901877e-03 7.13952780e-01 4.54767793e-01
1.62789211e-01 -2.55318284e-01 8.08039308e-02 -1.49277127e+00
3.69548827e-01 -3.45666081e-01 -2.39229470e-01 -1.07993841e+00
-1.17113066e+00 2.50313282e-01 1.75249413e-01 -1.32287419e+00
1.29541293e-01 -9.38596249e-01 -7.93560088e-01 8.55849028e-01
-2.25979853e+00 -1.07710326e+00 -5.58324754e-01 9.67259049e-01
1.50685012e-01 1.37682810e-01 6.68512642e-01 8.73539150e-02
-4.34884310e-01 -2.04561546e-01 5.51917732e-01 -3.10218155e-01
7.07137764e-01 -1.01170480e+00 -3.15213144e-01 1.10583580e+00
-1.71439305e-01 4.71842498e-01 8.87293875e-01 -1.94739014e-01
-1.78616047e+00 -1.03741693e+00 4.37060714e-01 -7.90177137e-02
1.62190273e-01 -3.58101755e-01 -4.73503351e-01 5.28640687e-01
2.54214734e-01 -8.49378484e-05 5.10255396e-01 -2.73431689e-01
-3.22862059e-01 -5.14262736e-01 -1.04357231e+00 3.86666298e-01
1.06075060e+00 -3.83554429e-01 -3.39675516e-01 7.04656720e-01
1.13174073e-01 -2.35292017e-01 -2.59208828e-01 5.38880467e-01
5.22842467e-01 -1.22334635e+00 1.32265890e+00 -2.95806706e-01
5.48613548e-01 -3.71109396e-01 -1.91724017e-01 -1.29853928e+00
7.62375817e-02 -8.53606701e-01 4.27794874e-01 1.02936411e+00
1.54067397e-01 -1.00013244e+00 8.38931382e-01 3.26989800e-01
-1.70675501e-01 -2.74808168e-01 -2.58803129e-01 -6.08584523e-01
-2.93957651e-01 -2.55591750e-01 -9.03283730e-02 5.90146363e-01
-4.24096882e-01 4.20995653e-01 -5.69614947e-01 5.67979097e-01
1.53152454e+00 6.27708733e-01 1.04285598e+00 -1.25571990e+00
-5.93392968e-01 6.08949959e-02 -2.86407638e-02 -1.20743918e+00
2.27215409e-01 -6.03030324e-01 4.21927154e-01 -1.41108763e+00
2.71069348e-01 -6.94132566e-01 3.08405966e-01 -1.94603711e-01
2.08132908e-01 4.32408482e-01 8.32936987e-02 3.48497331e-01
-3.30280751e-01 3.17087293e-01 1.20304286e+00 2.45723426e-02
-3.02904546e-01 4.56210583e-01 -5.71630716e-01 9.28110480e-01
6.63098514e-01 -1.78099558e-01 -6.88803732e-01 -6.05583429e-01
5.17240107e-01 9.54794958e-02 9.19238389e-01 -9.62536216e-01
2.67329752e-01 -1.89377770e-01 -8.28809440e-02 -1.29391342e-01
7.22735226e-01 -9.26095724e-01 1.79007813e-01 2.99207065e-02
6.59955740e-02 -9.74251747e-01 -2.87023336e-01 6.28334999e-01
3.68062378e-04 -3.48626614e-01 1.00210905e+00 -3.82971376e-01
-6.30044222e-01 6.27160594e-02 -3.03539604e-01 -8.14881474e-02
9.59222853e-01 -1.01695210e-01 -6.28050625e-01 -2.81210393e-01
-6.73214316e-01 -1.78614154e-01 7.88664579e-01 -2.83263445e-01
8.15487266e-01 -6.97956383e-01 -4.52160388e-01 3.46035451e-01
2.46827513e-01 9.36913341e-02 3.13323259e-01 9.16966259e-01
-7.01103449e-01 1.93111539e-01 -3.60365920e-02 -7.36538887e-01
-1.63492644e+00 2.83241183e-01 5.84531963e-01 -3.37327011e-02
-8.28811824e-01 4.57122713e-01 5.41472375e-01 -4.24177647e-01
-2.19010144e-01 -2.62183510e-02 -6.41854554e-02 -4.19482321e-01
4.65868890e-01 3.14174265e-01 1.19517237e-01 -6.41463101e-01
7.58002773e-02 1.10655665e+00 2.97690094e-01 -1.34843528e-01
1.57181430e+00 -4.61690098e-01 -1.93646461e-01 2.33940080e-01
1.16704941e+00 3.30428451e-01 -1.37197220e+00 -4.06798303e-01
-7.34748900e-01 -5.61047375e-01 3.44917253e-02 -4.38013226e-01
-1.02548873e+00 9.12249088e-01 9.31484997e-02 1.31439924e-01
1.39917052e+00 8.16776827e-02 4.61491078e-01 6.54707789e-01
2.75288343e-01 -8.89514089e-01 1.44166559e-01 1.86886087e-01
5.84185898e-01 -1.38482916e+00 1.17735714e-01 -8.18832219e-01
-3.35391641e-01 1.13780415e+00 7.26511851e-02 -1.12337880e-01
6.81163251e-01 3.26401204e-01 1.21074833e-01 -3.24805945e-01
-4.03689891e-01 -2.29004592e-01 -5.40842451e-02 7.99817562e-01
7.82377794e-02 -2.65387475e-01 -2.06143960e-01 -2.38452982e-02
2.49644324e-01 2.90257223e-02 8.87504637e-01 7.76410103e-01
-3.45752746e-01 -1.37001646e+00 -4.17468935e-01 -3.92700806e-02
-2.72758782e-01 9.17054340e-02 -1.95044011e-01 5.93420982e-01
-8.94179791e-02 8.57047796e-01 -1.14987105e-01 1.74700037e-01
3.09595436e-01 -5.40011413e-02 8.29780161e-01 -7.34908283e-01
1.32763088e-01 6.99842274e-02 7.65367746e-02 -6.06490135e-01
-1.20049500e+00 -7.59911060e-01 -7.40258157e-01 1.60091057e-01
-1.88906148e-01 -1.70481458e-01 8.88543129e-01 8.66859734e-01
1.55234039e-01 2.05460578e-01 9.38154221e-01 -9.52991843e-01
-2.35299677e-01 -3.34645689e-01 -9.18051243e-01 5.29335201e-01
6.25403523e-01 -6.25683904e-01 -6.00663722e-01 2.09344789e-01] | [10.060291290283203, -2.835787773132324] |
18d48913-7b54-4934-9d52-bb0db77a9e35 | improved-diffusion-based-image-colorization | 2304.11105 | null | https://arxiv.org/abs/2304.11105v1 | https://arxiv.org/pdf/2304.11105v1.pdf | Improved Diffusion-based Image Colorization via Piggybacked Models | Image colorization has been attracting the research interests of the community for decades. However, existing methods still struggle to provide satisfactory colorized results given grayscale images due to a lack of human-like global understanding of colors. Recently, large-scale Text-to-Image (T2I) models have been exploited to transfer the semantic information from the text prompts to the image domain, where text provides a global control for semantic objects in the image. In this work, we introduce a colorization model piggybacking on the existing powerful T2I diffusion model. Our key idea is to exploit the color prior knowledge in the pre-trained T2I diffusion model for realistic and diverse colorization. A diffusion guider is designed to incorporate the pre-trained weights of the latent diffusion model to output a latent color prior that conforms to the visual semantics of the grayscale input. A lightness-aware VQVAE will then generate the colorized result with pixel-perfect alignment to the given grayscale image. Our model can also achieve conditional colorization with additional inputs (e.g. user hints and texts). Extensive experiments show that our method achieves state-of-the-art performance in terms of perceptual quality. | ['Tien-Tsin Wong', 'Chengze Li', 'Minshan Xie', 'Jinbo Xing', 'Hanyuan Liu'] | 2023-04-21 | null | null | null | null | ['colorization'] | ['computer-vision'] | [ 3.63837987e-01 -2.21130297e-01 6.76372573e-02 -4.15012807e-01
-4.75356460e-01 -6.37562156e-01 4.86807555e-01 -3.00177872e-01
-1.61409661e-01 2.14896724e-01 6.43686578e-02 -1.89563856e-01
3.74937356e-01 -8.13662469e-01 -6.55388951e-01 -6.72152817e-01
5.94388068e-01 1.16577022e-01 3.41450483e-01 -2.96456516e-01
4.25119162e-01 3.07885677e-01 -1.24092221e+00 4.50626165e-01
1.10134614e+00 9.48979437e-01 4.02539074e-01 8.93994451e-01
-4.81494784e-01 6.35409057e-01 -5.36504865e-01 -3.44725579e-01
4.20913577e-01 -6.18189096e-01 -5.89877248e-01 4.43560332e-01
5.66759706e-01 -4.23562557e-01 -1.88648060e-01 1.45271039e+00
1.47268191e-01 1.19706169e-01 5.84901512e-01 -1.27250576e+00
-1.38462591e+00 2.99842387e-01 -9.24425602e-01 -3.27106684e-01
1.05744436e-01 3.12071472e-01 8.71480525e-01 -9.99671817e-01
7.46633708e-01 1.38213003e+00 1.02768809e-01 6.53223753e-01
-1.32490778e+00 -5.80842972e-01 4.50190872e-01 1.12747237e-01
-1.14062881e+00 -4.99393418e-02 1.04500914e+00 -1.74302340e-01
4.63670701e-01 1.75677001e-01 7.87482738e-01 8.50429714e-01
1.28926069e-01 9.28181708e-01 1.39905548e+00 -6.52484477e-01
4.19071674e-01 2.82265782e-01 -2.63421834e-01 7.73422539e-01
1.95060614e-02 -2.18623430e-02 -6.90845132e-01 3.34906518e-01
1.08238339e+00 1.05017550e-01 -1.42991394e-01 -4.22210634e-01
-1.04469562e+00 5.64961791e-01 7.82576323e-01 2.26727277e-01
-3.74769986e-01 3.92081112e-01 -1.58878908e-01 6.06285632e-02
4.99051780e-01 3.17568660e-01 -2.39296883e-01 1.24311810e-02
-9.03778493e-01 -1.47470459e-01 4.66888905e-01 9.70404804e-01
1.10534692e+00 2.16032729e-01 -1.58229560e-01 7.79775798e-01
3.81485522e-01 9.05256033e-01 3.27901170e-02 -1.21858644e+00
4.48134571e-01 7.07429707e-01 3.21658134e-01 -9.75071430e-01
9.52840224e-02 -1.41206682e-01 -7.96641767e-01 8.01463842e-01
4.23508108e-01 1.62201356e-02 -1.36431956e+00 1.54087508e+00
2.52974808e-01 -2.31986091e-01 -2.16209069e-02 1.10255706e+00
9.56069678e-02 9.06117380e-01 1.23118669e-01 3.28100026e-01
1.13697684e+00 -1.16050816e+00 -6.15996122e-01 -4.98557836e-01
4.53367122e-02 -1.09815121e+00 1.54054618e+00 5.17175376e-01
-1.08659983e+00 -6.10666513e-01 -1.09861541e+00 -2.93286443e-01
-4.94144529e-01 1.92353830e-01 4.59605753e-01 8.64626229e-01
-1.37469363e+00 2.63223559e-01 -7.49468029e-01 -5.07010460e-01
2.14498907e-01 -1.08322963e-01 -4.31137383e-02 -4.46211100e-01
-9.71669257e-01 6.35425389e-01 1.83190867e-01 1.89262539e-01
-9.26779449e-01 -5.86100042e-01 -4.96778280e-01 -5.76037243e-02
3.56712133e-01 -6.59644842e-01 9.30294514e-01 -1.61170673e+00
-1.97067928e+00 7.71532059e-01 -1.19063616e-01 1.12544224e-01
7.19470382e-01 -2.39221081e-01 -2.16892600e-01 4.32898790e-01
5.47632761e-02 1.09359658e+00 1.31722343e+00 -1.85167718e+00
-7.22598255e-01 -2.16357991e-01 9.36525241e-02 2.90024281e-01
-5.39401293e-01 -9.76087525e-02 -1.27582932e+00 -8.66365194e-01
2.49586888e-02 -1.06587005e+00 -1.68598518e-01 6.17614448e-01
-6.11077547e-01 2.21888572e-01 9.77762520e-01 -6.59193695e-01
9.87684369e-01 -2.09719801e+00 1.96081698e-01 4.03212219e-01
1.28485203e-01 9.59017202e-02 -4.18893069e-01 3.86216372e-01
8.23167211e-04 -2.18374096e-02 -2.25854948e-01 -3.69261295e-01
1.64286438e-02 1.14245243e-01 -2.93220669e-01 3.31339866e-01
9.17182118e-02 9.96083200e-01 -8.79538536e-01 -3.91576082e-01
4.43584979e-01 6.31120145e-01 -6.17237985e-01 4.00405556e-01
-6.15477383e-01 3.45077991e-01 -3.93673122e-01 4.20793772e-01
9.24118102e-01 -4.32041407e-01 2.00573459e-01 -4.57470000e-01
-1.24805711e-01 -4.81282949e-01 -1.02667987e+00 2.01599860e+00
-5.68862557e-01 5.77294946e-01 1.46999523e-01 -4.16568696e-01
8.07991326e-01 -1.66755736e-01 2.98700273e-01 -9.27235126e-01
3.13506238e-02 3.67421322e-02 -3.19655448e-01 -4.64498512e-02
6.91087067e-01 -5.94198294e-02 3.80795859e-02 6.18425310e-01
-3.68043184e-01 -4.70020860e-01 1.16426200e-01 5.74657321e-01
3.74571294e-01 4.17832881e-01 -3.75549912e-01 -1.73214391e-01
3.73121113e-01 -9.27655841e-04 2.36034542e-01 6.23926103e-01
4.61974218e-02 8.73765647e-01 3.45505476e-01 -1.27080202e-01
-1.23478937e+00 -1.10220349e+00 5.58179379e-01 1.22590697e+00
5.50873578e-01 -7.51342475e-02 -9.41137433e-01 -5.83744943e-01
-1.82469383e-01 9.61080372e-01 -9.13333237e-01 -1.42497569e-01
-3.15552413e-01 -3.42901498e-01 1.02953590e-01 6.56074703e-01
8.08732331e-01 -8.57578933e-01 -5.46324015e-01 -3.12852263e-02
-1.33975267e-01 -9.96202767e-01 -9.39686477e-01 -7.17674568e-02
-6.88826680e-01 -8.10337603e-01 -1.12480474e+00 -7.45178521e-01
1.05498755e+00 6.86776936e-01 8.75829637e-01 1.19689032e-01
-3.74431551e-01 6.50141597e-01 -4.41393316e-01 -9.79139581e-02
-5.18219233e-01 -3.38816911e-01 -4.34451342e-01 3.93722296e-01
9.35799628e-02 -9.16537270e-02 -1.19614971e+00 3.55470687e-01
-1.43662143e+00 6.29310787e-01 7.10134566e-01 5.33057570e-01
5.51731348e-01 1.63894333e-02 1.50771802e-02 -9.23333287e-01
4.58009213e-01 1.97467104e-01 -6.97298646e-01 6.33276880e-01
-8.03992748e-01 2.76230067e-01 6.42477334e-01 -4.58402038e-01
-1.59133923e+00 2.62447335e-02 2.24211782e-01 -4.80897933e-01
7.82177895e-02 2.22602800e-01 -2.63472974e-01 -1.13467880e-01
4.63193744e-01 3.38866502e-01 -1.27675101e-01 -2.88254261e-01
1.20044374e+00 3.64639670e-01 5.13820708e-01 -7.22067773e-01
1.02374971e+00 8.48663330e-01 -2.19014570e-01 -6.09192312e-01
-7.62624204e-01 -2.18892187e-01 -5.62174857e-01 -3.88421893e-01
1.22181606e+00 -7.10474610e-01 -4.61903602e-01 8.71562719e-01
-1.05514169e+00 -8.74809980e-01 -1.40084088e-01 6.34566769e-02
-4.76690173e-01 5.64346731e-01 -5.34702837e-01 -6.57760084e-01
-2.75214881e-01 -1.04201031e+00 1.03956950e+00 5.34699917e-01
3.44313174e-01 -1.04948723e+00 -2.36788020e-01 1.26151100e-01
7.73611248e-01 -6.85950071e-02 1.15046406e+00 3.34930480e-01
-8.92550826e-01 4.68005985e-02 -8.43140662e-01 5.52597940e-01
2.59704769e-01 2.39369094e-01 -8.62372994e-01 -1.76239148e-01
-3.78903329e-01 -3.07615638e-01 8.68690729e-01 3.40200216e-01
1.10093021e+00 -7.08335638e-02 -1.75227914e-02 6.47400022e-01
1.77741134e+00 2.08231971e-01 7.69067824e-01 3.24527115e-01
9.74008858e-01 5.23595452e-01 4.92644042e-01 3.40183705e-01
3.49262476e-01 5.21503627e-01 3.33797485e-01 -7.85459995e-01
-7.44869351e-01 -6.12690866e-01 2.77882010e-01 4.72217172e-01
1.54211789e-01 -2.64335006e-01 -7.22717643e-01 3.42963725e-01
-1.64029753e+00 -6.31215394e-01 1.09806187e-01 2.01743364e+00
9.80276525e-01 9.29422304e-02 -2.58542478e-01 -3.61384183e-01
7.74259150e-01 1.26112625e-01 -8.60636413e-01 -2.15648755e-01
-1.84049442e-01 -9.19744298e-02 6.25449359e-01 6.46352232e-01
-7.18051374e-01 1.43950903e+00 5.79938078e+00 8.15718353e-01
-1.41217041e+00 -3.91215794e-02 9.45726514e-01 3.05916518e-01
-7.13160038e-01 2.11233839e-01 -3.84187013e-01 3.06251347e-01
2.34634116e-01 -6.64048493e-02 8.32431257e-01 5.85714757e-01
3.20771784e-01 -3.23231518e-01 -8.25933695e-01 1.04384935e+00
2.21994758e-01 -9.95332122e-01 4.57197100e-01 -1.33982196e-01
1.26979101e+00 -3.19607973e-01 7.01210558e-01 -1.09116219e-01
7.07385540e-01 -6.72504425e-01 1.05427921e+00 6.49655044e-01
1.23103845e+00 -5.80213785e-01 -1.77464262e-03 -1.59320548e-01
-1.13873577e+00 8.37216005e-02 -3.41854215e-01 4.17713434e-01
1.98513508e-01 4.58816946e-01 -4.67243552e-01 3.57141554e-01
6.44225895e-01 5.86643159e-01 -8.02503645e-01 5.90859532e-01
-6.65389180e-01 4.92403507e-01 -1.73193425e-01 -1.79868133e-03
4.19456869e-01 -6.30318940e-01 -3.87192070e-02 1.07350898e+00
2.09559798e-01 4.09894139e-02 1.06013671e-01 1.21461439e+00
-8.28234702e-02 -8.60806741e-03 -1.46884590e-01 -2.16202229e-01
7.13598728e-02 1.33699870e+00 -9.87988770e-01 -5.35573304e-01
-6.35724306e-01 1.79783845e+00 4.79931384e-02 1.02659261e+00
-8.01508904e-01 -5.24226308e-01 4.38217461e-01 -4.91955206e-02
3.90426069e-01 -3.82535964e-01 -4.62473780e-01 -1.22122526e+00
-3.08500558e-01 -7.15954304e-01 1.01049393e-01 -1.38724208e+00
-1.27486730e+00 5.47208905e-01 -2.36821845e-01 -1.02345598e+00
1.60095558e-01 -7.98046470e-01 -6.62850618e-01 1.08850563e+00
-1.85759199e+00 -1.52063620e+00 -6.87044084e-01 8.36159170e-01
5.17781675e-01 2.12085173e-01 4.58400279e-01 5.77746555e-02
-2.86336452e-01 4.10637051e-01 2.59618700e-01 1.19894437e-01
1.07543468e+00 -1.42510378e+00 5.93135238e-01 1.17991197e+00
8.20770115e-02 6.87878013e-01 7.45256007e-01 -7.04947889e-01
-1.69577777e+00 -1.09315956e+00 2.94006646e-01 -2.58197248e-01
6.30256832e-01 -2.83917874e-01 -7.11373746e-01 3.40713680e-01
5.95777452e-01 -2.38608286e-01 2.24728793e-01 -2.92923331e-01
-6.59956098e-01 -3.38140905e-01 -7.16231287e-01 8.76877904e-01
6.93585753e-01 -6.90359056e-01 -5.41042723e-02 7.16593638e-02
5.35223722e-01 -1.81916028e-01 -3.23902488e-01 -3.27820182e-01
5.55432916e-01 -9.19874728e-01 9.39539552e-01 -2.80788481e-01
5.22764385e-01 -5.34660578e-01 -1.95087537e-01 -1.31454277e+00
-2.56306261e-01 -5.69390357e-01 4.84951198e-01 1.12071729e+00
3.94114554e-01 -3.54134858e-01 8.10549140e-01 1.12988329e+00
1.25273168e-01 -1.59949034e-01 -3.15846860e-01 -5.62448084e-01
1.66022927e-01 -3.13322395e-01 3.43138278e-01 6.55404925e-01
-3.30797166e-01 1.50431722e-01 -4.66715038e-01 2.06406102e-01
8.18721294e-01 5.29732764e-01 8.53867829e-01 -7.66866982e-01
-2.60394514e-01 -5.97414613e-01 9.35816392e-02 -1.39870179e+00
-2.10216612e-01 -7.08440661e-01 2.15121359e-01 -1.93193734e+00
4.32869405e-01 -4.11543041e-01 -3.52306843e-01 4.81615424e-01
-5.61298251e-01 4.59182769e-01 5.63849330e-01 9.66409370e-02
-7.83999026e-01 6.45531833e-01 1.67668700e+00 -4.98125702e-01
-2.17987835e-01 -5.12472570e-01 -8.60017896e-01 5.26541293e-01
6.79752886e-01 -1.72255546e-01 -6.14190698e-01 -8.21743190e-01
3.60552847e-01 -1.12361856e-01 3.01204354e-01 -7.53503203e-01
3.91839266e-01 -4.57751811e-01 6.66488409e-01 -3.70378047e-01
2.90327162e-01 -9.13896739e-01 -1.27259046e-01 3.21859270e-01
-4.15041327e-01 1.48348054e-02 1.37138050e-02 8.21148276e-01
-1.24203578e-01 1.17539272e-01 1.01492524e+00 -1.17024414e-01
-1.08128881e+00 3.44171643e-01 -1.42019749e-01 1.90246291e-03
8.83779347e-01 -2.15303525e-01 -3.47248882e-01 -6.46542609e-01
-4.34339315e-01 8.36720169e-02 1.05560291e+00 5.25941908e-01
8.16009521e-01 -1.17105258e+00 -5.22422373e-01 3.16273987e-01
3.53857398e-01 -4.22956795e-01 5.07920921e-01 4.04163837e-01
-6.78626716e-01 1.83502778e-01 -3.15691203e-01 -5.33188641e-01
-9.04943347e-01 8.01000118e-01 1.50961861e-01 1.19964138e-01
-6.76872909e-01 8.41358781e-01 6.58765078e-01 9.53859091e-03
1.28230840e-01 -4.80014145e-01 4.03965712e-01 -3.37155819e-01
4.46479946e-01 7.88481236e-02 -3.13011467e-01 -3.18652809e-01
-1.15431614e-01 8.96826267e-01 -9.23590809e-02 -6.92604542e-01
1.06600344e+00 -4.72484231e-01 -2.95742564e-02 1.49273232e-01
1.25333261e+00 1.12389818e-01 -1.85376704e+00 -3.75840038e-01
-3.93899441e-01 -8.73086274e-01 1.94709137e-01 -1.23362637e+00
-1.41062737e+00 1.13876843e+00 7.29317904e-01 -5.93625717e-02
1.32127130e+00 -2.02948689e-01 7.91605413e-01 5.16568460e-02
2.55333185e-01 -1.37003982e+00 7.26632178e-01 9.89642888e-02
7.66631007e-01 -1.13768315e+00 -1.42200097e-01 -2.59874284e-01
-1.20480204e+00 1.20310831e+00 6.38858080e-01 -4.68941815e-02
4.23575401e-01 1.92193463e-02 7.03986049e-01 -1.15959972e-01
-4.54889536e-01 -1.87167466e-01 3.40616107e-01 6.55908346e-01
1.84778631e-01 4.52268161e-02 1.60540044e-01 9.42661241e-03
3.44052106e-01 -1.33160681e-01 4.80414450e-01 5.84507346e-01
-4.90016162e-01 -1.19551051e+00 -4.27757144e-01 -1.02904156e-01
6.67657424e-03 -2.07591206e-01 -4.67238486e-01 4.69913572e-01
-1.76279798e-01 9.89353001e-01 -1.70711994e-01 -1.45330518e-01
1.59679219e-01 -1.39511377e-01 5.08197188e-01 -2.79601097e-01
-1.09062076e-01 4.89110142e-01 -4.27517831e-01 -8.59062016e-01
-3.29315603e-01 -2.17623815e-01 -1.28493738e+00 -2.07334504e-01
-2.59492341e-02 -7.14517385e-02 9.33146060e-01 4.71126497e-01
2.24503532e-01 4.62050378e-01 6.70200408e-01 -9.34268653e-01
-1.26528755e-01 -5.86312234e-01 -8.57579947e-01 8.93807948e-01
6.66628852e-02 -2.48302907e-01 -9.53172147e-02 4.86576229e-01] | [11.394251823425293, -1.0477451086044312] |
6c027f9c-70c7-476b-95e3-c234c0d82def | hyper-parameter-sweep-on-alphazero-general | 1903.08129 | null | http://arxiv.org/abs/1903.08129v1 | http://arxiv.org/pdf/1903.08129v1.pdf | Hyper-Parameter Sweep on AlphaZero General | Since AlphaGo and AlphaGo Zero have achieved breakground successes in the
game of Go, the programs have been generalized to solve other tasks.
Subsequently, AlphaZero was developed to play Go, Chess and Shogi. In the
literature, the algorithms are explained well. However, AlphaZero contains many
parameters, and for neither AlphaGo, AlphaGo Zero nor AlphaZero, there is
sufficient discussion about how to set parameter values in these algorithms.
Therefore, in this paper, we choose 12 parameters in AlphaZero and evaluate how
these parameters contribute to training. We focus on three objectives~(training
loss, time cost and playing strength). For each parameter, we train 3 models
using 3 different values~(minimum value, default value, maximum value). We use
the game of play 6$\times$6 Othello, on the AlphaZeroGeneral open source
re-implementation of AlphaZero. Overall, experimental results show that
different values can lead to different training results, proving the importance
of such a parameter sweep. We categorize these 12 parameters into
time-sensitive parameters and time-friendly parameters. Moreover, through
multi-objective analysis, this paper provides an insightful basis for further
hyper-parameter optimization. | ['Hui Wang', 'Aske Plaat', 'Mike Preuss', 'Michael Emmerich'] | 2019-03-19 | null | null | null | null | ['game-of-go'] | ['playing-games'] | [-2.19618633e-01 -2.02027261e-01 -1.31589115e-01 -1.77627087e-01
-5.92676342e-01 -5.87545812e-01 2.41630594e-03 -6.66057542e-02
-7.70701051e-01 8.51797760e-01 -4.62293446e-01 -5.04105389e-01
-6.50663793e-01 -8.23000789e-01 -5.13340890e-01 -7.37312853e-01
-4.41523314e-01 4.91978556e-01 4.62706774e-01 -6.75094903e-01
6.20604336e-01 1.04657359e-01 -1.67516840e+00 8.86348560e-02
9.89168346e-01 6.54702604e-01 2.58480012e-01 8.27558100e-01
1.15600385e-01 5.51350892e-01 -9.56664562e-01 -7.09783256e-01
4.05815244e-01 -2.79231697e-01 -9.53656316e-01 -4.10907120e-01
-6.70844987e-02 -1.61206812e-01 3.24612334e-02 9.02938902e-01
7.32984960e-01 4.39055651e-01 3.80428493e-01 -1.47120714e+00
3.18647735e-02 7.81153619e-01 -4.46767241e-01 4.94789153e-01
3.61300826e-01 5.91336310e-01 1.14025235e+00 -3.97396713e-01
2.73303390e-01 9.01638091e-01 7.05143869e-01 3.18067670e-01
-9.30583239e-01 -7.66552091e-01 7.39213675e-02 1.98909223e-01
-1.47629690e+00 -1.83561504e-01 4.26650226e-01 -3.39666128e-01
1.17269588e+00 5.35031140e-01 7.96687603e-01 9.35889304e-01
4.72542048e-01 5.88060200e-01 1.06024277e+00 -6.78630292e-01
2.65382320e-01 -4.08964530e-02 2.39746422e-02 7.39252746e-01
2.47447073e-01 1.89324036e-01 -3.99929971e-01 1.69685464e-02
7.92789936e-01 -5.79725981e-01 -1.13041475e-01 -2.32981965e-01
-9.47596252e-01 9.67258930e-01 3.05183858e-01 3.08886886e-01
-4.33764644e-02 4.67821658e-01 3.28820050e-01 5.96909702e-01
2.61801571e-01 9.73728120e-01 -3.61994773e-01 -6.78045332e-01
-6.11196041e-01 8.59072149e-01 8.58313918e-01 7.74371743e-01
6.37903631e-01 2.60186996e-02 -6.78284839e-03 8.69775593e-01
1.37186825e-01 9.79803279e-02 4.67580169e-01 -1.09374797e+00
8.28893244e-01 3.00322175e-01 1.19547211e-01 -1.03734076e+00
-6.73370898e-01 -4.81925726e-01 -4.39637154e-01 4.25759584e-01
4.94626969e-01 -4.38692600e-01 -6.14172161e-01 1.65391767e+00
8.82703289e-02 9.94508341e-02 -1.36671707e-01 8.82689178e-01
8.41289222e-01 4.09708172e-01 7.13263974e-02 1.62874222e-01
1.09337437e+00 -1.17395461e+00 -4.61402088e-01 -3.12461704e-01
9.08769488e-01 -6.78081632e-01 1.57168400e+00 7.90902674e-01
-1.49297631e+00 -2.76788414e-01 -1.11754155e+00 2.42761463e-01
-4.54527646e-01 1.66843198e-02 1.08399451e+00 9.29695487e-01
-1.05382764e+00 8.63611102e-01 -7.46149540e-01 -2.38473967e-01
-9.17372555e-02 8.76580119e-01 -6.08858988e-02 1.86743051e-01
-1.37554729e+00 9.81362522e-01 8.18343401e-01 -2.58432664e-02
-6.49966002e-01 -2.85669565e-01 -5.35250366e-01 2.22059846e-01
7.81330884e-01 -8.13871324e-01 1.40090454e+00 -5.97094476e-01
-1.82311714e+00 9.16791797e-01 4.76672620e-01 -3.34462315e-01
5.42408586e-01 -5.98260053e-02 -2.44911596e-01 -1.28550321e-01
-4.35136743e-02 6.53501987e-01 3.55727434e-01 -9.27573025e-01
-8.61675858e-01 1.03811927e-01 9.00506794e-01 4.37577575e-01
-3.12809385e-02 2.87929296e-01 -8.04957151e-01 -6.18955731e-01
-2.04178900e-01 -8.19214702e-01 -4.61698592e-01 -5.90468645e-01
-1.80788070e-01 -1.45049796e-01 -1.96829811e-01 -2.17329085e-01
1.88696635e+00 -2.05277181e+00 1.07877344e-01 4.34023350e-01
1.28367886e-01 1.72771394e-01 -7.25462586e-02 4.95036781e-01
-6.74103573e-02 4.62651402e-01 -4.26457748e-02 -2.04180162e-02
1.39140502e-01 2.78352201e-01 3.39633584e-01 1.57920808e-01
-2.29497895e-01 6.85923457e-01 -9.10181224e-01 -2.92576045e-01
1.69986084e-01 1.80476625e-02 -1.14217544e+00 7.19677433e-02
-1.21424031e-02 -1.48664545e-02 -5.83505273e-01 5.69461703e-01
3.77214432e-01 -4.58449610e-02 -4.44147289e-02 2.92378277e-01
-2.20350549e-01 2.79466987e-01 -1.31696272e+00 1.58039117e+00
-4.64562654e-01 2.52714485e-01 8.85779932e-02 -8.92636240e-01
6.39142275e-01 8.71364996e-02 3.31553727e-01 -5.80623865e-01
6.32718146e-01 2.89038271e-01 2.05880687e-01 -6.63843811e-01
8.12621355e-01 -3.66695859e-02 -1.49503499e-01 1.59034356e-01
-2.07613483e-02 -2.68298000e-01 6.53833926e-01 -6.05831780e-02
1.10849166e+00 -5.84684033e-03 1.78797513e-01 -2.78581798e-01
2.59535491e-01 1.76275864e-01 2.54267335e-01 9.06008065e-01
-1.49680972e-01 6.59576237e-01 7.60398388e-01 -1.27931580e-01
-8.13611925e-01 -7.38871336e-01 5.72759174e-02 1.58740735e+00
3.07485849e-01 -7.46495664e-01 -9.74581242e-01 -3.83112520e-01
-1.72022149e-01 8.52515578e-01 -5.62720954e-01 -2.20211774e-01
-7.49179840e-01 -1.08044994e+00 6.45417273e-01 2.82368451e-01
5.51595032e-01 -1.16846931e+00 -8.56594563e-01 2.21814036e-01
-2.47643784e-01 -4.76993740e-01 -3.11292231e-01 3.68353248e-01
-8.47396135e-01 -1.10165727e+00 -4.23724622e-01 -6.20947778e-01
2.93426692e-01 2.24741384e-01 1.53438151e+00 4.16886210e-01
-9.22796205e-02 2.70702928e-01 -5.47841787e-01 -3.78693163e-01
-8.32769945e-02 6.41396284e-01 -1.31041065e-01 -7.20979869e-01
-8.15776214e-02 -6.05367064e-01 -4.64760512e-01 5.82847834e-01
-8.17297518e-01 -6.83222199e-03 4.61234361e-01 8.29261541e-01
2.48708174e-01 2.64028132e-01 3.10199142e-01 -1.03982925e+00
1.09422171e+00 -5.47689259e-01 -7.00576186e-01 1.38414457e-01
-7.46813834e-01 2.48299055e-02 4.30682331e-01 -3.35528553e-01
-6.19197071e-01 -4.89081532e-01 -5.27860165e-01 -3.36065024e-01
2.27980062e-01 7.19427288e-01 -2.13335022e-01 -2.87534207e-01
7.08062887e-01 -2.27625400e-01 -3.00295860e-01 -4.99273092e-01
2.28050441e-01 2.11800203e-01 3.86941165e-01 -9.48079467e-01
5.17425060e-01 -1.93685651e-01 -3.33615333e-01 -2.88316309e-01
-4.76762056e-01 -4.26248193e-01 -9.21255127e-02 -2.46996179e-01
6.09240651e-01 -5.74547708e-01 -1.12496018e+00 4.05378848e-01
-6.22902513e-01 -5.56397319e-01 -2.22532645e-01 3.48624021e-01
-6.89902008e-01 2.79499330e-02 -2.97276497e-01 -5.95820606e-01
-1.63647398e-01 -1.42096484e+00 5.78454792e-01 3.27307492e-01
-3.35285366e-02 -1.10306633e+00 -1.59099121e-02 4.13307935e-01
3.44343305e-01 1.58994198e-01 6.77722514e-01 -3.67326498e-01
-3.67086440e-01 -9.58231613e-02 1.58441752e-01 1.22456200e-01
-3.28447312e-01 8.24134871e-02 -5.60437143e-01 -5.59189975e-01
-2.80699492e-01 -3.47314268e-01 6.41460896e-01 5.27310431e-01
1.40227735e+00 -2.93294340e-01 -2.27973506e-01 1.13181210e+00
1.53368902e+00 4.52456981e-01 7.07228482e-01 1.18176174e+00
2.96763867e-01 2.79859990e-01 9.42358911e-01 4.90983397e-01
3.57330769e-01 8.07161987e-01 8.05666268e-01 -8.81106332e-02
4.47060794e-01 2.94753276e-02 3.33600372e-01 6.95184529e-01
-6.91367745e-01 -6.62646711e-01 -8.07155192e-01 3.71964991e-01
-1.64699244e+00 -8.94156575e-01 -7.46287704e-02 2.27737284e+00
6.69972479e-01 5.62380612e-01 4.74147767e-01 2.79696137e-01
4.50699598e-01 5.14676943e-02 -1.82030275e-01 -6.73987031e-01
-3.81913073e-02 4.73353744e-01 8.07650924e-01 6.06866241e-01
-1.03503907e+00 1.08245504e+00 7.66872883e+00 1.18851328e+00
-1.04800236e+00 -8.34250897e-02 6.11534417e-01 -4.99240369e-01
-2.19726771e-01 1.01649977e-01 -8.34257901e-01 4.59444672e-01
8.59162390e-01 -4.60751086e-01 8.06717336e-01 7.56876707e-01
2.64480978e-01 -3.73331428e-01 -7.30922163e-01 9.49195147e-01
-2.37179339e-01 -1.11789882e+00 -2.64225751e-01 2.74999458e-02
4.00123775e-01 -1.30004615e-01 4.76057082e-02 7.95508206e-01
6.52504742e-01 -1.18787038e+00 8.42798173e-01 1.70257270e-01
5.74091852e-01 -1.18827510e+00 8.82847369e-01 3.08448911e-01
-9.45895314e-01 -1.97857380e-01 -3.65950376e-01 -3.97134006e-01
-6.71550259e-02 1.82416424e-01 -6.22340679e-01 8.08523536e-01
9.27222967e-01 1.57098591e-01 -6.63463116e-01 1.49104607e+00
-3.18792731e-01 7.16248393e-01 -3.78861159e-01 -1.08732633e-01
4.97817189e-01 -2.90117264e-01 4.02425468e-01 9.99937356e-01
3.26691270e-01 3.11743408e-01 2.48985812e-01 5.99891663e-01
2.34568790e-01 1.55346841e-01 -1.67773038e-01 1.77384719e-01
3.08986813e-01 1.08013129e+00 -7.68796206e-01 4.35972027e-02
-5.36295995e-02 5.97051799e-01 2.65647054e-01 2.77037919e-01
-1.30314434e+00 -7.84884870e-01 6.52471244e-01 1.61740884e-01
1.05730228e-01 -1.41449720e-01 -4.12185907e-01 -9.80161965e-01
-5.19498289e-02 -1.08135355e+00 5.43695271e-01 -7.51093090e-01
-8.47800553e-01 6.23854160e-01 5.60609162e-01 -1.10701752e+00
-4.05331016e-01 -6.98006034e-01 -7.86138117e-01 8.58091533e-01
-1.12834263e+00 -5.59454978e-01 -2.34321654e-01 2.98073769e-01
4.32720095e-01 -1.58723220e-01 5.27697265e-01 5.26828766e-01
-8.39058340e-01 9.55477834e-01 1.55309394e-01 -2.37612560e-01
5.71579874e-01 -1.29501641e+00 4.41073179e-01 5.17552018e-01
-2.85167873e-01 5.93803644e-01 9.80102539e-01 -1.48953691e-01
-9.79911327e-01 -4.34221953e-01 6.66884124e-01 -2.42196754e-01
5.92263997e-01 -2.43330851e-01 -7.58350730e-01 5.43098092e-01
1.67440876e-01 -5.08267283e-01 5.02080441e-01 3.54531586e-01
3.58209670e-01 -1.57668274e-02 -1.08995032e+00 7.58133948e-01
1.23599470e+00 1.90776423e-01 -3.10695678e-01 7.13672265e-02
5.24275899e-01 -8.58354032e-01 -9.15604472e-01 2.13894695e-01
4.96285379e-01 -1.44078457e+00 1.16549706e+00 -6.42014086e-01
3.46183360e-01 -2.74395421e-02 2.33632922e-01 -1.51555610e+00
-4.11601752e-01 -6.43952668e-01 3.29710931e-01 1.03697300e+00
5.19493103e-01 -7.50826418e-01 8.63333404e-01 5.97724319e-01
-4.06991661e-01 -1.12813985e+00 -8.86334777e-01 -8.01259220e-01
2.71932185e-01 -6.06554925e-01 7.05536604e-01 8.38290572e-01
6.30320758e-02 1.85844094e-01 -4.72617775e-01 -1.61185607e-01
1.53499678e-01 -2.21920803e-01 9.29644704e-01 -9.72563565e-01
-7.40553081e-01 -8.60721171e-01 -1.92522138e-01 -1.11775601e+00
-2.90865719e-01 -5.62567890e-01 -1.08813636e-01 -1.24143422e+00
-6.08014539e-02 -8.74085665e-01 -2.24718556e-01 8.27651501e-01
-4.09963280e-01 1.76856458e-01 4.19948995e-01 1.18199714e-01
-5.93380332e-01 1.99585766e-01 1.29334915e+00 -2.26859432e-02
-3.95164758e-01 2.67869413e-01 -8.76893103e-01 6.87709391e-01
9.33875680e-01 -4.68796581e-01 -5.70077598e-01 -7.22412586e-01
6.11494303e-01 6.90592676e-02 8.55373740e-02 -1.20117342e+00
-6.82732463e-02 -5.68257391e-01 -2.37332493e-01 -2.02471986e-01
3.09918970e-01 -5.19634843e-01 1.74347445e-01 3.52298588e-01
-1.00273475e-01 3.85052234e-01 4.62682217e-01 5.96343949e-02
-2.53439814e-01 -8.16670001e-01 6.44236743e-01 -2.27239668e-01
-7.23287642e-01 -1.38093783e-02 -5.68544805e-01 2.57849872e-01
1.00222683e+00 -5.80006361e-01 -4.31472249e-02 -3.55230987e-01
-8.81859064e-01 6.39331639e-01 4.68745142e-01 2.48760164e-01
2.55429834e-01 -9.78998840e-01 -5.31441092e-01 2.37160567e-02
-2.77611986e-02 4.01926674e-02 3.04147542e-01 7.25565612e-01
-1.02049804e+00 3.22849691e-01 -3.96897644e-01 -3.41565728e-01
-1.32315302e+00 2.56727099e-01 5.02252340e-01 -6.91825151e-01
-3.45539987e-01 1.02026045e+00 7.55397230e-02 -4.27563757e-01
3.02301675e-01 -6.03930235e-01 -4.00418103e-01 -2.24783227e-01
1.77935749e-01 6.27744913e-01 2.16952726e-01 -6.67317882e-02
-2.34473482e-01 4.30965930e-01 1.03494830e-01 -1.88156188e-01
1.35221493e+00 6.59490079e-02 2.14497998e-01 2.22053051e-01
6.21007144e-01 -8.29019696e-02 -1.03225791e+00 4.97415155e-01
-7.96231255e-02 -7.25200415e-01 5.33378385e-02 -8.82308781e-01
-1.14638460e+00 5.92316687e-01 5.33967316e-01 2.12644055e-01
1.40871334e+00 -4.00379241e-01 3.96658748e-01 3.27372819e-01
5.75791717e-01 -1.00560033e+00 -1.39310136e-01 8.24649453e-01
5.57810247e-01 -9.81749654e-01 7.17826784e-02 -4.60043579e-01
-6.40075207e-01 9.63092208e-01 8.83416355e-01 -1.06355280e-01
2.62608677e-01 1.81066453e-01 -5.10138944e-02 -2.56679475e-01
-8.44104588e-01 -2.18239754e-01 -7.90741667e-02 4.20822203e-01
4.60015237e-01 -8.78887922e-02 -8.37708235e-01 8.33178759e-01
-7.73286283e-01 1.64547488e-01 5.15366018e-01 1.08448696e+00
-5.16759694e-01 -1.38937485e+00 -6.61542892e-01 3.22924435e-01
-6.34289443e-01 -1.92560345e-01 -2.02295646e-01 1.28873324e+00
3.92141908e-01 9.27385032e-01 -9.29642469e-02 -6.05414867e-01
6.55751288e-01 -2.86220908e-01 6.16742909e-01 -3.22443247e-01
-1.27272153e+00 -1.65632099e-01 4.21563059e-01 -3.99659514e-01
-1.09163761e-01 -2.70970404e-01 -1.13897920e+00 -9.42996204e-01
-6.62782192e-01 4.95984495e-01 4.99220818e-01 7.98149407e-01
-1.49455303e-02 7.40774572e-01 4.43230510e-01 -8.49503577e-01
-4.63516355e-01 -7.80057490e-01 -5.93251050e-01 1.55035272e-01
-4.12781119e-01 -1.06629062e+00 -3.85602146e-01 -4.17755544e-01] | [3.4925832748413086, 1.4561840295791626] |
d098753c-1775-47d5-ae7a-d1bd20bb2eba | robust-gyroscope-aided-camera-self | 1805.12506 | null | http://arxiv.org/abs/1805.12506v1 | http://arxiv.org/pdf/1805.12506v1.pdf | Robust Gyroscope-Aided Camera Self-Calibration | Camera calibration for estimating the intrinsic parameters and lens
distortion is a prerequisite for various monocular vision applications
including feature tracking and video stabilization. This application paper
proposes a model for estimating the parameters on the fly by fusing gyroscope
and camera data, both readily available in modern day smartphones. The model is
based on joint estimation of visual feature positions, camera parameters, and
the camera pose, the movement of which is assumed to follow the movement
predicted by the gyroscope. Our model assumes the camera movement to be free,
but continuous and differentiable, and individual features are assumed to stay
stationary. The estimation is performed online using an extended Kalman filter,
and it is shown to outperform existing methods in robustness and insensitivity
to initialization. We demonstrate the method using simulated data and empirical
data from an iPad. | ['Santiago Cortés Reina', 'Juho Kannala', 'Arno Solin'] | 2018-05-31 | null | null | null | null | ['video-stabilization'] | ['computer-vision'] | [ 2.83090919e-02 -2.96630055e-01 -1.53416008e-01 -1.41228884e-01
4.42440845e-02 -8.98851573e-01 6.47939146e-01 -7.12516665e-01
-3.95969868e-01 5.33668399e-01 -3.72505695e-01 -1.16213925e-01
2.22531825e-01 2.75213625e-02 -9.30814207e-01 -4.35419261e-01
3.18003625e-01 -9.09468308e-02 1.38767898e-01 4.01998878e-01
4.23895389e-01 7.05305636e-01 -1.47727716e+00 -1.05726683e+00
6.20530069e-01 8.44186783e-01 4.76642102e-01 1.26703036e+00
7.02237785e-01 4.05228406e-01 -4.62010652e-01 -1.80114776e-01
5.71904480e-01 1.75398141e-01 -1.66816749e-02 8.94835889e-01
4.07188237e-01 -8.17520261e-01 -5.77850640e-01 1.22449338e+00
1.83706880e-01 -5.16145900e-02 3.45208883e-01 -1.36609709e+00
-1.76738381e-01 -6.62665844e-01 -4.30929363e-01 -2.37458609e-02
7.08602607e-01 2.57010162e-01 2.90477335e-01 -8.13815355e-01
7.55102932e-01 8.28523040e-01 8.33691537e-01 1.81123450e-01
-8.11880231e-01 -2.90561289e-01 5.75685315e-02 -1.48952886e-01
-1.51226962e+00 -6.56908274e-01 4.52498555e-01 -6.98851168e-01
7.25421309e-01 -5.19058667e-02 8.73459637e-01 7.83938169e-01
4.70557600e-01 3.19521874e-01 7.33946979e-01 -5.07783234e-01
6.90748990e-02 4.11487341e-01 1.21589422e-01 6.99869871e-01
7.99712658e-01 2.30467200e-01 -3.71596396e-01 -1.61030754e-01
1.42168355e+00 -1.24804219e-02 -5.39769769e-01 -1.00252330e+00
-1.29314041e+00 3.99487853e-01 -2.55389929e-01 -4.72069919e-01
-3.99399996e-01 2.23832548e-01 -2.78721482e-01 1.29462034e-01
7.44819492e-02 7.15594962e-02 -3.85173559e-01 -6.51600242e-01
-5.13892651e-01 -5.07835560e-02 9.37192023e-01 1.78521395e+00
5.03347337e-01 6.81632385e-02 7.47734368e-01 2.45275140e-01
6.96973741e-01 1.17713165e+00 6.17377877e-01 -1.19011402e+00
3.66763175e-01 2.80531555e-01 8.85532141e-01 -9.04720843e-01
-3.05184513e-01 -4.81559187e-02 -2.99382359e-01 1.55536518e-01
4.79437590e-01 -5.78957379e-01 -6.29215002e-01 1.30472124e+00
5.62837124e-01 5.37386596e-01 -5.03548868e-02 8.37091506e-01
3.41091365e-01 3.73066723e-01 -8.46526146e-01 -5.47625959e-01
9.16099370e-01 -5.55079043e-01 -9.12440002e-01 -3.05168420e-01
1.16253749e-01 -1.01325989e+00 6.79509521e-01 5.12315333e-01
-8.77191544e-01 -6.61277890e-01 -1.43818867e+00 1.87071916e-02
1.07728228e-01 6.95627272e-01 1.88143194e-01 8.94927025e-01
-9.37226832e-01 1.93621621e-01 -1.26275575e+00 -7.03389645e-01
-4.55430299e-01 6.99596167e-01 -3.06214690e-01 3.92567575e-01
-5.75889707e-01 9.86084759e-01 -2.51589417e-02 1.51947662e-01
-4.16375339e-01 1.38699859e-02 -9.47294474e-01 -3.24476749e-01
3.15967143e-01 -7.60417223e-01 1.58679163e+00 -7.99578071e-01
-2.31092143e+00 5.08761764e-01 -4.94980931e-01 -2.98781216e-01
6.10991299e-01 -6.10114753e-01 -3.74861717e-01 2.06575438e-01
-2.74816543e-01 2.19978288e-01 1.45019448e+00 -8.41198742e-01
-5.92753708e-01 -2.22756088e-01 -4.06531803e-02 7.58590579e-01
-1.76820084e-01 -3.31658363e-01 -1.06278694e+00 -4.84148115e-02
3.26846063e-01 -1.58971155e+00 -8.23198706e-02 1.73907459e-01
-4.40892577e-01 6.01184785e-01 9.06607807e-01 -4.86453593e-01
8.90219212e-01 -2.03858256e+00 -6.33892938e-02 2.05400094e-01
-1.28944293e-01 1.05690382e-01 5.00004172e-01 3.96537520e-02
1.88610777e-01 -6.40455484e-01 5.66533148e-01 -3.90960872e-01
-3.54569852e-01 5.21489419e-03 -2.69865245e-01 1.11991227e+00
-4.64952976e-01 4.18702543e-01 -7.18887329e-01 2.74720900e-02
7.68141687e-01 5.73741198e-01 -9.19885561e-02 3.31846535e-01
3.65369201e-01 4.31150794e-01 -2.75710493e-01 4.85397428e-01
7.01998651e-01 -3.43377560e-01 2.81882167e-01 -2.34517917e-01
-3.82054210e-01 1.17491007e-01 -1.88478315e+00 1.13528764e+00
-1.93909526e-01 1.04462230e+00 1.41491085e-01 -1.82693928e-01
7.37208545e-01 3.49258482e-01 4.00942951e-01 -9.50926915e-02
2.77268708e-01 -9.33963284e-02 -2.24117607e-01 -4.93048072e-01
8.92415166e-01 5.19488871e-01 3.31483871e-01 3.19143206e-01
-4.61392179e-02 -6.32386506e-01 -1.01531662e-01 8.82615596e-02
4.16812956e-01 4.56414580e-01 7.77078867e-01 5.32355793e-02
4.43933666e-01 -1.96656615e-01 4.39537317e-01 6.12625599e-01
-1.33929074e-01 6.10296249e-01 -8.28698575e-02 3.04902401e-02
-1.10934246e+00 -8.51615429e-01 -1.16373368e-01 -1.13550067e-01
7.37929463e-01 -3.11937541e-01 -8.95072937e-01 -4.53433655e-02
3.14507298e-02 1.40784323e-01 -1.58240229e-01 2.88978331e-02
-3.10234845e-01 -3.57119858e-01 -2.03019485e-01 4.73818839e-01
5.04656494e-01 7.70696923e-02 -9.88977194e-01 6.02188408e-02
1.01848252e-01 -1.55047071e+00 -8.90685260e-01 -4.09999341e-01
-9.06681180e-01 -1.50020623e+00 -4.77898479e-01 -3.63691688e-01
1.01587379e+00 1.03301585e+00 5.32590032e-01 -5.08660972e-02
1.07822299e-01 1.26887310e+00 2.10276902e-01 -6.72775626e-01
7.35849068e-02 -4.14978296e-01 7.23753870e-01 3.17381531e-01
1.35936320e-01 -1.43746054e-02 -5.34293354e-01 7.08924294e-01
-3.96757096e-01 1.76254481e-01 -7.72133917e-02 5.73631108e-01
6.64091229e-01 -1.08791880e-01 -4.10603762e-01 -3.87770414e-01
3.35767090e-01 -6.55740574e-02 -1.66109395e+00 8.06664248e-05
-7.45905638e-01 -4.18749720e-01 3.27474803e-01 -7.77138829e-01
-1.12972867e+00 6.63160264e-01 6.46345377e-01 -6.08618677e-01
-6.28571957e-02 1.45473570e-01 -2.22683206e-01 -6.28793776e-01
3.60840499e-01 1.06518768e-01 4.05900002e-01 -1.77667752e-01
3.07693869e-01 8.32651675e-01 8.52722943e-01 2.52133477e-02
1.01104569e+00 7.38334179e-01 1.87778816e-01 -1.46804678e+00
-1.47403523e-01 -7.12369382e-01 -1.03192627e+00 -3.92030001e-01
4.55064982e-01 -1.36953163e+00 -1.10595691e+00 9.77280200e-01
-1.04097891e+00 3.38621736e-02 7.23387301e-02 1.21726453e+00
-7.05857992e-01 6.36975408e-01 -2.56725878e-01 -9.61002886e-01
1.49204567e-01 -1.32802284e+00 9.40701723e-01 7.13262975e-01
2.91767903e-03 -1.03693092e+00 1.17095947e-01 -1.51357740e-01
-5.69841936e-02 2.13989113e-02 -2.54816562e-01 1.16603650e-01
-1.04203629e+00 -7.45984018e-01 2.01579347e-01 3.91960144e-01
4.34439003e-01 6.13599718e-01 -7.56595731e-01 -7.04286814e-01
3.74517381e-01 3.94253850e-01 -1.13958828e-01 8.27245474e-01
4.90715981e-01 -1.43948004e-01 -5.11672020e-01 1.07646000e+00
1.36828840e+00 5.14076531e-01 3.06628942e-01 3.93913090e-01
8.27407062e-01 3.90159935e-02 8.04398119e-01 5.27775347e-01
5.96666396e-01 9.07065511e-01 5.43288410e-01 3.17030966e-01
2.31204331e-01 -2.41137534e-01 5.71842909e-01 8.65420580e-01
-1.49345919e-01 -2.40455344e-02 -4.42576557e-01 2.96965152e-01
-1.69769847e+00 -6.22604847e-01 -5.01002073e-01 2.82478023e+00
5.78244403e-02 3.63541953e-02 1.38086662e-01 -2.35586807e-01
9.42937970e-01 -2.94783413e-01 -9.57352638e-01 2.53355200e-03
8.88597742e-02 -4.37832773e-01 1.16615415e+00 9.08789575e-01
-1.08523989e+00 6.29118562e-01 7.21375656e+00 -4.51435030e-01
-1.12175322e+00 -4.05628055e-01 -2.07661927e-01 -1.82068795e-01
4.29191768e-01 3.34390253e-01 -1.23791444e+00 6.29426479e-01
9.59449291e-01 -3.95796090e-01 5.19449651e-01 8.96275580e-01
2.93674916e-01 -6.16251409e-01 -9.83339667e-01 1.39446390e+00
2.90472150e-01 -9.67825532e-01 -5.39291680e-01 2.66686350e-01
7.04155564e-01 3.12230531e-02 1.18717678e-01 -4.02489364e-01
1.80414259e-01 -2.25173384e-01 6.07612789e-01 8.12746406e-01
7.22410798e-01 -4.57308203e-01 4.78993833e-01 4.68752444e-01
-1.10606873e+00 -5.21357022e-02 -4.30328518e-01 -3.39144439e-01
3.58906180e-01 1.80526927e-01 -1.09377575e+00 2.98774570e-01
3.48593593e-01 1.08668780e+00 -5.24412692e-01 1.33420956e+00
-2.58597344e-01 2.66786128e-01 -5.62563479e-01 6.43922389e-03
-3.66966099e-01 -7.47799635e-01 8.39048624e-01 5.66989839e-01
6.32165194e-01 5.91889620e-02 3.53021882e-02 -1.73704866e-02
2.86767930e-01 -2.93318897e-01 -8.87056530e-01 2.65200406e-01
6.32425547e-01 1.02397168e+00 -2.95565814e-01 -3.10995370e-01
-8.18220913e-01 1.01656830e+00 -3.60625088e-01 4.99057531e-01
-7.67751694e-01 -4.68556911e-01 8.66207004e-01 3.24622951e-02
5.07230818e-01 -8.60137582e-01 -1.70876924e-02 -1.68074465e+00
2.01041773e-01 -5.40497839e-01 -1.59665555e-01 -1.13453197e+00
-4.71393496e-01 2.47960001e-01 1.27724275e-01 -1.97182083e+00
-7.96182990e-01 -7.68500507e-01 -3.51796567e-01 7.75680959e-01
-1.19300497e+00 -8.64613831e-01 -5.12345493e-01 7.73376107e-01
2.82502919e-01 -1.91213652e-01 4.34531033e-01 -2.08582684e-01
-6.14877999e-01 3.07791293e-01 5.74077070e-01 -1.02456339e-01
6.93058074e-01 -1.12281060e+00 5.46489596e-01 1.28657424e+00
3.30491997e-02 1.06627548e+00 1.05635071e+00 -6.70868695e-01
-1.91794145e+00 -5.91034710e-01 4.71349090e-01 -8.56428146e-01
6.72827721e-01 -2.09843487e-01 -3.41742277e-01 1.16343582e+00
3.17961387e-02 1.99649841e-01 1.28468931e-01 -3.73960763e-01
3.20625693e-01 -1.90833598e-01 -9.36503112e-01 4.28965479e-01
5.56616604e-01 -6.12454534e-01 -3.95276964e-01 3.97048563e-01
3.26570213e-01 -1.26556349e+00 -6.02687240e-01 -1.28311157e-01
9.35690045e-01 -7.73018897e-01 8.74280930e-01 1.02384970e-01
-7.31837451e-01 -6.32693052e-01 -1.49719104e-01 -1.17786264e+00
-1.06884427e-01 -1.13106513e+00 -5.66744089e-01 9.33322132e-01
-1.20327823e-01 -9.29993987e-01 6.51818097e-01 9.46542919e-01
3.49161506e-01 1.41018137e-01 -8.89085531e-01 -9.02279317e-01
-8.55479836e-01 -3.27075124e-01 3.82318497e-01 5.22065639e-01
-9.45159867e-02 3.75868469e-01 -7.59027719e-01 9.43379521e-01
8.44281018e-01 -6.25783503e-02 1.54368603e+00 -1.23543596e+00
-2.65625298e-01 2.89315552e-01 -6.62953138e-01 -1.73367465e+00
-1.84980948e-02 2.29148448e-01 -5.05013391e-02 -8.38652134e-01
-1.12554997e-01 2.02516928e-01 4.31721628e-01 -4.14081305e-01
9.68328305e-03 -9.05679241e-02 1.17854074e-01 2.95752913e-01
-2.55198449e-01 1.07671008e-01 6.99156284e-01 3.76739085e-01
-5.02586186e-01 7.72797883e-01 -1.14252493e-01 1.14290023e+00
3.09639782e-01 -2.36884635e-02 -7.78848946e-01 -5.32197475e-01
4.99660932e-02 3.12517196e-01 4.77199316e-01 -1.09244430e+00
7.14794219e-01 -8.32499191e-02 6.08493865e-01 -6.91721618e-01
7.82105982e-01 -1.22605252e+00 5.62174201e-01 2.83020169e-01
3.71675938e-01 5.13898969e-01 3.72106768e-02 6.68017805e-01
1.06503710e-01 -1.46977782e-01 6.13696992e-01 1.55947253e-01
-5.32485783e-01 3.59045297e-01 -4.13556695e-01 -5.39064646e-01
1.08212805e+00 -8.71320665e-01 -1.76390633e-01 -9.83941555e-01
-5.62077284e-01 -1.52377874e-01 1.08812797e+00 3.29366565e-01
6.24177337e-01 -1.15716720e+00 1.61990315e-01 6.74381316e-01
-8.93628076e-02 -1.69162676e-01 -1.24540001e-01 8.52572799e-01
-5.62297523e-01 6.63086474e-01 8.07427391e-02 -1.02674067e+00
-1.56774604e+00 4.97062773e-01 3.88052672e-01 6.13160193e-01
-3.73431325e-01 2.49923587e-01 -2.36514956e-01 -1.01873234e-01
1.86085850e-01 -5.52540600e-01 5.78067973e-02 -3.21711481e-01
5.31979144e-01 5.77998519e-01 -8.40036348e-02 -9.54568446e-01
-2.48171225e-01 1.08519077e+00 1.37857571e-01 -2.33426124e-01
8.78031254e-01 -1.15019739e+00 3.09028417e-01 5.69539368e-01
8.20516706e-01 1.59378916e-01 -1.98187554e+00 -2.70170003e-01
-3.07276696e-01 -8.74945939e-01 4.38534953e-02 -9.53328982e-02
-7.31051385e-01 5.07816613e-01 8.59289289e-01 -1.16659209e-01
7.73433089e-01 -6.43813372e-01 2.99952507e-01 5.83611667e-01
6.07408106e-01 -1.27667367e+00 -2.81813234e-01 5.52385509e-01
3.18866253e-01 -1.21433949e+00 5.78326404e-01 -4.55160320e-01
-5.43910801e-01 1.11229742e+00 6.93812013e-01 -2.01586589e-01
8.83499682e-01 1.89625889e-01 1.33283690e-01 4.79259580e-01
-4.88791466e-01 4.65589762e-02 2.64410168e-01 7.58704364e-01
7.39931921e-03 -5.17017804e-02 8.95671621e-02 -2.32664749e-01
1.67053426e-04 1.11664310e-01 1.18484604e+00 1.00439155e+00
-3.33862334e-01 -7.81786859e-01 -8.02474976e-01 1.40354782e-01
-1.57062083e-01 2.80187339e-01 -7.70656019e-02 9.99259174e-01
-3.89223039e-01 9.88748252e-01 1.47713184e-01 -7.86411017e-02
4.86870289e-01 -3.67058039e-01 6.06293976e-01 -2.46068820e-01
1.04698867e-01 3.35269541e-01 -1.26096681e-01 -6.38579190e-01
-4.93098944e-01 -1.09332204e+00 -7.96214938e-01 -1.43836811e-01
-7.26516187e-01 -1.70320481e-01 9.86525357e-01 8.16695988e-01
4.02993709e-01 -1.43605649e-01 9.36465204e-01 -1.10221303e+00
-6.63813233e-01 -7.60664642e-01 -5.85016012e-01 -2.14765035e-02
8.34230304e-01 -8.59189630e-01 -4.58672047e-01 5.08289516e-01] | [7.929126262664795, -2.1915507316589355] |
f2b9c1fb-4f67-48b5-837d-1adb32ddc82f | f3net-fusion-feedback-and-focus-for-salient | 1911.11445 | null | https://arxiv.org/abs/1911.11445v1 | https://arxiv.org/pdf/1911.11445v1.pdf | F3Net: Fusion, Feedback and Focus for Salient Object Detection | Most of existing salient object detection models have achieved great progress by aggregating multi-level features extracted from convolutional neural networks. However, because of the different receptive fields of different convolutional layers, there exists big differences between features generated by these layers. Common feature fusion strategies (addition or concatenation) ignore these differences and may cause suboptimal solutions. In this paper, we propose the F3Net to solve above problem, which mainly consists of cross feature module (CFM) and cascaded feedback decoder (CFD) trained by minimizing a new pixel position aware loss (PPA). Specifically, CFM aims to selectively aggregate multi-level features. Different from addition and concatenation, CFM adaptively selects complementary components from input features before fusion, which can effectively avoid introducing too much redundant information that may destroy the original features. Besides, CFD adopts a multi-stage feedback mechanism, where features closed to supervision will be introduced to the output of previous layers to supplement them and eliminate the differences between features. These refined features will go through multiple similar iterations before generating the final saliency maps. Furthermore, different from binary cross entropy, the proposed PPA loss doesn't treat pixels equally, which can synthesize the local structure information of a pixel to guide the network to focus more on local details. Hard pixels from boundaries or error-prone parts will be given more attention to emphasize their importance. F3Net is able to segment salient object regions accurately and provide clear local details. Comprehensive experiments on five benchmark datasets demonstrate that F3Net outperforms state-of-the-art approaches on six evaluation metrics. | ['Jun Wei', 'Shuhui Wang', 'Qingming Huang'] | 2019-11-26 | null | null | null | null | ['dichotomous-image-segmentation'] | ['computer-vision'] | [ 2.75677294e-01 9.55072418e-02 -1.61835536e-01 -4.56864327e-01
-3.97306323e-01 3.57834734e-02 3.84485930e-01 2.10534304e-01
-4.04244810e-01 6.29829764e-01 4.05978829e-01 1.48885831e-01
8.89273584e-02 -8.80847514e-01 -7.21607268e-01 -7.80869007e-01
9.67095345e-02 -5.14409184e-01 8.32509875e-01 -3.63701195e-01
3.91936034e-01 2.91556239e-01 -1.64143860e+00 4.66692477e-01
1.02260876e+00 1.21291292e+00 7.19406843e-01 -1.52863283e-02
-3.12192559e-01 6.96294010e-01 -5.27895868e-01 -1.59193546e-01
2.23314986e-01 -4.18078452e-01 -6.19024038e-01 -1.72771681e-02
2.65531957e-01 -5.72393119e-01 -2.40217999e-01 1.36511576e+00
4.95241284e-01 -1.29289841e-02 3.29825729e-01 -1.09113848e+00
-7.28785336e-01 7.65648127e-01 -8.15302968e-01 4.36899781e-01
8.98886099e-02 2.65077919e-01 9.84461069e-01 -1.00507116e+00
2.86176682e-01 1.47186971e+00 3.80485475e-01 3.25091034e-01
-9.35336947e-01 -6.26397669e-01 6.42248392e-01 2.69068033e-01
-1.35604632e+00 -1.43770382e-01 9.56004083e-01 -4.31682393e-02
6.89924836e-01 3.16520065e-01 7.01302707e-01 6.97429657e-01
2.47395739e-01 1.18113995e+00 7.87652433e-01 -1.39456019e-01
-8.32103565e-02 1.30247042e-01 3.56759392e-02 8.01488519e-01
2.87122399e-01 1.79823022e-02 -4.23929006e-01 2.40131840e-01
9.00325894e-01 3.18781644e-01 -5.59347153e-01 -3.14631283e-01
-1.39950097e+00 7.92241693e-01 1.22029829e+00 3.48544836e-01
-5.01027524e-01 -1.03369094e-01 2.63918966e-01 -6.47383034e-02
1.59285128e-01 2.18824357e-01 -4.20030653e-01 4.10561323e-01
-8.40618551e-01 1.56019241e-01 2.69728675e-02 8.40642810e-01
1.24733400e+00 -1.19203806e-01 -6.65200651e-01 6.57871008e-01
4.22990620e-01 1.64918393e-01 6.62477911e-01 -7.75719821e-01
4.98480111e-01 1.12543511e+00 -1.28434286e-01 -1.34010661e+00
-4.10064757e-01 -8.17401528e-01 -9.20942187e-01 3.29604357e-01
-5.18121291e-03 -3.27375382e-02 -1.03667808e+00 1.63873482e+00
1.75958753e-01 -2.07317551e-03 -1.11636512e-01 1.33878767e+00
1.16143692e+00 7.36167133e-01 1.29912153e-01 4.51118723e-02
1.14313900e+00 -1.17841947e+00 -5.07596076e-01 -4.15030897e-01
3.33127141e-01 -7.48833120e-01 9.58748341e-01 -1.31870648e-02
-1.10847867e+00 -9.02218223e-01 -1.27866852e+00 -3.06538880e-01
-3.77513349e-01 2.13507950e-01 5.67618132e-01 1.05736163e-02
-9.94252086e-01 5.41891634e-01 -6.12588108e-01 3.98961268e-02
7.33838260e-01 4.34056193e-01 -9.58971903e-02 8.10208991e-02
-1.33354282e+00 8.60352933e-01 5.61499596e-01 2.87791908e-01
-6.71605647e-01 -6.97451055e-01 -1.03850782e+00 3.53371769e-01
3.00858229e-01 -5.94061434e-01 1.05167067e+00 -1.26347136e+00
-1.09578753e+00 3.16500455e-01 -2.85040170e-01 -3.48672330e-01
4.25506204e-01 -2.68171519e-01 -3.40104163e-01 1.04833096e-01
3.36615115e-01 1.27280343e+00 8.49154115e-01 -1.18138003e+00
-1.09980357e+00 -6.56311139e-02 1.32037342e-01 3.91391784e-01
-4.03147191e-01 -3.22583728e-02 -4.31907654e-01 -7.88783371e-01
3.62038404e-01 -3.32976073e-01 -3.51026088e-01 2.33840510e-01
-4.92825091e-01 -1.01762123e-01 1.05012357e+00 -3.85217607e-01
1.30548692e+00 -2.28855681e+00 -6.53651431e-02 -3.13779945e-03
3.24640155e-01 3.90584648e-01 -2.33968660e-01 -1.30429164e-01
-3.07878423e-02 2.19472665e-02 -5.05796731e-01 -9.68605280e-02
-2.56626189e-01 -9.83298123e-02 -2.27060363e-01 9.35759097e-02
8.08674276e-01 1.01484728e+00 -9.70139682e-01 -6.75039649e-01
4.11470622e-01 5.82904935e-01 -4.06461895e-01 -2.24631876e-01
-9.16003212e-02 1.43416449e-01 -7.13464499e-01 6.21918082e-01
9.87122238e-01 -2.42888346e-01 -5.16506553e-01 -3.76918226e-01
-4.31614220e-01 3.22544634e-01 -1.18239868e+00 1.58664930e+00
-2.28939205e-01 4.69920665e-01 -4.99841422e-02 -9.14329052e-01
1.14776802e+00 -1.70852050e-01 2.10992739e-01 -8.85936856e-01
3.78906548e-01 9.39252973e-02 1.16480462e-01 -1.42490268e-01
5.33222616e-01 2.27762252e-01 2.53157299e-02 -3.79323363e-02
7.36463219e-02 4.56113935e-01 9.75318532e-03 1.91434383e-01
7.81156540e-01 7.10220784e-02 1.84674591e-01 -2.90563881e-01
9.02755558e-01 -1.95186466e-01 1.06977880e+00 4.64996338e-01
-3.82152885e-01 9.28642452e-01 2.82156318e-01 -4.22871947e-01
-5.71772158e-01 -9.52014863e-01 -1.14349574e-01 8.47225547e-01
9.01535988e-01 -3.24059933e-01 -6.03812635e-01 -8.77861142e-01
-2.11803075e-02 3.81974041e-01 -6.84791028e-01 -5.68498313e-01
-6.03969634e-01 -6.20300233e-01 9.41718463e-03 6.20183170e-01
1.01145089e+00 -1.42024136e+00 -9.01573658e-01 5.29674172e-01
-3.19268286e-01 -6.58248425e-01 -6.32162094e-01 1.84370264e-01
-7.82597005e-01 -8.51068020e-01 -8.35973442e-01 -1.00110710e+00
8.53504062e-01 8.10021222e-01 8.00961435e-01 2.29399040e-01
-2.38679975e-01 -5.59158981e-01 -4.78526235e-01 -4.79354948e-01
4.07801755e-02 1.25432923e-01 -4.11040395e-01 2.12003678e-01
4.01770324e-01 -3.67687076e-01 -9.46526647e-01 3.68999392e-01
-9.73549664e-01 3.85000795e-01 1.05802429e+00 8.29990029e-01
6.06600344e-01 9.97204632e-02 6.20061755e-01 -3.70080352e-01
3.65089864e-01 -3.57221067e-01 -3.74134392e-01 1.65550977e-01
-2.65417427e-01 1.96817979e-01 8.15429688e-01 -1.95001751e-01
-1.16599226e+00 3.62580009e-02 3.82842459e-02 -3.76046449e-01
8.45479034e-03 1.83763862e-01 -4.61802185e-01 -1.38550550e-01
3.12509298e-01 3.61777455e-01 -1.26149461e-01 -4.44837302e-01
1.39343500e-01 4.71472383e-01 5.05105078e-01 3.10822390e-04
6.37418687e-01 3.77752841e-01 -2.28043720e-01 -2.25741997e-01
-9.00437951e-01 -2.09186971e-01 -4.88681555e-01 -1.59845814e-01
6.62063003e-01 -1.00313842e+00 -3.95241112e-01 5.76679170e-01
-1.06808782e+00 1.14589497e-01 -3.04050952e-01 3.90329778e-01
-1.38784721e-01 2.90622383e-01 -5.41655600e-01 -4.33639228e-01
-4.51669812e-01 -1.34343612e+00 1.01950622e+00 9.42417383e-01
1.67759284e-01 -4.56178397e-01 -5.23518145e-01 -8.25821683e-02
5.43458402e-01 7.97970593e-02 6.96618795e-01 -2.12877586e-01
-7.87741542e-01 1.27172947e-01 -6.50116265e-01 4.50817555e-01
4.33469325e-01 1.19530231e-01 -9.39283133e-01 -2.23628923e-01
-2.57109175e-04 -2.90872958e-02 1.53788698e+00 5.37685931e-01
1.20252085e+00 -3.15305233e-01 -4.86199051e-01 5.80579400e-01
1.37019217e+00 8.67975205e-02 6.27612948e-01 5.58264554e-01
7.09653735e-01 4.75574315e-01 8.48497808e-01 3.36894602e-01
3.55903953e-01 4.05883551e-01 7.52065659e-01 -5.12848496e-01
-3.00663710e-01 -2.96478808e-01 4.73736733e-01 3.83741736e-01
2.15210736e-01 9.77311134e-02 -3.49022299e-01 6.70876086e-01
-1.78148830e+00 -9.05569613e-01 -5.16314432e-02 2.09195852e+00
8.74203742e-01 4.76866126e-01 5.60779460e-02 1.72555387e-01
1.01977038e+00 4.79112834e-01 -6.47750556e-01 -1.92308143e-01
-5.25694728e-01 -5.56192081e-03 4.59216535e-01 2.36731246e-01
-1.32857585e+00 8.42915118e-01 4.93753481e+00 9.33122456e-01
-1.34562218e+00 -7.06258342e-02 8.15521896e-01 -6.77635372e-02
-4.65512723e-01 1.40891209e-01 -8.75051200e-01 8.11203122e-01
1.11935697e-01 1.10903699e-02 1.88252460e-02 7.04999566e-01
8.37898329e-02 -3.63443702e-01 -5.78960955e-01 7.54886270e-01
-1.15748830e-01 -1.24063265e+00 2.45152146e-01 -2.11540833e-01
8.38562787e-01 1.95840880e-01 1.73983753e-01 2.73801476e-01
9.95709151e-02 -7.07900405e-01 9.12563682e-01 5.38255334e-01
3.30552012e-01 -9.26553488e-01 8.41018200e-01 3.03816795e-01
-1.38012207e+00 -3.09037268e-01 -7.83858657e-01 -1.00183748e-02
1.48886353e-01 9.41811681e-01 -3.69486421e-01 6.77664459e-01
9.30283725e-01 9.03100908e-01 -7.70421684e-01 1.27598882e+00
-2.89392650e-01 1.77842602e-01 -3.72438014e-01 -2.34316327e-02
5.17279506e-01 1.55504923e-02 4.94195193e-01 1.15764236e+00
2.46996596e-01 -2.42034178e-02 7.80428275e-02 1.12881482e+00
1.23337647e-02 1.02757901e-01 -1.56898692e-01 5.78858316e-01
3.34438890e-01 1.51828945e+00 -7.51021087e-01 -4.69478905e-01
-6.28588438e-01 9.66199458e-01 2.94910729e-01 1.96436539e-01
-8.21683466e-01 -6.70599759e-01 6.24646366e-01 7.02061728e-02
6.35313332e-01 2.68648148e-01 -4.29518342e-01 -1.01050019e+00
2.85210401e-01 -4.21738833e-01 2.03166530e-01 -6.74221337e-01
-8.77183437e-01 6.44047737e-01 -2.71606356e-01 -1.59439659e+00
3.14620286e-01 -1.58672526e-01 -9.02291775e-01 1.08723295e+00
-2.04611015e+00 -7.88914800e-01 -4.55917209e-01 4.33509350e-01
6.53784037e-01 1.32356286e-01 2.14941025e-01 2.47390136e-01
-6.37342691e-01 5.95108926e-01 -1.54974312e-01 1.07618116e-01
5.58240712e-01 -1.08274782e+00 3.85668159e-01 1.03813708e+00
-2.05911636e-01 4.80356276e-01 3.64852816e-01 -6.22219741e-01
-6.90166116e-01 -1.38415146e+00 7.98982024e-01 1.47959098e-01
2.19694301e-01 -1.08411647e-01 -1.21193254e+00 2.19403878e-01
1.34576708e-01 2.44029179e-01 -7.40799727e-03 -4.62969631e-01
-1.36924371e-01 -2.88585961e-01 -1.03564012e+00 5.45915306e-01
9.94062603e-01 -7.51875490e-02 -6.90408707e-01 -1.57450706e-01
8.70712638e-01 -1.16344623e-01 -3.74652267e-01 8.18760335e-01
1.86699003e-01 -1.30938518e+00 8.13564837e-01 -6.90622777e-02
6.06459022e-01 -8.06266665e-01 2.22072244e-01 -1.18240702e+00
-6.96970999e-01 -3.41529578e-01 6.02956712e-02 1.44104004e+00
3.96660566e-01 -5.62356412e-01 5.11347890e-01 1.60060734e-01
-3.78550410e-01 -9.58442390e-01 -6.31803751e-01 -5.04727483e-01
-2.89538324e-01 1.22898199e-01 8.54732275e-01 6.52965903e-01
-1.27680331e-01 3.66264254e-01 -8.35903883e-02 1.41628072e-01
4.54073697e-01 4.59006339e-01 2.01883480e-01 -1.18748450e+00
8.25785771e-02 -8.55608582e-01 -5.26514053e-01 -1.26920986e+00
-2.16257140e-01 -7.84696043e-01 2.00013056e-01 -1.62203455e+00
4.53464657e-01 -5.61377525e-01 -7.59605765e-01 7.54186869e-01
-7.19457865e-01 2.57010728e-01 3.39730889e-01 1.61355123e-01
-6.01394713e-01 9.41754043e-01 1.73921132e+00 -3.00165206e-01
-2.84593821e-01 -1.50279179e-01 -1.09637964e+00 7.91148782e-01
8.85351658e-01 -2.80638784e-01 -3.48886162e-01 -4.48895335e-01
-2.25663871e-01 -3.46975446e-01 6.20945096e-01 -1.18728817e+00
2.35329106e-01 -4.04740870e-02 8.97979856e-01 -7.62340069e-01
1.94542054e-02 -6.79341197e-01 -2.71290898e-01 4.84949529e-01
-2.00992793e-01 -1.15447409e-01 1.82717413e-01 2.63360530e-01
-5.50824106e-01 -8.85242298e-02 8.40606987e-01 -4.02749442e-02
-1.08060539e+00 2.82311022e-01 -5.69069050e-02 -2.37892225e-01
1.02182472e+00 -4.46749628e-01 -3.89737010e-01 -8.36008117e-02
-4.18482095e-01 4.97617066e-01 4.65207994e-01 6.82803571e-01
8.08002591e-01 -1.34280026e+00 -7.14358509e-01 4.70722258e-01
7.50950398e-03 4.68336135e-01 4.91701335e-01 7.84140885e-01
-2.28261590e-01 2.96078771e-01 -4.99056846e-01 -5.25910318e-01
-9.31014538e-01 6.67701066e-01 3.03893238e-01 5.44642732e-02
-5.86239219e-01 9.95386899e-01 7.39922523e-01 -1.36497244e-02
5.99917285e-02 -5.93174517e-01 -4.43926990e-01 4.99522372e-04
7.01076090e-01 6.83707520e-02 3.40306200e-02 -8.16698313e-01
-4.71611321e-01 5.92728555e-01 -5.44362128e-01 3.02821934e-01
1.13215566e+00 -2.39555165e-01 2.30253488e-02 1.76148266e-02
1.29440844e+00 -2.25014284e-01 -1.82365644e+00 -5.29397607e-01
-3.58868182e-01 -4.49247479e-01 3.83861005e-01 -7.55384326e-01
-1.55335987e+00 9.52420771e-01 6.26120865e-01 -1.71602946e-02
1.51219046e+00 4.41958681e-02 9.87313986e-01 -9.36627835e-02
1.75690383e-01 -1.02689815e+00 -9.54461619e-02 3.49791318e-01
8.79948318e-01 -1.31170726e+00 -2.98828259e-02 -5.96415997e-01
-6.95030332e-01 8.52144480e-01 1.03541374e+00 -2.91122735e-01
4.63192225e-01 1.17374152e-01 1.07622007e-02 5.26453219e-02
-5.49078047e-01 -4.73199725e-01 3.42772335e-01 3.57818931e-01
1.42454386e-01 -1.18266262e-01 -5.15210807e-01 7.98095167e-01
-1.51099288e-03 -1.72068119e-01 2.95261651e-01 9.27159369e-01
-8.79771769e-01 -8.49279404e-01 -2.45267719e-01 6.22224748e-01
-3.24616611e-01 -2.32326925e-01 -2.28812441e-01 5.77844441e-01
4.83938724e-01 7.65363455e-01 1.18647464e-01 -4.98132259e-01
2.62036622e-01 -4.38242555e-01 7.12316632e-02 -3.85528922e-01
-6.42935932e-01 1.75190955e-01 -5.27615190e-01 -5.42972803e-01
-4.94111687e-01 -6.40085578e-01 -1.61738193e+00 3.77261452e-02
-5.95325053e-01 4.09380570e-02 2.81675577e-01 7.97009170e-01
4.72182989e-01 8.38858426e-01 9.17033315e-01 -9.62189853e-01
-3.05815071e-01 -9.46843803e-01 -4.04573351e-01 2.74982452e-01
5.77027023e-01 -8.42609048e-01 -2.82900512e-01 -1.78093001e-01] | [9.687987327575684, -0.46443894505500793] |
da54b901-10ca-4fca-8b44-4fac210cf39b | knowledge-aware-audio-grounded-generative | 2307.01764 | null | https://arxiv.org/abs/2307.01764v1 | https://arxiv.org/pdf/2307.01764v1.pdf | Knowledge-Aware Audio-Grounded Generative Slot Filling for Limited Annotated Data | Manually annotating fine-grained slot-value labels for task-oriented dialogue (ToD) systems is an expensive and time-consuming endeavour. This motivates research into slot-filling methods that operate with limited amounts of labelled data. Moreover, the majority of current work on ToD is based solely on text as the input modality, neglecting the additional challenges of imperfect automatic speech recognition (ASR) when working with spoken language. In this work, we propose a Knowledge-Aware Audio-Grounded generative slot-filling framework, termed KA2G, that focuses on few-shot and zero-shot slot filling for ToD with speech input. KA2G achieves robust and data-efficient slot filling for speech-based ToD by 1) framing it as a text generation task, 2) grounding text generation additionally in the audio modality, and 3) conditioning on available external knowledge (e.g. a predefined list of possible slot values). We show that combining both modalities within the KA2G framework improves the robustness against ASR errors. Further, the knowledge-aware slot-value generator in KA2G, implemented via a pointer generator mechanism, particularly benefits few-shot and zero-shot learning. Experiments, conducted on the standard speech-based single-turn SLURP dataset and a multi-turn dataset extracted from a commercial ToD system, display strong and consistent gains over prior work, especially in few-shot and zero-shot setups. | ['Philip C. Woodland', 'Paweł Budzianowski', 'Ivan Vulić', 'Chao Zhang', 'Guangzhi Sun'] | 2023-07-04 | null | null | null | null | ['zero-shot-learning', 'text-generation', 'zero-shot-slot-filling', 'slot-filling', 'speech-recognition', 'automatic-speech-recognition'] | ['methodology', 'natural-language-processing', 'natural-language-processing', 'natural-language-processing', 'speech', 'speech'] | [ 4.15320128e-01 5.82299113e-01 -1.21831350e-01 -2.57705957e-01
-1.48681068e+00 -4.07437146e-01 6.74404860e-01 -3.84837613e-02
-3.00110281e-01 9.02478814e-01 5.90742230e-01 -5.65316021e-01
4.79173847e-02 -6.23008847e-01 -3.26993972e-01 -4.85969305e-01
3.41374874e-01 1.08116651e+00 2.87227571e-01 -6.38893187e-01
-1.35643221e-03 -1.81063622e-01 -1.62332368e+00 3.29642147e-01
7.73592174e-01 7.88277090e-01 3.23837191e-01 9.53435481e-01
-6.49480402e-01 9.19669211e-01 -9.15206134e-01 -1.18513688e-01
5.19304872e-02 -7.74639845e-01 -1.37108588e+00 4.13373321e-01
-1.23270638e-01 -9.63913798e-02 -1.49324834e-01 4.81333077e-01
9.41864550e-01 5.19624233e-01 1.69129387e-01 -1.16320515e+00
-8.72980058e-02 1.10876119e+00 1.92428663e-01 1.03352375e-01
5.60608506e-01 1.75818995e-01 1.14343357e+00 -1.02563226e+00
6.74940348e-01 1.37259364e+00 6.55138850e-01 8.98391664e-01
-1.31802833e+00 -1.00243747e-01 -3.52869853e-02 8.06456506e-02
-1.27071321e+00 -1.02913153e+00 6.09076679e-01 -2.73034066e-01
1.45518041e+00 4.42587882e-01 5.53815961e-01 1.06065869e+00
-6.28197968e-01 1.08877397e+00 1.03950334e+00 -9.26678002e-01
5.62666774e-01 1.27374217e-01 2.58734189e-02 3.69749039e-01
-6.09727859e-01 -9.98676270e-02 -1.17691278e+00 -2.92732120e-01
4.04718518e-01 -7.21763194e-01 -1.02922134e-01 -9.60512459e-02
-1.20432603e+00 8.92699778e-01 -4.67268169e-01 1.46541238e-01
-3.82925034e-01 -2.52929032e-01 7.28409350e-01 4.66324627e-01
7.22404897e-01 4.98592764e-01 -6.45266593e-01 -9.91658390e-01
-1.27884197e+00 3.67619336e-01 1.09095490e+00 1.20445943e+00
6.96270645e-01 3.07843566e-01 -7.38611639e-01 1.24824190e+00
2.09125876e-01 3.66545051e-01 7.72321463e-01 -7.70870507e-01
6.39871418e-01 3.01833689e-01 1.39123872e-01 -7.19040334e-02
-3.45930874e-01 5.57943434e-02 -3.06741267e-01 -4.04594868e-01
4.73003864e-01 -2.59553879e-01 -1.14789104e+00 1.82301664e+00
6.54058456e-01 -2.24038772e-03 4.98299867e-01 8.20627391e-01
1.06437206e+00 6.96095645e-01 1.74001887e-01 -4.95685250e-01
1.61100543e+00 -1.06990290e+00 -1.13598347e+00 -3.65186214e-01
8.51001084e-01 -8.31313372e-01 1.44695926e+00 2.28581563e-01
-1.16527724e+00 -2.02915668e-01 -6.31909728e-01 -3.20952237e-01
-1.56795561e-01 3.38727683e-02 6.11138284e-01 9.37242091e-01
-1.15067470e+00 2.78104573e-01 -8.04391503e-01 -4.76326376e-01
5.87887503e-02 -2.33943984e-02 -3.29986289e-02 6.77181184e-02
-1.57022357e+00 7.19970405e-01 3.59978855e-01 -2.36001283e-01
-8.63127410e-01 -5.54578125e-01 -1.00490260e+00 -6.55587986e-02
9.71248806e-01 -4.15598005e-01 2.08196616e+00 -3.98842365e-01
-2.10720682e+00 7.22985685e-01 -4.72603232e-01 -7.65132427e-01
3.65837604e-01 -1.98594227e-01 -2.33261883e-01 9.97584164e-02
1.44346863e-01 6.72967851e-01 8.04092884e-01 -8.63205314e-01
-3.36705804e-01 -1.09526716e-01 -7.50559196e-02 6.55022085e-01
-2.42445357e-02 9.69239548e-02 -1.65168926e-01 -6.72857404e-01
2.20953092e-01 -8.09623301e-01 -9.34806317e-02 -4.67155069e-01
-6.90335095e-01 -5.19812286e-01 7.18428493e-01 -7.79251218e-01
1.48913634e+00 -1.92781889e+00 1.25837445e-01 -1.61511019e-01
-1.87738210e-01 4.36913073e-01 -2.28380989e-02 9.23319101e-01
2.65389144e-01 -1.03694960e-01 -2.07910955e-01 -7.00887263e-01
2.69317478e-01 5.06075203e-01 -6.85933053e-01 -1.80534080e-01
2.08692789e-01 9.83117819e-01 -9.26515043e-01 -5.51278174e-01
3.38421702e-01 3.17726105e-01 -4.89801139e-01 3.75169069e-01
-5.68205178e-01 3.14033687e-01 -5.55406511e-02 6.75143301e-01
8.20648000e-02 -2.19952129e-03 2.56132543e-01 1.38328031e-01
-1.05781041e-01 1.14502668e+00 -1.24313462e+00 1.98480618e+00
-5.81325412e-01 3.91632974e-01 -2.16240548e-02 -9.15360272e-01
9.96980846e-01 8.97742808e-01 2.07984433e-01 -6.43615842e-01
6.75488338e-02 3.13923120e-01 -2.78029263e-01 -2.42965281e-01
1.17603040e+00 -3.95426095e-01 -4.33720469e-01 7.53783882e-01
6.43417060e-01 -6.50456429e-01 2.79087424e-01 5.06003618e-01
1.09044290e+00 -7.14609493e-03 4.50969040e-01 9.79699269e-02
1.88078791e-01 2.11438328e-01 4.61712003e-01 9.30899262e-01
-2.38398716e-01 7.02901363e-01 4.12031561e-01 7.42279887e-02
-1.09086430e+00 -7.92112291e-01 -2.64146179e-02 1.52368057e+00
-1.63845792e-01 -7.83407390e-01 -1.03130770e+00 -3.96992236e-01
-3.52165312e-01 9.90284503e-01 -1.88904285e-01 -6.18299395e-02
-3.32361430e-01 -4.40230548e-01 9.02463198e-01 3.22887331e-01
2.88571566e-01 -1.27966976e+00 -4.88217711e-01 6.59819186e-01
-6.92102194e-01 -1.15796816e+00 -3.46736610e-01 6.13436043e-01
-5.84325492e-01 -7.50154614e-01 -6.63944304e-01 -5.04040360e-01
3.34625761e-03 2.27205977e-01 1.24405169e+00 -2.42706582e-01
-6.34791628e-02 5.66256285e-01 -7.89324403e-01 -2.84619689e-01
-7.46348917e-01 3.06949943e-01 4.26978432e-02 -2.63272971e-01
4.59198982e-01 -2.75007278e-01 5.63150421e-02 3.95730734e-01
-6.05053604e-01 3.28269362e-01 1.82599455e-01 1.26990163e+00
4.97575521e-01 -1.49675593e-01 8.54579806e-01 -8.78238082e-01
9.45173442e-01 -3.54187936e-01 -1.00963838e-01 1.95060521e-01
-5.04548371e-01 7.30287954e-02 2.82273948e-01 -3.69809628e-01
-1.29550016e+00 -4.99428734e-02 -4.57369596e-01 -1.68516532e-01
-1.92086712e-01 5.12204349e-01 -2.01600716e-01 5.01735330e-01
8.55574906e-01 5.24048567e-01 -2.57915189e-03 -6.02803707e-01
6.96932614e-01 1.09276628e+00 5.57989776e-01 -8.17622304e-01
4.49769408e-01 -9.92443860e-02 -5.79253793e-01 -1.22496581e+00
-8.98122549e-01 -7.49833763e-01 -3.41610014e-01 -1.49172813e-01
5.62443137e-01 -1.01371861e+00 -1.89065978e-01 6.09868169e-01
-9.80738699e-01 -8.80243421e-01 -8.61757636e-01 2.11015880e-01
-1.04901540e+00 3.04367423e-01 -6.47120953e-01 -1.33649409e+00
-5.46055496e-01 -9.94225800e-01 1.33554757e+00 -2.40657274e-02
-6.19297683e-01 -8.16303551e-01 6.03558421e-02 7.39565670e-01
5.15677214e-01 -5.24680078e-01 6.82666421e-01 -1.00862753e+00
-4.15409863e-01 2.41689429e-01 1.67918593e-01 1.12762630e-01
-5.81106618e-02 -5.02392352e-01 -1.29414880e+00 -8.23683739e-02
-2.36808017e-01 -8.59068155e-01 6.45414293e-01 1.92302167e-01
5.46507478e-01 -4.96504962e-01 1.13199905e-01 3.82243879e-02
6.02903843e-01 1.03490196e-01 5.45534611e-01 1.71077564e-01
3.96787137e-01 6.09802723e-01 1.00363445e+00 7.95079529e-01
6.51712060e-01 1.05489314e+00 -6.01092838e-02 1.16966262e-01
-3.83831918e-01 -5.13351977e-01 2.38867491e-01 9.98186886e-01
3.22121531e-01 -4.49165046e-01 -9.50404048e-01 9.30114150e-01
-1.99924839e+00 -9.74064827e-01 1.06308110e-01 2.07757783e+00
1.54056180e+00 -2.51042610e-03 4.10845399e-01 7.21527278e-01
6.97601855e-01 2.55144447e-01 -3.48907262e-01 -4.85352814e-01
-2.00105742e-01 4.56098020e-01 -6.73392788e-02 5.80720007e-01
-1.03360474e+00 1.44058454e+00 5.96192265e+00 1.19148290e+00
-8.24232042e-01 4.06797141e-01 3.36202770e-01 -8.42462480e-02
-4.42679554e-01 7.22823367e-02 -9.45256591e-01 3.24418634e-01
1.29878485e+00 -3.16727251e-01 6.45385444e-01 8.79978657e-01
1.45629272e-01 -2.96955138e-01 -7.44892538e-01 8.66467357e-01
-1.23206899e-01 -1.17932010e+00 -3.23764801e-01 -2.66390860e-01
3.42079252e-01 -1.36993453e-01 -1.50499895e-01 7.28218377e-01
5.19194782e-01 -8.21634829e-01 1.07105303e+00 9.54609662e-02
1.15654361e+00 -5.29080212e-01 5.02559185e-01 4.85624880e-01
-1.09135854e+00 1.42982915e-01 -6.26464635e-02 -1.78992867e-01
6.26652956e-01 5.79510570e-01 -1.59755003e+00 4.03719157e-01
3.88771594e-01 1.46906048e-01 4.44163047e-02 6.86049104e-01
-4.14586842e-01 1.14289892e+00 -4.49244618e-01 1.23810023e-01
2.36614525e-01 3.68363529e-01 6.81668639e-01 1.29570496e+00
2.10140184e-01 1.75788462e-01 3.75863701e-01 5.72963119e-01
1.28064781e-01 1.11627534e-01 -3.35819930e-01 -3.81934702e-01
1.04542387e+00 1.00585544e+00 -7.65153408e-01 -4.52099264e-01
1.11923432e-02 9.54406917e-01 1.11149162e-01 2.11211175e-01
-2.77754337e-01 -4.26490247e-01 6.64290726e-01 -8.98106843e-02
3.15211058e-01 -2.81060904e-01 -2.75717825e-01 -1.02182126e+00
-1.03687294e-01 -9.58510756e-01 3.47792685e-01 -7.25570738e-01
-9.59853053e-01 7.15583384e-01 6.38767928e-02 -1.04837477e+00
-1.09831250e+00 6.03383221e-03 -1.81414559e-01 1.08095825e+00
-1.43766630e+00 -9.65243518e-01 1.80101469e-02 5.63079953e-01
1.04694283e+00 -2.06342354e-01 1.24031568e+00 1.52705565e-01
-2.61094481e-01 6.49271250e-01 -3.70721549e-01 -3.58302593e-02
5.18688679e-01 -1.30324149e+00 8.40270817e-01 7.20763922e-01
3.71714950e-01 3.14041674e-01 8.55294764e-01 -6.21855259e-01
-1.19839883e+00 -9.44031298e-01 1.43393409e+00 -3.72312218e-01
5.52053511e-01 -6.73663437e-01 -1.05380237e+00 4.47250634e-01
-5.42770736e-02 -1.81187242e-01 7.48081088e-01 4.21583563e-01
-1.93236887e-01 5.04678905e-01 -1.11924958e+00 4.02307391e-01
1.12982488e+00 -1.09585023e+00 -1.01303589e+00 3.45182121e-01
1.15087163e+00 -9.74049449e-01 -6.04783654e-01 2.24582348e-02
1.25307366e-01 -6.85588419e-01 7.38547266e-01 -5.33895850e-01
-7.28406385e-02 -1.84490681e-01 -3.22757035e-01 -1.53505051e+00
2.29208201e-01 -1.16864049e+00 -3.10766280e-01 1.44226551e+00
3.13625962e-01 -2.82505274e-01 6.84500098e-01 5.42602241e-01
-6.32678270e-01 -5.15120864e-01 -1.40901136e+00 -6.96847856e-01
-4.06989723e-01 -8.67597640e-01 6.43874764e-01 7.91642547e-01
4.33234304e-01 6.50153518e-01 -6.85212314e-01 -2.71761417e-01
1.16403207e-01 -2.05720603e-01 8.42796445e-01 -1.04903626e+00
-3.91135335e-01 7.13885650e-02 -1.01846039e-01 -1.11058605e+00
5.48727550e-02 -7.35237896e-01 6.47862017e-01 -1.51162922e+00
-3.57662380e-01 -7.90485680e-01 1.76250026e-01 1.01964295e+00
-1.25486270e-01 9.00417417e-02 2.69972920e-01 1.38297021e-01
-7.73703575e-01 8.78769159e-01 8.02709103e-01 1.09119006e-01
-6.32749617e-01 -8.17061216e-03 -4.76856649e-01 2.84182250e-01
6.44354522e-01 -4.06550020e-01 -6.60409272e-01 -2.28558946e-02
9.89583787e-03 6.08798623e-01 1.92451119e-01 -8.28490257e-01
3.12620193e-01 -1.40306696e-01 -5.72569013e-01 -5.66461444e-01
7.21086681e-01 -1.95719421e-01 -3.75647396e-02 -2.11805508e-01
-5.01538813e-01 -5.21048605e-01 3.24310809e-01 3.54187340e-01
-2.72584051e-01 -2.95618981e-01 4.44972277e-01 -1.75856322e-01
-8.55369329e-01 -9.95977893e-02 -7.29372263e-01 6.76784337e-01
4.71095681e-01 -3.26230943e-01 -2.68282592e-01 -5.58013618e-01
-9.22366083e-01 1.11900814e-01 9.60869417e-02 5.19194961e-01
3.80792052e-01 -1.18073952e+00 -4.68871236e-01 4.52340722e-01
3.79521847e-01 9.34265330e-02 3.80852968e-01 6.20322287e-01
2.65286505e-01 6.34192288e-01 2.25506619e-01 -6.56744123e-01
-1.18741715e+00 -4.60624546e-02 1.23684257e-01 -3.35179389e-01
-5.88909030e-01 1.06114948e+00 -4.69006628e-01 -8.42049897e-01
4.18131858e-01 -1.36627421e-01 -1.22522041e-02 4.10547048e-01
5.28750777e-01 2.76540190e-01 5.11271596e-01 -7.24968135e-01
-1.87791795e-01 -1.77677527e-01 6.21133633e-02 -8.07105780e-01
1.11485827e+00 -4.49112833e-01 3.94916564e-01 7.68694222e-01
3.95521194e-01 -2.38053620e-01 -1.10190833e+00 -7.48170376e-01
2.15909496e-01 -3.73556763e-01 1.68600604e-01 -1.10491073e+00
-3.31517130e-01 8.51109385e-01 8.47825408e-02 4.04774010e-01
7.98486173e-01 1.32842138e-01 9.88742054e-01 5.64708054e-01
6.28685832e-01 -1.47445941e+00 1.38234735e-01 1.04509902e+00
6.31721795e-01 -1.00183535e+00 -4.87562954e-01 -3.03441852e-01
-1.09572291e+00 7.70392716e-01 4.26484287e-01 7.52311230e-01
1.59357339e-01 2.49500468e-01 3.81965458e-01 -8.45224634e-02
-1.08666682e+00 -6.05881095e-01 2.03310810e-02 7.96607614e-01
4.63516086e-01 2.01445073e-01 -2.56722093e-01 7.63161182e-01
-6.30916953e-01 1.50240868e-01 5.60828626e-01 1.16756201e+00
-7.26838529e-01 -1.31476521e+00 -3.79140079e-01 3.05604756e-01
-1.95692241e-01 -4.80964005e-01 -3.00400704e-01 4.12074417e-01
-2.74443865e-01 1.34955537e+00 4.13789824e-02 -2.18322933e-01
3.14476907e-01 8.12765300e-01 1.29658371e-01 -1.18308067e+00
-5.54524302e-01 1.79873288e-01 8.22371244e-01 -5.46103835e-01
-3.19895446e-01 -7.06733823e-01 -1.38803613e+00 -1.89411119e-02
-5.54074883e-01 5.51987469e-01 6.44820571e-01 1.02875340e+00
5.04595041e-01 5.13296306e-01 3.71146291e-01 -8.51761758e-01
-8.30046654e-01 -1.21250248e+00 -6.67956471e-01 1.08686738e-01
2.04030067e-01 -8.02843213e-01 -3.03377122e-01 -4.92379367e-02] | [13.095648765563965, 7.5842204093933105] |
6b0780f0-3046-464c-85d7-65a63a451a9c | posediffusion-solving-pose-estimation-via | 2306.15667 | null | https://arxiv.org/abs/2306.15667v2 | https://arxiv.org/pdf/2306.15667v2.pdf | PoseDiffusion: Solving Pose Estimation via Diffusion-aided Bundle Adjustment | Camera pose estimation is a long-standing computer vision problem that to date often relies on classical methods, such as handcrafted keypoint matching, RANSAC and bundle adjustment. In this paper, we propose to formulate the Structure from Motion (SfM) problem inside a probabilistic diffusion framework, modelling the conditional distribution of camera poses given input images. This novel view of an old problem has several advantages. (i) The nature of the diffusion framework mirrors the iterative procedure of bundle adjustment. (ii) The formulation allows a seamless integration of geometric constraints from epipolar geometry. (iii) It excels in typically difficult scenarios such as sparse views with wide baselines. (iv) The method can predict intrinsics and extrinsics for an arbitrary amount of images. We demonstrate that our method PoseDiffusion significantly improves over the classic SfM pipelines and the learned approaches on two real-world datasets. Finally, it is observed that our method can generalize across datasets without further training. Project page: https://posediffusion.github.io/ | ['David Novotny', 'Christian Rupprecht', 'Jianyuan Wang'] | 2023-06-27 | null | null | null | null | ['pose-estimation'] | ['computer-vision'] | [-5.17994873e-02 -2.14710116e-01 3.72573249e-02 -3.58096868e-01
-8.42568040e-01 -1.00858819e+00 9.11737084e-01 -3.74227166e-01
-4.82657939e-01 2.39562064e-01 2.46153578e-01 -1.68172881e-01
-1.55735105e-01 -3.15739930e-01 -9.44805622e-01 -5.35692692e-01
3.10329676e-01 7.07595885e-01 2.26964846e-01 -8.46080855e-02
5.05211651e-01 7.88147449e-01 -1.25047016e+00 -3.25693518e-01
3.68291408e-01 7.44503558e-01 4.00472492e-01 8.09467852e-01
3.38760585e-01 5.14389277e-01 8.69573653e-02 -6.92495108e-01
7.13782430e-01 2.18732096e-02 -7.96338558e-01 4.88655478e-01
1.02922285e+00 -5.11713624e-01 -5.86941719e-01 9.00938094e-01
2.45386511e-01 7.70140216e-02 5.69253445e-01 -1.31885278e+00
-2.13247061e-01 -1.16564766e-01 -8.09094727e-01 -4.72807772e-02
6.62202060e-01 3.26016426e-01 1.25146997e+00 -1.17378294e+00
1.08880568e+00 1.04151046e+00 8.98349524e-01 1.25443041e-01
-1.41508651e+00 -2.39380613e-01 1.44533455e-01 4.06625643e-02
-1.45799506e+00 -5.36538541e-01 7.01760888e-01 -9.09511566e-01
6.73977077e-01 1.48429558e-01 6.60447419e-01 1.15034437e+00
5.03750332e-02 7.92133391e-01 9.30146396e-01 -3.64303768e-01
-4.29887231e-03 3.36899376e-03 -1.45944327e-01 6.29021287e-01
4.10380960e-02 1.81297049e-01 -6.64426863e-01 -1.95757955e-01
1.05887556e+00 2.12817430e-01 -4.46771652e-01 -1.24201775e+00
-1.69306922e+00 8.54529917e-01 2.14565828e-01 -1.27120152e-01
-1.47658199e-01 3.19324791e-01 -2.06646413e-01 1.24456272e-01
2.09259629e-01 3.85624588e-01 -5.03329515e-01 -1.88578561e-01
-9.90530968e-01 3.19535762e-01 9.50653076e-01 1.38529563e+00
1.02343488e+00 -3.69211555e-01 5.35813570e-01 5.48916996e-01
7.03024924e-01 7.24292278e-01 1.44076198e-01 -1.49318445e+00
4.75663930e-01 2.34636635e-01 3.06048810e-01 -1.14812136e+00
-3.44029248e-01 -1.47888437e-01 -6.06198609e-01 2.92107344e-01
5.26809216e-01 -1.46198003e-02 -5.29805243e-01 1.47713137e+00
4.15638953e-01 2.54613519e-01 -2.19659969e-01 8.51345301e-01
4.12948936e-01 3.01701427e-01 -6.74018264e-01 5.85958362e-02
1.05489945e+00 -1.01295674e+00 -4.57680881e-01 -5.32037079e-01
2.13900611e-01 -1.21496499e+00 5.91996670e-01 4.79266793e-01
-9.99932170e-01 -3.06222528e-01 -1.02385128e+00 -3.25571895e-01
-8.28027874e-02 -1.88190062e-02 6.04617357e-01 4.82123435e-01
-1.39386415e+00 7.47398198e-01 -1.02048302e+00 -7.21362948e-01
7.64135867e-02 4.13726032e-01 -7.33758688e-01 -1.96304172e-01
-3.47192913e-01 9.05963421e-01 -6.22113161e-02 -1.40780918e-02
-7.52348840e-01 -5.95222771e-01 -1.02257335e+00 -3.65105063e-01
5.56166768e-01 -1.21773767e+00 1.37499702e+00 -6.76175773e-01
-1.51331675e+00 1.01008785e+00 -3.05157363e-01 -1.40676409e-01
9.22385573e-01 -7.06092000e-01 3.44081223e-01 1.71513617e-01
4.87955064e-02 7.48818874e-01 9.97504354e-01 -1.40286922e+00
-4.73032832e-01 -4.68638182e-01 3.35744256e-03 4.68467712e-01
1.45839065e-01 -1.89874694e-01 -9.54410732e-01 -7.02395439e-01
6.25356019e-01 -1.44300056e+00 -3.43484581e-01 4.08201635e-01
-3.41917396e-01 4.22659665e-01 7.33411193e-01 -5.21526992e-01
6.52039289e-01 -2.00314260e+00 7.05624521e-01 1.71202272e-01
2.25566953e-01 -2.42000684e-01 8.92064273e-02 7.16429949e-01
4.56221364e-02 -2.85800278e-01 -2.56168872e-01 -7.72362292e-01
6.80148751e-02 3.27590317e-01 -3.37229162e-01 9.27213609e-01
-2.24973693e-01 7.46383250e-01 -7.71667838e-01 -3.23232681e-01
5.98522365e-01 5.66966653e-01 -6.73388302e-01 2.97125995e-01
1.06149115e-01 7.66260922e-01 3.43062058e-02 4.49767441e-01
8.11834455e-01 -2.90274858e-01 -5.36933094e-02 -5.24697065e-01
-3.42712611e-01 -4.16049249e-02 -1.69111085e+00 2.33190656e+00
-1.96900263e-01 7.36828208e-01 3.39617282e-01 -4.22004282e-01
6.37417197e-01 1.35190457e-01 6.92743480e-01 6.15943298e-02
-7.75993103e-03 9.22976658e-02 -5.11522472e-01 -2.90195853e-01
6.32188797e-01 9.29937288e-02 2.18900919e-01 3.99087936e-01
4.28345919e-01 -6.70112371e-01 -1.48803949e-01 4.51183051e-01
9.06694591e-01 7.82224178e-01 5.03366888e-01 -1.87687263e-01
2.59845704e-01 -6.55887276e-02 5.94237864e-01 5.67532539e-01
-3.51106897e-02 1.22742987e+00 1.22379102e-01 -3.49913985e-01
-1.17974532e+00 -1.18491459e+00 -1.61762968e-01 4.72058356e-01
3.30894679e-01 -5.38767457e-01 -5.83405137e-01 -3.75257194e-01
8.62515271e-02 2.66090840e-01 -3.60804677e-01 3.88950408e-01
-3.57957989e-01 -4.26347733e-01 -3.58682685e-03 4.19672906e-01
3.08557868e-01 -4.44124728e-01 -4.68618572e-01 -1.30599001e-02
-2.22399235e-01 -1.42351913e+00 -7.63743699e-01 -2.34727129e-01
-9.80607331e-01 -1.21858680e+00 -7.42936730e-01 -5.16572058e-01
5.89412928e-01 8.06689978e-01 1.17713130e+00 -2.04168156e-01
-2.07715735e-01 1.16055679e+00 -1.45281836e-01 -6.21534046e-03
-9.57052633e-02 -8.22012275e-02 2.40504757e-01 2.45635137e-01
2.31606454e-01 -8.53344917e-01 -7.95164287e-01 6.34230018e-01
-6.55443192e-01 1.26364574e-01 6.03739440e-01 7.25699902e-01
8.19366455e-01 -5.05599439e-01 -4.61313337e-01 -7.09131300e-01
-2.27683447e-02 -2.05225557e-01 -9.61838961e-01 1.85514595e-02
-5.96172094e-01 -6.80406094e-02 -7.32816197e-03 -1.36181325e-01
-9.20070469e-01 7.11432755e-01 4.37958017e-02 -6.33163393e-01
-2.31843278e-01 1.63159326e-01 -2.02941298e-01 -3.62298310e-01
4.30071652e-01 1.66134223e-01 2.48687074e-01 -7.01855183e-01
6.71934903e-01 3.04687202e-01 8.26415658e-01 -4.81127471e-01
1.16081583e+00 1.13952410e+00 1.33070067e-01 -8.11686397e-01
-6.73353851e-01 -9.59192753e-01 -1.43428099e+00 -2.24192142e-01
9.10555363e-01 -1.17453146e+00 -6.57862246e-01 5.19384623e-01
-1.27450740e+00 -2.20356852e-01 -6.08920725e-03 7.72396803e-01
-1.06879127e+00 8.72883081e-01 -4.33516592e-01 -4.84824449e-01
7.94324726e-02 -1.41734934e+00 1.31415570e+00 7.80808106e-02
-2.86601782e-01 -1.16241431e+00 2.44233459e-01 4.48932558e-01
-7.62068946e-03 3.17290962e-01 -4.11682110e-03 -1.64705608e-02
-1.17533565e+00 -1.63727626e-01 6.43583983e-02 2.20037058e-01
-2.49577276e-02 2.47251749e-01 -1.05546856e+00 -5.01649380e-01
2.86361337e-01 2.39990521e-02 5.99696100e-01 4.06932890e-01
5.50420880e-01 -8.88665691e-02 -3.16073120e-01 1.26821852e+00
1.52360415e+00 -3.76717985e-01 5.96984148e-01 6.96772933e-01
1.02595174e+00 6.61118984e-01 6.61864042e-01 3.69767129e-01
6.64092541e-01 9.24963534e-01 7.13625193e-01 1.43883331e-02
8.22454169e-02 -3.42132062e-01 3.13003212e-01 7.99168944e-01
-1.17026538e-01 1.79916561e-01 -1.00464642e+00 4.55293536e-01
-2.04513955e+00 -9.05273378e-01 -3.22389036e-01 2.34647155e+00
3.11170280e-01 -2.00838104e-01 -2.62489058e-02 -2.60677665e-01
4.31431621e-01 1.85573801e-01 -5.60992241e-01 1.85973302e-01
-2.83614725e-01 -3.20947647e-01 8.66656065e-01 7.91383743e-01
-1.16412973e+00 7.72566974e-01 6.01682472e+00 2.18587011e-01
-7.30467916e-01 1.51651174e-01 1.43009827e-01 -8.36993475e-03
-1.42297104e-01 4.33548868e-01 -8.21275413e-01 4.90070656e-02
2.53706098e-01 -2.66711172e-02 6.09985352e-01 7.57118046e-01
-1.27208605e-01 -1.63791671e-01 -1.39953244e+00 1.33292210e+00
3.89469564e-01 -1.20534921e+00 -2.40755886e-01 3.79764140e-01
7.60619104e-01 5.93200684e-01 1.98167771e-01 -3.69347721e-01
5.18467188e-01 -6.91934407e-01 8.60143840e-01 6.05003953e-01
4.98372763e-01 -3.65391225e-01 4.18853700e-01 4.81268167e-01
-9.07021761e-01 1.14833117e-01 -3.46474141e-01 1.12909831e-01
4.61792916e-01 4.86397833e-01 -5.90321541e-01 7.68520594e-01
8.00049245e-01 1.12025046e+00 -6.43596590e-01 1.25903499e+00
-3.20462823e-01 1.63056612e-01 -7.03154504e-01 7.08182454e-01
1.33433461e-01 -8.16743195e-01 8.27102721e-01 1.03987396e+00
4.74619091e-01 -1.40882298e-01 -3.15235853e-02 5.26187122e-01
1.43223032e-01 -1.30514920e-01 -9.70836759e-01 5.62831521e-01
2.59101897e-01 1.47582757e+00 -5.80347836e-01 9.03032199e-02
-6.77621841e-01 1.25406384e+00 2.09193364e-01 3.70445132e-01
-5.72105706e-01 3.28368783e-01 7.04215229e-01 1.73323274e-01
6.30481780e-01 -7.09264159e-01 -1.83070123e-01 -1.59203565e+00
2.10259899e-01 -8.78107846e-01 3.28020185e-01 -9.93856907e-01
-1.22673762e+00 2.74399728e-01 3.96330506e-02 -1.65217376e+00
-2.38529116e-01 -7.21051991e-01 -4.70691472e-01 7.29074299e-01
-1.36577392e+00 -1.32939875e+00 -4.29508001e-01 6.60792708e-01
5.38791835e-01 3.12743708e-02 6.01186872e-01 1.22268587e-01
-3.13943148e-01 1.24722578e-01 3.83713216e-01 5.21310195e-02
1.03241086e+00 -1.49822283e+00 5.88789344e-01 1.06702828e+00
5.91210842e-01 8.41094673e-01 8.00691962e-01 -3.24304730e-01
-1.87710321e+00 -6.65284693e-01 5.45303762e-01 -1.09810174e+00
8.54166031e-01 -4.10413593e-01 -5.07560730e-01 1.27520621e+00
3.17461133e-01 3.54398116e-02 5.57485163e-01 9.17078480e-02
-5.63551486e-01 3.47058102e-02 -8.02871287e-01 5.61911941e-01
1.07684278e+00 -5.23441255e-01 -4.35582489e-01 5.08975267e-01
3.72320026e-01 -7.73254514e-01 -9.30520296e-01 5.09077087e-02
6.45712197e-01 -1.25576878e+00 1.31031382e+00 -1.57545403e-01
1.99998528e-01 -5.26733398e-01 -5.08526087e-01 -1.22920763e+00
-3.43520254e-01 -1.18601358e+00 -1.89027712e-01 1.24556398e+00
2.10320339e-01 -5.70645452e-01 6.58486426e-01 8.06400061e-01
1.39603987e-01 -4.85218197e-01 -9.20786142e-01 -6.75357997e-01
-1.20984972e-01 -4.67162311e-01 2.93165296e-01 1.01131654e+00
-4.08199996e-01 3.79891127e-01 -5.97389400e-01 4.88477349e-01
9.83855844e-01 9.45293754e-02 1.38495231e+00 -1.26781261e+00
-7.70836651e-01 -2.64337450e-01 -6.50644839e-01 -1.55734229e+00
-1.09237440e-01 -6.23699725e-01 -1.34901881e-01 -1.35207653e+00
1.93007529e-01 -2.32005477e-01 2.89462715e-01 4.71009798e-02
-1.69566348e-02 1.36180207e-01 3.73497427e-01 6.53180718e-01
-5.25566220e-01 3.72000962e-01 9.71139610e-01 1.44094899e-01
7.99525604e-02 3.52103487e-02 -4.70583647e-01 1.13571334e+00
4.74190891e-01 -4.33982044e-01 -2.83403963e-01 -8.21965635e-01
4.47729141e-01 1.48434833e-01 6.21930718e-01 -7.91274011e-01
5.31204343e-01 -1.88503146e-01 2.09879905e-01 -7.03509688e-01
6.78108811e-01 -1.06700766e+00 5.30408502e-01 1.01771258e-01
2.73503363e-01 5.04940987e-01 -7.39856958e-02 8.25198054e-01
-1.18851669e-01 -2.24435225e-01 6.92208350e-01 -1.86190888e-01
-8.38550150e-01 4.65191841e-01 3.79892997e-02 -6.84435433e-03
1.02327776e+00 -4.17681038e-01 -3.54607985e-03 -6.69872999e-01
-6.65415943e-01 2.22253561e-01 1.22954190e+00 4.17658150e-01
4.85921592e-01 -1.23871815e+00 -5.51241577e-01 1.02320991e-01
1.78508669e-01 3.95920306e-01 9.46640223e-02 1.14092028e+00
-7.92227507e-01 2.37307642e-02 8.07372406e-02 -1.23242688e+00
-1.32781816e+00 5.94338238e-01 2.88660198e-01 9.84763503e-02
-8.71416450e-01 8.33393753e-01 3.57005954e-01 -6.64765298e-01
7.45698437e-02 4.85706814e-02 2.02835292e-01 -3.06110252e-02
3.18983704e-01 3.60281914e-01 -9.88123715e-02 -1.23956716e+00
-3.38526398e-01 1.20241916e+00 -5.43910898e-02 -5.10571003e-01
1.58251238e+00 -5.47779739e-01 -1.62092328e-01 3.71465355e-01
1.33459330e+00 2.46342242e-01 -1.85486388e+00 -3.04496169e-01
2.78604012e-02 -8.21983814e-01 3.81435044e-02 -2.96383411e-01
-9.67832744e-01 8.45653772e-01 3.54463756e-01 -2.34776661e-01
7.64836311e-01 7.78038502e-02 4.69121307e-01 4.17928308e-01
5.80472469e-01 -9.24095452e-01 -1.63680822e-01 4.46413159e-01
9.94190037e-01 -1.35357296e+00 3.88358325e-01 -5.01996577e-01
-6.77213609e-01 1.12573469e+00 2.44708434e-01 -1.58695623e-01
7.29465485e-01 2.67103970e-01 3.42254490e-01 -2.96083719e-01
-4.13614988e-01 -1.67048573e-02 2.63086706e-01 6.93950653e-01
1.24705665e-01 -3.48777235e-01 1.99424773e-01 -1.46256968e-01
-3.12430084e-01 -2.59573847e-01 7.24925458e-01 1.01283062e+00
-6.32869899e-02 -1.19054258e+00 -6.17966771e-01 -3.84598561e-02
-1.73927516e-01 5.40342294e-02 -5.09181798e-01 8.41046989e-01
-2.40516394e-01 6.99005842e-01 -1.05331652e-01 -2.38022804e-01
3.57597768e-01 -9.49839726e-02 7.71705151e-01 -5.05834520e-01
-1.84519202e-01 3.10053259e-01 -2.68680125e-01 -9.51202333e-01
-7.62295008e-01 -1.20722699e+00 -6.98070288e-01 -2.35839814e-01
-3.36390793e-01 -2.47395366e-01 8.66543233e-01 7.02007115e-01
4.32823896e-01 -2.06926987e-01 5.38524270e-01 -1.33851266e+00
-6.66519821e-01 -6.97123528e-01 -5.18836915e-01 7.01558948e-01
5.70825219e-01 -7.94757724e-01 -5.50279975e-01 4.00425613e-01] | [8.118128776550293, -2.402538776397705] |
65e6fd8d-5014-4a77-b659-8963c0de8b0b | gpinn-physics-informed-neural-network-with | 2306.09792 | null | https://arxiv.org/abs/2306.09792v1 | https://arxiv.org/pdf/2306.09792v1.pdf | GPINN: Physics-informed Neural Network with Graph Embedding | This work proposes a Physics-informed Neural Network framework with Graph Embedding (GPINN) to perform PINN in graph, i.e. topological space instead of traditional Euclidean space, for improved problem-solving efficiency. The method integrates topological data into the neural network's computations, which significantly boosts the performance of the Physics-Informed Neural Network (PINN). The graph embedding technique infuses extra dimensions into the input space to encapsulate the spatial characteristics of a graph while preserving the properties of the original space. The selection of these extra dimensions is guided by the Fiedler vector, offering an optimised pathologic notation of the graph. Two case studies are conducted, which demonstrate significant improvement in the performance of GPINN in comparison to traditional PINN, particularly in its superior ability to capture physical features of the solution. | ['Haolin Li', 'Yuyang Miao'] | 2023-06-16 | null | null | null | null | ['graph-embedding'] | ['graphs'] | [ 4.46442355e-05 2.43525848e-01 1.30235747e-01 -1.84457172e-02
4.98393178e-01 -2.94328511e-01 4.79889423e-01 -6.27664328e-02
-2.58289218e-01 5.35795867e-01 6.22121170e-02 -3.97456914e-01
-8.22419584e-01 -1.17465937e+00 -6.52413487e-01 -6.84716225e-01
-5.78217208e-01 1.86249197e-01 -1.44887641e-01 -3.79347801e-01
4.42064017e-01 9.43076134e-01 -1.13817739e+00 -2.77404636e-01
6.63049817e-01 9.53300774e-01 1.19411767e-01 6.42768979e-01
-1.77535594e-01 5.14454603e-01 -4.04248357e-01 4.62840684e-03
3.19731265e-01 3.18104513e-02 -6.50808334e-01 -4.99807030e-01
4.79464680e-01 2.44004447e-02 -1.08094764e+00 1.01001966e+00
3.24212909e-01 5.60731173e-01 5.70364058e-01 -1.26290131e+00
-1.06544948e+00 2.86804676e-01 -1.88608676e-01 1.73668370e-01
-3.59003805e-02 6.16914034e-02 1.15196919e+00 -6.02317572e-01
6.73634350e-01 1.35751283e+00 1.02436399e+00 2.36071825e-01
-1.51862180e+00 -4.19643879e-01 4.47267555e-02 1.24330483e-01
-1.42626607e+00 2.03489706e-01 1.46760321e+00 -3.42238784e-01
1.02943659e+00 2.95455247e-01 9.43507195e-01 7.40125597e-01
7.96682775e-01 1.68850854e-01 6.79503500e-01 -5.17434120e-01
4.77408588e-01 -1.05950639e-01 2.95030028e-01 8.40343058e-01
3.49614441e-01 5.06559491e-01 -6.67917192e-01 2.42850743e-04
1.25453413e+00 -1.21230118e-01 -2.89028138e-01 -9.97062385e-01
-9.69279408e-01 1.01383781e+00 1.16240692e+00 3.01740021e-01
-3.37255657e-01 5.11226773e-01 3.06069046e-01 7.10565001e-02
3.72667700e-01 9.28432643e-01 -2.78983235e-01 1.43567741e-01
-6.49007082e-01 3.32005590e-01 6.52472913e-01 6.73511744e-01
7.34606147e-01 4.17277545e-01 7.39776567e-02 4.57034260e-01
3.38496864e-01 2.50268131e-01 2.59983987e-01 -1.01083338e+00
3.25576276e-01 1.06580198e+00 -4.39770490e-01 -1.85972476e+00
-6.94763124e-01 -5.96012294e-01 -1.11233020e+00 4.57377911e-01
8.55079219e-02 1.08108304e-01 -9.39590573e-01 1.79128778e+00
7.95139521e-02 8.93196464e-02 5.44342399e-02 8.36224198e-01
7.50391841e-01 7.88346231e-01 -6.12330511e-02 6.36877120e-01
1.00322700e+00 -8.13980401e-01 -3.77070159e-01 -1.93370849e-01
7.26399064e-01 7.55955428e-02 9.58196580e-01 1.05951428e-02
-8.48552227e-01 -5.68572462e-01 -1.37356734e+00 -7.77251571e-02
-9.34848905e-01 -1.24089621e-01 9.77262855e-01 4.71459121e-01
-1.41597235e+00 1.12306941e+00 -6.59677565e-01 -1.58885986e-01
1.76841468e-01 7.56463647e-01 -5.66508770e-01 2.89741874e-01
-1.33936536e+00 7.85646915e-01 9.02454674e-01 2.88885832e-01
4.49963771e-02 -7.98877835e-01 -1.10007977e+00 3.01399678e-01
6.03617132e-02 -7.26684928e-01 5.02939165e-01 -5.47030330e-01
-1.36918736e+00 1.39714688e-01 4.56882358e-01 -4.19061720e-01
7.81937689e-02 2.17780605e-01 -2.39526585e-01 3.13928008e-01
-2.87439793e-01 8.21779728e-01 7.38107800e-01 -8.68212044e-01
1.16928555e-01 -1.13243386e-01 3.54865521e-01 2.20784992e-01
-4.15734708e-01 -7.16110528e-01 -3.95608306e-01 -8.56550276e-01
3.77261192e-01 -9.69269633e-01 -4.42757934e-01 3.10208946e-01
-3.89983594e-01 2.86629028e-03 1.09599948e+00 -5.98227859e-01
9.05526519e-01 -2.23006558e+00 3.58389586e-01 5.16373158e-01
8.95687878e-01 3.14049333e-01 -4.16439682e-01 6.49874687e-01
-4.83691007e-01 3.27765457e-02 -9.79974773e-03 8.23459923e-02
2.34763503e-01 4.71405566e-01 -8.44874978e-02 4.85862285e-01
3.30186635e-01 1.33188045e+00 -6.96118593e-01 -1.98995396e-01
4.63827252e-01 6.88342810e-01 -7.37493932e-01 -2.03166649e-01
-6.03407212e-02 2.38194674e-01 -5.63570380e-01 -9.22044292e-02
7.88087130e-01 -2.93529451e-01 1.01263903e-03 -4.73685801e-01
2.01326497e-02 1.35104805e-01 -1.04671037e+00 1.58330536e+00
-3.31433266e-01 8.35875392e-01 -6.23848848e-02 -1.10721242e+00
1.18871891e+00 -5.96616007e-02 5.12376368e-01 -1.06872559e+00
2.38906771e-01 -3.53315353e-01 9.32716802e-02 -1.13157719e-01
7.07101583e-01 1.34859547e-01 7.49715120e-02 1.76209956e-01
2.06841528e-02 1.82980046e-01 5.10630459e-02 2.59811521e-01
1.10773861e+00 1.49643704e-01 -5.76891638e-02 -6.42303288e-01
4.25416350e-01 -1.16235539e-01 2.70969927e-01 7.39438832e-01
8.83788150e-03 1.65766522e-01 6.51696861e-01 -7.39510357e-01
-1.20031905e+00 -1.11588144e+00 9.11404472e-03 5.05399287e-01
2.18553379e-01 -3.93337697e-01 -6.90011859e-01 -4.13763404e-01
2.13290185e-01 6.84080839e-01 -8.92026186e-01 -6.47314310e-01
-7.73114800e-01 -4.43714976e-01 4.92448598e-01 5.88118255e-01
4.72726971e-01 -1.04901850e+00 -6.23090029e-01 6.57879114e-02
3.76815468e-01 -9.28795516e-01 -5.01777530e-02 3.74535233e-01
-1.16985011e+00 -1.00405824e+00 -2.99362451e-01 -6.22215569e-01
6.77196026e-01 -1.04957260e-02 8.27027082e-01 1.28777355e-01
-3.28371435e-01 2.85391927e-01 -8.00823197e-02 7.84426033e-02
-1.45408601e-01 4.75440100e-02 1.37786016e-01 -4.02008981e-01
4.70558405e-02 -8.95689249e-01 -4.66870427e-01 -4.14682727e-04
-8.24992001e-01 3.20701778e-01 5.37730396e-01 8.23393285e-01
2.38282070e-01 4.92545754e-01 2.85828650e-01 -2.62125790e-01
7.57713258e-01 -1.61279052e-01 -6.73387110e-01 -1.12386234e-01
-7.21060097e-01 5.34168065e-01 9.34878945e-01 -4.91026044e-01
-3.91825110e-01 -3.27068985e-01 2.69876003e-01 -5.77554286e-01
2.69322962e-01 7.24263430e-01 2.94574294e-02 -9.77309883e-01
4.54329818e-01 -3.33504826e-02 2.68184006e-01 -5.51929772e-01
4.69530761e-01 -6.68887869e-02 5.25260091e-01 -4.87821430e-01
8.54469895e-01 -8.13867375e-02 6.60737216e-01 -8.31101835e-01
-2.07734108e-01 2.86482647e-02 -8.49430501e-01 -1.42367601e-01
5.48707128e-01 -2.48889998e-01 -9.30871487e-01 2.44729832e-01
-1.00789893e+00 -1.13658033e-01 -4.16623980e-01 5.37396550e-01
-5.45336068e-01 2.84476072e-01 -5.67948222e-01 -5.73967338e-01
-3.13825995e-01 -1.14150858e+00 6.26317382e-01 1.87060684e-01
-1.19110979e-01 -1.56284940e+00 2.46184040e-03 -1.90200984e-01
6.13379955e-01 5.54487646e-01 1.54303706e+00 -4.80353475e-01
-6.56271517e-01 -3.62042904e-01 -6.85149848e-01 1.43451899e-01
-1.18338056e-01 -4.39754315e-02 -8.04371178e-01 -2.80173302e-01
8.97337683e-03 3.30784589e-01 5.40456772e-01 5.74640334e-01
1.00818646e+00 -5.37371874e-01 -3.63159478e-01 8.06167722e-01
1.65020895e+00 2.24131763e-01 4.50795233e-01 4.82053429e-01
9.55350578e-01 4.44747508e-01 -1.63719416e-01 -2.79352199e-02
6.26448914e-02 5.97266793e-01 5.46015501e-01 -2.62042195e-01
-2.40190998e-01 -3.16845477e-01 -6.98403493e-02 1.04342175e+00
1.84117064e-01 -8.80288109e-02 -1.07906699e+00 2.10854188e-01
-1.75712645e+00 -3.73868555e-01 7.43266419e-02 1.86231053e+00
8.83227363e-02 2.64424205e-01 -3.44301969e-01 3.37651014e-01
7.02140927e-01 3.70946556e-01 -6.85677290e-01 -7.67129481e-01
7.70017365e-03 2.73864776e-01 6.73325241e-01 3.72485787e-01
-8.11387300e-01 6.65092647e-01 6.93225908e+00 6.04781210e-01
-1.40362191e+00 -3.04399490e-01 2.27575362e-01 1.12079367e-01
-7.22450465e-02 -9.12204310e-02 -1.77261576e-01 2.94453681e-01
1.08179367e+00 -1.66507557e-01 8.20990443e-01 7.98813403e-01
1.74213633e-01 2.21057922e-01 -1.13405609e+00 9.17509317e-01
-1.27304628e-01 -1.74814391e+00 2.98159301e-01 3.90874177e-01
2.52397686e-01 -5.92756085e-02 1.94680467e-01 1.63170174e-01
-7.75908679e-02 -1.19328570e+00 6.69312775e-01 8.06073010e-01
5.49407840e-01 -1.04454100e+00 6.59699976e-01 1.29542530e-01
-1.11435378e+00 -1.46935910e-01 -3.48782569e-01 -3.36682707e-01
-1.69026479e-01 4.29447055e-01 -6.82088792e-01 6.57929718e-01
5.71606576e-01 6.14670515e-01 -7.40609348e-01 8.84244919e-01
2.20106915e-01 2.23197639e-01 -3.40725064e-01 -2.70890713e-01
7.01756477e-01 -6.20718956e-01 8.41785908e-01 7.05207944e-01
-4.30279188e-02 -1.27278045e-01 3.41642722e-02 1.41180754e+00
-1.07016526e-02 -9.93312597e-02 -9.41764772e-01 -4.50238675e-01
2.25094691e-01 1.05569363e+00 -8.60910535e-01 1.11994989e-01
5.27068647e-03 4.62723911e-01 4.53883648e-01 6.10978186e-01
-6.95714474e-01 -8.92222166e-01 4.36292917e-01 -9.27684829e-02
3.41532022e-01 -6.96969867e-01 -4.20366108e-01 -5.63234925e-01
1.66470200e-01 -4.08131450e-01 -1.86626852e-01 -9.51888144e-01
-9.91763830e-01 6.66387200e-01 8.78370255e-02 -7.50421643e-01
8.00371096e-02 -1.23808873e+00 -6.15319550e-01 1.18487573e+00
-9.62161660e-01 -1.05439711e+00 -1.32654354e-01 2.01048061e-01
-2.63199389e-01 -1.33715183e-01 8.02151799e-01 2.48832673e-01
-6.60906255e-01 4.31902915e-01 3.66105169e-01 2.18101501e-01
-2.04869643e-01 -1.24204350e+00 9.25276875e-01 5.16901791e-01
-4.20786366e-02 9.05256271e-01 6.81822121e-01 -6.60765111e-01
-1.88517666e+00 -1.03002357e+00 5.25113940e-01 -1.94276005e-01
7.79749274e-01 -4.97599155e-01 -8.97386372e-01 3.80473077e-01
-1.43235996e-01 -1.58327818e-01 2.94223040e-01 6.21140301e-02
-3.39942396e-01 -6.12471660e-04 -1.18699253e+00 9.53414083e-01
8.60105753e-01 -5.63606679e-01 -4.38997358e-01 1.78354189e-01
1.06238365e+00 -1.72127917e-01 -8.86417329e-01 3.52701783e-01
4.89185810e-01 -7.08823979e-01 1.36068857e+00 -5.87939441e-01
2.42826268e-01 -1.45189211e-01 -2.13215604e-01 -1.21098936e+00
-9.05629218e-01 -4.13612783e-01 -3.47961754e-01 6.89269364e-01
1.34022653e-01 -9.97852445e-01 1.10023725e+00 6.32781029e-01
1.50328167e-02 -1.10181415e+00 -1.07142544e+00 -9.68417227e-01
2.04862401e-01 -3.29709232e-01 7.53004074e-01 8.27298939e-01
-1.63862973e-01 1.90957054e-01 1.27553821e-01 2.83094168e-01
5.61212063e-01 -2.58700311e-01 3.22618574e-01 -1.41975963e+00
-5.95164858e-02 -6.44430995e-01 -1.20211244e+00 -5.21444798e-01
1.51657924e-01 -1.03175211e+00 -4.31730449e-01 -1.58496308e+00
-3.31406057e-01 -5.34852564e-01 -4.26145077e-01 4.48514432e-01
4.64829326e-01 2.91798234e-01 2.38228574e-01 -1.55443490e-01
-9.82183591e-02 9.39414740e-01 1.24053776e+00 -1.69109389e-01
-2.88992792e-01 -5.53124130e-01 -5.17396271e-01 6.21419966e-01
8.62834930e-01 -3.21736038e-01 -5.70586085e-01 -4.69527423e-01
1.79584295e-01 -1.05985746e-01 7.11547494e-01 -1.40802646e+00
2.14188010e-01 2.67979149e-02 5.89371502e-01 -6.35507464e-01
6.61876023e-01 -8.80018413e-01 2.11459175e-01 5.19560218e-01
-3.09253484e-01 5.47886014e-01 6.60905778e-01 5.80631912e-01
8.91825408e-02 -1.90901652e-01 6.34200037e-01 1.81813523e-01
-5.75576007e-01 2.41396666e-01 -1.28260925e-01 -2.84947664e-01
6.26662016e-01 -7.66659856e-01 -2.37119749e-01 -4.17885855e-02
-6.40622616e-01 -7.01672360e-02 5.34854293e-01 4.18306351e-01
5.77073991e-01 -1.57635534e+00 4.51461263e-02 4.75785941e-01
-1.94219574e-01 -1.36979207e-01 2.75901020e-01 5.70988297e-01
-9.81523097e-01 7.62439668e-01 -6.06919587e-01 -4.72035587e-01
-7.69411206e-01 6.93168759e-01 4.20803815e-01 -2.72074491e-01
-1.24123418e+00 4.00433481e-01 9.57690924e-02 -8.62417340e-01
2.33310267e-01 -4.76597309e-01 -1.09804198e-01 -3.81483704e-01
9.10182968e-02 4.73006427e-01 1.18127309e-01 -5.80274105e-01
-2.43363529e-01 5.95218301e-01 7.55925849e-02 8.09613988e-02
1.36666286e+00 1.77315325e-01 -3.74862373e-01 2.80856460e-01
1.32737339e+00 -3.84653509e-01 -1.14164519e+00 -1.64065212e-01
-1.36327744e-01 -1.67670518e-01 6.96182609e-01 -7.59739637e-01
-1.15442574e+00 8.61090004e-01 6.32810771e-01 2.37846792e-01
8.50781679e-01 -5.32078803e-01 5.90994358e-01 7.56637573e-01
2.07306460e-01 -9.89154577e-01 -2.12308727e-02 8.12851131e-01
8.61262679e-01 -3.64117026e-01 1.38984889e-01 -1.87974319e-01
8.96483380e-03 1.36586368e+00 5.54792643e-01 -6.74988508e-01
7.54489541e-01 -8.43997449e-02 -4.54286098e-01 -6.04112744e-01
-2.98762411e-01 4.91447955e-01 7.22346723e-01 5.77780306e-01
2.39394717e-02 -3.06714289e-02 1.31881878e-01 2.06367627e-01
-5.15238225e-01 -3.48584980e-01 2.67282993e-01 9.63191330e-01
-2.75062412e-01 -6.63323879e-01 -1.75442532e-01 4.72434491e-01
2.37296462e-01 -1.09400183e-01 -4.39845920e-01 1.03604126e+00
-1.96034759e-02 4.07162458e-01 2.28777245e-01 -5.70005596e-01
2.39869446e-01 -1.84309613e-02 3.98069739e-01 -1.54987589e-01
-3.49816680e-01 -5.27227938e-01 -2.12154344e-01 -7.78532267e-01
1.81910858e-01 1.04801498e-01 -1.22304392e+00 -5.41323006e-01
-2.04046816e-01 3.83829504e-01 1.03370690e+00 8.09199154e-01
8.29447925e-01 9.04062033e-01 3.35756034e-01 -1.21946418e+00
-2.64181882e-01 -4.05357361e-01 -6.09527290e-01 -3.83481868e-02
4.74176198e-01 -1.12299681e+00 -1.37604535e-01 -6.26056433e-01] | [6.915254592895508, 6.128830909729004] |
0830284d-f351-452c-bae4-fea3d6b18555 | domain-adaptive-multiple-instance-learning | 2304.03537 | null | https://arxiv.org/abs/2304.03537v1 | https://arxiv.org/pdf/2304.03537v1.pdf | Domain Adaptive Multiple Instance Learning for Instance-level Prediction of Pathological Images | Pathological image analysis is an important process for detecting abnormalities such as cancer from cell images. However, since the image size is generally very large, the cost of providing detailed annotations is high, which makes it difficult to apply machine learning techniques. One way to improve the performance of identifying abnormalities while keeping the annotation cost low is to use only labels for each slide, or to use information from another dataset that has already been labeled. However, such weak supervisory information often does not provide sufficient performance. In this paper, we proposed a new task setting to improve the classification performance of the target dataset without increasing annotation costs. And to solve this problem, we propose a pipeline that uses multiple instance learning (MIL) and domain adaptation (DA) methods. Furthermore, in order to combine the supervisory information of both methods effectively, we propose a method to create pseudo-labels with high confidence. We conducted experiments on the pathological image dataset we created for this study and showed that the proposed method significantly improves the classification performance compared to existing methods. | ['Tatsuya Harada', 'Masaru Kitsuregawa', 'Masanobu Kitagawa', 'Masashi Fukayama', 'Tetsuo Ushiku', 'Akihiko Yoshizawa', 'Hiroyuki Abe', 'Yusuke Mukuta', 'Yusuke Kurose', 'Shusuke Takahama'] | 2023-04-07 | null | null | null | null | ['multiple-instance-learning'] | ['methodology'] | [ 5.55845618e-01 -1.75243653e-02 -2.41835624e-01 -4.03212398e-01
-1.03632224e+00 -3.71745795e-01 2.06428871e-01 6.61019742e-01
-5.70905387e-01 8.16117823e-01 -2.62270898e-01 -1.30324081e-01
8.52629095e-02 -7.54520893e-01 -4.90600556e-01 -8.54989231e-01
5.50534844e-01 4.33592558e-01 6.16311550e-01 3.82458955e-01
3.49419326e-01 3.07491332e-01 -1.37198985e+00 4.71017808e-01
1.12969422e+00 8.02671731e-01 4.61084932e-01 3.51257980e-01
-3.62652898e-01 7.83085108e-01 -6.56333089e-01 -2.36684069e-01
1.78631637e-02 -4.62199271e-01 -8.90713871e-01 3.54778737e-01
1.18104890e-01 -2.49026299e-01 3.18714976e-01 1.08929968e+00
2.31249601e-01 -1.06828492e-02 7.14303672e-01 -9.97553587e-01
-1.09292373e-01 2.93571830e-01 -7.60762990e-01 1.42541513e-01
-1.22349538e-01 -2.92500015e-03 8.44887853e-01 -4.41328526e-01
6.58680260e-01 6.98488116e-01 5.77300310e-01 5.40847182e-01
-1.10536230e+00 -8.04959238e-01 1.26261979e-01 7.30778351e-02
-1.36333668e+00 -2.79719859e-01 6.66274130e-01 -5.02703369e-01
3.76426578e-01 1.94605455e-01 3.49203497e-01 8.44871640e-01
-1.23628311e-01 8.26033652e-01 1.19464421e+00 -6.32208049e-01
2.67568558e-01 4.17734861e-01 1.63076133e-01 7.24866927e-01
2.96778113e-01 -4.54295456e-01 -7.80946687e-02 -2.75126517e-01
5.67785203e-01 3.30081373e-01 -7.64418691e-02 -2.37120315e-01
-1.10082054e+00 6.51305914e-01 3.84458274e-01 6.25545561e-01
-5.31894006e-02 -2.22904235e-01 4.53508377e-01 -1.57643054e-02
4.99946117e-01 5.90351105e-01 -3.87002081e-01 6.45794496e-02
-9.30775762e-01 -3.39981824e-01 5.54390192e-01 3.74418020e-01
7.75487125e-01 -6.19408071e-01 -2.44601339e-01 1.01996577e+00
4.78358753e-02 3.28451432e-02 6.09099448e-01 -6.53992355e-01
2.61486113e-01 1.11546838e+00 2.58982014e-02 -8.94790888e-01
-5.46872437e-01 -5.10081708e-01 -8.38635623e-01 9.09151211e-02
7.86775589e-01 1.42248601e-01 -1.13211942e+00 1.61502278e+00
2.85313278e-01 3.95749956e-01 -1.52566517e-02 8.59781146e-01
6.44325435e-01 4.09290999e-01 2.10340977e-01 -3.19843382e-01
1.28776252e+00 -9.47729349e-01 -7.08311379e-01 -3.42346840e-02
1.13402128e+00 -6.01617396e-01 1.18638575e+00 3.52202296e-01
-5.25895715e-01 -3.92568409e-01 -9.40698802e-01 1.02524802e-01
-2.39079505e-01 5.12700617e-01 4.54184562e-01 4.77996349e-01
-6.36465311e-01 3.31461579e-01 -1.11978793e+00 -5.21993339e-01
5.41928649e-01 4.30908054e-01 -4.67991590e-01 -1.86226696e-01
-7.23047018e-01 7.63738155e-01 6.44100785e-01 -1.13941558e-01
-6.60833836e-01 -4.57998723e-01 -6.21853769e-01 8.44474286e-02
6.26302063e-01 -2.74780959e-01 1.05248475e+00 -1.18464744e+00
-1.21578848e+00 8.51388991e-01 -3.15780908e-01 -2.76418298e-01
4.64789361e-01 3.21881361e-02 -1.09157801e-01 2.78020799e-01
1.24252431e-01 6.10510647e-01 6.57734096e-01 -1.28595591e+00
-8.76941979e-01 -4.39453572e-01 3.74303013e-02 2.83984262e-02
-7.33415961e-01 -2.02950925e-01 -6.54050231e-01 -4.64146525e-01
5.22034355e-02 -9.92796540e-01 -4.09515679e-01 -1.26907632e-01
-3.33494246e-01 -1.93104237e-01 8.36915433e-01 -7.12967873e-01
1.07577276e+00 -2.46459293e+00 -1.02485627e-01 4.47491020e-01
2.04860374e-01 5.08062243e-01 4.36520316e-02 -9.05213282e-02
2.07692459e-01 2.96470702e-01 -4.92524445e-01 -3.14363182e-01
-5.46610117e-01 3.68418515e-01 2.05018744e-01 1.97324887e-01
2.72562355e-01 5.39216995e-01 -8.94391179e-01 -9.00167882e-01
5.94288185e-02 1.69523597e-01 -5.07096052e-01 2.66105831e-01
-2.78354913e-01 7.70761490e-01 -4.32982564e-01 5.51110625e-01
4.45784450e-01 -6.19837224e-01 2.93812692e-01 -2.31661633e-01
2.59673655e-01 -7.54843876e-02 -1.22063601e+00 1.53514767e+00
-5.02371550e-01 2.26557970e-01 -1.37837783e-01 -1.21498990e+00
9.14163470e-01 3.77178460e-01 5.79325914e-01 -4.33253348e-01
1.36190474e-01 3.34265292e-01 1.59778029e-01 -7.01463878e-01
-1.33905299e-02 -1.07090302e-01 1.87165484e-01 3.72357249e-01
-1.14230052e-01 1.13179423e-01 3.86485159e-01 -1.65519565e-02
1.31150579e+00 -1.97557777e-01 4.51757759e-01 1.19786605e-01
8.32845628e-01 2.04504967e-01 9.04558718e-01 5.18602133e-01
-2.73669243e-01 5.36669016e-01 6.29397929e-01 -2.84230262e-01
-8.60199988e-01 -5.47462821e-01 -1.83233216e-01 7.63979733e-01
1.84654340e-01 -4.48215514e-01 -7.73663044e-01 -1.10273969e+00
-8.37999135e-02 3.61902058e-01 -6.21323824e-01 -9.46331099e-02
-4.00922716e-01 -1.00690627e+00 5.05223095e-01 5.40222406e-01
5.44958115e-01 -7.71817565e-01 -4.38478172e-01 1.53967500e-01
-3.93358380e-01 -1.15034318e+00 -2.97962487e-01 1.30389139e-01
-1.03349268e+00 -1.19135320e+00 -5.82211018e-01 -8.19064379e-01
1.27606881e+00 1.58637777e-01 6.41945958e-01 5.06064057e-01
-3.30158770e-01 -2.19078317e-01 -5.61984718e-01 -4.14264768e-01
-5.22721827e-01 3.48767221e-01 -2.29289070e-01 3.10213506e-01
3.40384007e-01 -1.17486320e-01 -3.86274040e-01 4.16379184e-01
-1.13662660e+00 2.40689337e-01 9.96972919e-01 1.10631597e+00
8.01796317e-01 5.44438481e-01 7.42606461e-01 -1.34923387e+00
2.60467201e-01 -9.36667621e-02 -5.00845373e-01 2.60660470e-01
-5.88476956e-01 5.10439985e-02 7.18596458e-01 -4.40108448e-01
-1.15485477e+00 4.75273192e-01 -1.84400901e-01 -1.70612007e-01
-3.18838507e-01 6.55205309e-01 -8.68149325e-02 -8.52078497e-02
5.68429112e-01 -2.06143204e-02 2.87076205e-01 -4.86610740e-01
-2.29478434e-01 9.08068240e-01 3.59648108e-01 -3.17542255e-01
4.42323446e-01 5.39326847e-01 6.67945594e-02 -5.90221047e-01
-1.09633613e+00 -7.59947538e-01 -7.57494032e-01 6.00467920e-02
8.16965759e-01 -6.70640290e-01 -3.57506722e-01 3.50958198e-01
-8.13188016e-01 -2.35684469e-01 9.67874005e-02 6.23961270e-01
-1.17620364e-01 3.96660686e-01 -6.08157158e-01 -6.83848381e-01
-1.36802584e-01 -1.37880099e+00 9.70795989e-01 2.57721394e-01
-2.04426661e-01 -9.19372201e-01 -8.63141492e-02 6.28995121e-01
3.07546228e-01 3.25753689e-01 9.50064480e-01 -8.97260308e-01
-4.90870059e-01 -3.79573584e-01 -3.25009227e-01 3.36346686e-01
4.78889167e-01 1.17909990e-01 -9.30297971e-01 -3.35341990e-01
-2.81508500e-03 -2.26535305e-01 9.25864160e-01 2.04723671e-01
1.57204759e+00 -1.79585405e-02 -7.62658179e-01 2.65077055e-01
1.38565040e+00 2.22423181e-01 4.37236249e-01 4.25029159e-01
6.46019399e-01 6.74196243e-01 9.32608068e-01 1.69827744e-01
1.79542288e-01 7.40813971e-01 2.22201586e-01 -4.59689349e-01
-2.97300220e-02 -1.97220147e-02 5.96929491e-02 4.64731276e-01
4.88222651e-02 -9.29103121e-02 -1.02977026e+00 6.18578196e-01
-2.04204082e+00 -6.12979412e-01 -1.07514627e-01 2.23122311e+00
9.79818583e-01 2.74884760e-01 -6.79433346e-02 4.63886708e-01
7.97905385e-01 -5.00431240e-01 -3.36277515e-01 -1.92283781e-03
2.15450808e-01 2.19956245e-02 4.17992324e-01 2.66686678e-01
-1.19744849e+00 6.67764843e-01 6.04520035e+00 9.25505698e-01
-1.21778917e+00 2.35548586e-01 8.41581464e-01 1.48943961e-01
2.95697921e-03 -5.45179993e-02 -7.89707363e-01 5.44204116e-01
7.14113712e-01 1.40484139e-01 -2.02490598e-01 7.74093270e-01
2.40380943e-01 -4.68764901e-01 -1.11282575e+00 8.67279172e-01
1.31018043e-01 -1.08107781e+00 -1.19299740e-01 1.16981283e-01
6.87567532e-01 -4.33954418e-01 -2.96465546e-01 1.68476805e-01
-4.00261134e-02 -7.97252655e-01 -1.37145482e-02 2.46341065e-01
5.91990292e-01 -6.45425677e-01 1.28272343e+00 6.52361393e-01
-7.87010372e-01 -8.96674171e-02 -2.45685339e-01 1.49080768e-01
-1.90324113e-01 1.01107728e+00 -1.42580032e+00 4.61538553e-01
5.11165798e-01 6.19529903e-01 -8.29070330e-01 1.27738237e+00
-2.72898674e-01 7.97571719e-01 -2.88707852e-01 1.27389550e-01
-4.82204231e-03 1.48712128e-01 1.52082667e-01 1.17810845e+00
3.22502822e-01 -5.91069367e-03 5.56515515e-01 4.71784443e-01
-1.08642563e-01 3.48001748e-01 -4.84306902e-01 -2.13851538e-02
2.18784720e-01 1.56804597e+00 -1.19828784e+00 -3.86794388e-01
-4.15461361e-01 8.18416715e-01 3.76571864e-01 -3.44806798e-02
-7.12392807e-01 -2.69609362e-01 9.93187577e-02 1.79464519e-01
-3.64321433e-02 4.58925851e-02 -2.64363796e-01 -1.09736097e+00
-3.59114408e-02 -6.93974137e-01 7.24510610e-01 -3.17281961e-01
-1.24846017e+00 3.43245119e-01 -2.20776096e-01 -1.30541694e+00
-8.57290551e-02 -4.87088412e-01 -3.96279067e-01 5.54800510e-01
-1.57473063e+00 -1.18269360e+00 -5.25269151e-01 3.28821242e-01
4.24474925e-01 8.16275179e-02 7.69302666e-01 5.73162198e-01
-8.11914980e-01 6.99552357e-01 -5.94442971e-02 2.46904492e-01
1.16379404e+00 -1.21108568e+00 -4.70838487e-01 7.98310935e-01
9.82611673e-04 3.46380621e-01 3.11361492e-01 -6.09413505e-01
-8.39141428e-01 -1.10197258e+00 6.45364285e-01 -3.15929532e-01
3.22466433e-01 -1.67361721e-01 -1.32217205e+00 4.88155603e-01
-2.49119520e-01 1.12085111e-01 1.04272139e+00 1.51423529e-01
-5.74225858e-02 -2.68674374e-01 -1.36613262e+00 3.14289272e-01
5.07677555e-01 -2.57175416e-01 -4.22590792e-01 3.59351486e-01
4.45937157e-01 -5.03536224e-01 -7.83659399e-01 6.17510438e-01
2.61544317e-01 -6.00850523e-01 5.92807889e-01 -2.28481203e-01
2.52551794e-01 -7.83906281e-01 8.30605701e-02 -1.21785831e+00
-2.47250557e-01 1.26466408e-01 1.47870913e-01 1.53030789e+00
5.91909826e-01 -6.09387338e-01 8.95572782e-01 6.70158267e-01
-6.07115254e-02 -5.90633392e-01 -6.81215286e-01 -7.26329446e-01
-3.25535893e-01 -1.00309372e-01 4.02057141e-01 1.02917254e+00
8.16121697e-02 3.02403837e-01 -9.19568166e-02 2.95989871e-01
4.16895658e-01 1.57696158e-01 7.77525604e-01 -1.27346039e+00
-2.56526202e-01 -1.89829990e-01 -5.65896511e-01 -3.52819622e-01
1.52224034e-01 -1.11971462e+00 1.82029575e-01 -1.46074998e+00
7.21761823e-01 -8.14526558e-01 -6.35351181e-01 8.58891904e-01
-5.56244910e-01 4.95380342e-01 -9.47238728e-02 4.65664923e-01
-6.03311598e-01 -9.19394288e-03 1.23618960e+00 -2.24905148e-01
-2.48299822e-01 8.34786706e-03 -6.90361440e-01 7.35614419e-01
8.55917394e-01 -5.75016081e-01 -3.06250870e-01 -3.29712123e-01
-1.77622482e-01 -2.01195702e-01 8.57050344e-02 -1.05820012e+00
2.82126158e-01 -1.05774671e-01 5.60424030e-01 -4.53764260e-01
9.57667604e-02 -9.53562617e-01 6.17966428e-02 6.07344568e-01
-3.72661352e-01 -5.24989128e-01 1.18268535e-01 6.39706910e-01
-5.56275547e-01 -4.85516101e-01 1.03348100e+00 -1.76972806e-01
-6.67469621e-01 8.18895325e-02 -1.84124097e-01 -3.36142391e-01
1.36052954e+00 -5.28156310e-02 -3.20509106e-01 -2.73160245e-02
-6.09091461e-01 3.88509780e-01 6.65168941e-01 7.61867315e-02
3.51424217e-01 -1.00538278e+00 -5.67175031e-01 6.12810515e-02
4.65186715e-01 2.10738540e-01 7.58089200e-02 1.09768867e+00
-4.73590344e-01 2.54972041e-01 -1.69701040e-01 -9.17738378e-01
-1.51858044e+00 5.87075353e-01 1.01912417e-01 -7.52331555e-01
-4.91227955e-01 6.48116052e-01 2.78006881e-01 -4.19148684e-01
1.07993156e-01 -2.97812760e-01 -5.08697748e-01 5.64040318e-02
5.36565185e-01 7.13564903e-02 3.68051440e-01 -3.11726630e-01
-3.03647548e-01 4.29694057e-01 -3.77016395e-01 8.95953253e-02
1.12982070e+00 1.55849164e-04 -1.69534579e-01 3.93016279e-01
1.00451887e+00 6.74661901e-03 -9.17423427e-01 -2.33856753e-01
3.67622823e-01 -6.50169492e-01 1.19099405e-03 -7.88555205e-01
-1.08383310e+00 8.48033249e-01 6.38967156e-01 1.38367996e-01
1.41902590e+00 -1.12398721e-01 7.01857507e-01 2.86512852e-01
2.87996769e-01 -1.08004367e+00 1.46390468e-01 1.20095350e-01
2.78276592e-01 -1.65100527e+00 -9.50591415e-02 -7.75745213e-01
-6.73857331e-01 1.07323349e+00 9.15740252e-01 2.60734856e-01
3.95589769e-01 2.71916777e-01 1.57042906e-01 5.77903911e-02
-6.90613151e-01 -2.88145721e-01 1.11124046e-01 3.46062541e-01
4.97903883e-01 -6.30408376e-02 -5.45900404e-01 5.31722128e-01
3.59356821e-01 3.09451073e-01 5.09427965e-01 1.04438901e+00
-4.27340090e-01 -1.54868782e+00 -4.64879096e-01 8.47654343e-01
-9.29343760e-01 3.67962062e-01 -3.26258570e-01 5.46060741e-01
2.49088421e-01 9.55552101e-01 -2.79530995e-02 -2.37140283e-01
5.72967753e-02 1.08270101e-01 2.56407022e-01 -8.77782285e-01
-3.14090490e-01 2.23731264e-01 2.08660625e-02 -1.70912042e-01
-7.09185600e-01 -4.03794169e-01 -1.50283647e+00 2.73222774e-01
-6.38950229e-01 2.47476473e-01 4.89420474e-01 1.11009371e+00
2.65392929e-01 6.32509410e-01 5.86813509e-01 -2.70653367e-01
-1.65529594e-01 -9.62238431e-01 -4.80315238e-01 6.98386788e-01
1.33369818e-01 -8.15570414e-01 -2.29549631e-01 4.00079548e-01] | [15.018019676208496, -2.82610821723938] |
527f0b85-8aea-4a04-a76e-4768ff4e1020 | guiding-generative-language-models-for-data | 2111.09064 | null | https://arxiv.org/abs/2111.09064v2 | https://arxiv.org/pdf/2111.09064v2.pdf | Guiding Generative Language Models for Data Augmentation in Few-Shot Text Classification | Data augmentation techniques are widely used for enhancing the performance of machine learning models by tackling class imbalance issues and data sparsity. State-of-the-art generative language models have been shown to provide significant gains across different NLP tasks. However, their applicability to data augmentation for text classification tasks in few-shot settings have not been fully explored, especially for specialised domains. In this paper, we leverage GPT-2 (Radford A et al, 2019) for generating artificial training instances in order to improve classification performance. Our aim is to analyse the impact the selection process of seed training examples have over the quality of GPT-generated samples and consequently the classifier performance. We perform experiments with several seed selection strategies that, among others, exploit class hierarchical structures and domain expert selection. Our results show that fine-tuning GPT-2 in a handful of label instances leads to consistent classification improvements and outperform competitive baselines. Finally, we show that guiding this process through domain expert selection can lead to further improvements, which opens up interesting research avenues for combining generative models and active learning. | ['Alun Preece', 'Hélène de Ribaupierre', 'Jose Camacho-Collados', 'Asahi Ushio', 'Aleksandra Edwards'] | 2021-11-17 | null | null | null | null | ['few-shot-text-classification'] | ['natural-language-processing'] | [ 4.93416220e-01 4.00851935e-01 -6.60672426e-01 -4.94701296e-01
-8.87013912e-01 -3.13976884e-01 9.34357524e-01 5.47406733e-01
-4.43792224e-01 8.80632758e-01 2.73313642e-01 -1.69654891e-01
-1.00872830e-01 -8.09071720e-01 -2.93382734e-01 -7.13899612e-01
5.67936972e-02 1.06553733e+00 9.31351185e-02 -2.07183927e-01
2.37808302e-01 2.64571130e-01 -1.66037488e+00 4.74518508e-01
1.04654431e+00 4.17345673e-01 -3.06390852e-01 4.07266587e-01
-3.17445785e-01 7.93626845e-01 -6.79887593e-01 -5.19997656e-01
4.45133820e-02 -4.81623769e-01 -8.58147025e-01 3.53981256e-01
3.56275529e-01 9.54267383e-02 9.22679454e-02 6.16006076e-01
7.49976337e-01 2.82294214e-01 7.76628256e-01 -1.31502438e+00
-5.40187776e-01 9.58805263e-01 -5.08932292e-01 2.93061078e-01
1.44563075e-02 1.65493622e-01 1.06382918e+00 -9.21793699e-01
5.62820077e-01 1.23059261e+00 7.14159608e-01 5.59976220e-01
-1.57351422e+00 -7.58416414e-01 3.50188583e-01 1.41262218e-01
-1.04174542e+00 -6.69702053e-01 7.37424970e-01 -2.29437545e-01
1.04253983e+00 1.70508981e-01 5.12131631e-01 1.30790758e+00
-3.40578765e-01 9.90152895e-01 1.22521758e+00 -1.01453221e+00
4.14112240e-01 4.93152589e-01 5.83891749e-01 3.96596372e-01
4.71329868e-01 -1.25709146e-01 -7.41309822e-01 -5.18064976e-01
3.44981641e-01 -2.79314399e-01 -1.24012850e-01 -4.04240608e-01
-8.12881947e-01 1.36681604e+00 2.18936652e-01 3.95520955e-01
-3.60497117e-01 -1.25296250e-01 4.01725590e-01 2.16105193e-01
1.16856337e+00 9.90199625e-01 -5.49096465e-01 -1.18808709e-01
-1.14236128e+00 3.19313198e-01 8.11174989e-01 7.07453847e-01
5.56660116e-01 1.89545639e-02 -5.92915475e-01 1.16797352e+00
1.82225689e-01 2.11255886e-02 6.00452483e-01 -4.75179255e-01
5.90888739e-01 9.33282971e-01 -1.10146992e-01 -4.74446177e-01
-3.56591254e-01 -6.66303158e-01 -6.42640471e-01 1.19345710e-02
2.94256777e-01 -2.64810801e-01 -1.14323497e+00 1.59839904e+00
3.41542274e-01 8.61377344e-02 6.12757653e-02 4.18600053e-01
8.34070981e-01 4.76590216e-01 5.47218621e-01 -3.52357268e-01
1.20402133e+00 -9.20684338e-01 -7.64601231e-01 -6.63092017e-01
1.02307785e+00 -7.37402976e-01 1.16540504e+00 3.47067326e-01
-9.57891226e-01 -3.92907947e-01 -7.94548154e-01 2.00436026e-01
-3.32856238e-01 2.24571854e-01 1.01164412e+00 9.61564660e-01
-6.36150301e-01 4.97092724e-01 -7.82294571e-01 -4.11613047e-01
1.06794894e+00 4.43304896e-01 -1.28249675e-01 -1.61845356e-01
-1.22524786e+00 8.22579026e-01 4.91791517e-01 -4.12276775e-01
-6.23093426e-01 -8.53251934e-01 -7.47045040e-01 5.49695008e-02
5.66643894e-01 -6.02083564e-01 1.29194558e+00 -7.75844693e-01
-1.22360957e+00 9.64479446e-01 -4.61462028e-02 -8.61718535e-01
4.78423864e-01 -3.30774188e-01 -1.82789654e-01 -3.32684696e-01
7.71454647e-02 8.77695084e-01 6.12896979e-01 -1.14852750e+00
-5.29311061e-01 -3.73690963e-01 -2.19220385e-01 4.14592147e-01
-8.12506557e-01 1.24716312e-01 -2.92091608e-01 -7.88337111e-01
-1.76880002e-01 -9.03338969e-01 -5.88294625e-01 -6.58850074e-01
-3.85126293e-01 -4.41055745e-01 6.78093135e-01 -1.57585964e-01
1.21474314e+00 -1.90907550e+00 -1.49919480e-01 1.25548035e-01
7.24322572e-02 6.07267141e-01 -9.99932066e-02 4.20773089e-01
-1.39123201e-01 3.47224563e-01 -2.74744421e-01 -5.61910093e-01
-1.08336043e-02 3.44232023e-01 -4.40312982e-01 1.11164160e-01
6.43189728e-01 1.10424280e+00 -7.30131388e-01 -4.32556182e-01
9.59962308e-02 4.61756140e-01 -6.36239409e-01 1.80631727e-01
-4.44396138e-01 3.61658126e-01 -1.93868548e-01 1.89221710e-01
3.27547550e-01 -5.04706442e-01 1.20326400e-01 3.22984815e-01
2.69834340e-01 7.15626657e-01 -1.10929370e+00 1.25806773e+00
-3.30522895e-01 5.42469323e-01 -6.34521186e-01 -1.06158149e+00
1.09775412e+00 2.45259449e-01 1.39278144e-01 -6.80624366e-01
1.18499070e-01 -1.42418221e-03 1.96826801e-01 -2.89372951e-01
5.64561844e-01 -2.23741919e-01 -1.86301544e-02 6.06285751e-01
2.05566958e-01 -5.17784171e-02 6.51167572e-01 2.84000188e-01
8.90761733e-01 1.87714584e-02 4.57097471e-01 -2.07935303e-01
1.34096995e-01 3.52980882e-01 3.32623541e-01 1.07965124e+00
1.34833664e-01 6.37040019e-01 5.05102575e-01 -1.79369181e-01
-1.13554883e+00 -4.17862266e-01 -2.23766476e-01 1.49783933e+00
-5.10376990e-01 -3.72294724e-01 -6.58129752e-01 -9.22917545e-01
-1.18644081e-01 1.23025167e+00 -8.37652683e-01 -5.01284540e-01
-4.72579688e-01 -1.68707442e+00 4.20445383e-01 6.76797211e-01
3.70526701e-01 -1.23522687e+00 -3.85276645e-01 2.91956484e-01
3.22674774e-02 -1.10224545e+00 6.03467152e-02 7.15481699e-01
-1.21985781e+00 -9.34274495e-01 -6.54679716e-01 -6.85672462e-01
7.47151852e-01 -9.16779339e-02 1.46848726e+00 8.44722763e-02
-1.88366264e-01 2.36610144e-01 -7.95070231e-01 -7.81713128e-01
-7.10579216e-01 7.18314290e-01 -2.17474893e-01 -1.85791790e-01
6.94872499e-01 -3.17776680e-01 -1.64326489e-01 2.09446758e-01
-8.99088085e-01 1.60864279e-01 5.20014107e-01 8.61581326e-01
4.39289868e-01 4.36374471e-02 8.17248404e-01 -1.75467455e+00
9.95979369e-01 -3.38151395e-01 -3.52992296e-01 1.85027301e-01
-1.01566446e+00 2.37039775e-01 3.17946911e-01 -6.59564555e-01
-1.17037427e+00 -1.09650224e-01 -2.08266273e-01 7.56810755e-02
-2.46255308e-01 4.55934435e-01 -4.21594828e-02 1.02811329e-01
1.17409801e+00 -4.75856885e-02 -1.82230204e-01 -3.66473764e-01
2.85857439e-01 5.97653747e-01 4.88184243e-02 -4.51566696e-01
6.45312369e-01 1.74224898e-01 -2.53103256e-01 -5.81091702e-01
-1.34833908e+00 -5.63300312e-01 -7.34672129e-01 2.45630622e-01
3.96025270e-01 -9.48958337e-01 1.89343572e-01 2.50687152e-01
-6.94020271e-01 -5.64529538e-01 -6.84056520e-01 3.01498145e-01
-3.12913179e-01 4.68415543e-02 -3.95108163e-01 -9.24874425e-01
-5.85758865e-01 -9.67047453e-01 1.03716242e+00 3.02947462e-01
-5.98989308e-01 -1.15497410e+00 3.29056799e-01 5.09665549e-01
4.24788117e-01 -1.33330911e-01 9.42642570e-01 -1.44177055e+00
-2.04434484e-01 -3.45256001e-01 1.05652303e-01 2.89854884e-01
-3.52614038e-02 -2.50944227e-01 -1.17638087e+00 -3.41882199e-01
-2.53708720e-01 -5.44019878e-01 1.04831064e+00 3.46416533e-01
1.11285639e+00 -2.67075211e-01 -4.59059119e-01 1.74504355e-01
1.04942763e+00 1.80562615e-01 8.09133410e-01 4.72389996e-01
5.83020449e-01 6.32911563e-01 7.96130955e-01 4.22822356e-01
-7.36112818e-02 6.57087564e-01 5.13203964e-02 -3.39283496e-01
-2.49642178e-01 1.88681222e-02 -1.38176717e-02 3.44290972e-01
1.20118096e-01 -4.79497641e-01 -1.17211187e+00 5.37678301e-01
-1.61954081e+00 -7.67870069e-01 -3.43242913e-01 2.26018214e+00
1.23977017e+00 4.78667527e-01 2.73701161e-01 6.11538112e-01
5.82016170e-01 -1.93972126e-01 -3.04175109e-01 -2.62720734e-01
-1.35039404e-01 6.87782526e-01 2.54696906e-01 2.75092363e-01
-1.15048146e+00 9.92549181e-01 6.45282269e+00 8.61236513e-01
-8.59655619e-01 1.71548203e-01 1.02352178e+00 -5.80176935e-02
-6.87526315e-02 2.60747317e-02 -1.36572540e+00 3.35918635e-01
9.91818845e-01 -1.08339027e-01 -6.81204423e-02 8.49309921e-01
-2.50039008e-02 -1.30496651e-01 -1.10278797e+00 5.78318834e-01
1.74570456e-01 -1.29467940e+00 7.92768598e-02 2.56955326e-01
1.08438182e+00 4.33368003e-03 1.29350469e-01 7.12725461e-01
6.61270916e-01 -1.01617551e+00 2.81049728e-01 5.30503392e-02
3.76828104e-01 -8.61631811e-01 1.03379476e+00 5.09876251e-01
-4.23489571e-01 -2.04771664e-02 -2.53128946e-01 -5.39402813e-02
-1.08954031e-03 9.06403065e-01 -1.56932449e+00 2.93145537e-01
4.46246982e-01 3.83091360e-01 -1.06837320e+00 1.26573253e+00
-3.21122319e-01 1.18288839e+00 -2.69505918e-01 -1.34225190e-01
1.59045056e-01 2.49196991e-01 3.62662792e-01 1.30053449e+00
-1.12331785e-01 -9.79321748e-02 1.07092358e-01 7.71536946e-01
-2.30238810e-01 3.87560993e-01 -3.75393778e-01 -3.82298768e-01
5.13087571e-01 1.16122186e+00 -1.07155335e+00 -5.89828670e-01
-1.70489833e-01 5.16175508e-01 2.94924825e-01 1.80232227e-01
-5.79258502e-01 -9.05795023e-02 2.65696317e-01 4.07163948e-01
2.70982116e-01 1.31031126e-01 -5.17876804e-01 -9.90007043e-01
-3.76971751e-01 -1.12176216e+00 5.21703243e-01 -4.75888103e-01
-1.44701695e+00 4.79106992e-01 1.14659503e-01 -9.47994053e-01
-4.10926789e-01 -4.17741001e-01 -5.62443554e-01 7.72125304e-01
-1.38927150e+00 -9.07806933e-01 -3.70736748e-01 6.70270026e-02
7.13576257e-01 -3.53835136e-01 1.04404855e+00 2.01469377e-01
-6.85363829e-01 7.81910181e-01 7.00935200e-02 1.48723006e-01
7.56259441e-01 -1.27433443e+00 9.13173020e-01 8.32490265e-01
5.62098742e-01 5.71441352e-01 7.44240582e-01 -8.05297017e-01
-6.61192477e-01 -1.20915067e+00 8.62757385e-01 -6.70608878e-01
3.36556077e-01 -5.62976897e-01 -1.10562980e+00 6.54796362e-01
2.14691609e-01 -1.09983914e-01 1.07277179e+00 5.30442417e-01
-1.98085636e-01 2.45103076e-01 -1.09057164e+00 3.60322058e-01
8.38051140e-01 -1.86522771e-02 -5.39128900e-01 3.57793272e-01
5.37425637e-01 -3.54503006e-01 -6.27565920e-01 5.96950650e-01
3.94621789e-02 -6.95929170e-01 9.06411469e-01 -9.80868876e-01
4.74498123e-01 3.22007120e-01 3.72183681e-01 -1.43663871e+00
-3.82867157e-01 -3.41352671e-01 -1.43755957e-01 1.61421382e+00
7.77244389e-01 -5.62798738e-01 1.11658823e+00 5.58816075e-01
-4.18720534e-03 -7.99923182e-01 -4.95884031e-01 -5.05444169e-01
1.59732088e-01 -4.31582630e-01 4.28748369e-01 1.18005097e+00
-2.22183987e-01 9.20176566e-01 -2.91112930e-01 -4.11262184e-01
4.29591745e-01 -5.54646105e-02 8.57921660e-01 -1.68796718e+00
-1.96950585e-01 -3.94470900e-01 -2.06851184e-01 -5.73732734e-01
6.36583194e-02 -1.04334760e+00 -8.37665126e-02 -1.66368866e+00
3.90361279e-01 -7.83218980e-01 3.73306684e-02 8.81771386e-01
-6.11046910e-01 4.89469379e-01 1.15571050e-02 1.25750780e-01
-6.88278794e-01 5.29799998e-01 9.72947240e-01 8.28981102e-02
-4.68141258e-01 2.55176246e-01 -1.03619242e+00 4.61309016e-01
9.04377222e-01 -7.02816963e-01 -5.70501029e-01 -8.90048966e-02
2.74898410e-01 -5.91594279e-01 1.43620789e-01 -8.69034469e-01
-3.60234082e-02 7.24577457e-02 5.54683864e-01 -4.23857421e-01
2.17205361e-01 -4.87875283e-01 -1.54401258e-01 2.44523466e-01
-7.46108174e-01 -1.72219723e-01 4.50904101e-01 4.88544554e-01
-1.12413336e-02 -6.34714723e-01 7.62187541e-01 -1.29517168e-01
-4.80053157e-01 2.04486281e-01 -3.63765091e-01 5.00262678e-01
7.94569671e-01 -4.00044322e-02 -3.23988646e-01 -5.19467667e-02
-7.32356071e-01 8.85790661e-02 2.52218485e-01 5.22328675e-01
1.36923686e-01 -1.24918640e+00 -8.62003624e-01 3.12413037e-01
3.05894554e-01 1.43321216e-01 -2.07105383e-01 7.89531469e-01
4.94365431e-02 5.15795410e-01 8.87655392e-02 -6.48916602e-01
-1.54164195e+00 3.27032387e-01 7.75669441e-02 -7.75098860e-01
-4.23587680e-01 1.03633285e+00 -6.81992620e-02 -4.98285741e-01
3.39408278e-01 -1.62742455e-02 -5.09946764e-01 5.18193960e-01
3.74239236e-01 2.12251723e-01 6.07281864e-01 -1.75945774e-01
-8.84027332e-02 -7.76626319e-02 -6.16114438e-01 -2.94887763e-03
1.60374188e+00 3.13364357e-01 2.08407074e-01 3.67091984e-01
7.98876047e-01 -3.08709711e-01 -8.18617880e-01 -4.88304287e-01
2.55243897e-01 -4.13325548e-01 1.71337590e-01 -1.09771478e+00
-8.28656554e-01 7.13229060e-01 5.78440428e-01 3.89521569e-01
1.00160348e+00 1.41782612e-02 2.80788869e-01 3.05742502e-01
1.56796902e-01 -9.95689631e-01 2.67521977e-01 4.81549591e-01
4.34945136e-01 -1.51776063e+00 1.53068662e-01 -6.24639809e-01
-7.73239911e-01 7.70406425e-01 6.10628963e-01 6.79154694e-02
2.40833625e-01 3.18813503e-01 -1.16157636e-01 -2.39189282e-01
-1.01540339e+00 -3.54602903e-01 4.64202255e-01 5.53801358e-01
8.34834158e-01 -1.05313018e-01 -2.99466252e-01 3.35457087e-01
-2.49591321e-01 2.97239609e-02 2.98734456e-01 9.29134846e-01
-4.96246845e-01 -1.45842528e+00 -4.08838242e-01 9.57622528e-01
-5.20171762e-01 -4.00593579e-01 -6.04169071e-01 9.35765564e-01
3.67613658e-02 8.01778495e-01 7.88503066e-02 1.23727380e-03
2.86222845e-01 6.22044265e-01 5.54422021e-01 -1.27386141e+00
-7.91390002e-01 7.06721023e-02 4.43289191e-01 2.15615854e-01
-5.14576614e-01 -9.00514126e-01 -9.83586788e-01 6.47810996e-02
-7.83735871e-01 3.95069540e-01 4.11331207e-01 9.97345805e-01
2.49680996e-01 6.48268342e-01 2.53168732e-01 -4.48184103e-01
-5.71009994e-01 -1.34674346e+00 -1.61851555e-01 4.35527354e-01
-2.26698294e-01 -7.94269085e-01 -4.28876460e-01 -1.15249399e-03] | [10.838948249816895, 8.22099781036377] |
73ca19a0-ebf3-414a-93cf-80e71dbe5c5e | long-horizon-video-prediction-using-a-dynamic | 2212.14376 | null | https://arxiv.org/abs/2212.14376v2 | https://arxiv.org/pdf/2212.14376v2.pdf | Long-horizon video prediction using a dynamic latent hierarchy | The task of video prediction and generation is known to be notoriously difficult, with the research in this area largely limited to short-term predictions. Though plagued with noise and stochasticity, videos consist of features that are organised in a spatiotemporal hierarchy, different features possessing different temporal dynamics. In this paper, we introduce Dynamic Latent Hierarchy (DLH) -- a deep hierarchical latent model that represents videos as a hierarchy of latent states that evolve over separate and fluid timescales. Each latent state is a mixture distribution with two components, representing the immediate past and the predicted future, causing the model to learn transitions only between sufficiently dissimilar states, while clustering temporally persistent states closer together. Using this unique property, DLH naturally discovers the spatiotemporal structure of a dataset and learns disentangled representations across its hierarchy. We hypothesise that this simplifies the task of modeling temporal dynamics of a video, improves the learning of long-term dependencies, and reduces error accumulation. As evidence, we demonstrate that DLH outperforms state-of-the-art benchmarks in video prediction, is able to better represent stochasticity, as well as to dynamically adjust its hierarchical and temporal structure. Our paper shows, among other things, how progress in representation learning can translate into progress in prediction tasks. | ['Zafeirios Fountas', 'Qinghai Guo', 'Alexey Zakharov'] | 2022-12-29 | null | null | null | null | ['video-prediction'] | ['computer-vision'] | [ 9.84411538e-02 -2.60421224e-02 -4.32279646e-01 -9.54866633e-02
-1.86608687e-01 -6.76509798e-01 9.05839980e-01 -7.79453516e-02
2.21945956e-01 5.19234180e-01 9.47422683e-01 -1.75519530e-02
-2.82791197e-01 -5.37842393e-01 -8.69933844e-01 -9.00430381e-01
-7.40400910e-01 4.81224805e-01 3.29146862e-01 -2.78558675e-02
-8.43473822e-02 3.14632714e-01 -1.83016920e+00 8.08115602e-01
2.89705843e-01 9.45304990e-01 1.16192162e-01 8.88743937e-01
3.11506987e-01 1.46292579e+00 -1.40879676e-01 -2.75445431e-01
1.61619231e-01 -6.21492982e-01 -7.14736044e-01 2.75357902e-01
4.59882975e-01 -2.47705892e-01 -9.64959860e-01 5.14697373e-01
-1.86665863e-01 1.27727330e-01 7.41621256e-01 -1.24073708e+00
-6.39791489e-01 5.53052604e-01 -1.77356347e-01 3.65104258e-01
4.27255988e-01 3.50155234e-01 1.21667755e+00 -2.44757459e-01
8.24734271e-01 1.21918464e+00 6.24981046e-01 7.22783566e-01
-1.61812520e+00 -5.14524043e-01 5.63917994e-01 3.57939154e-01
-1.00308275e+00 -4.54102099e-01 6.01548851e-01 -1.09442854e+00
1.10002661e+00 1.15623727e-01 9.76024091e-01 1.61911547e+00
6.36037469e-01 9.13134396e-01 9.32714641e-01 -8.50982592e-03
1.52638227e-01 -2.63831913e-01 8.94005038e-03 7.98033476e-01
-1.62547022e-01 2.66982973e-01 -8.89813364e-01 -2.16494754e-01
7.97838151e-01 2.67471611e-01 -2.89867014e-01 -6.35573566e-01
-1.33610141e+00 6.04843676e-01 3.47307250e-02 3.52792770e-01
-4.28425938e-01 3.48289043e-01 4.16502744e-01 5.02063394e-01
4.58347023e-01 3.48137468e-01 -4.99187469e-01 -4.15160835e-01
-1.13169122e+00 3.14820468e-01 8.28624904e-01 6.07161939e-01
6.60363078e-01 1.89332739e-02 -2.96022296e-01 3.13205004e-01
8.04501548e-02 -1.10701352e-01 8.21437061e-01 -1.23273456e+00
2.28328630e-01 3.56530279e-01 3.06981921e-01 -1.14028633e+00
-1.55590966e-01 -3.57713938e-01 -9.35262799e-01 -9.90526825e-02
3.04677457e-01 1.18844144e-01 -9.06276524e-01 2.15966558e+00
-1.82754785e-01 7.80991614e-01 -5.86587377e-02 4.68384057e-01
1.69525102e-01 1.25618601e+00 2.29522407e-01 -6.29040360e-01
8.21586430e-01 -9.25028741e-01 -5.48028469e-01 7.23927543e-02
3.53502452e-01 -2.93739706e-01 6.92752063e-01 5.02331555e-01
-1.23261690e+00 -7.06525505e-01 -7.17458308e-01 1.55906916e-01
-6.01726323e-02 -3.77028048e-01 7.43617952e-01 1.20822094e-01
-1.35310292e+00 1.02511621e+00 -1.37006092e+00 -4.16850120e-01
2.41642073e-01 2.75477409e-01 -4.93757427e-01 3.67317200e-02
-1.21217978e+00 6.54034913e-01 3.76745850e-01 -1.25438586e-01
-1.40414429e+00 -7.37352490e-01 -8.08021843e-01 1.96123004e-01
8.42104033e-02 -8.82484496e-01 9.92963314e-01 -1.12246072e+00
-1.42323017e+00 6.00190043e-01 -4.93525118e-01 -8.70444000e-01
5.22735178e-01 -2.48617694e-01 -4.25593019e-01 8.02341551e-02
-3.55782062e-02 5.97963452e-01 1.02475786e+00 -1.11920667e+00
-7.17893183e-01 -1.92233548e-01 -2.64043976e-02 2.70415366e-01
-3.12464148e-01 -1.21278167e-01 -4.62598234e-01 -6.97867751e-01
3.31805423e-02 -1.05352020e+00 -3.78689885e-01 -3.25601697e-01
-1.80374179e-02 -2.73332417e-01 7.94093788e-01 -6.52630389e-01
1.59654403e+00 -2.19410443e+00 8.38035166e-01 -1.79683760e-01
4.75839287e-01 -2.62217700e-01 -5.27600162e-02 6.48661017e-01
-2.23515853e-01 1.78511500e-01 1.18275344e-01 -4.49699968e-01
-5.43915555e-02 5.20661294e-01 -7.42775917e-01 4.90666449e-01
7.12174773e-02 8.29900205e-01 -1.09973145e+00 -2.91554570e-01
5.33508435e-02 3.56081814e-01 -5.99326968e-01 3.95846665e-01
-5.12343705e-01 7.09859729e-01 -1.21902443e-01 2.26802081e-01
6.10173121e-03 -5.27414024e-01 3.82884026e-01 5.63675165e-02
-1.75196320e-01 3.67999494e-01 -8.01238060e-01 1.55926561e+00
-2.42716804e-01 9.38278556e-01 -3.97291392e-01 -1.12891603e+00
5.22380114e-01 5.86991251e-01 9.22146678e-01 -3.20174813e-01
-3.94147128e-01 -2.26864949e-01 -1.17400616e-01 -7.02184081e-01
4.92076606e-01 -2.92874753e-01 -1.20959051e-01 4.24676061e-01
3.32033217e-01 3.29064786e-01 3.16582650e-01 3.67916375e-01
1.26918876e+00 3.94195408e-01 9.69151929e-02 -5.95295355e-02
-3.82921211e-02 -1.63810298e-01 8.04795504e-01 8.19851100e-01
-1.87085137e-01 3.27139676e-01 9.06525433e-01 -7.36447036e-01
-1.19814193e+00 -1.37136805e+00 1.86948523e-01 1.17946744e+00
8.64414424e-02 -8.72781456e-01 -3.22251797e-01 -3.57977688e-01
-3.10461912e-02 3.98563623e-01 -1.04374909e+00 -4.34764534e-01
-7.08569586e-01 -6.16520524e-01 1.58564374e-01 3.60944360e-01
-3.48569602e-02 -9.54171956e-01 -6.36087775e-01 3.32057446e-01
-2.15203911e-01 -9.75191653e-01 -2.98998356e-01 1.65636644e-01
-1.10569215e+00 -7.04578042e-01 -4.57293868e-01 -5.47882438e-01
8.85469466e-02 4.90066735e-03 1.55267239e+00 -3.08711857e-01
-2.69026637e-01 6.08263612e-01 -3.76239300e-01 2.64218628e-01
-6.18753254e-01 -1.19615115e-01 2.74670422e-01 3.14856440e-01
5.94463758e-02 -1.07029009e+00 -6.91992283e-01 1.72919363e-01
-8.73996615e-01 3.86877984e-01 3.20903033e-01 7.57131755e-01
5.11704981e-01 3.18834662e-01 2.63080932e-02 -7.28131473e-01
2.75052428e-01 -9.92500424e-01 -3.26122582e-01 3.21742028e-01
-2.76015282e-01 2.29005069e-01 7.29490459e-01 -6.64498031e-01
-9.69274879e-01 1.78958640e-01 4.32443231e-01 -8.22502792e-01
-3.56126428e-01 5.06307065e-01 1.82390139e-01 5.36185086e-01
4.27393585e-01 6.09623194e-01 -1.28795400e-01 -3.75832975e-01
4.75188404e-01 3.47862160e-03 4.26453590e-01 -5.70278764e-01
6.01770997e-01 6.65211558e-01 2.60750763e-02 -7.56286204e-01
-9.98496950e-01 -3.39429796e-01 -8.82994831e-01 -4.17854816e-01
9.61281538e-01 -1.17446220e+00 -3.22110593e-01 3.40763658e-01
-9.51312244e-01 -6.93460166e-01 -5.39145708e-01 2.98517734e-01
-8.25397968e-01 2.87071258e-01 -1.03156710e+00 -7.00968742e-01
4.74093825e-01 -9.11455393e-01 8.98090422e-01 -8.89634490e-02
-5.46478152e-01 -1.20771229e+00 3.94370914e-01 1.08879268e-01
1.60961643e-01 4.83462930e-01 1.08713400e+00 -7.68397599e-02
-9.27580237e-01 -8.28279480e-02 3.59334677e-01 2.03198642e-01
7.54225701e-02 4.09050137e-01 -5.91822505e-01 -3.57778877e-01
2.46087890e-02 -3.00383359e-01 1.25390816e+00 5.37544906e-01
1.15250587e+00 -5.18249393e-01 -3.87041569e-01 6.50127232e-01
1.21193969e+00 1.20819911e-01 5.15413404e-01 5.41199259e-02
7.12818682e-01 8.81275594e-01 2.36184224e-01 5.93421221e-01
4.87976044e-01 7.11292207e-01 4.30172265e-01 3.94280225e-01
-9.65346843e-02 -4.82661903e-01 8.12949955e-01 1.01239586e+00
-3.96052867e-01 -2.42432088e-01 -9.05340075e-01 6.03845000e-01
-2.26431489e+00 -1.64750886e+00 3.75671983e-02 1.98363733e+00
8.31337154e-01 1.64602667e-01 4.73672718e-01 -1.65657312e-01
4.67918664e-01 4.71184522e-01 -4.86506552e-01 -1.85720831e-01
-1.28746971e-01 -1.04246281e-01 1.17700212e-01 4.85540122e-01
-1.21171057e+00 9.59436893e-01 7.12765408e+00 5.31140208e-01
-1.21897411e+00 -8.10922589e-04 7.36857772e-01 -3.23268265e-01
-3.86178136e-01 6.61869869e-02 -5.71731448e-01 6.69559598e-01
1.18876672e+00 -9.76579860e-02 4.45646077e-01 6.14491880e-01
3.31724286e-01 2.31439725e-01 -1.59197664e+00 8.35516810e-01
-1.39894485e-01 -1.64076102e+00 5.32748520e-01 8.25202987e-02
9.66950953e-01 9.22517553e-02 4.45852906e-01 4.23411846e-01
5.28099060e-01 -1.29767549e+00 1.02135372e+00 9.62165833e-01
3.77835006e-01 -4.33826178e-01 1.56809524e-01 6.14542782e-01
-1.36670613e+00 -6.02844179e-01 -2.51068741e-01 -4.99162316e-01
1.96567401e-01 2.87583828e-01 -2.16898069e-01 2.35888258e-01
7.96106696e-01 1.25702286e+00 -5.60937285e-01 7.62630105e-01
5.26971407e-02 8.94207776e-01 1.30904004e-01 2.90688485e-01
3.24709356e-01 -1.99522078e-01 5.85384011e-01 1.28609288e+00
2.72798479e-01 -2.13277563e-02 4.43190306e-01 6.44480824e-01
1.92242518e-01 -4.86186296e-01 -7.27830112e-01 -2.05794722e-01
1.12215973e-01 6.88244402e-01 -5.13530672e-01 -5.08921087e-01
-3.42522740e-01 1.18390751e+00 5.14884949e-01 5.22249341e-01
-1.00756025e+00 5.11003435e-01 9.06103969e-01 3.41502219e-01
4.18886244e-01 -5.70340395e-01 1.61845341e-01 -1.65233469e+00
-1.91627949e-01 -6.83087826e-01 5.52519381e-01 -7.10109115e-01
-1.37814486e+00 7.32198596e-01 1.02498598e-01 -1.31252432e+00
-5.44604778e-01 -4.10572648e-01 -4.27148551e-01 5.75316250e-01
-1.13693476e+00 -1.12459815e+00 -1.33293355e-03 7.18707621e-01
7.72013962e-01 -1.08750343e-01 8.58536184e-01 -3.48659456e-02
-5.52805543e-01 1.30614951e-01 3.43129158e-01 -9.34768617e-02
3.29352617e-01 -1.26786721e+00 3.37435812e-01 7.96767354e-01
4.59962636e-01 6.23092771e-01 9.97324049e-01 -5.83634794e-01
-1.14454627e+00 -1.15074492e+00 9.01559174e-01 -7.93998957e-01
1.20009136e+00 -5.02276301e-01 -8.92886758e-01 1.05201936e+00
1.13889404e-01 7.04761893e-02 8.92773628e-01 2.21621960e-01
-6.56343341e-01 4.29836512e-02 -3.20235431e-01 7.56235898e-01
1.19756711e+00 -8.64409864e-01 -5.43359399e-01 3.37177664e-01
8.51062834e-01 -1.28410280e-01 -8.83305311e-01 3.28977346e-01
8.84801626e-01 -1.42773032e+00 8.34386885e-01 -1.09560239e+00
8.00587952e-01 3.53252627e-02 -2.48754591e-01 -1.09257531e+00
-8.33107769e-01 -9.21170712e-01 -8.49778354e-01 8.42744946e-01
2.68538654e-01 -7.82759711e-02 1.02895927e+00 5.47086418e-01
7.28857657e-03 -9.77661908e-01 -7.73250103e-01 -9.55184519e-01
8.62796158e-02 -3.85842949e-01 2.25013092e-01 8.69544566e-01
-2.29155067e-02 2.53076017e-01 -8.01274419e-01 2.07945988e-01
4.69500422e-01 4.67919946e-01 4.79562163e-01 -1.22155297e+00
-7.25797892e-01 -5.13894677e-01 -6.52501643e-01 -1.33914995e+00
3.04656863e-01 -5.59043705e-01 -3.95290554e-02 -1.24853623e+00
4.52498138e-01 -9.73683596e-02 -5.80006897e-01 6.85445592e-02
8.12062696e-02 1.05679771e-02 3.76184195e-01 6.74853623e-01
-1.04512417e+00 5.14335632e-01 9.65419531e-01 -4.97300029e-02
-1.86801150e-01 3.26512232e-02 -2.67551631e-01 8.39010298e-01
5.83871901e-01 -4.17627186e-01 -5.79422176e-01 -3.23425233e-01
4.01533186e-01 4.33257639e-01 3.56574148e-01 -1.05758607e+00
6.32084236e-02 -2.89069086e-01 2.71627516e-01 -3.35127324e-01
5.24492919e-01 -6.94380820e-01 5.19233286e-01 5.65398633e-01
-6.70462191e-01 7.19078183e-02 -1.61992937e-01 1.02540016e+00
-3.19644123e-01 3.25503230e-01 5.16687691e-01 -4.23527062e-01
-8.27170789e-01 6.86149657e-01 -8.24408293e-01 -1.42144620e-01
1.06632411e+00 -3.23834330e-01 1.15598366e-01 -7.57229984e-01
-1.33231485e+00 1.16449110e-01 5.27563453e-01 5.70539653e-01
3.93428355e-01 -1.25861931e+00 -5.23366988e-01 -1.39994528e-02
-7.08288029e-02 -4.08157408e-01 6.18378997e-01 6.67960405e-01
-2.54499555e-01 3.56222719e-01 -3.25781673e-01 -8.37458611e-01
-1.18141973e+00 7.57814527e-01 2.99721450e-01 -7.34634697e-01
-5.97904205e-01 1.04122174e+00 3.78497124e-01 1.58549592e-01
3.80628735e-01 -3.62577468e-01 -3.64160329e-01 2.86003202e-01
2.41945371e-01 1.53318301e-01 -6.72900975e-01 -8.40079844e-01
1.21037355e-02 4.51117665e-01 -1.72245130e-02 3.39226723e-02
1.44257605e+00 -2.70419061e-01 -2.56352335e-01 1.13859332e+00
1.22251475e+00 -3.69807094e-01 -1.93747795e+00 -3.27799842e-02
2.15292573e-01 -3.61118942e-01 -3.29806328e-01 -4.37625349e-01
-6.72145844e-01 8.83538127e-01 3.72422874e-01 7.33669043e-01
8.65309775e-01 3.11595052e-01 7.29130149e-01 1.06350645e-01
3.49398345e-01 -6.85021222e-01 6.00021422e-01 6.49075031e-01
7.85459340e-01 -9.56466913e-01 -1.77265719e-01 -2.02851407e-02
-6.39761746e-01 1.10079956e+00 3.87423873e-01 -3.21265221e-01
7.61030972e-01 5.84317110e-02 -5.33269286e-01 -1.59145802e-01
-1.53744531e+00 8.10366049e-02 3.62953991e-01 5.20351291e-01
4.60381657e-01 1.22747578e-01 4.20148432e-01 4.68573421e-01
-1.81750372e-01 -9.87115800e-02 5.01063526e-01 6.85797095e-01
-3.27994078e-01 -9.12795961e-01 2.90642865e-03 4.67663348e-01
-4.06623125e-01 8.33103806e-02 -3.86080556e-02 5.64137161e-01
3.61794502e-01 5.53472757e-01 3.89169246e-01 -5.53446054e-01
-7.03270286e-02 1.52615234e-01 6.08776271e-01 -7.27615237e-01
-1.86339736e-01 9.26437229e-02 -1.32823065e-01 -9.63777900e-01
-5.33556044e-01 -1.07522798e+00 -6.28808260e-01 -2.27120861e-01
2.58109927e-01 1.48559988e-01 5.46152964e-02 8.65118504e-01
4.42241967e-01 5.38672090e-01 6.02443755e-01 -1.12831938e+00
-4.01883364e-01 -6.43719554e-01 -5.62775493e-01 6.64748728e-01
7.20388770e-01 -5.58472991e-01 -3.98225516e-01 7.37199366e-01] | [8.553470611572266, 0.6566185355186462] |
a839368f-64f6-4f63-8d2c-d1d5e275dd9d | a-novel-stereo-matching-pipeline-with | 2204.04865 | null | https://arxiv.org/abs/2204.04865v2 | https://arxiv.org/pdf/2204.04865v2.pdf | A novel stereo matching pipeline with robustness and unfixed disparity search range | Stereo matching is an essential basis for various applications, but most stereo matching methods have poor generalization performance and require a fixed disparity search range. Moreover, current stereo matching methods focus on the scenes that only have positive disparities, but ignore the scenes that contain both positive and negative disparities, such as 3D movies. In this paper, we present a new stereo matching pipeline that first computes semi-dense disparity maps based on binocular disparity, and then completes the rest depending on monocular cues. The new stereo matching pipeline have the following advantages: It 1) has better generalization performance than most of the current stereo matching methods; 2) relaxes the limitation of a fixed disparity search range; 3) can handle the scenes that involve both positive and negative disparities, which has more potential applications, such as view synthesis in 3D multimedia and VR/AR. Experimental results demonstrate the effectiveness of our new stereo matching pipeline. | ['Feng Liu', 'Jiazhi Liu'] | 2022-04-11 | null | null | null | null | ['stereo-matching-1'] | ['computer-vision'] | [ 1.55564278e-01 -6.07568443e-01 -6.88736662e-02 -3.06660026e-01
-1.33253813e-01 -4.11996722e-01 5.74991167e-01 -3.86635274e-01
-1.34553894e-01 4.95417207e-01 2.59037077e-01 -3.66543353e-01
2.70896733e-01 -9.50287700e-01 -3.76522660e-01 -4.89782214e-01
4.19089586e-01 3.60571891e-02 9.05270278e-01 -3.37451220e-01
5.07246315e-01 4.63781089e-01 -2.05336714e+00 2.43925244e-01
9.61998165e-01 1.08312488e+00 6.73644602e-01 3.15028250e-01
-1.99457645e-01 4.04379547e-01 -2.58692265e-01 5.97949177e-02
6.26956046e-01 -1.96593121e-01 -4.60900933e-01 1.54800296e-01
7.95191109e-01 -7.28336036e-01 -5.86106837e-01 1.22904646e+00
6.78456187e-01 -2.23516133e-02 1.35308638e-01 -1.14624667e+00
-2.55741298e-01 -3.47470909e-01 -8.29268575e-01 2.76960552e-01
9.38091636e-01 7.97729045e-02 5.67250788e-01 -8.63777995e-01
6.59291089e-01 1.35974038e+00 4.53742981e-01 5.11288226e-01
-8.82480383e-01 -1.07100773e+00 1.65687636e-01 2.87830055e-01
-1.36526108e+00 -2.23204494e-01 7.16957986e-01 -5.27942181e-01
8.50553632e-01 1.92232236e-01 9.42020357e-01 5.90924263e-01
1.81689456e-01 6.87860727e-01 1.17028630e+00 -6.61215410e-02
5.03876135e-02 -2.42445886e-01 -1.64431632e-01 3.94343793e-01
-4.46942598e-02 5.34065545e-01 -3.39867800e-01 2.98423823e-02
1.32737494e+00 1.43642202e-01 -5.21148086e-01 -6.43100560e-01
-1.47441959e+00 6.83020771e-01 3.56752157e-01 1.03090540e-01
7.71429390e-02 -2.76708245e-01 2.65042305e-01 3.01889390e-01
2.33206108e-01 4.40592542e-02 -1.00234039e-01 3.92904170e-02
-7.00341105e-01 4.18529719e-01 2.41704330e-01 1.40661371e+00
1.19571698e+00 -7.56653473e-02 1.16647951e-01 9.80935931e-01
1.25724688e-01 5.39754450e-01 3.53851408e-01 -1.20719361e+00
8.24663758e-01 7.18098760e-01 1.89523045e-02 -1.05469513e+00
-4.14892852e-01 9.10244957e-02 -9.94186223e-01 4.48312521e-01
-3.40957083e-02 1.37650982e-01 -8.55415761e-01 1.38532448e+00
4.41405505e-01 2.26225406e-01 -5.16357161e-02 1.13696873e+00
1.37471843e+00 6.84127212e-01 -5.26218057e-01 -7.51855746e-02
1.01510799e+00 -7.97413766e-01 -7.00311005e-01 -2.81839401e-01
2.21689582e-01 -1.14073503e+00 1.00040436e+00 1.47389889e-01
-1.19560027e+00 -8.78658414e-01 -1.19338274e+00 -4.78303105e-01
-1.21627867e-01 -4.75671291e-01 8.24695826e-01 5.50232589e-01
-1.30246878e+00 1.62414566e-01 -3.75561059e-01 -3.50682944e-01
-2.45764330e-02 5.01005888e-01 -3.65347385e-01 -4.01242346e-01
-1.21341860e+00 6.34092748e-01 3.25314492e-01 -7.71547779e-02
-2.94858575e-01 -6.10243678e-01 -1.33952045e+00 -1.75256267e-01
7.34145194e-02 -1.05754793e+00 9.88561451e-01 -6.38740063e-01
-1.28401768e+00 1.14762795e+00 -6.16653621e-01 2.78109133e-01
5.87572217e-01 -7.49363080e-02 -3.61626685e-01 5.95374741e-02
2.88437188e-01 1.10386705e+00 3.00989419e-01 -1.10909855e+00
-1.11546266e+00 -3.99675220e-01 4.33028847e-01 6.66878998e-01
3.12001314e-02 -1.44956782e-01 -9.21792150e-01 -3.92397076e-01
8.05122137e-01 -6.67542398e-01 -1.62043437e-01 2.58525521e-01
-2.74309397e-01 4.95160185e-02 9.31291163e-01 -1.04357287e-01
1.09576952e+00 -2.27402043e+00 -3.42014767e-02 1.40380457e-01
1.99343905e-01 1.16306350e-01 1.61139995e-01 3.17595690e-01
-6.16445020e-02 -3.05272490e-01 8.73960108e-02 -1.72854543e-01
-3.96435916e-01 8.95327255e-02 -2.83193260e-01 2.45296583e-01
-2.39768222e-01 4.60577071e-01 -9.08980548e-01 -7.09902763e-01
8.85800540e-01 2.63941407e-01 -7.70028055e-01 2.06362784e-01
1.97845921e-01 7.07080722e-01 -3.03962737e-01 6.48779929e-01
1.28056788e+00 -2.77213342e-02 -3.00030679e-01 -3.14422071e-01
-6.24326527e-01 2.55480409e-01 -1.61316383e+00 2.02248263e+00
-1.77243412e-01 6.74872518e-01 3.66224982e-02 -2.99444526e-01
1.03998494e+00 1.00957014e-01 5.58162451e-01 -9.98904526e-01
-2.80971900e-02 3.22111607e-01 -2.96278954e-01 -3.87661815e-01
6.61434054e-01 -5.36805056e-02 2.91411817e-01 5.27028553e-02
-3.36295903e-01 -5.62799990e-01 3.05039227e-01 -9.29227173e-02
6.57842517e-01 1.25316471e-01 1.12095274e-01 -3.49259019e-01
8.71293306e-01 -9.62162241e-02 1.07169330e+00 3.78110200e-01
-1.30454287e-01 1.13769996e+00 8.71748030e-02 -4.32664633e-01
-1.00858665e+00 -1.26264834e+00 -3.68371785e-01 3.47030729e-01
1.26580310e+00 -2.29399845e-01 -2.04750985e-01 -4.66913693e-02
-1.24718659e-02 4.37458120e-02 -9.34208483e-02 5.22363856e-02
-5.85200846e-01 -1.77957997e-01 1.24959536e-02 5.44688523e-01
1.19359660e+00 -6.51282191e-01 -3.76294643e-01 -1.13203764e-01
-4.65034515e-01 -1.24720848e+00 -7.42334247e-01 -4.37385321e-01
-1.16182983e+00 -1.13880205e+00 -7.83306658e-01 -1.23440254e+00
6.24061644e-01 1.13897753e+00 9.17805493e-01 3.19380723e-02
-1.61129788e-01 1.39667511e-01 -4.50043380e-03 -1.88664109e-01
1.39913127e-01 -2.48493344e-01 -4.62289862e-02 -2.97225714e-01
4.74842906e-01 -6.64255559e-01 -9.86089826e-01 9.23554838e-01
-8.35898519e-01 5.75015247e-01 1.45870820e-01 7.56968379e-01
7.05795527e-01 3.81520167e-02 1.44964000e-02 -6.23078883e-01
2.10481867e-01 -3.73277143e-02 -8.88778269e-01 -2.06390291e-01
-3.30482811e-01 -3.83378655e-01 3.49012256e-01 -2.44473219e-01
-1.19367325e+00 3.81708145e-04 -3.15909684e-01 -4.76631492e-01
-1.05162024e-01 -1.03201941e-02 -4.41637218e-01 -1.45391464e-01
2.37197042e-01 1.92233697e-01 -2.16756258e-02 -4.03200835e-01
-2.15440303e-01 6.72937691e-01 6.26670182e-01 -3.26637685e-01
8.49393666e-01 9.39246118e-01 1.49501070e-01 -6.55676544e-01
-2.32883364e-01 -9.25274968e-01 -6.49146557e-01 -2.74079204e-01
8.11196625e-01 -1.27022624e+00 -9.26970661e-01 6.96713388e-01
-1.14756227e+00 8.13268945e-02 1.37855873e-01 9.66831923e-01
-6.63055003e-01 7.36889660e-01 -7.34665334e-01 -2.66706437e-01
-2.57711858e-02 -1.44324625e+00 1.25299025e+00 4.99215335e-01
-3.04914098e-02 -9.83262122e-01 -1.54784946e-02 2.16746777e-01
3.18279326e-01 -2.93029696e-02 8.75155628e-01 3.66937071e-01
-1.00554073e+00 1.23394124e-01 -4.36728001e-01 -4.77621593e-02
2.65205055e-01 -8.20556059e-02 -1.05682778e+00 -2.79806942e-01
-1.20165564e-01 4.79935948e-03 5.78935981e-01 5.32576084e-01
1.04916394e+00 5.51078737e-01 -3.10927391e-01 1.21939838e+00
1.51901412e+00 4.47058201e-01 1.02785909e+00 5.37869573e-01
5.90188742e-01 8.44825804e-01 9.35953498e-01 2.47864112e-01
5.71387708e-01 8.23124290e-01 5.97809196e-01 -5.90523660e-01
-1.57396644e-01 -4.16569054e-01 8.56409669e-02 9.23261344e-01
-5.53471893e-02 1.91521764e-01 -5.90466440e-01 3.83461922e-01
-1.82357883e+00 -1.03252423e+00 -3.39530617e-01 2.39533019e+00
5.76286972e-01 6.39340430e-02 -9.65121239e-02 2.09546417e-01
1.14338863e+00 3.53814930e-01 -5.48569798e-01 -3.34864110e-01
-3.69170457e-01 7.21690208e-02 3.83890688e-01 6.60531938e-01
-9.65908706e-01 7.38786697e-01 6.88669205e+00 5.88816464e-01
-1.14685440e+00 -3.68583262e-01 1.64604455e-01 6.47836700e-02
-6.10468447e-01 6.90704584e-02 -9.30208385e-01 4.31377292e-01
-2.60685921e-01 -1.25936136e-01 1.38733625e-01 7.61237681e-01
2.61088789e-01 -4.84145373e-01 -1.25297511e+00 1.71364987e+00
7.24529754e-03 -1.27165663e+00 1.79406613e-01 2.08198354e-01
9.10845101e-01 -8.55149031e-02 2.66895052e-02 4.43506762e-02
-4.44468819e-02 -7.88748503e-01 5.16236305e-01 -3.15297917e-02
9.77459311e-01 -6.23278439e-01 6.95789218e-01 5.03942311e-01
-1.39550507e+00 2.05401957e-01 -7.55331874e-01 -3.47148895e-01
2.16154307e-01 6.09494030e-01 -1.91670090e-01 5.63965082e-01
8.62188399e-01 1.13498819e+00 -3.18991601e-01 1.39489222e+00
-6.14885800e-02 -4.59764034e-01 -1.96536139e-01 3.53162169e-01
-6.39017578e-03 -6.06232703e-01 4.17110890e-01 8.73489320e-01
6.09316349e-01 7.12719560e-02 1.70595005e-01 6.30756676e-01
3.95922154e-01 5.24412394e-02 -1.02948809e+00 8.90591145e-01
7.01484501e-01 6.44513786e-01 -4.04668123e-01 -3.02048534e-01
-8.76258016e-01 1.04172158e+00 -2.53352791e-01 3.90738428e-01
-6.40152335e-01 -3.81591469e-01 9.36792076e-01 1.69791132e-01
1.65744126e-01 -2.28637710e-01 -5.21714866e-01 -1.43802416e+00
2.80211031e-01 -6.19904339e-01 3.44206631e-01 -1.14415908e+00
-9.16385889e-01 4.66471523e-01 7.43865892e-02 -2.00057387e+00
-3.18459600e-01 -4.67914343e-01 -5.78444898e-01 1.15016913e+00
-1.89090574e+00 -8.04845929e-01 -8.10324252e-01 1.03573394e+00
6.33778512e-01 8.54550377e-02 5.08554876e-01 7.04910934e-01
-2.30886683e-01 2.19424367e-01 -1.28889829e-01 -1.55708402e-01
8.43815386e-01 -9.59609807e-01 3.53548735e-01 7.98057795e-01
-6.11104548e-01 5.77189922e-01 4.34607863e-01 -4.85598803e-01
-1.18636632e+00 -6.02712929e-01 1.00534093e+00 -9.38737765e-04
4.67115343e-02 -2.13262036e-01 -5.93268454e-01 4.94534999e-01
1.82022136e-02 -1.07991226e-01 4.55765456e-01 -4.78817187e-02
-2.49744371e-01 -3.98481846e-01 -1.12197077e+00 8.18415523e-01
1.56324935e+00 -5.99608898e-01 -4.95903224e-01 2.96444687e-05
4.89295244e-01 -9.52027202e-01 -6.12661541e-01 8.29445481e-01
6.22230053e-01 -1.72211158e+00 1.22561753e+00 8.08233619e-02
4.91446435e-01 -6.91463113e-01 -2.96025664e-01 -9.44888949e-01
-5.25528729e-01 -3.37709725e-01 4.29741085e-01 1.04446995e+00
-8.47243890e-02 -7.35243857e-01 8.10146689e-01 5.19369960e-01
-4.15765822e-01 -5.50848901e-01 -9.44976985e-01 -6.85749292e-01
-3.25235128e-01 -3.34984452e-01 8.42852235e-01 8.93036902e-01
2.07222532e-02 1.53992161e-01 -3.08063924e-01 2.48216584e-01
4.83651310e-01 7.10936666e-01 1.08087003e+00 -1.30529010e+00
-1.26000568e-01 -4.89966333e-01 -8.74738932e-01 -2.02368760e+00
-1.92133397e-01 -4.71477121e-01 -1.18889190e-01 -1.58677459e+00
3.04682314e-01 -5.31097889e-01 2.34663695e-01 -1.76173538e-01
-1.76092342e-01 2.35766858e-01 -1.23937331e-01 3.19817603e-01
-1.52391732e-01 4.32403475e-01 1.66451287e+00 1.26996497e-02
-2.30919927e-01 4.05923054e-02 -3.71952593e-01 8.05771053e-01
4.89577144e-01 2.02864096e-01 -6.36793733e-01 -6.87942743e-01
1.66829139e-01 2.47392803e-01 1.96341008e-01 -9.57591295e-01
3.35121363e-01 -4.64560360e-01 4.11194295e-01 -1.20389545e+00
6.06155336e-01 -8.81042659e-01 3.00320506e-01 4.30877358e-01
2.56344140e-01 4.23420489e-01 1.05405807e-01 1.99743956e-01
-6.92440450e-01 6.92423433e-02 8.63447428e-01 -1.33756384e-01
-1.23431838e+00 5.48436522e-01 -1.62036240e-01 -5.74843399e-02
1.09849107e+00 -1.12113905e+00 -2.65500695e-01 -4.72497165e-01
-2.76605099e-01 5.09404182e-01 1.04043996e+00 5.86375237e-01
8.79947186e-01 -1.73156989e+00 -4.18309420e-01 6.84679270e-01
3.35267216e-01 3.96243095e-01 4.06224519e-01 5.54404259e-01
-9.26840901e-01 5.68966568e-01 -4.90148067e-01 -1.13981164e+00
-1.42769945e+00 5.33002913e-01 3.90837371e-01 2.31216893e-01
-7.01739788e-01 5.93784213e-01 1.09074557e+00 -4.51116085e-01
3.05227578e-01 -2.25664303e-01 -3.14788997e-01 -2.97645688e-01
4.75389004e-01 4.25080180e-01 8.64081364e-03 -6.86898589e-01
-2.69128352e-01 1.21902037e+00 1.30003855e-01 -5.82452603e-02
8.49529028e-01 -5.60130000e-01 -5.99168427e-02 2.95810521e-01
1.22746515e+00 7.89071098e-02 -1.17558587e+00 -2.80950874e-01
-6.34471416e-01 -1.41717124e+00 -1.05854496e-01 -7.56407296e-03
-1.09979320e+00 1.15797126e+00 5.17439127e-01 -1.95269808e-01
1.36421919e+00 -2.58800209e-01 1.00410461e+00 -1.05977103e-01
7.10195661e-01 -1.05497730e+00 -1.70304000e-01 6.80096865e-01
6.01467013e-01 -1.24756229e+00 1.28443949e-02 -1.11507523e+00
-2.95534104e-01 1.15569866e+00 9.92047489e-01 6.86618537e-02
6.63729131e-01 1.56124949e-01 1.96184307e-01 -1.27513736e-01
-4.49798495e-01 -4.08979297e-01 1.92776501e-01 8.90966475e-01
5.73110461e-01 -4.05508608e-01 -4.15047586e-01 -2.58834630e-01
-1.31227911e-01 3.59639190e-02 3.17793220e-01 8.44031036e-01
-4.68556494e-01 -1.17011666e+00 -4.18324411e-01 1.03809252e-01
8.79046693e-03 -1.48674533e-01 -1.08078472e-01 7.86883414e-01
3.75228763e-01 1.00944817e+00 3.26947987e-01 -5.43533802e-01
6.94164217e-01 -5.96784294e-01 5.36602497e-01 -6.76134229e-01
-2.71595538e-01 1.92290246e-01 -1.42425835e-01 -8.02158713e-01
-5.86899459e-01 -4.29176331e-01 -1.13237739e+00 -7.46407390e-01
-2.83518523e-01 -2.19410360e-01 4.33120221e-01 5.84638596e-01
3.60722512e-01 2.11933888e-02 7.51412034e-01 -1.05002451e+00
8.81246552e-02 -5.65584481e-01 -5.70419967e-01 5.47570944e-01
3.22293401e-01 -8.17055523e-01 -3.31682950e-01 -9.70187038e-02] | [9.042820930480957, -2.4330644607543945] |
353b467c-d1cb-4bfc-87b1-4d94c65f5def | crime-prediction-with-graph-neural-networks | 2111.14733 | null | https://arxiv.org/abs/2111.14733v2 | https://arxiv.org/pdf/2111.14733v2.pdf | Crime Prediction with Graph Neural Networks and Multivariate Normal Distributions | Existing approaches to the crime prediction problem are unsuccessful in expressing the details since they assign the probability values to large regions. This paper introduces a new architecture with the graph convolutional networks (GCN) and multivariate Gaussian distributions to perform high-resolution forecasting that applies to any spatiotemporal data. We tackle the sparsity problem in high resolution by leveraging the flexible structure of GCNs and providing a subdivision algorithm. We build our model with Graph Convolutional Gated Recurrent Units (Graph-ConvGRU) to learn spatial, temporal, and categorical relations. In each node of the graph, we learn a multivariate probability distribution from the extracted features of GCNs. We perform experiments on real-life and synthetic datasets, and our model obtains the best validation and the best test score among the baseline models with significant improvements. We show that our model is not only generative but also precise. | ['Suleyman Serdar Kozat', 'Selim Furkan Tekin'] | 2021-11-29 | null | null | null | null | ['crime-prediction'] | ['miscellaneous'] | [-2.05393881e-01 2.43650705e-01 -1.64205730e-01 -3.48955780e-01
-6.25428915e-01 -1.73343852e-01 6.84724391e-01 1.42422058e-02
-7.10644852e-03 7.97690451e-01 6.46601319e-01 -3.53537709e-01
-2.89079808e-02 -1.34845829e+00 -9.09342051e-01 -4.34863478e-01
-5.50184727e-01 4.72413540e-01 3.30285937e-01 -1.33307651e-01
1.90914944e-02 5.62075078e-01 -1.04356039e+00 4.66419309e-01
8.26738775e-01 6.93194568e-01 1.64961040e-01 5.31990051e-01
8.14930871e-02 1.40309680e+00 -4.52437013e-01 -6.33902073e-01
-2.33131453e-01 -1.86335698e-01 -5.83863735e-01 -1.32723138e-01
1.06753483e-01 -2.88525522e-01 -1.04058278e+00 7.56795943e-01
3.63768786e-01 5.10700107e-01 8.84912729e-01 -1.02317703e+00
-1.13466275e+00 8.38118076e-01 -7.20939457e-01 4.03467387e-01
2.38086700e-01 -2.56337017e-01 9.73200798e-01 -6.34155631e-01
6.54123187e-01 1.20913219e+00 8.70978951e-01 2.17691064e-01
-1.21671295e+00 -4.02851820e-01 5.87950289e-01 3.96975100e-01
-1.74020267e+00 -4.04601209e-02 8.15646529e-01 -5.25587916e-01
1.48067129e+00 -1.60538435e-01 5.79232037e-01 1.47563124e+00
2.53483832e-01 6.45440280e-01 5.48348248e-01 4.03079502e-02
2.55272269e-01 -5.88588119e-01 2.50477910e-01 7.79103279e-01
9.12374482e-02 -1.09213805e-02 -4.02573436e-01 -1.47425145e-01
1.08732498e+00 4.36495036e-01 -2.58296225e-02 8.01461563e-02
-5.50690234e-01 1.03700721e+00 9.18906987e-01 3.87573421e-01
-5.45332372e-01 6.14100635e-01 1.38580680e-01 -2.11915568e-01
9.51500475e-01 -1.36033773e-01 6.97144270e-02 -1.23551190e-01
-9.27584350e-01 2.31719792e-01 3.76780182e-01 8.63244712e-01
7.52672434e-01 1.55634373e-01 -4.94410187e-01 7.91707158e-01
-1.43744782e-01 4.43108678e-01 -8.20770115e-02 -6.65074408e-01
7.73771405e-01 6.28050148e-01 -1.33973524e-01 -1.67211139e+00
-5.56775093e-01 -4.26224560e-01 -1.34434688e+00 -3.38105828e-01
-3.42476070e-02 -3.38578790e-01 -1.15538454e+00 1.83565998e+00
-9.87155735e-02 8.52147460e-01 -2.20517173e-01 5.00711560e-01
8.69274378e-01 9.31586146e-01 3.37177306e-01 1.01221614e-01
6.82020843e-01 -6.74537241e-01 -7.03949213e-01 -2.59447008e-01
6.97866261e-01 1.02543555e-01 7.46642351e-01 1.38548408e-02
-7.37941027e-01 -4.28150713e-01 -4.25338000e-01 -1.00026511e-01
-4.33468938e-01 6.50116801e-02 8.31904054e-01 2.61327952e-01
-1.38807535e+00 7.61135876e-01 -1.10571527e+00 -3.55108708e-01
8.07137251e-01 -7.15510771e-02 -4.00900722e-01 -1.64319634e-01
-1.27237236e+00 6.19450271e-01 2.41390094e-01 2.14354798e-01
-7.16641426e-01 -4.31059986e-01 -1.19566882e+00 3.14763844e-01
-7.02624246e-02 -4.61353600e-01 3.95819187e-01 -2.01732680e-01
-8.20587516e-01 4.24845219e-01 -3.26494843e-01 -6.17186248e-01
9.75167677e-02 -6.10089563e-02 -4.17332172e-01 -4.38310727e-02
3.83027822e-01 2.84422725e-01 6.01976275e-01 -1.00322843e+00
-5.07054687e-01 -3.27762187e-01 -1.42553421e-02 -1.21933363e-01
-2.46261060e-01 -4.26089726e-02 -4.72141176e-01 -8.87933969e-01
7.12472200e-02 -6.23171926e-01 -6.05219483e-01 -8.27251375e-01
-6.34830534e-01 -4.28918093e-01 5.83314180e-01 -9.89031613e-01
1.75026512e+00 -2.20276308e+00 1.10085435e-01 2.58640766e-01
3.59046072e-01 -6.35425076e-02 -1.54793099e-01 4.94025648e-01
-1.38149962e-01 1.84458241e-01 -2.61801511e-01 -6.44068420e-01
-5.48484363e-02 4.54525113e-01 -7.20086277e-01 5.12155056e-01
2.37906471e-01 1.39113700e+00 -1.03769529e+00 -3.33142519e-01
3.91697064e-02 8.62301648e-01 -7.18634546e-01 2.61021852e-01
-7.78453648e-02 1.86832562e-01 -6.16357803e-01 3.96212250e-01
5.92321336e-01 -6.01882219e-01 3.20343673e-01 9.98405889e-02
3.38205516e-01 2.23089173e-01 -9.09328699e-01 1.51801991e+00
-2.17517897e-01 5.51948488e-01 -4.00449783e-01 -1.21383238e+00
1.06147718e+00 1.93801612e-01 2.71713108e-01 -7.65408993e-01
-2.34066367e-01 -3.08969378e-01 -6.62089646e-01 -2.72673815e-01
7.01222777e-01 2.62137771e-01 -3.90994042e-01 4.43802357e-01
2.45152801e-01 3.78937930e-01 -8.56959727e-03 5.79010010e-01
1.46431053e+00 2.14057207e-01 -1.33210763e-01 -6.49833605e-02
-1.59610212e-01 -3.41120541e-01 5.35343409e-01 8.82494271e-01
2.21620157e-01 8.31522763e-01 9.98174906e-01 -7.63094723e-01
-8.27096641e-01 -1.22202480e+00 2.31326744e-01 9.18658078e-01
-2.09695995e-01 -6.97468221e-01 -5.00836968e-01 -6.83003366e-01
-2.14367852e-01 7.97201991e-01 -1.02673614e+00 -1.45594925e-02
-8.80701661e-01 -1.11373329e+00 4.58098650e-01 1.01706505e+00
4.02685165e-01 -1.19911349e+00 -1.79457530e-01 2.41512030e-01
-3.45571309e-01 -1.38280928e+00 -1.34949967e-01 -1.18228272e-01
-5.44689953e-01 -9.00037885e-01 -4.15828317e-01 -6.85290873e-01
5.37459671e-01 -4.99746017e-03 1.23088109e+00 -1.66641995e-02
4.56519499e-02 1.13095708e-01 -4.56773967e-01 8.75460133e-02
1.86270714e-01 2.84431428e-01 -3.53295892e-01 2.17610255e-01
1.26946166e-01 -1.25593877e+00 -3.18558574e-01 -3.93386632e-01
-6.60935223e-01 6.41521811e-02 8.14921334e-02 7.13860333e-01
7.33925164e-01 1.58129081e-01 6.51567757e-01 -1.12734020e+00
8.41324925e-01 -8.78913701e-01 -4.38925177e-01 2.76236922e-01
-1.58252537e-01 -7.35699385e-02 6.47306204e-01 -1.50896892e-01
-8.97891581e-01 -1.87718570e-02 -8.45977664e-02 -5.37617981e-01
-3.28393668e-01 8.20173562e-01 1.63884267e-01 4.11603272e-01
6.77269578e-01 2.02713117e-01 -6.67532742e-01 -4.73120928e-01
6.95526898e-01 1.12655498e-01 6.91727877e-01 -6.62682116e-01
5.15873194e-01 5.10810077e-01 6.81196898e-02 -6.57953501e-01
-9.16410744e-01 -1.08980447e-01 -6.75928771e-01 -1.05125405e-01
1.09809160e+00 -9.84650373e-01 -4.31553602e-01 3.52510571e-01
-1.53151083e+00 -6.65855229e-01 -9.10019577e-02 2.59072036e-01
-4.53574866e-01 2.83633918e-01 -9.79129553e-01 -9.66390371e-01
-2.23793224e-01 -5.85408211e-01 1.30068827e+00 9.71111134e-02
3.56833870e-03 -1.21161389e+00 2.05496252e-01 -2.03222007e-01
4.62768018e-01 9.20471132e-01 1.06837821e+00 -4.94508475e-01
-4.90616828e-01 -1.53106377e-01 -4.85298872e-01 -1.82940155e-01
2.20487695e-02 -1.49638265e-01 -8.49242926e-01 9.82034672e-03
-6.06571019e-01 -1.65872350e-01 1.43866396e+00 6.69156253e-01
1.54657161e+00 -4.05556887e-01 -6.21234000e-01 8.64403546e-01
1.35306358e+00 -2.20207930e-01 8.43930244e-01 1.08965542e-02
1.15134728e+00 3.70810270e-01 5.99592552e-02 5.26870191e-01
8.67590368e-01 5.04489064e-01 5.46116650e-01 -2.20279589e-01
-1.84380949e-01 -7.68068731e-01 3.21441442e-01 5.51801741e-01
-5.70929945e-01 -5.03875136e-01 -1.13573468e+00 9.28810656e-01
-2.39046216e+00 -1.40363979e+00 -1.28594637e-01 1.63502097e+00
2.92694896e-01 1.56338047e-02 1.93914413e-01 -1.79659039e-01
8.90730202e-01 6.29589736e-01 -1.15441300e-01 -4.09717470e-01
-3.58792722e-01 5.85285187e-01 2.99751192e-01 6.48232758e-01
-1.40385807e+00 1.37949359e+00 6.73301888e+00 9.25530791e-01
-8.69642556e-01 6.83018863e-02 9.83578682e-01 3.85061349e-03
-3.89616549e-01 -1.98827073e-01 -6.59840345e-01 3.45047116e-01
1.09736085e+00 1.01335816e-01 5.93679130e-01 7.10705340e-01
1.34347938e-02 2.74202317e-01 -6.05030417e-01 9.53151107e-01
-6.09014323e-03 -1.84162676e+00 2.85723776e-01 -6.70254156e-02
7.78465092e-01 3.30454201e-01 2.96004545e-02 4.81425345e-01
8.47152948e-01 -1.58588421e+00 5.35271406e-01 9.29191053e-01
7.82404184e-01 -1.03938210e+00 5.32949924e-01 4.26944613e-01
-1.41436994e+00 -4.07950673e-03 -6.45797312e-01 -3.64616096e-01
5.21734357e-01 5.68133354e-01 -6.84670627e-01 8.59386683e-01
7.31071055e-01 1.43910623e+00 -6.31931067e-01 6.16967380e-01
-6.13022208e-01 9.87232208e-01 -2.42629915e-01 8.50964561e-02
3.81388634e-01 -2.84274578e-01 1.92757934e-01 1.49681640e+00
4.87040192e-01 2.90042996e-01 3.13215226e-01 1.18204117e+00
-6.20460231e-03 -2.05081671e-01 -1.03009009e+00 1.16004534e-01
4.52271134e-01 9.23068583e-01 -6.69511378e-01 -2.22498953e-01
-2.64092296e-01 8.72335017e-01 1.23261738e+00 8.64324510e-01
-8.94392848e-01 -2.12167159e-01 4.27775085e-01 1.05334714e-01
6.07912183e-01 -3.99695188e-01 -2.90716171e-01 -1.24959898e+00
-4.26453650e-02 -1.02054544e-01 7.05226839e-01 -6.55968845e-01
-1.56581116e+00 1.13970029e+00 1.79872498e-01 -5.55907667e-01
-5.40441394e-01 -2.39314944e-01 -8.48095179e-01 9.23790872e-01
-1.33731854e+00 -1.56278217e+00 2.43322235e-02 8.51516724e-01
-7.12903067e-02 -1.82603732e-01 9.85538423e-01 2.81228721e-01
-7.01550066e-01 3.93620014e-01 -1.64285064e-01 7.15617895e-01
-3.22652496e-02 -1.11296129e+00 9.24952030e-01 1.17601740e+00
4.09111917e-01 3.65484446e-01 3.25428724e-01 -9.19378936e-01
-7.70880044e-01 -1.77413726e+00 1.28623343e+00 -5.70462048e-01
7.38445461e-01 -7.70399332e-01 -9.31112707e-01 1.20296657e+00
-3.34878229e-02 4.51116025e-01 4.16652054e-01 5.73109269e-01
-5.47884643e-01 1.78026065e-01 -9.04114306e-01 4.17569578e-01
1.37142336e+00 -6.93906426e-01 -3.27519804e-01 5.14917970e-01
8.83211553e-01 -3.01683813e-01 -7.60112822e-01 2.96724290e-01
-4.35622223e-02 -8.77667010e-01 9.19727266e-01 -8.27386558e-01
6.00092590e-01 2.56613307e-02 -3.26562524e-01 -1.25601196e+00
-8.94085467e-01 -3.50505382e-01 -3.91440809e-01 1.13808656e+00
5.56641579e-01 -6.30871952e-01 8.05817008e-01 5.37113786e-01
-3.91744301e-02 -1.00034761e+00 -1.13176465e+00 -5.32534838e-01
1.93286955e-01 -8.45361590e-01 9.43057418e-01 1.00661397e+00
1.03540398e-01 4.30637002e-01 -7.60070801e-01 5.70389688e-01
6.42622411e-01 3.80688906e-01 4.70018506e-01 -9.39423144e-01
-2.57027864e-01 -3.69771242e-01 -7.30140567e-01 -9.17836547e-01
5.45661747e-01 -9.88648355e-01 -4.35589731e-01 -1.95639324e+00
2.74254858e-01 -4.23303723e-01 -1.94314912e-01 7.31157005e-01
-3.19953710e-01 2.08684325e-01 -4.04677652e-02 -2.23016024e-01
-9.31125224e-01 6.39706254e-01 8.29515755e-01 -9.27289352e-02
-1.31472051e-01 -1.14345908e-01 -5.12230158e-01 5.33446014e-01
8.24719012e-01 -3.65226746e-01 -4.97406542e-01 -8.64002824e-01
4.46539134e-01 1.90909117e-01 6.35687649e-01 -8.27040315e-01
2.16298565e-01 -1.15216658e-01 5.27065814e-01 -6.95158839e-01
2.77044296e-01 -2.95226067e-01 1.59013301e-01 -1.21374197e-01
-1.70839846e-01 1.86580613e-01 -1.01586161e-02 7.97852099e-01
-2.07940161e-01 5.87355435e-01 2.98627108e-01 2.86240666e-03
-7.58099675e-01 8.69207323e-01 -7.82920867e-02 -1.43834073e-02
7.20989108e-01 2.24742129e-01 -4.67199981e-01 -7.98760235e-01
-1.12842155e+00 2.11484775e-01 8.41688886e-02 3.48813206e-01
8.46757770e-01 -1.56094146e+00 -9.40622926e-01 2.31457904e-01
-1.99167430e-01 1.29695073e-01 4.23009843e-01 3.17928284e-01
-2.42479652e-01 2.60359526e-01 6.79066926e-02 -3.95452619e-01
-3.99354398e-01 7.11309254e-01 2.98960716e-01 -6.65248871e-01
-9.73822057e-01 8.25802088e-01 2.01534092e-01 -3.88243079e-01
5.40407635e-02 -3.29704165e-01 -7.53631949e-01 -2.37409770e-01
4.70083803e-01 2.63197780e-01 -1.91935226e-01 -9.81537640e-01
-3.17262560e-01 3.49600255e-01 3.05747420e-01 1.37953699e-01
2.06674266e+00 1.10910960e-01 -1.94636717e-01 2.56885737e-01
1.05609667e+00 -2.00920194e-01 -1.10112238e+00 -3.97542983e-01
-8.80377591e-02 -2.89876670e-01 -1.13108344e-01 -4.57953781e-01
-1.38494790e+00 9.00140107e-01 -1.69857014e-02 3.48309219e-01
9.99450445e-01 3.62054288e-01 7.36453235e-01 5.38600497e-02
5.37882149e-01 -6.76449895e-01 -2.02628732e-01 8.98622274e-01
9.15790498e-01 -7.61146903e-01 -2.28557348e-01 -3.36978644e-01
-6.79932356e-01 8.70903790e-01 4.80298966e-01 -7.75128186e-01
1.13340878e+00 2.58218735e-01 -4.96647686e-01 -5.66555083e-01
-8.98018420e-01 -3.11218530e-01 2.82834917e-01 9.30701613e-01
3.60077918e-01 4.91967499e-01 -2.49981042e-03 1.15150261e+00
-2.27744475e-01 -1.92298934e-01 2.41785720e-01 3.62635046e-01
-2.79935658e-01 -8.47335458e-01 1.24138311e-01 6.42086804e-01
-3.92750889e-01 -2.46692255e-01 -1.57606192e-02 4.78853256e-01
8.29703826e-03 9.14797068e-01 2.54856884e-01 -7.65982449e-01
3.31733078e-01 -2.92366296e-01 1.66279837e-01 -6.98880553e-01
-2.24791676e-01 -2.91421469e-02 1.64062411e-01 -8.90679836e-01
-2.13394120e-01 -5.33763170e-01 -1.31911170e+00 -4.40128326e-01
1.64419755e-01 2.41515800e-01 -8.98648053e-02 7.82201886e-01
6.46000862e-01 6.49086714e-01 5.62925398e-01 -8.54895115e-01
1.85610712e-01 -1.01993108e+00 -8.85418296e-01 5.02763629e-01
2.87301749e-01 -6.72121108e-01 3.47916186e-02 -2.12076604e-01] | [6.685215473175049, 2.1044960021972656] |
9891d49b-8691-49f6-acf0-0e0b2da6f83e | forecasting-pandemic-tax-revenues-in-a-small | 2112.15431 | null | https://arxiv.org/abs/2112.15431v1 | https://arxiv.org/pdf/2112.15431v1.pdf | Forecasting pandemic tax revenues in a small, open economy | Tax analysis and forecasting of revenues are of paramount importance to ensure fiscal policy's viability and sustainability. However, the measures taken to contain the spread of the recent pandemic pose an unprecedented challenge to established models and approaches. This paper proposes a model to forecast tax revenues in Bulgaria for the fiscal years 2020-2022 built in accordance with the International Monetary Fund's recommendations on a dataset covering the period between 1995 and 2019. The study further discusses the actual trustworthiness of official Bulgarian forecasts, contrasting those figures with the model previously estimated. This study's quantitative results both confirm the pandemic's assumed negative impact on tax revenues and prove that econometrics can be tweaked to produce consistent revenue forecasts even in the relatively-unexplored case of Bulgaria offering new insights to policymakers and advocates. | ['Fabio Ashtar Telarico'] | 2021-12-22 | null | null | null | null | ['econometrics'] | ['miscellaneous'] | [-4.55605596e-01 3.49969923e-01 -7.70559072e-01 4.93388139e-02
-3.07746530e-01 -4.61512625e-01 1.20016658e+00 3.78419280e-01
-5.24954736e-01 1.11821902e+00 7.17438400e-01 -1.06707907e+00
-2.52791405e-01 -8.50633979e-01 8.05051008e-04 -5.95158100e-01
-5.27879968e-02 4.63418096e-01 -2.85256952e-01 -6.18400633e-01
5.08260012e-01 3.49018574e-01 -1.20962346e+00 -3.13441247e-01
9.10168111e-01 6.55488431e-01 1.28648520e-01 1.87388808e-01
-3.96520074e-04 6.57451808e-01 -4.48744178e-01 -3.90276700e-01
4.48383629e-01 1.10195346e-01 -4.72966582e-01 -3.39405596e-01
-1.16084777e-01 -6.36735976e-01 -1.14149243e-01 7.37261236e-01
8.75594318e-02 -1.59612954e-01 6.23836040e-01 -8.91216338e-01
-5.44214010e-01 5.42012274e-01 -2.92435884e-01 5.00196040e-01
2.79894352e-01 2.53764778e-01 7.41681933e-01 -4.81811613e-01
6.13129020e-01 9.09460366e-01 5.14681160e-01 4.70994748e-02
-8.60933244e-01 -6.66979969e-01 8.08177739e-02 1.76232859e-01
-1.16356230e+00 -4.25299376e-01 1.51548848e-01 -8.48041296e-01
9.65636849e-01 3.30062240e-01 9.65893745e-01 4.80186909e-01
5.23814082e-01 -5.01866899e-02 1.07328308e+00 -4.22812402e-01
6.32741451e-02 3.27884674e-01 -2.14213938e-01 1.73909939e-03
8.42642784e-01 3.52699488e-01 6.89788982e-02 -1.42557934e-01
4.34232414e-01 -1.37101188e-01 -1.32308185e-01 4.16771889e-01
-1.08664370e+00 1.07187688e+00 3.89636494e-02 4.45950627e-01
-8.82755756e-01 -9.58354995e-02 6.86055720e-01 8.41844603e-02
8.45185697e-01 -2.61619184e-02 -3.73679399e-01 -2.41512775e-01
-1.30134583e+00 7.44074285e-01 3.81824464e-01 3.57534707e-01
1.60761729e-01 6.21232212e-01 1.64235219e-01 8.96033049e-02
2.31037781e-01 1.08979058e+00 3.15086961e-01 -6.65247202e-01
8.07103992e-01 1.25503883e-01 7.30933189e-01 -1.15466177e+00
-3.39888692e-01 -4.60322857e-01 -4.00419146e-01 7.14539364e-02
7.49558687e-01 -1.68367371e-01 -4.33065414e-01 1.20599413e+00
-1.18208058e-01 -2.65709132e-01 9.59597379e-02 6.91991270e-01
1.11279137e-01 7.01416850e-01 2.51763135e-01 -7.90166974e-01
1.10181510e+00 -1.76099181e-01 -5.70566297e-01 1.43125698e-01
6.73444569e-01 -7.16730893e-01 3.10045809e-01 2.27870867e-01
-1.11539710e+00 -1.66756749e-01 -3.02463055e-01 6.08349025e-01
-2.61803150e-01 -6.62552714e-02 7.38895595e-01 7.22671330e-01
-8.77623439e-01 2.22488821e-01 -7.12742329e-01 -6.94280118e-02
1.96105108e-01 -2.04268526e-02 1.81391522e-01 2.61514157e-01
-1.25008428e+00 1.30056727e+00 4.24051881e-01 -2.82401759e-02
-5.15542865e-01 -5.43765664e-01 -5.29870689e-01 4.12455022e-01
1.39391258e-01 -2.05962330e-01 1.17519891e+00 -6.39762998e-01
-5.77979922e-01 3.12089801e-01 -1.06053138e-02 -9.03273344e-01
5.11121929e-01 3.93366188e-01 -7.03730881e-01 -1.82031810e-01
4.07831162e-01 -2.20764473e-01 -9.56242383e-02 -7.55259871e-01
-9.93471503e-01 -6.89456940e-01 -9.81755778e-02 -9.36353058e-02
-8.03396404e-02 4.43940938e-01 5.95568836e-01 -7.40722299e-01
-2.09920779e-01 -6.91976607e-01 -3.14703077e-01 -1.18732619e+00
5.59716485e-02 -9.50059593e-02 3.14647406e-01 -1.09659600e+00
1.53390193e+00 -1.28051734e+00 -8.92695546e-01 3.36427450e-01
-2.10678235e-01 3.48231524e-01 5.44850707e-01 8.07796180e-01
-1.58738300e-01 7.67492279e-02 1.52752057e-01 1.71165898e-01
2.64426414e-02 8.19999427e-02 -8.43548179e-01 6.01446092e-01
-1.06033623e-01 7.00804532e-01 -9.06686544e-01 8.81487057e-02
3.26959729e-01 1.26752704e-01 -3.29502106e-01 -5.50590575e-01
8.87574181e-02 3.23926270e-01 -4.07876492e-01 6.43418968e-01
9.12427962e-01 2.28764445e-01 6.51317835e-01 2.31520697e-01
-9.46245193e-01 6.40327156e-01 -5.79339564e-01 4.74153608e-01
-3.30835134e-01 6.97264373e-01 -1.67631745e-01 -1.24121833e+00
9.15952027e-01 4.35413986e-01 6.13469005e-01 -1.28217101e+00
1.40801832e-01 4.01566386e-01 6.37844652e-02 -5.83106399e-01
1.06174839e+00 -4.76932585e-01 1.04847113e-02 4.18392897e-01
-5.11210978e-01 2.40260083e-02 4.81236905e-01 3.99640575e-02
2.43841931e-01 -3.11207920e-02 5.26829541e-01 -7.99054384e-01
3.18728775e-01 1.19857959e-01 5.84268093e-01 -3.24288532e-02
3.78917456e-02 -2.89202064e-01 4.18248147e-01 -5.97894430e-01
-1.23805988e+00 -6.91164136e-01 -3.01476955e-01 5.73401749e-01
-6.73038900e-01 -1.45002007e-01 -4.74617422e-01 -3.85282308e-01
2.94316173e-01 1.22292173e+00 -4.65272963e-01 4.67238933e-01
-5.45547307e-01 -1.04356503e+00 3.77973586e-01 1.41041741e-01
3.60894889e-01 -5.95109761e-01 -1.03191173e+00 3.81143868e-01
-2.89454430e-01 -9.66687500e-01 1.34209886e-01 -3.19098294e-01
-7.67177582e-01 -1.20868576e+00 -7.98826277e-01 -3.44038289e-03
5.22082686e-01 3.58439177e-01 8.46403122e-01 -1.77739128e-01
2.07265064e-01 1.92152724e-01 2.41703223e-02 -9.97134924e-01
-4.27342594e-01 -1.95867077e-01 1.69545695e-01 -6.88102961e-01
5.71048379e-01 -3.06832045e-01 -4.93047923e-01 -8.43371749e-02
-6.22888923e-01 -2.76889414e-01 -2.01889411e-01 2.68951654e-01
-6.83561862e-02 3.30844641e-01 1.39390230e+00 -6.56668007e-01
7.53179848e-01 -8.58405948e-01 -1.06047308e+00 2.16362178e-02
-1.31747127e+00 -3.08977842e-01 5.02048612e-01 1.96077660e-01
-1.45756423e+00 -7.14706600e-01 1.21935382e-01 4.67250615e-01
-5.02237454e-02 1.10806930e+00 4.91853803e-01 3.44911754e-01
3.37059259e-01 1.29015163e-01 7.60379136e-02 -4.96424705e-01
1.37017714e-02 5.88208795e-01 8.13081086e-01 -4.73991305e-01
7.20261693e-01 6.28104687e-01 -8.60956386e-02 -5.94688296e-01
-2.78582603e-01 -5.39388537e-01 -2.36851946e-01 -6.61776543e-01
3.17953259e-01 -1.21901858e+00 -6.87094510e-01 1.53521672e-01
-8.60324681e-01 -1.35021001e-01 -9.14896429e-02 1.00303102e+00
-5.65445900e-01 2.25951582e-01 -2.83211917e-01 -1.46176326e+00
-2.79859662e-01 -9.49984074e-01 3.78179252e-02 1.86017975e-01
-2.79583544e-01 -1.51833141e+00 4.25584614e-01 4.17055875e-01
8.03249359e-01 5.68712115e-01 7.90183127e-01 -5.00451803e-01
-2.16911748e-01 -2.67430320e-02 -3.58313620e-01 1.52299076e-01
3.85179073e-01 3.24486017e-01 -6.80393100e-01 -3.27486515e-01
-1.41893979e-02 2.06358626e-01 7.68364429e-01 6.57369256e-01
2.41439819e-01 -1.06227577e+00 -1.99269459e-01 -9.95578170e-02
1.81405497e+00 4.59801316e-01 7.59022892e-01 9.72825468e-01
-2.66279280e-01 5.98747134e-01 1.02792215e+00 9.54782844e-01
9.71021652e-01 4.56347257e-01 6.81025147e-01 1.64252281e-01
3.53626639e-01 -1.21961035e-01 2.18752921e-01 8.04720044e-01
-1.06033385e+00 2.02022925e-01 -1.20309353e+00 9.57291722e-01
-1.49651182e+00 -1.57569087e+00 -2.72186548e-01 2.69066072e+00
5.03526270e-01 3.43336314e-01 5.57599664e-01 2.11693674e-01
2.56057590e-01 1.81208029e-01 4.53120053e-01 -8.74317586e-01
5.89138195e-02 1.97262019e-01 1.19898212e+00 6.37256265e-01
-3.38492274e-01 5.90732157e-01 6.94596434e+00 5.06852090e-01
-1.25497067e+00 -2.31702238e-01 6.91321075e-01 5.26030064e-02
-8.43031824e-01 1.66364655e-01 -8.50809634e-01 7.29557931e-01
1.56343067e+00 -1.08874547e+00 8.50450099e-02 5.53398252e-01
1.38519561e+00 -6.86312020e-01 9.94155332e-02 6.75921962e-02
-1.59588426e-01 -1.61208427e+00 -1.49224296e-01 8.28035712e-01
8.40214789e-01 3.67441066e-02 2.36880496e-01 3.25884253e-01
3.68170530e-01 -8.78141105e-01 8.55427802e-01 8.24017823e-01
7.31829703e-01 -1.20369458e+00 9.54098523e-01 4.69563752e-01
-9.36003089e-01 -4.61903751e-01 -4.85389709e-01 -7.28487730e-01
5.47449648e-01 7.72359490e-01 -1.33257473e+00 8.50049853e-01
2.06550211e-01 3.64672691e-01 -5.44320196e-02 7.12857664e-01
1.33980691e-01 9.35225606e-01 1.06764734e-01 5.54502547e-01
7.27570057e-01 -6.85381472e-01 3.04085732e-01 1.15505683e+00
7.44363546e-01 4.46450375e-02 -2.57225513e-01 1.72841758e-01
3.89603198e-01 2.17225999e-01 -7.62636125e-01 -1.48483902e-01
5.67213118e-01 6.41300738e-01 -3.66639972e-01 -3.59583497e-01
-6.60395622e-01 -2.14727804e-01 -1.13642983e-01 3.44383657e-01
-8.20027411e-01 7.53269196e-02 5.80707133e-01 8.25321794e-01
1.03388049e-01 -3.03922653e-01 -3.39171290e-01 -8.83464515e-01
-1.76964656e-01 -5.91725945e-01 1.59909829e-01 -3.51936579e-01
-2.66408622e-01 7.59127289e-02 4.44922984e-01 -7.66512752e-01
-7.50144064e-01 -3.42944682e-01 -7.29367733e-01 1.10159028e+00
-1.91246212e+00 -1.10505390e+00 4.52673405e-01 1.41087860e-01
2.33024597e-01 -1.61665678e-01 6.95438802e-01 1.12295263e-01
-3.84723023e-02 1.60295926e-02 6.93903685e-01 -3.72959256e-01
3.55128616e-01 -8.76513183e-01 5.26022255e-01 5.56871057e-01
-5.50139844e-01 5.66080213e-01 8.08477700e-01 -9.18466270e-01
-4.14200664e-01 -9.77228582e-01 1.73150933e+00 5.60438484e-02
8.99638176e-01 2.83359915e-01 -4.98882443e-01 8.89979362e-01
3.10839355e-01 -9.18016255e-01 4.38150704e-01 2.72024190e-04
-1.16411127e-01 8.28265119e-03 -1.22609961e+00 2.04997167e-01
1.34313554e-01 -1.07214719e-01 -5.14968395e-01 2.94434369e-01
-9.90715437e-03 1.65127516e-01 -1.09617484e+00 7.56418183e-02
9.04134631e-01 -1.14064980e+00 1.01503634e+00 -6.92557395e-01
1.39073223e-01 3.17479342e-01 -2.72321373e-01 -1.38391840e+00
-2.68402547e-01 -4.76982176e-01 3.79798830e-01 1.03329492e+00
1.43684506e-01 -9.29209471e-01 5.25955498e-01 5.26047409e-01
1.30110592e-01 -6.37265444e-01 -1.14384544e+00 -1.01315212e+00
3.89069617e-01 -3.32248449e-01 1.06624925e+00 1.28067255e+00
4.17249978e-01 -5.85067630e-01 -2.88739890e-01 -9.55486670e-02
5.48426092e-01 -9.48340446e-02 7.02973068e-01 -1.32055354e+00
3.54604870e-01 -4.64907438e-01 -2.46771082e-01 -2.34121248e-01
9.93674845e-02 -5.77009439e-01 -7.85380661e-01 -1.40576351e+00
3.67393911e-01 -5.76843381e-01 9.29798782e-02 1.73577845e-01
2.64699250e-01 -3.87024321e-02 5.29874921e-01 1.22912586e-01
4.51577842e-01 3.68401617e-01 1.11687922e+00 1.26115665e-01
4.21330333e-02 4.02853698e-01 -8.07517946e-01 7.42837191e-01
1.14964902e+00 -2.02916011e-01 -3.48238409e-01 -7.98692368e-03
6.92518055e-01 4.47940290e-01 5.00422478e-01 -6.95843816e-01
-1.32421911e-01 -1.16165376e+00 -5.75920828e-02 -8.57924163e-01
-4.74373884e-02 -7.01144457e-01 4.21052545e-01 7.32967198e-01
1.46361426e-01 5.26848495e-01 3.20325166e-01 8.10161326e-03
-1.54583052e-01 -1.98230624e-01 4.81761694e-01 -1.81197554e-01
-1.53233081e-01 1.43110201e-01 -7.06677854e-01 -6.44557327e-02
9.99060690e-01 -1.56702697e-01 -6.26322389e-01 -6.31406486e-01
-6.23222351e-01 1.19662598e-01 5.99150896e-01 -6.62665889e-02
1.21106908e-01 -1.21158731e+00 -1.09478045e+00 -1.44070819e-01
-2.48147309e-01 -6.22623622e-01 5.10433912e-02 9.78779137e-01
-4.60275710e-01 1.61059320e+00 -3.04700077e-01 2.30391547e-01
-9.53508556e-01 4.30605829e-01 -3.68883684e-02 -6.85025394e-01
-5.25152624e-01 -1.60634935e-01 -1.64584741e-01 1.47534668e-01
-3.60548943e-01 -3.58891249e-01 -4.13719177e-01 5.83116770e-01
6.44588768e-01 8.31333160e-01 7.17720576e-03 -1.05248690e+00
-1.19146921e-01 1.43639315e-02 3.85374110e-03 -1.41596466e-01
1.36840796e+00 -5.63443184e-01 1.89130068e-01 2.62383491e-01
5.05485356e-01 2.26382181e-01 -9.08481359e-01 4.00520153e-02
2.08930552e-01 -7.06756115e-01 2.79348910e-01 -7.89336979e-01
-1.07532811e+00 6.02475762e-01 8.23900849e-02 3.59427124e-01
8.76009285e-01 -6.24035776e-01 5.84545851e-01 -2.32443526e-01
3.59721482e-01 -1.34927976e+00 -8.44360173e-01 1.14281327e-01
8.02139163e-01 -7.06598878e-01 4.54435945e-01 5.12223393e-02
-6.03990138e-01 1.10014367e+00 -3.41907293e-01 1.41543344e-01
5.78401923e-01 -2.16228776e-02 -7.40412399e-02 8.38436633e-02
-7.63671219e-01 -1.02092378e-01 -2.41926134e-01 8.17817748e-01
4.10199791e-01 7.20743954e-01 -1.18599570e+00 4.52317029e-01
-3.75567168e-01 4.18552697e-01 1.04807866e+00 4.92472589e-01
-6.91488028e-01 -8.44904900e-01 -7.39410758e-01 4.66012955e-01
-7.81654835e-01 -3.06042016e-01 3.03501844e-01 1.36686957e+00
1.11651486e-02 7.42250204e-01 4.27998304e-01 8.23800862e-02
-2.01075345e-01 2.16560215e-02 1.52192220e-01 2.06960384e-02
-5.22774816e-01 3.04070294e-01 5.19542038e-01 -4.12150994e-02
-3.69456202e-01 -1.18705332e+00 -1.11993325e+00 -1.09479809e+00
-3.48190874e-01 4.58651423e-01 9.63595271e-01 1.07808566e+00
-1.82909712e-01 1.76067263e-01 1.03095996e+00 -5.40538788e-01
-9.61170137e-01 -9.82107043e-01 -8.13533485e-01 -1.04813695e-01
2.16299161e-01 -5.66173375e-01 -3.87127757e-01 -6.53501973e-02] | [5.734038352966309, 4.043146133422852] |
4986934a-d8ff-4313-b590-9130293bb5a1 | rethinking-cnn-based-pansharpening-guided | 2006.16644 | null | https://arxiv.org/abs/2006.16644v1 | https://arxiv.org/pdf/2006.16644v1.pdf | Rethinking CNN-Based Pansharpening: Guided Colorization of Panchromatic Images via GANs | Convolutional Neural Networks (CNN)-based approaches have shown promising results in pansharpening of satellite images in recent years. However, they still exhibit limitations in producing high-quality pansharpening outputs. To that end, we propose a new self-supervised learning framework, where we treat pansharpening as a colorization problem, which brings an entirely novel perspective and solution to the problem compared to existing methods that base their solution solely on producing a super-resolution version of the multispectral image. Whereas CNN-based methods provide a reduced resolution panchromatic image as input to their model along with reduced resolution multispectral images, hence learn to increase their resolution together, we instead provide the grayscale transformed multispectral image as input, and train our model to learn the colorization of the grayscale input. We further address the fixed downscale ratio assumption during training, which does not generalize well to the full-resolution scenario. We introduce a noise injection into the training by randomly varying the downsampling ratios. Those two critical changes, along with the addition of adversarial training in the proposed PanColorization Generative Adversarial Networks (PanColorGAN) framework, help overcome the spatial detail loss and blur problems that are observed in CNN-based pansharpening. The proposed approach outperforms the previous CNN-based and traditional methods as demonstrated in our experiments. | ['Gozde Unal', 'Ugur Alganci', 'Furkan Ozcelik', 'Elif Sertel'] | 2020-06-30 | null | null | null | null | ['pansharpening'] | ['computer-vision'] | [ 7.23170638e-01 -2.49683559e-01 4.48172214e-03 -6.31866008e-02
-6.85143769e-01 -8.72768164e-01 5.63139498e-01 -5.15129685e-01
-4.33360934e-01 8.69111657e-01 -8.34591389e-02 -1.98854864e-01
-1.12569734e-01 -1.33965802e+00 -8.67026925e-01 -1.02820098e+00
3.93330753e-01 -2.46649608e-01 1.45932958e-01 -6.33074403e-01
-1.01816297e-01 6.02275491e-01 -1.29474163e+00 7.76518136e-02
1.32621157e+00 7.02373922e-01 1.28071576e-01 8.13728750e-01
2.28582144e-01 5.21092296e-01 -5.76877654e-01 -2.23937958e-01
7.65417516e-01 -5.53407311e-01 -5.39951622e-01 3.51398259e-01
6.25894845e-01 -5.17824709e-01 -5.07551789e-01 1.42960131e+00
2.60850132e-01 1.43377095e-01 3.33292723e-01 -1.06986237e+00
-1.18346298e+00 3.64234716e-01 -1.04267156e+00 1.61056101e-01
-2.13290662e-01 1.34412616e-01 6.47870660e-01 -4.90909755e-01
2.91015029e-01 9.52248871e-01 8.34813595e-01 4.21844482e-01
-1.31718040e+00 -6.17549837e-01 -1.91642180e-01 -8.65528584e-02
-1.22681665e+00 3.11198831e-03 9.94069695e-01 -4.56374846e-02
2.87712783e-01 3.90285879e-01 8.09276700e-01 6.49947226e-01
2.00610414e-01 3.10858309e-01 1.62161946e+00 -5.03253937e-01
3.41832428e-03 -1.43088385e-01 -3.37009341e-01 3.51713419e-01
3.70104760e-01 4.99765575e-01 7.81897157e-02 2.08020568e-01
1.23296070e+00 2.27074668e-01 -5.96187294e-01 -1.26155853e-01
-9.51274872e-01 8.90374303e-01 1.03503072e+00 2.92736709e-01
-4.50071126e-01 1.02406286e-01 -4.68961000e-02 4.96525109e-01
5.17045677e-01 4.12315518e-01 -3.51018280e-01 5.94060957e-01
-1.17760122e+00 2.64930576e-01 2.35371977e-01 5.64778030e-01
1.12776887e+00 4.91558671e-01 6.78566918e-02 9.33279157e-01
-1.17093213e-01 6.32102191e-01 4.51338619e-01 -8.72653008e-01
3.49146664e-01 4.48318005e-01 1.64372772e-01 -1.01167440e+00
-2.19896734e-01 -6.27571344e-01 -1.15419674e+00 6.43794298e-01
2.03786641e-01 -3.64842296e-01 -1.24373472e+00 1.63276184e+00
2.11212844e-01 8.74852538e-02 4.30711776e-01 1.11776721e+00
4.64340031e-01 1.00534689e+00 -1.20465402e-02 -7.80343041e-02
1.09124935e+00 -1.16068339e+00 -7.07176983e-01 -3.26846451e-01
-2.97587439e-02 -8.86357307e-01 9.08933342e-01 3.79451126e-01
-9.19505715e-01 -7.45895326e-01 -1.36411440e+00 5.10838442e-02
-6.62307262e-01 1.01147056e-01 7.02422678e-01 8.89053822e-01
-1.15114522e+00 7.22051322e-01 -5.32185853e-01 -1.76264793e-01
3.56976837e-01 2.74574514e-02 -3.18266034e-01 -2.62418270e-01
-1.33944583e+00 8.15785408e-01 5.98705411e-01 4.05750692e-01
-7.38824129e-01 -6.29447579e-01 -6.43800557e-01 4.07244600e-02
3.36694062e-01 -4.44412410e-01 7.67495573e-01 -1.58772075e+00
-1.51089764e+00 5.53945601e-01 5.74207306e-01 -4.18570340e-01
5.66238225e-01 -1.77126631e-01 -7.09273219e-01 2.35227972e-01
-1.83002964e-01 6.71455383e-01 1.14209068e+00 -1.62108159e+00
-6.35500133e-01 -4.88274656e-02 2.91746557e-01 2.01639742e-01
-2.16876701e-01 -1.95026383e-01 -2.85827160e-01 -1.10679698e+00
1.63984239e-01 -7.81314194e-01 -3.84434938e-01 8.79161060e-02
-2.22955152e-01 7.21036196e-01 1.16648662e+00 -9.10684168e-01
5.96871018e-01 -2.06341219e+00 4.29937541e-02 6.33204132e-02
-6.96100621e-03 6.41842127e-01 -4.68060732e-01 2.68153876e-01
-4.06097561e-01 1.56699568e-01 -7.93464422e-01 2.53818054e-02
-4.21812147e-01 4.55662191e-01 -3.06075126e-01 5.05535960e-01
4.94905233e-01 9.17031050e-01 -8.51693034e-01 -2.60513484e-01
3.88439715e-01 8.85969639e-01 -3.41486990e-01 1.95173845e-01
-1.42398462e-01 4.84666795e-01 -1.76463485e-01 4.96507525e-01
1.49023199e+00 2.28367150e-01 -6.96362779e-02 -4.27721381e-01
-3.18983376e-01 -6.00209713e-01 -1.04348123e+00 1.43284810e+00
-5.89332938e-01 4.96571809e-01 2.61936873e-01 -1.07504904e+00
9.02980208e-01 1.35063171e-01 3.18771213e-01 -6.81393802e-01
-1.73008665e-01 1.89432746e-03 -1.80983394e-01 -1.83556646e-01
7.59376347e-01 -5.20640671e-01 2.34955162e-01 2.93269485e-01
-1.41498372e-01 -6.76590443e-01 -1.41478166e-01 -1.14244163e-01
4.52193022e-01 3.34708571e-01 5.74600473e-02 -7.69234449e-02
6.42001867e-01 1.29104972e-01 5.13677180e-01 5.97410083e-01
3.00871581e-02 9.74159956e-01 2.71472365e-01 -4.11591560e-01
-1.44036937e+00 -8.93737435e-01 2.48844996e-02 7.19687045e-01
1.51186496e-01 2.61064827e-01 -8.84320617e-01 -5.20670652e-01
-3.23257446e-01 4.21325892e-01 -7.93043733e-01 -1.99522693e-02
-6.08699083e-01 -1.28305435e+00 6.38164580e-01 4.26657259e-01
1.28232491e+00 -8.38462830e-01 -5.99889100e-01 1.30585164e-01
-1.45715475e-01 -1.15886915e+00 -3.45766366e-01 2.38375083e-01
-9.95310068e-01 -1.11601102e+00 -1.15455210e+00 -6.24822140e-01
5.67610145e-01 5.29723465e-01 7.74999082e-01 -3.05568147e-02
-3.12662661e-01 9.48270187e-02 -4.63540196e-01 -8.54652300e-02
-5.23364723e-01 -1.23810865e-01 -4.55950648e-01 2.31777191e-01
-2.26317585e-01 -6.88487589e-01 -8.15257609e-01 8.79255384e-02
-1.75821888e+00 2.25365773e-01 9.62024271e-01 1.00523984e+00
5.35146713e-01 6.86121762e-01 3.18337083e-01 -9.22086954e-01
4.43745285e-01 -1.40199423e-01 -7.57247865e-01 3.22469950e-01
-6.15476310e-01 -1.30452126e-01 9.57961559e-01 -4.73600626e-01
-1.38925600e+00 9.63132381e-02 -2.16435507e-01 -3.85814548e-01
-1.77112609e-01 4.14281934e-01 -1.08408816e-01 -6.59455240e-01
7.04103589e-01 5.36318958e-01 -1.22704007e-01 -4.89608377e-01
6.50661826e-01 4.64112461e-01 1.01612854e+00 -1.93700060e-01
1.52719772e+00 7.43682146e-01 1.29834525e-02 -7.96536505e-01
-7.63488650e-01 -1.23267502e-01 -6.57160521e-01 -1.07301444e-01
8.85043740e-01 -9.90373313e-01 -6.27956912e-02 7.84637213e-01
-8.75449359e-01 -5.03021955e-01 -2.57250130e-01 1.51080385e-01
-4.29876536e-01 5.74088037e-01 -5.20609498e-01 -6.21375740e-01
-4.29718673e-01 -7.73670554e-01 7.75834739e-01 6.06323600e-01
7.10610509e-01 -9.25576270e-01 1.14505820e-01 1.72559634e-01
9.56903160e-01 8.87377620e-01 8.62670362e-01 1.76708788e-01
-4.96623755e-01 -1.17341280e-01 -6.24685526e-01 8.02569032e-01
3.54064941e-01 -9.99638066e-03 -9.09638584e-01 -4.12401915e-01
1.38445795e-01 -2.50720441e-01 1.07472813e+00 3.39340001e-01
1.12833834e+00 -3.24355036e-01 1.67626753e-01 1.10293067e+00
2.03847218e+00 9.68884397e-03 1.06264067e+00 6.29968286e-01
7.14191020e-01 3.64786834e-01 3.95618737e-01 1.20701015e-01
-6.06959611e-02 4.87027317e-01 6.96851075e-01 -8.90492618e-01
-4.61431593e-01 -2.22399712e-01 7.24167228e-02 1.77339956e-01
-4.98727590e-01 -1.49840564e-01 -4.85388905e-01 3.95253271e-01
-1.47212911e+00 -9.66707647e-01 -1.62327901e-01 2.07720160e+00
9.08911645e-01 -2.09703431e-01 -1.48737401e-01 3.21950972e-01
9.35648978e-01 5.84081233e-01 -4.77480352e-01 -1.03304803e-01
-5.55499494e-01 5.92888057e-01 1.02315891e+00 4.77443904e-01
-1.35761058e+00 1.07035565e+00 6.20493317e+00 6.05021298e-01
-1.54381466e+00 2.02039123e-01 5.40551305e-01 4.40350622e-01
-2.77747244e-01 4.03239317e-02 -3.98861207e-02 1.49694905e-01
5.12604475e-01 7.92578012e-02 7.61615872e-01 5.67830980e-01
1.13804445e-01 -1.11356892e-01 -2.81020343e-01 8.14885497e-01
4.13812883e-02 -9.88236845e-01 2.18285635e-01 -1.40560463e-01
1.12163174e+00 -7.09921345e-02 3.94440651e-01 -1.57219358e-02
3.56294781e-01 -9.92142558e-01 4.45070088e-01 4.55171973e-01
9.99086261e-01 -8.41033578e-01 7.05495238e-01 9.20533165e-02
-1.08219242e+00 -2.23785862e-01 -6.02405906e-01 1.33446798e-01
5.06904200e-02 5.47072947e-01 -1.17979877e-01 9.74632025e-01
6.87073529e-01 5.17710745e-01 -7.17082083e-01 9.84130979e-01
-4.95566517e-01 5.09383678e-01 -4.72343713e-02 7.11674452e-01
3.47807527e-01 -5.02698541e-01 2.73397774e-01 1.11162353e+00
3.48751456e-01 2.81602174e-01 -1.56859562e-01 1.09514594e+00
-3.22204866e-02 -1.88855261e-01 -3.52700591e-01 -5.87637536e-03
-4.94380668e-02 1.37638831e+00 -7.40782857e-01 -2.26525038e-01
-4.58039969e-01 1.27740240e+00 -1.54906809e-01 7.20601439e-01
-8.90535891e-01 -5.51496267e-01 3.66172999e-01 -1.63128629e-01
4.47134227e-01 -1.02767430e-01 -1.73998341e-01 -1.29068875e+00
-2.88856536e-01 -1.01810300e+00 2.76355743e-01 -1.02549326e+00
-9.80642557e-01 6.78054452e-01 -1.01650558e-01 -1.31435835e+00
1.79765120e-01 -4.99161601e-01 -7.53823459e-01 1.04870713e+00
-2.26866245e+00 -1.55863845e+00 -8.14842284e-01 6.56542301e-01
4.35640633e-01 1.44075140e-01 6.58500671e-01 1.71586141e-01
-4.29241210e-01 2.33693019e-01 3.00614178e-01 1.78781509e-01
7.07653463e-01 -1.19689608e+00 4.20930028e-01 1.39018798e+00
-1.95985362e-01 2.09513947e-01 8.18395793e-01 -4.56549883e-01
-1.23415780e+00 -1.46800649e+00 2.19676271e-01 1.24828227e-01
4.62830454e-01 1.20589048e-01 -1.04567909e+00 5.73590279e-01
3.98141861e-01 2.32158706e-01 2.13953301e-01 -6.88175797e-01
-3.75207275e-01 -3.03473026e-01 -1.32504761e+00 4.33179229e-01
5.97687483e-01 -4.16601062e-01 -5.51899076e-01 1.18560888e-01
6.45268559e-01 -3.42400670e-01 -6.84980810e-01 4.52513546e-01
4.01805311e-01 -1.08103096e+00 1.12430692e+00 -1.95987940e-01
7.79553115e-01 -5.48700213e-01 -5.70579320e-02 -1.61030900e+00
-3.84893119e-01 -4.54148442e-01 3.70101392e-01 9.93463457e-01
2.87505448e-01 -7.56971359e-01 6.13536894e-01 1.17009468e-01
-1.30891083e-02 -1.25387564e-01 -4.52676594e-01 -5.89650691e-01
4.80550855e-01 5.54190651e-02 8.01393688e-01 1.18120098e+00
-7.96258092e-01 -2.67715782e-01 -7.28172898e-01 6.23996079e-01
8.83977532e-01 3.54696035e-01 6.54588759e-01 -8.86490464e-01
-4.16539639e-01 -3.41742277e-01 -2.42212061e-02 -6.85995519e-01
-2.25753114e-01 -3.72676104e-01 1.07259810e-01 -1.37356794e+00
1.36050388e-01 -3.15665126e-01 -2.55034506e-01 5.34068584e-01
-3.51087838e-01 9.10284340e-01 2.23034918e-01 1.57559291e-01
2.01334849e-01 4.59148675e-01 1.49716139e+00 -3.36213171e-01
-2.52722830e-01 -9.25528929e-02 -7.90345967e-01 4.98982280e-01
9.40853953e-01 -2.98789382e-01 -2.02890441e-01 -6.33871853e-01
1.07702315e-01 1.25517085e-01 6.87643647e-01 -1.15296721e+00
8.68908502e-03 -3.59010130e-01 6.61822081e-01 -2.75379747e-01
2.49616399e-01 -9.07302082e-01 4.12014753e-01 5.37891567e-01
-1.73008889e-01 -2.02927575e-01 3.76418054e-01 4.52532679e-01
-3.47114593e-01 -5.67543022e-02 1.23774040e+00 -2.98177809e-01
-7.39131391e-01 2.64204323e-01 -1.82270795e-01 -4.41144854e-01
8.07913601e-01 -1.77776083e-01 -5.46934903e-01 -4.34427887e-01
-5.44692278e-01 -2.36922994e-01 7.01115429e-01 2.40323737e-01
4.94746655e-01 -1.18457103e+00 -7.77142525e-01 3.49978179e-01
-1.41866490e-01 -8.55755135e-02 5.41346967e-01 5.39962173e-01
-8.99634063e-01 6.31555170e-02 -7.45349348e-01 -1.03093348e-01
-7.57486224e-01 6.78363800e-01 6.95607603e-01 -1.02957383e-01
-6.46443784e-01 4.15462494e-01 1.65913194e-01 -3.91815186e-01
-2.87046313e-01 -2.20491573e-01 -2.56928355e-01 -2.82077581e-01
4.63322371e-01 1.15317136e-01 7.15192109e-02 -6.60316229e-01
1.70031369e-01 8.50046039e-01 2.06711471e-01 -1.92899019e-01
1.59457481e+00 -1.76924184e-01 -1.99091271e-01 -1.01474926e-01
9.99977887e-01 9.40471143e-02 -1.50613129e+00 -3.02208155e-01
-5.47847331e-01 -6.78576350e-01 5.01413226e-01 -9.40709889e-01
-1.65702820e+00 7.94668615e-01 8.90084147e-01 2.77627587e-01
1.64308381e+00 -4.62394327e-01 7.48136401e-01 8.83840844e-02
-2.21592877e-02 -9.98273015e-01 -3.56245390e-03 1.55856013e-01
8.74023318e-01 -1.20172346e+00 1.51231185e-01 -5.02092779e-01
-4.73941714e-01 1.29999208e+00 5.40227890e-01 -3.91873419e-01
2.78007656e-01 1.88739225e-01 3.02485615e-01 5.90950549e-02
-1.19930431e-01 -3.36879492e-01 8.02357495e-02 8.67514670e-01
1.63739428e-01 4.44112457e-02 -2.46701822e-01 1.28019914e-01
-1.66658431e-01 -1.69110729e-03 7.28575230e-01 7.00759828e-01
-3.98979276e-01 -1.00936770e+00 -9.10888612e-01 -3.17978440e-04
-4.01080698e-01 -2.58300394e-01 -1.85682073e-01 8.96176100e-01
3.18714499e-01 7.91850328e-01 -6.51818514e-03 -2.22588807e-01
3.59424770e-01 -2.07108825e-01 3.21349442e-01 -1.45904183e-01
-4.69211251e-01 9.26878601e-02 -4.31297600e-01 -2.72622555e-01
-7.18453646e-01 -2.30100438e-01 -5.74605405e-01 -3.69483739e-01
-3.24329287e-01 3.83723155e-02 6.33936644e-01 7.24653006e-01
-2.17924073e-01 6.08219743e-01 8.76921952e-01 -1.07557547e+00
-4.34904069e-01 -9.32564378e-01 -7.82809258e-01 3.88011515e-01
6.60857737e-01 -2.61579812e-01 -4.76622522e-01 1.85784161e-01] | [10.20046329498291, -1.9765450954437256] |
4221eb8f-3f34-415c-8003-baaf47f8004c | study-and-observation-of-the-variations-of | 1809.06188 | null | http://arxiv.org/abs/1809.06188v3 | http://arxiv.org/pdf/1809.06188v3.pdf | Study and Observation of the Variations of Accuracies for Handwritten Digits Recognition with Various Hidden Layers and Epochs using Neural Network Algorithm | In recent days, Artificial Neural Network (ANN) can be applied to a vast
majority of fields including business, medicine, engineering, etc. The most
popular areas where ANN is employed nowadays are pattern and sequence
recognition, novelty detection, character recognition, regression analysis,
speech recognition, image compression, stock market prediction, Electronic
nose, security, loan applications, data processing, robotics, and control. The
benefits associated with its broad applications leads to increasing popularity
of ANN in the era of 21st Century. ANN confers many benefits such as organic
learning, nonlinear data processing, fault tolerance, and self-repairing
compared to other conventional approaches. The primary objective of this paper
is to analyze the influence of the hidden layers of a neural network over the
overall performance of the network. To demonstrate this influence, we applied
neural network with different layers on the MNIST dataset. Also, another goal
is to observe the variations of accuracies of ANN for different numbers of
hidden layers and epochs and to compare and contrast among them. | ['Mohammad Mahmudur Rahman Khan', 'Md. Abu Bakr Siddique', 'Zahidun Ashrafi', 'Rezoana Bente Arif'] | 2018-09-17 | null | null | null | null | ['stock-market-prediction'] | ['time-series'] | [ 2.92681038e-01 -4.21976060e-01 -9.40979868e-02 -7.13907406e-02
5.17996788e-01 -8.77717361e-02 3.14693362e-01 3.92371804e-01
-5.55131912e-01 7.77996540e-01 -4.25912887e-01 -1.85797527e-01
-4.99618828e-01 -6.15231216e-01 -1.99277326e-01 -7.01590955e-01
-6.06523156e-02 -1.48054855e-02 2.15376258e-01 -1.67280406e-01
7.44795144e-01 1.09137094e+00 -1.82489026e+00 -5.68149760e-02
7.08482444e-01 1.35107183e+00 -1.25533685e-01 2.24721268e-01
-6.18225560e-02 8.47470701e-01 -5.41593015e-01 -1.82704031e-01
1.70047268e-01 -1.65008202e-01 -1.97965056e-01 2.77463824e-01
-5.30866563e-01 1.34993240e-01 -1.54549405e-01 7.30605483e-01
3.87205869e-01 2.21380517e-01 7.41247952e-01 -9.52189803e-01
-6.59719467e-01 2.79502481e-01 -3.69657993e-01 3.38919431e-01
-5.78506617e-03 1.04525991e-01 4.55392897e-01 -7.57491589e-01
2.14441016e-01 7.71721423e-01 6.29096150e-01 3.22849691e-01
-7.48285413e-01 -7.26439595e-01 -2.73951113e-01 6.19027078e-01
-1.30468202e+00 -1.82849839e-01 6.81875944e-01 -4.16704237e-01
8.40982556e-01 1.52719408e-01 4.87019479e-01 7.20110476e-01
5.91297388e-01 5.06676853e-01 1.12548566e+00 -4.85394090e-01
3.76621962e-01 5.04857957e-01 2.54696578e-01 3.57736498e-01
4.90428776e-01 2.20877990e-01 -4.05859500e-01 1.98333010e-01
4.54004049e-01 4.03303385e-01 -5.78809343e-02 1.09156281e-01
-8.65126729e-01 6.73602998e-01 2.63824284e-01 6.86564207e-01
-6.25573635e-01 -2.52069235e-01 4.59994406e-01 4.56106693e-01
2.19609812e-01 7.26386189e-01 -5.42867780e-01 -7.00755268e-02
-6.48332715e-01 1.15051279e-02 6.80820584e-01 3.11513603e-01
2.58409530e-01 6.43863499e-01 2.54327744e-01 9.41628933e-01
5.87957818e-03 2.10227102e-01 1.35943902e+00 -5.15208006e-01
-5.93187213e-02 8.78783882e-01 -3.77431899e-01 -1.07087696e+00
-7.15935409e-01 -6.87725723e-01 -1.24134016e+00 3.31602156e-01
1.11423679e-01 2.50900052e-02 -6.66288912e-01 1.03049159e+00
7.03667179e-02 1.00809865e-01 2.87448704e-01 4.45832253e-01
6.39444232e-01 5.91679871e-01 -2.13103041e-01 -3.75897169e-01
9.71267700e-01 -6.24179602e-01 -5.96164823e-01 -1.86347097e-01
1.44057050e-01 -6.80706084e-01 5.37092149e-01 7.31815159e-01
-6.78528130e-01 -5.32629848e-01 -8.88041437e-01 4.97801840e-01
-7.67817438e-01 3.01624775e-01 6.23625755e-01 5.64057648e-01
-4.38284725e-01 9.00815427e-01 -6.17699564e-01 -4.35535342e-01
3.15578222e-01 5.83342969e-01 -1.94767118e-01 3.24769020e-01
-9.74553406e-01 1.16123354e+00 8.45262945e-01 2.19313875e-01
-1.34080380e-01 -1.93369538e-01 -5.20447135e-01 2.44139969e-01
8.89926180e-02 -1.66099176e-01 7.19696939e-01 -9.70144689e-01
-1.39226770e+00 4.56426233e-01 1.03815153e-01 -7.94227958e-01
1.27599999e-01 1.08528562e-01 -7.67539740e-01 -2.83192635e-01
-5.43531060e-01 2.72498250e-01 9.01474476e-01 -3.94593656e-01
-4.50322449e-01 -5.10309935e-01 -7.07026422e-01 -5.58166169e-02
-6.32698417e-01 -1.45845041e-01 2.36082509e-01 -7.29264200e-01
2.73570627e-01 -8.78082573e-01 -2.02429891e-02 -2.70208448e-01
-2.04735562e-01 -2.52840519e-01 8.61508369e-01 -6.62149429e-01
1.07309747e+00 -2.26707840e+00 -3.64169776e-02 5.74084044e-01
-2.31385961e-01 3.92404944e-01 2.32110992e-01 4.56948698e-01
-2.68929005e-01 -9.13325101e-02 -2.48386905e-01 2.40339607e-01
-5.36881626e-01 1.93375126e-01 -9.02656168e-02 4.04244602e-01
4.03572738e-01 6.06411219e-01 -1.52046174e-01 2.08461788e-02
3.58460307e-01 4.50889438e-01 -1.44592226e-01 -1.37924001e-01
-1.97457094e-02 3.28984916e-01 -2.22599998e-01 9.16071951e-01
1.44746691e-01 -2.93795407e-01 -7.67474473e-02 2.92434227e-02
-2.58373022e-01 -2.45786324e-01 -1.13215768e+00 7.28071749e-01
-1.94934770e-01 9.31934476e-01 -5.15182257e-01 -1.38579297e+00
1.37039995e+00 4.62469041e-01 5.18008113e-01 -8.55229676e-01
5.07683098e-01 5.06999254e-01 4.42491829e-01 -6.65045083e-01
3.32555681e-01 1.79417972e-02 4.58646297e-01 1.28653586e-01
-2.75272906e-01 3.84328187e-01 2.67393082e-01 -5.55368125e-01
6.38197482e-01 -4.22195733e-01 4.86867726e-01 4.72848862e-02
7.36907601e-01 -1.87991098e-01 3.18844318e-01 1.57290295e-01
4.80706543e-02 1.39057100e-01 2.00212717e-01 -5.95226109e-01
-1.01177239e+00 -3.98436368e-01 -2.61885434e-01 4.47078854e-01
-1.77915290e-01 5.12505889e-01 -3.40563744e-01 7.61336833e-02
2.18891293e-01 6.47872329e-01 -3.34713250e-01 -5.11912763e-01
-5.41490674e-01 -8.68502915e-01 3.83702338e-01 4.25222903e-01
7.72516549e-01 -1.53151715e+00 -6.90894604e-01 3.17940295e-01
5.62107682e-01 -1.12390721e+00 3.65076870e-01 1.02963813e-01
-1.43643463e+00 -9.15759087e-01 -6.40244603e-01 -6.20611310e-01
4.05521274e-01 -3.76906335e-01 6.00133061e-01 2.65700996e-01
-5.27456284e-01 -2.60559544e-02 -1.52814880e-01 -7.61261344e-01
-5.96832633e-01 3.95767897e-01 3.80101234e-01 2.04868272e-01
3.02373499e-01 -7.17333257e-01 -4.00317580e-01 2.42281958e-01
-9.86752510e-01 -4.36839908e-01 1.10365856e+00 7.38816381e-01
1.91885054e-01 4.88708109e-01 1.05085027e+00 -9.37167943e-01
8.84626687e-01 -4.60726082e-01 -7.38361835e-01 2.96398193e-01
-1.30077457e+00 3.74086313e-02 7.99709201e-01 -6.66720092e-01
-5.94961345e-01 -1.71191886e-01 -2.17661262e-01 -2.19796032e-01
-1.57565162e-01 7.94436753e-01 1.69726640e-01 -2.77042568e-01
7.15354860e-01 6.60815835e-01 5.20425200e-01 -5.17483115e-01
-4.13540721e-01 8.27439427e-01 4.99807328e-01 1.78364664e-01
4.88561422e-01 -1.58331506e-02 3.31138194e-01 -1.24657083e+00
-2.59153694e-01 -3.45966011e-01 -2.60976344e-01 -2.57592082e-01
3.77123415e-01 -2.97505438e-01 -9.90662575e-01 8.16068828e-01
-8.99775147e-01 2.15070724e-01 -1.79940522e-01 5.99559605e-01
-3.87123367e-03 3.03317398e-01 -3.86690944e-01 -1.16033638e+00
-6.24803066e-01 -8.71274590e-01 -1.66982651e-01 5.86885452e-01
-1.67499244e-01 -9.19383109e-01 -4.38663900e-01 1.41389057e-01
6.30533934e-01 4.02968258e-01 1.01418710e+00 -9.75220621e-01
-3.69738728e-01 -7.40696609e-01 8.04876462e-02 8.29622626e-01
3.06587219e-01 1.35644108e-01 -8.10769320e-01 -2.67053824e-02
2.60073155e-01 -6.08055200e-03 7.12604165e-01 3.19914639e-01
1.21789873e+00 -5.66333592e-01 -1.58790752e-01 2.53513634e-01
1.34054661e+00 9.72551703e-01 8.24870884e-01 5.71430981e-01
2.50739723e-01 6.09019339e-01 2.37148225e-01 3.17207575e-01
-2.33828366e-01 1.24122135e-01 4.41595256e-01 1.53966576e-01
4.71009582e-01 4.92939860e-01 8.81071389e-02 8.32450211e-01
-2.23453239e-01 -1.93798825e-01 -8.22825670e-01 2.65733659e-01
-1.36861777e+00 -9.63379860e-01 2.96938479e-01 2.23863721e+00
3.38715285e-01 3.46483439e-01 2.51657665e-02 1.06562221e+00
6.71781778e-01 -3.89134616e-01 -8.65946233e-01 -5.22594571e-01
-2.31285021e-01 1.48956031e-01 5.06482482e-01 -6.84931874e-02
-7.76964724e-01 3.85549337e-01 5.89569998e+00 7.33451843e-01
-1.79985392e+00 -3.51601183e-01 7.57549345e-01 1.38892740e-01
2.51947582e-01 -5.54157317e-01 -6.11417413e-01 6.99518919e-01
8.14946771e-01 1.53944991e-03 5.59511483e-01 8.65721285e-01
1.20118469e-01 -1.14019997e-01 -7.61645913e-01 9.01979744e-01
-3.38223316e-02 -1.32974470e+00 -5.17930426e-02 -1.06249992e-02
6.00474000e-01 -2.31327742e-01 2.80400634e-01 2.67891198e-01
-7.34248102e-01 -1.09834611e+00 2.61774242e-01 7.00172782e-01
2.40517169e-01 -8.88440728e-01 9.38404977e-01 5.93841970e-01
-5.89187086e-01 -7.01229811e-01 -1.04229331e-01 -3.87426078e-01
-2.02751487e-01 6.54375374e-01 -9.51411307e-01 4.22944099e-01
4.47605163e-01 4.34864521e-01 -4.41392839e-01 1.33250821e+00
2.34400123e-01 4.70712721e-01 -2.43227854e-01 -6.05222285e-01
2.45277341e-02 -2.01426104e-01 3.63571763e-01 6.95402265e-01
4.45325941e-01 9.27348062e-02 -3.83210599e-01 5.76994956e-01
5.27295470e-02 2.37219140e-01 -3.93906385e-01 -5.21586657e-01
3.49872321e-01 9.03310180e-01 -9.23234642e-01 -2.70082772e-01
-1.67831793e-01 6.13429844e-01 -1.40663266e-01 2.87179589e-01
-4.62633938e-01 -6.55050814e-01 3.04901272e-01 3.36756498e-01
2.10174263e-01 -2.06724375e-01 -6.83706820e-01 -7.40233779e-01
-1.67042166e-02 -8.13582242e-01 3.76924872e-01 -4.60919172e-01
-9.18998957e-01 5.87665975e-01 -2.31951341e-01 -1.30298316e+00
-4.48137879e-01 -9.63495255e-01 -6.94507241e-01 7.06863761e-01
-1.38505840e+00 -3.25193197e-01 -2.77062416e-01 2.20006093e-01
4.67021823e-01 -8.96050870e-01 6.25095546e-01 3.92252266e-01
-8.48546088e-01 4.07672912e-01 4.03295279e-01 1.04114488e-01
2.38917753e-01 -6.37836039e-01 2.00163107e-02 8.42272103e-01
1.58190392e-02 3.01851362e-01 5.10624528e-01 -4.05243903e-01
-9.83909249e-01 -7.74311423e-01 8.26648891e-01 4.53820258e-01
3.35954964e-01 3.16727072e-01 -9.39055502e-01 1.86813727e-01
-1.70689002e-01 -1.98931918e-01 3.67966920e-01 -2.90328771e-01
1.16952144e-01 -6.81081772e-01 -1.34253204e+00 5.04111409e-01
-2.46850736e-02 1.14867566e-02 -1.97390094e-01 8.11048076e-02
2.28086904e-01 -3.52085680e-01 -8.42128158e-01 5.76318920e-01
7.90474474e-01 -1.05225277e+00 8.62372041e-01 -3.57165724e-01
5.06376684e-01 -2.89598946e-02 2.99432397e-01 -8.39188099e-01
-1.03530139e-01 -1.26889646e-01 -2.17982620e-01 8.96340370e-01
5.74282646e-01 -1.19520390e+00 8.85402262e-01 6.93496704e-01
3.68885621e-02 -1.21039891e+00 -9.31695104e-01 -7.70238578e-01
-3.05425614e-01 -7.26267472e-02 4.40391570e-01 9.32178617e-01
-1.50288135e-01 2.32223511e-01 -3.03693175e-01 -1.33353561e-01
2.68569261e-01 -1.23644575e-01 1.97764575e-01 -1.67678952e+00
-2.65020370e-01 -6.45072877e-01 -8.16825628e-01 -3.18350613e-01
-2.84257382e-01 -5.12852669e-01 -4.40684706e-01 -1.30027544e+00
-5.53229511e-01 -4.17514384e-01 -5.61665952e-01 3.78910899e-01
1.44318908e-01 3.55744809e-01 1.73934162e-01 3.49839985e-01
1.42722845e-01 2.18591332e-01 9.19061124e-01 -9.23188552e-02
-4.55377519e-01 4.81267124e-01 -3.89664739e-01 6.69408679e-01
1.04423761e+00 -3.71027201e-01 -1.72978148e-01 -2.26358511e-02
1.05701976e-01 -7.81605542e-02 7.21270517e-02 -1.32738912e+00
4.16421652e-01 -5.75560965e-02 7.89848506e-01 -3.01556587e-01
3.36590469e-01 -1.05439687e+00 3.13760281e-01 1.00343049e+00
-4.88677546e-02 2.52058953e-01 2.91444868e-01 1.95853427e-01
-3.94065380e-01 -7.66839206e-01 6.96213961e-01 -8.01452436e-03
-7.83641458e-01 6.95857927e-02 -6.69768453e-01 -6.22001886e-01
1.24304986e+00 -6.73719049e-01 2.88310573e-02 -2.05121726e-01
-5.37386596e-01 -9.32842419e-02 -2.96985600e-02 4.44286227e-01
6.74683511e-01 -8.79427910e-01 -3.99521321e-01 3.94620568e-01
-2.40854129e-01 -6.18132688e-02 1.36571497e-01 8.90723348e-01
-6.68057859e-01 4.68233317e-01 -4.19089139e-01 -4.15538132e-01
-1.33457541e+00 5.27122438e-01 2.84305245e-01 -9.01343822e-02
-1.30715191e-01 4.56522584e-01 -6.45380616e-01 -2.80485265e-02
4.52414483e-01 -3.25159878e-01 -6.01454020e-01 -2.18094345e-02
3.02341551e-01 8.54620218e-01 3.38921040e-01 -9.06979665e-02
-3.45510542e-01 5.83762884e-01 -1.17295668e-01 2.75385976e-01
1.27703857e+00 4.05110449e-01 -1.62693635e-01 6.79747760e-01
8.31617773e-01 -3.89060676e-01 -5.36150813e-01 -5.42902723e-02
4.17854071e-01 7.58320838e-02 -9.23565999e-02 -7.20700026e-01
-1.03332770e+00 7.58232355e-01 8.44193041e-01 5.20930052e-01
1.34450722e+00 -7.00959146e-01 6.35053754e-01 8.71070862e-01
-6.46257997e-02 -1.13034856e+00 7.22702742e-02 4.94559795e-01
7.00863600e-01 -1.04685473e+00 1.77125067e-01 6.45056069e-02
-5.63978195e-01 1.50282443e+00 2.69689590e-01 -2.29433298e-01
6.77943587e-01 1.56245053e-01 -7.15076700e-02 1.40959933e-01
-5.34516990e-01 8.48164111e-02 3.31047744e-01 3.12574267e-01
4.71882612e-01 -3.23599875e-01 -5.19228518e-01 3.18703145e-01
-1.75588593e-01 2.92227119e-01 5.80274642e-01 9.34998095e-01
-5.60178638e-01 -9.35276806e-01 -4.43076193e-01 8.91513228e-01
-7.47466683e-01 -3.98450419e-02 -2.17956990e-01 8.11393559e-01
3.02647334e-02 6.52426541e-01 1.57029733e-01 -3.58981818e-01
3.16189498e-01 2.95313507e-01 1.90056711e-01 -1.38117447e-01
-6.33742809e-01 -3.24415386e-01 -3.50276083e-01 1.02973692e-01
-3.79963577e-01 -5.45888722e-01 -1.11070824e+00 -2.64636695e-01
-5.00782967e-01 1.78895593e-01 1.15597010e+00 1.12019408e+00
2.35711083e-01 6.33155465e-01 7.69101262e-01 -4.37826633e-01
-6.86533391e-01 -1.19852185e+00 -5.65657496e-01 1.05391353e-01
6.57916516e-02 -4.91299272e-01 -4.05628175e-01 -6.96825050e-03] | [8.277992248535156, 3.044825792312622] |
f5aa3b85-3ab9-4bcf-95e0-9b01fb13f1de | towards-holistic-surgical-scene-understanding | 2212.04582 | null | https://arxiv.org/abs/2212.04582v3 | https://arxiv.org/pdf/2212.04582v3.pdf | Towards Holistic Surgical Scene Understanding | Most benchmarks for studying surgical interventions focus on a specific challenge instead of leveraging the intrinsic complementarity among different tasks. In this work, we present a new experimental framework towards holistic surgical scene understanding. First, we introduce the Phase, Step, Instrument, and Atomic Visual Action recognition (PSI-AVA) Dataset. PSI-AVA includes annotations for both long-term (Phase and Step recognition) and short-term reasoning (Instrument detection and novel Atomic Action recognition) in robot-assisted radical prostatectomy videos. Second, we present Transformers for Action, Phase, Instrument, and steps Recognition (TAPIR) as a strong baseline for surgical scene understanding. TAPIR leverages our dataset's multi-level annotations as it benefits from the learned representation on the instrument detection task to improve its classification capacity. Our experimental results in both PSI-AVA and other publicly available databases demonstrate the adequacy of our framework to spur future research on holistic surgical scene understanding. | ['Pablo Arbeláez', 'Nicolás Fernández', 'Juan Caicedo', 'Jessica Santander', 'Mathilde Verlyk', 'Nicolás Ayobi', 'Isabela Hernández', 'Paola Ruiz Puentes', 'Natalia Valderrama'] | 2022-12-08 | null | null | null | null | ['atomic-action-recognition'] | ['computer-vision'] | [ 5.46743274e-01 5.54214180e-01 -8.69334698e-01 -1.81946948e-01
-1.29030538e+00 -7.84790695e-01 6.83936715e-01 2.52555907e-01
-1.68532372e-01 9.93079618e-02 9.24824774e-01 -4.60151792e-01
-2.35359907e-01 -2.21293673e-01 -6.27618015e-01 -6.95759833e-01
-1.39590114e-01 1.50154248e-01 -1.18659832e-01 -1.76145211e-02
2.12842718e-01 7.30315983e-01 -8.04328501e-01 8.42701912e-01
3.02495241e-01 1.05564189e+00 1.62705839e-01 9.12745237e-01
2.71165341e-01 1.58527541e+00 2.41014753e-02 -7.47773647e-02
3.16353947e-01 -4.90510732e-01 -1.17023087e+00 -3.61607969e-02
4.21246231e-01 -3.55935276e-01 -6.01032674e-01 7.17596948e-01
3.60653877e-01 -2.07815558e-01 6.84120357e-01 -1.05177462e+00
-1.12831816e-01 4.89087105e-01 -4.32409465e-01 1.70396581e-01
3.76642227e-01 6.67770147e-01 9.40457702e-01 -7.04082251e-01
1.11085761e+00 9.36579108e-01 8.78865302e-01 5.53753495e-01
-1.01044047e+00 -3.96259964e-01 8.64356756e-03 1.31406277e-01
-7.96988666e-01 -4.36323971e-01 6.71865761e-01 -6.75285876e-01
9.68891382e-01 3.90848547e-01 9.25507724e-01 1.38160062e+00
8.51397038e-01 1.31164789e+00 1.02472079e+00 -1.49585977e-01
3.06395367e-02 -3.80638272e-01 6.58733621e-02 1.28433514e+00
2.31317550e-01 4.94732231e-01 -5.03828824e-01 1.80249602e-01
1.10038209e+00 3.10920864e-01 -2.58010924e-01 -1.03861654e+00
-1.70177245e+00 3.41901153e-01 8.86937261e-01 1.59407884e-01
-6.62665784e-01 4.31442350e-01 8.20061386e-01 -1.55271888e-01
-3.42007369e-01 8.22225869e-01 -7.23333433e-02 -4.48030144e-01
-5.31594694e-01 -3.60644519e-01 7.42174685e-01 8.22149098e-01
4.05390650e-01 -3.03877354e-01 -7.07540393e-01 4.60182637e-01
-3.35106179e-02 1.40421083e-02 5.17358661e-01 -1.09660566e+00
2.28155360e-01 8.85053575e-01 -2.17603654e-01 -5.84560871e-01
-8.99001360e-01 -5.32810807e-01 -8.05392563e-01 4.50745821e-01
1.96477860e-01 3.26651663e-01 -1.30930388e+00 1.25419629e+00
-5.34916781e-02 1.75234694e-02 1.57646358e-01 6.80206835e-01
1.12530077e+00 -1.83498427e-01 2.13428676e-01 2.61634625e-02
1.51641440e+00 -1.64690781e+00 -7.18666196e-01 -6.31367743e-01
1.28477216e+00 -3.21566463e-01 1.00206316e+00 4.13543105e-01
-8.95012200e-01 -2.03877971e-01 -9.07303393e-01 -3.75417441e-01
-1.12829082e-01 6.57301843e-01 1.10593414e+00 2.04823807e-01
-8.30926001e-01 1.74732402e-01 -1.39075089e+00 -3.60873908e-01
7.77632773e-01 2.89728969e-01 -9.33829665e-01 -2.71596223e-01
-2.45544180e-01 1.08518231e+00 1.92344368e-01 2.39173949e-01
-1.84489799e+00 -8.45701098e-01 -1.36201847e+00 1.30346409e-04
6.37211561e-01 -7.76199818e-01 1.54969513e+00 -7.56177068e-01
-1.14421284e+00 1.27400315e+00 2.98658609e-02 -5.68126500e-01
3.95316631e-01 -4.60282452e-02 7.25618079e-02 6.07097387e-01
-1.49985671e-01 5.66266119e-01 4.93242383e-01 -1.17920911e+00
-3.01200628e-01 -5.48739016e-01 3.43375474e-01 3.21499944e-01
2.15171184e-02 -4.77128774e-01 -4.16859567e-01 -4.15695548e-01
8.63848701e-02 -1.24648273e+00 -5.56296289e-01 5.49719751e-01
-6.01693332e-01 2.89880097e-01 2.49446735e-01 -6.49540424e-01
1.09854066e+00 -2.39329433e+00 3.68977219e-01 -2.51277357e-01
4.39028054e-01 -1.46968551e-02 -1.27859429e-01 3.36124986e-01
-5.42010307e-01 -9.82515216e-02 -5.00615425e-02 -4.16340441e-01
-3.75476986e-01 4.99142438e-01 1.22778304e-02 5.53885162e-01
7.55995791e-03 1.25019181e+00 -1.13848841e+00 -8.08138728e-01
5.88862181e-01 -3.79679948e-02 -7.50608325e-01 4.92447764e-02
2.14685768e-01 6.25045061e-01 -4.14290279e-01 1.25749516e+00
8.05591643e-02 -3.76459152e-01 3.81038368e-01 -6.18491352e-01
5.66462278e-02 1.85368016e-01 -3.45759422e-01 2.51294041e+00
-6.15633965e-01 4.24850196e-01 2.61556357e-01 -9.04504001e-01
2.04332337e-01 3.14208150e-01 1.00141096e+00 -7.40212858e-01
1.66530550e-01 2.03694537e-01 8.96250457e-02 -5.45260191e-01
2.46857762e-01 -2.14435771e-01 -3.20624441e-01 -1.66944832e-01
3.76651585e-01 -2.08854303e-01 -5.25309294e-02 3.90530407e-01
1.67100537e+00 2.25806832e-01 1.06979287e+00 -5.36855049e-02
4.08861935e-01 6.77631259e-01 5.35698771e-01 9.95534182e-01
-7.54205525e-01 5.36052465e-01 9.37657714e-01 -5.31001091e-01
-6.24783456e-01 -1.22222424e+00 9.40314978e-02 7.68377423e-01
2.65976250e-01 -4.74879205e-01 -1.77040577e-01 -1.30676460e+00
1.87437106e-02 5.27115881e-01 -1.27834618e+00 -3.15735579e-01
-5.91108203e-01 -1.42241523e-01 4.80559856e-01 1.13333678e+00
4.51009199e-02 -1.07979751e+00 -5.76871634e-01 -7.87800401e-02
-1.26158774e-01 -1.23696125e+00 -2.81334102e-01 4.01490211e-01
-1.05767274e+00 -1.63935971e+00 -4.48932618e-01 -8.11177850e-01
8.15783501e-01 5.01160085e-01 1.06440210e+00 -1.39079869e-01
-9.30222094e-01 9.72647011e-01 -2.36832738e-01 -3.65360051e-01
-3.60135078e-01 -2.35149205e-01 -3.31196874e-01 -3.80962938e-01
-2.80249655e-01 -1.22235887e-01 -8.77204478e-01 1.06387235e-01
-7.00916111e-01 6.17637813e-01 1.33532715e+00 1.03618515e+00
6.50247693e-01 -6.77507937e-01 -1.32272124e-01 -9.28861320e-01
1.21101541e-02 -3.58642340e-01 -1.44284517e-01 3.44645917e-01
-3.23351860e-01 -1.37429878e-01 1.02231972e-01 -1.16320103e-01
-9.84800041e-01 5.88142693e-01 1.01421982e-01 -6.92790210e-01
-7.15161934e-02 6.34739280e-01 3.53387237e-01 -2.83338308e-01
8.09537768e-01 2.14523822e-01 2.86724716e-01 -1.92931056e-01
2.48683989e-01 1.40463665e-01 9.16556418e-01 -2.60487467e-01
5.68528175e-01 8.85666311e-01 4.05345351e-01 -3.87634903e-01
-1.25993967e+00 -8.84433329e-01 -7.01174080e-01 -3.50294441e-01
1.10554445e+00 -1.13099802e+00 -9.75243628e-01 1.01295926e-01
-7.47084856e-01 -5.72178245e-01 -5.51636100e-01 6.70517743e-01
-1.02280152e+00 3.32426876e-01 -7.61765420e-01 -5.89948654e-01
-4.75836456e-01 -1.49472690e+00 1.69273853e+00 4.65277582e-02
-1.89852282e-01 -9.89095271e-01 1.19484678e-01 6.61780536e-01
8.05595294e-02 6.12053335e-01 6.93625569e-01 -5.90679348e-01
-7.36419380e-01 -3.75181377e-01 -2.51973450e-01 6.76555857e-02
1.73733771e-01 -3.23749423e-01 -5.96949339e-01 -2.84030974e-01
-2.30643302e-01 -5.16991198e-01 9.38825071e-01 3.71869445e-01
1.46423256e+00 -1.63618103e-01 -7.26917565e-01 8.97429883e-01
1.24659026e+00 1.53961957e-01 7.97181606e-01 5.16958058e-01
8.06285501e-01 2.06595153e-01 1.06075549e+00 1.19878918e-01
3.13349545e-01 6.29527271e-01 9.16773975e-01 -4.69255984e-01
-5.14684856e-01 -1.55656353e-01 3.38664830e-01 3.09962362e-01
-1.02730833e-01 2.89299518e-01 -1.37314773e+00 7.27241099e-01
-1.75641418e+00 -9.30752635e-01 1.46720722e-01 1.98375070e+00
6.85015321e-01 -8.84166062e-02 -3.37913990e-01 -2.90989935e-01
-1.38015762e-01 3.29933852e-01 -4.35790926e-01 -2.13129148e-01
4.32208061e-01 8.65281150e-02 5.77619791e-01 1.85691997e-01
-1.35303903e+00 6.38266683e-01 6.32598257e+00 5.61463296e-01
-9.68553007e-01 -1.13957599e-02 4.15330827e-01 -1.72070518e-01
2.05133185e-01 -1.51923308e-02 -4.37732041e-01 -1.81868523e-01
4.82547700e-01 2.90351678e-02 8.38490203e-02 1.06528389e+00
-5.33667766e-02 -2.37129048e-01 -1.56967068e+00 1.04636014e+00
3.33024651e-01 -1.68502629e+00 -2.76816398e-01 1.18533626e-01
4.79931384e-01 -1.80422179e-02 -9.21649393e-03 7.39333570e-01
2.50175983e-01 -1.24820149e+00 1.93380326e-01 5.99084079e-01
8.30491364e-01 -1.36851773e-01 7.26368308e-01 1.24980234e-01
-1.24200594e+00 -4.40493792e-01 4.06183422e-01 2.80565321e-01
-2.50928223e-01 -1.49916157e-01 -1.25118291e+00 9.10270154e-01
3.55725378e-01 1.30480146e+00 -7.59915054e-01 1.03234041e+00
-3.06404740e-01 2.28652120e-01 1.08233534e-01 6.16572261e-01
4.02682692e-01 2.53673136e-01 4.02963817e-01 1.26646173e+00
-8.06382764e-03 1.52590990e-01 4.03775930e-01 2.73011059e-01
-1.69892348e-02 -2.30263367e-01 -8.03412259e-01 -5.51871121e-01
-1.12730846e-01 1.69384098e+00 -6.02534175e-01 -2.69832581e-01
-5.06032586e-01 1.00696349e+00 3.50544274e-01 -2.81671118e-02
-8.14405680e-01 1.19495928e-01 6.83951735e-01 -2.49252990e-01
-3.66640203e-02 -1.92312837e-01 -6.17173016e-01 -1.20436096e+00
-2.84795046e-01 -1.01189005e+00 8.68915617e-01 -9.49712574e-01
-6.51643932e-01 5.60247339e-02 -2.95259446e-01 -1.66908777e+00
-2.18828425e-01 -1.06685102e+00 -4.83079582e-01 2.19747797e-01
-1.46552908e+00 -1.74888015e+00 -9.31668460e-01 6.34736240e-01
8.40878129e-01 2.17097625e-01 1.03459203e+00 -2.22750366e-01
-6.05053067e-01 3.31309527e-01 -3.76546919e-01 4.80772346e-01
8.90202641e-01 -1.28783774e+00 -2.14366734e-01 6.13542736e-01
-8.78246650e-02 4.50917929e-01 2.67827570e-01 -6.47789419e-01
-2.11504292e+00 -8.84405494e-01 1.19383402e-01 -9.10179019e-01
7.19551027e-01 -6.74058273e-02 -2.92881876e-01 1.41354752e+00
-1.05566449e-01 3.21236759e-01 7.64190376e-01 2.76044965e-01
-4.39862221e-01 4.49446365e-02 -8.85367930e-01 6.95531011e-01
1.35951757e+00 -6.08913958e-01 -6.27730250e-01 7.43938148e-01
4.34621632e-01 -1.08323991e+00 -1.23476362e+00 9.58308578e-01
7.55221367e-01 -9.27999616e-01 1.38625550e+00 -6.68702424e-01
1.04329574e+00 1.03987539e-02 1.25355065e-01 -8.84534299e-01
-2.86309838e-01 -1.39208317e-01 -1.34826526e-01 2.64328867e-01
1.07303314e-01 -7.19281211e-02 8.08156729e-01 4.15809244e-01
-8.15372050e-01 -1.12277687e+00 -8.39332223e-01 -6.26232028e-01
-2.55359560e-01 -3.38921696e-01 -2.49297485e-01 8.52724373e-01
3.05500180e-01 -1.88276887e-01 -1.82002723e-01 2.52021044e-01
4.52289075e-01 2.92800933e-01 9.04408693e-01 -7.56017983e-01
-5.16957402e-01 -5.57100415e-01 -7.16055930e-01 -5.87941289e-01
-2.44483903e-01 -1.22312367e+00 1.72356978e-01 -1.95318007e+00
7.50140786e-01 8.87575895e-02 -5.45378804e-01 1.08038867e+00
-4.47734967e-02 2.84035951e-01 3.31453681e-01 4.34173733e-01
-8.01749110e-01 3.38447213e-01 1.56954432e+00 -2.05089509e-01
-9.13163498e-02 -2.50723869e-01 -6.74205422e-01 9.79463577e-01
2.74329513e-01 -3.62477422e-01 -1.47751778e-01 -2.59565055e-01
-1.00609317e-01 5.86292863e-01 5.90977550e-01 -1.08841777e+00
3.91661614e-01 -2.78580636e-01 4.25246954e-01 -5.25306642e-01
6.52890384e-01 -8.70393038e-01 -1.26097038e-01 1.05038393e+00
-4.29845750e-01 -4.80358273e-01 2.92354256e-01 7.89030075e-01
-2.88644314e-01 1.32231012e-01 7.34058619e-01 -3.53813052e-01
-1.00210774e+00 3.24143261e-01 -1.78208739e-01 -1.84409201e-01
1.32434607e+00 -1.72737837e-01 -7.43994355e-01 -1.18874781e-01
-1.14790726e+00 3.70535195e-01 5.83546579e-01 5.46320200e-01
6.84952319e-01 -1.07223356e+00 -3.49430948e-01 -2.54160948e-02
8.02646279e-01 -6.79354742e-02 7.26266325e-01 1.81911695e+00
-8.09747100e-01 6.07890606e-01 -3.55937034e-01 -7.19153225e-01
-1.47190118e+00 8.63484383e-01 5.68734825e-01 -9.09382880e-01
-1.04305661e+00 5.98281145e-01 9.00013328e-01 -5.11200130e-01
4.77114320e-01 -4.36191618e-01 -2.00774431e-01 -2.21801281e-01
3.09303910e-01 -1.88015357e-01 -9.61836353e-02 -1.65323511e-01
-6.08234048e-01 2.38776758e-01 -1.92472547e-01 1.96773544e-01
1.20666993e+00 2.65151888e-01 8.34989101e-02 2.79301137e-01
1.06308031e+00 -1.18427306e-01 -1.02938342e+00 -9.10221785e-02
1.03411851e-02 -3.02306235e-01 1.82495981e-01 -1.20588183e+00
-7.41988838e-01 7.96209216e-01 5.46050370e-01 -5.87994397e-01
1.31758273e+00 1.52254313e-01 4.44390714e-01 4.97953176e-01
5.13467789e-01 -6.97170019e-01 4.43691343e-01 4.38615322e-01
1.21644819e+00 -1.47376251e+00 1.99257612e-01 -7.06314623e-01
-1.06796885e+00 1.20705771e+00 6.75320566e-01 2.64174957e-02
2.70134479e-01 4.04070675e-01 1.54826157e-02 -5.45631349e-01
-6.93264842e-01 -3.11826110e-01 4.88440096e-01 3.19379508e-01
3.88907045e-01 1.74063325e-01 -1.22029722e-01 5.27112842e-01
1.68701366e-01 2.17168123e-01 5.58263481e-01 1.31421959e+00
-1.43831059e-01 -6.31577671e-01 1.22333348e-01 5.10827482e-01
-2.38286495e-01 -3.68002579e-02 -4.35565174e-01 9.50647473e-01
-3.55883509e-01 4.52283859e-01 -2.16998950e-01 -1.99905276e-01
4.19058532e-01 -4.53070663e-02 6.70623839e-01 -7.79252470e-01
-8.46074224e-01 -1.89888731e-01 5.03330231e-01 -1.53165638e+00
-1.55741975e-01 -5.37934959e-01 -1.42056704e+00 4.82907832e-01
3.28621030e-01 -3.32675666e-01 8.73670101e-01 6.95852458e-01
3.00971627e-01 1.21389437e+00 2.50854820e-01 -9.58936393e-01
-3.74819845e-01 -6.71472430e-01 -3.10442686e-01 3.95422161e-01
5.20184994e-01 -6.92224503e-01 -2.52891004e-01 -2.22477615e-01] | [14.074535369873047, -3.382964611053467] |
80ce7398-b99a-4538-a431-1cddf97c9d39 | supervised-anomaly-detection-based-on-deep | 1904.06034 | null | http://arxiv.org/abs/1904.06034v1 | http://arxiv.org/pdf/1904.06034v1.pdf | Supervised Anomaly Detection based on Deep Autoregressive Density Estimators | We propose a supervised anomaly detection method based on neural density
estimators, where the negative log likelihood is used for the anomaly score.
Density estimators have been widely used for unsupervised anomaly detection. By
the recent advance of deep learning, the density estimation performance has
been greatly improved. However, the neural density estimators cannot exploit
anomaly label information, which would be valuable for improving the anomaly
detection performance. The proposed method effectively utilizes the anomaly
label information by training the neural density estimator so that the
likelihood of normal instances is maximized and the likelihood of anomalous
instances is lower than that of the normal instances. We employ an
autoregressive model for the neural density estimator, which enables us to
calculate the likelihood exactly. With the experiments using 16 datasets, we
demonstrate that the proposed method improves the anomaly detection performance
with a few labeled anomalous instances, and achieves better performance than
existing unsupervised and supervised anomaly detection methods. | ['Tomoharu Iwata', 'Yuki Yamanaka'] | 2019-04-12 | null | null | null | null | ['supervised-anomaly-detection'] | ['computer-vision'] | [-3.32521200e-01 -1.41998425e-01 -1.44744068e-01 -6.35672033e-01
-3.18118304e-01 1.37211353e-01 3.54435354e-01 9.31011140e-02
-2.34063447e-01 5.51571071e-01 -4.72865105e-02 -1.46757141e-01
5.48463501e-03 -9.69227374e-01 -2.76372552e-01 -9.32225764e-01
-1.63562790e-01 4.32723522e-01 2.17561632e-01 3.44433606e-01
3.04556370e-01 3.17577809e-01 -1.36922443e+00 -2.88062841e-01
1.11107790e+00 1.37963212e+00 -3.42274159e-01 1.47424906e-01
-7.31595278e-01 8.72834206e-01 -8.24341714e-01 -7.22986162e-02
6.52129576e-02 -4.85462546e-01 -2.55529046e-01 1.08398087e-01
1.31206036e-01 -9.76395190e-01 -3.45750362e-01 1.43044257e+00
1.41246960e-01 2.68689126e-01 1.03092563e+00 -1.44559443e+00
-5.44058204e-01 2.23331392e-01 -1.09173024e+00 4.70586777e-01
-6.41437471e-02 -2.62486219e-01 8.23414385e-01 -9.04345572e-01
-1.56766087e-01 1.17008698e+00 5.74729145e-01 4.92933869e-01
-7.42223382e-01 -9.36975837e-01 3.86115849e-01 3.27354938e-01
-1.57834244e+00 -2.42536217e-01 7.35683262e-01 -2.40315959e-01
7.58331835e-01 -3.99539918e-01 4.62090999e-01 8.92404556e-01
7.22807124e-02 9.97611821e-01 4.52549815e-01 -2.20776096e-01
4.16608304e-01 -6.47436455e-02 8.06677416e-02 6.32315755e-01
5.20117521e-01 1.60346374e-01 -1.70707017e-01 -5.40442228e-01
8.83015871e-01 7.04069912e-01 -2.18562242e-02 -3.17690849e-01
-5.32884359e-01 1.01048410e+00 2.48903617e-01 2.50723153e-01
-4.61557478e-01 -1.48052191e-02 6.34996355e-01 9.72399041e-02
7.85611093e-01 -1.66632190e-01 -2.20659986e-01 -4.13580894e-01
-7.43808508e-01 4.50807529e-05 7.75884867e-01 8.33574355e-01
7.59843111e-01 6.09342039e-01 -4.36528176e-02 1.19935596e+00
6.86629117e-01 6.18594587e-01 7.04061866e-01 -5.85799754e-01
1.60633922e-01 9.11203384e-01 -9.83220562e-02 -9.89550233e-01
-1.09427206e-01 -2.46118546e-01 -9.89097238e-01 2.52197906e-02
2.68970937e-01 -3.91210347e-01 -9.73342538e-01 1.52534342e+00
2.32441038e-01 7.82646179e-01 7.37095401e-02 7.37481713e-01
4.36365336e-01 7.49814391e-01 2.30694473e-01 -3.78314286e-01
9.50864255e-01 -4.35090333e-01 -9.65976954e-01 -1.06993668e-01
8.86561990e-01 -4.01243657e-01 8.81472230e-01 4.47293639e-01
-4.43762571e-01 -4.82551083e-02 -1.03051639e+00 6.18262112e-01
-1.34313390e-01 -9.70891714e-02 7.33364820e-01 6.86767042e-01
-4.04036641e-01 3.73423100e-01 -1.13235581e+00 -2.88021088e-01
8.25592756e-01 4.89224754e-02 -2.50013798e-01 -7.96011984e-02
-1.05253112e+00 4.77914304e-01 8.01238775e-01 -9.23989043e-02
-8.05061281e-01 -2.08786964e-01 -9.02732670e-01 3.38926315e-01
2.70944566e-01 -3.75528354e-03 1.32308757e+00 -8.34148765e-01
-1.20198071e+00 3.95832509e-01 -2.51085162e-01 -6.11427903e-01
2.90963352e-01 -4.71693784e-01 -8.40543747e-01 7.98184797e-02
1.98429804e-02 1.09926611e-01 9.45426583e-01 -7.87702024e-01
-9.24845457e-01 -3.33659172e-01 -5.88460684e-01 2.22591385e-02
-6.95851386e-01 -2.03439713e-01 -1.51034415e-01 -5.60383737e-01
4.45010096e-01 -4.06044573e-01 6.94841426e-03 -1.59255251e-01
-2.61024237e-01 -3.60515654e-01 1.49618387e+00 -4.32944894e-01
1.32623863e+00 -2.51590657e+00 -7.21351147e-01 6.00284159e-01
1.91433609e-01 3.08099598e-01 2.00685978e-01 -3.16777155e-02
-2.94379462e-02 -7.65647143e-02 -7.20095992e-01 -9.67099145e-02
-1.14398651e-01 5.33119798e-01 -5.79552829e-01 5.28227270e-01
4.21034992e-01 3.52739125e-01 -8.42579603e-01 -5.70242882e-01
9.94799659e-02 1.42259210e-01 -5.63774824e-01 6.18493617e-01
1.07118189e-02 2.10307345e-01 -6.75712943e-01 7.56605208e-01
9.66043532e-01 -1.37326509e-01 -1.72155246e-01 3.55177224e-01
3.65689725e-01 -6.69084489e-02 -1.29300165e+00 1.18768740e+00
-9.71272662e-02 3.23970586e-01 -3.59150767e-01 -1.31241107e+00
1.40691876e+00 2.89622813e-01 5.99030435e-01 -4.82296079e-01
-2.73210276e-02 3.17758024e-01 1.38463322e-02 -4.63917345e-01
9.85392630e-02 -2.58780688e-01 3.41730379e-02 6.07897639e-01
9.71220434e-02 4.67827588e-01 -6.57347739e-02 2.40849063e-01
1.03182364e+00 -1.22473352e-01 3.66318971e-01 9.90959108e-02
6.66887224e-01 -4.37693328e-01 1.03058028e+00 8.02446604e-01
-3.86475652e-01 3.91216874e-01 8.64136636e-01 -4.97931004e-01
-8.32211435e-01 -1.56221318e+00 -4.77594733e-01 9.08165216e-01
9.17197987e-02 -1.88585237e-01 -6.43659174e-01 -1.13224494e+00
6.39910251e-02 8.48386407e-01 -2.95510828e-01 -8.29328001e-01
-3.49450946e-01 -9.80943143e-01 6.48672879e-01 7.67668962e-01
8.93607497e-01 -1.11686635e+00 -4.65836562e-03 6.64307624e-02
8.73375311e-02 -7.49356329e-01 -9.97488201e-02 1.38294762e-02
-1.14444339e+00 -9.62910414e-01 -5.62392294e-01 -4.69389737e-01
9.63950753e-01 -1.05970286e-01 7.30969965e-01 1.57084897e-01
-7.33876750e-02 1.89833030e-01 -3.63346547e-01 -6.27409101e-01
-1.79008007e-01 -1.31895825e-01 5.60114622e-01 1.66868880e-01
1.25025177e+00 -7.46402919e-01 -4.07751948e-01 2.45340645e-01
-1.07597649e+00 -7.11465061e-01 5.97909391e-01 9.22186315e-01
5.48908949e-01 1.60571516e-01 1.18082774e+00 -1.08199215e+00
7.84202576e-01 -9.67557311e-01 -5.98277688e-01 -1.66971579e-01
-9.17695403e-01 1.00732096e-01 6.37966692e-01 -4.59372789e-01
-1.34544408e+00 -2.95113534e-01 -2.78089225e-01 -6.03946686e-01
-5.23409963e-01 5.89436293e-01 -2.53199875e-01 2.93326765e-01
4.13834095e-01 4.24579978e-01 6.36246502e-02 -6.84021235e-01
-1.55387685e-01 9.54250336e-01 4.60899353e-01 -6.26736403e-01
5.95026374e-01 2.29630381e-01 -8.69234130e-02 -6.59223795e-01
-9.49757338e-01 -5.41214883e-01 -4.52266186e-01 5.00152260e-03
4.87868309e-01 -7.59662449e-01 -4.30782199e-01 8.11564267e-01
-7.70169675e-01 2.96237588e-01 -2.79100418e-01 6.43190742e-01
-3.01882088e-01 7.40742266e-01 -5.46474576e-01 -1.25830853e+00
-2.11587951e-01 -7.07735479e-01 7.87446260e-01 5.36234319e-01
-6.71438081e-03 -1.18429363e+00 1.66745797e-01 -3.98313254e-01
4.79430676e-01 4.27023880e-02 9.50300038e-01 -1.49021220e+00
-1.86435029e-01 -6.04542077e-01 -4.32923675e-01 7.56245375e-01
3.43244731e-01 6.44188549e-04 -9.07178521e-01 -2.04657689e-01
1.64090857e-01 -1.43002614e-01 7.81013310e-01 2.50707269e-01
1.64356303e+00 -2.84447968e-01 -2.21132457e-01 5.74474096e-01
1.04343915e+00 3.81889731e-01 8.20811629e-01 2.20575720e-01
7.65386820e-01 5.58331013e-02 6.34194434e-01 7.80912161e-01
1.11983754e-02 -3.40441316e-02 4.38928157e-01 2.13159099e-01
7.93280184e-01 -4.13945228e-01 4.14357007e-01 8.24989319e-01
3.10079306e-01 6.03098236e-02 -1.11245692e+00 5.79348207e-01
-2.08994055e+00 -9.80993092e-01 2.46172249e-02 2.35547638e+00
5.60486078e-01 3.81029308e-01 1.51508868e-01 1.26100674e-01
9.75996077e-01 1.26242414e-01 -8.58590722e-01 -4.16697979e-01
2.59201258e-01 2.43003685e-02 7.30920136e-02 3.87169025e-03
-1.16500485e+00 8.11441481e-01 7.21632719e+00 8.43887568e-01
-8.68310511e-01 -1.69532791e-01 6.21625543e-01 -5.04569039e-02
1.74770299e-02 -4.74730954e-02 -6.92176759e-01 8.59177351e-01
9.63054657e-01 -2.45941505e-01 -4.11229320e-02 1.49248230e+00
5.25669940e-02 -1.35450199e-01 -8.55443597e-01 9.53490674e-01
-4.40864125e-03 -6.02436662e-01 3.11150819e-01 1.32485271e-01
5.04377544e-01 -1.63237035e-01 1.03303909e-01 8.17567825e-01
1.49227843e-01 -9.40650940e-01 -2.39101425e-01 7.25417674e-01
4.63571191e-01 -1.28618705e+00 1.28721416e+00 6.29978955e-01
-9.25118208e-01 -1.99609056e-01 -7.51201808e-01 -1.02723211e-01
8.39585066e-02 1.03652322e+00 -8.10096025e-01 1.04742005e-01
6.44827843e-01 9.04627562e-01 -3.41339648e-01 1.18664920e+00
-1.55630767e-01 1.05438221e+00 -5.73160708e-01 9.90630910e-02
2.33968168e-01 -6.06327832e-01 5.79549789e-01 8.98155451e-01
4.64303464e-01 -1.48627102e-01 3.83457214e-01 9.31006074e-01
-1.28630534e-01 3.40349108e-01 -8.15543354e-01 -2.25947186e-01
6.75161779e-01 1.11364186e+00 -4.23779279e-01 -4.01137829e-01
-5.71939409e-01 7.25931525e-01 3.72656077e-01 4.28379178e-01
-6.70965731e-01 -7.50700772e-01 6.09516561e-01 4.44375770e-03
2.19401732e-01 3.81973423e-02 -2.58428138e-02 -1.19939125e+00
-5.77627821e-03 -4.84053701e-01 7.09701180e-01 -4.08617020e-01
-1.90200341e+00 8.68002847e-02 1.15609102e-01 -1.44598114e+00
-6.55305207e-01 -5.60623109e-01 -1.33140147e+00 8.02266717e-01
-1.08578384e+00 -4.17799771e-01 -2.12860018e-01 4.36970025e-01
1.94461662e-02 -7.57775486e-01 8.46010566e-01 3.53336841e-01
-8.07884037e-01 6.99304640e-01 3.12508613e-01 5.84255993e-01
6.96747780e-01 -1.32164979e+00 2.22313732e-01 8.77418518e-01
2.17449050e-02 5.79881430e-01 2.72872478e-01 -7.89738238e-01
-4.75218773e-01 -1.02745664e+00 2.58915991e-01 -2.51460895e-02
6.02513313e-01 7.93917626e-02 -1.59252727e+00 6.80520356e-01
-3.65458280e-01 3.28529447e-01 9.85556841e-01 4.04926658e-01
-3.42941612e-01 -1.44395635e-01 -1.51395524e+00 3.29920173e-01
6.55074418e-01 -4.61675465e-01 -7.39144266e-01 6.93026856e-02
4.14808780e-01 -1.57441601e-01 -5.82912624e-01 5.74410856e-01
2.64378637e-01 -1.04910934e+00 4.84333813e-01 -7.24458754e-01
1.32732093e-01 -3.22132081e-01 8.70111510e-02 -1.18980706e+00
-1.48535013e-01 2.01773509e-01 -1.01967359e+00 1.32293057e+00
1.17977791e-01 -1.06888819e+00 9.51835454e-01 6.32025182e-01
1.05762593e-01 -7.53221452e-01 -1.10795105e+00 -6.75231695e-01
-1.47510111e-01 -4.96144414e-01 7.49697328e-01 1.00719154e+00
-1.04079634e-01 -6.31387681e-02 -3.84966344e-01 1.63495034e-01
4.99441832e-01 -3.87697488e-01 5.90666234e-01 -1.62236226e+00
-2.43581720e-02 -2.08337978e-01 -9.06449556e-01 -1.08353627e+00
3.62339228e-01 -5.50552130e-01 3.92959453e-03 -1.12405884e+00
2.60378420e-01 2.63389274e-02 -8.76984358e-01 5.01637876e-01
-3.99217963e-01 -1.87521160e-01 -6.07774317e-01 3.24020833e-01
-8.10357094e-01 9.15233433e-01 7.57740736e-01 1.37078837e-01
-6.25290424e-02 1.84259593e-01 -5.50958872e-01 1.22047520e+00
8.89437258e-01 -6.24261439e-01 -4.02366161e-01 2.59560570e-02
-1.24245793e-01 -3.88801217e-01 2.87069492e-02 -1.12376952e+00
7.59270117e-02 -5.50137013e-02 7.88504362e-01 -7.07151353e-01
-3.46649811e-02 -8.42476785e-01 -6.27491713e-01 2.18730822e-01
-7.28359297e-02 -1.24012865e-02 1.40045688e-01 1.07706392e+00
-4.95160699e-01 -4.78581429e-01 7.38778472e-01 1.47480950e-01
-7.94103980e-01 7.50465751e-01 -5.72754920e-01 1.09161027e-01
9.97792840e-01 -1.10903829e-01 -8.98763388e-02 -5.81443071e-01
-5.19074202e-01 4.23976600e-01 2.55749792e-01 3.48441601e-01
9.04340923e-01 -1.83710134e+00 -5.35096645e-01 5.16221821e-01
3.99323493e-01 2.85692662e-01 1.77329436e-01 7.30623484e-01
-4.69728768e-01 -1.79904848e-01 -1.41825229e-01 -9.10357833e-01
-6.75172389e-01 2.85879612e-01 4.46387649e-01 -1.94031492e-01
-7.19867408e-01 4.67954040e-01 3.22293490e-01 -3.90224546e-01
3.17355901e-01 1.98904738e-01 -3.20660681e-01 -4.17052865e-01
8.83132219e-01 3.72816086e-01 -2.92803019e-01 -3.75485331e-01
-3.32081109e-01 1.49311006e-01 -7.53085554e-01 1.55006468e-01
1.02339983e+00 8.57525617e-02 -1.79068267e-01 7.73920417e-01
1.11200476e+00 -3.06048840e-01 -1.14525378e+00 -4.71647084e-01
2.73307443e-01 -6.48410261e-01 1.85307890e-01 -3.35807949e-01
-1.13177752e+00 1.04126430e+00 7.41607130e-01 3.72576118e-01
1.10237575e+00 -2.68167049e-01 9.66675162e-01 6.53346419e-01
1.20424218e-01 -1.28002369e+00 2.37271503e-01 8.35574746e-01
1.56163216e-01 -1.47786129e+00 -7.65064405e-03 4.22124863e-02
-6.44788802e-01 1.04115939e+00 1.12476158e+00 -4.94864017e-01
9.05472159e-01 1.66504756e-01 1.37739694e-02 -1.29833236e-01
-4.36259955e-01 1.18708007e-01 2.58622825e-01 7.28390634e-01
3.78615975e-01 -1.35915235e-01 -8.16270411e-02 8.55820715e-01
2.70357996e-01 -3.61332238e-01 2.14814231e-01 8.34086478e-01
-8.54081571e-01 -8.97425056e-01 -3.13984275e-01 1.13740218e+00
-6.88332200e-01 3.47180925e-02 -2.47772038e-01 5.24257600e-01
-1.88177809e-01 6.40987933e-01 7.00570166e-01 -2.04549402e-01
1.31944552e-01 5.56816399e-01 -1.47809982e-01 -6.68486297e-01
2.82236487e-01 -1.30230501e-01 -2.87701219e-01 -3.63769174e-01
5.41851036e-02 -4.34277505e-01 -1.61575055e+00 -3.75902563e-01
-5.18234253e-01 4.05216575e-01 2.76320010e-01 1.24330854e+00
1.65986300e-01 3.78296942e-01 7.28544772e-01 -2.09115595e-01
-7.38850176e-01 -1.21495104e+00 -1.14993083e+00 5.54739177e-01
2.05698967e-01 -9.21921670e-01 -7.36111403e-01 -2.93535113e-01] | [7.607344627380371, 2.404714584350586] |
f2c38523-aee1-424e-807a-89c5f7cbbeab | mvpnet-multi-view-point-regression-networks | 1811.09410 | null | http://arxiv.org/abs/1811.09410v1 | http://arxiv.org/pdf/1811.09410v1.pdf | MVPNet: Multi-View Point Regression Networks for 3D Object Reconstruction from A Single Image | In this paper, we address the problem of reconstructing an object's surface
from a single image using generative networks. First, we represent a 3D surface
with an aggregation of dense point clouds from multiple views. Each point cloud
is embedded in a regular 2D grid aligned on an image plane of a viewpoint,
making the point cloud convolution-favored and ordered so as to fit into deep
network architectures. The point clouds can be easily triangulated by
exploiting connectivities of the 2D grids to form mesh-based surfaces. Second,
we propose an encoder-decoder network that generates such kind of multiple
view-dependent point clouds from a single image by regressing their 3D
coordinates and visibilities. We also introduce a novel geometric loss that is
able to interpret discrepancy over 3D surfaces as opposed to 2D projective
planes, resorting to the surface discretization on the constructed meshes. We
demonstrate that the multi-view point regression network outperforms
state-of-the-art methods with a significant improvement on challenging
datasets. | ['Yan Lu', 'Jinglu Wang', 'Bo Sun'] | 2018-11-23 | null | null | null | null | ['3d-object-reconstruction', '3d-object-reconstruction-from-a-single-image'] | ['computer-vision', 'computer-vision'] | [ 8.59057009e-02 3.66709709e-01 3.19785714e-01 -4.91166264e-01
-7.44177878e-01 -5.41342676e-01 6.55017436e-01 -5.21923602e-01
1.61702946e-01 4.31056440e-01 -1.73630133e-01 1.16379812e-01
2.52990425e-01 -1.23337424e+00 -1.54153299e+00 -4.61684495e-01
1.52153909e-01 1.12594032e+00 -1.55712262e-01 -1.03933424e-01
1.09776445e-01 9.14094508e-01 -1.78414667e+00 3.77221048e-01
5.76922536e-01 1.03619421e+00 2.00584397e-01 5.85529625e-01
-1.35513693e-01 2.91807540e-02 -1.03189588e-01 -6.00102544e-01
7.59461045e-01 -4.67414856e-02 -5.54320931e-01 4.42806840e-01
1.00731635e+00 -5.08752584e-01 -1.19994327e-01 1.05130446e+00
2.35048663e-02 -4.64033723e-01 7.62516916e-01 -1.22167790e+00
-8.69498909e-01 3.82584408e-02 -7.88292646e-01 -6.85394049e-01
1.60633937e-01 -6.55717179e-02 1.02024913e+00 -1.28837502e+00
1.02736771e+00 1.51454651e+00 7.23032713e-01 4.10988003e-01
-1.42457330e+00 -3.30252141e-01 1.18489489e-01 -2.52994895e-01
-1.49496448e+00 -1.51569843e-01 1.30693436e+00 -7.64586151e-01
7.75465250e-01 8.37026387e-02 9.53817785e-01 1.11556935e+00
2.64267594e-01 3.02263767e-01 9.56974447e-01 -1.49014637e-01
1.36868626e-01 -2.05329675e-02 -3.25103641e-01 8.58260393e-01
1.57714829e-01 -1.16846040e-01 -4.31414098e-01 -2.66247720e-01
1.52024484e+00 4.29771692e-01 -7.04315975e-02 -9.81826186e-01
-1.27070177e+00 9.24608350e-01 6.66778266e-01 -2.99946573e-02
-4.46522653e-01 2.45691523e-01 -1.11228995e-01 1.75542057e-01
9.12922859e-01 2.02074185e-01 -3.66627961e-01 4.60185498e-01
-7.31279433e-01 2.52399594e-01 6.71591401e-01 1.31675851e+00
1.21089339e+00 1.33234531e-01 5.09768069e-01 4.56503987e-01
6.47767603e-01 6.45737529e-01 -2.91007876e-01 -1.30468488e+00
5.84712029e-01 9.54857945e-01 1.39529914e-01 -1.10484028e+00
-1.08145751e-01 -2.97730029e-01 -1.23797023e+00 6.97558999e-01
6.94356337e-02 1.57305747e-01 -9.32032049e-01 1.47576880e+00
3.30156296e-01 2.71314532e-01 -1.19530879e-01 9.07357156e-01
6.25969350e-01 6.00247979e-01 -6.98650062e-01 1.57790273e-01
1.23072183e+00 -6.23875380e-01 -1.58139214e-01 6.14681169e-02
1.04520984e-01 -3.93297553e-01 7.19987214e-01 5.48659205e-01
-1.59934175e+00 -6.64124429e-01 -1.19232821e+00 -4.73944157e-01
-1.49849176e-01 6.01219013e-02 2.58938372e-01 -1.29876122e-01
-1.37819719e+00 7.67125547e-01 -1.08161688e+00 1.05157018e-01
5.89199066e-01 2.79218614e-01 -5.31648636e-01 1.51083708e-01
-6.00533485e-01 6.51828766e-01 -1.24151558e-01 1.42027125e-01
-8.96041572e-01 -9.04552400e-01 -9.09703612e-01 3.07733454e-02
2.24004705e-02 -1.48186195e+00 7.13604093e-01 -9.37760890e-01
-1.35873222e+00 1.29403603e+00 -1.94126129e-01 -1.90531179e-01
7.81521261e-01 -1.11878701e-01 4.74515408e-02 1.49723396e-01
1.31239429e-01 8.22439432e-01 9.73428071e-01 -2.00922298e+00
-3.55324388e-01 -8.46896827e-01 1.44277029e-02 3.19510162e-01
3.59751999e-01 -6.50453568e-01 -4.45505887e-01 -1.95132360e-01
6.79485083e-01 -9.19080436e-01 -2.63612628e-01 5.37030220e-01
-8.21647346e-01 -9.42126289e-03 8.11801195e-01 -5.87648749e-01
2.65694559e-01 -2.02468204e+00 6.99730515e-01 2.57703990e-01
6.97477698e-01 -4.07099575e-01 -1.36700959e-03 2.24272907e-01
-9.92247909e-02 2.79835820e-01 -2.95317709e-01 -8.32249701e-01
-3.12912576e-02 2.03894168e-01 -2.63805360e-01 6.76105440e-01
3.11574817e-01 8.11583161e-01 -6.29051626e-01 -1.86401345e-02
3.24729502e-01 8.74268889e-01 -6.78496301e-01 2.98346490e-01
-5.05361497e-01 6.84826851e-01 -4.43730712e-01 3.83656353e-01
1.09448326e+00 -5.99685073e-01 -1.89573780e-01 -4.61952060e-01
-2.42041796e-01 1.78442895e-01 -1.03630376e+00 2.30155230e+00
-6.10527396e-01 2.47475013e-01 8.82564262e-02 -7.56990790e-01
1.08420253e+00 3.16743582e-01 6.37549520e-01 -1.74456626e-01
-2.18912456e-02 3.79998207e-01 -4.51572239e-01 8.69978592e-03
1.30002245e-01 -1.50187239e-01 2.09054708e-01 1.74962550e-01
2.89094169e-02 -5.98944962e-01 -4.74739879e-01 -2.99466811e-02
6.11230612e-01 3.55023384e-01 -7.59624168e-02 -7.11091459e-02
3.51357102e-01 -1.59151316e-01 3.07496130e-01 2.15902880e-01
7.95378447e-01 1.29807603e+00 5.38754523e-01 -8.45898330e-01
-1.76072478e+00 -1.39497113e+00 -2.67588437e-01 6.65247217e-02
1.63168937e-01 -1.08497646e-02 -6.86170518e-01 -5.14845788e-01
2.92805225e-01 5.59329450e-01 -6.61647499e-01 2.71624386e-01
-6.15262926e-01 -5.79239577e-02 1.37905717e-01 3.40582371e-01
4.70416665e-01 -6.49611831e-01 -5.40186822e-01 -1.18765593e-01
1.26794428e-02 -1.00854158e+00 -2.18239859e-01 2.73698438e-02
-1.11341989e+00 -1.04644489e+00 -7.24044442e-01 -7.44363368e-01
9.64264810e-01 3.92594188e-01 1.47246814e+00 -7.31855109e-02
1.79591104e-01 3.78130466e-01 1.52202114e-01 -1.97985962e-01
-5.19858122e-01 -5.89488484e-02 1.10864438e-01 2.96899170e-01
1.18164793e-01 -1.32510996e+00 -5.88746905e-01 1.37026578e-01
-7.02272534e-01 6.51916921e-01 3.48937213e-01 7.18223691e-01
1.31950176e+00 -4.03925687e-01 2.61865631e-02 -8.23413193e-01
-1.22417495e-01 -6.87157989e-01 -9.26662028e-01 3.28831095e-03
-3.03475827e-01 -7.15884473e-03 5.38449049e-01 8.78682509e-02
-6.19694114e-01 3.79408062e-01 -1.00404449e-01 -1.26615655e+00
-2.35278279e-01 -8.55921023e-03 -2.58112133e-01 -3.86960693e-02
4.53588903e-01 1.56488866e-01 1.95711538e-01 -7.08850741e-01
6.06977880e-01 1.85769111e-01 3.12974423e-01 -4.41019565e-01
9.50668216e-01 8.74399424e-01 3.71249259e-01 -6.36321366e-01
-7.07930624e-01 -1.27829716e-01 -1.24858153e+00 -1.33314103e-01
1.22828293e+00 -1.19624150e+00 -6.75392151e-01 4.06346768e-01
-1.81723297e+00 3.61095816e-02 -3.05058658e-01 1.57689005e-01
-1.00276053e+00 1.28584951e-01 -4.61747110e-01 -3.91892940e-01
-1.96028084e-01 -1.23298979e+00 1.73182797e+00 -3.45109403e-01
8.42168257e-02 -9.55930769e-01 8.83375853e-02 1.39138669e-01
-1.14721864e-01 6.35681152e-01 8.98261368e-01 -1.57869995e-01
-1.23209321e+00 1.22396126e-01 -1.76282942e-01 4.27378684e-01
4.03998122e-02 2.23347113e-01 -1.01164269e+00 -2.41134748e-01
4.11459267e-01 -2.19238833e-01 5.82204938e-01 5.16213775e-01
1.26979470e+00 -3.63735139e-01 -3.57256234e-01 1.28866851e+00
1.79422450e+00 -6.70555234e-02 5.20612121e-01 9.30641592e-02
1.29949427e+00 5.42009413e-01 -1.71717644e-01 2.45708734e-01
5.39431989e-01 7.12871253e-01 1.02112651e+00 -1.70619816e-01
2.18049400e-02 -6.72793031e-01 -7.17408061e-02 9.05037224e-01
-3.77588779e-01 -1.21332519e-01 -8.96416008e-01 3.17856580e-01
-1.46553469e+00 -6.99506640e-01 -4.86293733e-01 2.13974500e+00
2.95530081e-01 -1.02117389e-01 -1.97129831e-01 -3.61071229e-01
6.34466052e-01 2.08314046e-01 -8.59330893e-01 -1.93531841e-01
-1.85260877e-01 -1.88779607e-02 3.70178044e-01 6.34088099e-01
-7.06986964e-01 6.54082835e-01 5.67558861e+00 1.92474455e-01
-1.18163562e+00 6.66608289e-02 7.64895320e-01 -8.57092906e-03
-9.47456241e-01 -1.42845094e-01 -7.10009575e-01 2.41098329e-01
5.12638032e-01 1.98093921e-01 4.67667729e-01 9.02319670e-01
-1.94399357e-01 4.67884660e-01 -1.62420213e+00 1.22337735e+00
2.10767016e-01 -1.74129403e+00 4.99297351e-01 4.52627838e-01
9.34957802e-01 4.65657234e-01 1.97865754e-01 -2.72928715e-01
4.14658993e-01 -1.07323515e+00 9.42084849e-01 7.28879392e-01
1.22600961e+00 -5.12163401e-01 1.77107841e-01 7.17685342e-01
-8.19578767e-01 5.97899318e-01 -4.96234089e-01 8.32449868e-02
3.19804043e-01 6.28944099e-01 -6.37621045e-01 8.83093297e-01
7.17641532e-01 1.07263446e+00 -6.26620352e-02 4.34208870e-01
1.34612069e-01 -5.88399917e-02 -5.05096853e-01 6.30085588e-01
1.88240603e-01 -8.69861484e-01 6.56250715e-01 4.36806709e-01
8.56265008e-01 7.41476491e-02 -5.11784740e-02 1.73015678e+00
-2.94909328e-01 -2.50077933e-01 -1.33276701e+00 5.18986642e-01
3.06724370e-01 1.23757792e+00 -4.94270653e-01 -2.77058482e-01
-4.90971923e-01 1.01912308e+00 5.67383647e-01 2.92709082e-01
-5.65637887e-01 2.81354070e-01 6.95471108e-01 4.46985036e-01
3.96588922e-01 -4.31385815e-01 -6.23436272e-01 -1.35411060e+00
3.53789926e-01 -5.35461187e-01 -4.58445609e-01 -1.21302688e+00
-1.61141181e+00 9.81406331e-01 -4.20766324e-02 -1.69696534e+00
-1.20901808e-01 -6.16887987e-01 -3.87427479e-01 1.28986669e+00
-1.31499505e+00 -1.27669919e+00 -3.77599746e-01 5.04998565e-01
4.13345784e-01 -6.98501617e-02 8.03344309e-01 2.57834028e-02
8.57343003e-02 -3.10363136e-02 1.48680937e-02 -5.73374107e-02
1.97748199e-01 -1.29393601e+00 1.04439437e+00 4.83935088e-01
3.96737754e-01 3.98631394e-01 1.22661151e-01 -6.84415221e-01
-1.65431452e+00 -1.22237623e+00 6.59144461e-01 -8.93268704e-01
1.70124635e-01 -6.72579825e-01 -1.09143722e+00 1.07167268e+00
1.53114498e-01 2.15149730e-01 1.95867211e-01 -1.38136491e-01
-4.55326587e-01 1.14723258e-02 -1.17310309e+00 4.07979816e-01
1.37483835e+00 -4.06218678e-01 -2.84697056e-01 3.86680901e-01
7.53195822e-01 -8.99776876e-01 -9.08676088e-01 2.08551347e-01
3.07795584e-01 -1.28515387e+00 1.14771724e+00 -5.65150201e-01
9.79813874e-01 -2.97495067e-01 -2.95868725e-01 -1.61194730e+00
-3.85337383e-01 -3.60174179e-01 5.61958971e-03 7.44456172e-01
2.11769983e-01 -5.36638379e-01 9.19602752e-01 2.34640345e-01
-4.92062926e-01 -9.36716795e-01 -1.01313806e+00 -3.81090939e-01
3.13933879e-01 -3.00812811e-01 8.97561014e-01 9.44688261e-01
-7.99416780e-01 4.15463120e-01 -2.87784636e-01 5.34011245e-01
8.74156654e-01 3.14938754e-01 8.80887389e-01 -1.75191164e+00
-1.00837380e-01 -2.96724528e-01 -5.36469698e-01 -1.46305788e+00
2.18558133e-01 -1.09561050e+00 -1.27109870e-01 -1.58486414e+00
-1.22829288e-01 -6.03631794e-01 3.96930397e-01 1.65922046e-02
4.51133668e-01 1.92811593e-01 1.72187880e-01 3.03212225e-01
-1.29508838e-01 6.58107817e-01 1.77239752e+00 -8.32687691e-02
1.10542722e-01 -1.33357141e-02 -4.40810382e-01 1.08700860e+00
3.14496338e-01 -2.94927150e-01 -3.36443871e-01 -1.07957220e+00
5.53196132e-01 5.86971819e-01 7.85212934e-01 -7.28482187e-01
5.45066223e-02 -1.56182488e-02 6.56624436e-01 -1.04922605e+00
8.70207131e-01 -1.14870322e+00 7.92543530e-01 -4.17964272e-02
-1.05441578e-01 2.80679077e-01 -2.22909048e-01 7.53615439e-01
-1.30003989e-01 1.77299529e-01 6.03555083e-01 -4.18564677e-01
2.59552617e-02 8.15974593e-01 3.75520110e-01 -2.12684557e-01
9.57439482e-01 -4.05882061e-01 1.35019153e-01 -1.92523494e-01
-7.94195771e-01 -3.72341339e-04 1.10155129e+00 4.38399136e-01
8.35841060e-01 -1.75460839e+00 -8.84488463e-01 8.10218096e-01
-5.42296171e-02 1.07606530e+00 2.29048744e-01 3.18922549e-01
-8.59900534e-01 2.30559856e-01 -3.52842689e-01 -1.24437153e+00
-7.94588208e-01 4.47024345e-01 6.32483900e-01 4.84750140e-03
-1.16928923e+00 7.72227526e-01 6.52980804e-01 -6.24583900e-01
-7.48644676e-03 -5.69807172e-01 2.08348528e-01 -4.02194768e-01
1.08235683e-02 1.19202800e-01 4.45336662e-02 -7.75008142e-01
3.18712518e-02 1.09713495e+00 2.34084159e-01 -1.60443485e-01
1.63969672e+00 7.24914223e-02 -3.08293998e-01 7.80205488e-01
1.52157795e+00 -8.23477209e-02 -1.67410982e+00 -1.21982425e-01
-5.70797861e-01 -5.19205213e-01 -9.27398056e-02 -2.95584232e-01
-1.32834685e+00 1.11720037e+00 1.98833883e-01 3.08227897e-01
5.94956040e-01 7.40384236e-02 7.59782791e-01 2.33896360e-01
8.03172708e-01 -3.95953178e-01 -1.47211015e-01 4.77555275e-01
1.34064186e+00 -1.20096636e+00 -1.90228716e-01 -6.54772401e-01
-3.61915708e-01 1.30762160e+00 4.56381857e-01 -7.77064979e-01
8.83355677e-01 1.19911246e-01 -2.53603429e-01 -6.68269694e-01
-6.50116801e-01 4.07776505e-01 5.73093712e-01 5.94968081e-01
8.93493518e-02 8.45229626e-02 4.77240562e-01 2.17626348e-01
-4.43751186e-01 -9.98014882e-02 3.42931300e-01 2.64207482e-01
-1.44790873e-01 -8.88047755e-01 -3.03144157e-01 4.28348213e-01
3.92638408e-02 9.19993445e-02 -3.62317592e-01 6.46128535e-01
3.02404433e-01 9.67433304e-02 7.67836213e-01 -4.14317191e-01
4.71754849e-01 -2.10519936e-02 6.55844748e-01 -8.43673706e-01
-1.20288536e-01 1.05130106e-01 -2.95268059e-01 -5.29846191e-01
-3.30468446e-01 -6.62098527e-01 -1.02651083e+00 -3.66205722e-01
1.90749139e-01 -1.73814982e-01 7.87351549e-01 6.30819857e-01
7.78364480e-01 2.57580340e-01 1.01583683e+00 -1.52670348e+00
-4.41158265e-01 -5.78092813e-01 -7.03507841e-01 4.71664727e-01
6.12197816e-01 -7.79327095e-01 -5.25878668e-01 1.86426193e-01] | [8.593944549560547, -3.491241931915283] |
07c288e8-ae89-47bf-9e86-5d8aabc11162 | online-hybrid-ctc-attention-end-to-end | 2307.02351 | null | https://arxiv.org/abs/2307.02351v1 | https://arxiv.org/pdf/2307.02351v1.pdf | Online Hybrid CTC/Attention End-to-End Automatic Speech Recognition Architecture | Recently, there has been increasing progress in end-to-end automatic speech recognition (ASR) architecture, which transcribes speech to text without any pre-trained alignments. One popular end-to-end approach is the hybrid Connectionist Temporal Classification (CTC) and attention (CTC/attention) based ASR architecture. However, how to deploy hybrid CTC/attention systems for online speech recognition is still a non-trivial problem. This article describes our proposed online hybrid CTC/attention end-to-end ASR architecture, which replaces all the offline components of conventional CTC/attention ASR architecture with their corresponding streaming components. Firstly, we propose stable monotonic chunk-wise attention (sMoChA) to stream the conventional global attention, and further propose monotonic truncated attention (MTA) to simplify sMoChA and solve the training-and-decoding mismatch problem of sMoChA. Secondly, we propose truncated CTC (T-CTC) prefix score to stream CTC prefix score calculation. Thirdly, we design dynamic waiting joint decoding (DWJD) algorithm to dynamically collect the predictions of CTC and attention in an online manner. Finally, we use latency-controlled bidirectional long short-term memory (LC-BLSTM) to stream the widely-used offline bidirectional encoder network. Experiments with LibriSpeech English and HKUST Mandarin tasks demonstrate that, compared with the offline CTC/attention model, our proposed online CTC/attention model improves the real time factor in human-computer interaction services and maintains its performance with moderate degradation. To the best of our knowledge, this is the first work to provide the full-stack online solution for CTC/attention end-to-end ASR architecture. | ['Yonghong Yan', 'Pengyuan Zhang', 'Gaofeng Cheng', 'Haoran Miao'] | 2023-07-05 | null | null | null | null | ['speech-recognition', 'automatic-speech-recognition'] | ['speech', 'speech'] | [ 2.12226123e-01 -7.48948231e-02 -1.28001407e-01 -3.05691749e-01
-1.26949215e+00 -2.97029316e-01 4.47676718e-01 -2.78948337e-01
-5.40575385e-01 4.05443907e-01 4.32692498e-01 -9.41752076e-01
3.53330165e-01 -2.93338802e-02 -6.12212360e-01 -5.49575269e-01
2.71661401e-01 4.93166685e-01 2.53228068e-01 -3.10256541e-01
6.43545985e-02 3.15071911e-01 -1.16076684e+00 4.82046634e-01
7.16122210e-01 9.74689007e-01 6.79472804e-01 1.33697712e+00
-1.69207871e-01 1.02763641e+00 -7.16794729e-01 -4.09562111e-01
-6.37014359e-02 -5.71453393e-01 -9.32408988e-01 -1.80548951e-01
-8.49789008e-02 -4.76151556e-01 -6.57396197e-01 5.72338641e-01
8.88655186e-01 2.86504507e-01 1.17907383e-01 -7.62249529e-01
-7.70571172e-01 8.15087676e-01 -4.46902663e-01 7.48185277e-01
3.50616246e-01 1.90453753e-01 1.01162088e+00 -1.09315574e+00
7.36453235e-02 1.15914941e+00 2.83007503e-01 7.93394208e-01
-6.22999310e-01 -5.86719930e-01 4.49646950e-01 6.16436005e-01
-1.31720841e+00 -1.12199461e+00 5.70889354e-01 4.52454202e-02
1.81260097e+00 5.42322576e-01 5.18126547e-01 1.13806641e+00
-8.04832950e-02 1.17695940e+00 5.66288948e-01 -5.20578444e-01
1.42267674e-01 -2.97965407e-01 2.39293471e-01 3.79841536e-01
-7.32780218e-01 5.45043051e-02 -8.70127857e-01 1.02046490e-01
5.00333786e-01 -2.31683720e-02 -3.41127872e-01 7.54212022e-01
-1.23881650e+00 3.36909831e-01 9.44475159e-02 3.70854765e-01
-5.43189108e-01 1.78789392e-01 5.50214767e-01 5.71058810e-01
6.14628315e-01 -2.61880219e-01 -6.57881081e-01 -7.60936379e-01
-1.03445685e+00 -4.78891760e-01 6.61140144e-01 1.20999277e+00
2.10421592e-01 5.16305268e-01 -3.54474396e-01 1.20200872e+00
2.30342984e-01 5.70878327e-01 1.11197555e+00 -4.25631464e-01
8.96405637e-01 -6.13581426e-02 -2.09459379e-01 -1.87078327e-01
1.59799099e-01 -5.51810443e-01 -8.68502855e-01 -4.12470698e-01
-5.80160730e-02 -2.36444131e-01 -1.09620023e+00 1.36193001e+00
6.53037503e-02 5.48711598e-01 2.73885071e-01 1.14702415e+00
6.08429849e-01 1.12692821e+00 -1.52224749e-01 -4.68921602e-01
1.09783018e+00 -1.43658876e+00 -9.13670003e-01 -2.90661424e-01
7.50868320e-01 -8.93040478e-01 1.14447343e+00 2.34130904e-01
-1.44346452e+00 -4.62189794e-01 -9.09768462e-01 -3.16171527e-01
3.72885801e-02 4.07352328e-01 -2.84612598e-03 4.03785557e-01
-1.24595249e+00 2.58852124e-01 -1.16433465e+00 -2.79594988e-01
-1.59018654e-02 5.57508767e-01 -3.86205837e-02 8.09968486e-02
-1.14702356e+00 7.15364873e-01 8.52226019e-02 5.69066882e-01
-1.11333990e+00 -4.11885291e-01 -5.62450528e-01 2.22428024e-01
3.48839223e-01 -3.80568117e-01 1.83000386e+00 -1.29025161e+00
-2.33979106e+00 3.17073762e-01 -9.40765679e-01 -7.29413986e-01
7.59303719e-02 -3.65904629e-01 -4.60459590e-01 -5.51867411e-02
-5.82748115e-01 3.66604865e-01 8.51844490e-01 -6.67496204e-01
-6.44922316e-01 -2.18121946e-01 -2.71872222e-01 4.37737137e-01
-5.06401896e-01 6.55559361e-01 -5.23054779e-01 -7.27914393e-01
7.29930624e-02 -7.60523617e-01 -5.29030859e-02 -6.98963702e-01
-1.93013251e-01 -3.92697453e-01 8.01650524e-01 -1.22322023e+00
1.76975012e+00 -2.20380473e+00 2.96927810e-01 -2.52491415e-01
-1.52711421e-01 7.81570077e-01 -2.73297042e-01 4.69475687e-01
-1.31293312e-01 -1.71497129e-02 -1.89828455e-01 -8.44627857e-01
-1.65412411e-01 2.38218844e-01 -6.59114957e-01 2.46844009e-01
1.67703480e-01 9.81822729e-01 -7.65903115e-01 -2.95330256e-01
2.13192463e-01 5.50084651e-01 -3.46840948e-01 6.16017520e-01
-3.59946936e-02 3.43919426e-01 -1.58920109e-01 6.17511749e-01
1.69479713e-01 7.79913589e-02 -1.07633388e-02 2.59582251e-01
-2.70388812e-01 1.06651354e+00 -6.18196011e-01 1.64541459e+00
-8.45800340e-01 7.75034964e-01 2.25951925e-01 -8.08909118e-01
9.80110049e-01 9.23805296e-01 -5.66512719e-02 -8.13318491e-01
7.76603371e-02 5.20907044e-01 8.86085033e-02 -3.34073186e-01
7.26713181e-01 1.56586692e-02 3.73207480e-01 4.21455622e-01
2.20354330e-02 3.63323152e-01 -3.91372830e-01 1.61223501e-01
1.31939375e+00 -1.07244387e-01 1.25138491e-01 3.43956262e-01
8.53613794e-01 -3.50278288e-01 4.85666484e-01 3.18287075e-01
-2.70737141e-01 8.27093244e-01 3.50012369e-02 -1.20680004e-01
-1.21716380e+00 -7.05403805e-01 6.20514631e-01 1.42244160e+00
-3.77052903e-01 -3.23894948e-01 -7.60869861e-01 -5.03653169e-01
-6.19839430e-01 7.35790551e-01 1.17121659e-01 1.10167257e-01
-1.04427505e+00 -2.92002529e-01 6.84147716e-01 6.89180493e-01
2.59795278e-01 -1.29419374e+00 -1.95444047e-01 5.12237668e-01
-4.39128280e-01 -1.13782132e+00 -1.17567825e+00 2.33957276e-01
-7.68009067e-01 -2.04702616e-01 -1.03911853e+00 -9.55199838e-01
3.84820521e-01 5.52701652e-01 6.45211935e-01 4.32538092e-02
2.96432048e-01 3.91270928e-02 -7.61021435e-01 -4.93372083e-02
-3.74113142e-01 5.34585655e-01 1.46472640e-02 3.23030919e-01
3.76592636e-01 -5.54290771e-01 -5.64318120e-01 4.07259345e-01
-4.58402276e-01 1.27422780e-01 6.92136943e-01 8.94033670e-01
5.35983562e-01 -6.52351260e-01 8.65487754e-01 -3.09855044e-01
6.66165709e-01 -4.19761240e-01 -3.63707513e-01 3.60655427e-01
-3.59680921e-01 -2.18527988e-02 1.07977617e+00 -5.83581388e-01
-1.01560187e+00 -7.67196119e-02 -7.31771946e-01 -9.37287092e-01
6.79405406e-02 4.99484897e-01 1.46865705e-02 3.58244270e-01
5.10786399e-02 6.75745964e-01 -1.97749883e-01 -6.80564404e-01
2.98306763e-01 1.67269087e+00 5.19951582e-01 -1.78190678e-01
3.34639907e-01 1.73844036e-03 -8.30592930e-01 -9.68395233e-01
-5.81661999e-01 -8.08796883e-01 -3.09830517e-01 -8.77512842e-02
7.60717154e-01 -9.21106458e-01 -7.09700704e-01 4.75810677e-01
-1.57258093e+00 -8.48631918e-01 -5.67472801e-02 6.91788256e-01
-5.14678359e-01 4.20515895e-01 -9.48409379e-01 -1.30744946e+00
-9.67944801e-01 -1.26237965e+00 1.15376258e+00 -1.90377817e-01
-1.12044767e-01 -6.41427457e-01 -1.22947916e-01 5.45463324e-01
6.28890276e-01 -7.89817154e-01 1.92405924e-01 -7.15251505e-01
-6.30252659e-01 3.36907990e-02 -1.31416202e-01 5.85003734e-01
-2.31023192e-01 -1.68328211e-01 -1.06227589e+00 -6.30091727e-01
1.31117225e-01 -1.52890950e-01 7.17193902e-01 3.42479527e-01
1.18144774e+00 -6.19782865e-01 -2.56770886e-02 6.77514672e-01
8.98584068e-01 7.02777624e-01 7.21659184e-01 4.77871969e-02
7.19529986e-01 3.78862113e-01 6.25725269e-01 3.19212943e-01
3.01655114e-01 9.02486861e-01 -1.29980519e-02 2.21610330e-02
-3.96389663e-01 -3.25776011e-01 1.19083452e+00 2.15104961e+00
-7.15442374e-02 -5.44277310e-01 -8.35438192e-01 7.67264485e-01
-1.86502612e+00 -9.35740292e-01 -2.21156459e-02 2.22671127e+00
7.78523088e-01 9.91663933e-02 4.39483076e-02 1.66135535e-01
8.45580518e-01 1.68533221e-01 -4.21343386e-01 -9.33315814e-01
8.01876262e-02 4.18598235e-01 5.21618426e-01 8.88566136e-01
-4.91790056e-01 1.10689390e+00 5.04255104e+00 1.34342086e+00
-1.38933539e+00 7.50651062e-01 5.73930740e-01 -3.82443845e-01
-2.22436920e-01 1.55590743e-01 -9.66470957e-01 5.18952668e-01
1.88181412e+00 -1.30124509e-01 8.18857491e-01 7.04834700e-01
5.83994508e-01 4.34084952e-01 -9.87535655e-01 1.08976960e+00
1.98231533e-01 -9.92079318e-01 -4.04177383e-02 -9.14333537e-02
2.03952342e-01 3.83205175e-01 -1.20631801e-02 4.01683122e-01
4.55637388e-02 -8.52885067e-01 8.96149874e-01 6.74448758e-02
1.16751838e+00 -7.48280644e-01 7.08506942e-01 4.98967826e-01
-1.52579677e+00 -2.17009529e-01 -2.12988943e-01 -1.94202177e-02
6.03389204e-01 1.58894628e-01 -9.84413981e-01 5.23139119e-01
4.41467732e-01 5.68458498e-01 -5.01819775e-02 7.49341190e-01
-2.52114564e-01 1.29037905e+00 -2.74370551e-01 -1.97861865e-01
4.15554792e-01 6.61039054e-02 7.17685163e-01 1.50347400e+00
4.50067729e-01 2.83060759e-01 -2.05186069e-01 1.72181591e-01
-1.28664017e-01 7.54549503e-02 -9.22640190e-02 -1.21291943e-01
6.20934904e-01 8.32649529e-01 -2.60994762e-01 -4.31380332e-01
-4.73358423e-01 1.34463012e+00 5.04444182e-01 5.11848867e-01
-9.72993195e-01 -5.75038731e-01 4.43687171e-01 -3.46166231e-02
5.07873178e-01 -5.51798284e-01 -1.48926780e-01 -1.25142741e+00
1.95581630e-01 -1.00814188e+00 1.87229872e-01 -9.46231723e-01
-9.28389072e-01 1.02873111e+00 -5.04319847e-01 -1.06466186e+00
-1.65153712e-01 -1.76398739e-01 -5.91364443e-01 1.06932461e+00
-1.78554881e+00 -1.02106869e+00 9.26170200e-02 6.09436989e-01
1.50084543e+00 -3.20761293e-01 6.66600466e-01 6.26632214e-01
-9.53302979e-01 9.05844688e-01 1.04187936e-01 1.27220035e-01
5.21659553e-01 -9.79359925e-01 8.03881586e-01 1.21087468e+00
1.71851620e-01 5.41753173e-01 2.69262135e-01 -5.08271456e-01
-1.56673062e+00 -1.11016667e+00 1.34383845e+00 -6.78640828e-02
6.48698986e-01 -4.63955164e-01 -1.03429747e+00 8.92791390e-01
6.01745129e-01 -1.94789112e-01 3.88549358e-01 -1.16237774e-01
-1.89900458e-01 -3.45039904e-01 -4.57174271e-01 7.06282914e-01
1.10820246e+00 -9.55256999e-01 -5.94607174e-01 1.49260521e-01
1.36174250e+00 -4.43360180e-01 -6.73510075e-01 1.16779111e-01
4.58324641e-01 -5.31846523e-01 5.95917642e-01 -2.46677011e-01
1.70391127e-01 -2.59089619e-01 -2.74749070e-01 -1.17900121e+00
-2.03599870e-01 -1.44255328e+00 -4.91036773e-01 1.30926085e+00
6.04311049e-01 -6.98859990e-01 6.48069978e-01 2.67726630e-01
-1.00948000e+00 -9.08343792e-01 -1.28406549e+00 -7.84435451e-01
-2.82313436e-01 -6.84963226e-01 6.15344286e-01 5.31403959e-01
9.94084254e-02 5.19294620e-01 -5.16780853e-01 2.59449989e-01
7.26829544e-02 -2.82846957e-01 4.62149948e-01 -2.61702269e-01
-5.98122656e-01 -3.11909050e-01 -6.22147974e-03 -1.87079799e+00
8.84860903e-02 -9.81955290e-01 3.19748998e-01 -1.31985974e+00
-8.28931034e-02 -2.48771131e-01 -4.24304664e-01 4.27998841e-01
-1.86284445e-02 -2.54170299e-01 2.36062989e-01 3.65848631e-01
-6.55733168e-01 9.46028113e-01 8.69504869e-01 1.29216328e-01
-4.20490205e-01 -1.22962683e-01 -1.23955473e-01 1.52115792e-01
9.07676578e-01 -5.15190601e-01 -3.52787048e-01 -9.29102957e-01
-2.57489443e-01 4.98935461e-01 -1.00549258e-01 -7.57184505e-01
8.43255401e-01 2.67423857e-02 -3.02260280e-01 -8.53377461e-01
4.91726339e-01 -5.31661987e-01 -7.29631931e-02 2.99849510e-01
-4.99460757e-01 4.42823231e-01 6.24420866e-02 4.63679850e-01
-4.53544468e-01 1.50544150e-02 6.58347189e-01 2.15966716e-01
-4.35934037e-01 3.60117555e-01 -6.95201278e-01 -6.27554283e-02
7.02073276e-01 -1.51956812e-01 -2.45669663e-01 -4.12463337e-01
-7.38018572e-01 2.58864701e-01 -1.82208613e-01 4.02611107e-01
9.63636160e-01 -1.05411851e+00 -8.66560459e-01 3.41722012e-01
-2.08066002e-01 4.40738071e-03 3.21124882e-01 1.16385126e+00
-3.98035407e-01 8.47901404e-01 5.28430879e-01 -4.88902241e-01
-1.58418500e+00 3.58142972e-01 2.53822088e-01 -2.00096667e-01
-5.75239539e-01 1.05308735e+00 -1.73335075e-01 -1.20337009e-01
5.01914680e-01 -2.48174459e-01 1.01888895e-01 -3.73245746e-01
5.53797185e-01 4.65934247e-01 3.34608763e-01 -5.74566483e-01
-2.95677453e-01 1.86208710e-01 -3.66361856e-01 -6.05393469e-01
1.17750227e+00 -4.72419888e-01 2.68932153e-02 5.94852924e-01
1.10982883e+00 -1.68097213e-01 -9.88056064e-01 -1.68060839e-01
-1.22004308e-01 -2.50349373e-01 2.92738974e-01 -6.78758323e-01
-1.03898382e+00 1.25460351e+00 3.08803350e-01 6.87147826e-02
1.24335992e+00 -1.47033349e-01 1.62628591e+00 3.50661576e-01
1.27394214e-01 -1.11433613e+00 -8.04518834e-02 9.15985823e-01
1.04698241e+00 -7.52453148e-01 -6.72501326e-01 -1.74546391e-02
-7.05538988e-01 1.04943573e+00 5.64778686e-01 1.73742279e-01
4.17807400e-01 4.03698862e-01 6.90470114e-02 4.79333907e-01
-1.41347694e+00 -2.63666391e-01 3.52116935e-02 2.19320267e-01
6.39004886e-01 5.10579124e-02 -4.07270938e-01 6.56252146e-01
-1.88043743e-01 5.66555783e-02 3.02540421e-01 7.98985779e-01
-3.91330242e-01 -1.19728005e+00 -2.03111485e-01 2.62777686e-01
-5.97825348e-01 -5.43891728e-01 -2.41824582e-01 -5.24567254e-02
-4.38170850e-01 1.21322799e+00 9.09063369e-02 -6.31526589e-01
2.83774287e-01 2.52025157e-01 7.22204372e-02 -6.18197858e-01
-1.06247199e+00 4.43985075e-01 2.02205479e-01 -3.84910583e-01
5.98982461e-02 -4.95947480e-01 -1.34782934e+00 -3.26109797e-01
-6.10448360e-01 2.50378639e-01 8.85473847e-01 9.72336650e-01
7.32018888e-01 7.30504096e-01 8.39951754e-01 -7.73710370e-01
-6.89445019e-01 -1.32534671e+00 -5.75088076e-02 -8.54043216e-02
6.61831200e-01 2.20399890e-02 -5.47659159e-01 4.83646207e-02] | [14.489850997924805, 6.827888011932373] |
a40c618b-600a-4568-8948-14a6e37a4419 | semantic-graph-parsing-with-recurrent-neural | 1910.00051 | null | https://arxiv.org/abs/1910.00051v2 | https://arxiv.org/pdf/1910.00051v2.pdf | Semantic Graph Parsing with Recurrent Neural Network DAG Grammars | Semantic parses are directed acyclic graphs (DAGs), so semantic parsing should be modeled as graph prediction. But predicting graphs presents difficult technical challenges, so it is simpler and more common to predict the linearized graphs found in semantic parsing datasets using well-understood sequence models. The cost of this simplicity is that the predicted strings may not be well-formed graphs. We present recurrent neural network DAG grammars, a graph-aware sequence model that ensures only well-formed graphs while sidestepping many difficulties in graph prediction. We test our model on the Parallel Meaning Bank---a multilingual semantic graphbank. Our approach yields competitive results in English and establishes the first results for German, Italian and Dutch. | ['Sorcha Gilroy', 'Federico Fancellu', 'Adam Lopez', 'Mirella Lapata'] | 2019-09-30 | semantic-graph-parsing-with-recurrent-neural-1 | https://aclanthology.org/D19-1278 | https://aclanthology.org/D19-1278.pdf | ijcnlp-2019-11 | ['drs-parsing'] | ['natural-language-processing'] | [ 3.20662707e-01 8.60984147e-01 -1.97480112e-01 -4.39896584e-01
-5.51727295e-01 -9.43709254e-01 1.37664810e-01 1.98688731e-01
3.07534300e-02 8.62392366e-01 3.14458728e-01 -9.30004537e-01
1.11288972e-01 -1.16031754e+00 -8.15490603e-01 -2.33264551e-01
-3.28994125e-01 7.71557152e-01 5.23400068e-01 -3.15557897e-01
-6.61800653e-02 2.94230402e-01 -1.13119102e+00 4.34700131e-01
6.72542810e-01 1.31521910e-01 5.46161413e-01 9.86521006e-01
-8.21538091e-01 1.04554319e+00 -3.93647760e-01 -1.02939594e+00
-1.15292549e-01 -7.62828290e-01 -1.24121380e+00 -2.31744707e-01
2.11791053e-01 3.21141660e-01 -2.53244460e-01 1.13907683e+00
2.54641492e-02 -2.33861208e-01 3.53769630e-01 -1.20785534e+00
-7.04080939e-01 1.05258405e+00 -2.05465123e-01 5.33285178e-02
7.05163002e-01 -4.88540173e-01 1.83735633e+00 -2.08096951e-01
1.18897057e+00 1.52220321e+00 7.61799872e-01 7.13080943e-01
-9.60780025e-01 -3.00893903e-01 4.13653851e-01 2.05151856e-01
-8.73839736e-01 -5.40398099e-02 6.99956179e-01 -1.86771259e-01
1.77776194e+00 1.82083577e-01 7.67842472e-01 9.06000793e-01
5.46375334e-01 7.03385472e-01 6.87336266e-01 -6.76664889e-01
7.43032843e-02 -4.45304245e-01 4.72285151e-01 1.29097569e+00
3.64716381e-01 -1.03903390e-01 -4.60747302e-01 -7.24486038e-02
5.94838262e-01 -5.59998393e-01 -1.00498693e-02 -3.71849120e-01
-7.12697744e-01 1.12882721e+00 2.95186937e-01 3.22434545e-01
-9.06405523e-02 3.10194284e-01 6.66535497e-01 4.14449573e-01
3.98322284e-01 1.33042753e-01 -7.63127685e-01 -1.46121129e-01
-2.12708637e-01 1.83319673e-01 1.33017075e+00 1.15846515e+00
7.86374629e-01 1.71530306e-01 6.95984840e-01 7.58207083e-01
2.13363230e-01 4.64412123e-02 4.19345230e-01 -6.61030114e-01
6.77282453e-01 7.25424647e-01 -3.40764642e-01 -1.03779852e+00
-7.24858701e-01 -1.61477923e-03 -3.76129001e-01 -1.70817301e-01
7.17449069e-01 -1.60584450e-02 -9.02242661e-01 1.84812689e+00
3.58083583e-02 -1.50577262e-01 3.70328218e-01 6.05808973e-01
8.86578023e-01 7.09177673e-01 5.33111453e-01 -1.67168349e-01
1.33971453e+00 -8.76252592e-01 -7.01927423e-01 -7.03952670e-01
1.19745469e+00 -4.10083979e-01 9.77630198e-01 1.86437532e-01
-9.85963166e-01 -1.44028723e-01 -9.50631440e-01 -1.93180323e-01
-5.07471621e-01 -5.96712112e-01 1.18373346e+00 7.54982233e-01
-1.35093832e+00 7.78951526e-01 -8.34931433e-01 -7.42577076e-01
-2.99684014e-02 2.44420588e-01 -6.84965372e-01 6.23430014e-02
-1.16634071e+00 9.99330282e-01 9.07045245e-01 -1.19102679e-01
-1.15178108e-01 -1.47272602e-01 -1.40720952e+00 1.64524615e-01
2.90188313e-01 -7.10367441e-01 1.27992129e+00 -1.20795226e+00
-1.11804724e+00 1.19900525e+00 -5.02810240e-01 -6.79555595e-01
2.97523066e-02 2.25709267e-02 -5.03455281e-01 4.60547470e-02
8.91357958e-02 5.61014950e-01 2.24169731e-01 -1.09203506e+00
-4.76994723e-01 -5.42817473e-01 -1.49519229e-02 1.68604508e-01
1.87583134e-01 2.13899776e-01 -2.74946421e-01 -5.54769337e-01
4.92761493e-01 -9.97761309e-01 -4.63973582e-01 -6.72219455e-01
-5.71328402e-01 -4.40426648e-01 4.17541981e-01 -1.09952772e+00
1.24809277e+00 -1.65403867e+00 6.72421455e-02 1.80530891e-01
2.00683773e-01 -8.68030936e-02 -2.56886959e-01 7.94134796e-01
-4.83522266e-01 3.76729071e-01 -2.09032506e-01 2.48818219e-01
6.09998778e-02 8.84171069e-01 -2.18914747e-01 4.96519990e-02
4.43110913e-01 1.32037866e+00 -9.97765839e-01 -4.97833252e-01
-1.03127724e-02 -1.00167669e-01 -6.01066709e-01 7.86306858e-02
-7.20771849e-01 -4.51126546e-02 -4.88622874e-01 5.60846686e-01
3.54690194e-01 -5.73967874e-01 1.20755816e+00 2.44528562e-01
3.19133192e-01 7.79438436e-01 -9.59646523e-01 1.81931090e+00
-2.10885838e-01 4.24884170e-01 -2.35562429e-01 -1.48321545e+00
1.00647032e+00 8.20369422e-02 2.18287986e-02 -7.90970266e-01
-1.89860851e-01 2.79650927e-01 1.78869158e-01 -4.06345487e-01
3.68045628e-01 -4.07294124e-01 -3.86572033e-01 2.37294883e-01
2.51455903e-01 -2.12996706e-01 3.98275018e-01 4.81378496e-01
1.23344946e+00 6.07165217e-01 6.91072822e-01 -3.22447985e-01
2.18473807e-01 4.29349959e-01 7.19703555e-01 5.71271956e-01
4.19390500e-02 3.46895367e-01 1.16703773e+00 -7.80811548e-01
-1.04203188e+00 -1.03845406e+00 5.51872671e-01 1.30831254e+00
-5.55566289e-02 -8.65478814e-01 -8.31721723e-01 -1.10648620e+00
-2.51556396e-01 8.13657522e-01 -2.21527442e-01 1.23695716e-01
-1.11499870e+00 -6.74695432e-01 5.82338095e-01 6.29153967e-01
-1.96282178e-01 -1.30673826e+00 -8.40288103e-02 6.75717235e-01
-1.68831408e-01 -1.41764522e+00 -6.68821484e-02 6.02221727e-01
-1.03035331e+00 -1.35117185e+00 -3.34737115e-02 -1.48191893e+00
4.92708057e-01 -1.08612776e-01 1.67579532e+00 2.29067802e-01
-1.42633334e-01 2.00594799e-03 -3.35367084e-01 -2.77762741e-01
-8.88389409e-01 1.51159331e-01 -4.96365875e-01 -7.46517420e-01
3.24372351e-01 -6.18115127e-01 1.62762702e-01 -2.55229808e-02
-5.18558741e-01 1.96716085e-01 5.35225756e-02 9.58249629e-01
4.96511579e-01 -1.65396869e-01 6.82894707e-01 -1.51714075e+00
4.27689314e-01 -3.01811159e-01 -7.94488251e-01 5.89511216e-01
-5.41340172e-01 5.36366105e-01 8.34790111e-01 1.03507631e-01
-1.02973306e+00 4.29316431e-01 -5.43060005e-01 2.92108655e-01
-6.02403879e-02 7.42114246e-01 -3.00292194e-01 1.61847889e-01
5.53682983e-01 3.75978872e-02 -9.44965929e-02 -4.01549935e-01
7.07865953e-01 5.28483652e-02 4.65385556e-01 -6.55729413e-01
4.15980875e-01 8.46465118e-03 1.49249494e-01 -8.30461621e-01
-6.42425597e-01 -1.13611355e-01 -5.65110266e-01 4.96506244e-02
1.05620313e+00 -7.75725245e-01 -5.85066140e-01 -2.97492463e-03
-1.54549742e+00 -4.59271431e-01 5.86450286e-02 8.19726884e-02
-6.39733195e-01 8.95487905e-01 -1.05458534e+00 -6.64664686e-01
-2.84759402e-01 -9.11582410e-01 9.49420929e-01 -7.04159513e-02
-5.93249798e-01 -1.44957733e+00 9.43526719e-03 9.49748382e-02
-1.89808741e-01 2.04856724e-01 1.62940538e+00 -9.76159096e-01
-6.68419480e-01 1.42735526e-01 -3.05146635e-01 3.03233098e-02
-1.21669881e-02 -1.56176016e-01 -6.07689440e-01 1.23365402e-01
-5.95993042e-01 -2.80382335e-01 7.02616692e-01 2.08285257e-01
8.39265049e-01 -2.43160889e-01 -4.74250197e-01 3.72050136e-01
1.47629976e+00 3.12029570e-01 6.10105395e-01 2.80097574e-01
8.69874835e-01 1.01178706e+00 2.86976844e-01 -1.42183036e-01
6.66610718e-01 2.22369671e-01 3.97313774e-01 9.90717933e-02
-3.47964227e-01 -8.64343941e-01 3.87788087e-01 1.23552978e+00
6.58277944e-02 -5.85634232e-01 -1.16645670e+00 5.01100659e-01
-1.93102777e+00 -6.53508902e-01 -8.32932889e-01 1.56761253e+00
5.46238184e-01 3.27315807e-01 7.88485110e-02 1.41003300e-02
7.21611142e-01 1.98355541e-01 -9.42466483e-02 -9.15823936e-01
-3.09638172e-01 7.61065483e-01 6.11348689e-01 8.24969590e-01
-8.47934902e-01 1.68067086e+00 7.28668404e+00 4.84606117e-01
-5.93036413e-01 3.36074010e-02 5.20299375e-01 6.62330568e-01
-6.84038639e-01 4.27734166e-01 -6.27249062e-01 1.22809738e-01
1.23501635e+00 -2.46369913e-01 3.71879637e-01 9.83409345e-01
-2.24589691e-01 1.20156050e-01 -9.59456325e-01 7.83827901e-01
-2.43410647e-01 -1.52687204e+00 2.46047184e-01 -2.12944210e-01
1.36566803e-01 9.36392695e-02 -7.72833645e-01 2.96588480e-01
1.13446069e+00 -1.02166700e+00 6.00940526e-01 -5.55169843e-02
5.41133881e-01 -7.31831193e-01 5.49056053e-01 1.60962358e-01
-1.41445816e+00 4.35807183e-02 -4.79327053e-01 -2.45201290e-01
3.60975325e-01 4.40564841e-01 -8.98056269e-01 8.23958397e-01
3.75334680e-01 7.98723400e-01 -3.00690264e-01 3.25991213e-01
-7.39716291e-01 6.55356109e-01 -3.55073251e-02 -3.53719801e-01
2.75054485e-01 -4.72114444e-01 2.33119652e-01 1.55948710e+00
2.47503936e-01 1.43681839e-01 2.70063937e-01 4.72439498e-01
2.64257826e-02 3.39382440e-01 -1.18990183e+00 -5.03720820e-01
1.57355890e-01 8.99255991e-01 -1.06686771e+00 -3.18100035e-01
-6.36611283e-01 1.17812705e+00 8.52838993e-01 1.95227623e-01
-5.43550491e-01 -2.60660052e-01 5.71364641e-01 -2.52287060e-01
1.52225226e-01 -4.62157875e-01 -2.78570503e-01 -1.21583092e+00
4.57543395e-02 -8.42463970e-01 1.08449841e+00 -9.51136053e-01
-1.23261476e+00 7.34234452e-01 -2.91308135e-01 -3.34764481e-01
-7.42111921e-01 -1.02624178e+00 -4.33155060e-01 6.91474617e-01
-1.27367294e+00 -1.26125193e+00 2.76947618e-01 3.10303062e-01
5.14527619e-01 4.93592173e-02 1.16195452e+00 -4.55472022e-02
-2.19547197e-01 2.47178853e-01 -5.24583817e-01 2.85253972e-01
8.69219452e-02 -1.60607874e+00 1.22977662e+00 9.12881851e-01
5.72528601e-01 3.27150553e-01 7.19265223e-01 -9.00829732e-01
-1.52032721e+00 -8.92570257e-01 1.70677137e+00 -3.41306359e-01
1.19474840e+00 -5.97214937e-01 -1.09473169e+00 1.25976229e+00
1.32730275e-01 -2.70400643e-01 4.50716704e-01 3.61471474e-01
-5.75914383e-01 4.69458997e-01 -6.03110552e-01 6.23793304e-01
1.83017218e+00 -6.23622358e-01 -7.47127414e-01 6.46724045e-01
1.08885622e+00 -4.38504875e-01 -7.60837078e-01 9.79119837e-02
3.84958923e-01 -8.68938386e-01 9.16386306e-01 -1.09883547e+00
3.69705051e-01 8.76819715e-02 -2.25108206e-01 -1.37962556e+00
-2.16898307e-01 -7.33662307e-01 2.27092922e-01 9.10374761e-01
8.07395518e-01 -8.15415084e-01 1.28760231e+00 2.49464348e-01
-3.21533740e-01 -2.76532531e-01 -8.62207770e-01 -9.06868815e-01
2.04583168e-01 -6.51372612e-01 4.75292563e-01 1.15838766e+00
5.62077522e-01 8.82575274e-01 -7.59137943e-02 1.93214834e-01
4.60207611e-01 3.62010628e-01 3.23313624e-01 -1.33620119e+00
-3.43952805e-01 -3.22111964e-01 -5.09045362e-01 -8.80865216e-01
8.94701958e-01 -1.38252342e+00 -5.21808118e-02 -1.93139696e+00
3.61132100e-02 -1.80898368e-01 1.70750007e-01 7.82597661e-01
-7.11324066e-02 -2.88501292e-01 4.21483852e-02 -3.29648346e-01
-4.97881740e-01 9.10495818e-02 9.56402242e-01 3.00586820e-02
-6.51926324e-02 -5.04275084e-01 -6.31973505e-01 8.45936954e-01
9.45073664e-01 -5.40772438e-01 -6.31061435e-01 -5.31992197e-01
7.47909367e-01 6.90369725e-01 8.67772102e-02 -3.37033957e-01
-2.84517799e-02 -1.47151321e-01 -1.67174459e-01 -4.72417831e-01
-5.24124950e-02 -5.01298249e-01 4.23028797e-01 6.63285077e-01
-2.30546743e-01 5.57915688e-01 1.00629941e-01 5.62321305e-01
-2.43085116e-01 -2.69491643e-01 4.46850568e-01 -6.69314265e-01
-1.15278256e+00 -3.51357870e-02 -3.57657075e-01 2.86263019e-01
6.37081265e-01 -3.09581399e-01 -3.31137657e-01 -4.19090450e-01
-1.16499734e+00 -1.50583521e-01 4.32096392e-01 5.86018562e-01
5.48620403e-01 -8.37668240e-01 -5.08379698e-01 1.39455453e-01
1.29264876e-01 -2.55431324e-01 -1.92341834e-01 -1.01320110e-02
-8.96417737e-01 3.74710828e-01 -2.40415186e-02 -2.49285594e-01
-1.31248915e+00 6.20400369e-01 1.48384795e-01 -5.17440736e-01
-7.96574652e-01 9.34050858e-01 2.47776985e-01 -9.23295557e-01
-1.31875679e-01 -3.71973157e-01 -1.45245239e-01 -3.41643602e-01
-4.32557724e-02 -8.29716548e-02 1.86210930e-01 -5.74945867e-01
-3.78921211e-01 3.38828385e-01 1.43678665e-01 2.98988074e-01
1.40385127e+00 2.62929890e-02 -5.27987599e-01 1.44878998e-01
1.18352866e+00 -1.43168673e-01 -6.68501616e-01 7.22695440e-02
9.35495138e-01 8.23190659e-02 -5.47222733e-01 -5.26900887e-01
-7.68559575e-01 6.94822550e-01 -1.09893374e-01 5.11956096e-01
7.22541273e-01 4.53332365e-01 1.05614829e+00 5.04280090e-01
6.18514895e-01 -8.40852499e-01 -3.00658166e-01 8.59236598e-01
4.60376859e-01 -9.20789421e-01 -4.38544273e-01 -1.10692656e+00
-6.14241838e-01 1.29709911e+00 4.25461620e-01 -1.59115836e-01
3.42907488e-01 3.75079215e-01 -7.46692047e-02 -3.83080274e-01
-8.30801070e-01 -3.53506029e-01 -3.43386717e-02 7.41617620e-01
6.25810564e-01 3.75321984e-01 -6.13189936e-01 6.77474439e-01
-6.65899575e-01 -2.89528489e-01 5.51850438e-01 9.07939672e-01
-4.15165484e-01 -1.55469084e+00 4.83523123e-02 2.56357908e-01
-5.88772118e-01 -4.22672749e-01 -6.76684260e-01 1.05901992e+00
-3.79071087e-01 9.55829740e-01 -1.59142315e-01 -2.75007218e-01
3.79135013e-01 4.65298504e-01 6.79472089e-01 -7.49210179e-01
-4.23164815e-01 -1.11546144e-01 1.12430203e+00 -8.19335699e-01
-1.80143625e-01 -4.46614385e-01 -1.76912355e+00 -3.00646245e-01
-6.94983676e-02 2.77513444e-01 6.23054743e-01 7.68748701e-01
2.68113554e-01 3.79650712e-01 3.58762592e-02 -2.07793206e-01
-1.59609526e-01 -4.70638871e-01 -7.07824349e-01 6.01267278e-01
-2.76933938e-01 -1.91716939e-01 -1.86607942e-01 2.76735485e-01] | [10.334688186645508, 9.457886695861816] |
f7820324-39f6-4ea7-a414-6df7c6fbe9d3 | hybrid-dynamic-contrast-and-probability | 2109.14157 | null | https://arxiv.org/abs/2109.14157v1 | https://arxiv.org/pdf/2109.14157v1.pdf | Hybrid Dynamic Contrast and Probability Distillation for Unsupervised Person Re-Id | Unsupervised person re-identification (Re-Id) has attracted increasing attention due to its practical application in the read-world video surveillance system. The traditional unsupervised Re-Id are mostly based on the method alternating between clustering and fine-tuning with the classification or metric learning objectives on the grouped clusters. However, since person Re-Id is an open-set problem, the clustering based methods often leave out lots of outlier instances or group the instances into the wrong clusters, thus they can not make full use of the training samples as a whole. To solve these problems, we present the hybrid dynamic cluster contrast and probability distillation algorithm. It formulates the unsupervised Re-Id problem into an unified local-to-global dynamic contrastive learning and self-supervised probability distillation framework. Specifically, the proposed method can make the utmost of the self-supervised signals of all the clustered and un-clustered instances, from both the instances' self-contrastive level and the probability distillation respective, in the memory-based non-parametric manner. Besides, the proposed hybrid local-to-global contrastive learning can take full advantage of the informative and valuable training examples for effective and robust training. Extensive experiment results show that the proposed method achieves superior performances to state-of-the-art methods, under both the purely unsupervised and unsupervised domain adaptation experiment settings. | ['Xinbo Gao', 'Nannan Wang', 'Jingyu Zhou', 'De Cheng'] | 2021-09-29 | null | null | null | null | ['unsupervised-person-re-identification'] | ['computer-vision'] | [-1.05898611e-01 -3.52794677e-01 -5.65897450e-02 -3.99940163e-01
-5.18692851e-01 -1.40632372e-02 5.72237730e-01 -2.82661226e-02
-5.24256229e-01 5.74330986e-01 4.73110341e-02 4.55388933e-01
-4.23610926e-01 -6.01041555e-01 -1.63731292e-01 -1.09326363e+00
7.31386542e-02 7.18787849e-01 2.67294168e-01 1.75945774e-01
-7.78909847e-02 7.46121556e-02 -1.73122871e+00 -1.06805205e-01
1.30773270e+00 7.87994802e-01 1.70851395e-01 3.71325821e-01
7.14352876e-02 4.77877170e-01 -6.45435870e-01 -1.71475798e-01
1.14317141e-01 -5.56394637e-01 -1.65525541e-01 4.13844019e-01
2.21458927e-01 -8.24556425e-02 -2.88149148e-01 1.32477796e+00
6.90434098e-01 5.36277592e-01 7.62211680e-01 -1.34001386e+00
-5.18761575e-01 3.07741702e-01 -8.11645985e-01 4.34915543e-01
1.69212326e-01 1.64989866e-02 4.72655356e-01 -6.84198439e-01
1.77020267e-01 1.19751430e+00 6.15520597e-01 5.75686455e-01
-1.15247881e+00 -9.70985711e-01 4.18799520e-01 5.40323734e-01
-1.70986724e+00 -2.50912279e-01 9.07040954e-01 -6.77183211e-01
3.63667160e-01 1.38622075e-01 5.20561755e-01 9.20693517e-01
-4.59247857e-01 7.22010076e-01 1.17118180e+00 -2.35380843e-01
2.06860706e-01 4.74797666e-01 4.20118541e-01 3.16119820e-01
3.56635511e-01 2.47398674e-01 -1.77547440e-01 -4.65838201e-02
5.18679500e-01 4.16416198e-01 -2.45518684e-01 -3.34832162e-01
-1.18086708e+00 4.58915889e-01 2.79642731e-01 4.18141663e-01
-2.32768893e-01 -4.18861210e-01 2.62307853e-01 -1.50571149e-02
3.51476490e-01 -2.49652052e-03 -2.69924879e-01 -2.26863706e-03
-1.15185356e+00 1.95676878e-01 4.23380315e-01 7.96429515e-01
7.94921756e-01 -1.49567306e-01 -2.86928087e-01 9.41149116e-01
2.95548916e-01 7.52371073e-01 9.48252082e-01 -5.22912562e-01
3.13819975e-01 7.06075668e-01 1.01331197e-01 -1.29189301e+00
-4.18550938e-01 -6.61589384e-01 -1.25709569e+00 -5.18188998e-02
4.21346903e-01 -2.87887841e-01 -6.43034816e-01 1.86550629e+00
4.54319924e-01 7.10585475e-01 7.55687729e-02 8.86876225e-01
7.51723468e-01 6.97295904e-01 2.91964531e-01 -6.82248056e-01
1.12974918e+00 -7.54729152e-01 -7.93828249e-01 -1.48224488e-01
1.50159866e-01 -3.50021571e-01 7.05012619e-01 4.32532221e-01
-6.82934582e-01 -1.12494934e+00 -1.18067062e+00 5.97190142e-01
-2.23231733e-01 3.19348246e-01 3.08572929e-02 7.04708993e-01
-5.96277118e-01 3.49699199e-01 -5.68848550e-01 -4.27171677e-01
2.41539419e-01 2.92174459e-01 -3.08620811e-01 -1.17891632e-01
-1.07328761e+00 4.89424616e-01 9.03214216e-01 2.20255792e-01
-5.35173118e-01 -4.48514819e-01 -4.62104261e-01 6.27555372e-03
4.36273098e-01 -4.15721864e-01 7.46315718e-01 -9.63769376e-01
-1.17304921e+00 7.63525307e-01 -1.49818629e-01 -2.12868229e-01
5.08595705e-01 1.32526848e-02 -8.75760436e-01 1.19743913e-01
2.10452944e-01 2.95118392e-01 8.79464805e-01 -1.27956784e+00
-1.10101116e+00 -6.00953400e-01 -5.91541350e-01 4.95877713e-01
-5.68593442e-01 -1.57089755e-01 -6.92254364e-01 -8.40561450e-01
1.08658835e-01 -7.92396486e-01 5.48587330e-02 -4.36838090e-01
-3.51610214e-01 -2.70921111e-01 9.23965812e-01 -7.80715644e-01
1.41379762e+00 -2.39571643e+00 1.60402626e-01 3.95326555e-01
3.17395598e-01 4.02369380e-01 2.16865897e-01 -1.54897526e-01
-2.14129120e-01 -3.10050398e-01 -2.31022760e-01 -2.24383861e-01
-1.31582871e-01 -5.59963770e-02 2.17229679e-01 5.41507423e-01
-1.91871703e-01 3.44244838e-01 -9.96836066e-01 -8.49181056e-01
5.28137445e-01 2.62712330e-01 -2.91328013e-01 4.53667581e-01
1.75227359e-01 7.53686607e-01 -3.09775203e-01 3.11569065e-01
8.43118966e-01 -1.50863808e-02 1.18936062e-01 -3.48994672e-01
7.08136559e-02 -5.17308295e-01 -1.63453746e+00 1.17368376e+00
1.90850303e-01 3.06809753e-01 -1.00200683e-01 -1.33642089e+00
1.00672185e+00 3.22357528e-02 7.25044549e-01 -5.90267777e-01
-1.73754200e-01 2.08148658e-02 -1.26797065e-01 -5.37861109e-01
2.18552887e-01 -2.22866505e-01 -5.60435392e-02 2.68643379e-01
1.39155701e-01 6.51163340e-01 2.33113751e-01 -1.53572205e-02
4.51392680e-01 -2.01411638e-02 2.95758307e-01 -2.33736634e-01
8.14532101e-01 -2.38224849e-01 9.01606858e-01 6.92053020e-01
-4.21392381e-01 4.83243108e-01 -1.27441427e-02 -4.25133891e-02
-8.17395449e-01 -1.16338241e+00 -2.41233602e-01 9.61843967e-01
6.04225874e-01 1.10369036e-02 -9.01101112e-01 -6.98007941e-01
-1.55741826e-01 7.24170685e-01 -4.02526647e-01 -4.01194274e-01
-3.29630136e-01 -1.27337670e+00 3.48565668e-01 3.22239876e-01
1.12662709e+00 -9.71389472e-01 -9.79872569e-02 1.53077886e-01
-4.48488325e-01 -8.85402858e-01 -5.25612414e-01 -9.63822827e-02
-6.87630653e-01 -1.16974151e+00 -1.01357293e+00 -9.74067986e-01
7.37078190e-01 4.55607653e-01 6.30070627e-01 1.53491367e-02
-1.30141079e-02 4.22026277e-01 -3.48750949e-01 -1.58795848e-01
-2.45624259e-01 -1.45146415e-01 5.11595964e-01 5.48657596e-01
1.02754402e+00 -5.27158856e-01 -4.33603019e-01 7.68089533e-01
-5.28048396e-01 -1.18318468e-01 4.06148642e-01 9.18026984e-01
6.28283918e-01 8.04511845e-01 8.25338006e-01 -6.78000748e-01
4.75915343e-01 -5.15064359e-01 -4.30298775e-01 2.90959150e-01
-7.77190745e-01 -6.97216988e-02 4.90692735e-01 -7.67201841e-01
-1.32858670e+00 -1.44060791e-01 1.53977752e-01 -3.44656259e-01
-4.08990860e-01 3.67886871e-01 -7.96784580e-01 2.20921576e-01
6.21423244e-01 5.73157489e-01 4.06239294e-02 -4.03635085e-01
1.49294719e-01 1.01168442e+00 9.94748533e-01 -5.43286800e-01
9.77689445e-01 4.42525357e-01 -5.33856750e-01 -8.43780041e-01
-7.02535331e-01 -9.24618602e-01 -7.74520934e-01 -4.00781155e-01
9.07336950e-01 -9.88499939e-01 -5.62969208e-01 8.99659634e-01
-6.23759687e-01 -9.74601042e-03 -4.13639218e-01 5.61268330e-01
-2.48124376e-01 8.07301998e-01 -1.80545956e-01 -9.21465516e-01
-4.03519161e-02 -1.04057336e+00 6.89379632e-01 7.48423576e-01
1.40316024e-01 -9.17402148e-01 2.15046871e-02 2.30432764e-01
1.05199665e-02 2.60643631e-01 4.94358331e-01 -9.96491253e-01
-1.11952372e-01 -3.75207484e-01 -2.81389534e-01 4.38181370e-01
4.20114130e-01 -3.65232289e-01 -1.04418921e+00 -4.30995464e-01
9.24581811e-02 7.32779130e-02 7.69461751e-01 4.08176452e-01
9.61916149e-01 -2.22034827e-01 -5.62089920e-01 5.66491485e-01
1.18473673e+00 3.45974505e-01 5.33183634e-01 4.48742300e-01
8.90181124e-01 6.34508193e-01 7.29668617e-01 5.55381596e-01
4.19092178e-01 7.36888707e-01 -4.66715805e-02 -6.20486550e-02
3.64412293e-02 -3.49694878e-01 3.08167487e-01 7.73594499e-01
-1.61569506e-01 -6.08604923e-02 -5.98215222e-01 4.26226288e-01
-2.17615080e+00 -1.40046620e+00 1.09936081e-01 2.68047142e+00
6.80002809e-01 9.33660269e-02 5.01430869e-01 3.94598901e-01
1.50512743e+00 -2.52900831e-02 -6.57432199e-01 4.37271357e-01
-2.18105763e-01 -4.66141194e-01 2.96295434e-01 1.47267163e-01
-1.49094737e+00 5.58239877e-01 5.09944916e+00 1.14929748e+00
-7.40987837e-01 2.18288094e-01 6.57949567e-01 2.10663289e-01
2.89962232e-01 -3.87052447e-01 -8.87927234e-01 9.84205782e-01
7.95893192e-01 -1.82070658e-01 4.04924303e-01 8.98756683e-01
2.14490443e-01 -2.00225398e-01 -1.03807783e+00 1.71446574e+00
3.10145736e-01 -6.97187841e-01 -1.11461788e-01 -8.01592227e-03
8.06679070e-01 -5.59021831e-01 8.47164690e-02 4.37633395e-01
1.04570545e-01 -7.84545362e-01 5.76707959e-01 8.91251743e-01
7.18605697e-01 -8.63988638e-01 9.53292668e-01 6.35351121e-01
-1.24557352e+00 -4.50410306e-01 -4.04495418e-01 1.98928013e-01
-4.55873087e-03 6.22821331e-01 -3.01912487e-01 6.82008028e-01
1.04051602e+00 1.02888191e+00 -7.29839563e-01 1.19255650e+00
6.41883239e-02 5.94144881e-01 -3.12453896e-01 1.65956169e-01
-3.07650805e-01 -4.13652331e-01 6.36346817e-01 1.21755469e+00
1.84921071e-01 2.97117531e-01 4.78772432e-01 6.00458026e-01
3.53486001e-01 -3.84377725e-02 -3.71779092e-02 2.95679033e-01
5.93566179e-01 1.01393843e+00 -5.78131974e-01 -6.41740799e-01
-2.39520326e-01 9.75285470e-01 1.86481550e-01 5.02615988e-01
-7.60677159e-01 -3.14171195e-01 3.65434349e-01 1.59899205e-01
3.40383440e-01 -8.37695822e-02 -2.15576719e-02 -1.32984912e+00
3.16265337e-02 -8.47777367e-01 8.21855128e-01 -4.77942169e-01
-1.80362010e+00 3.22834671e-01 4.54644501e-01 -1.52990806e+00
-3.19335997e-01 -3.07611197e-01 -6.27185702e-01 6.11188710e-01
-1.31039977e+00 -9.88285005e-01 -8.69564533e-01 9.28090036e-01
3.04391891e-01 -7.12578177e-01 5.07911801e-01 5.50041914e-01
-9.41670418e-01 9.24860001e-01 5.61945379e-01 3.08042347e-01
1.02157211e+00 -1.11666095e+00 -4.42462206e-01 1.16390455e+00
-1.94504708e-01 4.80720222e-01 5.77301860e-01 -8.28836381e-01
-7.46573865e-01 -1.21920645e+00 4.16393608e-01 -1.74761385e-01
2.40918100e-01 -1.23384908e-01 -9.60534513e-01 3.77946585e-01
-2.76226103e-01 -1.35328129e-01 6.07166171e-01 4.27767076e-02
-8.66998583e-02 -5.40802717e-01 -1.31616724e+00 3.70416403e-01
8.44760239e-01 -2.28537112e-01 -8.16195548e-01 2.61435509e-01
2.90957838e-01 -2.00439133e-02 -7.83121169e-01 3.58389169e-01
1.84532613e-01 -1.12711954e+00 1.07561028e+00 -1.68861017e-01
-1.86614990e-01 -6.80932879e-01 6.73333788e-03 -1.21826613e+00
-6.94523215e-01 -2.90396094e-01 -8.69269744e-02 1.79714894e+00
-1.18109420e-01 -7.22265005e-01 5.30791879e-01 6.30314767e-01
2.16773644e-01 -5.12821190e-02 -9.28173065e-01 -7.22517610e-01
-3.05587262e-01 -2.43033186e-01 5.13020456e-01 9.95339096e-01
4.75337841e-02 2.07746282e-01 -5.46320975e-01 3.87703419e-01
1.15050209e+00 -1.08751610e-01 8.28983426e-01 -1.55853713e+00
-3.37295800e-01 -4.06729877e-01 -7.03893363e-01 -1.00976300e+00
-5.22356015e-03 -7.37375557e-01 2.12199047e-01 -1.18447208e+00
6.71000659e-01 -6.06555283e-01 -5.13213992e-01 8.33789408e-02
-7.00936198e-01 -7.50922412e-02 1.12972796e-01 6.95779026e-01
-9.39220309e-01 5.62885284e-01 6.90927505e-01 -4.42330301e-01
-5.11083603e-01 1.42697468e-01 -7.18725979e-01 7.08682358e-01
5.87767601e-01 -2.44553655e-01 -4.81325656e-01 1.14794113e-01
-5.32653928e-01 -1.71108171e-01 4.83765423e-01 -1.50292790e+00
4.91945356e-01 -1.01767682e-01 6.72013462e-01 -6.11011386e-01
6.30133972e-02 -8.76434386e-01 1.83600202e-01 2.21909836e-01
-2.23777026e-01 -3.99309397e-01 -2.83282027e-02 9.20455277e-01
-2.89055020e-01 -1.04532473e-01 1.09213316e+00 -1.88285932e-01
-9.40168262e-01 4.01376724e-01 -1.43717706e-01 2.19156027e-01
1.11897612e+00 -6.53728724e-01 -1.15065880e-01 -2.93664366e-01
-8.33441973e-01 3.79365146e-01 3.94526213e-01 2.32088581e-01
4.84544992e-01 -1.35994637e+00 -8.19273710e-01 3.35693687e-01
2.34944865e-01 7.01315105e-02 7.73235500e-01 7.39761412e-01
-1.05598774e-02 1.56275317e-01 -1.42834350e-01 -8.64492714e-01
-1.01380813e+00 1.01656747e+00 4.31699008e-01 -2.74904132e-01
-5.92123091e-01 4.20979142e-01 4.46082801e-01 -3.26745600e-01
4.37276721e-01 2.59765089e-01 -6.31729841e-01 1.00723058e-01
8.15295100e-01 6.40924692e-01 -3.48529190e-01 -9.12167788e-01
-3.38202536e-01 7.38521755e-01 1.01479180e-02 4.92978133e-02
8.95137548e-01 -4.71920401e-01 2.40596253e-02 4.58828032e-01
9.28525090e-01 -1.96454734e-01 -1.19473279e+00 -5.05757272e-01
-4.53230068e-02 -3.67234200e-01 -7.68689290e-02 -6.24509513e-01
-9.21558559e-01 6.35544002e-01 1.16403270e+00 1.12278633e-01
1.20822263e+00 -1.17642403e-01 7.06688643e-01 3.28229427e-01
3.20750386e-01 -1.50354636e+00 1.48975030e-01 -3.44311073e-02
2.89788425e-01 -1.38385594e+00 1.05438858e-01 -2.40028679e-01
-7.31232464e-01 8.89183640e-01 8.13877821e-01 -1.42080635e-01
7.48323083e-01 -5.10845184e-02 -6.00067377e-02 1.51819855e-01
-3.81487496e-02 -1.80056691e-01 4.07024384e-01 1.11238766e+00
-1.23524122e-01 2.20265806e-01 -4.79713194e-02 1.08868515e+00
-7.08449399e-03 -6.50887638e-02 1.03039160e-01 3.30172479e-01
-5.06908476e-01 -6.63811386e-01 -7.91612804e-01 5.13981760e-01
-1.18958000e-02 3.12442273e-01 7.17720315e-02 6.87062919e-01
5.81729054e-01 1.20837414e+00 1.87068820e-01 -7.26858675e-01
2.11773708e-01 3.35577340e-03 1.41455218e-01 -3.48896563e-01
-2.49998629e-01 2.77323484e-01 -2.49870211e-01 -1.17680341e-01
-8.36622417e-01 -8.99661422e-01 -1.11726391e+00 -1.95239916e-01
-4.41291660e-01 2.34974280e-01 3.62246521e-02 1.09093201e+00
1.88778549e-01 4.10447985e-01 7.17963696e-01 -8.60725880e-01
-4.11858290e-01 -1.13429809e+00 -7.45643973e-01 7.71899164e-01
4.68155481e-02 -9.31583583e-01 -5.84825516e-01 2.70881087e-01] | [14.878098487854004, 1.1332310438156128] |
6776961c-2047-45e3-aa89-f610ff3775dc | applying-unsupervised-keyphrase-methods-on | 2303.08928 | null | https://arxiv.org/abs/2303.08928v1 | https://arxiv.org/pdf/2303.08928v1.pdf | Applying unsupervised keyphrase methods on concepts extracted from discharge sheets | Clinical notes containing valuable patient information are written by different health care providers with various scientific levels and writing styles. It might be helpful for clinicians and researchers to understand what information is essential when dealing with extensive electronic medical records. Entities recognizing and mapping them to standard terminologies is crucial in reducing ambiguity in processing clinical notes. Although named entity recognition and entity linking are critical steps in clinical natural language processing, they can also result in the production of repetitive and low-value concepts. In other hand, all parts of a clinical text do not share the same importance or content in predicting the patient's condition. As a result, it is necessary to identify the section in which each content is recorded and also to identify key concepts to extract meaning from clinical texts. In this study, these challenges have been addressed by using clinical natural language processing techniques. In addition, in order to identify key concepts, a set of popular unsupervised key phrase extraction methods has been verified and evaluated. Considering that most of the clinical concepts are in the form of multi-word expressions and their accurate identification requires the user to specify n-gram range, we have proposed a shortcut method to preserve the structure of the expression based on TF-IDF. In order to evaluate the pre-processing method and select the concepts, we have designed two types of downstream tasks (multiple and binary classification) using the capabilities of transformer-based models. The obtained results show the superiority of proposed method in combination with SciBERT model, also offer an insight into the efficacy of general extracting essential phrase methods for clinical notes. | ['Maryam Lotfi Shahreza', 'Matthias Samwald', 'Nasser Ghadiri', 'Hoda Memarzadeh'] | 2023-03-15 | null | null | null | null | ['entity-linking'] | ['natural-language-processing'] | [ 4.30103242e-01 9.50690582e-02 -2.70424604e-01 -1.42184108e-01
-5.61411262e-01 -6.18580103e-01 2.16027424e-01 1.13780797e+00
-5.70825160e-01 9.51110423e-01 5.49948037e-01 -3.61296684e-01
-5.71826637e-01 -6.64884746e-01 8.32089335e-02 -6.08660758e-01
2.53380295e-02 5.46416938e-01 -1.87354758e-01 -1.01307645e-01
4.51371074e-01 5.09278893e-01 -1.16727138e+00 5.47521651e-01
9.40122187e-01 6.37273729e-01 2.37066865e-01 4.24002290e-01
-7.56943405e-01 7.38113463e-01 -5.55884242e-01 -3.82889986e-01
-1.34018147e-02 -5.67979872e-01 -9.39952672e-01 -1.51384175e-01
-3.98692131e-01 -9.42318793e-03 3.41810614e-01 1.14928854e+00
4.75204468e-01 -6.83441907e-02 8.40948105e-01 -7.35749900e-01
-2.18110591e-01 7.39555597e-01 -2.35330954e-01 3.39193970e-01
5.37661314e-01 -3.37519646e-01 9.70968068e-01 -7.28401542e-01
8.16578984e-01 7.63968289e-01 5.31913757e-01 2.36856893e-01
-7.04829633e-01 -5.31030476e-01 -4.36218642e-02 -2.96556745e-02
-1.44257724e+00 -2.00195789e-01 5.52431107e-01 -5.94567955e-01
9.49817061e-01 3.41683596e-01 5.14714956e-01 5.96946478e-01
5.60747743e-01 2.93160141e-01 9.21113610e-01 -6.73207462e-01
-6.73140138e-02 5.09583533e-01 5.07158816e-01 5.69541156e-01
4.86553252e-01 -5.60769558e-01 -3.20805073e-01 -4.39711750e-01
1.50075734e-01 3.00037742e-01 -4.29400206e-01 4.00731087e-01
-1.16042078e+00 5.99865615e-01 -1.10351890e-01 8.33075047e-01
-5.05900502e-01 -5.62265694e-01 7.62590408e-01 -6.23484328e-02
6.78194556e-05 5.56659758e-01 -5.17662287e-01 -9.45901349e-02
-8.93233776e-01 -6.22721128e-02 9.46661115e-01 9.98420298e-01
2.42070436e-01 -6.17412865e-01 -3.67660940e-01 4.77374315e-01
2.44475588e-01 1.81588441e-01 9.69668031e-01 -8.30789953e-02
4.94583279e-01 1.06191003e+00 1.20637398e-02 -1.28093779e+00
-5.03052413e-01 -4.50455487e-01 -1.00832725e+00 -5.14026523e-01
1.55743733e-02 -7.68529475e-02 -9.22811329e-01 1.32228196e+00
2.06602260e-01 -6.92408532e-02 4.14291292e-01 5.61108708e-01
1.03917718e+00 5.73661149e-01 5.33410192e-01 -7.69741178e-01
2.04129505e+00 -3.97186756e-01 -1.17874491e+00 2.35807985e-01
6.78463578e-01 -1.12243724e+00 4.47755218e-01 2.84131467e-01
-6.80105686e-01 -2.93785691e-01 -8.64214063e-01 -1.10881962e-01
-6.10128701e-01 3.42962623e-01 4.51404840e-01 4.12641495e-01
-4.34339643e-01 4.87247139e-01 -6.59738898e-01 -5.55985332e-01
2.41314888e-01 2.89185882e-01 -4.60868686e-01 1.74775973e-01
-1.35199845e+00 8.68211091e-01 7.68863261e-01 1.71634242e-01
-1.38783872e-01 -6.39308274e-01 -5.51363409e-01 2.49951884e-01
2.73232728e-01 -7.82682180e-01 8.78149152e-01 -6.31626666e-01
-8.61844659e-01 7.68797934e-01 -2.89052337e-01 -2.49651983e-01
1.65262386e-01 -1.00144215e-01 -7.30907202e-01 2.78566539e-01
2.80155540e-01 5.88473119e-02 2.91197747e-01 -5.57865381e-01
-9.36692357e-01 -4.00863558e-01 -2.92541176e-01 3.29427958e-01
-5.28991520e-01 2.32406572e-01 -3.42562526e-01 -8.20489407e-01
2.66960293e-01 -7.36350477e-01 -2.69996047e-01 -3.55943680e-01
-5.23932219e-01 -1.26725584e-01 4.45539534e-01 -9.45660949e-01
1.75566304e+00 -1.96274412e+00 -1.00737214e-01 4.50950950e-01
3.31768453e-01 3.54083866e-01 5.36719859e-01 8.20618749e-01
-1.10951722e-01 6.15107417e-01 -2.68028587e-01 2.28951007e-01
-4.09585059e-01 1.39373049e-01 -2.29008615e-01 4.83545326e-02
1.98717639e-01 5.61885774e-01 -6.75324976e-01 -1.04451346e+00
-2.30781529e-02 4.36052054e-01 -2.34594256e-01 1.42024338e-01
8.11869204e-02 3.88600677e-01 -8.76467288e-01 6.63176596e-01
3.84123564e-01 -2.43103564e-01 3.20906132e-01 -6.07493281e-01
-3.94236036e-02 3.87728930e-01 -1.14858603e+00 1.38205135e+00
-1.57667443e-01 2.21637227e-02 -2.30308533e-01 -1.07100224e+00
7.22810984e-01 8.07829857e-01 9.54440773e-01 -2.80951858e-01
3.11038077e-01 2.46948317e-01 -7.51438737e-02 -1.03531146e+00
3.51212680e-01 -3.88242096e-01 3.68588865e-02 1.10612385e-01
-7.82114789e-02 3.02096307e-01 4.28519666e-01 1.69654414e-01
1.05929720e+00 -2.57510602e-01 1.11841893e+00 -2.20132038e-01
9.70274448e-01 3.12924325e-01 6.38565719e-01 3.80783170e-01
9.65726450e-02 3.46482396e-01 3.97832006e-01 -2.06728309e-01
-6.43196106e-01 -6.13311648e-01 -4.46597159e-01 2.99053609e-01
-1.44678056e-01 -6.55034065e-01 -3.73097062e-01 -5.52363276e-01
-1.94992036e-01 4.52727020e-01 -3.93653154e-01 4.25161310e-02
-3.83278668e-01 -8.51005018e-01 7.45773494e-01 2.34624580e-01
2.28796259e-01 -9.18219626e-01 -7.73128271e-01 6.28844440e-01
-3.06566507e-01 -1.09996569e+00 -4.27185446e-01 4.01178449e-01
-9.80839014e-01 -1.34923828e+00 -5.43671429e-01 -8.92432749e-01
7.80261457e-01 -2.25886449e-01 7.84403682e-01 9.38884467e-02
-5.11020660e-01 -5.50104901e-02 -7.22265840e-01 -6.35168672e-01
-3.86056244e-01 3.05694669e-01 -2.07053021e-01 -2.10709006e-01
7.02021241e-01 -4.30791378e-01 -4.91491526e-01 -1.66193187e-01
-1.12637424e+00 4.03187089e-02 8.61361444e-01 7.48014331e-01
6.20183349e-01 1.88243285e-01 5.69935024e-01 -1.19375658e+00
1.05185831e+00 -5.22310734e-01 -1.85708836e-01 5.16699731e-01
-9.39430654e-01 3.65174860e-01 7.40848660e-01 -2.58528739e-01
-9.33568478e-01 1.57485858e-01 -3.46369445e-01 1.56185463e-01
-2.54401177e-01 1.08579111e+00 4.13739681e-02 5.30193090e-01
4.03918654e-01 2.95290619e-01 -9.74640548e-02 -4.90295619e-01
4.35636975e-02 9.31194544e-01 1.49378642e-01 -3.21111232e-01
3.47426057e-01 1.22333743e-01 1.85869187e-01 -7.69654632e-01
-6.00892186e-01 -9.75099921e-01 -6.74793005e-01 1.76556289e-01
1.18112338e+00 -6.46412134e-01 -7.20317066e-01 2.57365871e-02
-1.10850489e+00 8.38813365e-01 -1.40054822e-02 7.07142413e-01
1.17169693e-01 5.53861320e-01 -6.59670711e-01 -6.55025780e-01
-9.26276028e-01 -9.99415874e-01 7.94361174e-01 3.48794669e-01
-4.31465983e-01 -8.14980865e-01 4.17427495e-02 7.82993436e-02
7.68604651e-02 1.43117577e-01 1.44307196e+00 -1.15281725e+00
-3.27394009e-02 -2.46499494e-01 -1.52670257e-02 -5.64669147e-02
8.07334125e-01 -5.78814223e-02 -5.19974589e-01 7.98899308e-02
2.55212873e-01 2.34100461e-01 5.31967342e-01 2.65458256e-01
8.35812807e-01 -5.82386732e-01 -6.44179463e-01 3.98599476e-01
1.50110328e+00 7.59974003e-01 4.59180593e-01 1.89619139e-01
4.85937387e-01 5.85206032e-01 6.66923285e-01 4.87233251e-01
8.43424797e-02 2.81741023e-01 -2.49314234e-01 3.02834846e-02
3.69205743e-01 -1.56920508e-01 -1.81131735e-02 1.04330993e+00
2.14148592e-02 -1.71409220e-01 -1.06219566e+00 4.57586735e-01
-1.53327024e+00 -8.44126105e-01 -1.83793530e-01 1.91921258e+00
1.26504123e+00 1.31292135e-01 -2.60814697e-01 3.01288217e-01
5.80268204e-01 -4.26423967e-01 -6.39714524e-02 -3.63620222e-01
1.55962676e-01 4.70413804e-01 4.78091985e-01 2.41843209e-01
-8.01151633e-01 6.54360056e-01 5.14497662e+00 6.84600949e-01
-1.24769199e+00 -1.46286443e-01 4.73452240e-01 2.18187317e-01
-9.17317569e-02 -6.70973435e-02 -1.09915936e+00 4.03796673e-01
7.85613894e-01 -4.30254251e-01 -2.56568938e-01 6.61750138e-01
4.73680496e-01 -1.37402803e-01 -1.01497769e+00 9.00728822e-01
-4.98388819e-02 -1.18280768e+00 2.84712881e-01 -1.67862445e-01
3.53056312e-01 -5.99566400e-01 -3.87546599e-01 -6.25463501e-02
-2.59224147e-01 -1.05403686e+00 9.79014114e-02 6.82112396e-01
6.16014540e-01 -5.67058921e-01 1.27141774e+00 2.44849846e-01
-1.41481578e+00 8.07568952e-02 -2.54343569e-01 1.88594401e-01
1.68530643e-01 7.05449283e-01 -1.36229265e+00 9.81502771e-01
4.44125772e-01 5.27236044e-01 -3.41297358e-01 1.03750432e+00
-8.95082429e-02 4.87103224e-01 -2.84910113e-01 -1.83542609e-01
7.11296946e-02 -1.38541371e-01 2.78860450e-01 1.45097923e+00
5.92601120e-01 7.05423713e-01 1.89661577e-01 6.36458874e-01
4.50955369e-02 1.03822947e+00 -6.03920877e-01 -6.21457458e-01
4.76028919e-01 1.28401887e+00 -1.07066631e+00 -3.80715609e-01
-3.35664272e-01 6.59538627e-01 -2.59419531e-01 3.10523715e-02
-5.15799761e-01 -6.06637001e-01 3.32444608e-01 2.28910699e-01
-4.99363728e-02 1.22577727e-01 -2.65252948e-01 -9.37367797e-01
5.65052629e-02 -9.58251774e-01 6.94947183e-01 -3.98469627e-01
-1.13484204e+00 7.86079466e-01 -1.33672962e-02 -1.42325890e+00
-1.86939672e-01 -5.81749201e-01 -1.89668357e-01 9.96677279e-01
-1.27266443e+00 -9.66334641e-01 -1.68208316e-01 6.03140295e-01
4.26972687e-01 -1.19607307e-01 1.13050735e+00 6.01493359e-01
-4.68210101e-01 3.30508471e-01 -5.95301352e-02 3.88012648e-01
8.54165614e-01 -8.74425232e-01 -3.21301937e-01 6.62065983e-01
3.55332233e-02 1.22825420e+00 6.14383519e-01 -9.10693884e-01
-9.47863340e-01 -7.81912267e-01 1.44836473e+00 -5.71265481e-02
3.50674063e-01 2.39226863e-01 -9.44689691e-01 4.89659488e-01
7.85837695e-02 -4.85106200e-01 1.08303356e+00 -9.94000435e-02
1.91341892e-01 -6.37236983e-02 -1.09472919e+00 4.69250530e-01
5.75431406e-01 -4.82649982e-01 -1.05224574e+00 2.67295837e-01
5.82309008e-01 -1.55812621e-01 -9.91940916e-01 4.63242739e-01
3.42299789e-01 -4.15586233e-01 7.71097720e-01 -9.21730161e-01
3.87274563e-01 -3.45075667e-01 1.49501264e-01 -9.42041695e-01
-4.87097725e-02 -2.69089371e-01 3.60336065e-01 1.45538342e+00
6.61533535e-01 -4.97825176e-01 5.06753445e-01 7.45711863e-01
3.83661613e-02 -7.72902131e-01 -5.84203720e-01 -6.67909756e-02
-3.54562461e-01 -1.35044947e-01 3.18833441e-01 1.14912879e+00
5.19917309e-01 4.98807579e-01 -7.40156099e-02 1.31761149e-01
8.82006809e-02 6.41798303e-02 1.16515286e-01 -1.31658697e+00
1.15378231e-01 -2.92175412e-01 -4.52121675e-01 -4.45326507e-01
-1.85678512e-01 -9.64591146e-01 -1.36147276e-01 -1.79543233e+00
3.86248976e-01 -5.07543325e-01 -5.37585795e-01 5.79188526e-01
-3.28832686e-01 -4.36726868e-01 -1.83478266e-01 4.06832308e-01
-8.86212513e-02 5.05639650e-02 8.35169315e-01 -1.81685761e-01
-4.37295705e-01 1.19157925e-01 -7.71252751e-01 5.66685081e-01
7.55919337e-01 -8.89735699e-01 -5.16257823e-01 8.36710855e-02
4.68575686e-01 2.42516831e-01 -1.71452284e-01 -7.09788561e-01
4.89265591e-01 -3.09879124e-01 1.12254851e-01 -5.63670754e-01
-2.28152037e-01 -1.13026965e+00 2.22769409e-01 5.75125515e-01
-3.90910536e-01 4.32045817e-01 2.61755675e-01 2.25963950e-01
-5.35821140e-01 -7.14676321e-01 3.58637154e-01 -3.66778880e-01
-6.23659849e-01 -1.23550728e-01 -5.25705934e-01 1.12391211e-01
9.88655746e-01 -5.91925420e-02 5.08549288e-02 6.48967549e-02
-7.89245486e-01 5.53380512e-02 -2.14378033e-02 2.56174505e-01
6.83928907e-01 -8.75542343e-01 -5.76828122e-01 -6.38468936e-02
2.15179443e-01 -1.02886319e-01 2.22763941e-01 9.79506493e-01
-8.06042612e-01 7.81018376e-01 -3.11623096e-01 -2.12015837e-01
-1.71795714e+00 6.04021430e-01 6.34224638e-02 -6.79417968e-01
-4.88695413e-01 4.85157907e-01 -1.17010035e-01 7.06673414e-03
1.61877662e-01 -7.83369720e-01 -9.07726705e-01 4.90957916e-01
5.53388059e-01 -1.57786801e-01 4.02349442e-01 -5.10819018e-01
-5.85263312e-01 6.45029128e-01 -2.40925536e-01 4.08890322e-02
1.24386227e+00 -1.08142085e-02 -5.23012400e-01 1.92124888e-01
1.05142176e+00 4.11097050e-01 7.78256208e-02 -9.64476690e-02
4.73583728e-01 -2.53263637e-02 -7.28288665e-02 -8.08165610e-01
-7.29904532e-01 6.06168270e-01 5.51246643e-01 -9.50595587e-02
1.36027169e+00 -3.02723140e-01 5.17709315e-01 4.49309826e-01
1.80755123e-01 -1.00594819e+00 -4.56722587e-01 3.40707898e-01
5.14648438e-01 -1.10594022e+00 1.49469465e-01 -7.67590523e-01
-5.97198486e-01 1.36824071e+00 1.28917426e-01 5.71634293e-01
9.15452719e-01 4.78235632e-01 1.79108322e-01 -3.27451110e-01
-4.79656160e-01 -1.97860241e-01 3.50278825e-01 1.50366649e-01
8.92620206e-01 -1.61781594e-01 -1.22828913e+00 1.03477740e+00
-5.24837598e-02 4.35164958e-01 2.71198273e-01 1.06403399e+00
-3.07863235e-01 -1.36851215e+00 -3.81913483e-01 6.65225387e-01
-1.10930264e+00 -5.40566564e-01 -2.81043708e-01 5.79038680e-01
4.32197243e-01 9.90208745e-01 -4.44647193e-01 -2.81537503e-01
2.60018617e-01 4.87123340e-01 -2.83774473e-02 -8.45580578e-01
-8.59121501e-01 9.23220739e-02 1.04834080e-01 -5.72120175e-02
-5.71647286e-01 -4.79228556e-01 -1.71733963e+00 1.86984047e-01
-3.59116852e-01 6.98025048e-01 4.67805028e-01 1.07342875e+00
3.27830940e-01 6.76340699e-01 9.02980044e-02 2.31847242e-01
-4.38295692e-01 -1.02781129e+00 -2.21066415e-01 6.49728596e-01
9.49889049e-03 -2.67269552e-01 -8.73972699e-02 4.36019361e-01] | [8.488511085510254, 8.675358772277832] |
758960b2-9713-468d-8dcf-dd4e5df100a2 | exploration-of-interpretability-techniques | 2006.02570 | null | https://arxiv.org/abs/2006.02570v3 | https://arxiv.org/pdf/2006.02570v3.pdf | Exploration of Interpretability Techniques for Deep COVID-19 Classification using Chest X-ray Images | The outbreak of COVID-19 has shocked the entire world with its fairly rapid spread and has challenged different sectors. One of the most effective ways to limit its spread is the early and accurate diagnosis of infected patients. Medical imaging such as X-ray and Computed Tomography (CT) combined with the potential of Artificial Intelligence (AI) plays an essential role in supporting the medical staff in the diagnosis process. Thereby, five different deep learning models (ResNet18, ResNet34, InceptionV3, InceptionResNetV2, and DenseNet161) and their Ensemble have been used in this paper to classify COVID-19, pneumoni{\ae} and healthy subjects using Chest X-Ray images. Multi-label classification was performed to predict multiple pathologies for each patient, if present. Foremost, the interpretability of each of the networks was thoroughly studied using local interpretability methods - occlusion, saliency, input X gradient, guided backpropagation, integrated gradients, and DeepLIFT, and using a global technique - neuron activation profiles. The mean Micro-F1 score of the models for COVID-19 classifications ranges from 0.66 to 0.875, and is 0.89 for the Ensemble of the network models. The qualitative results depicted the ResNets to be the most interpretable models. This research demonstrates the importance of using interpretability methods to compare different models before making the decision regarding the best-performing model. | ['Andreas Nürnberger', 'Oliver Speck', 'Georg Rose', 'Petia Radeva', 'Nirja Desai', 'Rahul Mishra', 'Rupali Khatun', 'Valerie Krug', 'Chompunuch Sarasaen', 'Sebastian Stober', 'Soumick Chatterjee', 'Suhita Ghosh', 'Fatima Saad'] | 2020-06-03 | null | null | null | null | ['pneumonia-detection', 'interpretability-techniques-for-deep-learning'] | ['medical', 'miscellaneous'] | [-2.24239323e-02 2.29484871e-01 -1.76140529e-04 -1.40375242e-01
2.60712564e-01 -1.10965818e-01 2.23769531e-01 3.52105111e-01
-5.08728445e-01 7.24322915e-01 9.75561664e-02 -4.67215925e-01
-8.36324215e-01 -4.42363858e-01 -1.64094433e-01 -6.67091846e-01
-2.77945757e-01 8.39815378e-01 -1.64946079e-01 -1.46943703e-01
1.26679748e-01 7.33038187e-01 -1.19470382e+00 4.74480897e-01
9.09819663e-01 9.14387226e-01 4.10844982e-01 7.14295030e-01
5.04148453e-02 1.13058984e+00 -5.39634407e-01 -1.93295121e-01
-1.58966742e-02 -3.62902850e-01 -8.20934236e-01 -4.25083786e-01
-2.68319547e-01 -4.89176095e-01 4.48259637e-02 7.52690434e-01
5.58166385e-01 1.05065797e-02 9.38355029e-01 -1.26130068e+00
-5.92927992e-01 3.71263862e-01 -3.24109793e-01 8.57146204e-01
8.09890181e-02 2.48931602e-01 6.21198952e-01 -6.24375165e-01
7.37705290e-01 1.23601520e+00 8.90110195e-01 5.12816548e-01
-7.95001030e-01 -7.10408330e-01 1.98845584e-02 5.05055845e-01
-1.13990378e+00 3.38737071e-01 3.98890436e-01 -7.43677020e-01
1.30003035e+00 4.22944874e-01 9.08925116e-01 1.17116952e+00
7.39062846e-01 1.52512953e-01 1.29723716e+00 -1.22741751e-01
6.74092025e-02 2.23132133e-01 4.09215212e-01 8.05571258e-01
6.41355872e-01 1.78972289e-01 3.76179218e-02 -2.02353656e-01
6.25028312e-01 6.96775973e-01 -3.36264133e-01 4.39791381e-01
-1.16538095e+00 9.95441437e-01 9.70372438e-01 7.08353341e-01
-8.02481890e-01 -6.55253008e-02 4.50769722e-01 2.47222006e-01
6.50309145e-01 1.03313911e+00 -4.93442625e-01 2.10574031e-01
-6.49663806e-01 9.60178599e-02 5.39419472e-01 8.02624226e-02
3.03354740e-01 1.21985875e-01 -2.52529562e-01 7.01330662e-01
4.52499300e-01 5.13443410e-01 6.32011354e-01 -7.61250734e-01
2.22436845e-01 7.68824816e-01 -2.51725078e-01 -1.47318280e+00
-9.86212194e-01 -5.15472114e-01 -1.25973248e+00 2.86375135e-01
-1.24688588e-01 -4.44937885e-01 -1.20205307e+00 1.41001236e+00
-9.20902342e-02 1.13025755e-01 -5.59613928e-02 9.37868953e-01
9.80294466e-01 6.52911663e-01 4.35126275e-01 5.38797714e-02
1.40060437e+00 -1.00340760e+00 -7.22174704e-01 -2.06634700e-01
5.38459301e-01 -2.09259510e-01 5.16492903e-01 3.81690592e-01
-7.46138811e-01 -4.64917034e-01 -1.07430100e+00 5.77587664e-01
-6.43864989e-01 -1.28567711e-01 3.54005665e-01 3.06545317e-01
-1.31963742e+00 7.14559197e-01 -7.09917128e-01 -6.04732871e-01
4.59028870e-01 5.70763290e-01 -1.49603665e-01 -7.52895847e-02
-1.33377206e+00 1.58740401e+00 5.63805878e-01 1.59400746e-01
-9.66097534e-01 -5.35801291e-01 -3.94045591e-01 7.51034468e-02
-1.50868595e-01 -1.07390761e+00 6.68635964e-01 -1.09967005e+00
-7.47138560e-01 8.24387491e-01 1.93450764e-01 -5.79223692e-01
4.18678045e-01 -8.00974667e-02 -4.14483845e-01 6.15178287e-01
-7.76035786e-02 9.33329463e-01 7.05414832e-01 -1.27929211e+00
-6.14034176e-01 -3.02560389e-01 9.79754552e-02 2.47053266e-01
-2.51908720e-01 3.60686183e-01 2.84541637e-01 -6.95950508e-01
-5.20079657e-02 -9.52326238e-01 -3.07378501e-01 -2.26440936e-01
-4.52038258e-01 -4.54799831e-01 9.50736225e-01 -1.02092671e+00
1.02122140e+00 -2.08728099e+00 -1.98249463e-02 3.09226453e-01
9.09311414e-01 4.49256748e-01 1.36758938e-01 8.28683227e-02
-3.53244334e-01 6.20062470e-01 -2.20439732e-01 1.15636224e-02
-3.85999173e-01 2.94905603e-01 3.58007520e-01 3.62159640e-01
3.77678126e-01 9.06784892e-01 -6.88330114e-01 -5.37463009e-01
4.77420270e-01 7.57078469e-01 -3.11382115e-01 2.93483198e-01
7.27183595e-02 6.92309678e-01 -4.86593604e-01 5.28539240e-01
4.89621162e-01 -7.80803621e-01 -1.16624720e-01 1.08384356e-01
1.70092613e-01 -2.36657470e-01 -5.31089723e-01 8.31948936e-01
-3.48331183e-01 7.76830912e-01 7.76923671e-02 -9.32596684e-01
5.75850844e-01 7.34425664e-01 6.46996975e-01 -4.70771343e-01
4.17575151e-01 2.43244529e-01 3.79865229e-01 -1.07181680e+00
-7.61039108e-02 -1.28781512e-01 4.76381958e-01 5.49378693e-01
-1.55749515e-01 3.66370082e-01 -4.51645069e-02 -2.76921038e-02
1.10567713e+00 -5.84140778e-01 3.87111396e-01 -3.36419582e-01
3.54539037e-01 1.41158447e-01 1.63225040e-01 7.46950150e-01
-5.22744656e-01 5.82543910e-01 3.10870051e-01 -7.47429848e-01
-9.22407925e-01 -9.68469381e-01 -1.35727525e-01 8.59405637e-01
-5.54542653e-02 3.52053136e-01 -8.43074203e-01 -6.04533613e-01
-2.11744443e-01 8.07148695e-01 -8.34642529e-01 -3.07879031e-01
-5.66346049e-01 -8.11693966e-01 6.32800698e-01 4.14322317e-01
6.06610894e-01 -1.60844779e+00 -1.04007816e+00 5.58897965e-02
-2.42618874e-01 -6.98057473e-01 3.05645734e-01 3.72964650e-01
-9.41947281e-01 -1.18435824e+00 -8.35128903e-01 -9.25723076e-01
7.87729383e-01 -1.00419067e-01 1.05362630e+00 5.22736430e-01
-4.37665790e-01 2.49148533e-01 -2.50616968e-01 -5.91382802e-01
-4.59252775e-01 7.17862323e-02 9.28005110e-03 -4.13667500e-01
3.50319654e-01 -2.98293054e-01 -6.70672536e-01 -2.44486891e-02
-7.94075251e-01 8.19130614e-02 6.59116149e-01 6.62987292e-01
2.73894697e-01 7.23614022e-02 6.97581410e-01 -8.17189336e-01
9.69625056e-01 -9.71729040e-01 1.74001649e-01 1.85336426e-01
-8.72193694e-01 -4.07956064e-01 5.54917216e-01 -1.82638332e-01
-9.30604160e-01 -5.70392549e-01 -2.56954879e-01 -6.69927418e-01
-4.99161839e-01 5.81638396e-01 6.04008079e-01 8.19809139e-02
7.70318866e-01 -1.99277550e-01 4.61062416e-02 -2.44828135e-01
-1.56755686e-01 6.95862651e-01 1.84547022e-01 1.53011099e-01
2.92101443e-01 2.88159162e-01 -4.04360034e-02 -8.84553850e-01
-5.83086014e-01 -2.96862870e-01 -4.66381609e-01 -5.70379376e-01
1.62231696e+00 -7.15219915e-01 -6.62542224e-01 5.67526221e-01
-1.38718522e+00 4.33273464e-02 -2.81917751e-02 6.72749281e-01
4.34028618e-02 -7.01591969e-02 -7.34667301e-01 -5.94062924e-01
-7.92132080e-01 -1.30126691e+00 3.89528424e-01 3.23714197e-01
-5.69462121e-01 -1.34595025e+00 -1.19559709e-02 4.02595073e-01
8.01907659e-01 5.82358539e-01 1.23676991e+00 -1.19987905e+00
-2.10963920e-01 3.19912215e-03 -5.26660442e-01 5.47422886e-01
1.46900177e-01 -5.51588461e-02 -9.35862780e-01 -1.29914939e-01
4.07161057e-01 -2.68091351e-01 9.14494336e-01 8.35282981e-01
1.09969401e+00 -4.75998849e-01 -5.70739985e-01 4.72742081e-01
1.45024061e+00 7.59576321e-01 2.34890983e-01 2.85746902e-01
8.80396008e-01 6.54322386e-01 8.76492485e-02 2.48035654e-01
2.65757233e-01 -3.29389386e-02 8.09830427e-01 -3.39261651e-01
3.34275886e-02 4.61567402e-01 1.93520654e-02 8.47464442e-01
-5.45246601e-01 -5.16346216e-01 -1.26570940e+00 3.75211358e-01
-1.46159351e+00 -9.15023267e-01 -2.52952307e-01 1.46569932e+00
2.19879255e-01 1.28837079e-01 -2.67217338e-01 1.20696098e-01
9.25509751e-01 -5.26593812e-02 -4.19830233e-01 -7.09142685e-01
-6.98424727e-02 2.10583240e-01 1.42089278e-01 1.97558850e-01
-1.16709912e+00 2.23611251e-01 6.61611843e+00 3.56922060e-01
-1.38616920e+00 3.55822802e-01 1.11866891e+00 5.31589016e-02
2.31134333e-02 -6.25130773e-01 -2.31741607e-01 6.25773191e-01
1.12707770e+00 2.81647652e-01 3.80108088e-01 8.99505198e-01
4.77583498e-01 -3.61903086e-02 -6.44863605e-01 7.94131517e-01
1.72540396e-01 -1.30358028e+00 -5.78431785e-03 -2.21639320e-01
7.99943566e-01 5.45055032e-01 5.54566979e-02 1.55866265e-01
1.49288625e-01 -1.29527676e+00 3.07458133e-01 6.80463850e-01
5.99145114e-01 -5.97631574e-01 1.19333565e+00 2.44435042e-01
-8.06609094e-01 -3.20914656e-01 -6.37227893e-02 -3.22335958e-03
1.35031883e-02 3.99545610e-01 -1.28285182e+00 3.27228159e-01
1.10690153e+00 4.76693124e-01 -5.06820619e-01 8.35841477e-01
-2.00924516e-01 7.43511140e-01 -2.77178615e-01 -3.71934563e-01
4.59099382e-01 5.52957132e-02 6.81882560e-01 1.41007698e+00
1.62957251e-01 1.53563961e-01 1.50786802e-01 8.60398650e-01
2.56017536e-01 -1.36339758e-02 -8.05816472e-01 1.95716955e-02
1.08976975e-01 1.07961690e+00 -1.10145700e+00 -5.71713448e-01
-6.34495839e-02 7.73066163e-01 -7.79228657e-02 3.07412565e-01
-9.67790008e-01 -2.59697407e-01 3.42565715e-01 7.82403126e-02
-3.86769330e-04 3.26907426e-01 -7.36621737e-01 -4.30943906e-01
-4.72717702e-01 -8.39526892e-01 5.08259296e-01 -1.06671107e+00
-1.32829142e+00 1.10524321e+00 2.13220417e-01 -7.79779136e-01
-2.80459851e-01 -7.49477386e-01 -6.80459321e-01 9.60883200e-01
-1.28060508e+00 -8.34202588e-01 -5.73971868e-01 4.50275451e-01
4.31814730e-01 -2.20369071e-01 9.20183659e-01 3.49806398e-01
-6.26720905e-01 -6.24379031e-02 5.37084788e-02 7.23825991e-02
1.64429788e-02 -1.08082449e+00 -1.46256834e-01 4.83670086e-01
-6.25647247e-01 4.75398421e-01 4.74600375e-01 -8.20097387e-01
-3.47922772e-01 -1.21458542e+00 1.02835846e+00 -1.34981245e-01
2.15854138e-01 2.97512501e-01 -8.63877058e-01 4.79563296e-01
6.64597929e-01 -3.96541953e-01 5.64163804e-01 -4.54127312e-01
2.41996095e-01 3.11386764e-01 -1.50702667e+00 4.02865320e-01
6.58388615e-01 -2.42674246e-01 -8.20797384e-01 4.74508703e-01
8.79133403e-01 8.25872738e-03 -8.01343381e-01 5.32950163e-01
1.92490160e-01 -1.09405971e+00 9.70551252e-01 -6.49895966e-01
6.63645983e-01 1.59236491e-01 2.13267565e-01 -1.38925123e+00
-7.78227448e-01 1.04733437e-01 8.93247128e-02 5.15640259e-01
5.61776996e-01 -1.07113934e+00 3.96174133e-01 3.23187530e-01
-3.36171001e-01 -1.12739074e+00 -6.56893969e-01 -2.81730860e-01
-2.19314262e-01 -3.30523819e-01 3.98513854e-01 1.21481574e+00
-3.82619143e-01 1.68150723e-01 -1.22424155e-01 -3.08586136e-02
2.12585732e-01 -3.76149178e-01 -1.01059482e-01 -1.55175889e+00
-4.15375642e-02 -6.54720247e-01 -3.78137946e-01 -9.22183320e-02
-2.02998042e-01 -1.05404294e+00 -3.16149563e-01 -2.16209269e+00
2.45216191e-01 -3.83747995e-01 -7.57726908e-01 6.98910058e-01
-1.59107253e-01 1.67438284e-01 2.23816350e-01 3.77925277e-01
-3.04306805e-01 -1.01901114e-01 1.39126325e+00 -1.89583957e-01
-4.65017296e-02 2.69716289e-02 -5.44845581e-01 9.63345408e-01
1.27370155e+00 -7.48300552e-01 -5.59057772e-01 -4.19557571e-01
2.03420341e-01 -5.73005565e-02 4.47011948e-01 -1.14838338e+00
1.28693390e-03 -1.10535972e-01 6.79942489e-01 -6.54172659e-01
3.12629044e-01 -1.00407112e+00 3.53223592e-01 1.21263158e+00
-2.83522099e-01 6.23384178e-01 1.82911992e-01 1.64291695e-01
-7.76642114e-02 -3.42493176e-01 8.52999270e-01 -4.30232048e-01
-5.65356135e-01 1.46202520e-01 -8.21673810e-01 1.99016795e-01
1.21987915e+00 -2.15698063e-01 -3.78461957e-01 -2.04086944e-01
-9.84044850e-01 4.93074246e-02 -1.04639279e-02 3.42642486e-01
8.57035518e-01 -8.74768555e-01 -7.68042505e-01 1.90526456e-01
-2.25773960e-01 -4.67372946e-02 4.43493336e-01 1.14716434e+00
-1.33284724e+00 6.09191060e-01 -5.82332253e-01 -7.19100535e-01
-1.13482559e+00 4.55923080e-01 5.95106483e-01 -6.43313169e-01
-5.21314979e-01 8.31982195e-01 1.85091853e-01 -2.62235105e-01
2.17905179e-01 -4.70474690e-01 -7.00775266e-01 1.44108489e-01
2.22588301e-01 7.90538311e-01 6.60059452e-02 -5.86739540e-01
-5.35378158e-01 3.02589357e-01 6.56253542e-04 2.82726735e-01
1.38719797e+00 1.46026939e-01 -5.10228813e-01 2.96343803e-01
1.18419635e+00 -7.92467773e-01 -5.29716492e-01 4.52497602e-01
-1.06013894e-01 1.09360985e-01 1.01114763e-03 -1.21501219e+00
-1.19142938e+00 9.00596261e-01 1.02500904e+00 5.79978406e-01
1.17413306e+00 -1.20520070e-01 4.74375069e-01 4.36941683e-01
-1.03852496e-01 -8.90190303e-01 1.34177342e-01 4.19076443e-01
8.52219701e-01 -1.23696792e+00 -1.70669764e-01 4.90402840e-02
-7.56012976e-01 9.21253681e-01 6.50119781e-01 -1.88439026e-01
1.02604151e+00 6.43701926e-02 3.15575093e-01 -7.22325504e-01
-6.62806273e-01 3.22529078e-01 2.58545578e-01 7.31987655e-01
2.94834375e-01 2.81765521e-01 -2.08910793e-01 4.39620972e-01
7.55008981e-02 -4.53081429e-02 7.46586919e-02 8.18006158e-01
-3.96552861e-01 -2.61034161e-01 -6.32779419e-01 1.08246899e+00
-8.18578243e-01 -8.85455832e-02 -4.30925667e-01 8.91827762e-01
6.07560933e-01 1.00997245e+00 9.41957533e-02 -6.50172114e-01
9.60010365e-02 4.75059114e-02 -1.22002520e-01 -3.98807108e-01
-1.17995012e+00 -3.37120652e-01 -3.52983288e-02 -1.62315249e-01
-5.61332941e-01 -3.69914293e-01 -1.54344046e+00 -3.43277991e-01
-2.83842713e-01 -1.16782248e-01 7.27696896e-01 1.04449260e+00
3.11210930e-01 1.09465122e+00 4.17510480e-01 -7.43299425e-01
-2.53504187e-01 -1.13870406e+00 -2.79211044e-01 4.00819093e-01
3.52448225e-01 -5.99603176e-01 -5.93418300e-01 -1.84417844e-01] | [15.53707504272461, -1.7382797002792358] |
daee3ff7-ea19-45f3-ab88-1d56abb543f9 | fast-submodular-function-maximization | 2305.08367 | null | https://arxiv.org/abs/2305.08367v1 | https://arxiv.org/pdf/2305.08367v1.pdf | Fast Submodular Function Maximization | Submodular functions have many real-world applications, such as document summarization, sensor placement, and image segmentation. For all these applications, the key building block is how to compute the maximum value of a submodular function efficiently. We consider both the online and offline versions of the problem: in each iteration, the data set changes incrementally or is not changed, and a user can issue a query to maximize the function on a given subset of the data. The user can be malicious, issuing queries based on previous query results to break the competitive ratio for the online algorithm. Today, the best-known algorithm for online submodular function maximization has a running time of $O(n k d^2)$ where $n$ is the total number of elements, $d$ is the feature dimension and $k$ is the number of elements to be selected. We propose a new method based on a novel search tree data structure. Our algorithm only takes $\widetilde{O}(nk + kd^2 + nd)$ time. | ['Yitan Wang', 'Zhao Song', 'Lianke Qin'] | 2023-05-15 | null | null | null | null | ['document-summarization'] | ['natural-language-processing'] | [ 1.53614745e-01 1.54963523e-01 -7.18326628e-01 -9.00608152e-02
-5.74709177e-01 -9.88803208e-01 -4.20177639e-01 5.30126929e-01
-4.91574287e-01 6.31964564e-01 -3.08176368e-01 -2.80008465e-01
-3.50275010e-01 -8.69349897e-01 -6.64936900e-01 -8.30336571e-01
-2.98951685e-01 7.37641633e-01 5.30613005e-01 -2.95027550e-02
2.72587866e-01 5.18110335e-01 -1.08842015e+00 -5.29191315e-01
6.97143137e-01 1.37471700e+00 2.74317563e-01 5.27842402e-01
-1.79989234e-01 2.51928002e-01 -5.86235464e-01 -1.10746853e-01
7.53981650e-01 -2.26429328e-01 -8.16056013e-01 6.08123124e-01
4.30944702e-03 -4.02586102e-01 -6.74428582e-01 1.36657906e+00
4.77359235e-01 1.72780976e-01 -2.04194803e-02 -1.48127282e+00
-3.52624118e-01 6.10941350e-01 -8.05908680e-01 1.79385230e-01
2.47864753e-01 -1.44062415e-01 1.12181723e+00 -3.32770646e-01
9.01830554e-01 8.41709971e-01 -1.90163448e-01 3.91902477e-01
-9.82549429e-01 -6.21205032e-01 4.20975149e-01 2.70994931e-01
-1.27199566e+00 -9.44908559e-02 7.89460778e-01 -2.05597151e-02
4.59900528e-01 5.41953325e-01 6.64916158e-01 -2.25534096e-01
-1.35451809e-01 9.22532082e-01 5.52179098e-01 -1.04279459e-01
7.12850869e-01 1.17085941e-01 -3.82260196e-02 5.35670519e-01
4.51334566e-01 -3.56951296e-01 -6.21973574e-01 -4.13367391e-01
2.30481401e-01 1.05739608e-01 -4.94131923e-01 -5.65878451e-01
-7.48339772e-01 9.33347285e-01 3.65075767e-01 1.31770922e-02
-2.62923837e-01 3.67218643e-01 1.15539700e-01 6.32621825e-01
1.60370678e-01 3.24426204e-01 -5.68741500e-01 -1.92924991e-01
-9.07884181e-01 4.58201319e-01 9.18506324e-01 1.00523269e+00
8.33577573e-01 -2.88431257e-01 1.65243745e-01 5.46692431e-01
-3.10114231e-02 6.50491416e-01 -2.20500659e-02 -1.14074922e+00
7.41167843e-01 8.46430361e-01 2.25507900e-01 -1.24255931e+00
-1.98237911e-01 4.97491062e-02 -5.04166901e-01 -7.53762275e-02
1.84055611e-01 -3.75793874e-01 -7.12129056e-01 1.53027093e+00
9.42286730e-01 -4.10738170e-01 -2.71938592e-01 8.85803998e-01
4.94648933e-01 8.85887563e-01 -9.61921155e-01 -9.06991005e-01
8.77002656e-01 -6.98205769e-01 -7.56578505e-01 -1.93477631e-01
5.86145580e-01 -3.23252171e-01 3.17005485e-01 5.49454153e-01
-1.24877024e+00 3.14739436e-01 -1.14249074e+00 9.56937298e-02
-2.52511859e-01 -3.88110489e-01 4.92380977e-01 7.40245223e-01
-9.86689448e-01 3.57920676e-02 -6.28872454e-01 -4.67176773e-02
2.72451133e-01 8.89889777e-01 -7.62543008e-02 -4.35876966e-01
-7.10904181e-01 2.26719573e-01 3.73845160e-01 5.00458628e-02
-7.76580870e-01 -2.28305325e-01 -6.64778531e-01 -6.71616197e-02
1.14356112e+00 -1.96893394e-01 9.86275613e-01 -2.76070178e-01
-1.00714648e+00 7.46053338e-01 -3.83215934e-01 -4.09155339e-01
3.64333063e-01 3.64366204e-01 8.19847286e-02 3.24615806e-01
1.06364220e-01 4.24820483e-01 6.71990335e-01 -9.88152742e-01
-9.12657142e-01 -1.00671458e+00 3.02868694e-01 4.29340988e-01
-6.08537853e-01 -1.01936586e-01 -9.72804666e-01 -4.49350178e-02
6.10436559e-01 -1.08434081e+00 -5.60197771e-01 3.87704670e-01
-5.76317489e-01 -2.02498406e-01 1.02371693e+00 -4.12162513e-01
1.55715668e+00 -2.29890370e+00 3.18413436e-01 7.21669197e-01
4.26446557e-01 -1.29354820e-01 8.77245963e-02 5.90120256e-01
6.14967704e-01 9.24591199e-02 -3.07681680e-01 -7.31569901e-02
-1.53339468e-02 3.10064048e-01 -1.52908295e-01 6.69830441e-01
-8.54429185e-01 5.12772858e-01 -5.88022470e-01 -2.15857983e-01
-3.36689740e-01 -4.01250333e-01 -5.10786772e-01 -2.51069188e-01
-5.31009257e-01 -2.86834955e-01 -6.16238654e-01 6.53874695e-01
1.13330615e+00 -3.66314530e-01 5.04468977e-01 4.19456601e-01
-5.40204858e-03 -3.75393629e-02 -1.77958870e+00 1.47243512e+00
1.89986906e-03 4.60699528e-01 6.36290133e-01 -9.95858967e-01
7.25680470e-01 -8.14417750e-02 9.57653761e-01 -5.64296067e-01
1.73578769e-01 4.36677277e-01 -3.62097740e-01 -2.31047541e-01
6.70872390e-01 2.72497296e-01 -2.59015530e-01 7.84210443e-01
-7.00396061e-01 -7.32056275e-02 3.56168479e-01 3.04362893e-01
1.37664139e+00 -9.86699879e-01 1.05702981e-01 -6.65396964e-03
3.15396577e-01 2.79807180e-01 6.84686005e-01 6.16688609e-01
-6.14775755e-02 2.72882819e-01 9.82745886e-01 -2.60194957e-01
-5.56380093e-01 -4.99463081e-01 1.06333882e-01 1.00020766e+00
9.54103410e-01 -1.56727523e-01 -7.23261833e-01 -6.96720302e-01
5.24383605e-01 3.77067178e-01 -5.51662683e-01 9.41431336e-03
-3.19625229e-01 -5.16523540e-01 -9.65344980e-02 1.27863944e-01
4.69538480e-01 -7.42807627e-01 -7.33296812e-01 2.78502852e-01
9.29476693e-02 -1.10994971e+00 -1.12609076e+00 2.47240037e-01
-9.65546072e-01 -1.03096366e+00 -4.13980186e-01 -9.07716811e-01
1.06343246e+00 6.84357762e-01 3.35655272e-01 3.29646817e-03
-2.35347569e-01 3.02350670e-01 -2.62618661e-01 -3.54201794e-01
1.08077370e-01 1.33891925e-01 -8.36504847e-02 -1.51798874e-01
1.68457255e-01 -1.03628159e-01 -5.74032307e-01 3.60499799e-01
-1.26895285e+00 -6.44817650e-01 9.38717052e-02 3.62991393e-01
1.04135776e+00 6.64710224e-01 2.74781287e-01 -9.23164725e-01
5.83229244e-01 -4.68839020e-01 -1.17824483e+00 1.84589416e-01
-7.42754400e-01 -2.39046961e-02 4.50581044e-01 -3.01085234e-01
-2.70081967e-01 3.12242001e-01 3.84123534e-01 -2.95218021e-01
6.80948794e-01 7.69354105e-01 -4.48625624e-01 -1.78636521e-01
1.62230328e-01 5.48045814e-01 1.37568340e-01 -2.57339150e-01
2.18094438e-01 7.49936700e-01 2.35479146e-01 -1.07796408e-01
7.67055273e-01 8.39847922e-01 2.06092805e-01 -7.42194891e-01
-3.97677600e-01 -5.64187050e-01 -1.05966501e-01 7.78530538e-02
3.81523967e-01 -6.08477950e-01 -1.07727754e+00 5.40685892e-01
-9.05184031e-01 -1.14283927e-01 -4.20392454e-01 2.33300142e-02
-2.04550564e-01 4.38168198e-01 3.79050635e-02 -8.82073939e-01
-3.19095254e-01 -1.08861101e+00 5.88734925e-01 2.49135688e-01
3.39271694e-01 -7.40627050e-01 -1.15742952e-01 5.73529363e-01
1.60848707e-01 3.10811639e-01 8.75178754e-01 -4.99974877e-01
-1.01370811e+00 -5.13536155e-01 -1.19109184e-01 3.19111673e-03
1.90528795e-01 -4.07578886e-01 -9.91433337e-02 -7.44343877e-01
5.08687384e-02 -1.29345760e-01 4.83067125e-01 3.78640682e-01
1.49861407e+00 -7.84482837e-01 -5.10110259e-01 5.68492234e-01
1.41059017e+00 6.28812253e-01 3.90506566e-01 2.35051274e-01
3.88043106e-01 2.06565887e-01 8.80161941e-01 1.01762486e+00
5.82234085e-01 5.67045391e-01 9.52464044e-01 2.42758408e-01
6.18501365e-01 -9.10137966e-02 3.05678904e-01 3.60748768e-01
6.45775497e-01 -7.91418910e-01 -4.74880338e-01 8.25534642e-01
-1.89175487e+00 -6.22325063e-01 1.58321619e-01 2.48914957e+00
8.46007049e-01 6.81381673e-02 2.57733017e-01 1.60484150e-01
7.74178743e-01 3.55457783e-01 -1.31039381e+00 -6.04016066e-01
6.95716590e-03 -1.90063089e-01 1.13898468e+00 5.49212813e-01
-6.85260534e-01 6.38473570e-01 5.39229774e+00 7.07233071e-01
-1.19386828e+00 -1.62268847e-01 6.48927689e-01 -7.51615584e-01
-6.63013399e-01 6.12219833e-02 -9.70054865e-01 6.04050815e-01
2.87025243e-01 -6.49153471e-01 7.86424100e-01 9.85054374e-01
1.45183891e-01 -6.70355916e-01 -1.02709150e+00 1.31153595e+00
2.41737023e-01 -1.38697493e+00 -2.61537999e-01 6.97003126e-01
1.07798851e+00 -1.69571027e-01 1.01147451e-01 -3.78091872e-01
5.26982360e-02 -7.14434445e-01 3.65074754e-01 -4.03823376e-01
7.30594873e-01 -1.01357317e+00 3.63685042e-01 7.53863811e-01
-1.16394508e+00 -4.32372183e-01 -4.29617435e-01 3.15508366e-01
2.35436961e-01 7.27745116e-01 -7.02847481e-01 2.85055131e-01
8.00473332e-01 1.16320476e-01 9.04856175e-02 1.08732855e+00
1.83031455e-01 -3.71631049e-02 -1.04412568e+00 -4.69944119e-01
2.34494522e-01 -2.27693468e-01 8.23002815e-01 3.91899526e-01
2.50806063e-01 4.21015263e-01 5.18412292e-01 2.43389755e-01
-5.21310151e-01 1.17226169e-01 -4.66516256e-01 -2.55384624e-01
9.46458876e-01 9.35625017e-01 -7.42773235e-01 9.79911238e-02
-9.66285095e-02 1.07595336e+00 -6.39064284e-03 5.49236313e-02
-6.67502046e-01 -6.76790059e-01 5.78062415e-01 3.86730433e-01
6.73004508e-01 -4.90920275e-01 -4.85354394e-01 -5.95208704e-01
5.73357522e-01 -6.74553990e-01 7.54392862e-01 -4.09504585e-02
-6.34598136e-01 -1.70619246e-02 -1.44688785e-01 -8.66770744e-01
4.84171771e-02 -1.86814651e-01 -2.91953206e-01 3.19843292e-01
-1.24819136e+00 -2.68839568e-01 -2.67733276e-01 6.53335273e-01
1.57662034e-01 -1.33826043e-02 1.90642729e-01 1.29783407e-01
-4.01237726e-01 7.09465683e-01 4.87012446e-01 -3.01035106e-01
1.19818926e-01 -9.83885884e-01 -2.46793583e-01 8.28922868e-01
-1.67862758e-01 1.45186514e-01 5.93688548e-01 -5.38191795e-01
-1.97657084e+00 -6.13587379e-01 9.61204171e-01 1.90505356e-01
3.21582735e-01 -2.63165444e-01 -4.49425906e-01 2.94706911e-01
-3.28298926e-01 1.63617224e-01 6.14435971e-01 -5.09931922e-01
2.05978185e-01 -6.79183066e-01 -1.63950741e+00 3.94122899e-01
1.00443363e+00 -1.08300216e-01 3.29859197e-01 5.51733315e-01
7.43618608e-01 -7.56282449e-01 -6.18524671e-01 1.69576511e-01
2.52519131e-01 -5.54240465e-01 4.36728805e-01 -2.67081618e-01
-9.92050171e-02 -4.35094297e-01 -3.33416998e-01 -8.17519367e-01
7.72936419e-02 -1.12456679e+00 -5.20693600e-01 6.24411881e-01
5.29811144e-01 -7.26071894e-01 1.38653839e+00 9.54192877e-01
4.51369762e-01 -1.18673313e+00 -1.44081831e+00 -6.93829060e-01
-3.10971677e-01 -1.11913711e-01 7.51122594e-01 6.81632638e-01
8.39991793e-02 -1.03552409e-01 -3.31236959e-01 4.69739556e-01
5.59145510e-01 6.20505393e-01 9.34110224e-01 -9.51545954e-01
-4.54357088e-01 1.63895972e-02 -2.67391473e-01 -1.83229387e+00
-4.16404516e-01 -6.51443183e-01 -1.67497069e-01 -1.68553483e+00
3.96579504e-01 -6.47263110e-01 3.01559120e-02 5.46692133e-01
1.52529240e-01 -3.01213264e-01 3.60333651e-01 1.23871692e-01
-7.96630144e-01 4.31602091e-01 1.33186853e+00 -1.59812704e-01
-6.87484741e-01 2.50642031e-01 -8.69803727e-01 2.76515841e-01
5.48278511e-01 -6.22012973e-01 -5.28792143e-01 -6.78794801e-01
4.34700608e-01 5.77727377e-01 -2.40710333e-01 -2.97508866e-01
6.95900083e-01 -6.33830070e-01 -3.09263527e-01 -9.32282209e-01
2.68290460e-01 -1.15761447e+00 3.35105658e-02 7.84971893e-01
-1.17219463e-01 1.39243558e-01 -1.01566158e-01 7.97215521e-01
-2.43822917e-01 -4.34988886e-01 7.82391310e-01 8.27063322e-02
-5.55743575e-01 8.66112947e-01 -5.87667972e-02 3.19070406e-02
1.69950891e+00 -6.09943271e-01 -3.47963005e-01 -8.05668414e-01
-4.76018637e-01 1.08928502e+00 5.39234996e-01 3.15789551e-01
7.69690335e-01 -9.60945845e-01 -2.90732384e-01 -1.24138385e-01
-2.82421950e-02 5.85083008e-01 3.61404270e-01 7.15958178e-01
-6.68832898e-01 2.93585360e-01 4.94388402e-01 -3.32776487e-01
-1.37260032e+00 5.16906440e-01 1.23845533e-01 -2.80268669e-01
-6.66485503e-02 1.11128056e+00 -2.11998582e-01 -1.30327821e-01
5.89988053e-01 1.86807662e-01 2.74128377e-01 2.81995893e-01
3.08225214e-01 8.25086534e-01 -4.14508432e-02 -1.42545160e-02
-4.53747630e-01 2.10619479e-01 -3.68884236e-01 -2.56321222e-01
1.33341181e+00 -3.48847002e-01 -5.47028661e-01 -3.83849703e-02
1.34391141e+00 7.68983439e-02 -9.74096835e-01 -4.42789644e-01
-3.07961434e-01 -6.89538360e-01 1.27119452e-01 -3.60541850e-01
-1.48083091e+00 2.15041235e-01 5.46469152e-01 4.43351060e-01
1.31406271e+00 2.90020615e-01 1.27344024e+00 5.74888766e-01
8.65406334e-01 -1.67144668e+00 5.90259731e-02 3.05416524e-01
5.62002361e-01 -8.33689809e-01 4.06467199e-01 -5.32986343e-01
-4.28909928e-01 9.38500166e-01 5.83837152e-01 8.69346410e-02
7.87488818e-01 2.16201857e-01 -3.31498533e-01 -2.32435092e-01
-6.65640891e-01 1.07964873e-01 -2.78627932e-01 -1.75981410e-02
-6.00376487e-01 1.49704382e-01 -6.37063146e-01 3.13205451e-01
-1.18827127e-01 -4.44399342e-02 6.24630034e-01 1.26740873e+00
-9.47697937e-01 -1.19803607e+00 -3.70205998e-01 6.33620381e-01
-4.83435988e-01 3.43575895e-01 -5.10704815e-01 2.23299041e-01
4.04477306e-02 1.02586091e+00 1.05488867e-01 -2.14490175e-01
1.35575786e-01 -6.17685437e-01 2.45987639e-01 -6.15102351e-01
-1.59641474e-01 -1.27078235e-01 -2.14787066e-01 -4.20914650e-01
9.28944424e-02 -6.49220347e-01 -1.54954660e+00 -5.10360301e-01
-4.65206921e-01 2.55288213e-01 9.13451850e-01 5.94499528e-01
4.47448939e-01 -2.50955671e-01 1.40604699e+00 -8.26313570e-02
-7.69773126e-01 -1.82014450e-01 -9.70954180e-01 -1.76232085e-01
2.16459438e-01 -1.92550421e-01 -3.13093543e-01 -3.53154987e-01] | [6.552679538726807, 4.891228199005127] |
888c319a-9039-49c3-aa1f-5d7c69e6c950 | vit-ret-vision-and-recurrent-transformer | 2208.07929 | null | https://arxiv.org/abs/2208.07929v2 | https://arxiv.org/pdf/2208.07929v2.pdf | ViT-ReT: Vision and Recurrent Transformer Neural Networks for Human Activity Recognition in Videos | Human activity recognition is an emerging and important area in computer vision which seeks to determine the activity an individual or group of individuals are performing. The applications of this field ranges from generating highlight videos in sports, to intelligent surveillance and gesture recognition. Most activity recognition systems rely on a combination of convolutional neural networks (CNNs) to perform feature extraction from the data and recurrent neural networks (RNNs) to determine the time dependent nature of the data. This paper proposes and designs two transformer neural networks for human activity recognition: a recurrent transformer (ReT), a specialized neural network used to make predictions on sequences of data, as well as a vision transformer (ViT), a transformer optimized for extracting salient features from images, to improve speed and scalability of activity recognition. We have provided an extensive comparison of the proposed transformer neural networks with the contemporary CNN and RNN-based human activity recognition models in terms of speed and accuracy. | ['Arslan Munir', 'Hayat Ullah', 'James Wensel'] | 2022-08-16 | null | null | null | null | ['gesture-recognition', 'activity-recognition-in-videos'] | ['computer-vision', 'computer-vision'] | [ 7.21551001e-01 -3.11266363e-01 -3.41266930e-01 -1.37559026e-01
-2.05715317e-02 -1.46637991e-01 7.33860672e-01 -2.62110084e-01
-4.45327014e-01 4.68568295e-01 7.27249503e-01 6.94237873e-02
-1.29813358e-01 -7.34943151e-01 -3.25170875e-01 -7.58954346e-01
-2.76984662e-01 8.02204087e-02 3.33385170e-01 1.31586427e-02
3.37948948e-01 7.02243209e-01 -1.88374126e+00 3.49270701e-01
3.31662536e-01 1.17591488e+00 -2.21241474e-01 9.44602847e-01
2.96532474e-02 1.77058721e+00 -6.92565262e-01 1.05218939e-01
-1.47315022e-02 -6.89174533e-01 -6.94054127e-01 1.84133992e-01
-3.88280749e-02 -1.12495638e-01 -7.63460636e-01 3.32279950e-01
3.95725548e-01 4.25674021e-01 5.08110344e-01 -1.08169663e+00
-2.08483741e-01 1.61028668e-01 -1.08429678e-01 8.59030485e-01
6.48788929e-01 1.09967224e-01 5.03148139e-01 -6.40327275e-01
4.07880187e-01 6.27698064e-01 8.22998047e-01 5.19112945e-01
-4.85933006e-01 -3.94307077e-01 -1.96390390e-01 8.18297684e-01
-1.22028399e+00 -4.15463328e-01 8.85680497e-01 -4.89445210e-01
1.35658479e+00 1.73239753e-01 1.44732797e+00 1.52518857e+00
3.33472461e-01 1.20928776e+00 6.35811508e-01 -2.66023308e-01
1.57225057e-01 -4.43851978e-01 1.91945776e-01 7.16801643e-01
7.61281177e-02 2.18115866e-01 -1.04763734e+00 9.65780541e-02
1.04751062e+00 4.39540148e-01 -2.15120301e-01 -1.85353890e-01
-1.21906984e+00 5.59614539e-01 2.03403860e-01 4.81618434e-01
-8.42730045e-01 2.30413824e-01 5.78144789e-01 1.15331814e-01
1.09596059e-01 1.38726294e-01 -3.41010056e-02 -7.98089743e-01
-1.05373669e+00 1.16688088e-01 7.82015979e-01 4.54284668e-01
2.57350415e-01 5.53190708e-01 -6.12297297e-01 7.67908692e-01
1.62288234e-01 1.27170652e-01 1.08649564e+00 -7.71247327e-01
2.20768362e-01 1.22434044e+00 -1.98684275e-01 -8.26656282e-01
-4.41204786e-01 -4.28289026e-01 -7.78095365e-01 6.51963428e-02
3.52188289e-01 8.74128938e-02 -8.44004631e-01 1.17484713e+00
3.51192951e-02 5.14453948e-01 1.33235648e-01 9.48942006e-01
8.08955193e-01 6.20453954e-01 1.19740508e-01 -9.80342180e-02
1.45063674e+00 -1.09874463e+00 -6.65543258e-01 -2.62031645e-01
5.04189491e-01 -2.35800803e-01 5.65622389e-01 4.97343987e-01
-9.64291811e-01 -6.48721755e-01 -1.18751788e+00 -7.88356587e-02
-5.85264146e-01 3.62059563e-01 6.08415902e-01 6.04566872e-01
-6.98693871e-01 6.04542077e-01 -1.14645934e+00 -5.25220215e-01
4.36679602e-01 4.73295152e-01 -2.88549572e-01 4.82581317e-01
-9.81422544e-01 8.51378322e-01 2.32887313e-01 5.88693500e-01
-1.06890857e+00 -1.61755010e-01 -1.01520944e+00 1.41431421e-01
1.65701553e-01 -3.65721256e-01 1.25153053e+00 -1.47762716e+00
-1.57643545e+00 7.45691597e-01 -1.30958036e-01 -9.87112164e-01
4.72222775e-01 -3.55248451e-01 -5.14650345e-01 1.91673726e-01
-2.28561282e-01 7.73355663e-02 6.69579148e-01 -1.22832365e-01
-8.19346011e-01 -4.69449073e-01 -3.33419710e-01 4.07460064e-01
-3.19023073e-01 1.38352558e-01 -3.27930897e-01 -8.67906570e-01
-1.52764991e-01 -8.16432178e-01 4.02907990e-02 -2.05853313e-01
-4.04935926e-02 -5.11610448e-01 1.32391965e+00 -8.20863783e-01
1.21562231e+00 -1.71953452e+00 1.43332809e-01 1.91778257e-01
1.74588978e-01 7.27301836e-01 4.64235991e-02 3.86824250e-01
-1.45131439e-01 -6.42550349e-01 8.44813287e-02 9.81042609e-02
-4.12007511e-01 5.79526901e-01 3.04406658e-02 4.98895705e-01
2.17423230e-01 1.17321634e+00 -6.24632120e-01 -4.23469573e-01
5.47787249e-01 7.06447959e-01 -1.25201559e-02 5.11796057e-01
1.02247536e-01 4.01070416e-01 -4.41050321e-01 8.61489952e-01
-2.82275468e-01 -1.54815197e-01 2.52681047e-01 -1.13591187e-01
-3.47426057e-01 3.35248202e-01 -1.07285428e+00 1.32245469e+00
-2.76273359e-02 1.04999971e+00 -5.01172841e-01 -1.43513846e+00
1.01304913e+00 5.71123123e-01 1.02643645e+00 -9.37258482e-01
3.47569525e-01 -2.52956688e-01 -1.16942346e-01 -8.18005025e-01
6.14471436e-01 3.77244443e-01 -2.62389723e-02 6.76208973e-01
2.94523418e-01 9.11943614e-01 5.15911937e-01 -3.23603809e-01
1.47296405e+00 4.32832927e-01 5.64862132e-01 2.05120191e-01
6.91402435e-01 5.02744131e-03 8.21169496e-01 4.65598017e-01
-4.37066317e-01 2.99009979e-01 2.97702640e-01 -1.04877353e+00
-9.05846536e-01 -9.45089221e-01 5.58591068e-01 1.16312420e+00
-9.26329345e-02 -2.48721704e-01 -7.13408768e-01 -3.94738197e-01
-7.01508403e-01 4.20819707e-02 -7.77518213e-01 -2.86466181e-01
-9.48907971e-01 -5.58871150e-01 9.02070880e-01 1.16988623e+00
1.06299901e+00 -1.68567204e+00 -1.29297233e+00 2.93958157e-01
-4.39962924e-01 -9.66685832e-01 -4.42235589e-01 2.70514995e-01
-8.85450304e-01 -1.53539085e+00 -5.60446143e-01 -8.79578292e-01
3.93794775e-01 1.46986738e-01 9.21985686e-01 -3.59472632e-02
-5.25501013e-01 5.75928092e-01 -4.84748214e-01 -5.28524101e-01
1.79781690e-02 2.03666970e-01 -5.02679572e-02 3.55584234e-01
7.65144825e-01 -7.50832379e-01 -7.26549625e-01 4.23002332e-01
-6.96156621e-01 -8.48511606e-02 8.60633492e-01 7.19048262e-01
3.70213598e-01 -3.52204405e-02 -1.29172787e-01 -4.92260575e-01
7.17631400e-01 -1.60105646e-01 -2.63780892e-01 4.59600002e-01
-4.22413975e-01 6.96945935e-02 2.89338380e-01 -7.53033936e-01
-9.55248058e-01 5.26948750e-01 4.71255481e-02 -4.94585156e-01
-1.00486323e-01 4.32428032e-01 2.92750925e-01 -4.88867536e-02
7.75327384e-01 8.88777375e-01 8.21534023e-02 -3.14142376e-01
-3.30705792e-01 5.98353267e-01 8.03971410e-01 -1.47751406e-01
2.34358788e-01 2.98342586e-01 -7.29618296e-02 -1.11601651e+00
-6.72544062e-01 -7.63795853e-01 -8.11417043e-01 -8.03743243e-01
1.05606771e+00 -8.41984928e-01 -9.97012436e-01 9.45571482e-01
-8.53669882e-01 -3.62131774e-01 -4.37396109e-01 5.67855895e-01
-7.13429689e-01 1.63475215e-01 -8.41620862e-01 -9.75349367e-01
-6.81548178e-01 -6.63474858e-01 8.68098557e-01 5.71840286e-01
-4.44704384e-01 -1.00527143e+00 4.43341643e-01 6.98376894e-01
3.13391656e-01 5.25670767e-01 2.41082430e-01 -6.84755743e-01
-4.65732872e-01 -3.39038432e-01 1.96865976e-01 2.39291891e-01
1.75877750e-01 -9.73409712e-02 -8.24868917e-01 5.75598553e-02
-1.32287905e-01 -2.24096552e-01 6.12978816e-01 5.79689741e-01
1.01664484e+00 -4.30305034e-01 -2.15505362e-01 3.71099025e-01
9.29007888e-01 5.60500860e-01 1.16285670e+00 2.08848909e-01
6.73338771e-01 3.58796000e-01 3.59800607e-01 4.51452434e-01
1.07893474e-01 6.88725412e-01 1.73642099e-01 -4.50985581e-02
-7.22000152e-02 -4.45941478e-01 6.70069993e-01 6.62145197e-01
-9.21726346e-01 -1.37721580e-02 -8.57328355e-01 5.60994387e-01
-2.27929783e+00 -1.58059251e+00 1.18162535e-01 2.02988410e+00
2.28505909e-01 -3.86446007e-02 5.98510802e-01 6.00646973e-01
5.65386415e-01 1.90261796e-01 -5.19608080e-01 -4.83430803e-01
-2.53152587e-02 3.55785191e-01 4.88941014e-01 -1.62453160e-01
-1.22562122e+00 7.53284931e-01 7.22971344e+00 5.28547525e-01
-1.20166612e+00 -7.93003589e-02 1.60202920e-01 -1.37892216e-01
7.11628854e-01 -4.87513691e-01 -4.75598603e-01 3.19045305e-01
1.21738219e+00 1.26187086e-01 3.90862942e-01 9.39701498e-01
5.31305015e-01 -7.20337704e-02 -9.22874331e-01 1.19926250e+00
4.20031160e-01 -1.37730920e+00 1.10483542e-01 1.81476176e-01
2.83562154e-01 -4.19353470e-02 -3.22759539e-01 3.92697483e-01
1.68359175e-01 -1.23611593e+00 5.45013189e-01 9.54022110e-01
9.64420512e-02 -6.54103100e-01 8.23686600e-01 3.96550894e-01
-1.77541602e+00 -4.74880725e-01 1.74729466e-01 -5.16103685e-01
2.78859735e-02 -5.41568138e-02 -7.52340794e-01 5.74538261e-02
9.32004750e-01 1.17098498e+00 -6.10373557e-01 9.08490479e-01
-2.17226878e-01 6.36072457e-01 1.75314378e-02 -4.13403153e-01
2.96860486e-01 -2.62035042e-01 4.67840493e-01 1.24912190e+00
3.16383749e-01 3.73866484e-02 7.12914765e-02 4.27776277e-01
1.01667494e-01 -2.87194669e-01 -6.66999459e-01 -4.52316701e-01
5.20956703e-02 1.09769332e+00 -6.11056685e-01 -3.99355412e-01
-3.52253348e-01 9.74518538e-01 1.37517154e-01 1.22480899e-01
-7.07600415e-01 -4.37302947e-01 7.88657606e-01 2.70370007e-01
3.61358732e-01 -3.13502818e-01 3.00309472e-02 -1.13967323e+00
1.43676475e-01 -8.01632285e-01 7.75406599e-01 -9.31582153e-01
-7.48873353e-01 4.10006911e-01 -1.31582469e-01 -1.71681237e+00
-8.68057847e-01 -7.08902419e-01 -8.40187788e-01 3.47139508e-01
-9.56827700e-01 -1.33144069e+00 -5.88933110e-01 9.43422914e-01
8.91361356e-01 -3.44700783e-01 5.58624089e-01 2.60081887e-01
-6.86954618e-01 2.25789025e-01 -5.33802688e-01 9.08074200e-01
2.19449252e-02 -9.90672171e-01 1.37032345e-01 1.08952761e+00
3.40818673e-01 4.65297401e-01 5.00043452e-01 -4.88180637e-01
-1.41074347e+00 -1.10708225e+00 9.81743217e-01 -4.62635756e-01
2.74001390e-01 -8.96282792e-02 -6.33018315e-01 8.79823923e-01
6.97752461e-02 -1.28714610e-02 8.55375409e-01 -6.32139072e-02
-1.06665000e-01 -1.23362131e-01 -7.89119482e-01 6.39353931e-01
1.25005424e+00 -4.84332949e-01 -8.27164292e-01 -8.77306163e-02
-7.75725394e-02 -2.80573338e-01 -9.18005347e-01 3.11977088e-01
1.01631391e+00 -1.12416220e+00 9.93866205e-01 -6.07299745e-01
1.89002201e-01 -3.29052538e-01 2.65882343e-01 -8.93467426e-01
-3.51493061e-01 -4.51851726e-01 -6.83502793e-01 5.67055225e-01
-5.41418493e-02 -2.76616693e-01 1.40987933e+00 4.11486685e-01
-1.90416291e-01 -5.84075511e-01 -9.11432028e-01 -4.89963055e-01
-9.06519234e-01 -4.17817980e-01 3.15316021e-01 6.91728175e-01
8.30216408e-02 4.21156436e-01 -7.69544840e-01 -1.73105389e-01
3.66936356e-01 -1.56773999e-01 6.82029188e-01 -1.26392174e+00
-5.51777519e-02 -5.14682174e-01 -1.06605768e+00 -9.29111481e-01
-1.95230797e-01 -2.40891352e-01 -1.03454746e-01 -1.42664385e+00
5.19342460e-02 5.45415819e-01 -4.95438606e-01 7.20324397e-01
5.94732761e-01 4.16079491e-01 -1.39179587e-01 1.43336117e-01
-7.76121914e-01 4.62551028e-01 8.60395789e-01 -2.79419601e-01
-5.59368551e-01 5.39645493e-01 -3.41556698e-01 7.34509349e-01
7.12301075e-01 -1.09851323e-01 -4.22904998e-01 2.38576606e-02
-6.77180216e-02 1.91031694e-01 6.22672141e-01 -1.59124291e+00
5.52869320e-01 -1.80814341e-01 8.75647962e-01 -7.13741422e-01
5.25297523e-01 -6.71128035e-01 2.76631832e-01 9.26049471e-01
-4.81357187e-01 2.82527596e-01 -2.53925353e-01 5.36699951e-01
-4.43974644e-01 8.84878635e-02 4.47642714e-01 -2.70540774e-01
-1.17646289e+00 4.30792242e-01 -1.07605624e+00 -3.50621641e-01
9.93763387e-01 -1.08105731e+00 -1.20010622e-01 -3.34080994e-01
-7.73369372e-01 -1.78082082e-02 -2.01968968e-01 7.74677038e-01
9.03512001e-01 -1.51467681e+00 -2.82620579e-01 2.89289385e-01
2.28975207e-01 -4.72907245e-01 2.27901936e-01 8.72354031e-01
-6.64882541e-01 6.12824798e-01 -8.47385824e-01 -5.10972023e-01
-1.65415502e+00 3.93976793e-02 7.17121899e-01 -6.03333116e-01
-7.23413110e-01 2.19106987e-01 -5.27777076e-01 -7.68819153e-02
4.22664672e-01 -2.94827580e-01 -9.54760909e-01 -7.35596493e-02
8.74231219e-01 8.72136652e-01 2.20068134e-02 -8.67249250e-01
-3.43566567e-01 4.30160344e-01 2.24959254e-01 5.59634790e-02
1.36203480e+00 3.05376709e-01 1.69468775e-01 4.18553710e-01
7.40034342e-01 -8.41766179e-01 -1.33976090e+00 -2.23769993e-01
1.92489460e-01 -1.16313510e-01 4.12879959e-02 -4.91645277e-01
-1.08662224e+00 7.91508675e-01 8.72611701e-01 -3.89646133e-03
1.31917119e+00 -3.38388920e-01 8.87754023e-01 5.94088852e-01
3.59741896e-02 -1.59505057e+00 5.07484913e-01 4.57937300e-01
6.89870000e-01 -7.51765072e-01 -1.55333951e-01 1.20495304e-01
-7.41052270e-01 1.28613067e+00 6.74922764e-01 -3.16635072e-01
5.52003562e-01 5.52092791e-02 -8.96418095e-02 -3.79201829e-01
-8.02492023e-01 -3.13622147e-01 5.16396046e-01 8.64068270e-01
4.03872311e-01 -1.13693580e-01 -9.84695852e-02 3.81482840e-01
-7.07352012e-02 6.34131849e-01 3.78315955e-01 1.27988887e+00
-5.28380454e-01 -7.68585205e-01 -3.11194032e-01 7.23546207e-01
-3.56658638e-01 5.08059442e-01 -4.81854171e-01 4.82188463e-01
1.99052364e-01 8.36357832e-01 3.31890196e-01 -5.23253679e-01
4.44124997e-01 6.40220866e-02 4.12332565e-01 -2.95126349e-01
-9.41179931e-01 -2.99391270e-01 1.97809517e-01 -9.77757275e-01
-9.40608919e-01 -7.51709461e-01 -9.19217885e-01 -2.38720953e-01
3.37843001e-01 1.11521251e-01 4.40298796e-01 1.30341697e+00
7.28115514e-02 6.87413812e-01 2.86064893e-01 -8.75867844e-01
6.48765415e-02 -1.08467519e+00 -5.58315396e-01 2.73618370e-01
1.73017144e-01 -7.05564260e-01 1.36576086e-01 3.62492979e-01] | [8.034012794494629, 0.4710472524166107] |
7621c368-7870-4ddd-8a04-8504800cb765 | g-sto-sequential-main-shopping-intention | 2306.14314 | null | https://arxiv.org/abs/2306.14314v1 | https://arxiv.org/pdf/2306.14314v1.pdf | G-STO: Sequential Main Shopping Intention Detection via Graph-Regularized Stochastic Transformer | Sequential recommendation requires understanding the dynamic patterns of users' behaviors, contexts, and preferences from their historical interactions. Most existing works focus on modeling user-item interactions only from the item level, ignoring that they are driven by latent shopping intentions (e.g., ballpoint pens, miniatures, etc). The detection of the underlying shopping intentions of users based on their historical interactions is a crucial aspect for e-commerce platforms, such as Amazon, to enhance the convenience and efficiency of their customers' shopping experiences. Despite its significance, the area of main shopping intention detection remains under-investigated in the academic literature. To fill this gap, we propose a graph-regularized stochastic Transformer method, G-STO. By considering intentions as sets of products and user preferences as compositions of intentions, we model both of them as stochastic Gaussian embeddings in the latent representation space. Instead of training the stochastic representations from scratch, we develop a global intention relational graph as prior knowledge for regularization, allowing relevant shopping intentions to be distributionally close. Finally, we feed the newly regularized stochastic embeddings into Transformer-based models to encode sequential information from the intention transitions. We evaluate our main shopping intention identification model on three different real-world datasets, where G-STO achieves significantly superior performances to the baselines by 18.08% in Hit@1, 7.01% in Hit@10, and 6.11% in NDCG@10 on average. | ['Chao Zhang', 'Jin Li', 'Ming Wang', 'Chaosheng Dong', 'Yan Zhao', 'Xin Shen', 'Yuchen Zhuang'] | 2023-06-25 | null | null | null | null | ['sequential-recommendation'] | ['miscellaneous'] | [-1.33235723e-01 -3.49339306e-01 -5.04937291e-01 -6.22423053e-01
-2.03532860e-01 -4.68895793e-01 4.36793983e-01 2.13691905e-01
-3.24003458e-01 -2.99192309e-01 6.33304358e-01 -4.58135426e-01
-1.51284337e-01 -9.58346307e-01 -7.52305567e-01 -3.77035290e-01
-4.74132597e-02 4.70101446e-01 1.79870334e-02 -3.87393028e-01
1.43791467e-01 -1.85983092e-01 -1.27038646e+00 3.56082976e-01
1.03956747e+00 1.11011803e+00 2.57202744e-01 1.69464290e-01
-2.72102237e-01 5.98052084e-01 2.16786101e-01 -8.30727339e-01
1.53318092e-01 -3.49325866e-01 -1.05178460e-01 2.97202587e-01
-2.37843871e-01 -2.84336984e-01 -5.61982334e-01 9.94944036e-01
3.27518545e-02 5.39649308e-01 7.43711352e-01 -9.75859225e-01
-1.48983312e+00 1.07185340e+00 -3.80932689e-01 -1.69114932e-01
5.26797354e-01 -2.28184968e-01 1.46605551e+00 -1.19796181e+00
3.68735969e-01 1.30455780e+00 6.03247106e-01 1.61551327e-01
-1.35504043e+00 -4.63151902e-01 7.43772030e-01 1.60971001e-01
-1.33631265e+00 -8.24035779e-02 9.41030443e-01 -3.24184835e-01
8.18403721e-01 2.21558094e-01 4.67741460e-01 1.42662287e+00
1.19642444e-01 1.33284795e+00 3.77687335e-01 -2.09231392e-01
1.83214575e-01 5.22934198e-01 7.97764778e-01 3.37044597e-01
1.88142255e-01 -2.95652151e-01 -3.58746827e-01 -1.57057300e-01
5.68356514e-01 9.14480984e-01 8.18230025e-03 -5.66347957e-01
-8.56278419e-01 1.26792610e+00 4.21528399e-01 5.20304851e-02
-6.28444135e-01 -2.47915760e-01 3.94449413e-01 1.65695712e-01
4.42186326e-01 -8.37766081e-02 -5.56164265e-01 -1.01090670e-01
-3.86966676e-01 1.80592477e-01 8.05312693e-01 1.36225104e+00
3.66583407e-01 -1.97085321e-01 6.74537793e-02 1.04048932e+00
9.32420313e-01 3.57117742e-01 7.69223571e-01 -1.52787209e-01
4.17294085e-01 8.01727951e-01 1.90173775e-01 -1.29261732e+00
-1.33672237e-01 -5.73602736e-01 -6.61192894e-01 -9.93108869e-01
3.08933377e-01 1.17924184e-01 -7.20641732e-01 1.47251308e+00
5.23018390e-02 3.01740263e-02 -2.13036761e-01 8.05101991e-01
6.08174920e-01 7.67044365e-01 1.60357803e-01 -1.81184620e-01
1.43981743e+00 -9.81528163e-01 -7.33978570e-01 -2.94044703e-01
6.85596406e-01 -8.00175786e-01 1.54204381e+00 4.04461145e-01
-7.25436985e-01 -6.77777946e-01 -5.34648836e-01 -9.77321193e-02
-3.95482898e-01 2.97203213e-01 7.88183272e-01 6.76153958e-01
-3.74471694e-01 2.98127830e-01 -8.90172660e-01 -5.02704024e-01
1.09030351e-01 2.71923393e-01 2.10367203e-01 -4.32781845e-01
-1.31420791e+00 3.07224452e-01 1.45381037e-02 2.38693357e-02
-2.08066806e-01 -5.89451194e-01 -9.89560068e-01 1.73990801e-01
4.68561053e-01 -4.25062925e-01 9.56109643e-01 -5.40208459e-01
-1.43523812e+00 4.22718853e-01 -3.76977742e-01 -3.74275416e-01
3.26481164e-01 -4.87655103e-01 -7.88876474e-01 -7.11418390e-01
1.03625573e-01 -4.34847698e-02 6.69865847e-01 -1.11030996e+00
-6.72544479e-01 -6.32184088e-01 1.68702565e-02 1.76734418e-01
-6.67943060e-01 -2.57602334e-02 -1.03231728e+00 -7.31122553e-01
3.73254567e-01 -1.22678840e+00 -3.33852172e-01 -5.99805415e-01
-3.24995279e-01 -6.77678227e-01 6.60415888e-01 -6.46949410e-01
1.44304800e+00 -2.57058144e+00 2.50925086e-02 1.91775769e-01
5.33790402e-02 -1.77418143e-01 -5.11487797e-02 5.21017373e-01
2.57927239e-01 6.50072396e-02 2.80516744e-01 -5.35952926e-01
6.54399276e-01 2.76389241e-01 -6.56643927e-01 3.90191019e-01
-4.32189733e-01 1.12658417e+00 -7.86999404e-01 1.06764264e-01
3.81345212e-01 5.65496266e-01 -8.96184862e-01 2.11977109e-01
-5.54012135e-02 -1.47752792e-01 -6.58752203e-01 5.36055326e-01
6.84925735e-01 -3.82244349e-01 5.18121839e-01 -3.57093334e-01
1.88587785e-01 5.31877160e-01 -1.20468080e+00 1.56950772e+00
-7.75949359e-01 -2.14004979e-01 -2.96853751e-01 -9.24623966e-01
9.03145492e-01 -8.22519809e-02 5.89229047e-01 -6.83159411e-01
6.65402338e-02 -5.60124554e-02 -1.43376186e-01 -1.48619279e-01
7.30564356e-01 9.17785615e-02 -5.96271038e-01 5.24096549e-01
-1.45801194e-02 6.23963177e-01 6.47410303e-02 3.15440565e-01
6.06132805e-01 9.08280239e-02 1.20391183e-01 -5.79308867e-02
1.79939881e-01 -3.83887321e-01 5.51237285e-01 7.24810064e-01
-4.60876990e-03 3.82597774e-01 2.24295080e-01 -4.78008121e-01
-5.81149042e-01 -1.21927595e+00 -6.95688799e-02 1.72896254e+00
4.51049030e-01 -6.92923129e-01 -1.03258207e-01 -8.86711419e-01
2.73100019e-01 1.27468204e+00 -5.93531489e-01 -4.67998356e-01
-4.25457984e-01 -9.21822965e-01 -2.72105813e-01 5.98570704e-01
5.96069694e-02 -8.52180362e-01 1.15958109e-01 5.00669718e-01
-1.45995811e-01 -1.17353988e+00 -1.17372453e+00 1.58438742e-01
-8.07700098e-01 -6.54499412e-01 -3.69485199e-01 -8.16475332e-01
5.62121451e-01 5.16553044e-01 1.01115525e+00 -4.12634850e-01
2.41980046e-01 3.26845497e-01 -6.45701170e-01 -5.19971922e-02
1.17308840e-01 3.13488673e-03 4.15912598e-01 5.46942234e-01
1.09989631e+00 -5.43767512e-01 -7.33734488e-01 7.27685988e-01
-8.51841390e-01 -2.86651641e-01 3.37984413e-01 7.92467892e-01
5.83702624e-01 3.47573906e-01 2.05272004e-01 -1.17019045e+00
9.50983882e-01 -9.94515181e-01 -3.35965186e-01 1.81019425e-01
-8.93435359e-01 -1.66630909e-01 1.02534628e+00 -7.74232447e-01
-1.01028621e+00 8.40031952e-02 -2.28296250e-01 -2.84219176e-01
-6.10869043e-02 7.14578450e-01 -2.90628821e-01 5.95923245e-01
2.55928606e-01 5.59433281e-01 -4.44477975e-01 -7.07774222e-01
7.08376169e-01 7.74221480e-01 3.16706635e-02 -3.36179346e-01
4.54535902e-01 1.67433307e-01 -8.48860204e-01 -7.01712012e-01
-7.17861414e-01 -8.97963643e-01 -3.22027087e-01 2.13155806e-01
5.26791036e-01 -8.48601222e-01 -7.12951660e-01 4.51047122e-02
-6.68718219e-01 6.46582851e-03 -3.67471933e-01 7.03276873e-01
-2.05534533e-01 5.44804096e-01 -9.56157923e-01 -7.75643349e-01
-2.00707763e-01 -1.28443122e+00 8.97136211e-01 3.35879400e-02
-4.40093100e-01 -1.23133433e+00 -1.69501767e-01 4.19104993e-01
4.08997834e-01 -5.01640439e-01 9.48365629e-01 -1.06139159e+00
-3.48330051e-01 -6.24208748e-01 -2.13351190e-01 2.44846553e-01
5.14892161e-01 -4.36506093e-01 -4.06880319e-01 -2.03782901e-01
1.95168719e-01 8.56052786e-02 7.08795190e-01 6.53295934e-01
9.94717658e-01 -2.55071729e-01 -4.22312587e-01 5.53302348e-01
1.23596716e+00 5.49725056e-01 4.19510216e-01 -3.08516156e-02
9.04349148e-01 4.98742968e-01 4.91337717e-01 6.56323969e-01
9.02829945e-01 7.66622961e-01 1.75949410e-01 4.08347040e-01
2.26034403e-01 -8.69353175e-01 6.83503747e-01 1.34595299e+00
1.85195938e-01 -4.39287275e-01 -4.58436638e-01 4.73868579e-01
-2.09336948e+00 -6.93527520e-01 -2.90295184e-01 2.24658608e+00
3.14774185e-01 2.43139967e-01 2.86682278e-01 -3.79505932e-01
6.82776749e-01 -4.52884212e-02 -8.19492996e-01 -3.14500421e-01
2.18201175e-01 -2.27847427e-01 5.30203879e-01 3.57598543e-01
-1.09705877e+00 1.05347097e+00 5.29564333e+00 7.35257268e-01
-6.89915240e-01 1.29136547e-01 5.40708542e-01 9.33456197e-02
-8.63547146e-01 -1.11432403e-01 -1.08699083e+00 8.10319543e-01
8.73889983e-01 -6.92842752e-02 7.05442429e-01 1.05599701e+00
3.17793757e-01 4.71591055e-01 -1.14763379e+00 1.02006721e+00
1.52608499e-01 -7.56965876e-01 1.24306053e-01 3.10713828e-01
5.11518955e-01 -1.33800328e-01 4.69239175e-01 9.39708769e-01
8.03015888e-01 -6.60873652e-01 5.51326990e-01 4.00104403e-01
2.68288225e-01 -8.09062600e-01 7.32553959e-01 3.85350883e-01
-1.46535361e+00 -2.39934847e-01 -5.15186071e-01 1.10283457e-02
4.54784244e-01 6.31331563e-01 -3.88121158e-01 5.46261847e-01
5.85018277e-01 1.11112392e+00 -2.23050103e-01 4.50088710e-01
5.85344657e-02 9.50154126e-01 -4.41292971e-01 -3.06279629e-01
2.20808625e-01 -8.33840430e-01 4.35681939e-02 1.06220472e+00
5.44176817e-01 -5.36487326e-02 4.87724304e-01 9.37172353e-01
-2.47707471e-01 5.40000200e-01 -5.68799794e-01 -2.60544360e-01
1.39826402e-01 9.98132765e-01 -6.32327676e-01 -1.19213820e-01
-8.75901818e-01 1.26712024e+00 6.58190399e-02 4.59314525e-01
-1.04282534e+00 -2.24294826e-01 8.01355779e-01 1.91288963e-01
5.99308252e-01 -2.51506746e-01 -5.92780448e-02 -1.59052503e+00
1.87843800e-01 -6.50930762e-01 4.56055850e-01 -2.05451578e-01
-1.74917626e+00 3.99544835e-01 -3.18491220e-01 -1.09293318e+00
-1.70540363e-01 -4.28508431e-01 -4.52952355e-01 7.42127180e-01
-1.16829932e+00 -1.10290086e+00 1.74627244e-01 6.16561711e-01
9.34947789e-01 -9.18451101e-02 6.82052314e-01 4.75355268e-01
-6.58217251e-01 8.80285203e-01 3.45586598e-01 9.24581140e-02
4.58294183e-01 -1.13495505e+00 7.05603480e-01 6.25798762e-01
3.64132047e-01 1.32252395e+00 5.51254690e-01 -6.85504675e-01
-1.93726146e+00 -1.17457926e+00 1.14550805e+00 -6.41696453e-01
8.82016420e-01 -7.47917354e-01 -8.02234411e-01 9.67262924e-01
-9.19952691e-02 -5.34771383e-03 9.83967841e-01 7.56553590e-01
-3.27157825e-01 -8.86487961e-02 -7.88697660e-01 9.14656103e-01
1.30569482e+00 -5.21414459e-01 -5.37883162e-01 5.06063163e-01
9.69315827e-01 5.11624403e-02 -9.16348577e-01 6.07302673e-02
7.30285525e-01 -6.08714402e-01 1.17279494e+00 -6.02251947e-01
1.71879545e-01 4.43155169e-02 -5.00506163e-01 -1.23936296e+00
-8.09626460e-01 -3.79377693e-01 -3.10846657e-01 1.22093844e+00
3.76971871e-01 -6.43999159e-01 9.03297663e-01 8.49382699e-01
-2.03968361e-02 -6.73346817e-01 -2.96083778e-01 -5.10907531e-01
-2.04139143e-01 -7.93779135e-01 6.95633292e-01 8.61725271e-01
2.00822622e-01 5.96195877e-01 -6.12905622e-01 1.92660213e-01
5.83708763e-01 5.74871361e-01 7.23553538e-01 -1.05618691e+00
-4.78758931e-01 -3.50305498e-01 -9.09972936e-02 -1.83037949e+00
3.90045866e-02 -1.06850767e+00 -1.83580950e-01 -1.47886956e+00
1.68846652e-01 -4.21211034e-01 -7.46654749e-01 3.40915546e-02
-2.27913216e-01 -1.99280903e-01 7.33220726e-02 2.69847721e-01
-8.51340175e-01 7.25160241e-01 9.51028287e-01 -1.19381472e-01
-5.46440125e-01 4.05185044e-01 -1.09877706e+00 6.62559390e-01
5.77586949e-01 -3.50078911e-01 -7.98313498e-01 -2.57934451e-01
3.96036029e-01 -2.81589236e-02 -3.08622450e-01 -3.39506119e-01
1.91840962e-01 -1.60984010e-01 -9.65636373e-02 -5.14157772e-01
1.83308154e-01 -1.01746881e+00 1.13986462e-01 2.03076616e-01
-5.02759457e-01 1.33103430e-01 -2.79127449e-01 1.02938485e+00
-9.67218056e-02 1.22537464e-02 -1.14986263e-02 -1.20464107e-02
-7.71216393e-01 5.78058481e-01 -5.06587029e-01 -1.07546233e-01
7.48236120e-01 -7.21891373e-02 3.37016851e-01 -5.26935756e-01
-9.03741300e-01 1.40853256e-01 2.43593082e-01 9.18635309e-01
6.57293856e-01 -1.45804048e+00 -3.07721883e-01 4.51417416e-01
3.89225066e-01 -4.59825277e-01 4.19022024e-01 6.47775829e-01
2.05012888e-01 4.79613423e-01 3.70508999e-01 -4.63957638e-01
-8.36803436e-01 9.54351664e-01 -2.38780394e-01 -4.48321670e-01
-5.47197998e-01 8.75016809e-01 4.21326220e-01 -8.90271544e-01
3.21238488e-01 -6.66990519e-01 -4.01820749e-01 1.51945502e-01
1.71181783e-01 8.86799172e-02 -1.93463087e-01 -6.71324372e-01
-3.72109771e-01 3.97795528e-01 -5.31718493e-01 3.75119925e-01
1.43009388e+00 -3.79485399e-01 2.97345251e-01 5.58394313e-01
1.44625330e+00 2.87112355e-01 -1.03976142e+00 -6.26920164e-01
9.15240347e-02 -5.13840437e-01 1.18585169e-01 -3.35956395e-01
-9.88532722e-01 6.11573696e-01 4.74891335e-01 5.37146449e-01
8.06748927e-01 2.23944187e-02 1.26611137e+00 1.58009052e-01
4.97113377e-01 -9.94453549e-01 -1.54857710e-01 4.75459993e-01
4.54898149e-01 -1.27407539e+00 -3.33703369e-01 -4.62723523e-01
-1.03083229e+00 8.76514256e-01 2.13611543e-01 -2.88491935e-01
1.12819719e+00 -2.08149388e-01 -1.92532450e-01 -3.78655493e-02
-6.14882469e-01 -1.20433874e-01 4.51540291e-01 1.98089838e-01
6.26029313e-01 4.66132373e-01 -3.44783038e-01 1.40749574e+00
-5.55655919e-02 1.20920837e-02 6.16239980e-02 6.67933047e-01
-9.90407094e-02 -1.29697680e+00 1.70358047e-01 7.86745369e-01
-3.46316457e-01 -2.99530983e-01 1.16414472e-01 3.51868480e-01
-4.42195535e-02 1.01284516e+00 -2.95904409e-02 -8.20044518e-01
8.23602021e-01 1.45305336e-01 -7.98341036e-02 -7.53484309e-01
-2.59131759e-01 3.79312009e-01 -9.64701697e-02 -5.37033319e-01
1.15163125e-01 -9.48359668e-01 -1.04542446e+00 -3.58067006e-01
-4.62897837e-01 2.59711713e-01 4.42052722e-01 7.66448319e-01
4.99037445e-01 4.96197641e-01 8.13343883e-01 -4.25616562e-01
-8.08227777e-01 -8.04759324e-01 -9.00540292e-01 8.87547553e-01
-1.25816226e-01 -6.11159980e-01 -4.27652836e-01 -7.00106323e-02] | [10.154829978942871, 5.631142616271973] |
b423c813-9eaa-49bc-a7da-b9a0dbbd851f | face-hallucination-by-attentive-sequence | 1905.01509 | null | https://arxiv.org/abs/1905.01509v1 | https://arxiv.org/pdf/1905.01509v1.pdf | Face Hallucination by Attentive Sequence Optimization with Reinforcement Learning | Face hallucination is a domain-specific super-resolution problem that aims to generate a high-resolution (HR) face image from a low-resolution~(LR) input. In contrast to the existing patch-wise super-resolution models that divide a face image into regular patches and independently apply LR to HR mapping to each patch, we implement deep reinforcement learning and develop a novel attention-aware face hallucination (Attention-FH) framework, which recurrently learns to attend a sequence of patches and performs facial part enhancement by fully exploiting the global interdependency of the image. Specifically, our proposed framework incorporates two components: a recurrent policy network for dynamically specifying a new attended region at each time step based on the status of the super-resolved image and the past attended region sequence, and a local enhancement network for selected patch hallucination and global state updating. The Attention-FH model jointly learns the recurrent policy network and local enhancement network through maximizing a long-term reward that reflects the hallucination result with respect to the whole HR image. Extensive experiments demonstrate that our Attention-FH significantly outperforms the state-of-the-art methods on in-the-wild face images with large pose and illumination variations. | ['Liang Lin', 'Qingxing Cao', 'Keze Wang', 'Yukai Shi', 'Guanbin Li'] | 2019-05-04 | null | null | null | null | ['face-hallucination'] | ['computer-vision'] | [ 3.87269169e-01 4.04624879e-01 1.21477336e-01 -4.30899084e-01
-9.74258721e-01 2.65514672e-01 2.84115344e-01 -8.24083030e-01
1.39668822e-01 7.26896465e-01 4.37447727e-01 5.74320734e-01
2.67363526e-02 -8.84093165e-01 -8.97361636e-01 -8.27941418e-01
-3.57652642e-02 1.39479041e-01 -1.57394677e-01 -3.65932524e-01
-4.43170127e-03 7.07489789e-01 -1.77443767e+00 4.81957674e-01
5.89902520e-01 1.01572537e+00 4.66722608e-01 4.90862608e-01
3.58239561e-01 1.04562533e+00 -3.20290864e-01 4.37055901e-02
2.96143353e-01 -8.32824707e-01 -7.93316424e-01 6.01774514e-01
5.66538811e-01 -5.62079430e-01 -4.92938995e-01 9.35081601e-01
7.43518591e-01 4.08404261e-01 6.68676272e-02 -6.96070254e-01
-1.20754457e+00 2.52440631e-01 -9.99997258e-01 4.97695088e-01
5.08942544e-01 2.32278392e-01 5.84986448e-01 -1.16717803e+00
8.34850669e-01 1.62610912e+00 3.05727988e-01 9.02477324e-01
-1.41722929e+00 -6.80260837e-01 1.90468803e-01 8.77308771e-02
-1.39516640e+00 -7.56488085e-01 9.93760884e-01 1.90421343e-02
8.30194175e-01 -1.41039431e-01 3.70610356e-01 9.95145857e-01
2.26544753e-01 2.91636080e-01 1.29607630e+00 -1.75551936e-01
1.30115733e-01 -5.78022227e-02 -6.30530417e-01 9.01145577e-01
-5.19856811e-01 3.20558816e-01 -7.10520983e-01 -3.79648395e-02
1.38407099e+00 5.37886377e-03 -3.87637466e-01 -1.45438239e-01
-8.41436207e-01 6.20606184e-01 6.66414201e-01 1.80343777e-01
-9.29254115e-01 2.26676110e-02 -1.38642535e-01 3.32555741e-01
6.18751049e-01 3.87373388e-01 -2.49808446e-01 6.13622844e-01
-9.41229284e-01 2.00196445e-01 1.95243552e-01 4.56357062e-01
1.07940602e+00 5.19039273e-01 -5.42188227e-01 1.08497131e+00
-6.52353317e-02 2.77945608e-01 2.22013786e-01 -1.41244423e+00
4.53381911e-02 1.57840759e-01 2.10133016e-01 -7.48097956e-01
3.69219333e-02 -6.34353995e-01 -8.92212212e-01 4.33306426e-01
-3.74788195e-01 -2.94661969e-01 -1.03733516e+00 2.24346399e+00
4.76891547e-01 6.70510590e-01 1.71283349e-01 1.20897365e+00
9.74774241e-01 8.16849351e-01 8.48884284e-02 -8.39776933e-01
1.07394338e+00 -9.27148104e-01 -8.38570476e-01 -1.98990121e-01
-6.65885687e-01 -3.89876634e-01 6.20906472e-01 6.82942942e-02
-1.70713878e+00 -1.07720673e+00 -8.28324795e-01 -5.04767895e-02
2.60455668e-01 -2.37546396e-03 1.22829758e-01 -1.43677011e-01
-1.54157436e+00 6.53934956e-01 -3.84165913e-01 -1.29168585e-01
6.46174848e-01 4.10679579e-01 -4.69255656e-01 -2.41968468e-01
-1.14056504e+00 7.85529792e-01 -1.18010193e-01 3.24381232e-01
-1.48968482e+00 -6.73292577e-01 -8.87079358e-01 1.62214562e-01
3.02479416e-01 -7.41726339e-01 1.02078331e+00 -1.26475489e+00
-1.70583355e+00 7.22930551e-01 -4.69956934e-01 -8.82660449e-02
5.67532592e-02 -4.28383425e-02 -6.05433524e-01 3.76182675e-01
1.04564719e-01 9.48106945e-01 1.50310886e+00 -1.53506708e+00
-5.25257468e-01 -6.11128628e-01 -2.83718616e-01 6.29043341e-01
1.07075147e-01 2.50537127e-01 -5.22294760e-01 -5.40472448e-01
-4.77194935e-02 -3.30380410e-01 -3.26173812e-01 -1.04107372e-01
3.35014686e-02 4.65722792e-02 8.83641124e-01 -8.50037217e-01
8.78935575e-01 -2.15286875e+00 7.05490172e-01 -1.50726348e-01
1.59197479e-01 1.32902920e-01 -5.81768572e-01 -7.48346373e-02
-4.05864894e-01 -2.56045908e-01 -1.85963616e-01 -4.27700043e-01
-6.39281750e-01 1.07668675e-01 -4.77296144e-01 4.35657024e-01
6.81792438e-01 8.49102199e-01 -8.41922343e-01 -4.37279046e-01
5.10777570e-02 1.17775655e+00 -5.91749251e-01 7.59494543e-01
-2.42426172e-01 8.49129975e-01 -3.11522812e-01 5.58566630e-01
8.08126211e-01 -3.02752942e-01 2.11164758e-01 -2.56630272e-01
-1.14399560e-01 -2.14010149e-01 -1.02180707e+00 1.67135048e+00
-3.60944748e-01 1.93938106e-01 4.30584401e-01 -6.91670895e-01
1.16798329e+00 4.86535877e-01 5.42584240e-01 -9.12374377e-01
-1.50927231e-01 -2.07941085e-01 -4.60456967e-01 -4.47955638e-01
4.98515397e-01 -3.54071707e-01 4.28000659e-01 4.37972009e-01
3.84384900e-01 3.41242254e-01 -4.14408267e-01 -1.90750465e-01
8.40885460e-01 4.27181929e-01 1.06335513e-01 1.28685459e-01
6.26685321e-01 -7.04837382e-01 8.92411292e-01 3.13003242e-01
-1.68268830e-01 8.71476948e-01 3.34414810e-01 -3.57937604e-01
-9.70930874e-01 -1.00106287e+00 -7.13166455e-03 1.36547208e+00
1.01316780e-01 2.38039121e-01 -8.05237889e-01 -3.41991544e-01
-2.94160843e-01 2.65062124e-01 -9.89278316e-01 -2.12726220e-01
-7.59904087e-01 -5.53463697e-01 -9.87770185e-02 3.61522645e-01
8.09752345e-01 -2.00929046e+00 -4.55975652e-01 1.99838489e-01
-2.22052962e-01 -9.69607115e-01 -6.51991844e-01 -1.62815765e-01
-6.24590874e-01 -5.72578907e-01 -7.38534033e-01 -1.09899426e+00
7.35852122e-01 2.62740165e-01 1.05960345e+00 -5.08402064e-02
-3.92722696e-01 1.47006109e-01 -7.52623752e-03 3.78635913e-01
-2.19751582e-01 -3.56093228e-01 -2.11623713e-01 5.88251054e-01
-3.21124911e-01 -6.82919502e-01 -8.44573557e-01 1.35879546e-01
-7.48004854e-01 3.02435423e-04 7.36177325e-01 9.95919406e-01
1.09382141e+00 1.86528087e-01 8.36965024e-01 -8.59079957e-01
4.58543241e-01 -5.59753656e-01 -4.12426293e-01 2.70363927e-01
-2.87534058e-01 2.34070256e-01 4.41289991e-01 -4.04342145e-01
-1.67726564e+00 1.58452332e-01 -1.73276942e-02 -9.58499193e-01
-4.87790294e-02 -3.27497683e-02 -4.41010922e-01 -2.16273200e-02
4.48373497e-01 4.61162120e-01 1.74556077e-01 -3.70999306e-01
4.31108892e-01 2.69531727e-01 1.01375902e+00 -4.40011382e-01
7.09461272e-01 4.93325114e-01 -1.63823351e-01 -5.96898735e-01
-9.85407889e-01 -3.04993298e-02 -4.89575624e-01 -2.85676152e-01
1.04354060e+00 -1.15341616e+00 -7.50061750e-01 2.57209629e-01
-9.30873930e-01 -3.69660348e-01 -5.65977573e-01 -2.70253643e-02
-8.09166312e-01 -8.95395577e-02 -6.87285304e-01 -9.20837581e-01
-5.02022862e-01 -1.11003292e+00 1.32899797e+00 6.00259960e-01
3.11538219e-01 -5.56253016e-01 1.05092339e-01 3.16819608e-01
6.14515066e-01 4.98210967e-01 6.62692964e-01 4.03877348e-01
-7.17291772e-01 4.15383130e-01 -3.72785836e-01 2.97601491e-01
1.60831302e-01 -2.34137982e-01 -1.16831446e+00 -5.83859801e-01
9.75710601e-02 -5.25662541e-01 8.92216027e-01 5.53389788e-01
1.28969371e+00 -4.98008311e-01 2.68708915e-02 7.45274365e-01
1.46772671e+00 7.64763132e-02 9.51669633e-01 -1.63849056e-01
5.59441745e-01 6.69129908e-01 6.55556262e-01 4.88696754e-01
1.80752933e-01 8.57182384e-01 6.28584325e-01 -4.70146179e-01
-4.29771990e-01 -4.84477162e-01 6.26726270e-01 -2.77980808e-02
-3.18065286e-01 2.95287460e-01 5.20994999e-02 4.68744963e-01
-1.65315449e+00 -1.52875113e+00 8.59689057e-01 2.15800619e+00
9.54205513e-01 -3.87373149e-01 6.20195828e-02 -4.17245030e-01
9.86768246e-01 5.20245075e-01 -9.40960109e-01 -3.02833498e-01
-1.72674239e-01 5.93837023e-01 -1.10922456e-01 7.21036673e-01
-7.33974159e-01 1.20297647e+00 6.21044302e+00 4.99422163e-01
-1.28224933e+00 2.60545999e-01 9.63009000e-01 -1.10855177e-01
-3.31169844e-01 -3.10982227e-01 -7.24165916e-01 1.01550058e-01
7.61194348e-01 1.29210502e-02 1.08235562e+00 6.54366016e-01
2.55084008e-01 2.80232370e-01 -7.99018443e-01 8.66615057e-01
3.99766117e-01 -1.23580503e+00 2.38277614e-01 5.04648350e-02
9.81492281e-01 -1.74932018e-01 6.34087086e-01 2.80279875e-01
2.76924878e-01 -1.32389665e+00 3.57542396e-01 8.66231561e-01
1.21959519e+00 -1.16368473e+00 1.80102438e-01 -3.57010327e-02
-1.25209665e+00 -4.09673423e-01 -5.31985700e-01 3.52060020e-01
-4.53259051e-02 9.08995345e-02 -5.90091407e-01 3.72374564e-01
8.78395498e-01 7.30653286e-01 -3.36560786e-01 2.77335554e-01
-2.33394593e-01 7.03392550e-02 4.55031872e-01 9.90596592e-01
-2.34020695e-01 -5.86621501e-02 6.23652875e-01 6.83937609e-01
1.51598871e-01 5.55638075e-01 7.41771236e-02 1.27461421e+00
-3.22268039e-01 -3.32133561e-01 -2.70166665e-01 4.00889993e-01
3.80107254e-01 1.58508825e+00 -1.22195520e-01 -1.37634024e-01
-2.58297414e-01 1.29728270e+00 7.32964098e-01 6.70900941e-01
-7.08944380e-01 2.19456796e-02 8.36610436e-01 3.38184744e-01
7.53971219e-01 3.45348954e-01 2.41777375e-01 -9.48497593e-01
-2.46742994e-01 -9.82086539e-01 4.19712424e-01 -1.16041803e+00
-1.09433186e+00 1.30926454e+00 -3.36594343e-01 -9.50085402e-01
-4.58171338e-01 1.63026169e-01 -5.81527948e-01 1.17716157e+00
-1.96399844e+00 -1.22450447e+00 -4.03136820e-01 1.00683343e+00
8.04414570e-01 -2.94154316e-01 7.41762757e-01 -5.39428443e-02
-7.71653831e-01 5.20898402e-01 -4.72076178e-01 -6.00653514e-02
7.37238407e-01 -9.32112157e-01 1.20256968e-01 8.40310633e-01
-2.04255626e-01 4.51319396e-01 6.19905114e-01 -5.99513292e-01
-1.27498519e+00 -1.49567592e+00 5.99808753e-01 -1.17494769e-01
-3.38729285e-02 -7.28813931e-03 -1.23462284e+00 6.11421168e-01
3.85247648e-01 6.16098642e-01 4.20232303e-02 -2.65370786e-01
-3.89294595e-01 -4.73030567e-01 -1.53334296e+00 4.39318001e-01
1.06775904e+00 -6.41359270e-01 -4.25479919e-01 2.24451832e-02
7.45102525e-01 -5.03820360e-01 -1.02116966e+00 5.57285726e-01
3.88736933e-01 -1.07733428e+00 1.09493065e+00 -4.28217828e-01
7.01757789e-01 -4.37651336e-01 -1.08965255e-01 -1.26364374e+00
-8.58941019e-01 -7.69792795e-01 -2.11436093e-01 1.06397295e+00
6.53846115e-02 -3.11512172e-01 5.15285194e-01 3.14356297e-01
-1.61107238e-02 -7.60592878e-01 -8.57694387e-01 -1.13349020e-01
-3.22331339e-01 4.53561187e-01 8.71949852e-01 7.88838804e-01
-3.25546414e-01 5.02396703e-01 -7.93257236e-01 5.99503696e-01
9.87146556e-01 4.64478850e-01 3.92394304e-01 -8.83227289e-01
-4.07044530e-01 -9.41335484e-02 3.02860402e-02 -5.89904428e-01
4.54053581e-01 -5.46082854e-01 2.42440730e-01 -1.18260098e+00
6.05340302e-01 6.51460364e-02 -4.81857866e-01 6.35047197e-01
-3.34647089e-01 5.63626647e-01 -5.19194528e-02 1.62575334e-01
-7.20629215e-01 8.41669679e-01 1.83519983e+00 2.40872353e-01
-4.01696801e-01 -2.66627282e-01 -1.01624775e+00 2.71218270e-01
4.27603453e-01 -3.56396325e-02 -5.01479805e-01 -3.25695425e-01
-2.49383181e-01 1.01568401e+00 4.27031279e-01 -8.17025483e-01
1.02669418e-01 -3.14575493e-01 9.65805054e-01 -2.76045710e-01
6.27325237e-01 -4.46986765e-01 3.38598937e-01 1.61605120e-01
-7.05869019e-01 -1.06093243e-01 -4.26174030e-02 6.71817958e-01
-1.87011331e-01 6.29537344e-01 1.38397920e+00 -3.16614240e-01
-6.94023252e-01 7.86407173e-01 6.18167445e-02 -1.23401597e-01
8.93151760e-01 -1.35732442e-01 -2.65829712e-01 -4.50245142e-01
-1.04692030e+00 1.00103110e-01 4.61166263e-01 6.58571064e-01
1.09994328e+00 -1.30399930e+00 -1.09724784e+00 7.19850898e-01
-2.52451181e-01 2.14038938e-02 7.67841399e-01 2.48223290e-01
1.74356639e-01 -4.32026498e-02 -8.20010245e-01 -2.96514481e-01
-1.17710948e+00 7.67342508e-01 7.12980092e-01 -2.55155444e-01
-7.96727657e-01 7.54125535e-01 5.48463762e-01 -1.16606206e-01
7.13189170e-02 4.18680876e-01 -6.98177040e-01 -3.91266376e-01
8.97350669e-01 2.25038290e-01 -2.56798685e-01 -1.04675746e+00
-7.84742013e-02 6.76188707e-01 -3.03503871e-01 -2.62810111e-01
1.63141966e+00 -3.38899463e-01 -3.19113612e-01 -4.59837615e-02
9.48834240e-01 -3.66226196e-01 -1.96133101e+00 -5.05373657e-01
-7.67589688e-01 -6.08582377e-01 3.02244097e-01 -8.03467274e-01
-1.67234623e+00 3.70155573e-01 7.37446964e-01 -5.14869511e-01
1.72625840e+00 1.81958303e-01 5.57672799e-01 -2.96237439e-01
3.29546750e-01 -8.02523673e-01 5.80546618e-01 1.31060079e-01
1.44402850e+00 -1.03986180e+00 3.12159285e-02 -3.18698615e-01
-9.10132766e-01 6.66311562e-01 1.13155305e+00 -3.02113205e-01
5.71787000e-01 1.30496427e-01 -2.77962625e-01 -2.47102216e-01
-1.21428037e+00 -2.80245602e-01 7.27792829e-02 7.71159589e-01
2.64946908e-01 -2.08386093e-01 2.97520459e-01 3.38899910e-01
7.01169372e-02 2.45551258e-01 3.05785030e-01 4.74225342e-01
-5.73912680e-01 -7.10075617e-01 -4.67186272e-01 8.24349821e-02
-3.66856188e-01 1.54426605e-01 -3.90311927e-02 2.52450705e-01
2.60988206e-01 8.52062404e-01 2.86046445e-01 -2.74289787e-01
2.26455823e-01 -2.60578722e-01 7.17238307e-01 -7.79993832e-01
-4.72137272e-01 3.37233633e-01 -4.79732960e-01 -9.01083887e-01
-4.63341117e-01 -6.08366489e-01 -1.25144601e+00 -1.79018393e-01
1.18984252e-01 2.01501306e-02 9.34496894e-02 6.61542594e-01
6.25475705e-01 7.98874795e-01 1.29342246e+00 -1.14153147e+00
-3.46699595e-01 -8.85461390e-01 -7.82628179e-01 1.58785373e-01
7.83931732e-01 -5.18176734e-01 -6.32533357e-02 1.59094825e-01] | [12.759881973266602, -0.1190842017531395] |
85f6c84c-ad6d-4062-8284-de25b358bb8d | improving-the-quality-of-dental-crown-using-a | 2303.02426 | null | https://arxiv.org/abs/2303.02426v1 | https://arxiv.org/pdf/2303.02426v1.pdf | Improving the quality of dental crown using a Transformer-based method | Designing a synthetic crown is a time-consuming, inconsistent, and labor-intensive process. In this work, we present a fully automatic method that not only learns human design dental crowns, but also improves the consistency, functionality, and esthetic of the crowns. Following success in point cloud completion using the transformer-based network, we tackle the problem of the crown generation as a point-cloud completion around a prepared tooth. To this end, we use a geometry-aware transformer to generate dental crowns. Our main contribution is to add a margin line information to the network, as the accuracy of generating a precise margin line directly,determines whether the designed crown and prepared tooth are closely matched to allowappropriateadhesion.Using our ground truth crown, we can extract the margin line as a spline and sample the spline into 1000 points. We feed the obtained margin line along with two neighbor teeth of the prepared tooth and three closest teeth in the opposing jaw. We also add the margin line points to our ground truth crown to increase the resolution at the margin line. Our experimental results show an improvement in the quality of the designed crown when considering the actual context composed of the prepared tooth along with the margin line compared with a crown generated in an empty space as was done by other studies in the literature. | ['Francois Guibault', 'Farida Cheriet', 'Julia Keren', 'Ying Zhang', 'Ammar Alsheghri', 'Farnoosh Ghadiri', 'Golriz Hosseinimanesh'] | 2023-03-04 | null | null | null | null | ['point-cloud-completion'] | ['computer-vision'] | [ 2.68112391e-01 8.62722337e-01 2.11882040e-01 -3.06126624e-01
-7.30944335e-01 -1.27354041e-01 1.68388575e-01 5.45417853e-02
-2.84878388e-02 5.79117298e-01 8.35641176e-02 -4.86523099e-02
-1.83530390e-01 -1.06539154e+00 -9.84884262e-01 -5.23739934e-01
3.05078983e-01 6.65434182e-01 1.58663794e-01 -3.07097375e-01
2.43543059e-01 7.94668972e-01 -1.92386496e+00 1.90425143e-01
1.08733690e+00 7.87904799e-01 1.20164685e-01 1.61941215e-01
-1.30225927e-01 2.75192130e-02 -5.87694585e-01 -5.31729281e-01
2.50563949e-01 2.80324779e-02 -4.01956350e-01 2.47970432e-01
7.50703037e-01 -3.95887464e-01 6.14167452e-01 7.94601679e-01
3.50420177e-01 -3.42443168e-01 7.46657252e-01 -1.08195770e+00
-6.43378794e-01 5.85331440e-01 -4.80447590e-01 -9.65517938e-01
2.73342758e-01 -1.30947437e-02 7.39797413e-01 -1.03847873e+00
1.00508785e+00 1.23114407e+00 8.99982274e-01 6.27297878e-01
-1.34957230e+00 -6.57550454e-01 -8.69502425e-02 -1.50749967e-01
-1.39168203e+00 -4.45236742e-01 1.11841345e+00 -4.42299366e-01
2.63895452e-01 1.69143423e-01 1.22787607e+00 6.00050807e-01
2.30011627e-01 4.00809079e-01 9.76206422e-01 -7.20635891e-01
5.48193336e-01 1.09719403e-01 1.62259594e-01 5.49096286e-01
1.21199958e-01 3.74001443e-01 3.71948536e-03 -2.54982412e-01
9.44308221e-01 -1.41898960e-01 -2.20704541e-01 -2.15087160e-01
-6.54354870e-01 7.49677360e-01 7.66486168e-01 2.08654717e-01
-7.20271349e-01 -4.18995172e-02 -2.03449935e-01 -1.59550384e-01
4.01538134e-01 2.80574828e-01 -3.06557506e-01 4.39329386e-01
-1.20810103e+00 3.92776698e-01 9.75943267e-01 8.64025950e-01
9.43095803e-01 -1.67453185e-01 8.47140476e-02 8.42179000e-01
6.61605835e-01 4.67430711e-01 2.13361122e-02 -1.20460153e+00
9.02058855e-02 8.73495162e-01 3.89759168e-02 -1.05966914e+00
-4.13795054e-01 -2.67061830e-01 -7.09622025e-01 7.77599752e-01
6.11717343e-01 3.34876478e-02 -1.15645659e+00 1.36857784e+00
8.69059801e-01 -7.17082694e-02 -1.07526228e-01 6.33345127e-01
7.36691833e-01 2.93275982e-01 -2.01688632e-01 -1.95894316e-01
1.61569464e+00 -7.53536284e-01 -6.73176408e-01 -1.02956042e-01
3.93940687e-01 -1.13335490e+00 1.23551548e+00 6.35704815e-01
-1.12357378e+00 -4.68129694e-01 -1.18948460e+00 -3.40115219e-01
-3.61512713e-02 3.85925174e-01 4.21378881e-01 4.43361878e-01
-1.10087204e+00 8.21622670e-01 -6.81387067e-01 4.66818288e-02
3.72482836e-01 3.79447132e-01 -3.63853514e-01 2.42946558e-02
-7.97387302e-01 8.44664812e-01 -9.43315551e-02 3.43704581e-01
-2.46089816e-01 -8.95703375e-01 -8.32594931e-01 -3.01850498e-01
1.55210391e-01 -8.53130400e-01 1.19693756e+00 -8.64152849e-01
-1.93080592e+00 7.80571043e-01 -3.21289413e-02 -1.61093324e-01
7.37145364e-01 -2.33507201e-01 -8.06316808e-02 -1.29664764e-01
3.48641038e-01 9.30896580e-01 5.03775418e-01 -1.68220043e+00
-2.07164884e-01 -6.46572828e-01 -2.56512344e-01 -3.00507873e-01
1.90346166e-01 -6.11701608e-01 -4.77813095e-01 -5.50222397e-01
4.86452252e-01 -1.02827024e+00 -3.46211016e-01 8.39650452e-01
-9.91944313e-01 -3.65990698e-01 4.47626472e-01 -9.37117457e-01
7.51201510e-01 -2.17735481e+00 -4.58841085e-01 5.86992383e-01
-1.59033667e-03 -2.26645041e-02 5.06604500e-02 3.69195908e-01
3.51819322e-02 -5.77878281e-02 -4.69390243e-01 -6.78533673e-01
-2.06638619e-01 2.59703398e-01 3.37275025e-03 3.55018139e-01
1.01783723e-01 5.91707885e-01 -5.72266877e-01 -7.43017852e-01
1.56530514e-01 9.94587898e-01 -7.05701947e-01 1.38548659e-02
-5.88604152e-01 2.15205207e-01 -3.60030204e-01 7.52464235e-01
1.06514776e+00 -1.29748462e-02 6.78404719e-02 -7.25778341e-01
-2.66523689e-01 -3.82969789e-02 -1.45685220e+00 1.67838585e+00
-6.79067016e-01 1.42799929e-01 1.55659944e-01 -2.52673030e-01
1.41169989e+00 4.04864758e-01 3.19239229e-01 -5.74356198e-01
-1.56744961e-02 5.81116855e-01 -3.17909241e-01 -4.37278956e-01
4.53994930e-01 -2.26283371e-01 3.77546638e-01 2.79608876e-01
-3.71955186e-01 -7.89394259e-01 -2.13872746e-01 -3.29827935e-01
9.05507952e-02 6.13237202e-01 1.10471606e-01 -2.49460787e-01
3.24889123e-01 -2.67872661e-02 4.61682796e-01 2.94972267e-02
4.94757086e-01 1.00657892e+00 3.23882818e-01 -5.15816212e-01
-1.20537353e+00 -9.10567641e-01 -3.67956966e-01 2.64991075e-01
-1.25527129e-01 2.93320157e-02 -1.18492925e+00 -1.20303988e-01
1.29008451e-02 8.80021870e-01 -7.14544773e-01 1.95941582e-01
-9.23510253e-01 -2.56114274e-01 1.07896350e-01 3.91405970e-01
2.77058333e-01 -9.84647691e-01 -3.53793770e-01 3.86136770e-01
-6.01233691e-02 -8.53570282e-01 -5.86022735e-01 -1.22367367e-01
-1.05058002e+00 -1.14788103e+00 -8.05444777e-01 -1.05113554e+00
8.86281192e-01 -4.99749810e-01 9.59854007e-01 4.44406509e-01
-2.22982287e-01 -4.09960806e-01 2.44962305e-01 -5.39959133e-01
-6.73964202e-01 -1.66089296e-01 -2.83636719e-01 -2.37881802e-02
-1.10020474e-01 -7.94294178e-01 -7.65351236e-01 3.39926332e-01
-7.31497526e-01 3.80521774e-01 3.44754219e-01 4.13225949e-01
1.10870111e+00 -2.25280032e-01 5.78277230e-01 -6.60314023e-01
5.13562918e-01 -1.45668641e-01 -7.82784224e-01 4.86882851e-02
-7.18398631e-01 3.27355832e-01 4.20192748e-01 -4.73367721e-01
-8.64915252e-01 4.42772150e-01 -5.82820892e-01 -7.74685368e-02
-1.31223887e-01 1.93283260e-01 -3.00488800e-01 2.78362721e-01
8.31953764e-01 -3.09568495e-01 4.62753594e-01 -7.70494044e-01
7.23394811e-01 5.82601309e-01 7.16489613e-01 -6.48632050e-01
6.42029762e-01 6.25481069e-01 6.79204389e-02 -6.81956172e-01
-5.73182285e-01 1.83177978e-01 -6.20116234e-01 -2.59838730e-01
7.59093285e-01 -3.88017684e-01 -1.10876942e+00 4.17957395e-01
-1.57549608e+00 -2.04368770e-01 -5.91636479e-01 1.91028565e-01
-4.88569289e-01 2.76328146e-01 -5.25333405e-01 -8.16915631e-01
-7.88230836e-01 -1.15772688e+00 1.44694996e+00 2.95450449e-01
-1.71254665e-01 -5.96535385e-01 2.25324795e-01 4.13831145e-01
6.08082861e-02 6.05687201e-01 9.18937147e-01 3.83906476e-02
-6.35001123e-01 -2.00910762e-01 1.16506636e-01 4.49569076e-01
2.64267355e-01 6.48024380e-01 -1.04067600e+00 3.35626990e-01
8.32282566e-03 2.57061422e-01 2.76134580e-01 6.63083017e-01
4.76340622e-01 -4.40727264e-01 -5.15220881e-01 4.59908009e-01
1.38475955e+00 1.50220945e-01 7.97002435e-01 3.87047857e-01
4.78534967e-01 1.08411205e+00 4.12015140e-01 6.06390610e-02
5.89083731e-01 8.47230971e-01 5.04116535e-01 -1.20914720e-01
-5.65268159e-01 -4.54465032e-01 5.75561598e-02 7.44125783e-01
-2.44803399e-01 1.24241635e-01 -8.18445444e-01 5.85151255e-01
-1.35175765e+00 -4.35475230e-01 -4.68703866e-01 2.31634474e+00
1.11342430e+00 1.65333703e-01 7.66847888e-03 4.60974485e-01
7.43165553e-01 -6.88966691e-01 -4.29477960e-01 -4.02852863e-01
1.84817657e-01 4.55359340e-01 2.41131842e-01 8.79042149e-01
-2.81324476e-01 6.68493509e-01 5.95131540e+00 5.74667215e-01
-1.43825459e+00 -3.75427693e-01 5.00192881e-01 1.76589116e-01
-5.64813733e-01 -9.24532786e-02 -8.12363803e-01 2.85139799e-01
5.91890514e-01 1.47536084e-01 1.96037456e-01 9.94700432e-01
4.17072564e-01 -3.88932191e-02 -1.23494875e+00 5.24219871e-01
-1.36477813e-01 -1.42605245e+00 -1.25808284e-01 2.93099225e-01
5.94128907e-01 -4.68603253e-01 -1.16078556e-01 -2.92150766e-01
8.36058483e-02 -9.19922173e-01 1.01803279e+00 8.28623652e-01
7.44902432e-01 -5.02244473e-01 4.78544027e-01 3.58026922e-01
-1.02750087e+00 4.34173048e-01 5.47485352e-02 1.37186453e-01
4.43004668e-01 1.03243423e+00 -1.32029104e+00 4.12713259e-01
4.32659566e-01 7.25541040e-02 -2.21315652e-01 1.08823264e+00
-5.02830923e-01 1.16385162e-01 -6.47470415e-01 1.26374379e-01
-2.02501148e-01 -3.61405939e-01 2.65462399e-01 3.93856317e-01
4.34112549e-01 -2.51789123e-01 -1.56595871e-01 1.43089652e+00
6.31878749e-02 4.89140749e-01 -2.55071849e-01 6.68949425e-01
6.49353147e-01 1.04067099e+00 -6.41185760e-01 -1.44370183e-01
1.95192546e-01 6.14250302e-01 9.67807025e-02 5.42860329e-02
-4.31936055e-01 -4.07002568e-02 3.54349047e-01 4.92633611e-01
1.67480975e-01 8.49306360e-02 -6.49185359e-01 -4.12788391e-01
5.18695652e-01 -5.87025762e-01 -2.50292003e-01 -1.12247586e+00
-1.14893806e+00 1.02355540e+00 -6.89896271e-02 -1.24549854e+00
-3.77675802e-01 -4.01352555e-01 -5.50068140e-01 8.72107208e-01
-1.28608108e+00 -1.42647231e+00 -3.47086012e-01 2.77721494e-01
1.55252948e-01 5.56634605e-01 9.61800396e-01 2.53898442e-01
-2.20754310e-01 4.38475311e-01 -2.92463273e-01 -1.62459284e-01
4.22313720e-01 -9.52324986e-01 3.62043381e-01 1.69592500e-01
-9.34915245e-02 6.26441002e-01 7.45065570e-01 -8.99497569e-01
-8.43240321e-01 -7.58061409e-01 1.04685044e+00 -7.71559998e-02
7.99852610e-02 -2.40270346e-01 -9.01255846e-01 4.92859244e-01
-1.95006415e-01 -1.17484480e-01 3.37094009e-01 -1.83275118e-01
-1.64498508e-01 -1.59081444e-01 -1.65439475e+00 6.97846234e-01
6.22931540e-01 -2.37389550e-01 -6.60342276e-01 3.11261892e-01
8.61876249e-01 -5.34976065e-01 -1.06491077e+00 4.60099280e-01
9.66377676e-01 -8.71791601e-01 1.22478199e+00 2.01706842e-01
4.34849799e-01 -4.46332723e-01 1.88580602e-01 -1.13953388e+00
2.94932216e-01 -3.79906565e-01 3.31598818e-01 1.14103317e+00
4.52218801e-01 -6.07135296e-01 1.22975445e+00 7.66078830e-01
-5.23522794e-01 -1.06492865e+00 -1.07998490e+00 -4.11149532e-01
2.99227655e-01 -3.95771921e-01 9.78247046e-01 5.99363446e-01
-4.16003108e-01 -1.59958005e-02 1.26572242e-02 2.56562501e-01
8.26979399e-01 3.86890471e-01 7.07074046e-01 -1.69505465e+00
1.77156962e-02 -2.04591945e-01 -8.30844641e-02 -8.88460457e-01
-2.41715774e-01 -6.75608754e-01 2.64821768e-01 -1.87425399e+00
-4.29831773e-01 -9.51871812e-01 5.50929189e-01 4.59286511e-01
1.03169292e-01 1.56147838e-01 5.98877333e-02 3.12281307e-02
7.85454631e-01 3.73747259e-01 1.91442597e+00 -3.12210266e-02
-6.51648283e-01 2.47601107e-01 -5.89682460e-01 1.02126861e+00
5.62951386e-01 -5.28512478e-01 -1.77876458e-01 -3.21782589e-01
-3.53002362e-03 2.42118433e-01 3.90620857e-01 -8.19144309e-01
2.53454834e-01 4.12834398e-02 5.07719554e-02 -8.48650455e-01
7.24089801e-01 -1.14824438e+00 6.96405709e-01 6.00395858e-01
7.64947617e-03 -2.34439552e-01 6.00629710e-02 -3.04746628e-02
1.83596283e-01 -3.49323273e-01 7.13740826e-01 -1.32176606e-02
1.62068024e-01 5.63563295e-02 2.09640443e-01 -4.27092105e-01
8.17048192e-01 -7.41746664e-01 -4.93007302e-02 -3.66908088e-02
-1.03450298e+00 -1.47264481e-01 6.33037448e-01 1.54972911e-01
5.34030616e-01 -1.22485721e+00 -6.89878285e-01 4.90289301e-01
-1.17484726e-01 5.84564626e-01 -4.47723689e-03 4.51074570e-01
-7.12998331e-01 -9.98445153e-02 -6.64152205e-02 -7.29787588e-01
-1.05319333e+00 3.65760267e-01 6.83624268e-01 1.69895843e-01
-7.79994726e-01 5.30822635e-01 -2.14245439e-01 -6.39353871e-01
1.70802087e-01 -1.04088485e+00 -2.04485223e-01 1.74910247e-01
2.02996328e-01 4.11489874e-01 2.97755331e-01 -5.24430156e-01
-1.74189538e-01 1.30330229e+00 1.69398457e-01 -1.50369808e-01
1.61646342e+00 3.20275158e-01 -2.13852167e-01 1.94525152e-01
1.11662447e+00 3.46592098e-01 -1.10656238e+00 9.71835256e-02
-1.19762897e-01 -1.27804369e-01 -1.03596874e-01 -7.68775403e-01
-1.34103537e+00 5.30430317e-01 7.00451314e-01 6.01673722e-02
7.80738175e-01 7.71930665e-02 9.70552504e-01 -2.17040718e-01
4.94537473e-01 -1.01330054e+00 -1.61727145e-01 -1.22951597e-01
1.27748334e+00 -7.37400770e-01 -6.05691224e-02 -1.06412089e+00
-1.81546897e-01 1.24984109e+00 4.01778638e-01 -2.36528546e-01
9.62256014e-01 4.06418473e-01 3.32051337e-01 -4.60178137e-01
-9.24289674e-02 2.59676129e-01 4.98548388e-01 7.11792707e-01
4.23791587e-01 9.12720934e-02 -4.35718149e-01 4.55348998e-01
-9.56661820e-01 5.54318249e-01 3.89582962e-01 8.39405179e-01
-2.91156888e-01 -1.30312860e+00 -7.02350616e-01 1.17218539e-01
-5.00363186e-02 -1.90835208e-01 2.25379150e-02 7.75793970e-01
5.43120980e-01 9.75406647e-01 2.61160791e-01 3.88468541e-02
8.55940700e-01 -1.59615755e-01 3.72760415e-01 -5.31958640e-01
-6.27363026e-01 2.11520433e-01 1.22655444e-01 -3.78665656e-01
-2.06741169e-01 -4.62779790e-01 -1.64404547e+00 -1.95594087e-01
-6.50223136e-01 -2.80960202e-02 1.14806473e+00 7.25547194e-01
2.87625164e-01 2.39525303e-01 4.59941328e-01 -1.05981541e+00
-3.96443456e-01 -7.67474711e-01 -2.54406363e-01 4.12794113e-01
3.29791218e-01 -5.39815068e-01 -4.99194533e-01 1.73032925e-01] | [13.062817573547363, -1.6135456562042236] |
7f649d32-8ca9-4cf3-b28d-3a1273eb51cb | comparison-of-model-free-and-model-based | 2212.08801 | null | https://arxiv.org/abs/2212.08801v1 | https://arxiv.org/pdf/2212.08801v1.pdf | Comparison of Model-Free and Model-Based Learning-Informed Planning for PointGoal Navigation | In recent years several learning approaches to point goal navigation in previously unseen environments have been proposed. They vary in the representations of the environments, problem decomposition, and experimental evaluation. In this work, we compare the state-of-the-art Deep Reinforcement Learning based approaches with Partially Observable Markov Decision Process (POMDP) formulation of the point goal navigation problem. We adapt the (POMDP) sub-goal framework proposed by [1] and modify the component that estimates frontier properties by using partial semantic maps of indoor scenes built from images' semantic segmentation. In addition to the well-known completeness of the model-based approach, we demonstrate that it is robust and efficient in that it leverages informative, learned properties of the frontiers compared to an optimistic frontier-based planner. We also demonstrate its data efficiency compared to the end-to-end deep reinforcement learning approaches. We compare our results against an optimistic planner, ANS and DD-PPO on Matterport3D dataset using the Habitat Simulator. We show comparable, though slightly worse performance than the SOTA DD-PPO approach, yet with far fewer data. | ['Jana Kosecka', 'Gregory J. Stein', 'Arnab Debnath', 'Yimeng Li'] | 2022-12-17 | null | null | null | null | ['problem-decomposition', 'pointgoal-navigation'] | ['miscellaneous', 'robots'] | [-3.27257290e-02 4.81354803e-01 -4.05319557e-02 -3.07954490e-01
-1.13601446e+00 -8.01835656e-01 6.03312254e-01 2.57298261e-01
-8.00935447e-01 1.18652153e+00 5.58560014e-01 -2.11312711e-01
-7.08590984e-01 -8.83493185e-01 -1.01885521e+00 -6.25214458e-01
-7.28564799e-01 1.09135616e+00 4.26014483e-01 -3.04500818e-01
4.47658598e-01 2.95349360e-01 -1.43808794e+00 -1.67053998e-01
6.31233692e-01 8.23944569e-01 5.16007662e-01 9.36214328e-01
1.87110990e-01 9.39300597e-01 -1.20060340e-01 2.37313196e-01
5.24355054e-01 -1.68518871e-01 -1.18880332e+00 -1.08364597e-01
9.21335295e-02 -6.71930432e-01 -4.17318285e-01 7.55641818e-01
5.44256270e-01 7.17676103e-01 4.88594055e-01 -1.38571846e+00
-3.67421471e-02 4.54444706e-01 -1.09490231e-01 -7.66241457e-03
6.57446921e-01 3.84143323e-01 7.47353613e-01 -9.54574570e-02
8.51300776e-01 1.60221529e+00 6.02145731e-01 5.37951827e-01
-1.20156562e+00 -1.10272065e-01 5.13003409e-01 2.82056034e-01
-8.50014448e-01 -1.77171588e-01 3.12826365e-01 -1.58090621e-01
1.33031452e+00 -1.52820155e-01 9.39841866e-01 1.37376714e+00
4.17462945e-01 1.04730546e+00 1.64452612e+00 -1.44575164e-01
1.01285923e+00 -6.70858979e-01 -2.31784061e-01 8.99603128e-01
-3.09378244e-02 7.16953516e-01 -5.07563114e-01 -2.61160702e-01
8.27070415e-01 -1.91909358e-01 3.15827839e-02 -1.11742842e+00
-1.06001186e+00 8.29226077e-01 5.15566885e-01 -3.47615153e-01
-6.11460865e-01 7.70090103e-01 1.19396105e-01 6.19210526e-02
-2.62117647e-02 3.26366842e-01 -5.72069585e-01 -5.77615976e-01
-6.20038033e-01 1.01282394e+00 1.24895465e+00 1.13128245e+00
7.18575120e-01 -2.17439249e-01 -2.32368670e-02 7.97740817e-02
5.41397929e-01 4.66529369e-01 1.25338718e-01 -1.74586284e+00
5.28763771e-01 1.54352680e-01 7.30235219e-01 -3.81451905e-01
-8.25422704e-01 -1.34514108e-01 7.47841643e-03 7.81685293e-01
5.36558449e-01 -4.28188622e-01 -1.28853261e+00 1.70433652e+00
4.37403560e-01 2.25089546e-02 5.57010233e-01 8.61408591e-01
2.63513356e-01 5.14753759e-01 1.76094264e-01 3.80370498e-01
7.37446070e-01 -1.22974229e+00 -4.26869482e-01 -4.42471117e-01
3.51245850e-01 -1.29255086e-01 8.01534891e-01 6.08390629e-01
-9.13450420e-01 -2.90589124e-01 -1.01286912e+00 -6.65660501e-02
-4.17909741e-01 -5.72598457e-01 8.68486345e-01 4.50109333e-01
-1.47348809e+00 9.32443619e-01 -1.30871499e+00 -6.83135390e-01
5.72043896e-01 3.52098048e-01 -2.88068831e-01 -1.77432746e-01
-8.31599057e-01 1.14241219e+00 6.67868435e-01 -3.85628402e-01
-2.06028557e+00 -4.58590910e-02 -9.50205863e-01 8.02313462e-02
7.60373950e-01 -8.68236899e-01 1.77227211e+00 -2.93577909e-01
-1.86177158e+00 4.35776919e-01 1.12985164e-01 -1.04254651e+00
7.18945205e-01 -4.99249756e-01 3.35046887e-01 3.23649108e-01
3.25332969e-01 1.25014257e+00 2.09506944e-01 -1.52048075e+00
-1.09119284e+00 -4.20772582e-01 6.47302389e-01 8.69597375e-01
6.94757402e-01 -7.11812496e-01 -1.39294237e-01 1.53026313e-01
2.62084931e-01 -1.14785337e+00 -1.06023145e+00 -1.66719221e-02
-2.52436012e-01 7.19538191e-03 3.70531648e-01 -5.15442669e-01
4.01140451e-01 -1.52675414e+00 3.69225770e-01 -3.81138735e-02
-9.67006758e-02 -5.56065500e-01 -1.01253197e-01 9.00229692e-01
5.89597702e-01 -8.41628760e-02 -4.54907179e-01 -4.99202549e-01
5.01488984e-01 8.87151420e-01 -4.86068040e-01 5.62567234e-01
-3.85827661e-01 7.42478788e-01 -1.30830097e+00 -4.58369344e-01
4.17536318e-01 8.87399632e-03 -6.78885341e-01 5.63584119e-02
-6.29848421e-01 6.95541680e-01 -7.90645242e-01 7.13356078e-01
4.22305137e-01 4.47750419e-01 3.08801532e-01 7.56317794e-01
-4.74104673e-01 3.68007690e-01 -1.20748520e+00 2.54482031e+00
-3.36220294e-01 2.27823466e-01 1.01556376e-01 -8.48842978e-01
5.37288189e-01 1.09898910e-01 4.92360950e-01 -7.21463561e-01
-1.64166570e-01 2.17581719e-01 -3.57146293e-01 -4.55466211e-01
6.26047194e-01 -1.23062976e-01 -4.04163599e-01 1.34417832e-01
3.29399109e-01 -5.53089797e-01 9.26095992e-02 -1.21959858e-01
1.49516153e+00 1.36466837e+00 4.54449475e-01 -5.51408827e-01
-8.17360133e-02 7.96952188e-01 3.56708437e-01 1.41398358e+00
-6.97776973e-01 3.59262943e-01 4.83779371e-01 -5.89648902e-01
-8.64019990e-01 -1.41847348e+00 1.09721646e-01 9.59288538e-01
7.27497399e-01 -1.40016645e-01 -8.50815356e-01 -7.63298929e-01
1.60485506e-03 1.02090859e+00 -7.22263575e-01 3.75168592e-01
-6.54241145e-01 -4.46765691e-01 4.63265568e-01 5.98383546e-01
6.59347832e-01 -1.04465866e+00 -1.43697071e+00 4.58612263e-01
-2.02731416e-01 -8.86165261e-01 5.10943115e-01 6.86816752e-01
-1.05328238e+00 -1.13253725e+00 -3.99425000e-01 -4.47479099e-01
2.58213013e-01 8.44871253e-02 1.04338753e+00 -5.71129739e-01
1.92383021e-01 1.04993534e+00 -5.12943864e-01 -5.60650706e-01
-4.64200303e-02 -3.40191089e-02 1.65464934e-02 -8.39629233e-01
4.29099659e-03 -5.89666843e-01 -6.60996675e-01 3.35841030e-02
-5.91562927e-01 2.52129227e-01 4.30759519e-01 6.34597480e-01
9.00057316e-01 1.37016177e-01 -5.62282559e-03 -3.25142890e-01
5.17384768e-01 -5.46148837e-01 -8.58819246e-01 -9.95187908e-02
-4.35796976e-01 5.01031816e-01 3.33832115e-01 1.23961464e-01
-1.20163310e+00 4.04330194e-01 -2.78485686e-01 2.43053492e-02
-5.98482907e-01 3.66611123e-01 4.75667007e-02 -1.70371175e-01
6.34364426e-01 1.56791195e-01 -3.44139636e-01 -3.91129524e-01
4.61082757e-01 1.25656545e-01 4.23657686e-01 -9.89904761e-01
4.98566419e-01 1.11938775e+00 3.17838550e-01 -3.05806577e-01
-6.39840066e-01 -3.42915595e-01 -5.59643269e-01 -2.63193041e-01
1.02689743e+00 -8.97836804e-01 -8.17130744e-01 1.59698650e-01
-8.88215482e-01 -1.14385664e+00 -6.88586354e-01 5.73682785e-01
-1.68458760e+00 7.79426023e-02 -3.59020948e-01 -1.07787395e+00
1.81546062e-01 -1.27511907e+00 1.33366001e+00 3.37470651e-01
1.77436292e-01 -1.00405276e+00 5.99687397e-01 9.53872800e-02
2.51467407e-01 8.73564661e-01 5.41091621e-01 -4.57343787e-01
-9.85086381e-01 2.55856663e-01 1.51746422e-01 -3.76255333e-01
-4.43565130e-01 -8.58511329e-01 -9.34646726e-01 -3.67418736e-01
-1.96478032e-02 -4.65156227e-01 9.78750527e-01 6.68337107e-01
6.24937057e-01 -2.36812711e-01 -5.60251236e-01 5.42524874e-01
1.83771443e+00 4.94668216e-01 6.29924655e-01 1.24482453e+00
1.23968869e-01 8.42671573e-01 1.12190711e+00 5.51004350e-01
8.12040806e-01 3.96212965e-01 1.29113305e+00 2.72368222e-01
2.52434701e-01 -6.13004148e-01 4.02602375e-01 -3.05170983e-01
-2.41890252e-01 -4.41071361e-01 -1.18501592e+00 6.68263674e-01
-2.41045976e+00 -9.72209692e-01 1.88375726e-01 1.92949438e+00
1.84337407e-01 1.91851661e-01 1.56254679e-01 -3.17858815e-01
9.12691951e-02 2.70359665e-01 -7.70676255e-01 -2.73011625e-01
8.84853601e-02 1.04381680e-01 1.06564629e+00 9.09010291e-01
-1.23911428e+00 1.26034939e+00 6.58806229e+00 4.22681659e-01
-2.21627295e-01 3.41228515e-01 4.27914932e-02 -1.04364373e-01
-4.12824079e-02 6.21476352e-01 -7.60817826e-01 -8.44613016e-02
9.27425623e-01 7.88660124e-02 8.59313488e-01 1.24820685e+00
2.19867826e-01 -9.48182106e-01 -1.13443673e+00 6.26829445e-01
-2.25711972e-01 -1.15159702e+00 -3.74733299e-01 3.95208448e-01
5.83568633e-01 5.58046579e-01 -1.42681420e-01 7.57761419e-01
1.40085554e+00 -9.93964493e-01 1.34228802e+00 5.85675418e-01
5.03957756e-02 -8.20114076e-01 5.20473897e-01 8.11682940e-01
-9.96477664e-01 -5.77965915e-01 -3.95543247e-01 -6.14428341e-01
4.01166677e-01 -2.43626446e-01 -9.87845540e-01 7.88977861e-01
1.01243198e+00 3.40395272e-01 -1.98053449e-01 1.19260883e+00
-5.93468130e-01 2.71108210e-01 -6.00064397e-01 -1.95922241e-01
1.05014622e+00 -2.63198376e-01 7.66830742e-01 7.48196602e-01
2.19048783e-01 -1.13448882e-02 7.12388039e-01 6.75732493e-01
5.48389614e-01 -4.46337461e-01 -6.98567033e-01 3.42202306e-01
1.48344323e-01 6.77826583e-01 -8.17583859e-01 -1.99179202e-01
1.23481885e-01 9.69882369e-01 5.13408899e-01 4.22646791e-01
-8.19990993e-01 7.31815472e-02 6.53148890e-01 -1.89483911e-01
4.26300585e-01 -5.23339987e-01 -3.20947506e-02 -6.54268682e-01
-2.18327790e-01 -4.97408807e-01 3.81562352e-01 -8.71959090e-01
-7.65282631e-01 3.25054079e-01 5.49487591e-01 -1.01669729e+00
-3.98150116e-01 -6.61547244e-01 -1.24245413e-01 6.64579391e-01
-1.86358833e+00 -1.29840398e+00 -2.37785876e-01 3.57605428e-01
6.92363262e-01 2.27238551e-01 1.08047760e+00 -5.59180498e-01
1.28883377e-01 -4.89069045e-01 3.42600733e-01 -2.73176163e-01
9.87083837e-02 -1.82836342e+00 7.04295456e-01 8.21741283e-01
-1.91627488e-01 1.36123821e-01 9.71903026e-01 -8.50799620e-01
-1.47606099e+00 -7.61151195e-01 4.39206101e-02 -6.43671870e-01
4.80030835e-01 -2.28013098e-01 -1.94068164e-01 1.11453509e+00
3.16571742e-01 -3.58804852e-01 2.23272219e-01 8.15363880e-03
2.08680511e-01 3.59818041e-01 -1.35092235e+00 8.60374093e-01
1.48331547e+00 3.78599167e-02 -9.32335615e-01 3.13393235e-01
7.19983399e-01 -1.04216290e+00 -3.85990828e-01 3.28475952e-01
6.04988277e-01 -1.29130793e+00 9.03695703e-01 -5.99119127e-01
1.95556879e-01 -3.26419473e-01 -6.45296395e-01 -1.44332051e+00
-3.26431483e-01 -5.83536208e-01 -1.88293476e-02 5.80657661e-01
8.85183588e-02 -5.10892212e-01 9.74100411e-01 3.86970311e-01
-5.22862196e-01 -6.91946268e-01 -1.41704071e+00 -8.66597414e-01
6.75401464e-02 -5.06615102e-01 5.42543888e-01 1.99942276e-01
-1.73189700e-01 -1.55335203e-01 -3.19742799e-01 4.85729188e-01
1.00119603e+00 2.47087330e-01 9.05292809e-01 -1.00407851e+00
-3.45255733e-01 -6.33697733e-02 -2.87377656e-01 -1.24805439e+00
9.34284702e-02 -5.49571693e-01 6.47779882e-01 -2.29434776e+00
-1.88699305e-01 -5.72653353e-01 -1.33539170e-01 5.87667346e-01
4.87927407e-01 -5.84564090e-01 1.77274048e-01 -1.11307269e-02
-1.06905210e+00 7.23952949e-01 1.20816720e+00 1.14438929e-01
-5.11556208e-01 -6.68905675e-02 -4.23813134e-01 9.33288157e-01
9.42363262e-01 -7.16304541e-01 -4.39389110e-01 -6.58986807e-01
1.82183459e-01 4.95837599e-01 5.01909256e-01 -1.48983848e+00
3.60814929e-01 -4.37865198e-01 3.28832924e-01 -7.07040429e-01
6.00945652e-01 -8.37845743e-01 3.00000352e-03 9.24897134e-01
-3.16558510e-01 1.67382453e-02 3.51618886e-01 1.11417019e+00
2.47280896e-01 -3.86042982e-01 4.28914607e-01 -9.25695837e-01
-1.35077143e+00 1.41249776e-01 -7.16946363e-01 2.55169235e-02
1.18823254e+00 -3.25041205e-01 -3.35587561e-01 -4.87912804e-01
-8.27831328e-01 6.35769486e-01 7.23634124e-01 1.89536244e-01
6.51643991e-01 -7.15648949e-01 -3.57995033e-01 -3.32665980e-01
-1.15975380e-01 3.34229320e-01 4.16746616e-01 6.17791295e-01
-9.21204209e-01 4.78869200e-01 -6.62700534e-01 -4.82006818e-01
-5.48010588e-01 5.75864077e-01 6.01415157e-01 -6.54629469e-01
-7.66274393e-01 5.31419039e-01 -3.70660075e-03 -9.17745292e-01
4.52819407e-01 -4.29067552e-01 -1.23198777e-01 -3.24072182e-01
8.75295699e-03 5.64151764e-01 -3.83997530e-01 -2.06791312e-01
-2.58008778e-01 2.47478515e-01 4.63731945e-01 -8.70788217e-01
1.60961807e+00 -4.03100222e-01 5.08294046e-01 3.71405065e-01
3.06192875e-01 -3.88689339e-01 -2.14985108e+00 1.15057334e-01
1.59089521e-01 -2.43303686e-01 1.63650438e-01 -1.07486045e+00
-8.17913264e-02 7.41813183e-01 9.28658485e-01 -9.20445099e-02
6.93716049e-01 -1.16421163e-01 5.28279185e-01 8.50015879e-01
1.34230971e+00 -1.38363278e+00 -1.21386491e-01 5.96132219e-01
7.65479922e-01 -1.22171485e+00 -4.24212180e-02 1.15291394e-01
-7.29071438e-01 8.97276878e-01 5.80374658e-01 -3.41975212e-01
2.07528353e-01 2.80880839e-01 -1.42304972e-01 -2.20492899e-01
-6.58658206e-01 -8.32579672e-01 -5.38587630e-01 1.17047155e+00
-7.32243657e-01 1.55208886e-01 2.09233478e-01 1.99689791e-01
-3.32207054e-01 6.04643449e-02 5.32752097e-01 1.71058476e+00
-9.42933261e-01 -8.23160350e-01 -4.49514002e-01 -1.88978761e-02
-1.11503199e-01 1.31992862e-01 -2.38097906e-01 1.13716972e+00
4.68564406e-02 9.56102073e-01 -1.71320304e-01 -7.33342990e-02
3.27423364e-01 -1.10849805e-01 9.22583938e-01 -5.19999623e-01
-2.28292584e-01 -1.12113632e-01 2.76392460e-01 -1.11311257e+00
-5.97759843e-01 -8.81032348e-01 -1.71207237e+00 9.29456577e-02
2.53598988e-01 2.05396079e-02 9.06027436e-01 1.12610853e+00
2.24287286e-01 4.20968860e-01 8.95585865e-02 -1.42285895e+00
-7.20235646e-01 -6.45097733e-01 -4.81947422e-01 -9.47719142e-02
3.55260313e-01 -1.00993168e+00 -1.31993055e-01 -3.83337051e-01] | [4.609020233154297, 0.7913588881492615] |
d9f2e112-e940-4ed7-a332-5cb6893e595c | a-framework-for-unified-real-time | 2302.11768 | null | https://arxiv.org/abs/2302.11768v1 | https://arxiv.org/pdf/2302.11768v1.pdf | A Framework for Unified Real-time Personalized and Non-Personalized Speech Enhancement | In this study, we present an approach to train a single speech enhancement network that can perform both personalized and non-personalized speech enhancement. This is achieved by incorporating a frame-wise conditioning input that specifies the type of enhancement output. To improve the quality of the enhanced output and mitigate oversuppression, we experiment with re-weighting frames by the presence or absence of speech activity and applying augmentations to speaker embeddings. By training under a multi-task learning setting, we empirically show that the proposed unified model obtains promising results on both personalized and non-personalized speech enhancement benchmarks and reaches similar performance to models that are trained specialized for either task. The strong performance of the proposed method demonstrates that the unified model is a more economical alternative compared to keeping separate task-specific models during inference. | ['Paris Smaragdis', 'Michael M. Goodwin', 'Jean-Marc Valin', 'Devansh Shah', 'Ritwik Giri', 'Zhepei Wang'] | 2023-02-23 | null | null | null | null | ['speech-enhancement'] | ['speech'] | [ 6.38195455e-01 2.97119707e-01 -1.08953185e-01 -5.59081435e-01
-1.05414283e+00 -3.08527291e-01 8.65826368e-01 -8.80415589e-02
-7.33432531e-01 5.29291213e-01 7.60680377e-01 -2.30129480e-01
1.04710020e-01 -5.05166054e-01 -7.87078798e-01 -7.39905715e-01
2.01806173e-01 -1.19350776e-02 1.74234167e-01 -3.41528147e-01
-6.71829358e-02 2.55618453e-01 -1.74301922e+00 4.00943071e-01
8.42234194e-01 7.93963790e-01 4.62977648e-01 8.78363490e-01
9.07290950e-02 3.91855568e-01 -8.15214157e-01 -3.99413735e-01
1.08753480e-01 -3.50358307e-01 -6.14706576e-01 1.51928648e-01
4.81651127e-01 -4.52971727e-01 -5.03386319e-01 8.19082916e-01
9.85354662e-01 6.71149492e-01 3.45409274e-01 -9.76348102e-01
-7.36170471e-01 8.79533887e-01 -1.22030430e-01 3.78468931e-01
2.11922899e-01 -3.44695374e-02 1.04264188e+00 -8.66401196e-01
4.07271177e-01 1.10517895e+00 5.09280503e-01 9.10446346e-01
-1.29299295e+00 -3.22401106e-01 3.26277137e-01 1.04263954e-01
-1.05719376e+00 -1.01435506e+00 8.01627994e-01 -8.69036000e-03
1.12563729e+00 2.55359471e-01 1.42538682e-01 1.47285187e+00
-3.08341116e-01 8.76471519e-01 1.00116313e+00 -4.60925728e-01
2.61975676e-01 3.56586397e-01 1.03916056e-01 3.65950406e-01
-3.97472858e-01 4.35627073e-01 -7.62963712e-01 -5.11979870e-02
3.48802894e-01 -3.14392924e-01 -5.16575634e-01 6.52649105e-02
-1.14346135e+00 5.85892975e-01 1.24463648e-01 5.27917206e-01
-3.87426704e-01 2.46381596e-01 6.38184726e-01 3.48021924e-01
6.81793094e-01 3.22301477e-01 -5.07855773e-01 -3.38702232e-01
-1.32890296e+00 1.02335811e-01 4.08837199e-01 7.50824451e-01
6.38920963e-01 4.35425878e-01 -9.03786659e-01 1.14151382e+00
8.24409798e-02 3.38857174e-01 6.62180424e-01 -9.39193904e-01
4.10210997e-01 -1.41544312e-01 -1.13271177e-02 -3.69426966e-01
-1.35360852e-01 -7.87928045e-01 -5.02038777e-01 1.53419733e-01
1.47308812e-01 -3.42732012e-01 -1.07399297e+00 2.32464361e+00
1.25798762e-01 5.75008094e-01 2.69723386e-01 6.29687130e-01
7.24745452e-01 9.33829367e-01 4.20991540e-01 -1.32434234e-01
1.37846708e+00 -1.27708960e+00 -1.10788870e+00 -1.47202522e-01
4.18533176e-01 -7.40869999e-01 1.41214240e+00 1.69632003e-01
-1.13446987e+00 -6.88763916e-01 -1.17231476e+00 -2.00562060e-01
-4.26763117e-01 2.12253034e-01 3.19192141e-01 9.44274902e-01
-1.37821579e+00 7.11691022e-01 -6.66907668e-01 -2.09250867e-01
9.11913067e-02 1.35843202e-01 -2.58520186e-01 1.65507138e-01
-1.44507861e+00 9.70398724e-01 3.02958369e-01 -2.43454754e-01
-9.49265420e-01 -9.05889750e-01 -9.34479415e-01 5.12081265e-01
6.72675073e-02 -7.52718806e-01 1.41792679e+00 -9.60185945e-01
-1.99000359e+00 4.65327650e-01 -3.20112675e-01 -4.84624535e-01
2.34654173e-01 -3.18964541e-01 -6.00503147e-01 1.00302979e-01
-3.66726696e-01 8.30887437e-01 1.12122154e+00 -1.16212654e+00
-6.20248616e-01 -1.23814791e-01 2.83839822e-01 3.21193129e-01
-9.58570719e-01 6.82839006e-02 -4.95763034e-01 -9.28556681e-01
-4.60750639e-01 -6.51681423e-01 -3.51553448e-02 -1.55046970e-01
-2.70720899e-01 -1.24008492e-01 1.11957693e+00 -8.88506353e-01
1.35697508e+00 -2.34527731e+00 1.86954215e-01 -1.22843057e-01
-1.53889373e-01 4.95203823e-01 -5.79391360e-01 2.50402480e-01
-5.27914390e-02 1.39924005e-01 -3.44631970e-01 -1.09663546e+00
2.74933130e-01 2.87164211e-01 -1.88867286e-01 4.74250168e-02
4.04238462e-01 5.66951632e-01 -8.02646756e-01 -2.01643229e-01
3.00305873e-01 1.04905677e+00 -7.87271142e-01 3.75050068e-01
-9.68837664e-02 3.48764807e-01 2.90130731e-03 2.19629750e-01
6.20808661e-01 6.72064200e-02 8.78639221e-02 -2.67051101e-01
-8.46105721e-03 6.50308669e-01 -1.13709331e+00 1.75906837e+00
-9.15597260e-01 6.62978530e-01 2.41936862e-01 -1.00227809e+00
6.06815696e-01 7.48790085e-01 1.56685665e-01 -6.37332320e-01
-2.50795763e-02 -1.63903639e-01 -2.37325132e-01 -3.65593523e-01
8.43921244e-01 -2.36788273e-01 1.43153310e-01 4.81090158e-01
4.88988370e-01 1.27667069e-01 1.29874393e-01 6.59452304e-02
1.15469873e+00 1.69290856e-01 5.05248010e-02 -2.37439498e-01
6.31555557e-01 -8.15574944e-01 4.58115935e-01 8.73345494e-01
-3.39123577e-01 4.92413193e-01 5.09439074e-02 2.89062440e-01
-1.01434886e+00 -1.25769579e+00 -1.68135300e-01 1.90054357e+00
-1.85584247e-01 -2.89841145e-01 -8.78209651e-01 -6.49828374e-01
-3.86226594e-01 9.40828621e-01 -5.08509934e-01 -4.71808910e-01
-6.01386607e-01 -6.46507740e-01 7.42700994e-01 6.12562716e-01
3.39893341e-01 -9.35062468e-01 -2.17844903e-01 1.99058458e-01
-3.44795972e-01 -1.27362406e+00 -9.19518769e-01 3.47126126e-01
-7.47297049e-01 -1.68503404e-01 -9.49569345e-01 -8.23731065e-01
4.68351334e-01 1.93641528e-01 9.28033590e-01 -2.89594457e-02
3.60692948e-01 4.59657252e-01 -4.65363890e-01 -4.78794053e-02
-6.05958700e-01 2.70132244e-01 -2.44522523e-02 1.83463573e-01
-1.21680081e-01 -7.56098032e-01 -4.01612401e-01 3.98091301e-02
-1.00405848e+00 -9.60605145e-02 4.86954510e-01 1.06160176e+00
2.79228508e-01 -1.25126868e-01 1.03736842e+00 -5.75116575e-01
9.21482146e-01 -2.40814328e-01 -1.69662401e-01 2.69651294e-01
-5.42625368e-01 4.36573565e-01 6.13939404e-01 -5.59585750e-01
-1.59597600e+00 -8.50111172e-02 -7.82855570e-01 -2.92938948e-01
-2.48030886e-01 2.95919448e-01 -3.56843412e-01 1.40693456e-01
5.05250692e-01 2.05627352e-01 -1.42290249e-01 -6.22585416e-01
6.41303539e-01 7.35725820e-01 5.38018048e-01 -6.22803330e-01
6.08219981e-01 9.04964283e-02 -4.83746260e-01 -9.38870311e-01
-6.68126523e-01 -3.55361730e-01 -1.79373577e-01 -9.71374735e-02
8.11551273e-01 -9.35828269e-01 -2.16766819e-01 2.88731784e-01
-1.25764799e+00 -4.62743402e-01 -3.69979024e-01 5.01053810e-01
-5.66641867e-01 5.79186916e-01 -6.67530715e-01 -8.74213696e-01
-5.45862436e-01 -1.29928768e+00 1.17246473e+00 4.85725328e-02
-1.04398141e-03 -1.11739373e+00 8.26193914e-02 2.16146007e-01
8.83332849e-01 -3.66398394e-01 8.48925352e-01 -7.90117145e-01
-1.80748105e-01 1.27079353e-01 2.89674774e-02 7.88242459e-01
2.14312762e-01 -2.80357618e-02 -1.63713145e+00 -2.67468005e-01
-7.57049248e-02 -4.65820879e-02 1.08777010e+00 4.10000086e-01
1.22517288e+00 -2.04914048e-01 2.60035079e-02 5.83589494e-01
1.06780946e+00 1.37746856e-01 7.56293356e-01 1.70327559e-01
1.89626902e-01 4.36063439e-01 3.40757698e-01 5.60384870e-01
1.94658205e-01 9.86694336e-01 1.97476238e-01 -2.05094010e-01
-4.96469378e-01 -2.38241851e-01 7.70298839e-01 7.64090896e-01
3.10920682e-02 -5.21550894e-01 -2.69250423e-01 6.01021528e-01
-1.56226492e+00 -1.18807101e+00 4.42586213e-01 2.15078282e+00
1.06897664e+00 8.85775164e-02 1.82933897e-01 2.86173820e-01
9.99918222e-01 3.85221034e-01 -1.77712291e-01 -6.38931155e-01
-3.13149184e-01 6.64523721e-01 1.39903918e-01 7.54257739e-01
-1.14111710e+00 8.87261868e-01 7.22420549e+00 9.26124454e-01
-1.20504916e+00 4.65381861e-01 4.46764797e-01 -2.15633735e-01
-4.82328624e-01 -4.24491793e-01 -6.86715543e-01 4.55803007e-01
1.34840035e+00 -1.14757247e-01 5.90425432e-01 7.63479114e-01
4.76083815e-01 4.73375112e-01 -1.01673579e+00 6.34672403e-01
-3.56483944e-02 -1.25527334e+00 5.11197001e-02 -2.42404118e-01
5.85759461e-01 -3.31980400e-02 3.55576396e-01 6.32636189e-01
5.28407991e-02 -6.62344873e-01 8.17471683e-01 1.57390788e-01
6.78555429e-01 -6.09369040e-01 5.88858068e-01 1.79111719e-01
-1.10813367e+00 -8.26924741e-02 -3.75232957e-02 1.45623431e-01
5.69331706e-01 4.87637013e-01 -6.96687043e-01 4.32454079e-01
4.75899279e-01 2.64838696e-01 -3.03685546e-01 8.62491727e-01
-4.15474236e-01 6.72563195e-01 -2.40759939e-01 2.37630874e-01
1.23386696e-01 1.37784064e-01 6.46979511e-01 1.63729513e+00
5.41534185e-01 -2.96180665e-01 -2.28667289e-01 6.13234103e-01
-2.55230784e-01 -2.90880892e-02 -3.90365273e-01 -9.30026546e-02
5.36623001e-01 1.27896535e+00 -1.59182563e-01 -5.48019230e-01
-4.98619646e-01 1.10291851e+00 3.52651149e-01 5.47703683e-01
-1.14907968e+00 -3.37976128e-01 9.77583945e-01 -1.68627605e-01
6.03702426e-01 -3.04644499e-02 -1.85521215e-01 -1.03764081e+00
-1.39826447e-01 -7.75155544e-01 2.15188861e-01 -7.17410088e-01
-1.04477274e+00 1.00506926e+00 -1.61183178e-01 -7.81019688e-01
-3.45132381e-01 -4.49496537e-01 -7.65953243e-01 8.67374003e-01
-1.68983090e+00 -1.05874872e+00 -7.44666904e-02 5.70714414e-01
6.24550104e-01 6.82660714e-02 8.31166983e-01 7.95983732e-01
-5.86897016e-01 1.12942374e+00 -1.86310075e-02 -3.49750340e-01
9.32173550e-01 -1.30982280e+00 2.25161254e-01 1.10583067e+00
2.75974944e-02 5.04463792e-01 9.98084366e-01 -2.14897662e-01
-9.26335216e-01 -1.12435555e+00 1.02716899e+00 1.15444465e-03
3.60814393e-01 -3.76188070e-01 -9.09423828e-01 6.16137505e-01
6.89268053e-01 -2.35348284e-01 7.83327818e-01 2.58723587e-01
-3.56237173e-01 -1.00336201e-01 -1.29375076e+00 5.68978429e-01
1.10292280e+00 -7.91035831e-01 -6.65297270e-01 9.91642997e-02
1.27404320e+00 -2.56085485e-01 -8.67279947e-01 4.11632389e-01
2.60657728e-01 -7.20515549e-01 1.02585673e+00 -5.79379320e-01
3.24958950e-01 -9.08610672e-02 -4.15070593e-01 -1.55297983e+00
-3.17993015e-01 -7.44009733e-01 -1.70516938e-01 1.52950084e+00
6.70207858e-01 -4.61818516e-01 5.81022799e-01 4.49184567e-01
-7.89485514e-01 -5.38854659e-01 -1.02069116e+00 -8.38104606e-01
-2.45146081e-02 -5.33751428e-01 6.70026779e-01 7.42774487e-01
-3.63605097e-02 2.60346532e-01 -6.85077131e-01 3.30167115e-01
3.03951055e-01 -4.26815212e-01 3.46805602e-01 -6.29190207e-01
-6.14939749e-01 -5.43522000e-01 -3.07399649e-02 -1.15039849e+00
4.64967787e-01 -9.38367546e-01 2.40125239e-01 -1.22894216e+00
-6.07257672e-02 -1.64686829e-01 -6.43491566e-01 4.44478035e-01
-4.46328670e-01 1.76688790e-01 6.94209710e-02 -5.54900110e-01
-3.23260903e-01 8.56370389e-01 1.05283320e+00 -6.21472672e-02
-2.41421878e-01 -2.17156019e-02 -7.73883283e-01 2.93232203e-01
8.02394927e-01 -2.22119555e-01 -5.17926157e-01 -6.84290826e-01
-2.64535397e-01 -9.58409607e-02 1.90804392e-01 -9.22398746e-01
2.02504128e-01 1.41165748e-01 -1.15302823e-01 -3.00133586e-01
7.43380368e-01 -5.67208111e-01 -4.24583815e-02 1.38150662e-01
-7.00291574e-01 -1.68051153e-01 3.70450646e-01 5.17877996e-01
-3.13749492e-01 -3.91093761e-01 9.60194290e-01 3.65556449e-01
-6.82317197e-01 2.03586414e-01 -5.68627775e-01 -8.07237178e-02
6.16792679e-01 -3.26573313e-03 -3.82570922e-01 -5.56206405e-01
-1.00564396e+00 -1.63669541e-01 8.08063373e-02 4.82749552e-01
5.87991059e-01 -1.37760198e+00 -6.23796582e-01 1.78641185e-01
3.30711640e-02 -7.55758822e-01 5.63395321e-01 6.73904359e-01
2.83504277e-01 2.09732056e-01 -1.52610317e-02 -2.96285033e-01
-1.29553878e+00 5.02400160e-01 3.62202734e-01 -3.96854073e-01
-4.21746254e-01 8.25458288e-01 1.09547928e-01 -5.66665828e-01
4.91778612e-01 -2.03213021e-01 -1.33407250e-01 -1.19387992e-01
6.42516494e-01 3.79726410e-01 3.06754619e-01 -5.15407681e-01
-2.29588449e-01 3.56050134e-02 -4.09740284e-02 -4.73072290e-01
1.37977314e+00 -2.41329864e-01 3.35181504e-01 9.79963243e-02
1.10053945e+00 1.02776349e-01 -1.35287035e+00 -3.67882878e-01
-2.15655535e-01 -2.75717616e-01 4.98306572e-01 -8.96807909e-01
-1.10256648e+00 9.25710738e-01 6.64418161e-01 7.37142712e-02
1.43489254e+00 -1.18864672e-02 7.38548934e-01 2.12888017e-01
-7.78453276e-02 -1.29166734e+00 2.01388329e-01 6.48412466e-01
8.46469998e-01 -9.56690550e-01 -5.19573390e-01 -2.74578571e-01
-6.08357787e-01 7.62715042e-01 4.68994766e-01 2.21826792e-01
6.02123082e-01 5.29380679e-01 -1.17416017e-01 3.30748588e-01
-8.45329404e-01 -4.05048907e-01 2.91428775e-01 7.66166270e-01
6.66347444e-01 -3.14639434e-02 -2.95072377e-01 7.10884869e-01
-1.27782658e-01 -1.82404578e-01 2.80517519e-01 8.13995183e-01
-5.42196989e-01 -1.37608278e+00 -1.09234378e-01 1.26107976e-01
-4.59794402e-01 -4.63575333e-01 -7.21942307e-03 5.49074769e-01
1.07428469e-02 1.17704892e+00 -5.63521928e-04 -4.52413529e-01
4.50494409e-01 3.51633519e-01 4.66118872e-01 -4.97564286e-01
-8.65605235e-01 1.57325760e-01 5.02967715e-01 -3.74258995e-01
-4.42969203e-01 -3.85876983e-01 -9.76406634e-01 -2.50559568e-01
-3.51898253e-01 2.98722312e-02 8.05421948e-01 1.01561964e+00
5.96241951e-01 9.85180795e-01 6.20102048e-01 -9.90110397e-01
-8.35009217e-01 -1.08991730e+00 -4.64743793e-01 5.29181659e-01
4.72955018e-01 -7.49654055e-01 -6.27255023e-01 1.56411201e-01] | [14.895088195800781, 6.037303447723389] |
b6342512-ea33-4804-a77c-f676b42fd6a0 | auto-mvcnn-neural-architecture-search-for | 2012.05493 | null | https://arxiv.org/abs/2012.05493v1 | https://arxiv.org/pdf/2012.05493v1.pdf | Auto-MVCNN: Neural Architecture Search for Multi-view 3D Shape Recognition | In 3D shape recognition, multi-view based methods leverage human's perspective to analyze 3D shapes and have achieved significant outcomes. Most existing research works in deep learning adopt handcrafted networks as backbones due to their high capacity of feature extraction, and also benefit from ImageNet pretraining. However, whether these network architectures are suitable for 3D analysis or not remains unclear. In this paper, we propose a neural architecture search method named Auto-MVCNN which is particularly designed for optimizing architecture in multi-view 3D shape recognition. Auto-MVCNN extends gradient-based frameworks to process multi-view images, by automatically searching the fusion cell to explore intrinsic correlation among view features. Moreover, we develop an end-to-end scheme to enhance retrieval performance through the trade-off parameter search. Extensive experimental results show that the searched architectures significantly outperform manually designed counterparts in various aspects, and our method achieves state-of-the-art performance at the same time. | ['Jinxing Li', 'Hongren Wang', 'Zhaoqun Li'] | 2020-12-10 | null | null | null | null | ['3d-shape-recognition'] | ['computer-vision'] | [-3.77469927e-01 -5.19027770e-01 -1.40218228e-01 -4.34135139e-01
-6.46079183e-01 -5.82411408e-01 5.79105437e-01 -5.38498521e-01
-1.68072313e-01 -1.75396875e-02 1.98541597e-01 -8.62600133e-02
-3.84148568e-01 -7.52842307e-01 -5.73689222e-01 -7.08271861e-01
2.03295365e-01 6.87867582e-01 -1.60241619e-01 -1.72016144e-01
2.15221256e-01 9.48555470e-01 -1.26286924e+00 2.51577824e-01
3.95037562e-01 1.37447262e+00 1.83731884e-01 3.27156603e-01
-1.60496220e-01 3.22254568e-01 -3.11947633e-02 -5.98660231e-01
5.39139390e-01 1.84514731e-01 -5.17473102e-01 6.60137385e-02
6.06152713e-01 -5.00584245e-01 -4.19808805e-01 9.80799437e-01
1.12906146e+00 -1.10132784e-01 5.26735425e-01 -9.68821585e-01
-1.08360910e+00 2.53824532e-01 -5.64390898e-01 2.59747982e-01
9.07040015e-02 1.40784800e-01 1.31178737e+00 -1.57580960e+00
3.69785070e-01 1.39362609e+00 4.55653071e-01 4.92009193e-01
-8.53040755e-01 -5.54479837e-01 2.95860201e-01 2.86634475e-01
-1.38946426e+00 -5.01279831e-01 1.27740300e+00 -9.58640575e-02
1.32404566e+00 5.18994033e-02 8.74992371e-01 1.18121016e+00
8.66680890e-02 1.36308849e+00 8.19933772e-01 -2.08738551e-01
-2.52500504e-01 -3.53880614e-01 -4.55405921e-01 9.31159616e-01
8.21433738e-02 2.33019635e-01 -6.16334438e-01 -2.46585179e-02
1.01174343e+00 2.79645264e-01 -2.56737351e-01 -1.08246708e+00
-1.20234299e+00 6.63108528e-01 5.95035493e-01 2.93430746e-01
-5.14402390e-01 -7.22732116e-03 3.42435390e-01 2.07910344e-01
4.89500761e-01 2.92847872e-01 -5.49690604e-01 2.00834125e-01
-6.95828080e-01 9.75759923e-02 5.00225008e-01 8.97651851e-01
5.11809528e-01 3.13555866e-01 -3.31444182e-02 1.13986003e+00
4.82970893e-01 6.64134681e-01 4.20124888e-01 -6.05332434e-01
6.79092169e-01 1.08851957e+00 -2.55963743e-01 -1.07047033e+00
-5.72620392e-01 -7.62703836e-01 -1.16667140e+00 -3.14822756e-02
-8.49040002e-02 2.41567537e-01 -9.59335089e-01 1.41187954e+00
3.53430718e-01 -1.28709808e-01 1.12794312e-02 1.27037573e+00
1.05760610e+00 2.15207711e-01 -4.31158155e-01 1.77233428e-01
1.20360231e+00 -1.22304738e+00 -2.36530453e-01 -1.05547972e-01
1.74360469e-01 -7.26752639e-01 9.35101330e-01 2.93040246e-01
-1.39999044e+00 -6.41096950e-01 -1.00289559e+00 8.02183375e-02
-3.48625571e-01 3.45941663e-01 6.74901068e-01 4.00032878e-01
-1.05303812e+00 4.08603966e-01 -8.26595128e-01 -8.43806490e-02
6.56418145e-01 4.19039160e-01 -3.80444258e-01 -6.23209327e-02
-9.09024358e-01 6.34685874e-01 1.77793443e-01 4.25151229e-01
-9.61639106e-01 -5.75604022e-01 -9.13902104e-01 3.55428249e-01
4.15627241e-01 -1.16028309e+00 1.09914255e+00 -4.30991530e-01
-1.47543013e+00 1.12289572e+00 -1.03829885e-02 -4.63100038e-02
2.82505631e-01 -2.09766671e-01 -3.62277538e-01 2.77767599e-01
-9.01290774e-02 6.48763597e-01 9.71366882e-01 -1.33911061e+00
-3.92479926e-01 -8.29177320e-01 1.24106981e-01 4.61311489e-01
-6.41334593e-01 -2.23129570e-01 -9.24740136e-01 -7.31292248e-01
5.38209558e-01 -7.92790353e-01 -2.41466448e-01 -4.35235202e-02
-2.82075107e-01 -3.44097137e-01 8.02119136e-01 -1.27505705e-01
8.60958576e-01 -1.91040134e+00 3.42480063e-01 2.87492454e-01
3.73988420e-01 3.77028912e-01 -4.32587475e-01 3.47681165e-01
9.72203463e-02 -2.51409300e-02 1.30170003e-01 -3.50750208e-01
6.64105564e-02 8.49244371e-02 -2.21359000e-01 3.10318172e-01
1.76967025e-01 1.36653662e+00 -5.48177660e-01 -4.01605368e-01
1.90752253e-01 5.30233562e-01 -5.70215762e-01 3.03214490e-01
6.73469752e-02 2.39288405e-01 -7.71056354e-01 1.12123728e+00
7.41954029e-01 -7.29676366e-01 2.14979406e-02 -5.81303537e-01
1.28565878e-01 -1.12415276e-01 -9.21083987e-01 2.08380270e+00
-5.52851856e-01 -6.76288619e-04 1.33023247e-01 -1.09232402e+00
1.24794590e+00 1.66570261e-01 6.41411841e-01 -7.81852245e-01
1.27116635e-01 2.60584652e-01 -1.54114023e-01 -4.26419675e-01
2.90255934e-01 1.52490571e-01 2.48509049e-01 5.46362519e-01
2.37436414e-01 -1.17534161e-01 -2.50800431e-01 -2.62732089e-01
6.72434628e-01 2.00182170e-01 2.86362141e-01 7.21775070e-02
8.77900898e-01 -5.01866400e-01 4.39948499e-01 5.60489178e-01
-1.67694241e-01 6.79396749e-01 4.86363750e-03 -9.40896153e-01
-9.89141107e-01 -9.22022760e-01 -1.24443583e-02 9.04649377e-01
2.46996552e-01 -1.02737062e-02 -3.56769145e-01 -9.17743683e-01
1.10960290e-01 1.58140257e-01 -3.05733085e-01 -5.42150289e-02
-7.48159766e-01 -5.24353802e-01 5.14451444e-01 8.21203589e-01
7.39117801e-01 -9.32585895e-01 -6.44998729e-01 2.31850132e-01
1.41717508e-01 -1.31213963e+00 -6.53195143e-01 -1.17357977e-01
-1.12024617e+00 -9.77058232e-01 -1.10266769e+00 -9.62880492e-01
5.56772411e-01 7.16819942e-01 1.07156467e+00 9.50581133e-02
-1.04332313e-01 7.56799936e-01 -3.09393615e-01 -2.68499345e-01
1.70864150e-01 4.03189301e-01 -2.56541148e-02 2.28462085e-01
4.28148270e-01 -1.02205706e+00 -1.00987887e+00 5.56176245e-01
-7.85221636e-01 -2.01037332e-01 9.87532377e-01 1.08115947e+00
8.30380857e-01 -5.57389200e-01 5.35181165e-01 -2.91789591e-01
4.98817295e-01 3.68788913e-02 -6.72016442e-01 5.60079217e-01
-8.17498863e-01 8.53379741e-02 6.04681909e-01 -8.74835551e-02
-7.64898062e-01 1.23933680e-01 -3.19917709e-01 -1.24819136e+00
-3.15755814e-01 5.12063146e-01 -4.23553824e-01 -4.61264431e-01
2.41069123e-01 7.03823984e-01 1.41630083e-01 -6.25081241e-01
3.35347563e-01 3.55505556e-01 2.41897061e-01 -3.82257611e-01
6.73625708e-01 7.04478443e-01 1.42079834e-02 -4.32886839e-01
-1.11272085e+00 -4.99081731e-01 -6.88183427e-01 -1.46862656e-01
7.03499615e-01 -1.13111126e+00 -9.84742522e-01 4.83141392e-01
-1.10936749e+00 2.40465999e-01 2.08162129e-01 4.75909352e-01
-7.36225843e-01 4.30745542e-01 -4.54305649e-01 -5.51581740e-01
-9.23051894e-01 -1.33532500e+00 1.39105213e+00 1.61177427e-01
3.75256836e-01 -9.09189582e-01 -1.00109041e-01 4.80264366e-01
5.16430497e-01 -1.42621219e-01 1.09692991e+00 -8.34580660e-01
-8.72578859e-01 -1.95171416e-01 -4.75665450e-01 1.13135405e-01
1.09298481e-02 -5.30812383e-01 -9.50196087e-01 -5.46365082e-01
1.00483736e-02 -4.28091407e-01 9.61057305e-01 3.24514478e-01
1.56309187e+00 -1.49108112e-01 -2.10528046e-01 9.82961893e-01
1.46746969e+00 1.42995358e-01 3.69753063e-01 3.04507494e-01
8.97135615e-01 2.62191683e-01 1.75982401e-01 4.78764176e-01
5.75852513e-01 7.94969320e-01 8.53263080e-01 3.03055122e-02
4.36953194e-02 -2.31944576e-01 7.53172161e-03 1.30656445e+00
-2.15954959e-01 -1.98142126e-01 -9.84532714e-01 6.69071794e-01
-1.73769808e+00 -8.71611357e-01 5.54013669e-01 1.68030941e+00
3.38759243e-01 6.21962734e-02 -7.07359761e-02 -2.80288875e-01
4.27713275e-01 5.99348664e-01 -9.23172295e-01 1.15973921e-02
-3.09085608e-01 6.04021885e-02 1.66578591e-01 8.32223222e-02
-1.26309109e+00 8.92559648e-01 5.52208853e+00 1.01108921e+00
-1.32388127e+00 -1.59871906e-01 5.51020503e-01 -1.72514215e-01
-4.56265092e-01 -3.85038525e-01 -8.96927953e-01 -2.02561572e-01
6.83638304e-02 3.34495544e-01 4.65895146e-01 9.06074524e-01
-1.08285196e-01 6.95677161e-01 -9.63872850e-01 1.58628356e+00
4.82960939e-01 -1.43521345e+00 5.16584277e-01 1.94560513e-01
7.83833027e-01 2.43041396e-01 2.59739339e-01 1.26186565e-01
-4.01451029e-02 -8.83806765e-01 6.38643503e-01 5.16244054e-01
5.87223947e-01 -8.14162731e-01 5.48974514e-01 3.27219784e-01
-1.39342046e+00 -1.97935283e-01 -3.83482218e-01 4.13907886e-01
2.41987422e-01 4.66861188e-01 -5.68576872e-01 8.99645150e-01
6.58621371e-01 8.98218513e-01 -5.32299578e-01 1.10126925e+00
-3.15024704e-02 1.12770289e-01 -1.42214701e-01 -1.51174352e-01
6.13129735e-01 -2.37785697e-01 8.31603169e-01 7.25593865e-01
5.76956332e-01 -8.00663456e-02 2.12290347e-01 9.35499549e-01
-1.21620581e-01 2.09478185e-01 -9.07093227e-01 2.69403607e-02
4.04289693e-01 1.44014370e+00 -5.71121454e-01 -7.87302777e-02
-5.47455013e-01 8.88576269e-01 5.33949494e-01 3.24860334e-01
-4.83474582e-01 -1.78960800e-01 4.76769030e-01 -3.47596139e-01
9.87873554e-01 -3.38445961e-01 -3.11258197e-01 -1.35041857e+00
2.87530363e-01 -1.10880017e+00 3.22565496e-01 -7.36051023e-01
-1.53163838e+00 7.09582508e-01 -3.55685204e-01 -1.51288116e+00
-8.96841437e-02 -9.01489258e-01 -3.75032246e-01 5.71475804e-01
-1.70364165e+00 -1.62084246e+00 -2.17326239e-01 6.18341625e-01
6.86837018e-01 -8.29088748e-01 8.74467492e-01 3.89348537e-01
-3.78112823e-01 8.45977962e-01 -6.78828135e-02 3.48809898e-01
4.81586188e-01 -9.85952139e-01 5.64248145e-01 4.06331420e-01
6.10550523e-01 5.18568873e-01 -2.14393362e-01 -3.47355813e-01
-2.03270793e+00 -9.79640067e-01 5.09977818e-01 -3.00937086e-01
2.56669700e-01 -1.66642725e-01 -6.81772947e-01 2.78522342e-01
2.92404681e-01 1.29168034e-01 7.24039137e-01 2.11411744e-01
-6.10381722e-01 -2.34058648e-01 -7.72027373e-01 6.09611452e-01
1.48700547e+00 -5.60271859e-01 -4.03776050e-01 -2.99229175e-02
6.16236806e-01 -6.40516818e-01 -9.76340473e-01 7.57142723e-01
8.70144904e-01 -1.08508921e+00 1.43140829e+00 -8.84302080e-01
5.39672613e-01 -1.51097834e-01 -5.25359690e-01 -1.28841984e+00
-4.89245057e-01 -2.29700819e-01 -3.38966876e-01 8.74868631e-01
4.54249531e-01 -4.91495728e-01 1.03624785e+00 8.26189741e-02
-3.09994787e-01 -1.49189878e+00 -9.69367087e-01 -7.78075874e-01
-1.61671788e-02 -2.48992905e-01 9.99105692e-01 8.03630114e-01
-7.07151771e-01 5.44191837e-01 -3.58389318e-01 2.38767967e-01
6.98054194e-01 1.00174701e+00 7.54624307e-01 -1.16151583e+00
-2.72266090e-01 -8.44911218e-01 -4.63031650e-01 -1.56465745e+00
2.74507821e-01 -1.15423954e+00 -4.30920899e-01 -1.33018148e+00
3.39909852e-01 -3.14861119e-01 -3.90857875e-01 4.25171942e-01
-3.37177813e-02 1.48767129e-01 4.14973289e-01 2.76520342e-01
-8.00895393e-01 1.07805920e+00 1.80714595e+00 -2.74789929e-01
-6.54915646e-02 5.09339608e-02 -5.99072695e-01 7.97167122e-01
6.36249781e-01 3.06757074e-02 -2.97216237e-01 -9.59275961e-01
2.83822715e-01 1.17949486e-01 6.00445330e-01 -6.75616622e-01
4.47170556e-01 -1.43461730e-02 6.59315050e-01 -1.12006366e+00
5.50765753e-01 -1.00597370e+00 -2.56637216e-01 1.40373379e-01
-1.29778534e-01 4.06135440e-01 2.43000407e-02 7.47168124e-01
-5.05902886e-01 -1.86218992e-02 5.56377351e-01 -4.71234679e-01
-6.84494734e-01 8.88430655e-01 2.15274155e-01 -2.34819539e-02
6.21202290e-01 -3.13169301e-01 -8.97457898e-02 -4.06734318e-01
-6.36763632e-01 3.46957237e-01 1.39708683e-01 5.40062726e-01
1.08519435e+00 -1.87209928e+00 -4.01001811e-01 4.08995450e-01
1.70757025e-01 7.03997239e-02 3.41841578e-01 8.13733041e-01
-2.49336779e-01 6.95436835e-01 -1.85977980e-01 -8.70827317e-01
-1.02546072e+00 5.95319927e-01 3.65373909e-01 -5.23057342e-01
-5.12695491e-01 8.31796050e-01 2.63854951e-01 -8.71335745e-01
3.96697611e-01 7.29431137e-02 -2.88382381e-01 1.37224317e-01
2.36934409e-01 5.61899096e-02 2.70526856e-01 -5.03516257e-01
-3.15512657e-01 1.11736274e+00 -2.77961910e-01 1.18766680e-01
1.73369014e+00 -3.76474336e-02 -7.92162586e-03 3.58090922e-02
1.43571246e+00 -3.13754082e-01 -1.06295490e+00 -5.36108792e-01
-1.95191994e-01 -4.56803739e-01 9.29379389e-02 -6.12051010e-01
-1.58162105e+00 9.85117316e-01 6.72821045e-01 1.63511112e-02
1.16945207e+00 -3.10979430e-02 1.07779300e+00 8.25664818e-01
3.61425906e-01 -8.55220020e-01 3.20597082e-01 8.76099467e-01
1.18757582e+00 -1.35900569e+00 -2.35532857e-02 -9.33619663e-02
-5.24663329e-01 1.24504566e+00 8.80527437e-01 -1.16102122e-01
9.52212989e-01 -5.85946850e-02 1.76423982e-01 -7.01055884e-01
-7.71602690e-01 -3.33133996e-01 6.29816234e-01 3.94920647e-01
6.26207814e-02 -2.09025845e-01 1.62421420e-01 6.56471729e-01
2.68206019e-02 -3.67900848e-01 -2.63430893e-01 6.97553396e-01
-2.57145494e-01 -1.09566736e+00 -1.44083351e-01 4.87926066e-01
-4.13190752e-01 -4.37686220e-02 -5.02676606e-01 6.20068669e-01
-2.74490356e-01 3.63487095e-01 -1.80194423e-01 -4.58274841e-01
6.96153939e-01 1.12424128e-01 6.10683203e-01 -2.15405136e-01
-6.13928974e-01 4.81131792e-01 -3.11731666e-01 -6.31916046e-01
-6.97996378e-01 -5.32959998e-01 -6.93670630e-01 -2.14550439e-02
-5.32530367e-01 -2.84742326e-01 4.26346958e-01 9.21484709e-01
8.29096079e-01 1.12762511e-01 9.10302401e-01 -1.04790163e+00
-9.25898194e-01 -6.63313448e-01 -2.86434472e-01 1.50796682e-01
2.61173069e-01 -7.56915867e-01 1.14783175e-01 -4.19317335e-01] | [8.17464542388916, -3.7847399711608887] |
b4d23413-f052-4484-84da-524bfbf3601b | hierarchical-reinforcement-learning-for-ris | 2301.02771 | null | https://arxiv.org/abs/2301.02771v1 | https://arxiv.org/pdf/2301.02771v1.pdf | Hierarchical Reinforcement Learning for RIS-Assisted Energy-Efficient RAN | Reconfigurable intelligent surface (RIS) is emerging as a promising technology to boost the energy efficiency (EE) of 5G beyond and 6G networks. Inspired by this potential, in this paper, we investigate the RIS-assisted energy-efficient radio access networks (RAN). In particular, we combine RIS with sleep control techniques, and develop a hierarchical reinforcement learning (HRL) algorithm for network management. In HRL, the meta-controller decides the on/off status of the small base stations (SBSs) in heterogeneous networks, while the sub-controller can change the transmission power levels of SBSs to save energy. The simulations show that the RIS-assisted sleep control can achieve significantly lower energy consumption, higher throughput, and more than doubled energy efficiency than no-RIS conditions. | ['Melike Erol-Kantarci', 'Steve Furr', 'Raimundas Gaigalas', 'Majid Bavand', 'Medhat Elsayed', 'Long Kong', 'Hao Zhou'] | 2023-01-07 | null | null | null | null | ['hierarchical-reinforcement-learning'] | ['methodology'] | [-4.88390774e-02 4.03498769e-01 -7.31483757e-01 -2.62346901e-02
3.32139105e-01 -3.21456730e-01 -1.72644481e-01 -4.80828494e-01
8.50294977e-02 1.02684593e+00 -1.10443331e-01 -4.50838923e-01
-4.07635391e-01 -1.13828301e+00 7.46653304e-02 -1.24739039e+00
-3.23323309e-01 6.40743300e-02 3.15935254e-01 -1.73587471e-01
5.90073131e-02 7.70691693e-01 -1.06630027e+00 -7.10984886e-01
7.89329410e-01 1.29612279e+00 2.73008585e-01 2.18819305e-01
1.48448244e-01 5.31033218e-01 -6.36878610e-01 3.13472688e-01
-7.42015988e-02 -7.41355598e-01 -3.62929642e-01 -1.82823032e-01
-8.87964308e-01 -8.75602588e-02 -5.46929717e-01 5.20015776e-01
6.37354553e-01 9.05856639e-02 2.20299780e-01 -1.44403708e+00
7.57686570e-02 6.71920955e-01 -5.59982657e-01 3.65741253e-01
-1.35228410e-01 1.07920140e-01 5.90836108e-01 7.68770352e-02
2.07954481e-01 8.26573193e-01 -7.06890970e-02 6.37269676e-01
-7.62547493e-01 -1.01075888e+00 1.89275537e-02 4.63192046e-01
-1.22243285e+00 -9.14117515e-01 8.97797287e-01 4.78844643e-01
6.10061169e-01 5.07009983e-01 1.06150365e+00 3.44208091e-01
6.59557581e-01 1.26130119e-01 5.79529285e-01 -8.08494925e-01
6.62962794e-01 -1.56232491e-01 -3.27645898e-01 9.08018112e-01
6.64053261e-01 3.70704263e-01 2.50444543e-02 1.03124268e-01
1.01260591e+00 -2.06010669e-01 -1.89834628e-02 -3.31922084e-01
-8.96179855e-01 4.08757925e-01 7.96108067e-01 4.58239108e-01
-6.61529541e-01 6.79345012e-01 -1.06099270e-01 9.30706561e-02
-1.66927859e-01 5.74704468e-01 -2.24229112e-01 -1.33285616e-02
-7.05828011e-01 -6.89098477e-01 6.59843802e-01 1.09964752e+00
4.12338346e-01 4.40679163e-01 -4.83040005e-01 5.10692656e-01
6.79491580e-01 6.86637044e-01 9.12000686e-02 -1.31335795e+00
-9.22318324e-02 5.40907443e-01 6.89444840e-02 -5.33801377e-01
-8.95756006e-01 -9.77024496e-01 -1.32142341e+00 -5.49848787e-02
-3.54181707e-01 -6.79139733e-01 -5.52827537e-01 1.71478999e+00
1.76800296e-01 1.78001598e-01 1.84942529e-01 4.98219043e-01
4.24041361e-01 9.44716334e-01 1.02640815e-01 -1.02006900e+00
1.28646219e+00 -9.60440576e-01 -1.01115787e+00 -2.73936808e-01
1.08690970e-01 -2.14694038e-01 3.28296363e-01 -1.27309822e-02
-1.40970242e+00 -3.02604437e-01 -1.56868339e+00 6.60026014e-01
6.26270473e-02 3.09139937e-01 5.08569121e-01 1.05567801e+00
-1.09152353e+00 2.61499226e-01 -9.03353393e-01 -3.82863462e-01
5.51630795e-01 7.01945901e-01 6.85594261e-01 1.75903827e-01
-1.10839176e+00 5.45231521e-01 8.45339522e-02 -1.13071702e-01
-7.13158667e-01 -2.50621825e-01 -3.46037447e-01 4.18479830e-01
6.96473539e-01 -8.09430957e-01 1.15905261e+00 -4.66351211e-01
-2.32291985e+00 7.95317367e-02 -5.41106351e-02 -2.02915028e-01
-3.77246618e-01 6.06468618e-01 -8.41542184e-01 3.14588785e-01
-4.70876783e-01 3.73228639e-01 4.59604889e-01 -1.07635379e+00
-6.30010903e-01 -1.79595068e-01 3.90242226e-02 6.78509250e-02
-2.18399003e-01 -1.47211879e-01 -1.86649814e-01 -2.80416131e-01
2.94903845e-01 -1.04840910e+00 -6.42241359e-01 -4.63928789e-01
-3.42391402e-01 -3.74241710e-01 8.40402186e-01 1.14290252e-01
1.61406720e+00 -1.98515677e+00 2.62906581e-01 6.67902529e-01
1.25583798e-01 2.32374653e-01 2.28246123e-01 -6.75895391e-03
5.68949461e-01 2.21330792e-01 3.68690312e-01 3.58172268e-01
-2.58383274e-01 3.64662737e-01 2.55078405e-01 -8.87025241e-03
-3.06326896e-01 6.27863169e-01 -7.37657011e-01 -4.68911022e-01
2.54165351e-01 1.10126589e-03 -3.65155011e-01 1.52923495e-01
5.01062237e-02 4.83234048e-01 -9.54246879e-01 7.15519309e-01
5.44218123e-01 -3.98434490e-01 7.34512985e-01 -4.97979075e-01
-2.55675584e-01 -4.65580784e-02 -8.86170864e-01 9.53453600e-01
-6.24809384e-01 1.37581885e-01 4.30686623e-01 -7.64465511e-01
9.36545312e-01 2.80414075e-01 7.43242621e-01 -1.07525098e+00
4.06525999e-01 2.26306885e-01 2.83970177e-01 -3.11199218e-01
3.97057757e-02 -2.73604304e-01 -2.27305159e-01 4.66960073e-02
-1.51866719e-01 1.96115285e-01 1.32072061e-01 -3.80316019e-01
1.42741358e+00 -2.22347856e-01 5.25704265e-01 -3.76190811e-01
7.95393765e-01 -4.60112363e-01 1.10331833e+00 4.56266284e-01
-4.64218616e-01 -6.23073578e-01 2.33562067e-01 4.81582209e-02
-4.94886398e-01 -1.27734542e+00 2.50839323e-01 8.36564541e-01
8.43232274e-01 -6.58188434e-03 -5.10337412e-01 -2.84915507e-01
-5.80407739e-01 1.03145945e+00 1.94752574e-01 -6.45863891e-01
-4.91466016e-01 -7.68145561e-01 1.18495621e-01 2.48475939e-01
1.08352375e+00 -9.19235051e-01 -8.45304310e-01 5.99703312e-01
4.15873304e-02 -1.15877080e+00 -3.93983811e-01 4.17072743e-01
-9.12131071e-01 -6.10584378e-01 1.78822070e-01 -9.22942758e-01
4.72048104e-01 2.35280693e-01 9.64993238e-01 9.22144428e-02
1.20681375e-01 7.67035037e-02 -2.95998544e-01 -2.72834837e-01
-2.97384292e-01 4.05161917e-01 6.72519282e-02 -1.62128583e-01
-2.31707349e-01 -8.84535670e-01 -9.68772829e-01 7.67644525e-01
-2.84068406e-01 1.61971897e-01 9.17052865e-01 7.08258525e-02
7.67841041e-01 6.98773265e-01 1.18348515e+00 -2.39542931e-01
2.68808194e-02 -4.40240830e-01 -8.18251729e-01 4.33484644e-01
-7.50808835e-01 1.38222545e-01 8.20808947e-01 2.48866871e-01
-9.86489713e-01 8.85547772e-02 9.09643918e-02 9.43143200e-03
3.79983783e-01 -1.51289478e-01 -8.91096413e-01 -3.09280008e-01
2.22566109e-02 1.20122805e-02 -2.01299995e-01 -1.92273315e-02
-5.22706956e-02 6.00755692e-01 1.29590452e-01 -2.37510443e-01
9.17445064e-01 3.26962382e-01 8.13847601e-01 -7.60514140e-01
-5.85588396e-01 2.23682210e-01 8.75582695e-02 -5.74570715e-01
6.44000769e-01 -7.98727810e-01 -1.01171505e+00 -2.56013751e-01
-4.80646372e-01 -4.98820186e-01 -3.49902511e-02 5.27529120e-01
-5.47217190e-01 -2.62889981e-01 -2.32555032e-01 -8.16169679e-01
-5.87121904e-01 -9.86628413e-01 2.43224457e-01 1.02309024e+00
7.93169513e-02 -6.75812840e-01 -3.41551542e-01 2.74622329e-02
8.57827067e-01 3.96951467e-01 1.14164436e+00 2.43292734e-01
-9.21140194e-01 2.61964113e-01 -1.14980405e-02 -7.28753060e-02
4.18052346e-01 -4.13358510e-02 -3.50694388e-01 -5.44194996e-01
-2.04225048e-01 4.73290414e-01 2.56232649e-01 8.06718171e-01
1.31338096e+00 -4.01722133e-01 -9.49720442e-01 7.40672588e-01
1.64236474e+00 9.47320402e-01 1.03294897e+00 2.73521900e-01
1.71463281e-01 -1.26557648e-01 6.25341058e-01 6.77979648e-01
3.51562440e-01 7.89939642e-01 8.67132962e-01 6.29701242e-02
-9.67058912e-02 4.32309993e-02 5.75992048e-01 6.90956473e-01
-2.02953175e-01 -8.28727782e-01 -2.14240327e-01 -2.43937299e-01
-1.45536041e+00 -9.17593896e-01 2.50998884e-01 2.07288742e+00
4.78616357e-01 3.27372581e-01 1.35012686e-01 1.84335306e-01
8.06520760e-01 1.15410261e-01 -9.20169294e-01 -3.34965110e-01
1.27723232e-01 2.10521936e-01 8.78813565e-01 1.69038743e-01
-5.45747697e-01 6.27632201e-01 5.78396320e+00 8.17133665e-01
-9.94383991e-01 4.46412638e-02 5.37107885e-01 -3.13252658e-01
-3.93023714e-02 -7.76045248e-02 -4.48333830e-01 6.92560852e-01
1.35634077e+00 -3.68664533e-01 7.57103920e-01 5.12728989e-01
7.75736988e-01 -9.40744206e-02 -6.27447188e-01 8.24831188e-01
-5.60964644e-01 -1.38793182e+00 -2.86034435e-01 3.22420865e-01
5.52564502e-01 -2.19080344e-01 -4.62059110e-01 3.29491138e-01
1.99917331e-01 -5.09784818e-01 2.07929119e-01 7.90746152e-01
8.43924105e-01 -1.26989782e+00 7.00311244e-01 2.28903219e-01
-1.37620461e+00 -7.40428090e-01 -2.65378803e-01 1.88385591e-01
2.57626951e-01 7.27231145e-01 -2.59738028e-01 6.76917911e-01
4.49831009e-01 3.80700469e-01 -1.44754604e-01 1.07432675e+00
-3.03854972e-01 6.97868288e-01 -3.69097114e-01 -5.29048443e-01
-1.18814647e-01 -3.81942958e-01 5.64602435e-01 4.19880718e-01
5.48676908e-01 5.91841698e-01 6.87162206e-02 4.15401042e-01
-4.22973454e-01 -3.01731378e-01 4.70859483e-02 1.88143745e-01
1.21084893e+00 1.62583494e+00 -9.45941865e-01 -1.28916681e-01
-3.00016142e-02 6.09379411e-01 -3.04457933e-01 2.13236004e-01
-1.10591865e+00 -8.67386341e-01 6.43829703e-01 2.62461394e-01
1.88044861e-01 -3.17144513e-01 -3.44342530e-01 -3.78191292e-01
-3.53949487e-01 -1.79074481e-01 2.07497463e-01 -6.88543797e-01
-5.50986350e-01 2.88115889e-01 -3.25604320e-01 -1.28327465e+00
2.93778151e-01 -1.90062076e-02 -9.24411774e-01 8.28059912e-02
-1.70780075e+00 -4.41418439e-01 -6.44230723e-01 2.54678965e-01
1.86668053e-01 -3.27354372e-01 6.26920521e-01 3.10044289e-01
-1.17167807e+00 5.33965409e-01 3.50863904e-01 -3.78806263e-01
7.97857940e-02 -7.98973441e-01 -5.33677280e-01 7.03990638e-01
-7.42213607e-01 2.43584048e-02 4.60946679e-01 -1.68245688e-01
-1.96126568e+00 -1.03095460e+00 5.89428321e-02 5.92484057e-01
1.21083580e-01 1.34524524e-01 1.80633441e-02 1.37505770e-01
4.19940591e-01 -8.97973403e-02 7.66220927e-01 -4.56845164e-01
8.25578392e-01 -8.86799574e-01 -1.48495638e+00 9.47203815e-01
1.21568215e+00 4.90935415e-01 2.83597797e-01 9.00992192e-03
7.93822229e-01 -1.39772683e-01 -9.01127756e-01 5.98139822e-01
3.21350962e-01 -6.58086240e-01 8.27403009e-01 -6.38407841e-02
-3.08416784e-01 -5.74260414e-01 -2.54577786e-01 -1.40661716e+00
-9.14172947e-01 -9.30681825e-01 -6.05506659e-01 1.22751582e+00
1.19578764e-01 -4.46561366e-01 9.26863968e-01 -1.05829656e-01
-3.72133553e-01 -8.54278386e-01 -1.36184001e+00 -8.70399833e-01
-5.84528148e-01 2.37008825e-01 8.01638722e-01 1.04762346e-01
2.79566020e-01 7.42858648e-01 3.22760874e-03 6.43207788e-01
6.36090457e-01 -6.53361976e-02 3.04952085e-01 -1.25043535e+00
-1.00404851e-01 -5.87679684e-01 -1.50043428e-01 -9.40268576e-01
8.53421763e-02 -7.79687107e-01 -1.42949536e-01 -1.92449284e+00
-2.64603198e-01 -7.15069771e-01 -6.26394808e-01 4.69316602e-01
3.01073819e-01 -1.10933911e-02 -6.35643974e-02 -2.28774250e-02
-1.01366782e+00 8.16951156e-01 1.43763280e+00 1.56060502e-01
-5.32012463e-01 5.26096106e-01 -8.50282311e-01 5.49410701e-01
1.24581599e+00 -2.43212074e-01 -6.93559229e-01 2.33454958e-01
-6.37371764e-02 6.75205946e-01 -1.59120753e-01 -1.44334328e+00
1.90209836e-01 -5.65417528e-01 3.99294406e-01 -4.09297138e-01
-2.41581202e-02 -1.30612767e+00 1.12230025e-01 1.11139226e+00
4.37515117e-02 -4.88116622e-01 1.59755014e-02 7.30423272e-01
4.94304061e-01 2.21698254e-01 1.12076223e+00 3.28369558e-01
-3.74935657e-01 4.05655205e-01 -1.01712191e+00 -3.12462538e-01
1.46703255e+00 -1.80777073e-01 -2.87878394e-01 -2.83947825e-01
-6.09812975e-01 6.50456250e-01 -7.75581598e-02 2.16805026e-01
4.49258566e-01 -1.28297198e+00 -5.66127710e-02 7.17950463e-02
-3.06399077e-01 -4.64166462e-01 2.74200320e-01 7.88000882e-01
-3.22703302e-01 4.59911227e-01 -3.78533959e-01 -2.08007723e-01
-1.02276492e+00 1.58447623e-01 8.57675314e-01 -4.14191246e-01
-2.83353887e-02 3.14248413e-01 -6.02404118e-01 2.40802124e-01
-6.74788840e-03 -3.13359313e-02 -3.85921031e-01 -3.91135752e-01
3.05581670e-02 1.06172359e+00 -1.50926128e-01 6.70083910e-02
-6.28602684e-01 4.73021775e-01 4.30422455e-01 2.74573326e-01
1.34467804e+00 -6.63387775e-01 -2.26110011e-01 -1.20754674e-01
6.20138943e-01 -7.61213189e-04 -1.03367114e+00 2.52368078e-02
1.88324656e-02 -9.94962081e-02 8.03758800e-01 -8.61041188e-01
-1.50242567e+00 -3.83427255e-02 1.03033900e+00 4.27911639e-01
1.81462741e+00 -1.52285984e-02 8.59469354e-01 4.20071214e-01
8.99127901e-01 -1.26151741e+00 3.15884322e-01 1.58823237e-01
2.46128052e-01 -5.86964428e-01 1.58115238e-01 -5.82030416e-01
-5.55749750e-03 1.36638844e+00 9.87989247e-01 1.93284467e-01
8.73694837e-01 2.88064688e-01 -5.92032373e-01 -1.77118480e-01
-7.30270565e-01 -2.60429442e-01 -4.99961436e-01 4.50505584e-01
3.56395654e-02 3.38523239e-01 -3.20160955e-01 6.32310987e-01
2.94907894e-02 1.28925875e-01 6.56215906e-01 9.73901510e-01
-1.07561123e+00 -1.28968287e+00 -1.11352995e-01 6.85313761e-01
-4.30873819e-02 5.03157914e-01 3.20937544e-01 4.02568400e-01
2.76775450e-01 1.43142796e+00 1.76284552e-01 -5.85514545e-01
3.50363970e-01 -4.36793774e-01 5.61418235e-01 -4.18779343e-01
9.20345634e-02 1.08381093e-01 2.32309982e-01 -5.65910220e-01
-4.26529735e-01 -2.78819114e-01 -1.80528331e+00 -3.13945025e-01
-3.41335237e-01 3.94038975e-01 6.73632383e-01 1.00880742e+00
8.51126611e-01 1.30066013e+00 1.65135312e+00 -5.29401600e-01
-1.48329437e-02 -4.87255126e-01 -7.57066607e-01 -9.42175627e-01
2.74353832e-01 -7.62246013e-01 -4.00756687e-01 -5.47552645e-01] | [5.943614482879639, 1.593382477760315] |
b0ed9f71-77df-48bb-a99c-b3fd545430af | lif-seg-lidar-and-camera-image-fusion-for-3d | 2108.07511 | null | https://arxiv.org/abs/2108.07511v1 | https://arxiv.org/pdf/2108.07511v1.pdf | LIF-Seg: LiDAR and Camera Image Fusion for 3D LiDAR Semantic Segmentation | Camera and 3D LiDAR sensors have become indispensable devices in modern autonomous driving vehicles, where the camera provides the fine-grained texture, color information in 2D space and LiDAR captures more precise and farther-away distance measurements of the surrounding environments. The complementary information from these two sensors makes the two-modality fusion be a desired option. However, two major issues of the fusion between camera and LiDAR hinder its performance, \ie, how to effectively fuse these two modalities and how to precisely align them (suffering from the weak spatiotemporal synchronization problem). In this paper, we propose a coarse-to-fine LiDAR and camera fusion-based network (termed as LIF-Seg) for LiDAR segmentation. For the first issue, unlike these previous works fusing the point cloud and image information in a one-to-one manner, the proposed method fully utilizes the contextual information of images and introduces a simple but effective early-fusion strategy. Second, due to the weak spatiotemporal synchronization problem, an offset rectification approach is designed to align these two-modality features. The cooperation of these two components leads to the success of the effective camera-LiDAR fusion. Experimental results on the nuScenes dataset show the superiority of the proposed LIF-Seg over existing methods with a large margin. Ablation studies and analyses demonstrate that our proposed LIF-Seg can effectively tackle the weak spatiotemporal synchronization problem. | ['Wenbing Tao', 'Hongsheng Li', 'Xiao Song', 'Xinge Zhu', 'Hui Zhou', 'Lin Zhao'] | 2021-08-17 | null | null | null | null | ['lidar-semantic-segmentation'] | ['computer-vision'] | [ 5.84793501e-02 -6.09643877e-01 -1.44663170e-01 -2.93466181e-01
-7.45543242e-01 -3.83836716e-01 5.66354036e-01 -5.82475662e-02
-5.79604149e-01 6.61046445e-01 -5.13644814e-01 -8.61017704e-02
-2.31895640e-01 -8.57723594e-01 -4.97362673e-01 -8.65446687e-01
4.45606560e-01 1.93313658e-01 6.79900944e-01 -2.75482178e-01
2.04673216e-01 6.40007794e-01 -2.03892946e+00 -3.15321892e-01
1.37002301e+00 1.18554282e+00 5.69148660e-01 1.85869366e-01
-4.23614919e-01 1.31501943e-01 2.62474213e-02 -1.52853280e-01
3.41306120e-01 -4.50554155e-02 3.65448073e-02 2.27028187e-02
5.01498699e-01 -2.03055188e-01 -1.83176711e-01 1.13234854e+00
4.02762324e-01 2.40110476e-02 3.77511412e-01 -1.41069984e+00
-7.08434060e-02 -3.29241790e-02 -1.14335370e+00 2.33550936e-01
4.14381027e-02 8.29345509e-02 4.54296768e-01 -1.00747609e+00
3.35049659e-01 1.03652596e+00 7.22722948e-01 5.72388396e-02
-8.51298928e-01 -9.70015466e-01 1.81076020e-01 1.72792539e-01
-1.85118985e+00 -4.44738239e-01 9.95410860e-01 -3.99402946e-01
4.01191592e-01 6.66149929e-02 5.55125892e-01 5.75703800e-01
1.59082800e-01 3.94196928e-01 1.35362136e+00 -2.15539843e-01
-7.68387318e-02 3.27748507e-01 7.91408941e-02 5.39070487e-01
5.62037706e-01 3.47779304e-01 -5.72154224e-01 2.09249437e-01
5.80840945e-01 4.26154673e-01 -1.64132208e-01 -3.49319696e-01
-1.15204751e+00 4.42118257e-01 4.68085587e-01 4.32750553e-01
-2.50830948e-01 -6.08642697e-02 8.10133442e-02 -1.33024663e-01
3.56857806e-01 -2.10104868e-01 7.88356364e-02 -8.86071846e-02
-9.60407853e-01 7.88255930e-02 1.16085894e-01 1.11201656e+00
1.14920521e+00 1.01193838e-01 3.07571530e-01 5.33405721e-01
3.12173665e-01 9.80692089e-01 1.78727120e-01 -8.37790549e-01
6.48493767e-01 6.96801126e-01 1.25998899e-01 -1.16308773e+00
-3.61904949e-01 -2.27603927e-01 -9.03880179e-01 2.67306775e-01
1.91462174e-01 4.45889980e-02 -7.67269611e-01 1.55937791e+00
4.99430954e-01 5.76451480e-01 3.08665074e-03 9.92200971e-01
8.16969812e-01 5.44765770e-01 2.35543698e-02 -3.72303993e-01
1.14045894e+00 -4.91317958e-01 -8.89477253e-01 -2.30436906e-01
2.30364010e-01 -8.38954091e-01 6.17976546e-01 1.27564415e-01
-8.38047981e-01 -8.69200110e-01 -1.30786586e+00 5.62205650e-02
-4.97413248e-01 1.71781704e-01 3.53298455e-01 4.81082022e-01
-8.63668025e-01 4.55058478e-02 -6.58317327e-01 -4.40915316e-01
1.96770042e-01 4.07616436e-01 -4.33285415e-01 -2.64502674e-01
-1.06784999e+00 8.04961979e-01 4.52778757e-01 4.60388869e-01
-2.29856327e-01 -4.73035991e-01 -6.60989285e-01 -2.78616846e-01
5.41258097e-01 -4.73556697e-01 7.65295267e-01 -3.81909996e-01
-1.12904453e+00 4.69872415e-01 -5.27855814e-01 -2.08001837e-01
5.36080301e-01 -1.43051878e-01 -5.32415807e-01 2.33289778e-01
3.67627233e-01 7.04299688e-01 6.95743740e-01 -1.58215022e+00
-1.27883983e+00 -5.67675233e-01 -1.28112182e-01 5.01827538e-01
-2.99743623e-01 -4.36202884e-01 -7.68260598e-01 -2.15538397e-01
5.51174402e-01 -9.32235956e-01 -1.70736149e-01 2.68526524e-02
-4.46948456e-04 -1.63005278e-01 1.28959405e+00 -8.23912174e-02
1.12978768e+00 -2.33999062e+00 -1.46015346e-01 1.73491552e-01
1.70826986e-01 3.63927275e-01 1.51411891e-01 2.85982013e-01
1.63740411e-01 -4.49927114e-02 -3.14499646e-01 -4.37676072e-01
-3.86873066e-01 3.67527723e-01 -1.65434822e-01 6.09589636e-01
1.64905161e-01 6.32578254e-01 -8.29856455e-01 -9.84276056e-01
7.33752966e-01 6.13726318e-01 -7.59570897e-02 5.55752516e-02
1.85936078e-01 7.29100466e-01 -8.04499090e-01 7.02268898e-01
1.13768542e+00 3.80549729e-01 -2.74829417e-01 -4.61592942e-01
-6.81078911e-01 -4.04150635e-01 -1.42258132e+00 1.71714401e+00
-3.09865981e-01 3.46461445e-01 2.59631664e-01 -8.63093615e-01
1.18713081e+00 2.76210785e-01 6.55251443e-01 -1.04288304e+00
1.28194824e-01 5.67529202e-01 -3.83274734e-01 -5.05488276e-01
5.77977061e-01 -2.50312865e-01 -2.26000503e-01 -8.13733414e-02
-2.82463402e-01 -4.14515078e-01 1.56563103e-01 8.13049451e-02
4.95905340e-01 2.89445877e-01 1.24646820e-01 -1.46379337e-01
8.62106860e-01 1.11428693e-01 9.62293863e-01 5.49579442e-01
-3.59575421e-01 6.38992786e-01 -2.45953828e-01 -9.53059271e-02
-6.61846519e-01 -8.84826124e-01 -2.67780215e-01 2.48851553e-01
1.02357817e+00 -1.46858349e-01 -2.90503561e-01 -2.76290894e-01
9.95886475e-02 4.05665308e-01 -1.98358074e-01 -4.77805287e-02
-6.96095586e-01 -3.70669246e-01 4.13061708e-01 6.04699373e-01
1.08249497e+00 -3.13194841e-01 -7.08557010e-01 2.04409555e-01
-2.90382564e-01 -1.51397908e+00 -2.95696229e-01 4.98657906e-03
-9.51653063e-01 -8.65656435e-01 -2.99344093e-01 -5.32131553e-01
3.73692781e-01 1.15682447e+00 4.41827208e-01 1.56771079e-01
-1.73001532e-02 2.34499395e-01 -1.72747076e-01 -3.50849867e-01
1.29773557e-01 -1.62879303e-01 2.12667018e-01 2.55082756e-01
4.13595021e-01 -8.01697969e-01 -4.89419401e-01 4.46979940e-01
-7.66179085e-01 3.06709018e-02 6.80124283e-01 4.94662076e-01
7.23885000e-01 3.86827499e-01 4.88764286e-01 -4.29009467e-01
2.56547242e-01 -3.26112449e-01 -8.76407683e-01 1.42385393e-01
-6.47643805e-01 -2.35748187e-01 5.18152297e-01 -1.17094457e-01
-1.31070292e+00 3.96063179e-01 6.82341233e-02 -7.27407932e-01
-3.13826174e-01 4.68766600e-01 -4.54818368e-01 -2.51189530e-01
1.43431783e-01 2.45666280e-01 -3.16304192e-02 -4.48803395e-01
4.45280403e-01 7.52567828e-01 9.33676660e-01 -5.25639892e-01
1.14090407e+00 8.93081784e-01 2.33154401e-01 -9.79563415e-01
-3.66330475e-01 -7.84117699e-01 -9.42742944e-01 -4.54120845e-01
9.37213957e-01 -1.15452182e+00 -8.25470805e-01 5.02883792e-01
-1.09417796e+00 5.04107594e-01 -1.23563364e-01 7.71975338e-01
-2.50934809e-01 5.58217049e-01 -1.24965616e-01 -1.15221465e+00
-1.45855465e-03 -1.36069643e+00 1.22174227e+00 6.65159702e-01
3.82536948e-01 -6.07197344e-01 -1.10655218e-01 4.82519716e-01
1.97700977e-01 3.80842030e-01 3.34227651e-01 -2.89843660e-02
-9.64581132e-01 -1.46129444e-01 -6.10187292e-01 1.93019826e-02
1.50267690e-01 1.96935549e-01 -1.12892628e+00 -1.00516520e-01
1.21366002e-01 1.19359940e-01 8.41666877e-01 1.44990236e-01
6.61605358e-01 4.11015242e-01 -6.20015323e-01 6.47961497e-01
1.64970267e+00 5.00228405e-01 3.09028268e-01 1.72666907e-01
8.85878801e-01 5.01398265e-01 1.12815726e+00 2.12290332e-01
7.14356184e-01 7.75726736e-01 5.63617349e-01 -3.03296566e-01
-1.13278262e-01 -1.75566241e-01 2.33501270e-01 1.08729696e+00
-2.58828163e-01 -3.39439102e-02 -7.88833678e-01 6.38539255e-01
-2.03892803e+00 -8.05085421e-01 -4.94034916e-01 2.25960612e+00
3.82006913e-01 2.82701224e-01 -1.78040430e-01 1.35603890e-01
9.14545715e-01 1.79500431e-01 -4.88154203e-01 -9.89216790e-02
-3.18589032e-01 -8.89486596e-02 8.13036144e-01 3.85469377e-01
-1.02124643e+00 7.40935266e-01 5.08390141e+00 1.10773087e+00
-1.30712640e+00 1.81239650e-01 -6.29100055e-02 9.47762653e-02
-2.95326352e-01 2.07792014e-01 -1.04183626e+00 5.79791665e-01
5.81948340e-01 1.35757355e-02 1.36515900e-01 4.68284607e-01
3.77288610e-01 -6.09550714e-01 -6.81856930e-01 1.14552784e+00
9.54266451e-03 -1.18859720e+00 -1.81267515e-01 1.81952581e-01
5.33631027e-01 1.38044031e-02 -6.62743673e-02 -3.91427465e-02
-1.20185040e-01 -5.65075874e-01 9.44838464e-01 6.43362403e-01
8.68219793e-01 -7.50793993e-01 7.16740012e-01 6.81226671e-01
-1.94691980e+00 -1.03189692e-01 -1.87178999e-01 3.89247090e-02
3.96181196e-01 7.37639248e-01 -4.32411253e-01 1.28968525e+00
8.65016162e-01 8.68480325e-01 -5.79926729e-01 9.59611535e-01
9.43982601e-02 8.56249556e-02 -6.40889823e-01 3.47677469e-01
2.80760616e-01 -7.16319442e-01 7.05098033e-01 8.54239523e-01
6.79992139e-01 3.58864307e-01 4.69767302e-01 8.68541241e-01
2.97256023e-01 -2.13068455e-01 -1.00267494e+00 2.73047328e-01
8.29978645e-01 1.38069785e+00 -7.20471859e-01 -2.30110392e-01
-6.35686159e-01 3.63329709e-01 5.59777841e-02 3.01498801e-01
-8.23864937e-01 -3.28461468e-01 4.87641811e-01 2.06048429e-01
3.33567560e-01 -6.62858009e-01 -5.23323178e-01 -9.50835049e-01
2.61731803e-01 -3.00131977e-01 8.36861804e-02 -8.26599479e-01
-9.64448988e-01 4.69186753e-01 3.23973566e-01 -1.78239751e+00
8.71454179e-02 -1.06022522e-01 -5.48605084e-01 9.39286709e-01
-2.06406736e+00 -1.48423743e+00 -6.61959052e-01 8.71719539e-01
4.02064443e-01 1.70908585e-01 1.59524530e-01 5.70647180e-01
-6.79556251e-01 7.53707066e-02 3.38399112e-02 -3.53491724e-01
6.57087564e-01 -8.40984643e-01 -2.10612729e-01 1.22305608e+00
-1.98647201e-01 4.85327333e-01 3.96688372e-01 -7.69063532e-01
-1.43838215e+00 -9.88573611e-01 7.56702900e-01 -1.46404743e-01
3.25610787e-01 -2.09459633e-01 -9.09871221e-01 2.56224632e-01
2.37590462e-01 6.16473258e-02 5.36203831e-02 -3.33482265e-01
8.68973359e-02 -8.04030716e-01 -9.72222507e-01 4.19715762e-01
9.85212028e-01 -3.93857777e-01 -6.03160501e-01 -9.75074917e-02
6.23079717e-01 -3.38138610e-01 -7.98482597e-01 8.90362382e-01
5.00944376e-01 -1.15555418e+00 8.65536511e-01 4.09488946e-01
-1.25093013e-01 -1.00408030e+00 -2.91753709e-01 -8.86719108e-01
3.30573656e-02 -2.88279980e-01 3.52406591e-01 1.58384526e+00
-2.62228418e-02 -9.67982888e-01 4.10451591e-01 3.31751287e-01
-2.38750786e-01 -4.29829746e-01 -1.29065979e+00 -7.16009557e-01
-3.06477487e-01 -5.49306035e-01 6.97338820e-01 8.00801814e-01
-3.61119598e-01 2.96188354e-01 -2.84027427e-01 4.81935263e-01
7.88451970e-01 3.91280115e-01 8.25636685e-01 -1.56644177e+00
4.75624263e-01 -2.94803947e-01 -3.66750151e-01 -1.10284019e+00
-5.34514189e-02 -4.99957770e-01 3.82211328e-01 -1.40404320e+00
-4.39011445e-03 -8.93985868e-01 -2.42159918e-01 8.27099606e-02
-1.79383487e-01 3.32091689e-01 2.05782175e-01 4.87017244e-01
-4.04750884e-01 6.70104086e-01 1.19262469e+00 3.01526450e-02
-1.33055523e-01 -1.61885843e-01 -4.30225939e-01 7.00382590e-01
4.83734697e-01 -3.64964962e-01 -5.91634572e-01 -5.36951363e-01
-3.48605886e-02 2.48535961e-01 3.36635590e-01 -1.26979828e+00
7.48219490e-01 -2.86589682e-01 1.28159016e-01 -1.29758406e+00
6.73897028e-01 -1.35199869e+00 2.95539409e-01 1.04718789e-01
5.06777108e-01 1.73257038e-01 2.39623547e-01 8.36841106e-01
-5.85431218e-01 1.16752006e-01 7.82139421e-01 7.82023091e-03
-1.06531048e+00 4.26735193e-01 -1.58282563e-01 -3.27800244e-01
1.31434143e+00 -8.33446801e-01 -3.67539346e-01 -2.13931754e-01
-2.70503014e-01 5.05495727e-01 5.45692205e-01 5.25468707e-01
7.09562302e-01 -1.41318774e+00 -4.18210506e-01 5.26421726e-01
2.74042606e-01 3.38108331e-01 4.57584262e-01 1.32446897e+00
-3.01908582e-01 5.14855206e-01 -2.07352251e-01 -1.16781664e+00
-1.11520600e+00 4.55395192e-01 2.40103319e-01 3.25939991e-02
-5.07031918e-01 4.57070082e-01 2.65395701e-01 -3.20547134e-01
-9.65934023e-02 -3.30213130e-01 -3.38876873e-01 1.31296694e-01
2.40057092e-02 3.86520624e-01 1.29689872e-01 -1.18208218e+00
-6.62533879e-01 1.37105262e+00 2.65198648e-01 -1.57185316e-01
1.02755749e+00 -7.88694501e-01 -5.67224883e-02 5.55628121e-01
1.03340244e+00 1.56508669e-01 -1.25517237e+00 -3.49347502e-01
-2.57317990e-01 -5.83472610e-01 2.32039914e-01 -8.69473442e-02
-1.28817475e+00 1.19854963e+00 7.96190202e-01 1.06333345e-01
1.25598466e+00 -3.92024040e-01 9.85530674e-01 -6.22622259e-02
5.53351223e-01 -1.05478859e+00 -3.58060300e-01 2.38181621e-01
3.24944973e-01 -1.31878614e+00 1.81054130e-01 -7.67978728e-01
-4.96224999e-01 1.05015051e+00 8.30599666e-01 1.64306581e-01
6.51474893e-01 3.15168232e-01 4.01041284e-02 -1.51405573e-01
-3.55694145e-01 -5.23190677e-01 1.92285001e-01 5.79389393e-01
-1.27090618e-01 -1.88424811e-01 -3.22533876e-01 3.40399355e-01
1.52083859e-01 1.39261454e-01 2.63816059e-01 1.15208101e+00
-5.29469490e-01 -1.00124991e+00 -6.35705173e-01 9.37025547e-02
1.74148813e-01 2.41163164e-01 4.82931994e-02 1.07190216e+00
7.01653183e-01 1.12439942e+00 1.25780180e-01 -5.81981063e-01
5.35569012e-01 -2.52302647e-01 2.03749776e-01 -1.44950852e-01
-1.56885773e-01 3.23805660e-01 -1.17066145e-01 -6.11137092e-01
-8.68453145e-01 -6.95700586e-01 -1.29023898e+00 -2.80109853e-01
-6.11602783e-01 1.72896430e-01 7.60380268e-01 1.08483708e+00
4.88084614e-01 2.63278902e-01 8.49212527e-01 -1.01682091e+00
-9.53008831e-02 -4.83401686e-01 -6.76577806e-01 9.16519985e-02
3.75822246e-01 -1.10056186e+00 -3.60000104e-01 -1.10255152e-01] | [8.210474967956543, -2.5262129306793213] |
33ebd1f9-f16f-4ca5-935e-2def70b399da | a-visual-attention-grounding-neural-model-for | 1808.08266 | null | http://arxiv.org/abs/1808.08266v2 | http://arxiv.org/pdf/1808.08266v2.pdf | A Visual Attention Grounding Neural Model for Multimodal Machine Translation | We introduce a novel multimodal machine translation model that utilizes
parallel visual and textual information. Our model jointly optimizes the
learning of a shared visual-language embedding and a translator. The model
leverages a visual attention grounding mechanism that links the visual
semantics with the corresponding textual semantics. Our approach achieves
competitive state-of-the-art results on the Multi30K and the Ambiguous COCO
datasets. We also collected a new multilingual multimodal product description
dataset to simulate a real-world international online shopping scenario. On
this dataset, our visual attention grounding model outperforms other methods by
a large margin. | ['Mingyang Zhou', 'Yong Jae Lee', 'Runxiang Cheng', 'Zhou Yu'] | 2018-08-24 | a-visual-attention-grounding-neural-model-for-1 | https://aclanthology.org/D18-1400 | https://aclanthology.org/D18-1400.pdf | emnlp-2018-10 | ['multimodal-machine-translation'] | ['natural-language-processing'] | [-1.04367450e-01 -2.41488799e-01 -6.82138562e-01 -5.07919848e-01
-1.02397525e+00 -8.78912747e-01 8.22742283e-01 1.56896383e-01
-2.37374410e-01 1.07900605e-01 4.44086164e-01 -4.13555890e-01
5.65608561e-01 -2.41460174e-01 -1.01554775e+00 -2.60658503e-01
3.14496726e-01 5.82929373e-01 -3.40049803e-01 -4.09445643e-01
-9.20725167e-02 -5.73757812e-02 -1.00891423e+00 9.52036321e-01
7.12207615e-01 8.01558673e-01 3.89245927e-01 5.45430779e-01
-1.73395544e-01 7.99563169e-01 -1.29318267e-01 -1.07158232e+00
2.09409118e-01 -2.42198512e-01 -6.03183746e-01 2.10668966e-01
8.52026045e-01 -5.69334738e-02 -3.87965977e-01 9.63605046e-01
4.00384903e-01 -4.85274345e-02 5.89699149e-01 -1.62634528e+00
-1.72306454e+00 8.01943183e-01 -8.12146425e-01 -1.88208908e-01
5.66168308e-01 3.84967446e-01 1.52753735e+00 -1.33701468e+00
1.14797318e+00 1.42883921e+00 3.68394941e-01 1.00006670e-01
-1.63546145e+00 -4.58088040e-01 3.82154763e-01 3.33688200e-01
-1.41335309e+00 -3.21379527e-02 8.41900945e-01 -5.39246321e-01
1.05222654e+00 -8.98213759e-02 6.83627129e-01 1.44272733e+00
3.40938687e-01 1.10929525e+00 9.09798801e-01 -3.65138710e-01
-2.40005091e-01 2.96061039e-01 -1.23913540e-02 1.01672101e+00
-2.22919211e-01 -7.37862661e-02 -9.38650072e-01 4.50762808e-02
4.83989358e-01 -4.07855632e-03 -4.99769039e-02 -1.00180793e+00
-1.59385359e+00 1.06946528e+00 7.39291251e-01 -7.65589252e-02
-2.09888339e-01 3.52484465e-01 5.17523766e-01 4.51618493e-01
2.58428007e-01 4.15765107e-01 -3.77949685e-01 1.02630109e-01
-4.42437112e-01 1.80055071e-02 6.00158393e-01 1.44754744e+00
7.15406597e-01 -2.53809988e-01 -2.88043439e-01 7.99675107e-01
7.29718804e-01 1.05215251e+00 2.64954150e-01 -6.93952203e-01
1.07578015e+00 8.64391148e-01 8.78092423e-02 -8.94693971e-01
-1.55645281e-01 -1.70839384e-01 -2.36819476e-01 -1.12729147e-01
1.16348401e-01 2.01961607e-01 -9.87871110e-01 1.57249844e+00
7.88214281e-02 -3.37371558e-01 8.97852182e-02 1.11022270e+00
9.41877186e-01 7.59931862e-01 2.51903564e-01 3.11018258e-01
1.46399796e+00 -1.58202648e+00 -9.40120578e-01 -3.60191524e-01
8.18859875e-01 -1.13189459e+00 1.70310700e+00 -8.37552696e-02
-8.29442263e-01 -3.26686561e-01 -1.30008864e+00 -6.64673090e-01
-6.14430428e-01 4.34895098e-01 6.28739536e-01 1.40338570e-01
-7.15823054e-01 -4.04233709e-02 -6.04304671e-01 -5.68614244e-01
2.44817138e-01 -7.03103319e-02 -6.31638110e-01 -3.94092590e-01
-9.11797047e-01 1.00879562e+00 6.36486858e-02 5.27243949e-02
-9.65670764e-01 -5.69212258e-01 -1.43629205e+00 -2.24282816e-01
2.70429492e-01 -8.86349440e-01 1.27574193e+00 -1.28321242e+00
-1.38311458e+00 1.09664512e+00 -1.37526527e-01 -1.20926961e-01
3.55687648e-01 -3.06907386e-01 -4.61939156e-01 8.26791674e-02
3.58094186e-01 9.30778027e-01 7.46533275e-01 -1.66612911e+00
-2.51915574e-01 -4.29392874e-01 8.04525763e-02 4.13745582e-01
-1.68052077e-01 4.32905592e-02 -9.46968794e-01 -8.55153739e-01
-2.88901150e-01 -1.20582104e+00 6.30993247e-02 1.48108304e-01
-5.28728545e-01 9.92094427e-02 6.65322423e-01 -7.89513886e-01
8.25170100e-01 -2.27875948e+00 7.10572898e-01 2.36884728e-02
1.34432241e-01 -3.70610356e-01 -8.21861029e-01 6.92048490e-01
5.27799390e-02 1.73999239e-02 7.43441135e-02 -8.10675025e-01
5.48859239e-01 2.04385027e-01 -4.25800323e-01 3.81569594e-01
1.04635932e-01 1.70095336e+00 -8.21118951e-01 -5.55773675e-01
8.31815675e-02 5.58641613e-01 -3.87084395e-01 2.77640194e-01
-4.26859438e-01 1.36848733e-01 -1.16864510e-01 9.79209065e-01
4.10701752e-01 -6.73683465e-01 5.44308722e-01 -6.38092399e-01
1.47101000e-01 -1.27567723e-01 -5.96222520e-01 2.42695546e+00
-5.37488818e-01 6.59947872e-01 -1.35758638e-01 -3.42147261e-01
6.37368023e-01 5.64017333e-02 7.06888437e-02 -1.16635966e+00
5.21768406e-02 6.16192892e-02 -3.25476795e-01 -3.85864526e-01
8.76183271e-01 1.48440093e-01 -5.55497706e-01 4.88106340e-01
3.26779962e-01 5.17656244e-02 -1.04070328e-01 6.86467171e-01
4.67235118e-01 5.92860043e-01 5.24481349e-02 -1.16316587e-01
-2.89147794e-02 2.24085733e-01 2.56526798e-01 4.48127657e-01
-9.85620841e-02 5.38494647e-01 6.58122957e-01 -5.08416653e-01
-1.19274127e+00 -1.16923535e+00 3.62088799e-01 1.60879445e+00
5.88286340e-01 -7.64824867e-01 -4.04261351e-01 -9.45971906e-01
2.46124461e-01 6.87032878e-01 -9.89660919e-01 -1.71353773e-03
-3.16142380e-01 -1.98254630e-01 3.64425659e-01 6.71029985e-01
6.91308454e-02 -9.42257047e-01 -2.21283600e-01 -2.23224014e-01
-3.55205953e-01 -1.59435630e+00 -1.07894611e+00 -3.54338065e-02
-4.12864864e-01 -8.36094439e-01 -7.10932076e-01 -1.14022374e+00
5.51264882e-01 3.78326446e-01 1.37698269e+00 -3.23903620e-01
-1.70808494e-01 5.18252850e-01 -3.43232065e-01 -1.76723599e-01
-4.19202656e-01 1.95963636e-01 -3.41135204e-01 1.42384559e-01
2.63029277e-01 1.66819751e-01 -4.02155131e-01 3.42985064e-01
-7.33323395e-01 6.03632927e-01 4.42217976e-01 9.02049720e-01
8.14088166e-01 -1.07287502e+00 2.23812297e-01 -6.10176980e-01
5.57338774e-01 -4.26051468e-01 -6.04637861e-01 8.31617951e-01
-5.29466987e-01 2.40356386e-01 1.92252725e-01 -6.15007281e-01
-8.54296267e-01 2.43980378e-01 5.41731179e-01 -7.89798260e-01
1.91145465e-01 7.28939235e-01 -2.45657653e-01 1.56890854e-01
2.52674013e-01 -2.18185619e-01 -1.42369300e-01 -5.76670885e-01
1.21908498e+00 3.94835234e-01 4.99619454e-01 -4.21765834e-01
7.76035368e-01 3.90298694e-01 -2.67527282e-01 -3.82814765e-01
-5.26634157e-01 -2.61599422e-01 -7.08267808e-01 -9.07475352e-02
1.19971108e+00 -1.23618865e+00 -6.38030767e-01 1.96453810e-01
-1.29301405e+00 -4.48048443e-01 -3.80688831e-02 4.15969610e-01
-6.64029777e-01 3.09638917e-01 -8.59052956e-01 -7.97376856e-02
-2.41222277e-01 -1.39108276e+00 1.49462414e+00 -1.58451617e-01
-1.38219409e-02 -9.40120757e-01 1.46795988e-01 6.44445539e-01
1.61829993e-01 1.19031370e-01 1.15541399e+00 -3.65835965e-01
-7.00119138e-01 8.09113607e-02 -6.42255306e-01 -2.10590646e-01
-1.42060310e-01 -4.66163084e-02 -7.19291806e-01 -3.37569207e-01
-7.95336366e-01 -8.34123731e-01 8.87600362e-01 -7.35348091e-02
5.76135397e-01 -7.77252368e-04 -2.66279280e-01 6.97620273e-01
1.49023438e+00 -1.08510561e-01 3.86806339e-01 4.24039543e-01
1.28258395e+00 5.59173346e-01 6.19357347e-01 1.78651422e-01
9.71220672e-01 1.04936814e+00 7.11528063e-01 -6.68515086e-01
-2.37928137e-01 -8.27036977e-01 4.62978274e-01 9.99662042e-01
3.89394045e-01 -3.26350600e-01 -9.09520686e-01 6.86238229e-01
-2.33548021e+00 -5.67972422e-01 2.09198713e-01 1.78379643e+00
6.18304372e-01 -3.89056653e-01 9.13121253e-02 -8.89231503e-01
6.24290466e-01 2.05425456e-01 -4.37935859e-01 -5.59424460e-01
-5.24942458e-01 -4.00190651e-01 5.75438917e-01 6.35126710e-01
-1.13999844e+00 1.28226352e+00 6.67579794e+00 5.27527094e-01
-1.03665698e+00 4.85432923e-01 2.50394583e-01 -2.94795841e-01
-8.87291431e-01 -1.10846519e-01 -3.27309906e-01 1.18833818e-01
6.57204568e-01 -5.20295799e-02 6.51195109e-01 5.13377190e-01
-1.43024474e-01 1.97055399e-01 -1.29733431e+00 1.23947120e+00
6.61728799e-01 -1.49832618e+00 5.07137775e-01 2.98140973e-01
8.50619555e-01 2.62688130e-01 4.18810785e-01 3.41909140e-01
5.75305402e-01 -1.22076464e+00 1.22592115e+00 2.79337674e-01
9.76126909e-01 -7.27901280e-01 5.80054045e-01 -3.42236072e-01
-1.28414083e+00 4.98503745e-02 -1.28427995e-02 4.16430235e-01
6.07494116e-01 -3.29233795e-01 -4.46548104e-01 6.52033389e-01
5.84654689e-01 1.14870846e+00 -8.10106456e-01 4.73376602e-01
-4.10354793e-01 4.10666689e-02 2.51654118e-01 2.30027661e-02
5.70717454e-01 -2.90153801e-01 3.45141828e-01 1.37017024e+00
3.39323431e-02 -5.52354574e-01 4.44301069e-01 9.11281407e-01
-5.08196175e-01 5.35497308e-01 -9.26067889e-01 -5.02667606e-01
1.00853689e-01 1.34626245e+00 -4.61081028e-01 -2.59404063e-01
-9.70941007e-01 1.45194352e+00 6.40755177e-01 7.50196517e-01
-1.11854124e+00 -1.34243518e-01 7.65490413e-01 -2.44132057e-01
6.87936604e-01 -2.70396322e-01 -2.50488937e-01 -1.54090881e+00
1.31874261e-02 -1.03526878e+00 4.06773269e-01 -1.23539996e+00
-1.40028787e+00 6.67421818e-01 -1.37475044e-01 -1.17102921e+00
2.05258075e-02 -8.60576808e-01 3.38073410e-02 8.05853248e-01
-1.46752083e+00 -2.12645030e+00 -4.72890399e-02 6.64842784e-01
5.64562678e-01 -3.40778410e-01 9.03493404e-01 3.10310870e-01
-4.97166336e-01 8.80526721e-01 -5.66492528e-02 2.80945003e-01
1.11656725e+00 -1.11443722e+00 4.47652578e-01 5.64388216e-01
5.23266792e-01 7.08694100e-01 5.56158423e-01 -6.26273692e-01
-2.01098514e+00 -9.82077956e-01 1.00733972e+00 -7.87257969e-01
1.11370313e+00 -7.43113220e-01 -4.41720545e-01 1.27942097e+00
1.19208503e+00 -5.02638035e-02 8.65780473e-01 2.54548281e-01
-1.09513414e+00 1.23177342e-01 -6.24815881e-01 8.02328289e-01
8.95374119e-01 -1.07352531e+00 -5.29385746e-01 3.59772414e-01
1.06287479e+00 -5.70319355e-01 -9.58025813e-01 1.51865825e-01
7.74937093e-01 -2.39195973e-01 8.01719368e-01 -9.15697455e-01
8.17643583e-01 -4.06485230e-01 -6.10202014e-01 -1.49930584e+00
-2.59262294e-01 -4.52220380e-01 7.10963830e-03 9.55198348e-01
1.06373358e+00 -2.13947102e-01 2.51723707e-01 3.29055935e-01
3.72344162e-03 -5.46312273e-01 -7.51486361e-01 -6.84247732e-01
-1.01006836e-01 -2.82090098e-01 5.96077323e-01 1.23494196e+00
2.86061198e-01 1.00181794e+00 -5.98502815e-01 -2.66176313e-02
4.85394835e-01 6.14364743e-01 7.92211056e-01 -5.55688083e-01
-2.90320724e-01 -2.74922043e-01 -3.75167489e-01 -1.06904805e+00
3.85815978e-01 -1.27825201e+00 -2.40948141e-01 -1.64601254e+00
8.33727360e-01 2.13062733e-01 -3.42866898e-01 6.16640687e-01
1.24679454e-01 6.79870367e-01 7.09924221e-01 1.12451166e-01
-1.19521046e+00 7.34453142e-01 1.37122250e+00 -6.88147128e-01
2.01855577e-03 -1.11723149e+00 -6.45055532e-01 2.78383017e-01
3.71243775e-01 -2.43192405e-01 -3.85891974e-01 -1.05198705e+00
7.66082644e-01 -4.93972637e-02 3.86665434e-01 6.13697320e-02
2.62406841e-02 -1.95401296e-01 2.61069536e-01 -5.83487511e-01
4.85617906e-01 -7.82012343e-01 4.88443375e-02 5.62066734e-02
-7.11649120e-01 8.07223856e-01 2.47065023e-01 7.29190230e-01
-1.90856859e-01 4.53009725e-01 3.43379050e-01 7.03869015e-02
-8.50900769e-01 1.10340297e-01 -9.25578922e-02 4.95873801e-02
9.79079902e-01 1.59117460e-01 -8.81457865e-01 -4.23687696e-01
-7.62719870e-01 5.60920358e-01 8.92063022e-01 1.13171792e+00
5.61231375e-01 -2.00723791e+00 -7.05191135e-01 3.50459255e-02
1.04169011e+00 -8.58039737e-01 -4.70210463e-02 9.28414762e-01
-4.36519623e-01 5.21227598e-01 -3.50044698e-01 -7.72694767e-01
-1.35229421e+00 9.53379154e-01 9.61205289e-02 -3.13455239e-02
-5.23465574e-01 5.37357569e-01 4.87640023e-01 -7.17940688e-01
1.58840984e-01 -3.14646274e-01 2.78268103e-02 1.07858358e-02
3.37366760e-01 4.04489413e-02 -2.32420489e-01 -1.24317849e+00
-4.54363078e-01 7.35138416e-01 -8.78681466e-02 -6.56386554e-01
8.88862610e-01 -4.77894992e-01 -1.54636670e-02 7.63046265e-01
1.58284652e+00 1.00171100e-02 -1.09750164e+00 -5.08895755e-01
-3.05634513e-02 -4.51885760e-01 -1.00399218e-01 -1.16631627e+00
-1.03257310e+00 7.84887671e-01 6.02624416e-01 -3.58966798e-01
7.03276634e-01 4.85854626e-01 6.67813420e-01 2.62966275e-01
2.95341760e-01 -1.21909547e+00 5.11400223e-01 5.06929636e-01
1.24277413e+00 -1.67716277e+00 -1.81451976e-01 -4.18769389e-01
-1.42179275e+00 7.65790820e-01 6.16332531e-01 2.72563756e-01
2.96639621e-01 7.96737820e-02 6.17365360e-01 -4.28997308e-01
-9.75582063e-01 -2.40927875e-01 6.78197920e-01 3.31223220e-01
4.78361726e-01 5.00327885e-01 -8.22215453e-02 5.53182125e-01
1.82705298e-01 -3.28938603e-01 6.62240684e-02 7.85193980e-01
1.68378398e-01 -1.08954132e+00 2.12976383e-03 -3.25224519e-01
-1.40413404e-01 -3.58402133e-01 -6.58401251e-01 8.44440460e-01
-1.58384547e-01 8.62539351e-01 3.12982723e-02 -4.73830372e-01
4.17813778e-01 6.76743388e-02 6.85622275e-01 -4.21366900e-01
-6.47833705e-01 3.53662133e-01 1.28943250e-01 -1.01887810e+00
-2.76641130e-01 -5.57953477e-01 -1.04936028e+00 -1.87235788e-01
9.67111140e-02 -3.13975781e-01 8.42276216e-01 6.50066316e-01
6.70041203e-01 3.51792604e-01 3.64127368e-01 -7.31636167e-01
-1.95866629e-01 -7.15036035e-01 -4.47563529e-01 9.48456287e-01
2.30098724e-01 -4.93216276e-01 7.12062195e-02 2.26941869e-01] | [11.271248817443848, 1.5446293354034424] |
c8685d12-e3d5-4164-824a-499a90a18527 | planning-with-spatial-temporal-abstraction | 2210.15751 | null | https://arxiv.org/abs/2210.15751v2 | https://arxiv.org/pdf/2210.15751v2.pdf | Planning with Spatial-Temporal Abstraction from Point Clouds for Deformable Object Manipulation | Effective planning of long-horizon deformable object manipulation requires suitable abstractions at both the spatial and temporal levels. Previous methods typically either focus on short-horizon tasks or make strong assumptions that full-state information is available, which prevents their use on deformable objects. In this paper, we propose PlAnning with Spatial-Temporal Abstraction (PASTA), which incorporates both spatial abstraction (reasoning about objects and their relations to each other) and temporal abstraction (reasoning over skills instead of low-level actions). Our framework maps high-dimension 3D observations such as point clouds into a set of latent vectors and plans over skill sequences on top of the latent set representation. We show that our method can effectively perform challenging sequential deformable object manipulation tasks in the real world, which require combining multiple tool-use skills such as cutting with a knife, pushing with a pusher, and spreading the dough with a roller. | ['David Held', 'Chuang Gan', 'Yunzhu Li', 'Katerina Fragkiadaki', 'Zhiao Huang', 'Yunchu Zhang', 'Carl Qi', 'Xingyu Lin'] | 2022-10-27 | null | null | null | null | ['deformable-object-manipulation'] | ['robots'] | [ 2.18766361e-01 1.08973116e-01 -1.89383551e-01 5.64345680e-02
-2.05504388e-01 -8.06836009e-01 6.70495570e-01 1.61650866e-01
-7.76936710e-02 3.65263671e-01 3.01953673e-01 -1.15175977e-01
-6.05306089e-01 -8.96444738e-01 -7.48375177e-01 -5.85025847e-01
-3.84793848e-01 9.86178160e-01 6.37111843e-01 -2.11517990e-01
1.78721637e-01 8.04864943e-01 -1.51157701e+00 2.84765095e-01
6.54168069e-01 6.31508470e-01 5.38635492e-01 8.79974186e-01
2.24225800e-02 9.57537353e-01 -2.70633280e-01 1.39050558e-01
5.18686056e-01 1.98665723e-01 -1.02747273e+00 3.34150881e-01
5.49698733e-02 -6.83270633e-01 -4.56608593e-01 6.75414920e-01
-2.67421097e-01 7.09131896e-01 5.16029119e-01 -1.21108413e+00
-5.15307128e-01 4.27582026e-01 -2.66685307e-01 -2.72570133e-01
5.27518213e-01 4.99300569e-01 6.35169744e-01 -4.25725669e-01
7.86079645e-01 1.46504462e+00 3.21043760e-01 4.41476792e-01
-1.20909452e+00 -1.36471987e-01 5.47133684e-01 -5.50431237e-02
-9.25767660e-01 1.00244418e-01 6.07845306e-01 -8.97166014e-01
1.10552847e+00 3.50874186e-01 1.02165556e+00 1.10053158e+00
5.07509232e-01 9.19558942e-01 1.02957988e+00 4.12404872e-02
4.66059119e-01 -6.07645810e-01 8.50634575e-02 6.92351937e-01
-3.34420986e-02 2.74835348e-01 -4.94938344e-01 -3.93428504e-01
1.49162340e+00 7.25245476e-01 4.43230718e-02 -8.60143006e-01
-1.53602993e+00 5.13471365e-01 4.18362617e-01 -2.02969837e-04
-7.27699637e-01 4.95411426e-01 4.48634587e-02 2.83988193e-03
3.38500589e-01 5.75463712e-01 -4.78598982e-01 -2.17441931e-01
-7.62007117e-01 8.62173319e-01 4.65406448e-01 1.41926455e+00
5.18446147e-01 -1.26902163e-01 -4.54934448e-01 4.49084602e-02
1.82256609e-01 2.92389661e-01 -3.03069334e-02 -1.40148652e+00
6.00271225e-01 6.38331234e-01 5.95403016e-01 -8.16394031e-01
-4.66559380e-01 2.59816766e-01 -4.19235140e-01 6.31418705e-01
3.66648287e-01 1.02209255e-01 -1.43595481e+00 1.40100360e+00
5.29332161e-01 2.49024346e-01 -5.74777648e-02 1.01206088e+00
1.95093334e-01 7.01960981e-01 1.69163551e-02 -8.68355334e-02
1.23192453e+00 -1.06948888e+00 -7.19749510e-01 -1.27054662e-01
3.76141429e-01 -3.27542245e-01 1.12138438e+00 3.53915006e-01
-1.55095899e+00 -5.63022256e-01 -6.49924219e-01 -4.10018086e-01
-3.02844048e-01 -2.21286893e-01 8.59444559e-01 -2.49814719e-01
-7.86795676e-01 1.07244909e+00 -1.69504797e+00 -2.01708972e-01
1.73857272e-01 2.74877787e-01 -4.84325528e-01 -2.52402425e-01
-5.81914425e-01 9.82933760e-01 4.63194072e-01 2.71972090e-01
-1.43037856e+00 -4.60642993e-01 -1.00784183e+00 7.16000842e-03
9.32466328e-01 -7.83155620e-01 1.36796129e+00 -2.86621936e-02
-1.91014433e+00 2.85000682e-01 1.31923243e-01 -1.55703560e-01
5.46956360e-01 -7.02439487e-01 3.23322654e-01 -4.91277538e-02
1.30413296e-02 4.74418074e-01 7.90527165e-01 -1.21553612e+00
-2.96108097e-01 -6.45875096e-01 8.13415051e-01 4.54013199e-01
3.04450095e-01 -2.62302428e-01 -3.47644240e-01 -5.44786274e-01
6.47354603e-01 -1.37485778e+00 -6.27728522e-01 1.25459135e-01
-4.79940772e-01 -2.09483758e-01 8.80092323e-01 -5.85102916e-01
8.40357959e-01 -1.94023955e+00 1.17753828e+00 8.05011541e-02
1.07557207e-01 -7.15104938e-02 2.58207340e-02 7.21188128e-01
2.35511616e-01 -1.85704380e-01 -2.69515932e-01 -2.03589320e-01
1.97896183e-01 6.74032569e-01 -4.58879501e-01 3.07047695e-01
-4.57425490e-02 1.10736489e+00 -1.42524385e+00 -2.43449420e-01
3.94988894e-01 1.52761087e-01 -6.94897234e-01 1.82638973e-01
-8.36010993e-01 6.76523447e-01 -9.30505872e-01 6.09893799e-01
3.33189249e-01 -4.65442613e-02 2.30421707e-01 2.76904166e-01
-5.23234069e-01 3.72390538e-01 -1.19595420e+00 2.27723336e+00
-3.77638578e-01 7.08193937e-03 1.68345436e-01 -2.90524185e-01
3.97939086e-01 4.74807382e-01 6.11382008e-01 -6.09769970e-02
-3.31067920e-01 -1.60093948e-01 -2.56433249e-01 -6.40478909e-01
6.36453092e-01 -5.97820170e-02 -4.72835720e-01 3.86460423e-01
-1.66968971e-01 -8.96691382e-01 -9.98580903e-02 2.53851354e-01
1.43251038e+00 8.38535428e-01 2.56790161e-01 -2.14459285e-01
-1.96211413e-01 3.50285679e-01 4.66077656e-01 6.99665189e-01
1.01415716e-01 2.93969423e-01 4.23465431e-01 -5.73839247e-01
-9.37380075e-01 -1.43383920e+00 2.32112795e-01 9.92859244e-01
3.98211598e-01 -3.18163246e-01 -3.93234581e-01 -5.33803761e-01
2.22492471e-01 7.31338263e-01 -7.92799890e-01 -1.11758085e-02
-7.06436515e-01 1.37989327e-01 -6.98598027e-02 9.46970463e-01
1.51498854e-01 -1.13965118e+00 -1.29148233e+00 3.00286829e-01
3.38829085e-02 -8.42620909e-01 -3.45741153e-01 1.40534252e-01
-1.16306114e+00 -9.04418111e-01 -6.56747639e-01 -3.13095182e-01
6.87587321e-01 3.00639272e-01 7.74153650e-01 -1.19358608e-02
-2.68912256e-01 6.41409695e-01 -3.90105188e-01 -5.00058532e-01
-1.15791567e-01 -4.01650876e-01 1.90511972e-01 -5.16171932e-01
-4.90880489e-01 -6.54834032e-01 -2.44825915e-01 1.75496802e-01
-7.67983079e-01 5.25842845e-01 3.50917161e-01 4.43268776e-01
6.34823501e-01 2.19341561e-01 -5.22629440e-01 -4.90260303e-01
4.21556205e-01 -2.67585754e-01 -7.28016615e-01 2.08806902e-01
2.05885753e-01 1.12538770e-01 -3.84229980e-02 -7.68663406e-01
-1.15650284e+00 1.65162235e-01 5.42379022e-01 -8.71211827e-01
-1.48733780e-01 5.05609810e-01 5.23132496e-02 2.80957013e-01
3.63802016e-01 -1.68078281e-02 -3.24265927e-01 -5.42874694e-01
4.42335546e-01 -1.68163419e-01 3.93347919e-01 -1.06195366e+00
9.02133405e-01 6.67602420e-01 3.38642389e-01 -4.26170349e-01
-7.69191802e-01 -3.07556957e-01 -1.16121209e+00 -2.43902802e-01
1.15140653e+00 -6.04838967e-01 -9.95729208e-01 5.19057155e-01
-1.20397973e+00 -9.62979734e-01 -6.90524459e-01 6.02122426e-01
-1.07446826e+00 5.64262345e-02 -8.08070660e-01 -9.92184162e-01
4.21705008e-01 -1.26977444e+00 1.48187780e+00 -1.64035842e-01
-4.70413417e-01 -6.10071421e-01 4.18513454e-02 2.02055067e-01
1.05117582e-01 7.04795897e-01 9.07066107e-01 -8.50334391e-02
-1.20334911e+00 -1.87139097e-03 2.63170570e-01 -2.51001716e-01
2.67920196e-01 -1.28181949e-01 -3.25267911e-01 -2.98630863e-01
-1.88816246e-02 -2.51041085e-01 4.60401684e-01 4.07790959e-01
1.36076057e+00 -4.44280475e-01 -6.07908130e-01 2.52567083e-01
1.06118083e+00 2.52160251e-01 4.37000811e-01 1.70568246e-02
8.84549856e-01 7.72323847e-01 9.88839209e-01 3.93469274e-01
3.78757715e-01 1.04642117e+00 6.70944989e-01 2.99279541e-01
2.11740837e-01 -3.08493942e-01 3.55475396e-01 4.91385907e-01
-7.69908309e-01 -4.07646708e-02 -1.21331072e+00 6.73371613e-01
-2.13866496e+00 -9.78571415e-01 -7.23914504e-02 2.00739980e+00
5.25157571e-01 1.23291507e-01 2.69060545e-02 -2.99755652e-02
2.05276504e-01 1.82900548e-01 -7.73786485e-01 -2.63850063e-01
8.17483366e-01 -3.53965573e-02 4.35242862e-01 8.04674387e-01
-9.11900043e-01 1.09815013e+00 6.13743544e+00 2.29512453e-01
-7.48371720e-01 2.43287962e-02 -3.68916839e-01 -4.38856959e-01
-3.24572474e-01 2.38775954e-01 -3.91815424e-01 6.41762242e-02
3.48343313e-01 3.01033165e-02 6.57772958e-01 8.10242116e-01
-4.75772768e-02 -1.02933057e-01 -1.56390762e+00 3.84542406e-01
-5.84029138e-01 -1.10458970e+00 2.63152923e-02 2.37819076e-01
7.01111317e-01 -2.68851399e-01 1.46802098e-01 4.69143271e-01
1.09555149e+00 -9.19692576e-01 1.06494284e+00 8.18693995e-01
3.47571015e-01 -2.64384240e-01 7.83951059e-02 9.51905310e-01
-1.19617665e+00 -2.07074806e-01 -7.35577717e-02 -6.25703990e-01
5.48370600e-01 1.40609488e-01 -7.16487288e-01 6.34833574e-01
6.25682354e-01 4.39477205e-01 3.41891825e-01 5.51577926e-01
-3.89471412e-01 -3.38898748e-02 -5.30603647e-01 3.56322587e-01
3.56910378e-01 -3.34261566e-01 8.02789688e-01 5.29090881e-01
2.16293991e-01 7.25334823e-01 8.79566312e-01 9.18119192e-01
6.06850505e-01 -6.52991951e-01 -7.54916430e-01 -1.59392446e-01
2.01688662e-01 7.64434636e-01 -8.87670219e-01 -4.18688536e-01
-4.23717648e-02 8.97579908e-01 3.27419609e-01 5.21928787e-01
-7.69086838e-01 2.38121465e-01 9.82125044e-01 2.10084409e-01
1.20808654e-01 -1.13976312e+00 -2.09080234e-01 -1.08652711e+00
2.31395975e-01 -3.88406932e-01 1.04676358e-01 -1.05298758e+00
-6.77782953e-01 6.10193424e-02 7.60179281e-01 -1.14743769e+00
-1.97834551e-01 -4.22870427e-01 -3.34352523e-01 8.93286288e-01
-1.00894034e+00 -1.26967061e+00 -3.91302854e-01 6.94818616e-01
8.34291935e-01 5.31170428e-01 8.11363280e-01 -5.56553900e-01
-1.07390620e-02 -4.49338883e-01 -4.42478955e-01 -2.18804121e-01
-3.75222601e-02 -1.23475718e+00 6.69866562e-01 6.58138871e-01
-3.02642258e-03 8.38991702e-01 8.03948343e-01 -1.16846323e+00
-1.65050101e+00 -8.43498230e-01 3.05460006e-01 -9.49053764e-01
6.27852798e-01 -6.59338474e-01 -1.05138075e+00 1.37827849e+00
-3.99160892e-01 6.84781671e-02 -2.59991121e-02 2.32251808e-01
-2.49353535e-02 6.50002897e-01 -1.08067811e+00 8.66032422e-01
1.55569160e+00 -4.53175277e-01 -9.06139672e-01 6.36597574e-01
1.09610510e+00 -1.21351326e+00 -9.98454750e-01 6.46942735e-01
5.13507664e-01 -4.92644191e-01 1.23976696e+00 -1.06947803e+00
5.13188124e-01 -2.62805432e-01 -1.67860404e-01 -1.28935421e+00
-7.12888479e-01 -5.80773413e-01 -4.96445358e-01 5.17454088e-01
-1.01895824e-01 -2.75452256e-01 7.65022755e-01 8.85350049e-01
-3.52424800e-01 -7.83334315e-01 -7.83411682e-01 -9.91722047e-01
-9.89809334e-02 -4.72229004e-01 7.37945378e-01 8.07983577e-01
1.23881385e-01 -3.93882245e-01 -3.17391962e-01 7.61572003e-01
4.70687777e-01 3.85511905e-01 8.28480065e-01 -1.40492463e+00
-4.86439526e-01 -1.42108157e-01 -3.70601535e-01 -1.17390203e+00
1.33944795e-01 -5.61917543e-01 5.03750861e-01 -1.88921976e+00
2.85952296e-02 -6.13201439e-01 1.69617322e-03 8.17580819e-01
9.91997644e-02 -6.55899644e-01 5.97868264e-01 2.80542344e-01
-5.07309020e-01 6.73167348e-01 1.85432446e+00 -5.92326820e-02
-4.07701552e-01 -2.45083608e-02 6.05267026e-02 1.09532917e+00
4.09931093e-01 -5.06728053e-01 -7.13311493e-01 -7.80427516e-01
8.43812600e-02 8.84624183e-01 5.05701303e-01 -1.03193784e+00
1.48851901e-01 -1.10854352e+00 2.26563774e-02 -4.58804876e-01
1.00441146e+00 -1.04701877e+00 4.39614087e-01 5.21115541e-01
-4.74870980e-01 1.92154542e-01 3.70630980e-01 9.00768638e-01
2.10011661e-01 1.71543509e-01 2.24508554e-01 -4.62748796e-01
-6.30034387e-01 6.45008266e-01 -3.51740062e-01 -5.71595132e-01
1.55460942e+00 -1.68099850e-01 -1.35011300e-01 -9.67293233e-02
-1.37550783e+00 4.70329493e-01 8.37557375e-01 6.85512483e-01
6.56829357e-01 -1.18621039e+00 -3.71726662e-01 8.36299807e-02
-2.82422248e-02 7.36377418e-01 4.34472084e-01 6.18141413e-01
-3.39231402e-01 3.58478099e-01 -4.82542634e-01 -6.27890229e-01
-1.16127658e+00 9.52287912e-01 -6.49801716e-02 -5.64725459e-01
-1.04443634e+00 9.43293989e-01 4.69782799e-01 -6.11732900e-01
1.43465251e-01 -9.60273564e-01 6.49266988e-02 -3.19010258e-01
2.04483941e-01 5.84903657e-01 -3.27737182e-01 -2.86175489e-01
-2.59888142e-01 6.78809524e-01 1.29075319e-01 -3.35214972e-01
1.39666188e+00 2.92202175e-01 -1.25878513e-01 9.24143255e-01
4.00191605e-01 -2.67246783e-01 -1.82821953e+00 -2.60787576e-01
-1.81557968e-01 -7.52937078e-01 -1.56290382e-01 -5.36462665e-01
-3.56045306e-01 8.73303413e-01 -8.50281119e-02 1.83929488e-01
6.16304040e-01 2.45387584e-01 6.12308919e-01 4.96806979e-01
1.17748821e+00 -9.69091773e-01 4.41236526e-01 7.30383813e-01
1.41180348e+00 -7.23787725e-01 1.57150149e-01 -6.44427240e-01
-5.89190722e-01 8.12931657e-01 6.02443874e-01 -2.93747455e-01
5.12450039e-01 4.38299894e-01 -4.06882226e-01 -6.27296805e-01
-9.57078934e-01 -1.22356638e-01 3.13383818e-01 4.55910474e-01
-1.32279605e-01 4.95981932e-01 2.83110797e-01 1.49873734e-01
-1.65336147e-01 5.62336519e-02 4.71493781e-01 1.78798974e+00
-4.43199694e-01 -8.33116412e-01 -3.58033776e-01 3.05305392e-01
1.15944885e-01 3.81284416e-01 -1.05461203e-01 7.68711686e-01
1.08713724e-01 5.25377035e-01 2.73142040e-01 -2.17503697e-01
6.69003904e-01 6.39048070e-02 9.97016966e-01 -1.21835029e+00
-3.10154080e-01 -2.37637609e-01 -2.01492965e-01 -1.18569112e+00
-3.61847341e-01 -1.09028125e+00 -1.48280609e+00 -5.32216728e-02
6.73544556e-02 -1.97363824e-01 6.61718249e-01 1.10501432e+00
2.22868726e-01 8.23800921e-01 -1.60566285e-01 -1.58766687e+00
-6.84879541e-01 -6.87059939e-01 -5.79534233e-01 5.33813655e-01
5.44751942e-01 -1.20685482e+00 2.17296988e-01 6.69364259e-03] | [4.6679534912109375, 0.6883776783943176] |
a4b755a9-fee5-46b5-b857-b7297a85f723 | direct-attacks-using-fake-images-in-iris | 2111.00178 | null | https://arxiv.org/abs/2111.00178v1 | https://arxiv.org/pdf/2111.00178v1.pdf | Direct attacks using fake images in iris verification | In this contribution, the vulnerabilities of iris-based recognition systems to direct attacks are studied. A database of fake iris images has been created from real iris of the BioSec baseline database. Iris images are printed using a commercial printer and then, presented at the iris sensor. We use for our experiments a publicly available iris recognition system, which some modifications to improve the iris segmentation step. Based on results achieved on different operational scenarios, we show that the system is vulnerable to direct attacks, pointing out the importance of having countermeasures against this type of fraudulent actions. | ['Javier Ortega-Garcia', 'Julian Fierrez', 'Javier Galbally', 'Fernando Alonso-Fernandez', 'Pedro Tome-Gonzalez', 'Virginia Ruiz-Albacete'] | 2021-10-30 | null | null | null | null | ['iris-segmentation'] | ['medical'] | [ 4.57525820e-01 1.91030636e-01 -5.92921078e-02 -3.45783770e-01
1.91331446e-01 -7.08021343e-01 6.43786132e-01 8.72033983e-02
-4.34769392e-01 6.00066066e-01 -3.26733977e-01 -5.98354697e-01
-2.58437127e-01 -7.39589632e-01 -4.67857033e-01 -5.17650127e-01
-1.73380718e-01 4.85421777e-01 -7.81211630e-02 1.11888265e-02
8.02831352e-01 1.08645952e+00 -1.55112839e+00 3.00081611e-01
7.44499981e-01 5.47059357e-01 -1.08477831e+00 1.19454288e+00
2.97953308e-01 5.03422022e-01 -1.07493341e+00 -7.43563890e-01
8.96335304e-01 -5.86427152e-01 -8.90651464e-01 1.73600063e-01
6.62450373e-01 -6.97384238e-01 -3.76935601e-02 1.27402651e+00
4.46763217e-01 -5.33841252e-01 2.89565563e-01 -8.68025661e-01
-1.28758624e-01 4.07338977e-01 -4.26735461e-01 2.74450094e-01
5.72956622e-01 4.80296999e-01 7.96936732e-03 -1.73434019e-01
5.38803518e-01 1.07875407e+00 4.37929362e-01 6.99000657e-01
-1.27561009e+00 -6.55662179e-01 -7.72099078e-01 -3.20068240e-01
-1.52487743e+00 -6.12081647e-01 2.26775333e-01 -4.47455198e-01
7.19807506e-01 6.70098662e-01 5.79014063e-01 8.98667991e-01
2.19718605e-01 2.43188009e-01 1.92760015e+00 -7.58982718e-01
-8.29947516e-02 5.31306565e-01 2.36378402e-01 6.78703904e-01
7.92537510e-01 8.66241574e-01 -1.65081650e-01 -5.75803816e-01
9.55594778e-01 -3.20604771e-01 -1.37550831e-01 -1.55892866e-02
-6.93896770e-01 4.19490337e-01 2.58555613e-03 5.61829388e-01
-2.72681653e-01 -2.89216161e-01 6.24338984e-02 4.27264988e-01
5.83848655e-02 6.03809655e-01 -3.43372151e-02 -2.18424156e-01
-7.46362329e-01 -2.64264822e-01 9.59469676e-01 4.37203854e-01
3.36674035e-01 -3.84140819e-01 4.00879607e-02 2.01648548e-01
4.61301714e-01 6.55820191e-01 3.52053344e-01 -2.97464848e-01
7.54381791e-02 8.39080870e-01 2.55865127e-01 -1.18314445e+00
-1.88509852e-01 -1.48553878e-01 -4.56462353e-01 4.27065432e-01
6.58725917e-01 -1.81268141e-01 -1.21416175e+00 6.01926565e-01
2.03650951e-01 7.66163170e-01 3.08947235e-01 7.84671485e-01
5.56415379e-01 1.64445583e-03 -1.35981545e-01 -3.73513773e-02
1.09234869e+00 -9.70703140e-02 -6.59174144e-01 6.46568358e-01
4.88020122e-01 -1.16774905e+00 4.84804422e-01 9.62216079e-01
-8.23988676e-01 -3.22200686e-01 -1.01060915e+00 2.88007379e-01
-6.28917277e-01 3.60868692e-01 5.10154247e-01 1.86642206e+00
-8.42996478e-01 5.98755836e-01 -8.12169731e-01 -5.25359273e-01
3.02609593e-01 8.94182563e-01 -3.08392614e-01 3.76432955e-01
-8.03058922e-01 9.02339876e-01 2.68155187e-01 1.41099051e-01
-1.42505333e-01 -7.17147440e-02 -4.19107854e-01 -4.57225472e-01
-1.24335818e-01 -1.50223017e-01 7.97287405e-01 -1.09729517e+00
-1.86888909e+00 1.51297557e+00 8.41528699e-02 -9.29872274e-01
7.30448961e-01 1.24569520e-01 -9.37365711e-01 3.11342925e-01
-6.25227928e-01 -9.71300229e-02 1.12754750e+00 -9.64398980e-01
-5.97925961e-01 -4.69379008e-01 -1.11920521e-01 -6.55132234e-01
1.45202175e-01 5.05801737e-01 -6.19201660e-02 -5.65042257e-01
-2.38280222e-01 -1.04664040e+00 -8.13083798e-02 -6.44965649e-01
-5.87082148e-01 3.43673438e-01 5.79280436e-01 -5.65498233e-01
1.24692464e+00 -2.12238908e+00 -5.19158900e-01 1.02936518e+00
-9.71774384e-02 1.26923072e+00 1.59668997e-01 2.10669190e-01
-1.98273435e-01 4.34604317e-01 1.48592908e-02 2.43438967e-02
-4.09461111e-01 3.00429970e-01 -5.21480203e-01 7.74136305e-01
9.81192011e-03 5.14224470e-01 -5.66806972e-01 -3.28410417e-01
5.08957326e-01 3.27268332e-01 -1.00943513e-01 -2.52197944e-02
2.07853764e-01 5.14748812e-01 -5.50778210e-01 1.11376441e+00
9.25677776e-01 1.22895606e-01 1.42801866e-01 3.15959930e-01
-1.25578567e-01 2.04277337e-01 -1.26929605e+00 8.99230957e-01
1.35261402e-01 3.98029864e-01 -1.31023630e-01 -5.29165089e-01
1.06628859e+00 5.06120920e-01 8.43371004e-02 -3.07380289e-01
4.91975933e-01 3.96404892e-01 1.57908633e-01 -5.00635803e-01
4.66104865e-01 1.05724722e-01 4.38847959e-01 6.98050201e-01
-1.26217976e-01 1.41614899e-01 3.37243155e-02 -1.82771072e-01
6.76695347e-01 -1.24468863e-01 3.08934808e-01 -9.15951282e-02
9.26527858e-01 2.67638236e-01 1.95059344e-01 8.88245821e-01
-4.80346411e-01 3.00449610e-01 5.34441471e-01 -7.55137861e-01
-6.22401774e-01 -6.18842602e-01 -7.92685866e-01 -1.86643541e-01
4.70135286e-02 -1.49214849e-01 -1.08252978e+00 -8.92707646e-01
2.79149443e-01 2.17973158e-01 -6.23524666e-01 1.05584234e-01
-3.51739705e-01 -9.82660174e-01 1.21493089e+00 -3.35644841e-01
4.37895298e-01 -7.69354641e-01 -6.47974312e-01 1.08129807e-01
5.49244702e-01 -6.80541813e-01 8.25080350e-02 -6.50728762e-01
-1.04774308e+00 -1.77625418e+00 -3.34472746e-01 -1.85254112e-01
1.04119003e+00 -1.80800632e-01 7.47350514e-01 7.51494944e-01
-6.70917273e-01 4.32280302e-01 -2.24462599e-01 -7.45186865e-01
-9.33235347e-01 -3.46000135e-01 2.47309148e-01 6.26115203e-01
1.09192717e+00 6.01140447e-02 -3.85201842e-01 4.70071077e-01
-8.34333897e-01 -6.41430795e-01 4.51025188e-01 5.32938778e-01
3.75629932e-01 5.80142960e-02 -1.51314974e-01 -1.27584279e+00
5.51020324e-01 6.27298132e-02 -1.31411326e+00 2.82270759e-01
-9.36557651e-01 -1.64975181e-01 1.66626811e-01 -2.01419428e-01
-6.88931286e-01 2.86139637e-01 1.31049722e-01 -6.64270520e-02
-7.39863515e-01 -8.74857530e-02 2.25896820e-01 -1.08348846e+00
7.70081878e-01 1.17292823e-02 5.16008794e-01 -6.37113631e-01
-7.13213384e-02 1.25914788e+00 5.57905972e-01 -2.51807541e-01
7.12858737e-01 5.17957509e-01 1.83152318e-01 -9.58899200e-01
-7.26796314e-03 -4.17726099e-01 -5.03995419e-01 -1.66209206e-01
2.35777944e-01 -3.95552456e-01 -1.16954267e+00 1.26533937e+00
-8.74929011e-01 2.17217922e-01 4.67090756e-02 5.98788559e-01
-2.80231297e-01 5.82820117e-01 -5.27990162e-01 -8.40344131e-01
-2.03882739e-01 -1.10195386e+00 5.40870965e-01 6.19866550e-01
4.05121446e-02 -7.17162132e-01 3.28090847e-01 5.99856019e-01
3.26522768e-01 5.06666601e-01 1.49440825e-01 -7.45400667e-01
-6.51866257e-01 -5.71400106e-01 -1.11850137e-02 4.14478868e-01
4.13139015e-01 8.50711644e-01 -1.06762862e+00 -3.82933944e-01
1.09947205e-01 7.85309747e-02 6.20636523e-01 1.25776723e-01
7.97929287e-01 -3.86443615e-01 -4.60612655e-01 8.35190237e-01
1.55696070e+00 4.07665074e-01 1.16068113e+00 5.90307236e-01
6.25019148e-02 5.57868958e-01 6.84388280e-01 2.31060505e-01
-2.69975007e-01 5.12105048e-01 2.43346602e-01 -4.23678130e-01
2.89274424e-01 1.56239182e-01 -5.99987693e-02 -2.84906268e-01
-6.20479703e-01 1.59681514e-01 -1.02663398e+00 3.38016540e-01
-1.41831625e+00 -7.55906463e-01 -5.38341522e-01 2.66327739e+00
9.46674883e-01 3.09327580e-02 2.30034634e-01 2.10372056e-03
6.52071059e-01 -4.19306517e-01 -1.47243664e-01 -9.80932593e-01
-1.27387807e-01 7.88671076e-01 1.07020998e+00 7.89439857e-01
-1.32012773e+00 9.31509376e-01 7.68690729e+00 9.57934260e-02
-1.43435574e+00 -3.62545252e-01 4.55952317e-01 -7.23096132e-02
5.60589552e-01 6.26190053e-03 -7.85687566e-01 5.58500588e-01
1.31036866e+00 1.70109216e-02 3.84645075e-01 2.15367228e-01
1.71658576e-01 -4.27903742e-01 -5.78276753e-01 7.49739945e-01
1.41711950e-01 -1.21426785e+00 -2.38251388e-02 5.69369495e-01
4.53512222e-01 -2.43850082e-01 1.73157200e-01 -5.47208071e-01
2.03414917e-01 -1.18803215e+00 -2.27592990e-01 9.15961087e-01
7.73133457e-01 -1.03378212e+00 1.12007022e+00 -1.22554697e-01
-2.33342558e-01 2.67996401e-01 -2.88728893e-01 1.92815214e-02
-3.25318247e-01 4.05062407e-01 -1.21095419e+00 5.36917508e-01
3.86232257e-01 3.19003344e-01 -9.24423397e-01 1.21548355e+00
-1.70406640e-01 7.86250770e-01 -5.54044485e-01 2.77971119e-01
-1.37043878e-01 -4.10166025e-01 5.32247424e-01 1.03271008e+00
-4.17951755e-02 3.08204979e-01 -6.83727324e-01 7.66743541e-01
3.38728398e-01 1.77002251e-02 -8.59289646e-01 -4.88387585e-01
1.29829338e-02 8.06216300e-01 -5.71898103e-01 -3.19998443e-01
-4.03552890e-01 8.47284496e-01 -5.69911659e-01 -1.45345833e-02
-2.35904887e-01 -6.18632555e-01 7.68920660e-01 1.39979094e-01
3.12532298e-02 3.93251404e-02 -4.06453073e-01 -1.12300587e+00
-2.71570712e-01 -1.32228780e+00 3.70051801e-01 -2.78292716e-01
-7.77047992e-01 4.84955817e-01 -3.37492019e-01 -1.40417457e+00
-3.54049802e-01 -7.30934143e-01 -2.88251221e-01 1.32522452e+00
-1.12847078e+00 -1.01641929e+00 2.53599793e-01 7.26247907e-01
-6.50454819e-01 -7.35151768e-01 1.07576108e+00 -1.02905206e-01
-6.72854841e-01 8.84332001e-01 -7.25588724e-02 3.15968841e-01
8.27062726e-01 -1.08593524e+00 5.73207378e-01 1.21916878e+00
3.04870009e-01 1.06492615e+00 5.94071090e-01 -7.73908913e-01
-1.20427406e+00 -5.03957450e-01 9.93099153e-01 -7.25164890e-01
4.60219085e-01 8.42165649e-02 -8.12553227e-01 7.33431995e-01
2.38669962e-01 -1.07965387e-01 9.55819249e-01 -1.41307965e-01
-1.81095377e-02 1.35159001e-01 -1.79423010e+00 3.02045703e-01
1.32625327e-01 -5.50363183e-01 -7.99815297e-01 3.63636255e-01
-2.87469804e-01 -6.72386408e-01 -9.98417735e-01 1.60766006e-01
6.05319500e-01 -1.34243464e+00 8.15866113e-01 -7.55148709e-01
-1.29958078e-01 -6.42142475e-01 6.38621747e-01 -7.43848860e-01
5.13927877e-01 -1.05018044e+00 4.61509041e-02 1.06965017e+00
2.60471463e-01 -1.30888212e+00 8.37331414e-01 7.68960416e-01
9.19996440e-01 -9.82785597e-02 -8.15945446e-01 -5.85829735e-01
-3.42373997e-01 2.02854648e-01 9.14680123e-01 1.22587514e+00
7.16661364e-02 -5.83598018e-01 -4.39764142e-01 7.74577796e-01
7.53004372e-01 1.01136647e-01 1.13571560e+00 -1.22223365e+00
-2.62819499e-01 -2.42589697e-01 -9.99820471e-01 -1.92906424e-01
-5.06373346e-01 -1.96731448e-01 -7.72137582e-01 -3.06159109e-01
-4.35666621e-01 -2.16659114e-01 -2.28321373e-01 5.80026627e-01
1.16234936e-01 6.07029676e-01 9.16075148e-03 3.15301508e-01
3.46335322e-01 -6.69709325e-01 7.94892073e-01 3.09746433e-02
-3.39106739e-01 5.38681567e-01 -3.00179392e-01 6.32890761e-01
9.42215025e-01 -6.15789175e-01 2.09419608e-01 2.90370107e-01
-8.33974928e-02 -1.12217702e-01 3.90618891e-01 -9.80548978e-01
2.34379604e-01 -2.02269241e-01 2.21417338e-01 -3.74887496e-01
-1.67850181e-01 -9.78919566e-01 4.02802616e-01 9.62970495e-01
1.63895011e-01 -4.63738412e-01 4.26008195e-01 1.77728444e-01
-3.70783865e-01 -3.36763471e-01 8.80193770e-01 1.03577457e-01
-2.43590578e-01 -1.34307921e-01 -2.50010490e-01 -7.07350016e-01
1.12539816e+00 -7.51497447e-01 -6.57033503e-01 1.44044489e-01
-8.34044278e-01 -7.08592236e-02 1.15793455e+00 2.08896771e-01
3.97495985e-01 -6.39208198e-01 -5.83020747e-01 1.25684094e+00
1.91340104e-01 -7.21198499e-01 -2.46891722e-01 8.75924945e-01
-1.26030755e+00 5.93119919e-01 -5.68682969e-01 -4.49805498e-01
-1.97202599e+00 5.98739743e-01 8.88859332e-01 1.22793995e-01
-4.58795816e-01 5.60678720e-01 -8.15321147e-01 -1.55490428e-01
1.57775342e-01 -4.18125451e-01 -3.32757920e-01 -3.12780142e-01
9.53354657e-01 3.25147361e-01 3.19585979e-01 -7.78320849e-01
-3.43176931e-01 6.96629107e-01 -2.40713134e-01 3.41367126e-02
7.97215283e-01 1.08514607e-01 -5.55742860e-01 -2.68439054e-01
3.86938840e-01 5.83248258e-01 -5.65429330e-01 -2.54944079e-02
1.58301741e-01 -1.21733725e+00 -8.84071663e-02 -1.19919801e+00
-9.96035218e-01 4.88433510e-01 1.10267782e+00 4.75152254e-01
1.25680399e+00 -7.87697971e-01 2.91639388e-01 2.56364405e-01
6.28505945e-01 -9.32796597e-01 -1.02007449e+00 1.14750281e-01
3.78445536e-01 -1.17508161e+00 1.74284592e-01 -2.88910419e-01
-3.09237331e-01 1.34508693e+00 -3.33009511e-02 -1.93542808e-01
6.69444025e-01 2.27469727e-01 7.28302956e-01 -2.49800667e-01
-1.90306623e-02 -1.51772397e-02 3.68378848e-01 8.65375698e-01
1.87352881e-01 3.47682416e-01 -9.05816853e-01 -3.77690569e-02
-6.40121698e-02 5.21095991e-01 1.08244133e+00 9.85167086e-01
3.49043086e-02 -1.77280223e+00 -1.02167606e+00 4.91255850e-01
-9.30306792e-01 1.38871938e-01 -1.23126090e+00 7.99903452e-01
2.78751791e-01 1.00292170e+00 -1.79101884e-01 -4.32645559e-01
3.85901153e-01 -7.88026527e-02 3.60852540e-01 -3.32782030e-01
-1.27063811e+00 -1.76788807e-01 1.16101220e-01 -8.02839518e-01
-7.87393510e-01 -6.23513401e-01 -7.12717891e-01 -5.99557221e-01
-2.25346938e-01 2.85939220e-02 8.98160279e-01 6.90166175e-01
5.23091853e-01 -1.13036744e-01 5.89861512e-01 -1.56092390e-01
-5.71394145e-01 -7.08470881e-01 -1.10552359e+00 2.33793363e-01
6.06949270e-01 -2.87440896e-01 -5.87906539e-01 1.07408404e-01] | [3.7449095249176025, -3.6267409324645996] |
4cfa60ad-ebd1-4af0-8642-c2b7ae50fcdf | blind-image-quality-assessment-using-semi | null | null | http://openaccess.thecvf.com/content_cvpr_2014/html/Tang_Blind_Image_Quality_2014_CVPR_paper.html | http://openaccess.thecvf.com/content_cvpr_2014/papers/Tang_Blind_Image_Quality_2014_CVPR_paper.pdf | Blind Image Quality Assessment using Semi-supervised Rectifier Networks | It is often desirable to evaluate images quality with a perceptually relevant measure that does not require a reference image. Recent approaches to this problem use human provided quality scores with machine learning to learn a measure. The biggest hurdles to these efforts are: 1) the difficulty of generalizing across diverse types of distortions and 2) collecting the enormity of human scored training data that is needed to learn the measure. We present a new blind image quality measure that addresses these difficulties by learning a robust, nonlinear kernel regression function using a rectifier neural network. The method is pre-trained with unlabeled data and fine-tuned with labeled data. It generalizes across a large set of images and distortion types without the need for a large amount of labeled data. We evaluate our approach on two benchmark datasets and show that it not only outperforms the current state of the art in blind image quality estimation, but also outperforms the state of the art in non-blind measures. Furthermore, we show that our semi-supervised approach is robust to using varying amounts of labeled data. | ['Neel Joshi', 'Huixuan Tang', 'Ashish Kapoor'] | 2014-06-01 | null | null | null | cvpr-2014-6 | ['image-quality-estimation', 'blind-image-quality-assessment'] | ['computer-vision', 'computer-vision'] | [ 1.26058906e-01 -4.27089125e-01 -1.26339048e-01 -6.38169169e-01
-1.32092059e+00 -7.98465669e-01 3.78424436e-01 3.22985388e-02
-6.96288466e-01 6.22833550e-01 4.39092755e-01 -2.37521723e-01
-3.14207464e-01 -3.95557761e-01 -4.55746561e-01 -4.59484994e-01
-3.86034660e-02 1.66057125e-01 8.79076496e-02 5.60767874e-02
4.66766745e-01 4.68000025e-01 -1.66529310e+00 7.25512058e-02
1.07553613e+00 1.13541543e+00 2.23130062e-02 8.83187771e-01
2.93434441e-01 8.24878395e-01 -8.52824092e-01 -5.35765588e-01
6.64720118e-01 -5.68130493e-01 -7.92007327e-01 3.63028407e-01
8.73125017e-01 -4.35037732e-01 -1.30772069e-01 1.18638396e+00
8.84248793e-01 -5.10155049e-04 9.34948325e-01 -9.47577953e-01
-8.82994473e-01 1.99080165e-02 -2.26162925e-01 1.98126912e-01
3.19412023e-01 3.01977843e-01 9.49310541e-01 -7.67241299e-01
2.79353261e-01 1.01493669e+00 5.83114445e-01 6.23609424e-01
-1.22468734e+00 -4.07491118e-01 -2.72403777e-01 3.51797074e-01
-1.23734558e+00 -6.94319248e-01 4.33894008e-01 -6.27324104e-01
8.24600279e-01 1.80331931e-01 4.83382463e-01 7.99396753e-01
-4.47105527e-01 5.17248988e-01 1.75161386e+00 -7.71156967e-01
5.73081851e-01 3.12435985e-01 -6.86535314e-02 7.34245777e-01
3.51720527e-02 4.47011054e-01 -3.43765020e-01 5.63568110e-03
7.97191143e-01 -3.35994840e-01 -6.66144729e-01 -7.34866321e-01
-1.12633812e+00 6.65365279e-01 5.21817744e-01 1.76043704e-01
-1.17804080e-01 -1.25213742e-01 1.97338566e-01 7.72799671e-01
2.88071334e-01 8.46321523e-01 -4.56817597e-01 -3.64425331e-02
-1.49999511e+00 -1.67887568e-01 8.86349678e-01 6.96203768e-01
8.27121794e-01 -1.71604872e-01 -1.06103554e-01 1.06869376e+00
1.81104928e-01 6.11621082e-01 4.99069422e-01 -1.14207661e+00
2.80594260e-01 4.05276179e-01 3.71752173e-01 -5.47939718e-01
-1.22014187e-01 -2.74413943e-01 -5.49586415e-01 1.01236391e+00
7.42487073e-01 -1.50741497e-02 -1.33491838e+00 1.55171716e+00
-2.41325557e-01 -2.44383499e-01 -4.89673465e-02 1.12955248e+00
7.00208724e-01 4.02359396e-01 -1.83289260e-01 -2.97768325e-01
1.01013446e+00 -1.16022742e+00 -3.75861406e-01 -1.77122831e-01
2.00225152e-02 -8.75236630e-01 1.34333599e+00 7.25064635e-01
-1.19152558e+00 -5.99174082e-01 -1.28309703e+00 6.96001202e-03
-3.62418264e-01 3.12846959e-01 2.81067401e-01 1.07500064e+00
-1.54964328e+00 7.09164679e-01 -2.38061711e-01 -5.11440635e-01
3.83991987e-01 5.67899525e-01 -5.22760689e-01 -4.06302422e-01
-6.02755189e-01 1.20921946e+00 -3.73748094e-02 -3.21343631e-01
-9.51649487e-01 -4.53494012e-01 -6.72448993e-01 -1.37877807e-01
2.07046494e-01 -6.78506076e-01 1.35827613e+00 -1.36313069e+00
-1.52719128e+00 9.71082568e-01 -5.67193925e-02 -5.64321019e-02
5.30187130e-01 -1.29730970e-01 -5.03961086e-01 3.15208435e-01
-1.02732934e-01 5.43783128e-01 1.34330606e+00 -1.53909183e+00
-4.35475320e-01 -3.25237483e-01 7.81481341e-02 2.13910982e-01
-5.07853031e-01 3.29700887e-01 -6.12496257e-01 -7.13290393e-01
-1.73774645e-01 -7.75553405e-01 -4.01427746e-02 3.24647576e-01
-1.02966256e-01 -3.16763781e-02 2.33791515e-01 -7.00740337e-01
9.52466130e-01 -2.00323462e+00 2.43473172e-01 3.24737251e-01
1.62067235e-01 5.50622880e-01 -3.24850202e-01 1.82828307e-01
-8.50203186e-02 -5.30483946e-03 -3.58687699e-01 -2.50402123e-01
1.13331705e-01 1.71763971e-02 1.07436227e-02 5.96175671e-01
1.24183692e-01 7.05205381e-01 -8.24812353e-01 -5.53396404e-01
3.91119748e-01 4.40777332e-01 -4.81928796e-01 6.67488694e-01
1.91492453e-01 3.83107781e-01 1.59701869e-01 7.27236688e-01
5.12877703e-01 -2.79864550e-01 -1.71584368e-01 -4.84547526e-01
3.40898111e-02 2.42753297e-01 -1.35392272e+00 1.78026712e+00
-5.87000728e-01 6.46283150e-01 2.01317500e-02 -1.02179122e+00
6.70889080e-01 4.30474371e-01 1.25382543e-01 -8.12496483e-01
1.56213835e-01 4.00261998e-01 -1.90151960e-01 -7.25196600e-01
5.47695793e-02 -2.69662440e-01 2.52622485e-01 6.74036980e-01
5.40615976e-01 -3.72362971e-01 8.22409019e-02 -9.74901021e-02
1.22147834e+00 -6.33055419e-02 3.05800587e-01 -2.42383361e-01
6.47759199e-01 -2.29733437e-01 4.15758252e-01 6.24306202e-01
-3.96222532e-01 8.11322570e-01 5.81910349e-02 -3.13689828e-01
-1.33993936e+00 -1.35046101e+00 -1.30115807e-01 1.06038642e+00
2.10820571e-01 -2.00460151e-01 -9.06392217e-01 -7.76310682e-01
-1.15459688e-01 9.33132917e-02 -4.73357290e-01 5.49793914e-02
-1.66120455e-01 -7.34698594e-01 2.72949517e-01 5.33386350e-01
3.52676272e-01 -9.53146458e-01 -3.87958318e-01 -2.51128018e-01
-8.71872306e-02 -1.03541505e+00 -5.14839113e-01 1.33720070e-01
-7.68094778e-01 -1.20595300e+00 -1.06401658e+00 -9.60588872e-01
1.05839038e+00 2.07397580e-01 1.38129210e+00 -3.23238075e-02
-4.76450115e-01 7.16867328e-01 -2.89391428e-01 -1.15882821e-01
-4.13151324e-01 -3.59724432e-01 2.09882393e-01 -1.71921983e-01
3.14526260e-01 -3.77910912e-01 -7.91969299e-01 5.59960604e-01
-7.85840571e-01 -5.93903720e-01 8.91946971e-01 8.83781135e-01
4.82081294e-01 2.64118224e-01 6.08675539e-01 -4.28414106e-01
8.68607879e-01 1.97337359e-01 -6.35086894e-01 5.22604287e-01
-9.30375576e-01 3.56274724e-01 4.76289988e-01 -6.31191015e-01
-7.59703755e-01 -5.11091053e-02 2.67183244e-01 -8.28795731e-02
-2.46020928e-01 -3.19895223e-02 -2.29188815e-01 -6.39546871e-01
1.08968234e+00 -2.17908219e-01 -5.82068451e-02 -5.13405621e-01
5.97427249e-01 9.78530109e-01 8.44527066e-01 -3.53331029e-01
1.03637528e+00 2.41324738e-01 -2.45275766e-01 -5.44995070e-01
-6.96741760e-01 -7.45355427e-01 -5.79534829e-01 -1.98341429e-01
7.21675098e-01 -9.31413591e-01 -3.97530109e-01 5.44929206e-01
-7.41607606e-01 -4.01816994e-01 -4.70188856e-01 7.14096665e-01
-8.58285725e-01 5.48378348e-01 -5.76834023e-01 -6.71990097e-01
-4.10651118e-01 -1.23522890e+00 7.24733889e-01 1.42621025e-01
-3.91259976e-02 -7.73620605e-01 2.03332528e-01 5.18196940e-01
6.15001976e-01 -2.32718825e-01 7.71058619e-01 -1.92087606e-01
-4.79999542e-01 -1.90649435e-01 -6.29474998e-01 1.03319299e+00
4.76980567e-01 -3.56550187e-01 -1.15192533e+00 -5.22430658e-01
1.39387563e-01 -9.03114021e-01 8.93617868e-01 3.36993873e-01
9.67199504e-01 -2.82664269e-01 3.96046609e-01 7.24325955e-01
1.50980306e+00 -1.53934792e-01 5.81235647e-01 4.07343924e-01
2.55465120e-01 6.67551875e-01 2.48154879e-01 -4.81458195e-02
2.84530938e-01 6.92274988e-01 3.39841157e-01 -4.03370440e-01
-4.78935808e-01 -2.79302872e-03 3.63594383e-01 8.16233933e-01
-3.32534909e-01 4.57102098e-02 -6.59281015e-01 7.08102882e-01
-1.47965777e+00 -8.50440383e-01 4.40893471e-01 2.53308797e+00
1.04392159e+00 -9.72029641e-02 3.55457604e-01 6.06606543e-01
5.97741127e-01 -2.45899871e-01 -6.11325562e-01 -2.91389495e-01
-6.54292107e-02 5.57778895e-01 5.82636535e-01 5.32100976e-01
-1.23780537e+00 6.01153135e-01 7.59399462e+00 4.27680463e-01
-8.49678338e-01 5.03493026e-02 2.55328029e-01 -5.42498343e-02
9.36582163e-02 -1.60153404e-01 -2.60039493e-02 2.12545469e-01
8.49018574e-01 7.81285167e-02 9.00518477e-01 5.02202928e-01
-5.22320569e-02 -2.30533525e-01 -1.25042439e+00 1.37421083e+00
5.25850415e-01 -6.65679872e-01 -2.49437496e-01 -9.62937921e-02
7.87780464e-01 1.09737173e-01 2.87254393e-01 -1.26645848e-01
4.85494465e-01 -1.43287325e+00 5.36566854e-01 5.51247835e-01
1.12945724e+00 -4.99866813e-01 6.56771183e-01 3.10817435e-02
-7.05520213e-01 -3.07171732e-01 -5.41373909e-01 7.50684887e-02
-1.50149271e-01 5.42433321e-01 -5.29421091e-01 1.42420709e-01
8.30165029e-01 5.30182302e-01 -9.47074175e-01 1.69908905e+00
-5.53090870e-01 4.83385414e-01 -4.87694554e-02 2.78928697e-01
-1.47342354e-01 6.29987717e-02 1.04385063e-01 1.15450573e+00
4.30093884e-01 -1.34683326e-01 -1.11155652e-01 5.95596611e-01
-2.18478754e-01 2.30076298e-01 -3.72095287e-01 1.63117260e-01
1.77401051e-01 1.16407871e+00 -3.57966840e-01 -1.62119240e-01
-6.21219575e-01 1.15889478e+00 2.87947983e-01 4.12878215e-01
-2.48048455e-01 -5.16818583e-01 3.30776960e-01 4.07068245e-02
1.21507652e-01 -1.59269363e-01 -5.48925102e-02 -1.18793631e+00
2.12126285e-01 -1.22467804e+00 1.80060074e-01 -1.05004096e+00
-1.67163372e+00 7.98288643e-01 -2.64909416e-01 -1.48779297e+00
-3.22131038e-01 -9.42897320e-01 -2.48919159e-01 1.10106325e+00
-1.98376489e+00 -8.91507626e-01 -5.91778100e-01 9.38305140e-01
1.81196570e-01 -4.06285346e-01 1.07578778e+00 3.93241137e-01
-2.61379443e-02 7.72983193e-01 -8.36329386e-02 2.07562000e-01
1.48859155e+00 -1.59192681e+00 1.59484088e-01 9.39388752e-01
3.02942425e-01 3.41560662e-01 5.64186931e-01 1.85481664e-02
-1.06752884e+00 -6.91347718e-01 8.08888972e-01 -6.17340088e-01
4.60347712e-01 -8.73672292e-02 -7.68615961e-01 8.89339894e-02
3.11637133e-01 3.54332983e-01 9.47572112e-01 7.33627751e-02
-9.19340312e-01 -2.57204384e-01 -1.38377786e+00 2.14955881e-01
9.51216042e-01 -1.00423694e+00 -7.70012021e-01 2.61292934e-01
2.10762784e-01 -6.68936372e-02 -8.04795265e-01 3.96201730e-01
4.91838336e-01 -1.19049668e+00 9.86276329e-01 -3.67475688e-01
2.16667503e-01 -4.24481452e-01 -2.24367499e-01 -1.59028363e+00
-2.60104090e-01 -4.00533497e-01 3.38996127e-02 9.76846874e-01
5.42681396e-01 -2.66053736e-01 5.49372971e-01 6.63858533e-01
2.56565243e-01 -3.50455523e-01 -8.30191731e-01 -1.02181160e+00
1.12780131e-01 -2.82489419e-01 4.07818258e-01 7.55340278e-01
9.22866464e-02 1.43719405e-01 -5.76130211e-01 2.32717052e-01
9.86635268e-01 -4.54550870e-02 6.60417676e-01 -1.30405474e+00
-4.29775923e-01 -4.20822173e-01 -8.79726887e-01 -8.25773776e-01
-1.47748947e-01 -6.52879953e-01 1.22227333e-01 -1.76954138e+00
2.96613276e-01 -4.17393595e-01 -3.85472625e-01 4.43171442e-01
-1.81943208e-01 6.03690624e-01 -1.56874931e-03 3.91506135e-01
-7.34622478e-01 1.78255960e-01 1.13277948e+00 -2.86610484e-01
-1.59713760e-01 2.79838685e-02 -7.35412896e-01 5.98218679e-01
7.35402405e-01 -2.76794791e-01 -5.42376876e-01 -5.91970921e-01
2.73514271e-01 -1.89851493e-01 5.23451626e-01 -1.35250974e+00
4.39400882e-01 1.38524666e-01 6.13916218e-01 1.20433390e-01
2.76760876e-01 -8.20428669e-01 -3.86602759e-01 5.46921231e-02
-3.11400205e-01 -9.98366475e-02 -1.50491074e-01 3.88409585e-01
-2.94180274e-01 -4.44827288e-01 1.16413105e+00 -1.20220877e-01
-6.73575759e-01 1.70847744e-01 9.77995172e-02 2.14604974e-01
6.68077171e-01 -1.65816575e-01 -3.98039788e-01 -5.79506874e-01
-7.67476380e-01 -5.90834022e-02 9.18383539e-01 4.88169044e-01
6.03413999e-01 -1.30759478e+00 -6.51451826e-01 2.96686321e-01
4.50060576e-01 -4.61742729e-01 -2.07541868e-01 3.11125457e-01
-3.56149733e-01 5.89363873e-02 -3.18108112e-01 -4.25470114e-01
-1.26256967e+00 8.12851369e-01 4.83064413e-01 -2.18428262e-02
-4.55441684e-01 7.50741124e-01 -3.48471969e-01 -4.24850881e-01
6.56921685e-01 -1.89263284e-01 -1.78146333e-01 -1.53920367e-01
8.05414140e-01 2.72521824e-01 3.02651316e-01 -6.67556167e-01
-1.88184127e-01 9.43868160e-01 2.45197311e-01 -4.20428932e-01
1.25878501e+00 -8.84365737e-02 -1.28248408e-01 2.65644312e-01
1.21144474e+00 7.04642236e-02 -1.20278490e+00 -3.74776304e-01
6.25779554e-02 -7.35128880e-01 3.35294545e-01 -1.39690948e+00
-1.11032724e+00 8.90101969e-01 1.18582308e+00 -1.14735244e-02
1.60822344e+00 6.72357827e-02 4.05500472e-01 4.06906635e-01
4.78021145e-01 -1.24375832e+00 4.19893950e-01 -7.88102597e-02
8.16902041e-01 -1.64759624e+00 1.40337095e-01 -2.35563871e-02
-5.50844848e-01 9.47746634e-01 2.68293321e-01 -5.47172539e-02
7.04264998e-01 1.22579359e-01 6.53842986e-01 -4.80711758e-02
-2.04081684e-01 -4.35870260e-01 8.18298101e-01 1.03304684e+00
3.22583586e-01 2.45472044e-02 -1.40114293e-01 2.11846709e-01
-1.56654149e-01 1.17805913e-01 3.55573535e-01 7.54537284e-01
-5.23339331e-01 -1.43776345e+00 -4.89382029e-01 4.83106136e-01
-4.81138051e-01 -2.36594975e-01 -4.40544069e-01 2.80547470e-01
2.02937558e-01 1.36132526e+00 -4.41950530e-01 -3.77014220e-01
4.28817421e-01 -7.48077855e-02 8.62973094e-01 -5.90554237e-01
-4.16757047e-01 -1.13087185e-01 -1.76875174e-01 -6.12895250e-01
-9.93344486e-01 -3.69210511e-01 -5.50865293e-01 4.38261665e-02
-2.80521750e-01 4.05291356e-02 7.76597321e-01 7.84687638e-01
-5.86010627e-02 2.50514477e-01 8.04336786e-01 -8.05694640e-01
-6.84625208e-01 -9.66652691e-01 -6.20256126e-01 5.99561810e-01
7.92243242e-01 -5.59301555e-01 -7.10951149e-01 2.67092556e-01] | [11.911709785461426, -1.8090288639068604] |
22c92638-49c9-4a4a-86d6-5e8c5a1db910 | information-theoretic-safe-exploration-with | 2212.04914 | null | https://arxiv.org/abs/2212.04914v1 | https://arxiv.org/pdf/2212.04914v1.pdf | Information-Theoretic Safe Exploration with Gaussian Processes | We consider a sequential decision making task where we are not allowed to evaluate parameters that violate an a priori unknown (safety) constraint. A common approach is to place a Gaussian process prior on the unknown constraint and allow evaluations only in regions that are safe with high probability. Most current methods rely on a discretization of the domain and cannot be directly extended to the continuous case. Moreover, the way in which they exploit regularity assumptions about the constraint introduces an additional critical hyperparameter. In this paper, we propose an information-theoretic safe exploration criterion that directly exploits the GP posterior to identify the most informative safe parameters to evaluate. Our approach is naturally applicable to continuous domains and does not require additional hyperparameters. We theoretically analyze the method and show that we do not violate the safety constraint with high probability and that we explore by learning about the constraint up to arbitrary precision. Empirical evaluations demonstrate improved data-efficiency and scalability. | ['Jan Peters', 'Felix Berkenkamp', 'Julia Vinogradska', 'Carlos E. Luis', 'Alessandro G. Bottero'] | 2022-12-09 | null | null | null | null | ['safe-exploration'] | ['robots'] | [ 3.05675834e-01 2.85176277e-01 -3.47081423e-01 -1.23921834e-01
-8.41882229e-01 -8.84618044e-01 4.32451397e-01 5.49594939e-01
-7.85633981e-01 1.06795740e+00 -3.86079729e-01 -4.12880033e-01
-4.41050351e-01 -8.55112910e-01 -5.89934230e-01 -8.75830173e-01
1.14990652e-01 9.12351310e-01 5.99214673e-01 2.50592023e-01
4.65990245e-01 4.78444606e-01 -1.27168322e+00 -4.42837656e-01
1.16193485e+00 9.67024386e-01 -9.00808871e-02 3.90707910e-01
4.02217537e-01 2.61721900e-03 -5.22163808e-01 -5.45882322e-02
4.42645818e-01 -2.33075902e-01 -7.00536191e-01 1.14379756e-01
-3.15176547e-01 -2.95688242e-01 5.95184267e-01 1.36405563e+00
3.20011020e-01 2.85905242e-01 8.28519464e-01 -1.24100018e+00
5.63812256e-02 1.85255274e-01 -5.44358909e-01 3.66164111e-02
1.14692785e-01 6.21431693e-02 8.28138649e-01 -3.32419008e-01
6.12155259e-01 8.60525370e-01 6.06870592e-01 1.90070316e-01
-1.66228676e+00 -3.41320485e-01 3.07895988e-01 -1.93862140e-01
-1.76090372e+00 -1.94319636e-01 2.77240425e-01 -5.55043161e-01
5.55671334e-01 1.05937310e-01 4.07967120e-01 7.58088768e-01
3.14790487e-01 4.20343071e-01 1.01222718e+00 -5.06750166e-01
1.08925140e+00 2.07648963e-01 1.64290801e-01 5.04012823e-01
7.06841946e-01 2.17078090e-01 -1.61254019e-01 -6.22158945e-01
7.79310644e-01 -3.33694756e-01 -1.98709488e-01 -9.07739639e-01
-9.72972631e-01 1.19376481e+00 -1.98002696e-01 4.57266755e-02
-2.92632669e-01 -2.67725885e-02 1.33208722e-01 9.38450620e-02
1.50647819e-01 5.60919344e-01 -4.03122365e-01 -3.28988463e-01
-8.59663010e-01 4.69326735e-01 1.05009544e+00 9.79748309e-01
6.21229887e-01 -2.82156318e-01 -1.66641399e-01 4.02000338e-01
2.30688706e-01 4.17460471e-01 1.20474689e-03 -1.04186237e+00
1.89816177e-01 1.34411871e-01 8.02523255e-01 -6.23290122e-01
-3.97229046e-01 -4.56664413e-01 -3.73883337e-01 4.97041166e-01
7.35948682e-01 -4.46450323e-01 -8.15264583e-01 1.63824797e+00
5.30019045e-01 -1.80959895e-01 4.23291512e-03 5.99364817e-01
-5.08289695e-01 4.55011040e-01 -1.55022904e-01 -4.80688483e-01
1.12496948e+00 -3.38184297e-01 -5.67803025e-01 3.72011326e-02
5.07039309e-01 -3.96448493e-01 1.08551717e+00 9.19059694e-01
-1.01919115e+00 -4.92634401e-02 -1.27390337e+00 4.09121096e-01
5.61815836e-02 -8.64611268e-02 4.60662425e-01 9.46137667e-01
-6.75172389e-01 5.84224582e-01 -1.10503316e+00 -7.09568784e-02
2.51972795e-01 5.42086244e-01 1.03679141e-02 2.98465133e-01
-8.42953801e-01 9.68404651e-01 8.22249532e-01 -1.06396571e-01
-8.14909697e-01 -5.29432237e-01 -6.34109914e-01 1.21881172e-01
8.59170675e-01 -5.16076326e-01 1.35021055e+00 -4.88732636e-01
-1.59906840e+00 2.17500880e-01 -6.59350753e-02 -6.47178590e-01
9.65712845e-01 -2.41660058e-01 1.94297493e-01 2.29041278e-01
-2.78293695e-02 1.88632175e-01 7.03894973e-01 -1.16486359e+00
-6.41588748e-01 -2.10307956e-01 1.58593625e-01 3.05570632e-01
-9.13380757e-02 -2.70738065e-01 -5.17928600e-01 -3.63168657e-01
1.66228205e-01 -1.13952076e+00 -5.61128616e-01 1.02492228e-01
-6.23703122e-01 -4.78207618e-02 2.63569027e-01 -1.54467002e-01
1.22763813e+00 -1.99941730e+00 -5.14050201e-02 7.69832909e-01
-1.49481758e-01 -7.49998586e-03 3.58746022e-01 4.00406837e-01
5.86948633e-01 3.56950939e-01 -7.05463767e-01 -1.28361538e-01
2.28101507e-01 2.46449560e-01 -3.99502099e-01 8.45942914e-01
9.10280049e-02 3.11757803e-01 -8.55959892e-01 -4.67165649e-01
1.09585011e-02 2.90056199e-01 -8.63434911e-01 -9.26706418e-02
-4.58353400e-01 4.12718207e-01 -8.48021984e-01 1.97591469e-01
6.76227331e-01 -2.12992936e-01 3.84218574e-01 3.23811382e-01
-1.37075126e-01 1.16786137e-01 -1.71443129e+00 1.21679544e+00
-2.57543832e-01 1.29725868e-02 1.13966472e-01 -8.10475051e-01
8.04992497e-01 -5.48354983e-02 4.20392603e-01 -8.92883763e-02
1.85508505e-01 1.68814853e-01 -9.52590182e-02 -8.91964957e-02
2.43518203e-01 -4.52527463e-01 -2.51039952e-01 3.51110905e-01
-2.67665744e-01 -2.73242891e-01 1.81578800e-01 -3.55009466e-01
9.95115340e-01 2.32286543e-01 6.83623552e-01 -6.70850635e-01
1.66255862e-01 -1.57755554e-01 9.42854226e-01 9.91228700e-01
7.72767793e-03 4.99788821e-01 1.13198793e+00 5.25756441e-02
-8.44270825e-01 -1.08364153e+00 -6.09997153e-01 5.02922297e-01
2.71348566e-01 -1.36478052e-01 -6.43588483e-01 -7.45518088e-01
1.89943165e-01 1.05792797e+00 -7.52686918e-01 5.22982478e-02
-1.42984405e-01 -7.63274908e-01 2.13382527e-01 3.75818610e-01
3.25136632e-01 -3.68439227e-01 -1.00966477e+00 2.07598761e-01
1.70807868e-01 -8.87241900e-01 -2.20708802e-01 3.80554706e-01
-8.85371447e-01 -9.83154655e-01 -5.12690842e-01 -2.52449423e-01
7.85887897e-01 -5.49130797e-01 6.22971892e-01 -4.49177682e-01
1.24095149e-01 3.33280325e-01 2.42536678e-03 -5.24287522e-01
-2.91172713e-01 -1.42028332e-01 -2.71432474e-03 -9.18996856e-02
3.10824007e-01 -3.38155866e-01 -2.73613691e-01 4.81282651e-01
-6.82011604e-01 -3.37824106e-01 2.48693988e-01 7.39478171e-01
8.74145865e-01 5.58646381e-01 5.30601442e-01 -9.27473783e-01
6.80886209e-01 -4.63229686e-01 -1.35632479e+00 3.71950597e-01
-6.59969389e-01 6.17368519e-01 3.79927546e-01 -5.52196801e-01
-1.03103054e+00 1.89840928e-01 3.00814658e-01 -2.72472262e-01
-1.79183543e-01 4.38523233e-01 -2.64087617e-01 -4.05544452e-02
5.05397916e-01 -2.12124094e-01 -1.12369455e-01 -4.32845682e-01
3.59211452e-02 2.47150213e-01 3.59515220e-01 -1.09536147e+00
5.70070684e-01 4.84862983e-01 5.66431046e-01 -6.65062606e-01
-5.89302480e-01 -1.68582022e-01 -4.87285584e-01 1.84917033e-01
7.09154427e-01 -4.52946097e-01 -1.10854495e+00 -2.08596706e-01
-8.17426682e-01 -3.29631001e-01 -4.72120881e-01 6.69245780e-01
-9.25530076e-01 6.38730228e-01 -1.27673879e-01 -1.40353453e+00
1.75076261e-01 -1.16778481e+00 9.47030544e-01 2.45405287e-02
-4.41695243e-01 -9.87244308e-01 -3.14350612e-02 -2.47095302e-01
2.20486581e-01 4.62355316e-01 8.10949862e-01 -9.18914318e-01
-5.70748866e-01 -1.76568255e-01 6.41766191e-02 4.77267765e-02
-1.23379700e-01 4.92476933e-02 -4.85764652e-01 -2.74886668e-01
4.00847048e-01 -7.57640302e-02 6.88471556e-01 6.15959942e-01
1.24356067e+00 -5.46003044e-01 -4.29662168e-01 4.44462359e-01
1.39640653e+00 4.80332315e-01 4.38986480e-01 3.23623240e-01
-2.91751865e-02 6.65776134e-01 7.85251737e-01 9.17444825e-01
3.49688344e-02 7.69294500e-01 2.26011932e-01 4.38737631e-01
8.42590094e-01 -2.11332172e-01 1.00594237e-01 -1.83600292e-01
3.79144661e-02 -5.13058901e-01 -1.04376519e+00 5.41685581e-01
-1.93804848e+00 -4.43690628e-01 3.53103012e-01 3.02181578e+00
1.17764831e+00 5.58404386e-01 2.11228967e-01 2.11042508e-01
7.55454123e-01 -5.25881112e-01 -6.11982405e-01 -4.53265518e-01
2.88175553e-01 1.43305287e-01 8.52959692e-01 1.03107321e+00
-1.10074794e+00 5.05811095e-01 6.40465927e+00 8.56318533e-01
-6.72219515e-01 -1.57729313e-01 5.66389322e-01 -1.52341232e-01
-3.95853847e-01 2.74567544e-01 -1.07868063e+00 5.63155472e-01
8.20630729e-01 -3.61865610e-01 1.21351093e-01 8.34330618e-01
1.53647199e-01 -6.64553821e-01 -1.38107884e+00 5.25435805e-01
-5.75991631e-01 -8.50181103e-01 -3.91521424e-01 5.02038777e-01
6.41683877e-01 -5.44737935e-01 1.01751998e-01 -2.42865644e-02
6.12422109e-01 -9.85718429e-01 5.67791641e-01 5.49354792e-01
6.72681868e-01 -1.15291834e+00 6.17736459e-01 5.07499635e-01
-6.42392099e-01 -2.68145561e-01 -3.64409775e-01 3.15512978e-02
3.45112413e-01 7.42434919e-01 -9.48435783e-01 3.28391284e-01
3.01361442e-01 1.27481043e-01 -5.14650717e-02 1.37671471e+00
-3.39466840e-01 5.61640620e-01 -1.02121639e+00 -1.56965092e-01
1.71074480e-01 -3.82660359e-01 6.45531774e-01 9.73830163e-01
5.60846448e-01 4.91369106e-02 3.94210398e-01 9.17627692e-01
4.61343288e-01 -8.07529911e-02 -5.85439920e-01 9.78408977e-02
6.82908893e-01 4.66130942e-01 -1.02812219e+00 -1.10630184e-01
-1.37913108e-01 7.54300952e-01 5.08002564e-02 5.63933611e-01
-6.29159808e-01 -3.41277152e-01 4.22934383e-01 1.24526694e-02
5.90177000e-01 -3.28002095e-01 -7.04693139e-01 -8.56288373e-01
3.22504938e-01 -5.85862219e-01 5.77384174e-01 -6.48955852e-02
-1.15210879e+00 1.52515829e-01 5.43631434e-01 -1.06490719e+00
-5.85460186e-01 -5.18980145e-01 -2.61535019e-01 9.09585893e-01
-1.28200722e+00 -4.66743380e-01 3.77640098e-01 4.00421470e-01
-2.04821322e-02 2.94475734e-01 7.69805074e-01 -3.56098920e-01
-3.63599420e-01 4.33266193e-01 3.11737090e-01 -4.01791334e-01
4.21794444e-01 -1.32352769e+00 -1.35219276e-01 8.22293818e-01
-4.35093045e-01 7.97970176e-01 1.10703349e+00 -8.92418742e-01
-9.86607075e-01 -6.90674901e-01 7.27068365e-01 -3.70563328e-01
6.56748533e-01 -4.26703721e-01 -8.78057957e-01 5.01617014e-01
-2.80458212e-01 -6.89581931e-02 5.38497448e-01 2.98822641e-01
-2.16148987e-01 1.43215498e-02 -1.42552102e+00 5.46530902e-01
5.44114769e-01 -1.20827451e-01 -4.65627819e-01 1.73960075e-01
4.70192999e-01 -1.39720038e-01 -9.57271218e-01 6.41859472e-01
3.91606510e-01 -7.01254845e-01 7.38983512e-01 -4.61956918e-01
-2.23265111e-01 -4.48269337e-01 -2.04952091e-01 -1.01630926e+00
3.28824446e-02 -1.01311731e+00 -1.55432016e-01 9.64115202e-01
4.75422740e-01 -9.86905217e-01 6.30642712e-01 1.13408935e+00
3.40343982e-01 -7.51199305e-01 -1.29142511e+00 -1.15388620e+00
3.05302233e-01 -3.02866250e-01 5.82632482e-01 6.79312646e-01
2.51289278e-01 -7.25587681e-02 -1.63138092e-01 4.81016636e-01
8.99167597e-01 1.19737186e-01 3.68869573e-01 -1.49352074e+00
-7.08165824e-01 -5.15797734e-01 -2.84621239e-01 -1.00198269e+00
-1.03164107e-01 -2.20946386e-01 2.89154440e-01 -1.13774812e+00
1.17998585e-01 -7.82733262e-01 -2.01717541e-01 3.94335657e-01
-1.03744455e-01 -3.54087502e-01 -1.32612348e-01 -1.25000319e-02
-5.26818514e-01 5.11517644e-01 8.58017385e-01 4.51406538e-01
-4.91994828e-01 3.94533902e-01 -6.46689296e-01 8.39277685e-01
8.79877090e-01 -4.16899711e-01 -6.58853233e-01 3.78660932e-02
5.18823326e-01 2.29963079e-01 2.18462154e-01 -7.87556589e-01
7.63865858e-02 -5.48761368e-01 2.36266837e-01 -3.93298894e-01
1.95350498e-01 -8.16098869e-01 1.15672052e-01 3.26049685e-01
-4.63318199e-01 -4.10793394e-01 3.16481650e-01 8.42238188e-01
2.37038240e-01 -6.78486347e-01 9.34263051e-01 2.14689150e-01
-2.63896078e-01 1.69341844e-02 -4.20586795e-01 -3.87311056e-02
1.23089242e+00 -7.55884349e-02 6.48641884e-02 -4.64986622e-01
-8.98187757e-01 3.59540582e-01 8.16677868e-01 -1.11625977e-01
2.64353931e-01 -8.78844738e-01 -4.08853710e-01 1.52613267e-01
-7.20558763e-02 1.53603330e-01 -2.11912975e-01 5.85001230e-01
-2.60956973e-01 3.21790636e-01 1.57287404e-01 -5.86159110e-01
-7.06674278e-01 8.71930778e-01 3.08982253e-01 -3.86028916e-01
-5.16678095e-01 4.71317083e-01 1.66418299e-01 -4.42178547e-02
3.93090874e-01 -4.10382867e-01 1.12416267e-01 1.70352552e-02
4.06630844e-01 4.24461365e-01 -3.88817117e-02 -7.68310130e-02
-3.40102702e-01 4.46167797e-01 1.43231064e-01 -6.40152037e-01
1.02083910e+00 -1.76518351e-01 2.69483387e-01 4.69834954e-01
7.61471570e-01 1.43087298e-01 -1.66548777e+00 -1.40435562e-01
2.97907770e-01 -5.43693900e-01 5.38638048e-02 -8.21875274e-01
-4.47969586e-01 8.02774131e-01 2.74471790e-01 2.15496615e-01
9.87042785e-01 -3.17023128e-01 8.18591714e-02 5.55763602e-01
6.18277133e-01 -1.32658207e+00 -3.79002422e-01 3.47383678e-01
6.62641943e-01 -8.82583141e-01 1.62353814e-01 -4.55190808e-01
-7.53920496e-01 1.07155931e+00 3.72407347e-01 1.97153892e-02
7.75758684e-01 5.98463595e-01 -5.57314754e-01 1.95328981e-01
-6.36063755e-01 -2.27456108e-01 1.89839453e-01 6.00823939e-01
-3.46324295e-02 5.15686907e-02 -5.73093057e-01 5.55488646e-01
1.93490069e-02 1.52230099e-01 5.80564797e-01 1.13430846e+00
-6.49416149e-01 -1.06364822e+00 -4.54360783e-01 2.87501276e-01
-6.07326150e-01 8.55999514e-02 -1.01847753e-01 8.02830458e-01
-1.03128351e-01 1.01855111e+00 2.39515416e-02 1.71875969e-01
9.58964899e-02 1.41248077e-01 4.72016215e-01 -5.72552264e-01
4.92974482e-02 4.70319808e-01 2.76565015e-01 -5.62929034e-01
7.68862292e-02 -9.57109034e-01 -1.19421935e+00 -1.15033828e-01
-4.00058448e-01 4.58301574e-01 5.28009474e-01 9.41079617e-01
2.88105965e-01 1.45606488e-01 3.94626826e-01 -4.02190834e-01
-1.18150711e+00 -4.84843165e-01 -7.34055579e-01 -3.55340093e-01
4.07080352e-01 -9.64091301e-01 -4.72871333e-01 -2.73489863e-01] | [4.965944766998291, 3.0786631107330322] |
e86e741e-d9d5-4820-bb72-1598d1bb1d03 | 3rd-place-solution-for-pvuw2023-vss-track-a | 2306.02291 | null | https://arxiv.org/abs/2306.02291v2 | https://arxiv.org/pdf/2306.02291v2.pdf | 3rd Place Solution for PVUW2023 VSS Track: A Large Model for Semantic Segmentation on VSPW | In this paper, we introduce 3rd place solution for PVUW2023 VSS track. Semantic segmentation is a fundamental task in computer vision with numerous real-world applications. We have explored various image-level visual backbones and segmentation heads to tackle the problem of video semantic segmentation. Through our experimentation, we find that InternImage-H as the backbone and Mask2former as the segmentation head achieves the best performance. In addition, we explore two post-precessing methods: CascadePSP and Segment Anything Model (SAM). Ultimately, our approach obtains 62.60\% and 64.84\% mIoU on the VSPW test set1 and final test set, respectively, securing the third position in the PVUW2023 VSS track. | ['Huchuan Lu', 'Lu Zhang', 'Lihe Zhang', 'Zhenyu Chen', 'Jiawen Zhu', 'Xiaoqi Zhao', 'Ben Kang', 'Zeqi Hao', 'Shijie Chang'] | 2023-06-04 | null | null | null | null | ['video-semantic-segmentation'] | ['computer-vision'] | [ 3.14126700e-01 1.89836502e-01 -3.38727951e-01 -2.30360866e-01
-6.16897821e-01 -6.70893729e-01 2.97581702e-01 -6.32892072e-01
-3.92857462e-01 5.44277489e-01 -3.50704402e-01 -4.98186141e-01
3.70341778e-01 -4.25422817e-01 -7.33749270e-01 -5.30112088e-01
3.25973302e-01 3.26965988e-01 1.20543504e+00 1.67501830e-02
4.62181479e-01 3.50396842e-01 -1.21706378e+00 4.07310814e-01
8.47238898e-01 1.40522814e+00 5.14949441e-01 7.05875933e-01
-2.88825870e-01 2.57625073e-01 -7.41932988e-01 -1.45235091e-01
7.13783801e-01 -4.66196239e-01 -9.53060269e-01 3.76124740e-01
6.99115753e-01 4.15438786e-02 -1.63569868e-01 1.21509075e+00
2.34934255e-01 1.16661601e-02 7.74137318e-01 -1.43260050e+00
4.56369855e-02 3.86339277e-01 -1.32860887e+00 5.35501063e-01
8.33201259e-02 2.16047719e-01 6.71783805e-01 -6.39912188e-01
8.28443706e-01 1.16445982e+00 7.45175183e-01 2.42429405e-01
-7.75476813e-01 -8.42838764e-01 2.10247502e-01 5.63592792e-01
-1.34095716e+00 -2.38650054e-01 6.64982975e-01 -5.21119237e-01
6.01894736e-01 2.72481263e-01 8.80110741e-01 7.02556133e-01
9.13130492e-02 1.38024056e+00 1.20887470e+00 -1.33794606e-01
2.47125346e-02 8.80470127e-02 4.98268962e-01 8.11842322e-01
9.75802168e-02 -2.83136368e-01 -3.15454841e-01 5.51837802e-01
8.81201088e-01 -3.40449661e-01 -3.16829473e-01 2.65182052e-02
-7.44196892e-01 4.57154781e-01 4.36490476e-01 2.68173814e-01
-2.75614392e-02 4.89080884e-02 1.77514657e-01 -3.72432023e-02
4.96314652e-02 -7.09065655e-03 -5.31549096e-01 7.62482584e-02
-1.60435414e+00 1.64964229e-01 4.00931358e-01 1.26859891e+00
4.76458013e-01 1.20263338e-01 -2.65885890e-01 6.05259180e-01
5.05102158e-01 4.56817836e-01 2.05234110e-01 -1.09511971e+00
5.64757228e-01 6.37145042e-01 1.70592129e-01 -8.24368060e-01
-3.86369824e-01 -4.24437225e-01 -5.62337101e-01 2.06625417e-01
3.84580225e-01 -1.69738755e-01 -1.58120227e+00 1.04642200e+00
3.91065001e-01 4.85626280e-01 -1.93972588e-01 1.06587732e+00
1.01916754e+00 1.05426669e+00 1.74749941e-01 -2.43720040e-01
1.46074092e+00 -1.46231258e+00 -6.17458701e-01 -2.83912510e-01
1.90389618e-01 -9.82295930e-01 7.48676062e-01 5.57070494e-01
-1.19652164e+00 -8.95088136e-01 -1.20918512e+00 1.09970920e-01
-1.78575024e-01 1.54874459e-01 3.53457749e-01 4.80346203e-01
-1.18443441e+00 4.11646664e-01 -6.58576190e-01 -4.29791778e-01
6.44253731e-01 9.86728519e-02 2.60708593e-02 2.79911906e-01
-8.11530650e-01 4.65424448e-01 8.85424376e-01 1.72261968e-01
-9.00286913e-01 -5.50449729e-01 -3.77815932e-01 -1.26768366e-01
6.99602723e-01 -4.81557667e-01 1.14903855e+00 -9.60358918e-01
-1.13421547e+00 1.09297359e+00 -5.22484956e-03 -5.72047770e-01
8.36192131e-01 -1.24545008e-01 -3.46220225e-01 3.42270643e-01
6.25537112e-02 9.94296730e-01 7.61224389e-01 -1.43626893e+00
-1.19313288e+00 -2.88341194e-01 -2.99776196e-01 3.61348093e-01
1.60017580e-01 6.39456287e-02 -1.22506464e+00 -6.59692466e-01
4.68783885e-01 -8.54241848e-01 -1.82612121e-01 -1.54743716e-01
-9.80843008e-01 -2.67204076e-01 1.36577785e+00 -1.06306362e+00
1.13409984e+00 -2.24251294e+00 1.09121285e-01 4.52855051e-01
1.20097515e-03 7.95536578e-01 -3.48966429e-03 -4.12995033e-02
5.98376691e-02 2.74191886e-01 -3.86632055e-01 -1.59830540e-01
-2.05640852e-01 -5.91705404e-02 1.96405165e-02 2.79860139e-01
-2.19255343e-01 1.02416909e+00 -2.71038800e-01 -9.42845225e-01
3.88522595e-01 -1.33363549e-02 -4.16081905e-01 8.12408105e-02
-3.42746168e-01 3.02284241e-01 -5.98971605e-01 7.64034033e-01
9.56970215e-01 -1.05778709e-01 2.64642555e-02 -4.51628953e-01
-2.21391842e-01 -4.32422429e-01 -1.45285678e+00 1.82690609e+00
3.85301650e-01 7.17432022e-01 3.51803005e-01 -1.10266888e+00
7.90073276e-01 -8.25934205e-03 5.26866436e-01 -7.13052988e-01
3.80676389e-01 2.24451631e-01 -2.77683079e-01 -7.11513817e-01
5.35578847e-01 1.33012637e-01 1.39598355e-01 -3.83582085e-01
1.65614888e-01 -2.08733112e-01 4.12336022e-01 2.26933569e-01
8.13465118e-01 4.83184218e-01 1.05590381e-01 -6.30792081e-01
6.98445320e-01 3.37020040e-01 6.83701992e-01 7.73275137e-01
-5.40976465e-01 9.50059056e-01 5.07896960e-01 -9.89723727e-02
-7.59943843e-01 -1.17176366e+00 -1.59523726e-01 4.81676728e-01
6.26764596e-01 -7.60834962e-02 -1.19335043e+00 -9.19543624e-01
-2.22290650e-01 5.38921356e-01 -2.35944420e-01 1.70512468e-01
-5.13553858e-01 -4.97527152e-01 4.92535830e-01 6.05597436e-01
9.84317243e-01 -1.17612898e+00 -4.92292017e-01 4.28827927e-02
-2.45902240e-01 -1.50487041e+00 -5.14615417e-01 -9.98369902e-02
-7.58228779e-01 -1.48619843e+00 -1.03574443e+00 -1.25061750e+00
4.13226098e-01 3.09841752e-01 8.45943987e-01 3.23152766e-02
-4.28617626e-01 -1.52708739e-02 -3.00669104e-01 -2.42823124e-01
-2.55000181e-02 6.04016371e-02 -5.20639777e-01 -1.53369799e-01
1.40173454e-02 -2.34142005e-01 -9.54938531e-01 3.52668136e-01
-4.62544441e-01 4.46569711e-01 4.25014943e-01 2.11813867e-01
6.73736155e-01 9.20038894e-02 1.95027396e-01 -8.77837241e-01
9.62724015e-02 -1.08052015e-01 -8.65339816e-01 3.00946295e-01
-2.24268436e-01 -5.57961702e-01 3.17157686e-01 2.17704087e-01
-1.10004473e+00 2.14892715e-01 -3.35094184e-01 -3.23638856e-01
-3.73760700e-01 -1.39222339e-01 -4.53862101e-01 -1.26466736e-01
1.04390889e-01 2.65547246e-01 -2.34757811e-01 -4.98765677e-01
2.30299488e-01 7.95802414e-01 7.06414044e-01 -1.02568731e-01
8.17755163e-01 3.12933892e-01 -2.21426457e-01 -1.02226806e+00
-9.27941859e-01 -6.92911744e-01 -4.64369357e-01 -4.83267128e-01
1.46363127e+00 -8.51265848e-01 -7.01665521e-01 7.87240505e-01
-1.11543405e+00 -3.40799451e-01 -4.76293005e-02 2.61483323e-02
-5.10071993e-01 4.15108413e-01 -5.41598558e-01 -6.63476825e-01
-3.73919874e-01 -1.26170766e+00 1.12467194e+00 7.55232155e-01
-9.02572200e-02 -6.19249642e-01 -4.24726307e-01 8.70590091e-01
-1.42832369e-01 4.29746211e-01 6.32338285e-01 -5.71635962e-01
-7.95063257e-01 1.90151006e-01 -7.57909596e-01 6.29849792e-01
-4.31909651e-01 6.96729049e-02 -7.51882911e-01 -6.71017915e-03
-1.14489011e-01 1.36810899e-01 1.06872368e+00 8.24616373e-01
1.32921708e+00 1.94274485e-01 -4.96027261e-01 8.98406386e-01
1.57070804e+00 1.04732776e+00 8.38689446e-01 3.28469843e-01
9.63123322e-01 4.05563802e-01 7.90287077e-01 7.36942440e-02
2.54153341e-01 7.38564968e-01 2.76493341e-01 -4.78542835e-01
-5.82040310e-01 -2.15215504e-01 2.10359126e-01 6.69809282e-01
-4.06128205e-02 -4.82357830e-01 -9.94977832e-01 4.33663726e-01
-1.86062491e+00 -6.18754387e-01 -7.28843629e-01 1.65591156e+00
3.60788673e-01 4.89624113e-01 2.44927138e-01 1.58588246e-01
8.34841371e-01 2.23411411e-01 -3.79446685e-01 8.88605509e-03
-2.05660731e-01 1.78326011e-01 7.75829136e-01 4.51561034e-01
-1.27368772e+00 1.41500401e+00 6.52972126e+00 1.31566763e+00
-6.61683559e-01 5.49684316e-02 8.51617336e-01 1.54215291e-01
6.05981685e-02 -1.15397885e-01 -1.07709229e+00 7.89845347e-01
3.22443247e-01 3.32053006e-01 5.73385805e-02 4.56550360e-01
3.85037571e-01 -4.75147605e-01 -5.07271171e-01 1.13038027e+00
-1.22999935e-03 -1.30326390e+00 2.39976496e-02 -2.70949066e-01
9.76264119e-01 -2.14818671e-01 -1.10962555e-01 1.09292015e-01
-9.53200310e-02 -8.22027147e-01 1.04418933e+00 2.09242657e-01
5.79689443e-01 -8.36586356e-01 7.95766830e-01 1.54742435e-01
-1.46680701e+00 1.44300610e-01 5.01218848e-02 5.41201234e-01
5.84480941e-01 2.68417835e-01 -6.42061591e-01 5.84110022e-01
9.48742330e-01 7.14677095e-01 -6.08635366e-01 1.27226138e+00
-3.45877856e-01 7.41341352e-01 -3.22342724e-01 2.82310843e-01
5.46379626e-01 -5.08995593e-01 6.68357134e-01 1.29004776e+00
2.30705097e-01 2.74336785e-01 5.41498542e-01 5.12854517e-01
1.51901484e-01 -1.18323244e-01 -7.13484064e-02 7.98522159e-02
1.26205400e-01 1.09059954e+00 -1.45846677e+00 -5.60418069e-01
-2.26762351e-02 1.16061032e+00 -4.22834069e-01 6.59716189e-01
-1.35496354e+00 -2.63584405e-01 3.52589875e-01 2.70699561e-01
5.64514935e-01 -2.10811779e-01 -5.07009685e-01 -7.43254364e-01
-1.33785978e-01 -6.88793778e-01 3.41667891e-01 -8.66863191e-01
-5.21565974e-01 3.77676427e-01 -2.33819634e-02 -8.39406967e-01
4.71927047e-01 -6.88782573e-01 -6.14523172e-01 4.32428777e-01
-1.40826595e+00 -1.08771086e+00 -3.77845019e-01 6.00511789e-01
1.16039968e+00 -2.32900456e-01 -9.25386623e-02 4.85098392e-01
-7.23405361e-01 3.71246338e-01 -6.63578510e-02 8.63069072e-02
8.05671886e-02 -1.06711984e+00 5.35053015e-01 1.20600963e+00
1.05535701e-01 -8.30330700e-02 8.13787639e-01 -7.51907110e-01
-1.00890136e+00 -9.64298487e-01 2.86497384e-01 4.37132120e-02
3.82777423e-01 -1.45211607e-01 -5.93304932e-01 5.94681382e-01
4.60360587e-01 -2.75114059e-01 -1.57680106e-03 -4.32388395e-01
1.99304447e-01 1.31968305e-01 -1.21937585e+00 7.01578438e-01
1.45524800e+00 6.60558557e-03 -5.44246852e-01 1.68980554e-01
6.15913272e-01 -6.03638113e-01 -4.31361109e-01 5.43777883e-01
1.67924136e-01 -9.99720693e-01 1.11926556e+00 -1.83369651e-01
2.52087772e-01 -7.74478912e-01 -3.50876176e-03 -7.07782984e-01
7.70787224e-02 -6.34950817e-01 3.34036164e-02 1.28958917e+00
3.56146842e-01 -3.29694957e-01 1.26288176e+00 1.17148928e-01
-5.31767786e-01 -6.77373052e-01 -8.77827346e-01 -7.17247128e-01
-4.36174035e-01 -4.84688133e-01 -1.55979460e-02 6.10366166e-01
-7.43300498e-01 1.21928766e-01 -2.33666927e-01 9.12589207e-02
7.58884192e-01 1.74974605e-01 6.31835878e-01 -8.92495096e-01
-1.65107831e-01 -5.95558047e-01 -5.98543286e-01 -1.44281387e+00
-1.06321089e-01 -7.35629022e-01 1.58413455e-01 -1.89268231e+00
2.95942664e-01 -1.85937002e-01 -3.65435839e-01 3.43319893e-01
-9.91175920e-02 6.06145442e-01 5.42150617e-01 -1.83378056e-01
-8.43796849e-01 1.86833978e-01 1.58413434e+00 -2.22249120e-01
-1.41269729e-01 -1.95965869e-03 -4.55956489e-01 1.05448055e+00
9.80091870e-01 -2.92289257e-01 -4.65108991e-01 -8.28626379e-02
-5.36946774e-01 7.06312759e-03 1.85171694e-01 -9.88343537e-01
2.45128628e-02 -2.91704297e-01 5.38223863e-01 -1.21179104e+00
2.53432691e-01 -8.92874181e-01 2.69253492e-01 6.68207824e-01
2.08459213e-01 -1.31289527e-01 2.29402676e-01 3.33580494e-01
-2.29808956e-01 -3.41898084e-01 1.05562234e+00 -1.68634489e-01
-1.57681513e+00 2.02896804e-01 -2.11796492e-01 3.60171020e-01
1.56500053e+00 -7.01794505e-01 -1.19740710e-01 2.65387714e-01
-7.50354350e-01 7.05451906e-01 2.46877611e-01 5.84374905e-01
6.04171395e-01 -9.86065686e-01 -4.33944941e-01 1.83902770e-01
-3.72199178e-01 1.29785597e-01 4.88466501e-01 7.79729247e-01
-1.04250574e+00 3.23401153e-01 -3.08691770e-01 -9.48448598e-01
-1.55196500e+00 8.94558653e-02 2.50831187e-01 1.20580710e-01
-7.53033996e-01 1.19204128e+00 3.20614219e-01 1.85803957e-02
3.70727211e-01 -2.23163769e-01 -4.30470645e-01 2.48098478e-01
2.33629331e-01 4.89651859e-01 -2.63163537e-01 -6.57680571e-01
-4.92864609e-01 9.00391221e-01 -1.24410056e-01 -8.28138739e-02
7.99752176e-01 -1.14354931e-01 3.01448882e-01 1.59960866e-01
1.20771551e+00 -4.55452949e-01 -1.57425451e+00 4.31786031e-01
1.42016053e-01 -4.27391052e-01 -5.01437448e-02 -8.54519010e-01
-1.51085293e+00 6.70363545e-01 8.79500687e-01 9.23000202e-02
1.09703076e+00 -1.26713932e-01 1.03431714e+00 -2.25139007e-01
3.70977521e-01 -1.52167618e+00 -1.72998592e-01 2.96857476e-01
6.82821572e-01 -1.07846200e+00 -7.78730512e-02 -8.95581841e-01
-7.48720825e-01 7.85935640e-01 9.21907783e-01 -2.41329312e-01
4.51403081e-01 2.32120171e-01 1.70138299e-01 -9.13938433e-02
-8.24242383e-02 -1.95650622e-01 5.05668104e-01 5.04315436e-01
-4.66085896e-02 5.70331551e-02 -5.47655225e-01 4.58584011e-01
-1.67036057e-01 -1.37152344e-01 2.26478741e-01 7.66143620e-01
-7.65335739e-01 -6.56827331e-01 -4.08861160e-01 3.84261608e-01
-3.04571807e-01 1.00646563e-01 -4.19600457e-01 8.81120622e-01
4.35078025e-01 8.62655520e-01 -7.94961005e-02 -4.40163374e-01
4.91020411e-01 1.84773374e-02 3.90062064e-01 -2.01987103e-01
-6.15685284e-01 4.65025306e-01 4.45811711e-02 -8.20269525e-01
-4.04689491e-01 -6.72721446e-01 -1.70883155e+00 -9.40152556e-02
-2.20829889e-01 1.07184842e-01 6.53790057e-01 1.05852938e+00
1.17647918e-02 8.54785144e-01 2.92247474e-01 -7.83412755e-01
2.96023097e-02 -5.20397842e-01 -7.20036149e-01 3.57609242e-01
-1.38782198e-02 -4.61547554e-01 -4.12406176e-02 3.03499639e-01] | [9.168819427490234, -0.07967841625213623] |
07701bd1-81dd-495d-960d-66835094e7c2 | scene-to-patch-earth-observation-multiple | 2211.08247 | null | https://arxiv.org/abs/2211.08247v1 | https://arxiv.org/pdf/2211.08247v1.pdf | Scene-to-Patch Earth Observation: Multiple Instance Learning for Land Cover Classification | Land cover classification (LCC), and monitoring how land use changes over time, is an important process in climate change mitigation and adaptation. Existing approaches that use machine learning with Earth observation data for LCC rely on fully-annotated and segmented datasets. Creating these datasets requires a large amount of effort, and a lack of suitable datasets has become an obstacle in scaling the use of LCC. In this study, we propose Scene-to-Patch models: an alternative LCC approach utilising Multiple Instance Learning (MIL) that requires only high-level scene labels. This enables much faster development of new datasets whilst still providing segmentation through patch-level predictions, ultimately increasing the accessibility of using LCC for different scenarios. On the DeepGlobe-LCC dataset, our approach outperforms non-MIL baselines on both scene- and patch-level prediction. This work provides the foundation for expanding the use of LCC in climate change mitigation methods for technology, government, and academia. | ['Sarvapali Ramchurn', 'Christine Evers', 'Ying-Jung Deweese', 'Joseph Early'] | 2022-11-15 | null | null | null | null | ['multiple-instance-learning'] | ['methodology'] | [ 4.19055045e-01 -8.77366811e-02 -4.87983167e-01 -5.08658111e-01
-9.25104141e-01 -8.18093657e-01 7.32117772e-01 5.07473409e-01
-2.76836365e-01 6.89273179e-01 1.76955327e-01 -1.03754675e+00
1.54277340e-01 -1.36398280e+00 -8.80892277e-01 -5.60217321e-01
-1.45137936e-01 1.36215284e-01 2.72438437e-01 -1.74516082e-01
3.62612605e-02 4.79321569e-01 -1.64446378e+00 4.05019522e-01
1.22801840e+00 5.04684329e-01 4.13643897e-01 6.71852052e-01
-5.13864100e-01 3.82510781e-01 -5.05113304e-02 2.11795628e-01
4.65298086e-01 -3.29358250e-01 -7.80339241e-01 -1.10437430e-01
7.25126624e-01 -6.36947006e-02 4.89116937e-01 8.14694047e-01
2.48955116e-01 -1.13287725e-01 5.82061827e-01 -9.99229193e-01
2.23228242e-02 1.53967187e-01 -6.14896178e-01 -1.90637276e-01
-1.79660782e-01 1.82140544e-01 8.69463086e-01 -3.64916950e-01
4.32167441e-01 1.18128705e+00 1.11167347e+00 1.86096430e-01
-1.42094409e+00 -6.22417808e-01 5.16810417e-01 8.05407017e-03
-1.19206500e+00 -3.90462279e-01 4.55166042e-01 -7.22569466e-01
1.27963698e+00 7.24191666e-01 8.87480140e-01 2.37348035e-01
1.04303174e-01 6.82254195e-01 1.53691542e+00 -5.72151244e-01
3.09706032e-01 -8.83585438e-02 -5.67088090e-02 3.05574387e-01
2.39083737e-01 2.14578539e-01 2.52736211e-01 3.50784883e-03
4.95176494e-01 2.54133970e-01 -1.73612759e-02 -1.32061094e-01
-9.86587524e-01 7.32523203e-01 9.24356937e-01 1.21956401e-01
-4.33829576e-01 3.66702467e-01 1.09904714e-01 2.07331821e-01
1.01110256e+00 4.13802087e-01 -9.10222769e-01 6.20414764e-02
-1.29834235e+00 3.80263627e-01 4.68373746e-01 7.71273673e-01
1.18018830e+00 -2.13478372e-01 2.25961894e-01 6.56450927e-01
4.99817729e-01 1.03164864e+00 -1.79477766e-01 -8.93891156e-01
5.64974844e-01 1.01909983e+00 1.41199723e-01 -8.62132072e-01
-5.78974068e-01 -5.07900238e-01 -5.82723022e-01 4.22397345e-01
-4.11186926e-02 -2.12792173e-01 -1.31766641e+00 1.40999997e+00
4.54658270e-01 2.27930471e-01 1.01049477e-02 4.64684963e-01
5.12918413e-01 9.77866888e-01 6.79549992e-01 2.38679275e-01
9.67098355e-01 -8.17916572e-01 -4.50555295e-01 -5.36743462e-01
1.13163984e+00 -3.76215845e-01 9.72587049e-01 -3.11786622e-01
-5.11173904e-02 -4.96365339e-01 -8.63700390e-01 2.37269580e-01
-1.12266076e+00 -9.80422497e-02 7.99873948e-01 8.47219765e-01
-9.08868551e-01 4.76097256e-01 -8.35733891e-01 -7.53981769e-01
7.07975626e-01 -1.34565398e-01 -1.06012702e-01 -1.52547881e-01
-1.02560079e+00 1.15131652e+00 4.49970931e-01 1.59450620e-01
-6.54813230e-01 -1.07837844e+00 -9.29366291e-01 -1.38101906e-01
2.42043212e-02 -9.37630236e-02 7.16580451e-01 -7.76473403e-01
-7.52335191e-01 9.10916567e-01 1.72434282e-02 -4.87689555e-01
4.29255933e-01 -2.47196510e-01 -4.23034579e-01 -4.43256110e-01
4.97438371e-01 1.23478174e+00 3.48475665e-01 -1.48875904e+00
-1.09280694e+00 -2.19433308e-01 8.13122019e-02 3.27457845e-01
1.50813505e-01 -2.01411858e-01 -2.87950095e-02 -4.44573194e-01
1.10591628e-01 -1.07586658e+00 -8.14256847e-01 -1.70136429e-02
2.16865935e-03 -1.35139870e-02 1.08469892e+00 -9.56112802e-01
1.24161696e+00 -2.05148220e+00 -4.26070631e-01 2.29374319e-01
-2.30753690e-01 4.44946408e-01 -2.65830129e-01 5.22274852e-01
7.29526654e-02 7.43565559e-01 -7.51700819e-01 -1.08040161e-01
-1.16109036e-01 3.60631913e-01 -2.83540547e-01 4.06272978e-01
3.29720467e-01 9.89410877e-01 -8.42024803e-01 -4.77698624e-01
6.43384755e-01 2.03710645e-01 -4.61019725e-01 -1.02799341e-01
-7.38326967e-01 7.04658687e-01 -3.13665360e-01 7.82956958e-01
1.05483711e+00 2.31332764e-01 5.26340641e-02 2.06416085e-01
-8.14404428e-01 1.13220096e-01 -9.38450575e-01 1.38728809e+00
-6.66294456e-01 7.03171134e-01 -1.15662940e-01 -8.51962447e-01
8.73839080e-01 9.53168273e-02 4.20773655e-01 -9.42284763e-01
-5.46626568e-01 4.52626288e-01 -2.10317418e-01 -4.10025060e-01
1.34271234e-01 -1.06342211e-01 -2.18490362e-01 3.94444972e-01
-4.70465809e-01 -6.08281672e-01 -1.00287162e-01 -3.04972619e-01
5.53645134e-01 5.41349053e-01 2.15394512e-01 -6.86316907e-01
2.45150790e-01 8.16408038e-01 7.05074608e-01 7.29474783e-01
-4.00529563e-01 5.28442979e-01 1.75350472e-01 -8.89396310e-01
-1.20072246e+00 -5.36793053e-01 -6.08406007e-01 1.24089491e+00
-1.62466884e-01 -2.81287450e-02 -6.75024152e-01 -5.81291676e-01
4.23141241e-01 9.64504361e-01 -3.59446049e-01 3.36758196e-01
-4.70089138e-01 -1.27570593e+00 5.29172778e-01 3.75380546e-01
9.18156564e-01 -1.04021752e+00 -7.63674498e-01 3.57829154e-01
-3.91706914e-01 -9.15346086e-01 2.31582388e-01 1.52733684e-01
-1.02495396e+00 -1.05093169e+00 -4.69890684e-01 -5.07689655e-01
5.23929656e-01 5.44059217e-01 1.18116379e+00 2.25944202e-02
-3.34260434e-01 3.05609047e-01 -3.63619864e-01 -9.05302286e-01
-4.70832288e-01 4.67362374e-01 -4.57042456e-01 -2.88099557e-01
2.50503808e-01 -6.04335308e-01 -6.70413673e-01 1.46196380e-01
-8.26729357e-01 6.21366382e-01 6.03435755e-01 1.81816407e-02
6.95712626e-01 1.45662010e-01 6.69299662e-01 -1.10753667e+00
-1.80420294e-01 -5.29996216e-01 -7.93041945e-01 4.59659278e-01
-7.56728292e-01 -2.93328762e-01 3.56846750e-02 1.90114066e-01
-1.19572198e+00 5.00519514e-01 -1.69679925e-01 3.14975262e-01
-7.68476069e-01 9.83040214e-01 -2.85499573e-01 -1.51439786e-01
8.01520288e-01 -1.07765738e-02 -2.45331347e-01 -6.34420991e-01
6.10733032e-01 8.47619593e-01 2.72308916e-01 -3.47514361e-01
8.53982806e-01 5.10748684e-01 1.19716832e-02 -1.00708902e+00
-1.06752658e+00 -8.99413586e-01 -1.05750477e+00 -4.65822309e-01
8.29005539e-01 -1.42917669e+00 3.59743357e-01 6.78135753e-01
-7.63444126e-01 -1.06116104e+00 2.19363086e-02 1.38788208e-01
-2.52148032e-01 -1.09446436e-01 3.00241530e-01 -8.69731009e-01
-3.44314337e-01 -8.21807325e-01 1.26293957e+00 1.47991389e-01
1.18028678e-01 -1.17347097e+00 3.19906175e-01 4.01294231e-01
9.61587667e-01 9.19937968e-01 1.00007689e+00 1.50795892e-01
-8.03054273e-01 -3.65174741e-01 -4.11983311e-01 1.63731039e-01
4.98557270e-01 -1.65269133e-02 -1.12451172e+00 -3.90208513e-01
-4.80056763e-01 -8.56869444e-02 1.13421190e+00 4.68344629e-01
1.07688832e+00 -1.81796566e-01 -4.54695493e-01 6.76428139e-01
1.91495836e+00 4.42047827e-02 8.57929587e-01 7.03854024e-01
8.90575767e-01 8.50515664e-01 7.09486425e-01 -3.27463634e-03
7.44664669e-01 4.40458059e-01 6.37558699e-01 -8.84484172e-01
-2.59863108e-01 -7.59243667e-02 1.40390635e-01 1.44494161e-01
-2.32963324e-01 9.93356630e-02 -1.68726289e+00 8.95849228e-01
-2.04441714e+00 -1.03323543e+00 -6.08852088e-01 2.20369816e+00
9.07288074e-01 -1.80365399e-01 -2.24825829e-01 -9.26345736e-02
4.99585867e-01 4.03519899e-01 -6.24687552e-01 -8.84812474e-02
-2.36904517e-01 1.55949429e-01 1.23073685e+00 5.95472157e-01
-1.85518110e+00 1.32587504e+00 6.16015863e+00 6.05166435e-01
-1.33760011e+00 7.76024610e-02 7.06170797e-01 3.68810236e-01
-3.74844521e-01 3.47277254e-01 -9.49688256e-01 2.52053112e-01
1.15812516e+00 2.69089043e-01 2.25982800e-01 7.19956517e-01
6.31228149e-01 -4.03608114e-01 -3.42346668e-01 4.03111547e-01
-4.84911293e-01 -1.50875449e+00 -2.62018710e-01 1.77413493e-01
1.20520985e+00 6.36455834e-01 -3.19593430e-01 4.81623292e-01
4.31392699e-01 -8.39696169e-01 5.21148205e-01 5.26417255e-01
8.99600804e-01 -3.54569584e-01 7.22723961e-01 5.10300219e-01
-1.52560198e+00 -2.51652710e-02 -2.88410306e-01 -1.74753934e-01
-1.50394633e-01 5.88006258e-01 -5.92407525e-01 5.53364277e-01
9.07256901e-01 8.53812635e-01 -1.01307929e+00 1.29259527e+00
-3.02515894e-01 1.14915121e+00 -5.62286019e-01 4.87079203e-01
4.91746604e-01 -1.64831445e-01 2.87872463e-01 1.58798277e+00
4.14643139e-02 -1.54892445e-01 6.31749451e-01 6.25463367e-01
3.48361224e-01 1.71200469e-01 -7.69413173e-01 2.84436762e-01
4.05894458e-01 1.15798342e+00 -7.58183718e-01 -1.20428048e-01
-5.35544753e-01 6.41758978e-01 -2.54539773e-02 3.44546854e-01
-7.67925739e-01 -1.98243961e-01 5.92350304e-01 2.94539452e-01
1.04647186e-02 -2.56839901e-01 -6.20035052e-01 -7.99103260e-01
-1.91212952e-01 -6.31722271e-01 1.63172573e-01 -3.30959439e-01
-8.47072005e-01 -1.00036457e-01 1.86618298e-01 -1.36055374e+00
2.18184888e-01 -3.90143514e-01 -9.68978286e-01 1.03213608e+00
-2.31347322e+00 -1.99058306e+00 -6.03961587e-01 -3.17152999e-02
4.84040141e-01 2.98947006e-01 1.02467489e+00 1.59438312e-01
-5.54670334e-01 -1.60626620e-01 4.80704576e-01 -1.64019614e-01
5.32345951e-01 -1.51517522e+00 8.64787877e-01 9.54829574e-01
-2.12026075e-01 1.72140211e-01 2.91442871e-01 -8.18091989e-01
-8.15285444e-01 -2.29810452e+00 9.07762766e-01 -3.94751102e-01
2.81864852e-01 -3.52104872e-01 -8.83956909e-01 6.28240049e-01
-1.07952714e-01 -3.14644724e-02 8.24718118e-01 3.13994527e-01
-9.06815603e-02 -4.20841843e-01 -1.18622625e+00 3.07797462e-01
8.34506691e-01 -6.40466154e-01 -9.30099264e-02 4.96045887e-01
7.20005751e-01 -5.22783697e-02 -8.88402998e-01 8.11285317e-01
5.78322530e-01 -5.90071976e-01 8.10034871e-01 -3.62182140e-01
2.79189050e-01 -6.36749804e-01 -3.98713440e-01 -1.23727632e+00
-4.80145782e-01 6.48487732e-02 4.36755896e-01 1.30886471e+00
8.03703070e-01 -6.39655590e-01 7.11547554e-01 6.97120309e-01
-4.88924235e-01 -2.42513105e-01 -7.69371271e-01 -8.02201569e-01
6.93699718e-01 -5.93484640e-01 9.47895110e-01 1.12220383e+00
-6.38953149e-01 -2.11991996e-01 -1.20487191e-01 6.98953092e-01
6.06316686e-01 4.71609473e-01 8.93365026e-01 -1.51939714e+00
4.78659481e-01 -4.16674495e-01 -1.95141301e-01 -4.58308995e-01
1.23636778e-02 -9.24522340e-01 1.66540891e-01 -1.97706497e+00
1.58883855e-01 -1.09379685e+00 -1.97661847e-01 1.12273955e+00
-4.02687877e-01 1.03216633e-01 1.35669887e-01 4.77484256e-01
-3.88357460e-01 4.85219479e-01 7.84459770e-01 -4.30000067e-01
-4.44747627e-01 8.49884301e-02 -3.12641859e-01 5.24028838e-01
1.19821966e+00 -5.61735928e-01 -2.51667928e-02 -5.76845407e-01
2.30261564e-01 -5.99845111e-01 5.83366811e-01 -1.38866293e+00
-2.07210466e-01 -8.02908361e-01 3.82275045e-01 -1.05093467e+00
-3.40057522e-01 -9.65481997e-01 5.56875408e-01 6.01923347e-01
-7.57871345e-02 -4.09067452e-01 6.89813316e-01 4.59982276e-01
3.83922607e-02 1.38345612e-02 8.59462380e-01 -3.57220054e-01
-1.17713380e+00 3.59773010e-01 -4.87050772e-01 -4.79584813e-01
9.33452308e-01 -1.59201503e-01 -3.00401926e-01 8.32367316e-02
-3.54945719e-01 7.18425393e-01 7.80259132e-01 4.09914881e-01
-1.25730425e-01 -1.10244358e+00 -8.87136161e-01 2.24771351e-01
5.79857051e-01 4.26379353e-01 2.05056831e-01 4.85925376e-01
-8.83307099e-01 6.75601900e-01 -1.26322266e-02 -7.23559916e-01
-1.07608712e+00 7.86837786e-02 6.63334072e-01 -2.96828717e-01
-6.62738919e-01 3.73946846e-01 -6.43549860e-03 -1.45243502e+00
-2.43907779e-01 -2.80373693e-01 -2.71389425e-01 1.55191436e-01
1.30254626e-01 1.61684096e-01 1.52637154e-01 -7.37189233e-01
-3.27559620e-01 5.16573608e-01 4.84731168e-01 -3.93266007e-02
1.61493778e+00 -1.36426643e-01 -2.29194462e-01 4.99347270e-01
8.65965009e-01 -3.26048195e-01 -1.43394363e+00 -1.35311848e-02
4.62773651e-01 -4.77106899e-01 6.15013719e-01 -1.26326346e+00
-8.40247571e-01 7.92168021e-01 1.15382159e+00 8.31252113e-02
9.75730538e-01 -1.73583269e-01 5.02160490e-01 4.37133640e-01
5.55076122e-01 -1.06987035e+00 -7.43669271e-01 4.23000157e-01
9.75297868e-01 -1.73359609e+00 2.16001883e-01 -3.05462390e-01
-2.28712723e-01 7.60115981e-01 5.84801733e-01 2.19627440e-01
9.08801019e-01 -6.34595603e-02 3.46405566e-01 -2.77566910e-01
-3.35863680e-01 -7.33393967e-01 2.14659214e-01 7.95727074e-01
4.72686589e-01 8.44173789e-01 7.36487890e-03 -1.36145994e-01
1.59548908e-01 -2.04792172e-01 1.59344390e-01 1.19203639e+00
-8.74056756e-01 -1.30743027e+00 -5.63637257e-01 7.47213364e-01
7.00663310e-03 -4.00617003e-01 -1.91261530e-01 6.46272242e-01
5.67651689e-01 1.02582645e+00 9.85491574e-02 1.09758869e-01
2.67872959e-01 1.39944339e-02 -9.86742228e-02 -7.60933936e-01
-3.68639410e-01 1.49195679e-02 2.49377862e-01 -4.81752753e-01
-8.34217489e-01 -9.59283471e-01 -8.74557793e-01 -2.96125174e-01
-2.81951755e-01 -1.71260789e-01 1.03935194e+00 8.39140892e-01
6.21272206e-01 1.99259251e-01 7.34785736e-01 -1.04037845e+00
2.00789437e-01 -1.15215456e+00 -2.28488594e-01 2.96551138e-01
3.36944044e-01 -4.24058557e-01 -1.45558611e-01 1.58434972e-01] | [9.416248321533203, -1.4368367195129395] |
a2ae576b-6fb1-4f4a-92cd-b86c4ccf21f9 | object-counting-you-only-need-to-look-at-one | 2112.05993 | null | https://arxiv.org/abs/2112.05993v1 | https://arxiv.org/pdf/2112.05993v1.pdf | Object Counting: You Only Need to Look at One | This paper aims to tackle the challenging task of one-shot object counting. Given an image containing novel, previously unseen category objects, the goal of the task is to count all instances in the desired category with only one supporting bounding box example. To this end, we propose a counting model by which you only need to Look At One instance (LaoNet). First, a feature correlation module combines the Self-Attention and Correlative-Attention modules to learn both inner-relations and inter-relations. It enables the network to be robust to the inconsistency of rotations and sizes among different instances. Second, a Scale Aggregation mechanism is designed to help extract features with different scale information. Compared with existing few-shot counting methods, LaoNet achieves state-of-the-art results while learning with a high convergence speed. The code will be available soon. | ['Yabin Wang', 'Xiaopeng Hong', 'Hui Lin'] | 2021-12-11 | null | null | null | null | ['object-counting'] | ['computer-vision'] | [ 5.03892936e-02 -2.74311900e-01 -7.64360428e-02 -4.62352216e-01
-5.81919909e-01 -2.59101510e-01 6.51537836e-01 2.87315190e-01
-7.01725364e-01 4.49487925e-01 -1.69518396e-01 3.33348304e-01
-7.44821727e-02 -8.13239753e-01 -5.54112196e-01 -5.28496265e-01
-8.92064720e-02 5.55751622e-01 6.46042645e-01 1.76829785e-01
4.98471856e-01 3.40993464e-01 -1.80978668e+00 3.10005963e-01
5.76741397e-01 1.09235668e+00 3.08543831e-01 6.79572642e-01
-2.77090490e-01 1.17846215e+00 -5.32391071e-01 -3.81969661e-01
6.75055906e-02 -3.40950608e-01 -7.03960598e-01 1.89276010e-01
7.56899476e-01 -3.63647431e-01 -3.22983623e-01 1.19792032e+00
3.27890098e-01 4.55176651e-01 8.04972708e-01 -1.10991144e+00
-7.03381896e-01 3.96782100e-01 -8.94463420e-01 7.70694673e-01
9.80688184e-02 3.05194557e-01 1.09043312e+00 -1.19431734e+00
3.97490501e-01 1.14906800e+00 4.61347908e-01 6.43152535e-01
-1.02701926e+00 -6.96500778e-01 3.07021827e-01 5.49908638e-01
-1.41001010e+00 -2.08883747e-01 4.96864617e-01 -3.37106019e-01
1.02584684e+00 -6.95951656e-02 9.13519382e-01 7.46404707e-01
-1.97839394e-01 8.30663204e-01 7.46114254e-01 -3.91624004e-01
4.51211065e-01 -1.10914327e-01 4.73056853e-01 8.74178469e-01
5.76033235e-01 -2.33335227e-01 -6.63766205e-01 2.23505974e-01
6.61013067e-01 2.96848089e-01 3.46812546e-01 -4.75495696e-01
-1.12198997e+00 9.20219362e-01 8.38540554e-01 5.21092534e-01
-1.40719712e-01 4.68484789e-01 3.96820188e-01 -1.41143069e-01
5.69788992e-01 4.72944081e-01 -2.48257190e-01 3.32891382e-02
-1.03830492e+00 1.18096627e-01 5.47309220e-01 1.10296488e+00
7.58143544e-01 -2.13805273e-01 -5.21170139e-01 7.57424653e-01
-2.20378369e-01 2.62613714e-01 1.93623200e-01 -8.27960968e-01
4.23119992e-01 7.55996943e-01 1.23332022e-02 -8.51795733e-01
-4.96182948e-01 -3.67961586e-01 -7.73004174e-01 3.57383825e-02
4.47818696e-01 1.92432404e-01 -9.93110716e-01 1.62378144e+00
4.26645339e-01 5.02290249e-01 -4.38789725e-01 9.40077364e-01
8.72038066e-01 3.65767211e-01 1.60748109e-01 -2.38379359e-01
1.60336828e+00 -1.19004703e+00 -5.40874302e-01 -5.45215189e-01
3.89744610e-01 -4.99583900e-01 9.92796779e-01 1.00388631e-01
-1.05929685e+00 -6.77166820e-01 -1.04404271e+00 -3.29238117e-01
-6.22248650e-01 2.80331850e-01 7.12547958e-01 2.67302066e-01
-4.64414060e-01 7.39157379e-01 -8.22915375e-01 -4.25917536e-01
8.55565906e-01 1.37417674e-01 -7.13996291e-02 -2.04416767e-01
-8.17385972e-01 9.39653039e-01 6.22395694e-01 -1.08275950e-01
-7.68795967e-01 -5.64911783e-01 -8.49014223e-01 3.46774727e-01
6.83938861e-01 -7.29595006e-01 1.35969496e+00 -6.31505847e-01
-1.01769936e+00 9.85079706e-01 -2.59310752e-01 -3.44387025e-01
5.84084928e-01 -3.04129958e-01 1.33406557e-02 2.09434688e-01
5.46951354e-01 5.65352499e-01 7.59461164e-01 -1.02676380e+00
-9.11619067e-01 -5.21742284e-01 2.80636866e-02 4.37250212e-02
-4.32117313e-01 1.25328973e-01 -5.55230916e-01 -5.01926184e-01
4.20433059e-02 -5.66974401e-01 -5.03578149e-02 1.33763030e-01
-2.99662620e-01 -5.80599487e-01 6.94784224e-01 -2.04934999e-01
1.18071234e+00 -1.90751529e+00 6.75811768e-02 -2.54518956e-01
4.34281737e-01 1.73294663e-01 -1.44633844e-01 5.07808570e-03
2.59634126e-02 -4.92684133e-02 -1.22407861e-01 -5.27374327e-01
-1.76572636e-01 1.21850021e-01 -7.19847307e-02 5.67594230e-01
6.17414176e-01 8.98301721e-01 -1.26149118e+00 -8.93564284e-01
4.16693985e-01 1.93105921e-01 -3.89277339e-01 1.73653498e-01
-1.35602623e-01 3.86567980e-01 -1.70672834e-01 7.14216828e-01
6.82066023e-01 -6.33184552e-01 -2.22327963e-01 -1.27705514e-01
-1.25732943e-01 -1.69019569e-02 -1.28966355e+00 1.50145733e+00
-4.81800437e-01 3.92422915e-01 -4.34124976e-01 -8.51160884e-01
7.18750238e-01 -1.95078865e-01 2.64869273e-01 -7.59091556e-01
4.19403940e-01 1.03926368e-01 1.14649907e-02 -4.71533805e-01
3.42617571e-01 -1.15550831e-01 -1.21073380e-01 4.53044593e-01
5.37706792e-01 -1.77664444e-01 8.00536513e-01 3.45906675e-01
1.20101941e+00 -1.90620616e-01 6.50056303e-01 -2.87531447e-02
3.62084866e-01 -2.34890833e-01 5.99678993e-01 9.58396018e-01
-2.81687111e-01 7.04768300e-01 4.52943563e-01 -7.57756531e-01
-1.23606777e+00 -9.29388404e-01 -1.11786589e-01 1.45107913e+00
2.80010313e-01 -2.90523648e-01 -6.75988674e-01 -5.41441977e-01
-7.14858482e-03 7.33804226e-01 -1.05912340e+00 -1.92959726e-01
-5.81219375e-01 -4.71894979e-01 3.47664356e-02 1.00584221e+00
7.22363412e-01 -1.06383646e+00 -8.92197251e-01 1.91279888e-01
-9.68824327e-02 -1.16033959e+00 -6.50662065e-01 3.27862352e-01
-6.90774858e-01 -1.39283407e+00 -7.00789690e-01 -7.34373629e-01
7.54580438e-01 5.36824405e-01 1.22774887e+00 3.56934041e-01
-7.30430543e-01 1.01858594e-01 -3.13037068e-01 -5.62196851e-01
2.03354463e-01 1.19078323e-01 -7.83723146e-02 -7.09132552e-02
7.88221955e-01 -3.62316638e-01 -5.80448389e-01 1.96219221e-01
-6.07924104e-01 -1.03277020e-01 4.67826307e-01 9.74671960e-01
5.38136840e-01 -1.41585350e-01 3.76216382e-01 -7.11511493e-01
4.14958507e-01 -3.90997738e-01 -6.49056673e-01 3.16818953e-01
-2.80777901e-01 -8.25175196e-02 7.02711821e-01 -8.02031577e-01
-8.16107333e-01 1.38888001e-01 5.18041551e-01 -5.54686248e-01
-5.55335246e-02 -1.41453056e-03 6.72438592e-02 1.64549038e-01
5.85671663e-01 1.72578052e-01 -5.22332788e-01 -3.34673584e-01
4.27546114e-01 2.46495128e-01 5.27501583e-01 -3.29445988e-01
7.46978343e-01 6.32319152e-01 4.62893993e-02 -6.43855035e-01
-1.68776715e+00 -9.74780798e-01 -1.25021410e+00 -3.07175785e-01
9.22532201e-01 -7.87684321e-01 -7.71776497e-01 4.85451579e-01
-1.32666123e+00 -1.22555368e-01 -4.46789980e-01 2.92487413e-01
-6.54362798e-01 2.31311515e-01 -7.14075565e-01 -1.12051976e+00
-3.34921122e-01 -5.87468803e-01 1.24851203e+00 5.64083934e-01
-3.71684618e-02 -6.00136817e-01 1.03751987e-01 9.34638530e-02
2.05142885e-01 -1.62387826e-02 7.96044827e-01 -7.60499179e-01
-6.48403943e-01 -3.69965196e-01 -6.49650872e-01 2.21995130e-01
-3.05805858e-02 -8.17458984e-03 -8.80269229e-01 -3.10218781e-01
-1.98112056e-01 -5.02753496e-01 1.32311165e+00 2.82510012e-01
1.41018987e+00 -1.57501608e-01 -2.56489158e-01 4.55615848e-01
1.30559242e+00 -1.14320733e-01 3.28488201e-01 1.93863004e-01
5.96035004e-01 3.99995297e-01 7.12719083e-01 6.23477161e-01
2.64164984e-01 4.09068257e-01 4.87257093e-01 1.54577807e-01
-1.90004379e-01 -3.24984789e-01 -4.18228030e-01 7.08546162e-01
-2.43737489e-01 2.78009009e-03 -7.75192857e-01 6.61565542e-01
-1.97255003e+00 -1.44773686e+00 -1.47519661e-02 2.17591238e+00
4.13432866e-01 3.50991786e-01 2.40662619e-01 7.19308332e-02
9.30888772e-01 2.46713609e-01 -9.29151893e-01 -4.14711423e-03
7.89516419e-02 1.10940032e-01 3.58232826e-01 1.05298549e-01
-1.29699886e+00 9.59117413e-01 6.12192822e+00 7.76127219e-01
-5.28911233e-01 2.63032436e-01 6.84027076e-01 -2.79767722e-01
4.36398149e-01 -8.87664780e-02 -9.33392942e-01 4.96382445e-01
3.51526588e-01 -2.71889985e-01 4.17933494e-01 1.11384249e+00
-2.54041880e-01 -4.88851100e-01 -1.15810061e+00 8.85273874e-01
3.92164648e-01 -1.12452316e+00 -9.21646580e-02 -3.64484370e-01
7.06868529e-01 -7.50768259e-02 -2.21101552e-01 6.48422062e-01
2.44264916e-01 -8.30001235e-01 6.91729426e-01 6.94621205e-01
8.16094041e-01 -8.52101326e-01 7.84124494e-01 6.08484089e-01
-1.54859829e+00 -5.18957019e-01 -7.66346693e-01 -2.06917107e-01
1.41233623e-01 7.04656303e-01 -5.18394411e-01 7.08534867e-02
8.14572155e-01 7.10209012e-01 -7.82545805e-01 1.48086596e+00
-4.14174110e-01 1.48455724e-01 -2.00066462e-01 -4.18062299e-01
4.94445376e-02 -4.44250442e-02 2.37996563e-01 1.29370129e+00
3.13924551e-01 2.61137903e-01 2.32206464e-01 9.78498459e-01
-2.02853411e-01 2.21264698e-02 -4.99665439e-01 1.48283914e-01
4.95599806e-01 1.41237962e+00 -1.08715928e+00 -8.15070987e-01
-5.28824091e-01 9.76517856e-01 8.98782551e-01 -4.77499105e-02
-1.00916076e+00 -4.71250743e-01 2.76892483e-01 1.08601116e-01
5.61149657e-01 -1.29851624e-01 -3.09347123e-01 -1.24786007e+00
2.45788470e-02 -2.34570578e-01 6.22463703e-01 -8.98652971e-01
-1.69854319e+00 3.61169159e-01 5.44695929e-02 -1.19289446e+00
1.27117321e-01 -7.17916369e-01 -9.61295247e-01 4.64184344e-01
-1.38095725e+00 -1.08647752e+00 -7.16069341e-01 4.00679439e-01
8.63966584e-01 -9.59555507e-02 5.73761523e-01 1.41878784e-01
-5.48879921e-01 4.25421715e-01 -1.96559712e-01 4.00929481e-01
5.37804365e-01 -1.29940295e+00 4.10738975e-01 8.14461470e-01
3.44147027e-01 5.93045354e-01 4.44123268e-01 -7.92373478e-01
-7.87250638e-01 -1.19731128e+00 9.30607378e-01 -5.42539477e-01
7.91904807e-01 -4.98613447e-01 -1.04235315e+00 4.27350760e-01
-1.98425204e-01 6.95284903e-01 3.41138214e-01 1.51192829e-01
-8.12661648e-01 -3.23150237e-03 -9.52569127e-01 2.77423412e-01
1.31809676e+00 -4.68385756e-01 -6.91787899e-01 4.93496150e-01
5.58788955e-01 -2.87895739e-01 -4.52721596e-01 1.86624497e-01
3.75407457e-01 -9.72955585e-01 8.92984867e-01 -7.68538833e-01
6.35344207e-01 -2.68641055e-01 9.23895389e-02 -1.01032543e+00
-7.07470119e-01 -9.67644528e-02 -5.44037759e-01 9.97460485e-01
1.25035286e-01 -2.37939492e-01 7.36101687e-01 3.65036845e-01
2.31059030e-01 -6.42750978e-01 -1.16589010e+00 -9.76212323e-01
6.69400617e-02 -1.67630613e-01 4.64508533e-01 7.27142751e-01
3.25193554e-02 7.70088077e-01 -4.07546878e-01 -1.54490648e-02
9.06798244e-01 2.37609357e-01 6.48222446e-01 -1.49067175e+00
-9.91434827e-02 -4.23943311e-01 -5.58456600e-01 -8.70246053e-01
-3.65108959e-02 -7.06180573e-01 9.35027823e-02 -1.49269986e+00
1.09028196e+00 3.45062092e-02 -2.84139723e-01 5.00144899e-01
-5.00556171e-01 2.88776815e-01 5.31889141e-01 1.07259519e-01
-1.33691382e+00 5.02060652e-01 1.23622787e+00 -2.19146729e-01
4.05113101e-02 1.08166501e-01 -4.52403158e-01 8.05939257e-01
8.23919833e-01 -7.19142973e-01 6.58636168e-02 -3.16074401e-01
1.66739419e-01 -1.80363625e-01 5.11834025e-01 -1.32547462e+00
5.75164437e-01 -1.45289540e-01 7.37266243e-01 -9.77083921e-01
2.82562137e-01 -6.51802659e-01 -5.60783267e-01 4.55531925e-01
-2.97627538e-01 -1.29854888e-01 -6.98895454e-02 8.11536372e-01
-7.65890181e-02 -5.10633647e-01 1.11627495e+00 -5.35383940e-01
-7.61728168e-01 5.61973810e-01 8.74938723e-03 2.39526436e-01
1.12360227e+00 -2.64594197e-01 -6.05176389e-01 -2.07468700e-02
-6.71236992e-01 3.35089773e-01 2.81862378e-01 3.83339226e-01
5.49167156e-01 -1.48080885e+00 -5.14989793e-01 -2.45123189e-02
5.16248286e-01 2.60828614e-01 3.53021920e-01 7.71683276e-01
-9.83593389e-02 2.98766494e-01 -1.40535846e-01 -5.36955893e-01
-1.18253517e+00 1.09004486e+00 2.69898415e-01 -2.63425797e-01
-5.72764516e-01 1.00701177e+00 8.55539665e-02 -2.94555187e-01
3.82787734e-01 -1.78445548e-01 -3.43332857e-01 3.86425376e-01
1.08846009e+00 6.02679908e-01 -6.82491288e-02 -4.85605657e-01
-4.00740087e-01 7.18602657e-01 -1.41117513e-01 2.52636254e-01
1.35345995e+00 1.88816898e-02 8.45840294e-03 8.58642519e-01
1.19575179e+00 -4.52460766e-01 -1.40127790e+00 -4.65232253e-01
2.20766470e-01 -6.89708114e-01 -2.03626201e-01 -5.52204609e-01
-1.02738035e+00 1.19192159e+00 6.14395618e-01 1.40492067e-01
8.44783068e-01 2.50475854e-01 4.49330568e-01 5.66570759e-01
4.26221400e-01 -1.50106478e+00 8.01217973e-01 8.28136623e-01
6.60571396e-01 -1.56273198e+00 2.45892152e-01 -1.90819964e-01
-4.96210665e-01 1.28972411e+00 1.07458460e+00 -3.01169604e-01
4.74448770e-01 2.11039692e-01 -5.41455925e-01 -4.04449105e-01
-7.21267104e-01 -7.66813695e-01 2.92416781e-01 5.53151250e-01
3.10768872e-01 -1.71013534e-01 -1.25505432e-01 4.77432251e-01
1.75704330e-01 1.65507570e-02 3.28244179e-01 7.98367202e-01
-8.70489836e-01 -3.31955552e-01 -1.86808214e-01 7.05075204e-01
-1.80832192e-01 -8.51065740e-02 -3.52339029e-01 8.04329038e-01
2.48310372e-01 7.67608821e-01 4.65265930e-01 -8.44227299e-02
4.25372064e-01 3.27028669e-02 5.12653649e-01 -9.80257809e-01
-2.83100873e-01 -9.72624421e-02 -3.31014454e-01 -5.46577096e-01
-4.94867384e-01 -5.68962157e-01 -1.20652032e+00 -2.59735793e-01
-7.41654038e-01 -3.36395621e-01 3.24618667e-01 8.73168945e-01
2.93121208e-02 7.22679615e-01 5.75371027e-01 -1.02210796e+00
-6.77719712e-01 -1.10870719e+00 -7.22448528e-01 5.19776344e-01
2.40175694e-01 -9.88397419e-01 -4.90979254e-01 -2.10017920e-01] | [9.005268096923828, 0.5458279252052307] |
d2ad3a21-a106-46dd-98fb-fc637004d409 | a-practical-guide-to-multi-objective | 2103.09568 | null | https://arxiv.org/abs/2103.09568v1 | https://arxiv.org/pdf/2103.09568v1.pdf | A Practical Guide to Multi-Objective Reinforcement Learning and Planning | Real-world decision-making tasks are generally complex, requiring trade-offs between multiple, often conflicting, objectives. Despite this, the majority of research in reinforcement learning and decision-theoretic planning either assumes only a single objective, or that multiple objectives can be adequately handled via a simple linear combination. Such approaches may oversimplify the underlying problem and hence produce suboptimal results. This paper serves as a guide to the application of multi-objective methods to difficult problems, and is aimed at researchers who are already familiar with single-objective reinforcement learning and planning methods who wish to adopt a multi-objective perspective on their research, as well as practitioners who encounter multi-objective decision problems in practice. It identifies the factors that may influence the nature of the desired solution, and illustrates by example how these influence the design of multi-objective decision-making systems for complex problems. | ['Diederik M. Roijers', 'Peter Vamplew', 'Marcello Restelli', 'Gabriel Ramos', 'Ann Nowé', 'Patrick Mannion', 'Athirai A. Irissappane', 'Enda Howley', 'Fredrik Heintz', 'Richard Dazeley', 'Luisa M. Zintgraf', 'Timothy Verstraeten', 'Mathieu Reymond', 'Matthew Macfarlane', 'Johan Källström', 'Eugenio Bargiacchi', 'Roxana Rădulescu', 'Conor F. Hayes'] | 2021-03-17 | null | null | null | null | ['multi-objective-reinforcement-learning'] | ['methodology'] | [ 2.77788609e-01 2.40508318e-01 -6.24309063e-01 -2.79384911e-01
-7.73018241e-01 -4.41829085e-01 3.98009032e-01 4.49530244e-01
-7.83429563e-01 1.05841470e+00 2.78209358e-01 -6.40197575e-01
-6.97875917e-01 -6.39255822e-01 -7.43078589e-02 -6.19947016e-01
7.96187967e-02 8.78237009e-01 -2.19725780e-02 -4.25205201e-01
7.06667662e-01 2.09588200e-01 -1.39828873e+00 -2.83617526e-01
8.04487944e-01 6.73280180e-01 3.63070607e-01 7.03489900e-01
2.17278134e-02 8.53288233e-01 -7.18468904e-01 -7.28253424e-02
1.28197595e-01 -3.53901714e-01 -9.63509619e-01 4.57208633e-01
-3.02004039e-01 -5.52106321e-01 2.82939136e-01 7.91410983e-01
5.88822365e-01 4.85354602e-01 7.13978052e-01 -1.29843247e+00
-3.86427194e-01 4.17613477e-01 -2.82344222e-01 2.78755575e-01
5.26827812e-01 6.43727839e-01 1.06347048e+00 -1.20514959e-01
2.19451621e-01 1.57890940e+00 1.97252750e-01 3.77666354e-01
-1.39201307e+00 -1.71998218e-02 4.40250754e-01 1.68260232e-01
-1.09902179e+00 -5.30716360e-01 5.55248320e-01 -3.82511705e-01
9.89201605e-01 -6.24039806e-02 7.34689534e-01 4.15364534e-01
7.39330053e-01 6.93571568e-01 1.38167191e+00 -6.71768665e-01
5.78901052e-01 -9.06499848e-02 -3.01612765e-01 3.16498727e-01
2.44900897e-01 6.13081038e-01 -1.28045648e-01 -4.35766298e-03
5.55466712e-01 -1.50049582e-01 1.46832481e-01 -4.38315839e-01
-1.08717430e+00 1.19676542e+00 6.77286759e-02 3.77879262e-01
-7.99854040e-01 1.67134717e-01 1.17565744e-01 5.71859181e-01
-1.87854424e-01 1.00098991e+00 -2.02340081e-01 -3.26904953e-01
-6.43852830e-01 8.07124197e-01 8.15441906e-01 4.71662432e-01
6.80400074e-01 3.94231796e-01 -1.19367883e-01 6.10833406e-01
6.23506546e-01 1.95925772e-01 2.04995140e-01 -1.45139146e+00
4.01973128e-01 2.19858080e-01 7.98356950e-01 -7.10795939e-01
-5.03877878e-01 -1.54919475e-01 1.15180537e-01 1.06963730e+00
4.89792734e-01 -6.74022079e-01 -7.50286639e-01 1.32999706e+00
2.04305828e-01 -8.71630251e-01 3.43535393e-01 9.59214449e-01
-4.33413237e-02 6.01941049e-01 3.01039517e-01 -5.82520247e-01
1.00885403e+00 -8.61491919e-01 -9.31663156e-01 -8.49147797e-01
3.61781001e-01 -7.95117319e-01 9.61901665e-01 3.59943986e-01
-1.52709925e+00 -1.31701007e-01 -1.02988470e+00 4.05374974e-01
-4.43043336e-02 -5.80309451e-01 6.50430739e-01 6.39385462e-01
-8.27650845e-01 6.06922030e-01 -7.37603962e-01 -2.90741831e-01
-1.17257480e-02 4.65815574e-01 5.29998004e-01 -1.96553841e-01
-1.08255613e+00 1.65798342e+00 5.92132211e-01 1.56086773e-01
-1.02858758e+00 -2.35100925e-01 -8.16196978e-01 1.37161404e-01
1.06289983e+00 -7.34810829e-01 1.91831887e+00 -1.22475934e+00
-1.64713597e+00 8.46882537e-02 1.31874174e-01 -1.91774149e-03
3.86779964e-01 2.11721554e-01 -2.86679655e-01 -9.09072980e-02
2.24279940e-01 5.94073236e-01 6.96704447e-01 -1.21496534e+00
-1.08550549e+00 -1.95168987e-01 6.89423978e-01 1.01728046e+00
2.95554429e-01 2.15684667e-01 4.77919698e-01 -3.01674753e-01
-1.77646741e-01 -6.30948603e-01 -8.65283012e-01 -4.86355424e-01
7.87991360e-02 -2.59964347e-01 2.86322206e-01 -1.27354875e-01
1.29851830e+00 -1.67366135e+00 2.98695624e-01 4.54428457e-02
7.45319054e-02 -2.35162303e-01 -1.42010778e-01 9.54071283e-01
3.14740896e-01 1.06821708e-01 -9.42652225e-02 1.23654276e-01
3.17517459e-01 4.93757725e-01 1.73468590e-01 3.59819114e-01
3.20152640e-01 6.27193213e-01 -1.18841696e+00 -5.77833056e-01
4.14324492e-01 -3.49582165e-01 -4.88469988e-01 2.01488584e-01
-4.47835565e-01 1.72746330e-01 -9.22616601e-01 9.49720442e-01
-9.06361118e-02 1.01571925e-01 4.81572032e-01 6.04889512e-01
-4.41672415e-01 5.24486005e-01 -1.34300995e+00 1.07547498e+00
-3.93838704e-01 3.53338093e-01 3.08590591e-01 -1.18628359e+00
5.56216180e-01 4.88093942e-01 6.16676867e-01 -7.44974077e-01
3.15623760e-01 2.76934832e-01 5.43777883e-01 -4.92514610e-01
7.01090395e-01 -9.63928401e-01 3.89936939e-02 7.78412044e-01
-2.74001628e-01 -6.07380867e-01 6.12310648e-01 -4.75976944e-01
8.64489496e-01 2.65570953e-02 6.29289806e-01 -4.26216632e-01
1.22029908e-01 4.95822042e-01 6.85952723e-01 8.81250918e-01
-5.05730033e-01 -2.44060438e-02 6.77640200e-01 -4.13864613e-01
-8.68931234e-01 -6.51732028e-01 6.92458674e-02 1.10151470e+00
2.72410691e-01 1.81501105e-01 -2.84509450e-01 -4.22705173e-01
1.40194789e-01 1.22645152e+00 -3.40301365e-01 4.41232957e-02
-3.37167710e-01 -6.70981467e-01 -8.89405161e-02 3.07693690e-01
-2.86803246e-02 -1.18008947e+00 -1.15097606e+00 7.52798676e-01
1.87717211e-02 -6.48553610e-01 -2.60586947e-01 3.05361480e-01
-8.59679461e-01 -1.09384906e+00 -7.05876350e-01 -5.35424769e-01
5.49675465e-01 3.55655015e-01 9.88842368e-01 3.79379503e-02
1.44351020e-01 7.49517441e-01 -2.03492403e-01 -7.50568211e-01
-6.66433334e-01 -1.37743101e-01 7.87796453e-02 -4.52209473e-01
3.69824111e-01 -2.33365670e-01 -2.99582094e-01 3.94980133e-01
-9.41473722e-01 -2.04705432e-01 6.92271411e-01 9.17941689e-01
2.61011899e-01 4.74546909e-01 8.91474128e-01 -3.35785568e-01
1.40238965e+00 -4.34432209e-01 -7.93595552e-01 4.74049300e-01
-1.04327703e+00 1.25956714e-01 2.87883312e-01 -5.75569510e-01
-1.14721918e+00 -3.67898345e-01 2.12846324e-01 3.48781347e-02
-2.05128849e-01 1.06229639e+00 -4.75463122e-02 -1.51808366e-01
5.91901720e-01 3.32273846e-03 3.26368749e-01 1.75805956e-01
-1.25678793e-01 4.16279584e-01 -3.89820546e-01 -7.78878808e-01
4.60704923e-01 -2.72312284e-01 1.01926610e-01 -5.58639288e-01
-5.11630297e-01 -2.41879057e-02 -2.04748198e-01 -5.70170820e-01
6.67554200e-01 -4.94348556e-01 -8.01457465e-01 1.33178577e-01
-5.40970564e-01 -7.21282244e-01 -3.69415879e-01 6.63338840e-01
-1.19238019e+00 -5.69414757e-02 -1.15115643e-01 -1.13053775e+00
3.28764617e-01 -1.61362755e+00 4.40524966e-01 4.60905552e-01
-5.64551830e-01 -1.23859835e+00 2.73470469e-02 5.22841930e-01
5.03656209e-01 2.33411178e-01 1.17416716e+00 -3.09995294e-01
-4.91237998e-01 6.71867207e-02 3.89620394e-01 -9.90932062e-02
2.61262715e-01 2.37418581e-02 -2.40245208e-01 -3.84020716e-01
7.00887740e-02 -7.38184690e-01 9.11115482e-02 7.57143259e-01
3.32767099e-01 -4.31779981e-01 7.80914351e-02 -3.65856528e-01
1.58768356e+00 9.57737029e-01 2.42568970e-01 9.25694287e-01
-1.01742655e-01 9.68465030e-01 1.29493999e+00 7.52804399e-01
7.85168409e-01 5.21391928e-01 6.26431465e-01 9.97559354e-02
6.21612489e-01 3.82779092e-02 3.51570696e-01 2.56012946e-01
-1.47455826e-01 -3.43273729e-01 -1.15161324e+00 6.50962651e-01
-2.01637244e+00 -1.10355711e+00 6.08670413e-01 2.16570973e+00
6.03729844e-01 3.11453283e-01 4.85500932e-01 1.21474482e-01
7.54411876e-01 3.25092405e-01 -8.52300286e-01 -1.04759324e+00
3.56633693e-01 -2.23379180e-01 6.02037728e-01 8.65895987e-01
-8.31286609e-01 6.26637638e-01 8.01434994e+00 3.89561921e-01
-8.70663047e-01 -2.87897497e-01 5.54330289e-01 -2.52358079e-01
-4.88046199e-01 2.67472833e-01 -4.22864407e-01 1.64312020e-01
9.17311609e-01 -8.09597969e-01 8.18668187e-01 5.88544965e-01
7.77505696e-01 -7.48490274e-01 -1.10905063e+00 2.86527663e-01
-4.90881562e-01 -9.40751612e-01 -3.77670974e-01 3.39028925e-01
7.60447085e-01 -4.37054247e-01 -2.40072906e-02 3.74449223e-01
9.21531081e-01 -1.21287286e+00 9.85671282e-01 9.71867740e-02
7.37737566e-02 -1.03182876e+00 6.99492395e-01 6.40324473e-01
-7.02215493e-01 -7.05351233e-01 -9.21063647e-02 -7.91285574e-01
4.97277349e-01 -1.67345733e-03 -7.64349520e-01 3.12922299e-01
2.72015393e-01 1.64801180e-01 2.18140170e-01 1.04776525e+00
-1.15056440e-01 2.28724256e-01 -1.65188804e-01 -5.34077466e-01
9.37041461e-01 -3.71961683e-01 4.98923808e-01 5.65908194e-01
1.26524970e-01 4.13459897e-01 5.10243237e-01 5.71500957e-01
7.87307799e-01 -6.77410327e-03 -5.67847311e-01 -4.67803776e-01
4.72704381e-01 7.10427642e-01 -7.16820955e-01 2.73611825e-02
-4.04074401e-01 -5.18155955e-02 1.08155727e-01 4.85640883e-01
-5.06138027e-01 -1.88358739e-01 8.96017253e-01 4.12097313e-02
6.84763342e-02 -5.53528368e-01 -3.30531478e-01 -6.38346076e-01
-2.15262607e-01 -1.42145717e+00 3.12665403e-01 -4.97367382e-01
-8.18884909e-01 5.71605098e-03 3.32967788e-01 -1.01155472e+00
-5.88113785e-01 -3.56812894e-01 -7.19191730e-01 8.28764379e-01
-1.67286336e+00 -6.08053565e-01 4.25547600e-01 1.75678551e-01
8.38717520e-01 2.18407673e-04 5.50709486e-01 -1.89880624e-01
-5.07444859e-01 -9.75273550e-02 2.48169273e-01 -5.50588369e-01
3.89324754e-01 -1.15832603e+00 -3.12775254e-01 5.35225689e-01
-5.63815236e-01 2.39144415e-01 9.85129118e-01 -5.89787424e-01
-1.65071678e+00 -3.77938181e-01 7.83313930e-01 -1.21099325e-02
6.38651490e-01 6.27706349e-01 -4.27911371e-01 4.82632190e-01
3.87667775e-01 -9.42645967e-01 6.24157906e-01 2.70944983e-01
6.89553320e-01 5.00843376e-02 -1.41383719e+00 9.53626096e-01
3.66612941e-01 -1.47059023e-01 -8.73609364e-01 2.21346766e-01
2.74533212e-01 -3.82953256e-01 -9.27695334e-01 1.90521285e-01
4.69946384e-01 -7.03234732e-01 7.11427748e-01 -7.79860437e-01
5.06492376e-01 -1.89634427e-01 -1.73183292e-01 -1.68210447e+00
-7.30321169e-01 -6.03232801e-01 2.84359723e-01 6.96548700e-01
5.31774163e-01 -6.42288923e-01 6.07782423e-01 1.15169013e+00
-1.27562180e-01 -8.82514238e-01 -9.23679709e-01 -7.38504767e-01
2.11031020e-01 -1.63047418e-01 4.88124967e-01 8.05663109e-01
1.41770214e-01 2.17956066e-01 -2.76519746e-01 5.69447391e-02
6.85123742e-01 7.24088401e-04 3.13464850e-01 -9.11447167e-01
-1.72310695e-01 -7.57346094e-01 -8.39610621e-02 -7.92852759e-01
-1.97598591e-01 2.10958011e-02 3.24765325e-01 -1.81518757e+00
-4.18711483e-01 -6.43947959e-01 -2.71853179e-01 3.85186702e-01
1.08068530e-02 -7.25083709e-01 4.66663539e-01 8.24399665e-02
-4.56270993e-01 6.72186494e-01 1.58363259e+00 -8.51287171e-02
-5.66898644e-01 3.96103472e-01 -1.16562033e+00 8.77632558e-01
9.79628503e-01 -5.13509035e-01 -7.00894594e-01 -4.96437520e-01
2.81278580e-01 8.56522202e-01 -2.14090571e-01 -6.29083395e-01
3.01583081e-01 -1.58237886e+00 -2.78660432e-02 -2.17611536e-01
2.87639737e-01 -1.00631857e+00 3.64465751e-02 7.18445480e-01
-5.38089752e-01 5.44971526e-01 1.50904074e-01 3.49692285e-01
-3.14717412e-01 -9.06479418e-01 7.56297350e-01 -3.83303732e-01
-8.06017280e-01 -8.25238675e-02 -1.06474233e+00 3.57807316e-02
1.37866890e+00 -3.56785297e-01 9.59147140e-03 -6.54564917e-01
-7.08038092e-01 8.95813704e-01 4.59614277e-01 2.41731465e-01
5.23959041e-01 -1.03165996e+00 -5.98986208e-01 -5.27197897e-01
-2.39903152e-01 -1.02605470e-01 -1.13737341e-02 6.53394759e-01
-5.73927343e-01 6.44215167e-01 -5.61774671e-01 -9.32297483e-02
-9.18768167e-01 4.82153535e-01 5.46780169e-01 -4.47329432e-01
-9.77805927e-02 1.47193491e-01 -2.26954356e-01 -2.01511607e-01
2.31600538e-01 -1.35706663e-02 -3.99480730e-01 4.10753220e-01
3.33840340e-01 6.01015925e-01 -3.86934668e-01 -2.15547308e-01
-2.22979769e-01 3.71095181e-01 -3.60397100e-02 -7.24475324e-01
1.18402874e+00 -4.10900474e-01 3.75990957e-01 4.49352622e-01
2.07886308e-01 -6.64784491e-01 -1.44802678e+00 -2.27543056e-01
1.09806687e-01 -6.88834906e-01 4.26059425e-01 -8.57689917e-01
-5.33490539e-01 5.96431315e-01 1.89925566e-01 5.21679342e-01
1.17042983e+00 -3.97321790e-01 7.58104026e-02 7.31135011e-01
5.63212991e-01 -1.63244319e+00 2.51079082e-01 4.05517220e-01
1.02314186e+00 -1.24161923e+00 2.69255936e-01 1.44457445e-01
-1.08382678e+00 1.19558322e+00 5.54755390e-01 7.46235847e-02
2.87934989e-01 -6.34916276e-02 1.69900432e-01 6.32125437e-02
-1.07520425e+00 -3.92460942e-01 -2.17532381e-01 6.44494712e-01
1.27192393e-01 1.87159881e-01 -7.49181986e-01 -1.41140893e-01
2.76457891e-02 3.08862269e-01 8.99986267e-01 1.71263278e+00
-8.94867957e-01 -1.43519258e+00 -8.03254545e-01 4.89481539e-01
-5.04652977e-01 2.64563680e-01 -1.26306236e-01 8.87305617e-01
-2.04811618e-01 1.27330995e+00 7.06917346e-02 9.26711485e-02
2.56203681e-01 -9.24321115e-02 5.05611181e-01 -6.86318815e-01
-7.02883184e-01 3.46820831e-01 5.67735434e-01 -3.32350373e-01
-5.30982256e-01 -1.00050116e+00 -1.04782474e+00 -5.80945194e-01
-1.33135661e-01 1.76276460e-01 6.26287997e-01 1.20909882e+00
-9.99257341e-02 6.07170999e-01 7.09503233e-01 -8.11217904e-01
-1.18682969e+00 -6.89793050e-01 -6.24569356e-01 -4.10668284e-01
5.22886455e-01 -9.83664095e-01 -1.04615234e-01 -4.42684084e-01] | [4.220820426940918, 2.407278537750244] |
5690d413-0117-462b-a658-0cec69bae9a1 | scalable-resource-management-for-dynamic-mec | 2306.08938 | null | https://arxiv.org/abs/2306.08938v2 | https://arxiv.org/pdf/2306.08938v2.pdf | Scalable Resource Management for Dynamic MEC: An Unsupervised Link-Output Graph Neural Network Approach | Deep learning has been successfully adopted in mobile edge computing (MEC) to optimize task offloading and resource allocation. However, the dynamics of edge networks raise two challenges in neural network (NN)-based optimization methods: low scalability and high training costs. Although conventional node-output graph neural networks (GNN) can extract features of edge nodes when the network scales, they fail to handle a new scalability issue whereas the dimension of the decision space may change as the network scales. To address the issue, in this paper, a novel link-output GNN (LOGNN)-based resource management approach is proposed to flexibly optimize the resource allocation in MEC for an arbitrary number of edge nodes with extremely low algorithm inference delay. Moreover, a label-free unsupervised method is applied to train the LOGNN efficiently, where the gradient of edge tasks processing delay with respect to the LOGNN parameters is derived explicitly. In addition, a theoretical analysis of the scalability of the node-output GNN and link-output GNN is performed. Simulation results show that the proposed LOGNN can efficiently optimize the MEC resource allocation problem in a scalable way, with an arbitrary number of servers and users. In addition, the proposed unsupervised training method has better convergence performance and speed than supervised learning and reinforcement learning-based training methods. The code is available at \url{https://github.com/UNIC-Lab/LOGNN}. | ['Xuemin Shen', 'Tom Luan', 'Yilong Hui', 'Ruijin Sun', 'Wei Quan', 'Lianhao Fu', 'Nan Cheng', 'Xiucheng Wang'] | 2023-06-15 | null | null | null | null | ['edge-computing'] | ['time-series'] | [-1.93897069e-01 -4.19571340e-01 -6.04829073e-01 4.26110476e-02
1.17617466e-01 -2.04126179e-01 -3.16010565e-01 -3.99269789e-01
-3.30593020e-01 8.31188381e-01 -3.81188333e-01 -6.90248549e-01
-6.28151953e-01 -7.58782744e-01 -5.26976883e-01 -6.39009595e-01
-2.65861630e-01 6.19043589e-01 -4.58652340e-02 -4.10414450e-02
-1.67019114e-01 6.43179417e-01 -1.30905223e+00 -3.84346128e-01
8.35636377e-01 1.23067677e+00 4.11478341e-01 8.68942261e-01
1.00804761e-01 7.04960585e-01 -2.96769023e-01 6.51192069e-02
4.04759079e-01 -1.73797637e-01 -5.48447013e-01 -2.35234052e-01
-2.05157459e-01 -1.58781692e-01 -7.65214562e-01 9.03089166e-01
8.75390887e-01 4.18831706e-01 3.98470521e-01 -1.70378721e+00
-3.45458686e-01 5.04157662e-01 -3.26037079e-01 6.80809617e-01
-5.39797246e-01 1.68861449e-03 8.18983138e-01 -6.09024286e-01
4.97238100e-01 5.32490551e-01 6.79647982e-01 4.75457430e-01
-5.88367581e-01 -8.51503372e-01 2.78729886e-01 5.76974213e-01
-1.34359193e+00 -4.34738010e-01 6.76637888e-01 -6.16757870e-02
1.06172132e+00 1.66762143e-01 6.12427413e-01 7.31568933e-01
1.84430867e-01 6.39070272e-01 5.52591205e-01 -3.32965016e-01
2.56425321e-01 -1.59913376e-02 -1.18211515e-01 8.04271340e-01
4.40305203e-01 8.27702731e-02 -6.16707981e-01 1.26073599e-01
9.28221822e-01 2.58862436e-01 -4.60777208e-02 -1.97139800e-01
-8.09713125e-01 4.83451337e-01 6.43261135e-01 1.24021508e-01
-5.36326766e-01 7.08501041e-01 5.15624762e-01 4.33575392e-01
4.85825807e-01 3.23946774e-02 -8.76891077e-01 -4.13645774e-01
-7.49193013e-01 -3.11000347e-01 9.60831285e-01 1.23884130e+00
6.04007900e-01 5.26621521e-01 -8.96875262e-02 5.69634140e-01
1.16020337e-01 5.87607503e-01 3.49314868e-01 -1.05552912e+00
6.60315096e-01 4.10236031e-01 -6.80895597e-02 -1.15233397e+00
-9.31075871e-01 -9.96785820e-01 -1.22303724e+00 -2.21133325e-03
1.64610550e-01 -7.51611829e-01 -5.08912504e-01 1.61706865e+00
3.92569631e-01 4.96778548e-01 -1.40886664e-01 8.82896364e-01
4.26684409e-01 5.63545644e-01 4.52701263e-02 -3.67096394e-01
1.19975376e+00 -1.40355754e+00 -7.04138637e-01 -2.72234172e-01
8.41652334e-01 -3.33520263e-01 8.57470632e-01 -1.07016616e-01
-1.03345549e+00 -3.55333447e-01 -7.72162020e-01 3.03109616e-01
-3.85906219e-01 1.78167298e-01 1.01837838e+00 7.53243446e-01
-1.26019371e+00 5.33038557e-01 -7.35066414e-01 -1.92943826e-01
5.37902176e-01 9.65017438e-01 2.02529311e-01 2.33514309e-02
-1.27579284e+00 3.06357652e-01 5.97612381e-01 5.22769630e-01
-6.93360329e-01 -5.19519210e-01 -4.72865224e-01 5.44861257e-01
7.93349743e-01 -9.65183020e-01 1.15912116e+00 -1.15170050e+00
-1.67189801e+00 1.10329047e-01 -5.56905530e-02 -2.19415545e-01
2.68808156e-01 1.16692498e-01 -5.76355696e-01 1.48360908e-01
-2.27355242e-01 2.84930199e-01 8.28122437e-01 -7.34280944e-01
-8.51016045e-01 -2.68690556e-01 1.19568124e-01 5.09616971e-01
-8.30973506e-01 -1.36182621e-01 -7.05660224e-01 -4.96936202e-01
-1.83099419e-01 -9.99143839e-01 -2.71657497e-01 -2.54148424e-01
-1.53680101e-01 -1.14833184e-01 8.66885722e-01 -4.35285062e-01
1.59694755e+00 -1.96795797e+00 -1.95990987e-02 3.56725991e-01
5.05433798e-01 3.20426285e-01 -7.25632161e-02 2.43011877e-01
3.43916833e-01 1.32005543e-01 2.57781416e-01 -2.10553318e-01
-1.28138969e-02 2.79047608e-01 4.86147135e-01 2.10604101e-01
-3.16675574e-01 1.22727609e+00 -9.63163257e-01 -5.55857718e-01
-1.22200370e-01 3.66526276e-01 -6.22999489e-01 -5.02045639e-02
-6.86349869e-02 3.95779580e-01 -6.62877321e-01 8.57681155e-01
4.27079737e-01 -8.00094306e-01 6.13754988e-01 -6.47872612e-02
3.09738964e-01 -1.49719059e-01 -1.13411021e+00 1.02279365e+00
-8.83239686e-01 7.52378285e-01 3.23461294e-01 -1.35167384e+00
5.40657103e-01 5.18604755e-01 8.52398098e-01 -7.79311776e-01
3.51727635e-01 3.38801563e-01 1.53535232e-01 -6.64218247e-01
3.68762165e-01 1.69098958e-01 3.56131606e-03 5.55730164e-01
-5.69476075e-02 8.61308932e-01 1.73619717e-01 3.34669054e-02
1.06379044e+00 -1.56236410e-01 -6.32815063e-02 -2.26880133e-01
4.79525656e-01 -3.13199580e-01 7.85359681e-01 7.65804112e-01
-3.76925200e-01 -2.02019975e-01 5.08355498e-01 -4.11019653e-01
-9.79883969e-01 -5.69249034e-01 1.87702700e-01 1.63502693e+00
3.47101063e-01 -7.57283792e-02 -9.01952624e-01 -7.29172766e-01
-8.81396607e-02 2.27767304e-01 -3.13857704e-01 -1.94637790e-01
-6.76985323e-01 -8.30311835e-01 3.00532103e-01 6.05262876e-01
4.29724663e-01 -1.35459208e+00 -5.36366642e-01 3.12133223e-01
-2.57677168e-01 -1.42764318e+00 -7.63949454e-01 2.53908664e-01
-1.08676434e+00 -8.47827613e-01 -4.18841183e-01 -1.01551259e+00
7.88231015e-01 3.52248102e-01 1.11638200e+00 4.54738855e-01
1.97332986e-02 5.49191535e-01 -1.33210644e-01 -2.15001777e-01
-1.87398493e-02 9.25491571e-01 3.09258908e-01 8.77696946e-02
1.58753246e-01 -9.17205811e-01 -8.42904508e-01 5.85546911e-01
-6.49626970e-01 6.58058897e-02 7.64719188e-01 7.45864511e-01
6.84770048e-01 5.16255915e-01 1.10273397e+00 -1.05150676e+00
6.84044838e-01 -7.67527521e-01 -6.89003468e-01 2.84490764e-01
-1.21308804e+00 -4.31489982e-02 1.10784554e+00 -6.05522513e-01
-6.21218860e-01 -1.91484496e-01 3.49708855e-01 -6.50148094e-01
2.25475878e-01 5.94176412e-01 -2.15034544e-01 -2.33673409e-01
1.90146580e-01 8.49569365e-02 -9.29478854e-02 -1.85070783e-01
8.37244540e-02 7.50463486e-01 2.00907722e-01 -5.63444912e-01
6.12256527e-01 2.32374936e-01 3.92691165e-01 -5.37583232e-01
-6.45128906e-01 -4.66861069e-01 -3.20703506e-01 -4.99341190e-01
3.88623714e-01 -7.76216805e-01 -1.24230683e+00 3.28193992e-01
-7.27542758e-01 -8.39838803e-01 -1.17450304e-01 5.67131519e-01
-6.13546312e-01 9.21905562e-02 -6.83924615e-01 -7.81972468e-01
-7.46216953e-01 -9.92374301e-01 5.57899356e-01 5.97522855e-01
3.93026233e-01 -1.40906346e+00 -6.17371976e-01 2.69354999e-01
9.09429848e-01 -1.96082324e-01 8.23550224e-01 -4.88508284e-01
-7.22903013e-01 -2.74700254e-01 -5.47432840e-01 -1.23492358e-02
-6.52572419e-03 -4.14229184e-01 -3.91525507e-01 -6.30090654e-01
-1.55608937e-01 1.13380462e-01 2.80100793e-01 6.64075077e-01
1.50058758e+00 -4.68313098e-01 -4.46967959e-01 1.07251120e+00
1.61299086e+00 1.23040013e-01 2.69495070e-01 4.76674736e-01
9.60868955e-01 5.03561161e-02 2.46232182e-01 5.27380407e-01
3.76747936e-01 5.59609532e-01 6.78982139e-01 -5.79815432e-02
-5.54717146e-04 2.08215062e-02 2.53073156e-01 1.16379845e+00
-3.97697091e-01 -7.56128252e-01 -7.90626287e-01 1.21918097e-01
-2.12535977e+00 -5.70664763e-01 2.09976044e-02 2.10149312e+00
3.13354403e-01 4.39888448e-01 1.58125088e-01 -1.71542969e-02
8.29014599e-01 -9.36414395e-03 -1.07279420e+00 -3.69840294e-01
1.86578214e-01 -1.05698537e-02 9.80368137e-01 4.50292259e-01
-7.94382930e-01 9.93714869e-01 5.66749907e+00 9.43319023e-01
-1.24145997e+00 4.47310239e-01 6.51925087e-01 -3.06267440e-01
-2.99154241e-02 -2.13207275e-01 -8.72607708e-01 7.36039340e-01
1.29074526e+00 -1.93537757e-01 1.09118676e+00 9.16264594e-01
5.51533043e-01 2.63686568e-01 -6.03756547e-01 1.20980215e+00
-2.59152472e-01 -1.29795802e+00 -3.37620556e-01 9.44737494e-02
7.08625793e-01 3.23066801e-01 1.25377849e-02 5.19567966e-01
-9.39941034e-02 -9.41763997e-01 3.03151399e-01 6.30607963e-01
1.10576797e+00 -9.08777773e-01 8.78801942e-01 4.59702343e-01
-1.42999208e+00 -5.27848721e-01 -3.52882802e-01 -2.84782082e-01
1.33705258e-01 5.82735717e-01 -6.43198609e-01 4.76191223e-01
7.92463720e-01 5.13686776e-01 -4.03980762e-01 9.83260334e-01
5.80446832e-02 7.15569496e-01 -3.44940990e-01 -3.91201496e-01
8.30484927e-02 -3.32471341e-01 5.70935011e-01 1.05516016e+00
3.17935318e-01 -4.51531708e-02 2.48827234e-01 4.38437134e-01
-5.72079062e-01 1.10435590e-01 -2.65400410e-01 -1.36804283e-01
6.38775229e-01 1.62802160e+00 -1.01647782e+00 -1.53864950e-01
-2.29994088e-01 8.84421051e-01 5.09209335e-01 8.63882124e-01
-9.78442430e-01 -5.16244769e-01 4.65732336e-01 1.15627959e-01
4.12201047e-01 -4.08251375e-01 -1.64781258e-01 -8.03689897e-01
1.32139653e-01 -5.45180917e-01 3.55354875e-01 -6.08036935e-01
-9.55952048e-01 6.66980267e-01 -3.42650682e-01 -1.19985592e+00
-5.10570556e-02 -6.72720015e-01 -5.18427312e-01 6.04120195e-01
-1.70700312e+00 -8.41789722e-01 -4.49680328e-01 6.61346674e-01
4.77411330e-01 -4.72458243e-01 4.82904822e-01 8.31651926e-01
-1.12652886e+00 1.04191244e+00 5.19522727e-01 7.34796077e-02
1.80282459e-01 -1.01253498e+00 4.55144867e-02 7.62036204e-01
-1.50074065e-01 1.86523721e-01 2.91846663e-01 -4.34084207e-01
-1.64061248e+00 -1.23150992e+00 5.99262595e-01 3.45990770e-02
6.99870408e-01 -2.67780513e-01 -3.91325027e-01 5.20399809e-01
-1.57809064e-01 5.80868065e-01 6.66257441e-01 9.97647494e-02
3.33430618e-01 -3.18640620e-01 -7.49298215e-01 6.17730260e-01
1.45610452e+00 -3.48644316e-01 8.06005657e-01 7.11466432e-01
8.78835559e-01 -4.77840602e-01 -8.59777629e-01 4.04620171e-01
4.50454354e-01 -5.75195551e-01 7.99598992e-01 -8.71578515e-01
-8.34586397e-02 -1.04005672e-01 8.34862962e-02 -9.65840995e-01
-5.69455445e-01 -9.34634626e-01 -9.55784798e-01 7.76570439e-01
6.26994133e-01 -9.92477357e-01 1.24170804e+00 2.43524253e-01
-1.97282672e-01 -1.30979824e+00 -9.05693591e-01 -9.05543864e-01
-3.96971911e-01 -3.83492440e-01 5.37852585e-01 7.15825796e-01
-4.43585873e-01 5.42959273e-01 -4.91297841e-01 3.57815325e-01
4.95099664e-01 -1.14571396e-02 4.05819148e-01 -1.17402458e+00
-4.63293016e-01 -5.51927507e-01 -2.25291222e-01 -1.03944886e+00
1.94518909e-01 -1.16846633e+00 -2.82807499e-01 -1.52255750e+00
-5.27705662e-02 -1.07454932e+00 -8.14435661e-01 3.45269382e-01
-1.63284436e-01 -5.38451299e-02 1.23715572e-01 4.58205700e-01
-1.14792192e+00 3.76697421e-01 1.26621151e+00 1.69811681e-01
-4.60106134e-01 5.63382328e-01 -5.44007003e-01 4.80592608e-01
1.36445427e+00 -7.30892062e-01 -6.40990496e-01 -5.70459962e-01
6.78496480e-01 2.07029521e-01 2.96391838e-04 -9.97169018e-01
5.47316611e-01 -1.64763704e-01 2.40499690e-01 -7.60175437e-02
-4.02012654e-02 -1.19875216e+00 1.09219059e-01 4.26625490e-01
1.15007661e-01 3.75432193e-01 5.95510751e-02 8.08361113e-01
3.08537334e-01 -2.05541775e-01 2.62490630e-01 1.44351512e-01
-6.79528236e-01 1.04619658e+00 -4.37159657e-01 1.43536761e-01
9.06424880e-01 -2.83160895e-01 -1.16718479e-01 -5.14771879e-01
-7.48858571e-01 6.86480641e-01 5.21060498e-03 2.83550501e-01
2.82656550e-01 -1.42184258e+00 -3.54825348e-01 4.85635102e-02
-4.20072407e-01 -2.99288612e-02 5.52710354e-01 1.28243911e+00
-6.88280880e-01 3.47915769e-01 -8.63152742e-02 -4.01687413e-01
-7.28208661e-01 6.78340077e-01 7.83264577e-01 -6.65426791e-01
-3.96759868e-01 5.56136906e-01 -3.79122704e-01 -5.04075229e-01
4.72462684e-01 2.41731897e-01 -8.55809152e-02 -2.35160172e-01
5.37308753e-02 7.76515603e-01 2.62932666e-02 -3.01425010e-01
-1.94260880e-01 2.95214266e-01 1.80882543e-01 4.15654689e-01
1.25888765e+00 -3.84405971e-01 -1.04729898e-01 1.63167700e-01
1.12854171e+00 -3.60976070e-01 -1.17271304e+00 -4.64282393e-01
-2.58037634e-02 -6.32051229e-02 3.41179579e-01 -4.12431061e-01
-1.68992925e+00 5.85641801e-01 8.43549073e-01 3.38677585e-01
1.48230016e+00 -4.00083482e-01 1.11258280e+00 4.56107289e-01
2.84973413e-01 -1.49953794e+00 5.90529777e-02 5.66634238e-01
9.98267531e-02 -1.14892876e+00 -1.77768871e-01 -3.82362217e-01
-1.74800068e-01 1.21868181e+00 8.33194375e-01 2.52159357e-01
1.03359163e+00 2.85699546e-01 -2.46943757e-01 -2.68538684e-01
-6.70917749e-01 -1.19140469e-01 1.01212580e-02 4.66552168e-01
-1.26961153e-02 2.97806293e-01 -1.59468174e-01 6.54191434e-01
9.58636403e-02 1.91660285e-01 2.43811086e-01 7.75207579e-01
-3.11582357e-01 -9.75874305e-01 1.10630311e-01 8.17171872e-01
-4.71409827e-01 -2.69028872e-01 3.84150565e-01 5.60150743e-01
8.32449738e-03 7.16421366e-01 -1.17993755e-02 -4.16121244e-01
1.12814456e-01 -4.64845821e-02 1.30248904e-01 -2.51836151e-01
-3.18482757e-01 -2.71621227e-01 -8.53001624e-02 -5.11472881e-01
-1.30312666e-01 -1.00204095e-01 -1.38755441e+00 -5.24708033e-01
-5.38693845e-01 1.56480148e-01 6.81272328e-01 9.27952111e-01
8.64233315e-01 9.99402344e-01 8.87054563e-01 -8.03428054e-01
-3.27479988e-01 -6.81521118e-01 -6.85350597e-01 -1.13576073e-02
2.87580360e-02 -5.35520911e-01 -4.49998796e-01 -3.48861754e-01] | [6.08039665222168, 1.8661434650421143] |
fbe4d472-2b55-4542-9116-494cf827bd0c | statistical-relational-learning-and-neuro | 2306.13660 | null | https://arxiv.org/abs/2306.13660v1 | https://arxiv.org/pdf/2306.13660v1.pdf | Statistical relational learning and neuro-symbolic AI: what does first-order logic offer? | In this paper, our aim is to briefly survey and articulate the logical and philosophical foundations of using (first-order) logic to represent (probabilistic) knowledge in a non-technical fashion. Our motivation is three fold. First, for machine learning researchers unaware of why the research community cares about relational representations, this article can serve as a gentle introduction. Second, for logical experts who are newcomers to the learning area, such an article can help in navigating the differences between finite vs infinite, and subjective probabilities vs random-world semantics. Finally, for researchers from statistical relational learning and neuro-symbolic AI, who are usually embedded in finite worlds with subjective probabilities, appreciating what infinite domains and random-world semantics brings to the table is of utmost theoretical import. | ['Vaishak Belle'] | 2023-06-08 | null | null | null | null | ['relational-reasoning'] | ['natural-language-processing'] | [ 1.32896185e-01 8.44606221e-01 -4.88286227e-01 -4.42520827e-01
-5.44354141e-01 -4.98262286e-01 8.15374911e-01 3.83967429e-01
-4.29530025e-01 7.88014174e-01 2.91211456e-01 -6.73203409e-01
-7.67072320e-01 -1.07417119e+00 -6.00190163e-01 -5.43129325e-01
-2.48604774e-01 4.96970832e-01 8.68206695e-02 -1.27480805e-01
3.53828847e-01 3.66919100e-01 -1.47874463e+00 7.65254647e-02
8.08872819e-01 7.11175919e-01 -2.74442285e-02 9.39585268e-02
-4.34507549e-01 1.53651845e+00 -2.37415522e-01 -1.05380404e+00
-4.78555918e-01 -1.31222188e-01 -1.08735955e+00 -3.99923086e-01
1.57838613e-01 6.87464923e-02 -3.93410683e-01 1.17543054e+00
-1.76133931e-01 1.91216335e-01 9.58939850e-01 -1.02460003e+00
-1.04683220e+00 1.43658268e+00 8.88814181e-02 2.18426362e-02
4.96979088e-01 -2.43115455e-01 1.29868555e+00 -4.48765218e-01
3.98957759e-01 1.58871973e+00 7.59012818e-01 4.45182949e-01
-1.21061516e+00 -4.75067645e-02 3.83901358e-01 4.68156636e-01
-1.55250418e+00 -3.47479731e-01 7.46181250e-01 -4.56230968e-01
6.16753936e-01 2.98014313e-01 5.99209070e-01 1.06059074e+00
-9.65848682e-04 8.85908186e-01 1.14119351e+00 -8.27033281e-01
4.47071582e-01 3.47577810e-01 3.81861061e-01 5.58561504e-01
4.24690932e-01 2.43100971e-01 -6.62813246e-01 -1.10704169e-01
6.29517555e-01 -1.86366707e-01 -3.50819491e-02 -2.89442837e-01
-1.24068940e+00 8.62833261e-01 1.63086891e-01 6.85369134e-01
-2.69369304e-01 4.42012966e-01 2.30013847e-01 2.57064313e-01
9.41620618e-02 3.40232253e-01 -4.10634547e-01 -4.09771025e-01
-7.74129987e-01 2.00335503e-01 1.08186913e+00 6.90644562e-01
3.18465680e-01 -2.93825250e-02 4.77153808e-01 5.62544644e-01
7.99434721e-01 4.30768192e-01 2.09777161e-01 -1.26549721e+00
-1.91199079e-01 1.80877000e-01 2.21931696e-01 -1.13447487e+00
-7.40394965e-02 2.20385920e-02 -5.68387568e-01 1.38220593e-01
6.41578078e-01 1.22144364e-01 -3.00205976e-01 1.71382499e+00
-2.04878941e-01 -9.15307328e-02 4.12533849e-01 6.74794078e-01
7.83267140e-01 3.90650213e-01 5.69236040e-01 -4.14094567e-01
1.47711897e+00 -8.63593370e-02 -8.53860736e-01 -2.05737606e-01
6.01225853e-01 -4.39180493e-01 1.10462618e+00 5.54195702e-01
-1.13950098e+00 2.59270556e-02 -8.91784668e-01 -1.05336532e-01
-5.85607588e-01 -3.28857183e-01 1.17941368e+00 7.97304153e-01
-8.03121805e-01 8.34882557e-01 -1.11676347e+00 -4.82369572e-01
4.69962448e-01 -7.66537488e-02 -3.85667048e-02 4.99775726e-03
-1.69732475e+00 1.61050105e+00 6.51467741e-01 1.60115615e-01
-3.01951438e-01 -4.50725377e-01 -1.06176531e+00 -1.08116627e-01
7.85034239e-01 -4.61268783e-01 9.84960198e-01 -8.63812506e-01
-1.30676365e+00 1.02854156e+00 -2.95642972e-01 -6.10394418e-01
4.48505819e-01 -2.59212345e-01 -6.18489683e-01 7.04165176e-02
-2.40830496e-01 1.84897501e-02 3.91797662e-01 -1.36462092e+00
-3.73279631e-01 -3.63836080e-01 5.07089794e-01 3.14673692e-01
9.61681083e-02 1.03243150e-01 1.58741117e-01 -4.54163313e-01
6.49095058e-01 -3.29504609e-01 -6.07191473e-02 -1.64534807e-01
-2.26322487e-01 -6.78291082e-01 2.66486079e-01 -1.07406184e-01
1.16463506e+00 -2.07315159e+00 2.27484945e-02 3.22276384e-01
3.28215599e-01 -3.19725364e-01 5.09512901e-01 4.19049472e-01
-9.51075256e-02 4.66984689e-01 -6.00430295e-02 2.66142607e-01
5.70876598e-01 3.77586126e-01 -8.55267584e-01 4.11320239e-01
9.70635191e-02 9.93457794e-01 -1.23635542e+00 -8.14367056e-01
4.32502508e-01 5.02852023e-01 -6.12000898e-02 -2.63523608e-01
-9.05578509e-02 -2.15550065e-01 -6.57625318e-01 6.41560972e-01
3.27497989e-01 -3.27574998e-01 5.95052660e-01 -1.47577956e-01
2.69632824e-02 6.17417336e-01 -1.40822244e+00 1.11259699e+00
-2.49806121e-01 6.78996027e-01 -4.04562980e-01 -1.04489648e+00
6.86072469e-01 2.69317359e-01 -1.72939584e-01 -2.10409939e-01
2.59921461e-01 1.10391438e-01 1.14961900e-01 -3.86747837e-01
4.73470181e-01 -9.01867628e-01 -1.47195831e-01 5.16014755e-01
-5.28165586e-02 -4.33130026e-01 7.40817338e-02 3.38478029e-01
6.58644319e-01 4.56216007e-01 5.18584371e-01 -4.39716905e-01
4.58711356e-01 -1.33935288e-01 2.95540154e-01 8.11093330e-01
-1.32107183e-01 1.14287034e-01 8.53184342e-01 -5.88846266e-01
-4.96838838e-01 -1.77811265e+00 -6.46091223e-01 1.09986043e+00
2.01934025e-01 -4.34760869e-01 -3.36606771e-01 -2.89209872e-01
-1.00298986e-01 1.35855412e+00 -5.95625997e-01 -1.52375340e-01
-2.98169792e-01 -5.10763407e-01 5.96945882e-01 5.76901376e-01
-1.42424211e-01 -1.15358734e+00 -7.48614192e-01 -2.01874282e-02
-1.02205433e-01 -7.85724461e-01 4.94999915e-01 5.27475417e-01
-8.94992650e-01 -8.06594729e-01 -1.27093688e-01 -5.79098582e-01
3.35366607e-01 -2.74439871e-01 1.24482393e+00 5.29520288e-02
1.70080885e-01 6.47618830e-01 -2.88866192e-01 -5.17982244e-01
-5.93129575e-01 -3.38308752e-01 1.24376886e-01 -3.90446991e-01
9.53916132e-01 -8.37098777e-01 1.62594661e-01 -2.04008803e-01
-9.38935101e-01 -2.13066489e-01 1.93967238e-01 5.26315629e-01
3.13741684e-01 4.23944861e-01 3.75120550e-01 -1.30715442e+00
5.07872939e-01 -5.02559483e-01 -3.63198072e-01 6.66669071e-01
-4.48266715e-01 3.00347269e-01 3.16634625e-01 -3.45484376e-01
-1.21119320e+00 -5.60683012e-01 3.05462241e-01 1.15448376e-02
-2.28488296e-01 1.00814152e+00 -2.85420209e-01 3.45917583e-01
8.66061985e-01 1.68777540e-01 2.58033946e-02 -1.54651672e-01
8.48223805e-01 4.37133104e-01 4.65771377e-01 -1.14136505e+00
7.99899340e-01 4.32445467e-01 1.20533951e-01 -7.24456429e-01
-1.14933801e+00 2.42022797e-01 -7.31584191e-01 -1.83625087e-01
5.75914741e-01 -6.79816604e-01 -1.07776010e+00 2.67699033e-01
-8.47880244e-01 -2.67944872e-01 -5.89543104e-01 6.21311665e-01
-1.01866663e+00 3.49519491e-01 -5.00657141e-01 -1.38978648e+00
4.24264491e-01 -6.98525548e-01 4.05498564e-01 2.20464692e-01
-7.22926438e-01 -1.51067114e+00 -5.50852358e-01 1.33057341e-01
1.30990401e-01 2.84427971e-01 1.33478856e+00 -8.32905889e-01
-4.96735424e-01 -3.71595889e-01 -2.59317961e-02 1.25876859e-01
5.48375137e-02 2.20874310e-01 -1.02549946e+00 6.47928596e-01
2.89965302e-01 -8.46468151e-01 4.37656373e-01 2.62137562e-01
1.02434421e+00 -3.95246148e-01 -2.49988079e-01 3.28773558e-02
1.42803311e+00 1.81696549e-01 7.98070431e-01 2.97339469e-01
4.33201015e-01 1.01528025e+00 4.47207808e-01 2.70677477e-01
9.81780052e-01 8.15865993e-02 4.37211506e-02 5.31422019e-01
2.38673896e-01 -4.39774334e-01 2.55156010e-01 6.60515785e-01
-3.26446831e-01 8.80935490e-02 -1.03708732e+00 5.72602391e-01
-1.83662534e+00 -1.63068283e+00 -1.45498008e-01 2.34508181e+00
1.49719715e+00 5.82966626e-01 -1.93039894e-01 2.24812627e-01
7.61750638e-01 4.35974300e-01 -7.72641525e-02 -3.88439000e-01
-3.30682248e-01 1.57820031e-01 8.91838223e-02 8.24490309e-01
-1.05691695e+00 1.12080395e+00 7.09330988e+00 1.00408995e+00
-6.09832406e-01 -2.45771497e-01 4.28691834e-01 3.44071448e-01
-8.06505144e-01 3.71692538e-01 -3.40541571e-01 2.75048584e-01
1.17065156e+00 -3.67311627e-01 6.18680358e-01 8.67958486e-01
-2.98926890e-01 -4.71049190e-01 -1.51252043e+00 8.21489275e-01
-9.87595171e-02 -1.30532157e+00 -6.68674335e-02 -1.50831074e-01
2.82923281e-01 -2.97726125e-01 5.71158603e-02 1.04135670e-01
1.01603186e+00 -1.48578262e+00 1.13818526e+00 9.48514760e-01
3.24103475e-01 -7.05971956e-01 7.66297758e-01 4.34050977e-01
-7.03837633e-01 4.80612852e-02 -4.33471292e-01 -4.33174461e-01
3.43489379e-01 7.59424508e-01 -1.70422375e-01 3.74778599e-01
3.87746304e-01 5.22725761e-01 -2.46529579e-01 5.70881128e-01
-4.56933349e-01 5.65428078e-01 -4.43465024e-01 -2.80342013e-01
-4.36283946e-02 -1.90397844e-01 4.28669274e-01 1.03308666e+00
-1.90037549e-01 3.63459349e-01 -3.00736099e-01 1.29971731e+00
4.59946364e-01 -3.40391278e-01 -7.21996069e-01 -4.80787128e-01
7.04425395e-01 9.31325078e-01 -8.20089281e-01 -2.32296348e-01
-4.19810951e-01 2.30473474e-01 2.71204740e-01 3.17081749e-01
-7.59090185e-01 -4.98665959e-01 4.32835668e-01 -4.32017026e-03
-1.29328445e-01 -2.42249355e-01 -7.57234216e-01 -1.01865315e+00
-1.60470441e-01 -6.97242439e-01 2.81967580e-01 -8.02279055e-01
-1.53653884e+00 3.04745585e-01 5.21781027e-01 -6.34126306e-01
-3.22249085e-01 -8.72388542e-01 -3.95372748e-01 7.89637625e-01
-1.15445709e+00 -1.07369268e+00 3.85246128e-01 1.94574580e-01
-2.15705603e-01 1.59272775e-01 1.02667117e+00 -3.88100713e-01
1.04607180e-01 2.61224031e-01 -2.04276577e-01 1.90761879e-01
1.84295982e-01 -1.47178578e+00 1.27752230e-01 3.49343628e-01
3.58730108e-01 1.31957650e+00 9.98794258e-01 -4.98676330e-01
-1.47218478e+00 -3.51387113e-01 1.50891626e+00 -1.15690792e+00
1.19609511e+00 -3.93307060e-02 -1.04425728e+00 1.00567853e+00
-8.92624855e-02 -7.38768578e-02 8.03162336e-01 7.10104883e-01
-6.60374463e-01 1.08921528e-01 -1.20498502e+00 8.93376291e-01
7.75196493e-01 -8.71360362e-01 -1.54217434e+00 1.73717946e-01
5.53799331e-01 -2.56778687e-01 -9.67459261e-01 2.89023072e-01
7.55186915e-01 -8.41115117e-01 9.66593623e-01 -6.75695658e-01
3.34903121e-01 -1.69487342e-01 -4.39362735e-01 -6.30138099e-01
-2.05009252e-01 -5.35789371e-01 -1.37076136e-02 1.20831347e+00
3.46685559e-01 -7.92328477e-01 6.23197377e-01 1.15997028e+00
1.41840801e-01 -6.79816365e-01 -8.69671106e-01 -6.02504253e-01
5.65219343e-01 -1.17595720e+00 3.28546345e-01 1.09737515e+00
6.66104972e-01 6.99436441e-02 1.71116948e-01 -1.30091518e-01
8.33863914e-01 -7.32276514e-02 6.54134378e-02 -1.71754885e+00
-5.39347343e-02 -5.64399302e-01 -6.44674122e-01 -7.95596123e-01
2.43638098e-01 -1.02168036e+00 1.23545460e-01 -1.57669747e+00
1.72784626e-01 -5.75228512e-01 -4.81881231e-01 4.39294696e-01
6.42440142e-03 -1.00216046e-01 2.44378790e-01 3.10077723e-02
-6.11526489e-01 3.10921252e-01 9.05537367e-01 2.22775713e-01
1.25603855e-01 -9.70219374e-02 -1.24543905e+00 1.41837513e+00
6.21151149e-01 -4.14454192e-01 -5.60239136e-01 -1.12597965e-01
9.96170700e-01 2.96100639e-02 7.20658123e-01 -7.25826621e-01
1.99849486e-01 -7.64830053e-01 2.12490246e-01 -4.67711210e-01
3.47482800e-01 -7.70989299e-01 4.05669846e-02 1.98386744e-01
-6.37244940e-01 -3.40119630e-01 4.09379490e-02 4.06572729e-01
-2.57655662e-02 -6.52908683e-01 6.55599713e-01 -3.60781699e-01
-6.58562541e-01 -4.98276621e-01 -4.21458542e-01 4.32645500e-01
6.65769756e-01 -2.39043877e-01 -4.19258833e-01 -4.76114273e-01
-8.75063837e-01 7.45078828e-03 4.43817854e-01 -6.07154481e-02
6.21100664e-01 -1.20414400e+00 -3.21745634e-01 -3.77099872e-01
-1.07535765e-01 -2.17910290e-01 2.82760799e-01 7.94345856e-01
-3.67496520e-01 6.41629219e-01 1.63825274e-01 -1.43562585e-01
-5.61532199e-01 5.80213964e-01 3.34260762e-02 2.89570183e-01
-5.07182360e-01 1.13244355e+00 -1.05142206e-01 -3.11337918e-01
5.37788987e-01 -3.40304822e-01 -1.60313711e-01 2.81709969e-01
3.74233842e-01 4.19886172e-01 -4.94811594e-01 -5.13193667e-01
-5.39052129e-01 3.84601176e-01 8.05986747e-02 -3.27686191e-01
1.24702728e+00 -1.61914364e-01 -5.91162443e-01 1.54342818e+00
5.49599528e-01 1.40189543e-01 -7.95047760e-01 -3.96437854e-01
3.82668644e-01 -2.65250921e-01 -1.10538502e-03 -9.33531165e-01
-3.07775259e-01 8.19887638e-01 7.94868022e-02 7.00439572e-01
5.74720562e-01 6.70025349e-01 1.72967389e-01 6.83347702e-01
6.88312471e-01 -9.62626636e-01 -3.29668224e-01 5.67099214e-01
7.61911571e-01 -1.08098292e+00 3.23973894e-01 -3.90102565e-01
-1.01687646e+00 1.24265778e+00 2.18448803e-01 -5.81906289e-02
7.92857826e-01 2.63373703e-01 -8.36771727e-03 -2.97814429e-01
-7.89035141e-01 -2.56154001e-01 1.97044447e-01 8.91095340e-01
8.80129814e-01 4.29965377e-01 -3.80920380e-01 7.90052474e-01
-6.88866079e-01 1.91496953e-01 6.56635106e-01 7.58246243e-01
-6.50998294e-01 -1.12217832e+00 -3.95159990e-01 3.71132284e-01
-6.15494847e-01 -4.55553919e-01 -2.34605134e-01 8.53183687e-01
1.36627024e-02 8.27692568e-01 5.02191111e-02 -1.06018320e-01
-2.27118731e-01 4.29468125e-01 7.25231409e-01 -5.30418277e-01
6.54881299e-02 -1.85111165e-01 1.86040550e-01 -3.31649989e-01
-5.95962405e-01 -8.45891774e-01 -1.47359025e+00 -5.50036788e-01
-1.40593976e-01 2.94111878e-01 4.17834103e-01 1.27171397e+00
-4.40228552e-01 1.87512130e-01 6.25931565e-03 -1.96119800e-01
-8.62819374e-01 -7.41897821e-01 -9.07406569e-01 -1.71747729e-01
2.61334449e-01 -7.40640104e-01 -6.20575130e-01 1.52513355e-01] | [8.788926124572754, 6.643649101257324] |
bff7fd1a-abfd-4811-a55d-813cca52ca9f | contextual-response-interpretation-for | 2305.00577 | null | https://arxiv.org/abs/2305.00577v1 | https://arxiv.org/pdf/2305.00577v1.pdf | Contextual Response Interpretation for Automated Structured Interviews: A Case Study in Market Research | Structured interviews are used in many settings, importantly in market research on topics such as brand perception, customer habits, or preferences, which are critical to product development, marketing, and e-commerce at large. Such interviews generally consist of a series of questions that are asked to a participant. These interviews are typically conducted by skilled interviewers, who interpret the responses from the participants and can adapt the interview accordingly. Using automated conversational agents to conduct such interviews would enable reaching a much larger and potentially more diverse group of participants than currently possible. However, the technical challenges involved in building such a conversational system are relatively unexplored. To learn more about these challenges, we convert a market research multiple-choice questionnaire to a conversational format and conduct a user study. We address the key task of conducting structured interviews, namely interpreting the participant's response, for example, by matching it to one or more predefined options. Our findings can be applied to improve response interpretation for the information elicitation phase of conversational recommender systems. | ['Eugene Agichtein', 'Venugopal Vasudevan', 'Ankur Purwar', 'Kaustubh Dhole', 'Harshita Sahijwani'] | 2023-04-30 | null | null | null | null | ['marketing'] | ['miscellaneous'] | [ 5.99646389e-01 5.47242939e-01 -1.67501807e-01 -8.80056262e-01
-7.38024056e-01 -1.12321925e+00 3.17719072e-01 4.22149211e-01
-2.82198966e-01 3.77644747e-01 6.33030415e-01 -6.99701726e-01
5.74410753e-03 -5.27500570e-01 -9.00944024e-02 -3.52051228e-01
3.84640664e-01 8.89376879e-01 -1.86882108e-01 -3.48678410e-01
5.33930659e-01 1.28336206e-01 -1.23494816e+00 3.77220780e-01
5.03118217e-01 7.16987073e-01 3.83443356e-01 4.15521264e-01
-5.24282575e-01 5.64108908e-01 -2.84889847e-01 -6.44285917e-01
5.36553971e-02 -4.21244562e-01 -1.14781535e+00 5.57935834e-01
-1.64054200e-01 -5.05869269e-01 7.25890875e-01 9.35391605e-01
2.12617740e-01 5.93447387e-01 1.89809889e-01 -1.03437197e+00
-5.07511735e-01 1.05418503e+00 -3.00072044e-01 -2.63421535e-01
9.23360407e-01 -1.63071126e-01 1.31507504e+00 -8.63584876e-01
6.05151713e-01 1.36996174e+00 2.72508979e-01 4.38116789e-01
-1.26544988e+00 -5.72687387e-01 5.50453663e-01 -7.94497281e-02
-8.70359838e-01 -2.57919937e-01 7.80221879e-01 -6.41265035e-01
2.90338933e-01 5.20467579e-01 4.54996198e-01 1.06035292e+00
-2.53068656e-01 5.44968009e-01 9.96346533e-01 -2.48911127e-01
6.61542594e-01 8.89073372e-01 2.23259255e-01 -1.98746130e-01
-2.22956032e-01 -4.56023663e-01 -1.65371835e-01 -4.12993640e-01
4.17569846e-01 1.46504447e-01 -9.53331217e-03 -2.01061726e-01
-1.02220333e+00 1.26472425e+00 7.17637539e-02 3.49060237e-01
-7.22525001e-01 -4.80435312e-01 2.47677907e-01 4.94619787e-01
4.05728191e-01 8.28917801e-01 -4.75435346e-01 -2.36371636e-01
-4.19329971e-01 3.18741918e-01 1.39463341e+00 8.70870650e-01
7.79053152e-01 -7.37770140e-01 7.74430186e-02 8.77981544e-01
5.00070870e-01 1.04438946e-01 1.04505040e-01 -1.15773904e+00
1.27699375e-01 7.01960146e-01 8.47319305e-01 -1.11863279e+00
-4.03805643e-01 -2.54756749e-01 -2.23799273e-01 -1.19502358e-01
4.42835897e-01 -5.29927969e-01 -1.42525256e-01 1.21048439e+00
5.72337508e-01 -6.23190582e-01 -1.69836268e-01 1.29881334e+00
7.04453647e-01 5.52352011e-01 2.15055034e-01 -3.79998952e-01
1.74980843e+00 -6.21043563e-01 -6.51447952e-01 -5.27578115e-01
5.45802057e-01 -1.01117063e+00 1.18138897e+00 4.16357845e-01
-8.65013361e-01 -3.24658930e-01 -4.60476249e-01 1.12890102e-01
-3.22029330e-02 -3.89007092e-01 8.09969842e-01 7.96629369e-01
-7.23151922e-01 1.60604239e-01 -2.68644750e-01 -7.50358820e-01
-4.13084090e-01 5.97611427e-01 -6.28151819e-02 -1.73654541e-01
-1.24134147e+00 7.48590767e-01 -4.19211239e-01 2.60900557e-01
-1.31991178e-01 -2.57216454e-01 -7.19729841e-01 1.14763245e-01
6.06814444e-01 -3.84007603e-01 1.95918465e+00 -1.35512531e+00
-1.58635843e+00 5.83560109e-01 -2.20332876e-01 1.28309831e-01
6.28417879e-02 1.26800522e-01 -2.66069084e-01 -2.38031074e-01
2.93111086e-01 5.71700335e-01 3.90517831e-01 -1.25197089e+00
-6.67793155e-01 -4.46360677e-01 5.06709635e-01 4.83302057e-01
1.34598643e-01 4.52980161e-01 -2.05417931e-01 -4.06679660e-02
1.31232351e-01 -1.06889224e+00 -7.42514968e-01 -4.27229494e-01
9.76764932e-02 -3.45322907e-01 3.33054990e-01 -5.50424278e-01
8.02305281e-01 -2.14342809e+00 -1.66287467e-01 4.49384421e-01
1.85999051e-01 -8.50987732e-02 -2.01907024e-01 9.48809743e-01
6.28893003e-02 1.21438898e-01 2.31244430e-01 -1.38030291e-01
3.81089509e-01 -1.13749012e-01 -1.46475524e-01 -1.50272232e-02
-8.42492804e-02 5.63864112e-01 -1.09788096e+00 -1.11850686e-01
2.22373791e-02 1.91329330e-01 -6.43831134e-01 2.96542406e-01
-3.79453123e-01 7.13460743e-01 -7.14883327e-01 3.95165443e-01
4.75438416e-01 -1.45507410e-01 7.35856235e-01 -4.40099835e-02
-1.40961930e-01 7.52694786e-01 -1.14722264e+00 1.16590142e+00
-7.96627998e-01 4.84160185e-01 5.99927425e-01 -6.73336744e-01
1.20591915e+00 3.15047920e-01 5.37386417e-01 -5.25254250e-01
2.70302147e-01 -8.67999792e-02 2.24355191e-01 -5.49932301e-01
7.01925397e-01 -3.77694130e-01 -2.78292745e-01 1.02572632e+00
-5.59108794e-01 1.38058439e-01 2.77625192e-02 3.14959660e-02
9.45241928e-01 -2.85935581e-01 4.00091380e-01 -1.04606465e-01
4.23927248e-01 3.08233678e-01 5.11947095e-01 7.15094686e-01
-2.28253633e-01 1.75325155e-01 4.87208247e-01 -4.40345198e-01
-5.89147270e-01 -3.91140461e-01 1.91695020e-01 1.43577445e+00
5.49344942e-02 -2.58863479e-01 -6.40437722e-01 -4.94034111e-01
-2.52977848e-01 8.88425052e-01 -2.92612463e-01 3.01923037e-01
-2.07263634e-01 -5.15912361e-02 -3.18909198e-01 3.88383955e-01
8.12306851e-02 -1.05196047e+00 -5.00649035e-01 5.94588220e-01
-5.54963112e-01 -1.01578510e+00 -9.04622793e-01 -2.92308778e-02
-3.83341223e-01 -7.10630417e-01 -2.89498717e-01 -8.42222035e-01
6.90263987e-01 4.45568800e-01 9.32346880e-01 -1.18322648e-01
4.87171739e-01 6.01859152e-01 -6.63248897e-01 -3.64112049e-01
-4.23132062e-01 2.21046805e-01 -2.70809054e-01 2.14331254e-01
8.98234367e-01 -5.84142506e-01 -7.48184800e-01 9.41817224e-01
-8.74420583e-01 -9.40042064e-02 6.17514849e-01 3.15921634e-01
-5.67358732e-02 -1.68141872e-01 8.93343270e-01 -1.47407961e+00
1.17242265e+00 -6.55636609e-01 -4.65091079e-01 -4.47696038e-02
-5.37013471e-01 -2.35251755e-01 4.55224872e-01 -6.16529644e-01
-1.08359730e+00 2.35834435e-01 -3.81082505e-01 5.95446229e-01
-3.06426436e-01 9.36668158e-01 -2.53578007e-01 -8.75699371e-02
6.13838911e-01 -4.96818215e-01 3.80591713e-02 -2.50739694e-01
2.71115363e-01 1.23072267e+00 8.07975829e-02 -4.94646728e-01
7.17335701e-01 -6.25691339e-02 -7.04129875e-01 -5.61097205e-01
-5.73678017e-01 -8.74689758e-01 -3.51929158e-01 -4.85327810e-01
5.03840923e-01 -7.14136779e-01 -1.25086379e+00 -4.39897209e-01
-9.67902482e-01 -2.32444257e-01 1.32350409e-02 5.35447955e-01
-1.80595502e-01 -2.19146367e-02 -3.83570850e-01 -9.28155839e-01
-2.64926888e-02 -1.30796182e+00 7.81395495e-01 3.30866605e-01
-1.23587549e+00 -1.06313038e+00 1.47535733e-03 1.05736506e+00
6.21330142e-01 -2.36842811e-01 9.59168196e-01 -1.05451465e+00
-3.54571432e-01 -3.70594472e-01 6.47982880e-02 -2.89415926e-01
2.76522994e-01 -2.91501880e-01 -6.17791772e-01 -6.74450397e-02
2.28914618e-01 -1.57675222e-01 -4.72536273e-02 1.30484566e-01
5.42342067e-01 -4.07668144e-01 -2.35293865e-01 -4.28923517e-01
8.36294472e-01 6.54263973e-01 4.17324066e-01 2.09926501e-01
1.94427386e-01 1.36337602e+00 8.96865129e-01 4.79778141e-01
8.66941571e-01 5.22902668e-01 4.18364555e-02 -1.00803701e-02
8.40662241e-01 -2.04646170e-01 3.31906289e-01 6.81654334e-01
3.54505509e-01 -1.01147562e-01 -6.44424438e-01 5.94085574e-01
-1.79973519e+00 -6.18155599e-01 1.14949189e-01 2.04477096e+00
5.53174496e-01 -1.57403965e-02 4.96399969e-01 -3.95971611e-02
8.92317832e-01 -6.48197308e-02 -4.90662277e-01 -9.71007705e-01
6.80384755e-01 -2.47052610e-02 1.48886934e-01 6.58445060e-01
-4.62312669e-01 6.95041478e-01 6.12695789e+00 -2.20975935e-01
-9.59479034e-01 -2.85660833e-01 7.11268663e-01 2.17962936e-01
-1.01804829e+00 6.06075346e-01 -4.51569200e-01 9.08886418e-02
1.04041040e+00 -1.52960718e-01 7.64687240e-01 8.16189051e-01
5.71989179e-01 -2.72667974e-01 -1.51373148e+00 7.79536843e-01
-3.04216862e-01 -9.60396230e-01 -3.58132929e-01 1.39500573e-01
3.25708181e-01 -5.68904638e-01 -4.13479023e-02 3.90438408e-01
5.12769580e-01 -8.58876228e-01 2.39629745e-01 -1.65534198e-01
1.45079061e-01 -5.97332180e-01 7.73083448e-01 4.40517902e-01
-9.55996692e-01 -1.24852359e-01 -1.46865556e-02 -5.63814878e-01
2.42230430e-01 5.11687994e-01 -1.32978654e+00 -2.64547747e-02
2.65217930e-01 1.55140668e-01 -2.27887519e-02 5.60620785e-01
9.22303051e-02 5.80187261e-01 -9.99349877e-02 -5.37617981e-01
9.66104046e-02 -5.84597945e-01 1.69911951e-01 9.30889547e-01
1.19391888e-01 5.43906152e-01 2.71005660e-01 8.03979993e-01
-1.26428753e-01 4.21409667e-01 -4.09413040e-01 -3.75808418e-01
7.77985811e-01 1.61205995e+00 -9.11623418e-01 2.71275844e-02
-6.48766518e-01 6.64773345e-01 -1.15083270e-01 4.77311462e-01
-3.27387691e-01 -2.51286745e-01 6.87676430e-01 5.33397675e-01
1.84663638e-01 -1.61601100e-02 -1.71104565e-01 -7.10514545e-01
1.69763749e-03 -1.48642993e+00 8.34365338e-02 -7.10971713e-01
-1.14296007e+00 3.28699768e-01 -2.86792088e-02 -8.30806255e-01
-6.60597265e-01 -1.46039352e-01 -6.17387593e-01 8.03305030e-01
-8.24121594e-01 -5.70803761e-01 -1.80646837e-01 6.05687276e-02
6.26300871e-01 4.78775382e-01 7.44260192e-01 4.33014333e-01
-3.78438264e-01 1.31043360e-01 -3.91617596e-01 -1.99163761e-02
8.20282221e-01 -1.00861239e+00 4.89817053e-01 1.78093359e-01
-2.47534409e-01 9.48395610e-01 1.08667409e+00 -5.06575048e-01
-1.56633282e+00 -6.25925601e-01 1.29148734e+00 -1.74508885e-01
5.80485404e-01 -5.68647385e-01 -7.60238528e-01 8.43543053e-01
4.02841061e-01 -8.37216735e-01 1.26703620e+00 7.69544125e-01
1.72843382e-01 -2.29101088e-02 -1.17486286e+00 6.48874104e-01
6.05412960e-01 -6.77846432e-01 -4.76132005e-01 1.63086191e-01
6.41387224e-01 -2.64093757e-01 -9.32292759e-01 -2.31588840e-01
7.28094697e-01 -6.72230124e-01 3.35016131e-01 -3.14262122e-01
3.69520076e-02 -1.75362989e-01 -9.01103541e-02 -1.38865352e+00
-4.98527855e-01 -9.80386734e-01 1.01253974e+00 1.63891065e+00
8.05733681e-01 -7.69913316e-01 8.42221737e-01 1.61329031e+00
2.85417587e-01 -4.32355702e-01 -3.29435468e-02 1.54736727e-01
-3.77764165e-01 -3.51120412e-01 8.47240865e-01 9.47166085e-01
7.13896513e-01 8.68577778e-01 -1.12768181e-01 4.29279469e-02
1.87880546e-01 3.26632589e-01 7.67773211e-01 -1.36724257e+00
-2.44660944e-01 6.60889670e-02 2.22998455e-01 -1.18095851e+00
2.35767085e-02 -4.77853566e-01 3.68659824e-01 -1.57721210e+00
2.66555138e-02 -4.68547195e-01 1.51716009e-01 -5.79740182e-02
-1.05686888e-01 -2.89176136e-01 1.04567567e-02 3.05425320e-02
-5.69744051e-01 -3.18945535e-02 1.05870962e+00 2.81493496e-02
-7.03188539e-01 4.58554596e-01 -1.63723993e+00 5.97997129e-01
7.61647344e-01 -4.15840805e-01 -6.36525273e-01 -1.09499291e-01
7.32727468e-01 5.21622062e-01 -1.47693947e-01 -2.03255296e-01
4.95200813e-01 -5.74886680e-01 -3.19268554e-02 -3.36578578e-01
1.94293976e-01 -9.29412425e-01 5.78804612e-01 -1.77538499e-01
-7.92096317e-01 2.73085296e-01 -1.46878436e-01 2.78532684e-01
-2.93060124e-01 -4.04493004e-01 1.55788451e-01 -4.03556556e-01
-3.81637365e-01 -8.57066736e-02 -1.09152114e+00 -2.67786741e-01
8.11232328e-01 -3.12699556e-01 1.25836385e-02 -1.16072094e+00
-7.49170005e-01 4.89268988e-01 4.86200869e-01 3.92456502e-01
3.07020068e-01 -1.02845621e+00 -3.95839840e-01 -5.80861717e-02
9.86097753e-02 -5.20321429e-02 -2.72634067e-03 6.88519597e-01
1.03035513e-02 3.16696018e-01 -3.01448442e-02 -2.93325603e-01
-1.07239878e+00 3.98376286e-01 -2.45937668e-02 -6.16547950e-02
-6.17316626e-02 3.68498951e-01 3.08494687e-01 -6.39225662e-01
1.93166867e-01 -4.28525835e-01 -5.68023980e-01 4.38765585e-01
5.63583970e-01 6.51186407e-02 -3.97080816e-02 -4.10042465e-01
-3.60569417e-01 1.55924350e-01 -3.66521090e-01 -5.64537585e-01
1.25890911e+00 -5.58665693e-01 -6.30078390e-02 5.15133739e-01
8.93489003e-01 9.24786851e-02 -9.13196921e-01 -4.26317006e-01
3.07595611e-01 -4.75627869e-01 -1.52682558e-01 -6.73127532e-01
-6.46410882e-01 3.33809465e-01 2.97052544e-02 7.00774312e-01
9.59233820e-01 2.20620781e-01 8.77263725e-01 4.90696102e-01
3.07046384e-01 -9.91270483e-01 -6.54106885e-02 1.34992838e-01
7.26199269e-01 -1.31127572e+00 -4.22448397e-01 -5.60240269e-01
-9.85132337e-01 9.88243043e-01 3.07369590e-01 3.73051763e-01
3.83273304e-01 -7.82802049e-03 5.58351219e-01 -3.07355016e-01
-9.46354449e-01 -8.51790532e-02 5.56030199e-02 5.56150615e-01
9.26591873e-01 1.22269437e-01 -5.33403873e-01 7.15849102e-01
-3.25106949e-01 -9.55723226e-02 7.19477296e-01 8.75352859e-01
-5.01838684e-01 -1.46967566e+00 -4.03710753e-01 4.67217922e-01
-2.77647108e-01 1.80219170e-02 -9.52018321e-01 3.63639802e-01
-3.10195357e-01 1.76788509e+00 -6.14450015e-02 -4.73695576e-01
5.84318876e-01 -6.76936135e-02 8.58092681e-03 -9.45301533e-01
-1.03096747e+00 1.46253809e-01 6.86629236e-01 -3.35473806e-01
-5.04240096e-01 -1.05745769e+00 -9.70725298e-01 -5.08353591e-01
-3.70982409e-01 6.55785024e-01 8.93087566e-01 1.13663149e+00
1.62019297e-01 5.35795419e-03 1.00197303e+00 -5.82569659e-01
-5.85956454e-01 -8.52486312e-01 -3.95365506e-01 5.60991704e-01
2.97133595e-01 -2.01998293e-01 -2.53416181e-01 -4.77443077e-02] | [12.5521879196167, 7.778435707092285] |
091f4734-0bcd-4cd2-a395-2a55b0275dfa | clip4str-a-simple-baseline-for-scene-text-1 | 2305.14014 | null | https://arxiv.org/abs/2305.14014v1 | https://arxiv.org/pdf/2305.14014v1.pdf | CLIP4STR: A Simple Baseline for Scene Text Recognition with Pre-trained Vision-Language Model | Pre-trained vision-language models are the de-facto foundation models for various downstream tasks. However, this trend has not extended to the field of scene text recognition (STR), despite the potential of CLIP to serve as a powerful scene text reader. CLIP can robustly identify regular (horizontal) and irregular (rotated, curved, blurred, or occluded) text in natural images. With such merits, we introduce CLIP4STR, a simple yet effective STR method built upon image and text encoders of CLIP. It has two encoder-decoder branches: a visual branch and a cross-modal branch. The visual branch provides an initial prediction based on the visual feature, and the cross-modal branch refines this prediction by addressing the discrepancy between the visual feature and text semantics. To fully leverage the capabilities of both branches, we design a dual predict-and-refine decoding scheme for inference. CLIP4STR achieves new state-of-the-art performance on 11 STR benchmarks. Additionally, a comprehensive empirical study is provided to enhance the understanding of the adaptation of CLIP to STR. We believe our method establishes a simple but strong baseline for future STR research with VL models. | ['Yi Yang', 'Linchao Zhu', 'Xiaohan Wang', 'Shuai Zhao'] | 2023-05-23 | clip4str-a-simple-baseline-for-scene-text | https://arxiv.org/abs/2305.14014 | https://arxiv.org/pdf/2305.14014.pdf | null | ['scene-text-recognition'] | ['computer-vision'] | [ 4.95509326e-01 -3.16798955e-01 -3.60089630e-01 -4.08161551e-01
-8.26357782e-01 -5.87906182e-01 8.80898178e-01 -6.49202019e-02
-9.01709348e-02 1.22611642e-01 6.28903210e-01 -4.06273752e-01
5.43330610e-01 -4.47730571e-01 -8.98956418e-01 -5.12366474e-01
5.10734797e-01 1.08485743e-01 4.52737093e-01 2.15188973e-02
3.99241447e-01 1.60119101e-01 -1.49054217e+00 1.09799421e+00
6.02605581e-01 9.72882509e-01 5.33099771e-01 7.65730977e-01
-3.28381002e-01 1.15874994e+00 -1.61002606e-01 -4.82221335e-01
3.48947011e-02 -3.54126930e-01 -5.44776738e-01 2.64256090e-01
7.52266586e-01 -5.38372219e-01 -6.91630721e-01 6.86259747e-01
3.27776372e-01 -1.92122117e-01 6.97562993e-01 -9.66708660e-01
-8.95480156e-01 6.68147326e-01 -7.93808699e-01 5.06286956e-02
3.75792831e-01 1.75479934e-01 1.27124965e+00 -1.25580871e+00
6.63517773e-01 1.24673784e+00 6.83181345e-01 2.67938823e-01
-1.11967790e+00 -2.51199991e-01 4.12009597e-01 1.92390606e-01
-1.32811499e+00 -8.60202432e-01 5.32985270e-01 -6.57382190e-01
1.10549927e+00 2.76914448e-01 4.68239933e-01 1.52027571e+00
1.85259178e-01 1.44658053e+00 7.41415262e-01 -4.05449659e-01
-9.24871415e-02 -1.75659973e-02 6.41144812e-02 7.41525292e-01
5.75323142e-02 -1.41775027e-01 -9.73140657e-01 2.64285952e-01
5.71020603e-01 -2.17314780e-01 -5.17425239e-01 -2.95063406e-01
-1.21186483e+00 6.85661912e-01 1.97065890e-01 8.37640688e-02
6.71732649e-02 3.18290055e-01 4.47908789e-01 -7.82506540e-03
4.43756878e-01 9.30768922e-02 -1.63197473e-01 -3.01489592e-01
-1.30414343e+00 3.43675725e-02 5.43403149e-01 1.09241736e+00
4.41836834e-01 2.24500552e-01 -6.76404357e-01 9.18519855e-01
5.10110915e-01 6.37691081e-01 2.98862487e-01 -4.83487904e-01
8.39675725e-01 4.13652301e-01 -2.19800487e-01 -9.00114059e-01
-2.41850521e-02 -4.27848339e-01 -6.34274721e-01 -2.48227358e-01
2.43009701e-01 2.16487259e-01 -1.07344961e+00 1.24287653e+00
-1.25132456e-01 1.94189325e-01 -2.21941590e-01 7.64538765e-01
9.71492052e-01 7.52585769e-01 -8.36019963e-02 1.50636166e-01
1.27148950e+00 -1.36101151e+00 -5.85238278e-01 -5.49784660e-01
5.72746992e-01 -1.03966486e+00 1.11005819e+00 3.24509770e-01
-1.12367821e+00 -4.88136590e-01 -1.05866766e+00 -6.97031915e-01
-2.49227673e-01 5.79107940e-01 4.05200005e-01 3.67332309e-01
-1.11271822e+00 1.01087436e-01 -9.20743704e-01 -5.31463921e-01
5.80960691e-01 -1.98848858e-01 -5.20162582e-02 -3.38423729e-01
-7.84438491e-01 5.81441700e-01 1.24907330e-01 1.66770279e-01
-8.12671244e-01 -5.82608163e-01 -1.13009644e+00 1.83840189e-02
2.55311012e-01 -6.27141953e-01 1.25557458e+00 -9.04114306e-01
-1.38477707e+00 1.00026309e+00 -6.75746799e-01 -4.35143530e-01
6.30808830e-01 -2.64051527e-01 -2.23200589e-01 3.30979824e-01
1.72888562e-01 8.21369588e-01 1.22336388e+00 -1.33694148e+00
-7.03158557e-01 -1.72150970e-01 -3.33447427e-01 3.95500928e-01
-2.19683826e-01 6.71699177e-03 -1.18177497e+00 -7.85502791e-01
1.14378668e-01 -5.49948931e-01 3.92697960e-01 3.27797651e-01
-6.98166966e-01 3.09859738e-02 1.04270184e+00 -7.33014405e-01
1.16730380e+00 -2.42958999e+00 -4.22295816e-02 -2.63202757e-01
2.20675036e-01 1.38436198e-01 -1.31819189e-01 6.60023212e-01
1.77132577e-01 2.59699561e-02 -1.53931528e-02 -6.70552909e-01
6.82698041e-02 -7.54831880e-02 -9.34528351e-01 4.35259759e-01
2.86752731e-01 1.22601473e+00 -6.45855010e-01 -6.54247224e-01
5.07049143e-01 5.64604640e-01 -4.52201813e-01 6.47944659e-02
-4.70927536e-01 1.28290281e-01 -6.15206301e-01 8.01033080e-01
6.34762406e-01 -5.95488787e-01 -7.29690939e-02 -4.21039343e-01
-2.51069665e-01 3.22464764e-01 -7.81112373e-01 1.65143204e+00
1.38445776e-02 1.26269281e+00 -1.99554022e-02 -8.23636115e-01
6.95296049e-01 5.18821105e-02 2.04028189e-01 -9.28575099e-01
-3.41184158e-03 -1.91443041e-01 -5.00155270e-01 -6.60678744e-01
9.42850530e-01 3.04585159e-01 1.45154938e-01 2.25998104e-01
-2.01876760e-01 4.79085445e-02 1.21878505e-01 4.20444965e-01
9.77256238e-01 4.71972942e-01 -1.85336024e-02 9.82937813e-02
3.76538903e-01 4.81262542e-02 9.92296636e-02 1.11989450e+00
-1.58959940e-01 1.05268943e+00 5.83357155e-01 -2.87794173e-01
-1.07287192e+00 -1.28465796e+00 -3.35181415e-01 1.33151031e+00
2.68110812e-01 -6.23106182e-01 -5.03035188e-01 -5.58430195e-01
6.58973604e-02 8.21954846e-01 -5.52503586e-01 1.52780473e-01
-3.38150412e-01 -4.71427262e-01 9.28162575e-01 8.15493703e-01
6.52244687e-01 -7.32004166e-01 -4.15942192e-01 -1.66290000e-01
-4.47557747e-01 -1.61280131e+00 -6.04921460e-01 1.08008154e-01
-4.83997643e-01 -7.43494511e-01 -5.16360641e-01 -8.61564338e-01
4.59096432e-01 7.16944158e-01 1.13983846e+00 8.07783827e-02
-1.27999231e-01 6.14689767e-01 -5.07660508e-01 -1.30903855e-01
-3.10456544e-01 -1.59699067e-01 -5.42121649e-01 2.22388238e-01
2.38155738e-01 -2.03808397e-03 -4.41604733e-01 1.70648843e-01
-8.96079302e-01 7.82134116e-01 5.95940888e-01 7.71653831e-01
5.26377797e-01 -2.91464061e-01 1.95221808e-02 -6.15600884e-01
3.37396175e-01 -2.82202631e-01 -4.54345375e-01 3.71917725e-01
-2.70927221e-01 3.83694582e-02 5.28964281e-01 -1.26241624e-01
-1.19479060e+00 1.30100280e-01 -1.44463733e-01 -4.46426570e-01
-1.06068671e-01 7.72843242e-01 4.89446968e-02 2.49019310e-01
1.95095018e-01 7.85953820e-01 -4.60794687e-01 -4.54416245e-01
5.08437693e-01 9.07912433e-01 7.72803724e-01 -5.97335577e-01
6.33957088e-01 7.20813036e-01 -3.04981977e-01 -1.25106728e+00
-1.13903201e+00 -5.82393229e-01 -6.00446641e-01 -1.58510298e-01
1.01846170e+00 -1.32145572e+00 -3.31283569e-01 6.82941973e-01
-1.12921464e+00 -6.76220775e-01 1.08058415e-01 9.03379768e-02
-4.76476431e-01 6.78817689e-01 -7.21483946e-01 -6.17472351e-01
-2.07946777e-01 -1.13254356e+00 1.74021816e+00 1.57280713e-02
1.90717950e-01 -9.19783950e-01 -2.42489919e-01 6.44020736e-01
2.35570624e-01 -1.83295876e-01 9.02366877e-01 -4.51248676e-01
-8.78139853e-01 6.14317134e-02 -7.54926145e-01 8.91765803e-02
-2.71569312e-01 3.17211270e-01 -1.22249353e+00 -2.92021304e-01
-3.49634022e-01 -4.27580357e-01 1.26161742e+00 4.39468652e-01
1.32714665e+00 9.88244340e-02 -3.45086455e-01 9.06102300e-01
1.42754149e+00 -8.45704153e-02 9.47004139e-01 3.17876399e-01
1.05665779e+00 8.93003196e-02 2.96319157e-01 4.51873213e-01
7.58562684e-01 7.15530157e-01 2.41312057e-01 -1.26499578e-01
-4.86419857e-01 -6.76032662e-01 6.73046887e-01 7.60462165e-01
3.40393394e-01 -6.51870310e-01 -9.90758419e-01 3.88175786e-01
-1.86461616e+00 -9.91216540e-01 -3.71541291e-01 2.03759742e+00
7.36969709e-01 1.88541442e-01 -2.93408006e-01 -2.10162446e-01
5.83241105e-01 5.29465854e-01 -3.67510349e-01 -3.00500423e-01
-4.47084606e-01 -1.22073635e-01 4.92580950e-01 3.76744360e-01
-1.08746517e+00 1.31848705e+00 6.37442636e+00 8.88516784e-01
-1.37435305e+00 -2.90255934e-01 5.44964850e-01 4.53104004e-02
-2.55568683e-01 3.92706059e-02 -1.07248163e+00 4.09668684e-01
5.07228613e-01 1.22180775e-01 4.72022980e-01 7.50230193e-01
3.24928015e-01 -1.99960560e-01 -1.26359367e+00 1.09505534e+00
5.55182040e-01 -1.54214215e+00 3.40715736e-01 -1.24582872e-01
6.45757675e-01 4.74329203e-01 2.18207866e-01 2.84032375e-01
9.72703248e-02 -1.04685879e+00 1.02571177e+00 4.92984384e-01
1.05340767e+00 -1.84446037e-01 3.59786958e-01 2.18969434e-01
-1.47950590e+00 1.67688444e-01 -3.13770115e-01 1.23918876e-01
5.03368936e-02 4.01360035e-01 -7.09587395e-01 4.33839947e-01
6.73150837e-01 1.41621077e+00 -9.48926449e-01 9.05390680e-01
-2.99548984e-01 8.18612635e-01 3.90617102e-02 1.85670704e-01
3.74937743e-01 4.85123023e-02 5.19339979e-01 1.68867588e+00
3.53803746e-02 -4.22418475e-01 3.32105368e-01 1.01353276e+00
-8.10614675e-02 4.10633422e-02 -6.02108777e-01 -3.65847856e-01
3.78455251e-01 1.02497292e+00 -7.63745308e-01 -3.91516089e-01
-8.23017716e-01 1.30718613e+00 3.15856189e-01 6.30940855e-01
-8.48373175e-01 -1.68894276e-01 2.87642092e-01 1.22813799e-03
8.35345149e-01 -3.99777532e-01 -6.72126174e-01 -1.60525072e+00
7.88592771e-02 -8.54752302e-01 2.21556351e-01 -1.27535033e+00
-1.15928042e+00 1.99724331e-01 -2.82213300e-01 -1.01704681e+00
-3.24030332e-02 -8.39920104e-01 -4.11031693e-01 6.63192570e-01
-1.74550819e+00 -1.68442428e+00 -5.03445029e-01 6.88579142e-01
1.10982406e+00 -6.79508410e-03 4.45976108e-01 -1.38172545e-02
-7.35303938e-01 7.52415359e-01 5.01371741e-01 5.12447953e-01
7.60661602e-01 -1.04692698e+00 5.79539478e-01 1.14357972e+00
3.80674809e-01 2.85986423e-01 5.78184783e-01 -6.63404763e-01
-1.93927145e+00 -1.25435674e+00 8.51052403e-01 -6.48200154e-01
7.30570138e-01 -7.27166355e-01 -8.81424487e-01 1.00047946e+00
2.20975533e-01 -1.47216082e-01 2.40973562e-01 6.92799240e-02
-7.22960234e-01 1.37089103e-01 -4.53366399e-01 9.98450696e-01
8.72938931e-01 -7.54786670e-01 -5.85771620e-01 1.75285116e-01
6.84832811e-01 -7.22788692e-01 -4.84966934e-01 1.58849776e-01
6.90135896e-01 -1.14004683e+00 9.65407014e-01 -2.24012569e-01
1.00433230e+00 -1.35238364e-01 -3.91767859e-01 -7.25410521e-01
-2.46830940e-01 -4.05190498e-01 -1.83852389e-01 1.17973542e+00
3.49872231e-01 -2.78291553e-01 6.05172932e-01 1.90870926e-01
-4.69246656e-01 -7.17959344e-01 -6.52168930e-01 -5.90764701e-01
2.49640215e-02 -9.46053028e-01 1.00405894e-01 8.56409550e-01
-6.03187014e-04 5.43273211e-01 -5.27068734e-01 1.34259760e-01
5.42376518e-01 2.39210114e-01 8.48691463e-01 -7.94647157e-01
-3.24379712e-01 -6.21587038e-01 -2.10788146e-01 -1.88683236e+00
1.22311503e-01 -9.96577024e-01 2.18119234e-01 -1.73236477e+00
5.70062816e-01 1.99178476e-02 1.21794837e-02 4.86201972e-01
-1.69153482e-01 2.55903751e-01 3.41675192e-01 3.26404929e-01
-1.01553369e+00 6.16543412e-01 1.25693655e+00 -2.08627224e-01
-4.03152034e-03 -2.76462734e-01 -7.25678623e-01 7.03867912e-01
4.42730933e-01 2.10160892e-02 -3.83254111e-01 -9.87117946e-01
3.74724537e-01 -7.69670755e-02 5.04704297e-01 -6.73683107e-01
4.21067268e-01 -2.00578257e-01 5.78723848e-01 -1.16172135e+00
3.15288216e-01 -4.86063629e-01 -4.36146498e-01 -7.08612874e-02
-6.05923951e-01 -7.71917105e-02 2.51620680e-01 7.08053410e-01
-1.33053914e-01 -8.52584839e-02 6.52666688e-01 1.17531016e-01
-8.97294700e-01 1.73495546e-01 -5.01747012e-01 2.88151562e-01
5.12663245e-01 -4.79707599e-01 -8.81020188e-01 -5.16193926e-01
-1.04559757e-01 3.23726177e-01 5.93652546e-01 6.83862150e-01
8.30023110e-01 -9.05495346e-01 -8.01339865e-01 3.76037478e-01
4.56689417e-01 -4.82110307e-02 6.89303353e-02 9.92212772e-01
-3.95326823e-01 7.55733550e-01 2.59569049e-01 -9.99670029e-01
-1.13364553e+00 4.74245667e-01 1.67549610e-01 -9.58219469e-02
-1.01167750e+00 6.47795856e-01 5.79334199e-01 1.62738606e-01
5.25379539e-01 -4.02397513e-01 -1.03315245e-02 -2.49561802e-01
6.52916610e-01 4.32310849e-02 -5.65751456e-02 -8.09128881e-01
-1.19315833e-01 5.40318370e-01 -3.12490165e-01 -1.18641138e-01
1.09313369e+00 -4.86221015e-01 -8.32372159e-03 6.82965159e-01
1.04691458e+00 1.24599651e-01 -1.39484739e+00 -3.66815299e-01
-1.66668341e-01 -5.60402393e-01 3.01493764e-01 -9.02963936e-01
-8.30153346e-01 1.08873129e+00 7.44272992e-02 3.35862227e-02
1.03022456e+00 -1.39662288e-02 6.97988689e-01 3.23335081e-01
1.96113866e-02 -1.04463351e+00 2.95357764e-01 9.10591781e-01
8.44819307e-01 -1.24305665e+00 1.89592708e-02 -5.33769131e-01
-9.53261733e-01 1.21502662e+00 4.60462540e-01 2.45039254e-01
2.10266963e-01 4.87667769e-01 -8.96351561e-02 -1.48551524e-01
-1.03877115e+00 -3.63016814e-01 5.54623604e-01 5.46983540e-01
6.29731238e-01 -1.83582261e-01 9.08618644e-02 3.20513874e-01
-3.60251851e-02 -8.42818096e-02 4.95558918e-01 8.36818337e-01
-5.65025687e-01 -6.91701949e-01 -2.38122955e-01 5.01796186e-01
-3.09048265e-01 -6.13942087e-01 -4.25088793e-01 6.75792754e-01
-2.60827988e-01 9.80467200e-01 1.44023910e-01 -2.89630532e-01
1.02559291e-02 3.69282663e-02 4.23199028e-01 -4.87818152e-01
-3.10326159e-01 4.65595782e-01 1.59866109e-01 -5.58536410e-01
-4.72939201e-02 -6.65746748e-01 -1.01932561e+00 -2.11055472e-01
-1.93271235e-01 -5.03236949e-01 7.18536496e-01 1.09601939e+00
4.66126740e-01 3.60403299e-01 4.74047661e-01 -6.70390069e-01
-3.30601007e-01 -7.27717578e-01 -4.32827681e-01 3.13349754e-01
4.68085885e-01 -3.78129810e-01 -2.91644424e-01 5.77517986e-01] | [11.80916976928711, 2.0800909996032715] |
416e3545-9b9c-41fd-a654-f582ab1bc970 | overcoming-barriers-to-skill-injection-in | 2211.02098 | null | https://arxiv.org/abs/2211.02098v1 | https://arxiv.org/pdf/2211.02098v1.pdf | Overcoming Barriers to Skill Injection in Language Modeling: Case Study in Arithmetic | Through their transfer learning abilities, highly-parameterized large pre-trained language models have dominated the NLP landscape for a multitude of downstream language tasks. Though linguistically proficient, the inability of these models to incorporate the learning of non-linguistic entities (numerals and arithmetic reasoning) limits their usage for tasks that require numeric comprehension or strict mathematical reasoning. However, as we illustrate in this paper, building a general purpose language model that also happens to be proficient in mathematical reasoning is not as straight-forward as training it on a numeric dataset. In this work, we develop a novel framework that enables language models to be mathematically proficient while retaining their linguistic prowess. Specifically, we offer information-theoretic interventions to overcome the catastrophic forgetting of linguistic skills that occurs while injecting non-linguistic skills into language models. | ['Naren Ramakrishnan', 'Nikhil Muralidhar', 'Mandar Sharma'] | 2022-11-03 | null | null | null | null | ['mathematical-reasoning', 'arithmetic-reasoning'] | ['natural-language-processing', 'reasoning'] | [ 1.56436488e-01 5.98147213e-01 -8.98171142e-02 -2.91168362e-01
-5.47587872e-01 -8.37461531e-01 5.51496387e-01 5.51150739e-01
-5.61680973e-01 6.90196157e-01 3.31668615e-01 -1.05925155e+00
-2.93556958e-01 -1.25058019e+00 -9.66067970e-01 8.36748779e-02
1.50714844e-01 4.03965741e-01 -2.92746335e-01 -4.75066692e-01
2.11867437e-01 6.21475697e-01 -1.11425245e+00 3.13481122e-01
1.47574663e+00 3.14592123e-01 1.06528699e-01 4.51384306e-01
-3.76461387e-01 1.38598537e+00 -2.05400214e-01 -9.16630924e-01
1.30562514e-01 -2.69661844e-01 -8.01243246e-01 -6.54026151e-01
3.72385055e-01 -5.24165988e-01 -2.81914860e-01 8.32442045e-01
2.99837738e-02 1.43087074e-01 9.83572721e-01 -9.00914013e-01
-1.21595454e+00 1.15374768e+00 -1.61673296e-02 1.08655319e-01
5.70470095e-01 2.04302132e-01 1.12140119e+00 -7.45411813e-01
4.23548639e-01 1.34127367e+00 9.67219830e-01 4.96430606e-01
-1.36236262e+00 -6.46812916e-01 9.59320441e-02 -2.73699969e-01
-1.08927178e+00 -2.94306159e-01 5.42544007e-01 -5.38167953e-01
1.26995516e+00 -1.95555925e-01 8.23452711e-01 8.47797811e-01
3.14676166e-01 9.07669067e-01 1.25740755e+00 -7.34388590e-01
-7.80307874e-03 3.97015870e-01 3.20156455e-01 1.02545440e+00
5.93373120e-01 2.23341107e-01 -8.41569901e-01 3.06091994e-01
8.16653490e-01 -2.89898664e-01 4.26260568e-02 -2.37837464e-01
-1.08291245e+00 8.25599849e-01 2.85083890e-01 5.39099157e-01
-1.75611496e-01 1.56862378e-01 2.88699836e-01 7.23344922e-01
2.51015216e-01 1.08261371e+00 -6.89760625e-01 -1.18506260e-01
-9.17583704e-01 8.88743997e-02 9.13776338e-01 9.56639647e-01
4.87692118e-01 1.54394776e-01 -1.83980986e-01 3.61856252e-01
3.58936191e-01 3.94519031e-01 3.17572087e-01 -9.30071056e-01
6.54793322e-01 8.30864251e-01 -9.55586433e-02 -9.15154219e-01
-4.04915810e-01 -5.44219434e-01 -5.82437336e-01 2.45697033e-02
1.05718493e+00 -1.39642835e-01 -6.74634933e-01 2.09462047e+00
-1.58673912e-01 -2.48570427e-01 3.21870804e-01 3.62676859e-01
5.22037864e-01 6.79317534e-01 7.97195613e-01 -3.76417860e-02
1.07158840e+00 -6.38928711e-01 -3.68857175e-01 -5.65582871e-01
1.04909062e+00 -3.49755198e-01 1.56376231e+00 3.01090002e-01
-1.56458735e+00 -4.84077126e-01 -9.32834506e-01 -7.95676887e-01
-7.81883359e-01 -1.69409648e-01 1.28480148e+00 9.65361059e-01
-1.36059844e+00 7.31294811e-01 -7.16409802e-01 -1.75708994e-01
4.69359607e-01 3.66035312e-01 -2.88292289e-01 -2.24456325e-01
-1.58699548e+00 1.48360777e+00 5.42818606e-01 -1.51501477e-01
-6.11746967e-01 -1.32043755e+00 -1.18812299e+00 5.31925499e-01
1.35759771e-01 -1.13601816e+00 1.18615580e+00 -9.46192384e-01
-1.53975213e+00 1.02432632e+00 1.43900767e-01 -5.49640536e-01
5.45873165e-01 -1.70386791e-01 1.06985919e-01 -1.13466624e-02
-9.13423970e-02 5.42939126e-01 4.02900249e-01 -9.20655966e-01
-2.45987445e-01 -2.70385832e-01 6.03260159e-01 4.82959598e-01
-5.21673381e-01 4.90508601e-02 1.01499625e-01 -7.18709826e-01
-4.27252799e-02 -3.02708626e-01 -1.37306064e-01 1.02054849e-01
-1.18742704e-01 -2.66381264e-01 -3.08219731e-01 -8.36355031e-01
1.17385972e+00 -1.74000657e+00 2.03684881e-01 3.11181873e-01
2.40097776e-01 2.68923372e-01 -2.09634274e-01 4.61900353e-01
1.18659608e-01 4.95379031e-01 -6.53564185e-02 -1.95014119e-01
4.51856822e-01 -2.54211463e-02 -5.51211536e-01 8.75337124e-02
4.00846273e-01 1.41505611e+00 -1.08853245e+00 -3.98989379e-01
1.43982306e-01 2.76732594e-01 -1.06658745e+00 2.52314620e-02
-3.64747792e-01 3.31922710e-01 -3.37221891e-01 5.73269188e-01
3.07474613e-01 -2.13946238e-01 2.11046979e-01 3.10690492e-01
4.04120386e-02 6.31777585e-01 -5.51846862e-01 1.69078207e+00
-9.64108586e-01 3.74078304e-01 -1.48775935e-01 -9.48494256e-01
3.69090289e-01 -1.80198401e-02 -3.75372032e-03 -6.07931018e-01
4.26767580e-02 3.94128054e-01 5.20609915e-01 -4.26942378e-01
1.68337047e-01 -8.08355331e-01 -2.08349988e-01 6.14219308e-01
2.10725620e-01 -7.06473529e-01 3.92465383e-01 4.13999915e-01
8.22586954e-01 3.66036117e-01 3.93475026e-01 -3.69729966e-01
6.94764078e-01 1.60803229e-01 1.53505564e-01 9.10831153e-01
-8.67768750e-02 -2.73035407e-01 3.67818028e-01 7.84695670e-02
-1.15199065e+00 -1.28046334e+00 -4.89512049e-02 1.51250136e+00
-4.57638294e-01 -2.97107428e-01 -7.42637336e-01 -1.86149120e-01
1.40041187e-01 1.37215805e+00 -2.51587361e-01 -6.74131215e-01
-6.06808484e-01 -3.24047536e-01 9.52952564e-01 6.32731676e-01
4.36039120e-01 -1.07143927e+00 -3.18714917e-01 2.08079323e-01
2.95814872e-01 -9.04605746e-01 -5.26230484e-02 1.09164178e-01
-9.32435572e-01 -5.57717264e-01 -7.24087358e-01 -9.61117506e-01
7.60029972e-01 -3.41068178e-01 1.30247200e+00 3.12155992e-01
8.87600239e-03 5.66032588e-01 1.40042409e-01 -5.49023271e-01
-8.01625550e-01 2.42694780e-01 1.04831971e-01 -7.47389317e-01
7.12937772e-01 -6.40186727e-01 -1.26625318e-02 -5.42523324e-01
-7.56770372e-01 3.83243740e-01 6.50784016e-01 6.53866529e-01
1.27332732e-01 3.16965617e-02 8.11029017e-01 -9.55818057e-01
8.89788985e-01 -4.91799772e-01 -5.43242276e-01 4.93867248e-01
-4.56780940e-01 3.43119562e-01 1.05477464e+00 -4.79294330e-01
-1.33466220e+00 -1.24397829e-01 2.00712886e-02 1.45051539e-01
5.52965812e-02 1.05480456e+00 -1.30500449e-02 -4.43930253e-02
6.54909134e-01 4.71117884e-01 -1.80051476e-02 -2.68683523e-01
6.41777575e-01 9.68008488e-02 3.39800179e-01 -1.28456581e+00
1.23766184e+00 -1.16527297e-01 1.07811298e-03 -7.67669559e-01
-1.26260746e+00 5.43581247e-02 -6.43239558e-01 1.84963554e-01
7.82733142e-01 -1.10991216e+00 -8.64858806e-01 4.79105651e-01
-1.01336813e+00 -8.11931133e-01 -4.39113468e-01 5.16537666e-01
-9.17201221e-01 2.16952905e-01 -8.03844154e-01 -8.54245484e-01
-1.70478541e-02 -8.02853703e-01 4.63984936e-01 2.47634292e-01
-5.36322057e-01 -1.55989420e+00 -2.88678825e-01 3.92204106e-01
5.06439626e-01 -1.67109966e-01 1.75885391e+00 -8.15078914e-01
-7.46892154e-01 5.75773977e-02 -2.62287468e-01 2.94311285e-01
-9.75227207e-02 -2.82704234e-01 -7.26080537e-01 8.29171613e-02
2.46000639e-03 -8.15351725e-01 7.09333718e-01 7.82136545e-02
9.41513598e-01 -2.63931066e-01 -7.06753433e-02 6.85486853e-01
1.20906734e+00 -2.23254296e-03 3.34046781e-01 2.64708400e-01
4.17785376e-01 8.11011553e-01 1.26069084e-01 -2.05914810e-01
8.76902103e-01 -4.23171483e-02 -4.68686581e-01 1.07700624e-01
-8.69166031e-02 -9.69125390e-01 4.91959900e-01 8.02896678e-01
-4.95301094e-03 -2.76575573e-02 -1.44978166e+00 5.51355720e-01
-1.32554448e+00 -8.86432767e-01 2.49393091e-01 2.05437207e+00
1.75593102e+00 2.74839669e-01 -2.28617653e-01 -7.15784878e-02
5.73886037e-02 -3.98872375e-01 -5.51204205e-01 -6.17670178e-01
3.18402564e-03 5.36719322e-01 2.31206611e-01 7.73546398e-01
-7.37342954e-01 1.49488604e+00 6.77631521e+00 4.94715959e-01
-6.41008973e-01 -1.84064746e-01 6.44989133e-01 3.44787925e-01
-8.10483336e-01 -1.54589802e-01 -7.80092180e-01 4.31720447e-03
1.49543047e+00 -5.90950251e-01 7.45446146e-01 3.67454380e-01
2.16781013e-02 -1.84473962e-01 -1.58600438e+00 5.53335071e-01
-1.87586382e-01 -1.18127465e+00 4.11509514e-01 -3.83045852e-01
5.73057771e-01 -4.62949783e-01 3.72305632e-01 1.05570436e+00
7.16691792e-01 -1.73551118e+00 7.77950406e-01 6.28888905e-01
8.51264298e-01 -7.28212774e-01 2.20248297e-01 8.59693766e-01
-6.60069644e-01 -3.08503628e-01 -2.81855792e-01 -6.21489644e-01
5.26756607e-02 3.32546502e-01 -5.35091281e-01 2.21241072e-01
3.26006897e-02 4.46204275e-01 -7.91499317e-01 5.64025521e-01
-6.45092249e-01 4.79605436e-01 -3.46663654e-01 -1.64088428e-01
1.67386606e-01 -2.51276553e-01 -5.65960668e-02 9.92292464e-01
2.90994406e-01 5.47455668e-01 -2.39752065e-02 1.23428571e+00
-1.95271775e-01 3.62140030e-01 -9.05858576e-01 -7.10103810e-01
4.50841665e-01 5.24933159e-01 -4.46711898e-01 -3.99141163e-01
-6.91295385e-01 6.86001956e-01 7.78513432e-01 3.59945953e-01
-4.70431030e-01 -2.22459972e-01 3.91211450e-01 7.61753544e-02
-3.28522950e-01 -5.70125282e-01 -9.32465911e-01 -1.39824867e+00
-1.88731492e-01 -9.55524862e-01 1.32912919e-01 -8.79791617e-01
-1.42089760e+00 -7.36427307e-02 1.43448740e-01 -4.41023171e-01
-4.42463756e-01 -1.08948600e+00 -2.37389833e-01 1.24494934e+00
-1.55788076e+00 -1.29303205e+00 4.20920908e-01 5.11414111e-01
2.31146201e-01 -2.58673698e-01 9.66771245e-01 2.95321904e-02
-2.15208024e-01 8.37023199e-01 -1.07711621e-01 3.93924713e-01
3.15572262e-01 -1.30359674e+00 3.08553636e-01 8.34943950e-01
9.02169105e-03 1.59199667e+00 5.69711983e-01 -6.83736026e-01
-1.65424979e+00 -7.98483074e-01 1.53055632e+00 -7.02433407e-01
1.03210211e+00 -4.80119556e-01 -8.34661067e-01 1.22625375e+00
-5.21718757e-03 -4.71060693e-01 7.98692822e-01 1.98628306e-01
-5.65537095e-01 2.90091127e-01 -1.21179438e+00 1.01104760e+00
1.27552104e+00 -9.08371329e-01 -1.28103256e+00 5.08910298e-01
8.90073001e-01 -4.55851167e-01 -9.15586174e-01 9.23770890e-02
2.38397732e-01 -4.40268815e-01 1.28840494e+00 -1.06987095e+00
8.25644076e-01 1.35555893e-01 9.73296538e-02 -1.31457865e+00
-1.45885929e-01 -6.39351606e-01 -2.47226894e-01 1.15479672e+00
6.98645413e-01 -6.72093213e-01 5.55785537e-01 1.05045784e+00
-1.53917260e-02 -4.41725016e-01 -5.69209099e-01 -7.56736517e-01
1.22065592e+00 -5.73264718e-01 2.74226785e-01 1.10861623e+00
4.95407552e-01 4.68362242e-01 2.57934123e-01 3.36020999e-02
3.54802459e-01 4.60908040e-02 4.08589363e-01 -1.40478957e+00
-2.55455464e-01 -6.94904387e-01 -2.26481847e-04 -9.81306314e-01
7.60932326e-01 -1.58305228e+00 -2.19653666e-01 -1.50601029e+00
1.01502128e-01 -6.12292469e-01 -2.18122806e-02 6.97166145e-01
-5.80789968e-02 -2.06921265e-01 2.17522770e-01 -4.03066605e-01
-1.17019147e-01 2.97775328e-01 1.24610150e+00 4.34092283e-02
-9.73326117e-02 -1.20221525e-01 -1.17578542e+00 1.05563891e+00
8.32030594e-01 -2.60613531e-01 -7.18641341e-01 -5.56596577e-01
8.49907577e-01 1.21458292e-01 2.27149829e-01 -7.22569287e-01
3.09082776e-01 -3.93610120e-01 4.64968443e-01 -5.50631918e-02
5.38803339e-02 -7.12683141e-01 -4.80409563e-01 6.09685123e-01
-7.56879628e-01 -3.84761356e-02 6.68538690e-01 1.61481604e-01
1.00368328e-01 -4.72585112e-01 6.63112223e-01 -5.50777793e-01
-4.85887229e-01 -1.05532594e-02 -5.74382424e-01 4.12482649e-01
8.31852078e-01 -4.50051390e-02 -1.92612410e-01 -3.10464203e-01
-8.02995980e-01 2.15187356e-01 5.28160989e-01 1.32508680e-01
4.07201111e-01 -9.19946671e-01 -6.05889380e-01 9.77696553e-02
-2.35170409e-01 5.99955069e-03 4.17593531e-02 5.50713420e-01
-7.58495986e-01 8.81107748e-01 -2.22213224e-01 1.95731699e-01
-5.97704351e-01 9.40373778e-01 4.81222302e-01 -5.86877167e-01
-3.62496346e-01 1.09260798e+00 2.01094896e-01 -8.55652034e-01
2.48898700e-01 -7.52662301e-01 -1.60970256e-01 5.36958613e-02
4.60915297e-01 2.87704561e-02 -2.08536312e-01 -1.44089878e-01
-1.18545488e-01 4.27816331e-01 2.94521451e-02 -1.53044894e-01
1.36701536e+00 -9.08939987e-02 -1.78815067e-01 5.26239157e-01
7.07145572e-01 2.87655115e-01 -1.00553226e+00 -2.45832518e-01
2.82255381e-01 4.15846407e-02 -1.33362070e-01 -1.24940705e+00
-4.32362467e-01 1.22415388e+00 -2.90108919e-01 -2.09450603e-01
8.77365649e-01 -2.32334226e-01 4.45689410e-01 8.66186678e-01
5.19479334e-01 -9.13316488e-01 -1.30965754e-01 1.00773215e+00
7.01055765e-01 -1.04029667e+00 -5.17383143e-02 -4.71207023e-01
-3.28053027e-01 1.20618868e+00 7.38807142e-01 1.00593297e-02
4.97395337e-01 3.56176287e-01 -2.81127095e-01 5.00165566e-04
-7.71905184e-01 -2.59229511e-01 1.64274544e-01 6.46219850e-01
8.76714885e-01 -2.51462221e-01 -4.08295780e-01 8.81238103e-01
-7.96258032e-01 3.77108276e-01 6.23036146e-01 9.00081217e-01
-3.58012885e-01 -8.13795805e-01 -1.31919235e-01 2.80206650e-01
-4.31756616e-01 -8.72717619e-01 -2.68979728e-01 9.02694166e-01
5.17828502e-02 6.13508344e-01 2.48168837e-02 1.86166361e-01
1.96758494e-01 7.38116384e-01 8.98691356e-01 -1.03933370e+00
-8.07271361e-01 -7.35142529e-01 3.44360434e-02 -9.24514830e-02
-1.41267329e-01 -5.21350980e-01 -1.52190745e+00 -6.67986691e-01
3.49846482e-01 -2.59727657e-01 1.82968408e-01 1.19315076e+00
-6.55616075e-02 4.72514957e-01 -1.39582351e-01 -3.43788236e-01
-1.06522846e+00 -7.84685731e-01 -3.43556434e-01 1.28385112e-01
2.70913213e-01 -3.57568890e-01 -3.76133591e-01 3.08987554e-02] | [9.574441909790039, 7.346804141998291] |
4fc9c0de-ffdb-47ac-9790-ddd888b7e4ef | struct-mdc-mesh-refined-unsupervised-depth | 2204.13877 | null | https://arxiv.org/abs/2204.13877v1 | https://arxiv.org/pdf/2204.13877v1.pdf | Struct-MDC: Mesh-Refined Unsupervised Depth Completion Leveraging Structural Regularities from Visual SLAM | Feature-based visual simultaneous localization and mapping (SLAM) methods only estimate the depth of extracted features, generating a sparse depth map. To solve this sparsity problem, depth completion tasks that estimate a dense depth from a sparse depth have gained significant importance in robotic applications like exploration. Existing methodologies that use sparse depth from visual SLAM mainly employ point features. However, point features have limitations in preserving structural regularities owing to texture-less environments and sparsity problems. To deal with these issues, we perform depth completion with visual SLAM using line features, which can better contain structural regularities than point features. The proposed methodology creates a convex hull region by performing constrained Delaunay triangulation with depth interpolation using line features. However, the generated depth includes low-frequency information and is discontinuous at the convex hull boundary. Therefore, we propose a mesh depth refinement (MDR) module to address this problem. The MDR module effectively transfers the high-frequency details of an input image to the interpolated depth and plays a vital role in bridging the conventional and deep learning-based approaches. The Struct-MDC outperforms other state-of-the-art algorithms on public and our custom datasets, and even outperforms supervised methodologies for some metrics. In addition, the effectiveness of the proposed MDR module is verified by a rigorous ablation study. | ['Hyun Myung', 'Dong-Uk Seo', 'Hyunjun Lim', 'Jinwoo Jeon'] | 2022-04-29 | null | null | null | null | ['depth-completion'] | ['computer-vision'] | [-4.40366287e-03 -6.09992146e-02 -2.92394489e-01 -4.39374447e-01
-4.90149558e-01 -1.41846627e-01 5.79042017e-01 2.97950774e-01
-3.80140066e-01 8.11537921e-01 2.48619020e-01 2.77466804e-01
9.52168740e-03 -1.19422901e+00 -7.16702461e-01 -4.39863175e-01
-6.67629167e-02 6.68621361e-01 3.02633941e-01 -5.80329373e-02
5.83018899e-01 5.33938229e-01 -1.91070747e+00 1.71716213e-01
1.16371047e+00 1.14045012e+00 5.93800724e-01 -5.65380864e-02
-6.74977779e-01 4.45556074e-01 -1.27505839e-01 2.33585626e-01
3.25126141e-01 5.58014400e-02 -6.22318149e-01 8.92746150e-02
5.21433651e-01 -5.23922861e-01 -3.90142292e-01 1.06217730e+00
2.16230258e-01 2.23926269e-02 4.26463842e-01 -1.14587438e+00
-1.70158401e-01 -4.12332676e-02 -9.02061224e-01 -4.44154471e-01
7.09835887e-01 -1.71033159e-01 7.26933002e-01 -1.27914023e+00
8.39565039e-01 1.35335481e+00 7.98646510e-01 2.15731040e-01
-9.67876613e-01 -6.87465191e-01 6.31817132e-02 2.13423848e-01
-1.57507634e+00 -2.56372243e-01 9.28878009e-01 -3.85013103e-01
9.09366250e-01 -1.45475209e-01 1.02113581e+00 8.02077770e-01
3.17848295e-01 6.15939617e-01 9.76216972e-01 -2.83602506e-01
3.83780599e-01 -9.28869545e-02 -3.01546156e-01 9.38999176e-01
4.26931411e-01 1.47679031e-01 -1.03689766e+00 -2.27332592e-01
1.16727865e+00 3.05728912e-01 -4.74235237e-01 -7.86287606e-01
-1.41439855e+00 1.04916275e+00 7.43813753e-01 -5.93264215e-02
-3.03706706e-01 1.92897528e-01 1.24317475e-01 -8.61530541e-04
4.88595605e-01 3.01320642e-01 -2.28068471e-01 -9.14091095e-02
-9.93169308e-01 1.95929408e-01 6.00336134e-01 1.24640560e+00
1.62291169e+00 -1.41761556e-01 5.20755649e-01 6.33066714e-01
5.46270072e-01 5.53369939e-01 3.21291834e-01 -1.06858110e+00
5.97891927e-01 8.71624112e-01 1.04363687e-01 -1.24421954e+00
-5.04023671e-01 -1.81252379e-02 -1.09480071e+00 3.95128071e-01
7.95275345e-02 1.53478265e-01 -9.53595757e-01 1.15537488e+00
6.01674020e-01 2.80717105e-01 1.35900721e-01 9.99036908e-01
9.97554660e-01 6.01523399e-01 -4.63640809e-01 -1.08436748e-01
8.63330483e-01 -7.23620892e-01 -6.41634524e-01 -5.67882121e-01
5.08598745e-01 -5.31784892e-01 6.87145948e-01 4.49528694e-01
-5.97876072e-01 -4.02173012e-01 -1.23618031e+00 -5.98678887e-01
-3.23772244e-02 -1.76897466e-01 9.88906801e-01 -2.29316975e-05
-9.93388474e-01 4.78162259e-01 -8.62747788e-01 -2.55495012e-01
3.74981225e-01 1.84505075e-01 -7.79302001e-01 -5.58089435e-01
-9.34763253e-01 6.06743038e-01 3.18291217e-01 9.44863558e-02
-7.29831219e-01 -6.93082750e-01 -1.46859837e+00 -5.08744061e-01
1.21015497e-01 -7.67661095e-01 6.87289417e-01 -6.27312005e-01
-1.17810345e+00 7.09617555e-01 -6.90642059e-01 -2.71151006e-01
6.06499553e-01 -1.92934200e-01 2.91300952e-01 2.67433614e-01
4.61697191e-01 9.83756781e-01 6.36839747e-01 -1.44096255e+00
-9.76591885e-01 -5.05042970e-01 3.60139157e-03 4.40824777e-01
5.49593940e-02 -9.94116724e-01 -5.40340960e-01 -2.03737929e-01
1.14910305e+00 -6.34925663e-01 -3.53271276e-01 6.21713698e-01
-1.15195796e-01 2.04590317e-02 8.21126640e-01 -3.07956427e-01
7.03543305e-01 -2.20166945e+00 2.84527153e-01 3.34972739e-01
4.26794827e-01 -4.61846560e-01 1.66683152e-01 4.77498531e-01
6.37976587e-01 -2.42709860e-01 -2.52366453e-01 -8.88753772e-01
-3.39282125e-01 4.82916802e-01 -2.23160982e-01 8.64616990e-01
-1.26578957e-01 4.21907723e-01 -1.07146859e+00 -6.54922605e-01
6.11928284e-01 5.68998933e-01 -7.34903038e-01 4.59653810e-02
-1.56264991e-01 7.19036400e-01 -5.42731106e-01 1.00892556e+00
1.18763924e+00 -4.35324460e-02 -1.07442178e-01 -2.38545805e-01
-5.81141055e-01 1.05978854e-01 -1.26774645e+00 2.61248612e+00
-5.08854508e-01 5.85941017e-01 1.28497601e-01 -6.00230038e-01
1.31945717e+00 -1.73139721e-01 5.58837414e-01 -8.15079331e-01
-1.27111897e-01 5.32291055e-01 -5.85936427e-01 -1.76126778e-01
7.23390520e-01 6.23523444e-02 1.03702180e-01 -2.30185643e-01
-2.22868547e-01 -6.08155727e-01 -4.75663096e-01 1.26194611e-01
9.94942784e-01 2.89688528e-01 4.01843458e-01 -2.07734048e-01
2.70767987e-01 4.52477247e-01 9.76858616e-01 4.56146151e-01
1.26384109e-01 8.87670815e-01 1.66097462e-01 -6.72841251e-01
-9.85403061e-01 -7.52475142e-01 -3.43061209e-01 7.71517633e-03
9.24780846e-01 -4.99277949e-01 1.56681333e-02 -2.32430369e-01
5.10262728e-01 7.83332735e-02 -6.07093573e-01 1.61628783e-01
-3.94473076e-01 -7.25676119e-02 7.55753443e-02 3.77654672e-01
9.66759562e-01 -7.96119690e-01 -8.86329651e-01 2.34867632e-01
-4.40110654e-01 -1.13939142e+00 4.21315096e-02 1.62473038e-01
-1.05409908e+00 -1.06497967e+00 -4.88045126e-01 -8.20721686e-01
7.39635587e-01 5.91551602e-01 8.02772522e-01 5.79394177e-02
-2.48110875e-01 2.95206979e-02 -4.83309507e-01 -2.15482011e-01
2.54008412e-01 2.37038229e-02 7.51088932e-02 -1.77929655e-01
2.79342592e-01 -8.07341874e-01 -6.85291171e-01 2.09227443e-01
-5.58419228e-01 4.83386785e-01 5.86499572e-01 6.89760625e-01
1.08862281e+00 -1.99148923e-01 1.63582578e-01 -5.56277871e-01
-5.86561933e-02 -6.10090494e-01 -7.79134631e-01 -2.88153052e-01
-3.50236714e-01 2.01972172e-01 2.44563758e-01 -3.14879008e-02
-7.41788805e-01 5.48330069e-01 -1.14873849e-01 -7.21847951e-01
1.11068800e-01 6.63421750e-01 -1.90443635e-01 -5.27037978e-01
4.79277849e-01 3.38222951e-01 9.00568962e-02 -5.81197977e-01
2.84650493e-02 5.28369188e-01 3.93385768e-01 -3.76307219e-01
7.25644588e-01 1.14844251e+00 5.30505180e-01 -9.23905253e-01
-6.81801736e-01 -7.45482087e-01 -6.37348890e-01 2.30361596e-02
5.26033938e-01 -1.33386981e+00 -4.98212010e-01 5.08154809e-01
-1.33417284e+00 -1.27829686e-01 -1.07809260e-01 6.43033564e-01
-5.61548352e-01 7.23622859e-01 -2.68995434e-01 -7.46723890e-01
-1.42202780e-01 -1.15144336e+00 1.34511113e+00 1.07743829e-01
2.68341396e-02 -8.79988134e-01 8.97541568e-02 6.30061850e-02
8.82088393e-02 7.67206788e-01 5.71577072e-01 2.84211904e-01
-1.04509377e+00 4.58963290e-02 -3.17702532e-01 -1.30255193e-01
2.51867473e-01 -3.11005116e-01 -8.74543428e-01 -3.99889052e-01
-5.84378056e-02 -2.97341436e-01 9.32258725e-01 2.93804169e-01
8.03198576e-01 4.56051789e-02 -5.87585688e-01 1.29866517e+00
1.72652495e+00 -1.74925312e-01 7.05415547e-01 6.55449867e-01
1.07127810e+00 6.33222699e-01 1.15041900e+00 9.09098446e-01
6.83311582e-01 5.55627227e-01 9.78012860e-01 -1.77199528e-01
-8.90659466e-02 -6.20162368e-01 3.32322507e-03 4.55833077e-01
9.75985974e-02 1.73558816e-01 -1.14725304e+00 7.17302620e-01
-1.95401013e+00 -5.04432619e-01 -4.31412965e-01 2.19779801e+00
7.79034972e-01 -5.81326708e-02 -5.61842084e-01 1.50378317e-01
3.14701498e-01 2.07806706e-01 -5.84252715e-01 6.72911108e-02
-2.32947737e-01 3.13763022e-02 7.44791031e-01 9.13378060e-01
-7.94673681e-01 9.39268053e-01 5.00735760e+00 6.17753148e-01
-1.08489335e+00 -1.40968323e-01 -1.53888926e-01 8.10319036e-02
-6.97472751e-01 2.74413168e-01 -1.04500794e+00 2.69942611e-01
1.04947917e-01 -1.57052297e-02 1.54422924e-01 1.07616115e+00
2.67954797e-01 -8.35999131e-01 -1.12376177e+00 1.57100570e+00
5.06132655e-02 -1.63764405e+00 1.48106724e-01 3.87876749e-01
9.30580914e-01 3.09304118e-01 -3.36983055e-01 -9.96966884e-02
-3.65814306e-02 -9.99389887e-01 8.31030250e-01 5.47106862e-01
1.23136091e+00 -8.40301514e-01 7.41313517e-01 6.05886698e-01
-1.61082864e+00 1.06209576e-01 -7.13860452e-01 -3.97828430e-01
1.47220343e-01 1.20423245e+00 -7.09015727e-01 7.94853628e-01
7.91818202e-01 1.18643415e+00 -2.66862839e-01 1.28529179e+00
-1.48173839e-01 -3.27950835e-01 -5.62805951e-01 3.58258933e-01
3.13708514e-01 -5.10183647e-02 4.47795570e-01 7.24255919e-01
4.97878611e-01 -1.79925784e-01 4.48898256e-01 8.38267565e-01
1.47385985e-01 -4.84981090e-02 -9.51693773e-01 6.12604082e-01
8.11130643e-01 1.04443550e+00 -5.45892239e-01 -1.45606160e-01
-4.86241609e-01 1.00665534e+00 4.23765063e-01 5.09465784e-02
-4.42002505e-01 -4.79022443e-01 8.53231907e-01 3.08525354e-01
8.87969360e-02 -7.61214316e-01 -5.81563532e-01 -1.27752447e+00
1.79129019e-01 -4.97920752e-01 -1.75436825e-01 -5.62860847e-01
-8.87693644e-01 4.26695555e-01 -8.64586011e-02 -1.64518189e+00
-1.19036280e-01 -9.99906063e-02 -1.64833605e-01 1.08178687e+00
-2.10359120e+00 -1.05150008e+00 -1.30031013e+00 7.06203222e-01
6.44558549e-01 2.29208618e-01 6.32284760e-01 1.03968278e-01
2.00293884e-01 3.35727539e-03 7.67013356e-02 -1.95162475e-01
5.97533703e-01 -9.48402107e-01 4.34146523e-01 5.89414418e-01
-1.42033279e-01 4.80137825e-01 3.94632190e-01 -1.07697165e+00
-1.78028452e+00 -1.03535891e+00 7.13041961e-01 -5.50529435e-02
7.82381222e-02 -4.01936412e-01 -9.14575160e-01 4.98417974e-01
-4.70561683e-01 3.60724330e-01 6.26086220e-02 -1.83088705e-01
-1.20152190e-01 -3.41283940e-02 -1.29633582e+00 1.43100008e-01
1.35330582e+00 -4.51875597e-01 -4.16425049e-01 1.28912866e-01
5.92717052e-01 -8.76364946e-01 -6.38373554e-01 6.45880282e-01
6.15050495e-01 -1.30467010e+00 8.22298229e-01 5.66117644e-01
3.73089939e-01 -5.23403049e-01 -3.88704360e-01 -1.08277917e+00
-8.93747434e-02 -3.66963923e-01 -3.39243524e-02 9.54702437e-01
-1.32398486e-01 -7.02256978e-01 1.04823911e+00 2.19189927e-01
-3.56938183e-01 -6.74993694e-01 -1.24889040e+00 -5.38662493e-01
-3.21327657e-01 -2.41303369e-01 5.98585069e-01 9.40827549e-01
-7.04512522e-02 -1.35091990e-01 -2.89869130e-01 3.62511218e-01
9.39117193e-01 2.49045432e-01 1.07547259e+00 -1.56476068e+00
3.04130137e-01 1.74770445e-01 -7.06433713e-01 -1.53991461e+00
1.75048158e-01 -5.67274928e-01 2.64546603e-01 -1.98254585e+00
-1.41367137e-01 -9.87443089e-01 4.76890594e-01 4.18453217e-01
3.37218672e-01 1.83158457e-01 -2.62462944e-01 5.81476629e-01
-3.01992297e-01 9.30247962e-01 1.25799465e+00 -1.23426970e-02
-2.72296041e-01 -4.36946005e-01 -5.50236702e-02 8.95575225e-01
4.92977828e-01 -3.99210632e-01 -4.32184547e-01 -8.68871868e-01
2.92847157e-01 1.67286724e-01 2.85787672e-01 -1.31132877e+00
5.08998692e-01 -3.86505157e-01 4.06839520e-01 -1.00325274e+00
7.58173823e-01 -9.24661577e-01 2.36398757e-01 4.44131374e-01
4.23652738e-01 -8.49127248e-02 2.37589106e-02 8.05563867e-01
-4.86054152e-01 -8.47352818e-02 5.74300528e-01 -4.19397801e-01
-1.27766168e+00 5.88438690e-01 1.03964254e-01 -2.74502605e-01
9.94556010e-01 -6.86398327e-01 -1.33329332e-01 -3.01218182e-01
-2.07454413e-01 3.80593121e-01 1.16445518e+00 2.38853052e-01
1.31195033e+00 -1.30056369e+00 -5.78196228e-01 7.01077402e-01
3.88548553e-01 1.14273930e+00 1.93872616e-01 7.12992609e-01
-1.14411700e+00 2.46316195e-01 -2.83929676e-01 -1.05214608e+00
-8.48869979e-01 8.04316178e-02 8.03557560e-02 2.46192813e-01
-1.14851594e+00 7.92315841e-01 2.93291092e-01 -1.65465787e-01
3.08877528e-01 -5.82616210e-01 -3.11622513e-03 2.00822148e-02
3.41396213e-01 3.14404160e-01 1.18965007e-01 -6.27511501e-01
-5.74682057e-01 1.07777536e+00 3.17398459e-01 -9.10149664e-02
1.22132576e+00 -3.28487098e-01 -1.87278256e-01 5.51680624e-01
1.20335698e+00 2.18060315e-01 -1.51460183e+00 -3.47373366e-01
-2.74064332e-01 -9.99083996e-01 3.11752588e-01 -1.83702074e-02
-8.90291631e-01 8.97485554e-01 2.65085489e-01 -3.26571375e-01
7.27510989e-01 -1.04262292e-01 8.64089668e-01 3.26642215e-01
1.27229011e+00 -7.04982162e-01 -3.80196236e-03 6.76723540e-01
9.59675550e-01 -1.47480774e+00 3.96463096e-01 -7.67160058e-01
-3.88910204e-01 1.17983079e+00 6.94568932e-01 -1.44954771e-01
5.60659170e-01 2.29281768e-01 -8.50350484e-02 -2.10475743e-01
-2.92827487e-01 -2.15439871e-01 -6.02396466e-02 7.63111532e-01
-9.96259153e-02 -3.03721368e-01 -1.62152424e-01 -3.71429138e-03
-2.19292000e-01 8.00582990e-02 4.36054409e-01 1.24420428e+00
-7.91048586e-01 -9.39609230e-01 -4.16734964e-01 1.92034125e-01
3.45248103e-01 -4.69668880e-02 -5.19295633e-02 7.45567381e-01
2.88481593e-01 7.15392351e-01 2.91476220e-01 -5.29832065e-01
7.68881962e-02 -4.77673769e-01 4.55963254e-01 -8.00976336e-01
8.09963495e-02 -1.12269074e-01 -1.47662804e-01 -1.06649888e+00
-4.61111844e-01 -6.27201438e-01 -1.63107908e+00 -4.20717239e-01
-1.60436898e-01 8.45146626e-02 9.55728948e-01 7.63411939e-01
4.54227984e-01 -1.29472632e-02 6.53410792e-01 -1.44384742e+00
-2.54702847e-02 -7.78338671e-01 -6.53396368e-01 1.43652365e-01
6.81686819e-01 -1.17471921e+00 -5.55749714e-01 -4.31957066e-01] | [8.00857162475586, -2.379518747329712] |
7969ca55-7b3d-409f-a95a-841aef9133ca | analyzing-culture-specific-argument | null | null | https://aclanthology.org/2022.argmining-1.4 | https://aclanthology.org/2022.argmining-1.4.pdf | Analyzing Culture-Specific Argument Structures in Learner Essays | Language education has been shown to benefit from computational argumentation, for example, from methods that assess quality dimensions of language learners’ argumentative essays, such as their organization and argument strength. So far, however, little attention has been paid to cultural differences in learners’ argument structures originating from different origins and language capabilities. This paper extends prior studies of learner argumentation by analyzing differences in the argument structure of essays from culturally diverse learners. Based on the ICLE corpus containing essays written by English learners of 16 different mother tongues, we train natural language processing models to mine argumentative discourse units (ADUs) as well as to assess the essays’ quality in terms of organization and argument strength. The extracted ADUs and the predicted quality scores enable us to look into the similarities and differences of essay argumentation across different English learners. In particular, we analyze the ADUs from learners with different mother tongues, different levels of arguing proficiency, and different context cultures. | ['Henning Wachsmuth', 'Garima Mudgal', 'Mei-Hua Chen', 'Wei-Fan Chen'] | null | null | null | null | argmining-acl-2022-10 | ['culture'] | ['speech'] | [-3.14814538e-01 4.33239520e-01 -2.98133135e-01 -3.32490414e-01
-3.64865661e-01 -9.04671788e-01 7.71580756e-01 8.71316433e-01
-4.61285412e-01 6.04380071e-01 9.47577834e-01 -7.23336875e-01
-3.55367690e-01 -8.30357492e-01 -4.60822403e-01 -1.97986037e-01
7.40289330e-01 1.35009056e-02 -1.83261618e-01 -7.37508595e-01
8.10616374e-01 -2.49835864e-01 -1.59007990e+00 6.38970613e-01
1.82179070e+00 2.04449505e-01 2.22524121e-01 5.06046832e-01
-7.14106321e-01 1.30171525e+00 -9.90706682e-01 -8.22376668e-01
-4.29609090e-01 -8.33343804e-01 -1.04319596e+00 -2.42999732e-01
5.94819486e-01 -1.34294018e-01 3.42754066e-01 1.02672374e+00
2.51977175e-01 -3.01963091e-01 8.26201916e-01 -6.91010773e-01
-1.27674878e+00 1.37041676e+00 -1.01557612e-01 1.18334591e-01
5.94501078e-01 -3.89553249e-01 1.21523178e+00 -6.46573126e-01
8.88355613e-01 1.42474902e+00 5.94333768e-01 6.33731008e-01
-8.20967197e-01 -4.74913448e-01 1.66907579e-01 2.15842709e-01
-2.09403127e-01 -1.78422228e-01 9.69249010e-01 -8.94601345e-01
3.43988657e-01 -1.95438445e-01 1.22389793e+00 1.45461071e+00
-1.01626702e-02 9.79746819e-01 1.65105343e+00 -1.02152491e+00
1.33046061e-01 4.39212978e-01 7.66008556e-01 6.43128335e-01
3.39806437e-01 -3.95528853e-01 -7.87121296e-01 2.74968296e-01
1.37639672e-01 -5.42448759e-01 -1.66289225e-01 1.54989511e-01
-1.40177584e+00 1.37228251e+00 -6.21360429e-02 5.70233643e-01
2.67174281e-02 -4.14829224e-01 6.02452576e-01 7.83082247e-01
2.86039859e-01 6.61155522e-01 -4.78774548e-01 -6.29376829e-01
-1.56423345e-01 1.79362267e-01 9.05965805e-01 6.75908804e-01
1.03930995e-01 -1.23650646e-02 -4.37478982e-02 1.24272585e+00
6.78225160e-01 4.42026258e-01 7.84531951e-01 -8.52591097e-01
6.79359555e-01 1.10355198e+00 -2.29937628e-01 -7.01262236e-01
-5.05636893e-02 -1.69847190e-01 2.16362849e-02 1.64851874e-01
7.90066898e-01 -5.38617432e-01 3.23612057e-02 1.79779148e+00
3.39829326e-01 -7.66687155e-01 6.43588245e-01 6.70728564e-01
1.44614100e+00 3.13886374e-01 2.07587332e-01 7.58817121e-02
1.78256810e+00 -9.74210441e-01 -7.42226303e-01 -4.32812423e-02
1.21877694e+00 -1.13481271e+00 1.54563487e+00 1.75769448e-01
-1.35794866e+00 -5.57589173e-01 -1.25739312e+00 -4.78509873e-01
-4.86421943e-01 2.34120905e-01 3.84488225e-01 1.04084980e+00
-4.82103944e-01 1.91759899e-01 -3.54131818e-01 -1.54350728e-01
4.01188701e-01 -4.40523654e-01 1.29822925e-01 3.19693297e-01
-1.19275296e+00 9.69671607e-01 4.20112610e-01 -3.84603649e-01
-1.45562053e-01 -9.78908837e-01 -8.57124746e-01 -5.69682196e-02
5.74028902e-02 -4.11651492e-01 1.38229239e+00 -1.35857570e+00
-1.85491729e+00 9.20051396e-01 4.69036877e-01 7.11999321e-03
4.26117033e-01 -4.49625522e-01 -1.50797039e-01 8.38181525e-02
2.10019782e-01 1.55289680e-01 3.26490700e-01 -8.30579221e-01
-7.26214826e-01 -3.06453973e-01 2.51593351e-01 4.70052391e-01
-8.42961073e-01 3.27030510e-01 4.50574636e-01 -8.49425554e-01
4.36211415e-02 -6.84040487e-01 5.08490026e-01 -1.71133220e-01
4.37058359e-02 -7.77715683e-01 5.27178347e-01 -7.25848377e-01
1.11422861e+00 -1.92763901e+00 1.35372773e-01 7.75547698e-02
2.62182474e-01 1.05922975e-01 7.16979429e-02 5.18968165e-01
3.48184526e-01 4.96862113e-01 2.74666995e-01 2.83352673e-01
3.90631378e-01 3.55012864e-01 -1.49110675e-01 7.19504058e-02
3.58689539e-02 6.89207435e-01 -1.29191160e+00 -6.26342714e-01
-1.44850850e-01 3.23476195e-01 -4.73749965e-01 8.85024890e-02
-2.51029462e-01 3.99430931e-01 -6.93577409e-01 3.89359921e-01
-1.47439027e-02 9.43332165e-02 5.03333509e-01 4.69919145e-01
-6.64881229e-01 1.03694069e+00 -7.11990535e-01 1.43761921e+00
-6.96659744e-01 1.11999989e+00 -5.33378460e-02 -7.98548043e-01
7.98379660e-01 5.19777417e-01 -1.51953310e-01 -1.06904602e+00
3.86469245e-01 6.70078039e-01 8.20072949e-01 -7.10793376e-01
5.88635862e-01 -6.63610995e-02 -3.42431664e-02 1.03895450e+00
-1.60857230e-01 -4.18387234e-01 6.38595045e-01 2.28814617e-01
3.98660392e-01 2.70092994e-01 2.78290778e-01 -9.41745400e-01
5.76040685e-01 -5.45577854e-02 2.17614517e-01 3.15592885e-01
-2.36193854e-02 -1.42183572e-01 7.31530786e-01 -2.16438189e-01
-1.03530073e+00 -9.48832035e-01 -4.75136697e-01 1.23837590e+00
-1.46014482e-01 -3.61145884e-01 -1.00515735e+00 -6.35065615e-01
-2.80356884e-01 8.33793581e-01 -5.21021545e-01 1.07349522e-01
-8.81228387e-01 -2.37125516e-01 6.01551414e-01 4.52375323e-01
6.60106540e-01 -1.26448727e+00 -1.25702119e+00 2.08096504e-01
-3.74805748e-01 -8.95699859e-01 1.66117839e-04 -1.23054564e-01
-8.75097334e-01 -1.17623842e+00 -4.23562318e-01 -9.89178300e-01
4.48069423e-01 5.90233766e-02 1.32038641e+00 7.55789876e-01
2.42178768e-01 3.95368844e-01 -6.76230431e-01 -1.29897654e+00
-1.03739941e+00 8.74414891e-02 -4.96388450e-02 -7.96647370e-01
4.53295261e-01 -1.19432911e-01 -1.58612981e-01 8.40105489e-02
-7.88985014e-01 2.44998172e-01 2.48631671e-01 8.25850129e-01
-4.57664654e-02 -3.88993084e-01 8.95473719e-01 -9.13369060e-01
1.37361968e+00 -5.94473958e-01 -3.72061849e-01 6.08814240e-01
-7.56328404e-01 2.38000974e-01 7.31124163e-01 -4.43961501e-01
-1.57411885e+00 -1.20962334e+00 -1.85558140e-01 7.19295561e-01
-5.80181479e-02 8.19201171e-01 3.72141719e-01 2.29225278e-01
8.09256554e-01 -2.76750892e-01 2.91449547e-01 -9.70331728e-02
1.89869180e-01 6.77795887e-01 8.25745612e-02 -1.56393647e+00
3.97264868e-01 -1.13286942e-01 -4.16960508e-01 -1.25062048e+00
-1.28138757e+00 -3.73522304e-02 -5.56875229e-01 -4.74624664e-01
9.56980228e-01 -9.16208565e-01 -9.20823336e-01 3.79085988e-01
-1.03588057e+00 -4.39584434e-01 -2.25579634e-01 9.71817553e-01
-2.92114407e-01 4.25932974e-01 -9.98643994e-01 -6.21173203e-01
-3.21637481e-01 -1.11507416e+00 4.47441101e-01 5.38871229e-01
-6.65202737e-01 -1.56872213e+00 3.47645909e-01 8.56589437e-01
7.73902610e-02 1.91300571e-01 1.76692891e+00 -7.49872744e-01
2.02205136e-01 7.08000004e-01 1.11470543e-01 3.65628511e-01
6.97410181e-02 3.47643256e-01 -6.80486143e-01 2.17547432e-01
3.40208948e-01 -8.15330923e-01 2.45540008e-01 1.53247252e-01
3.44558328e-01 -4.35209274e-01 4.85484272e-01 6.91243187e-02
1.08311749e+00 3.49779092e-02 3.04357678e-01 9.07230794e-01
4.12559301e-01 1.08493960e+00 6.76497400e-01 6.85554668e-02
5.62620044e-01 1.29320592e-01 -1.07769571e-01 3.15223217e-01
-1.55232683e-01 -1.91724047e-01 7.66214669e-01 1.81573796e+00
-2.70601451e-01 -6.08236156e-02 -1.09614193e+00 7.64902592e-01
-1.56905735e+00 -9.27579582e-01 -6.26631379e-01 1.65884066e+00
1.09911895e+00 -7.34879375e-02 9.17354748e-02 3.17966104e-01
4.28140342e-01 6.19011447e-02 -7.16991797e-02 -1.05528545e+00
-3.43534112e-01 3.54424983e-01 -3.24430972e-01 4.99822617e-01
-2.17070907e-01 6.40428722e-01 5.59185934e+00 4.45072204e-01
-7.42795825e-01 -1.97858766e-01 3.94482970e-01 1.77877210e-02
-7.54325330e-01 -1.55116200e-01 -4.21532571e-01 3.11488003e-01
8.39916825e-01 -1.02095112e-01 -6.57159537e-02 5.18470466e-01
3.10215820e-02 2.82749031e-02 -9.38572645e-01 3.59726101e-01
5.25495084e-03 -1.15650308e+00 -2.46525928e-01 1.14675827e-01
1.04548335e+00 -2.86034107e-01 7.85998330e-02 4.95279253e-01
6.33933902e-01 -8.17936420e-01 1.35779357e+00 -3.69432680e-02
7.28480220e-02 -5.09365976e-01 6.74448013e-01 4.50217873e-01
-5.24297476e-01 -1.64342239e-01 -1.43481120e-01 -7.77728021e-01
-3.42675626e-01 1.11741863e-01 -7.14417458e-01 2.08330572e-01
6.03010297e-01 4.45037991e-01 -4.31558818e-01 4.22573425e-02
-9.16180313e-01 1.10424817e+00 4.37212922e-02 -7.90540159e-01
2.35188335e-01 -8.25026453e-01 2.81447858e-01 8.28354895e-01
4.14168060e-01 2.33632997e-01 -9.55681950e-02 6.49508774e-01
-1.08594306e-01 7.68655300e-01 -4.07270074e-01 -3.01822126e-01
6.82066441e-01 9.05919611e-01 -6.31443262e-01 -3.00122112e-01
-8.78069758e-01 3.01712185e-01 6.82805359e-01 9.55099612e-02
-6.70522690e-01 -4.60695028e-02 6.29885912e-01 -1.08680673e-01
-9.41497386e-02 -3.17748070e-01 -7.13209987e-01 -9.86887634e-01
-5.21588549e-02 -1.23160875e+00 3.90542746e-01 -4.70525146e-01
-1.47005475e+00 1.53799132e-01 -2.20708758e-01 -6.54224098e-01
-6.13390207e-02 -9.57255661e-01 -8.53062689e-01 7.99394965e-01
-1.34908092e+00 -9.24889982e-01 -1.93719715e-01 3.85320604e-01
7.98769832e-01 -3.52862000e-01 6.49378955e-01 -1.08665019e-01
-4.49246615e-01 3.76372427e-01 1.72555998e-01 4.01090771e-01
7.20472157e-01 -1.46388519e+00 -6.92384988e-02 4.97465700e-01
3.35345924e-01 8.59312892e-01 5.47457814e-01 -5.00264525e-01
-1.13375711e+00 -1.67934477e-01 1.03831017e+00 -5.84594727e-01
1.04701877e+00 -1.55561427e-02 -8.89218450e-01 3.05594742e-01
1.09667456e+00 -8.16403568e-01 1.37237406e+00 6.20608747e-01
-3.50182861e-01 6.73611522e-01 -7.73025095e-01 1.09379816e+00
1.07921195e+00 -5.40439606e-01 -1.31144905e+00 1.15995497e-01
3.68658930e-01 -3.77312034e-01 -1.30734587e+00 -1.78517565e-01
7.62676895e-01 -1.18615687e+00 7.78477907e-01 -9.44851696e-01
1.32228303e+00 2.24932790e-01 2.05637515e-01 -1.34077978e+00
-9.81005747e-03 -1.10962361e-01 1.62661448e-01 1.30782640e+00
4.64695662e-01 -6.16490781e-01 3.23055238e-01 3.40307683e-01
-4.80187356e-01 -8.06239486e-01 -4.56906050e-01 -2.75474012e-01
9.05640185e-01 -1.11430548e-01 6.08232141e-01 1.56786013e+00
3.96581054e-01 7.84703672e-01 5.12486935e-01 -4.52361226e-01
2.72449166e-01 4.15773958e-01 6.27673090e-01 -1.84026134e+00
8.19461718e-02 -1.07460523e+00 3.02500963e-01 -5.05670846e-01
6.45423770e-01 -9.87003922e-01 -2.02277124e-01 -1.72030437e+00
-2.31585890e-01 -3.82497430e-01 3.46632242e-01 1.61333546e-01
-4.02975351e-01 -2.60356575e-01 2.44760409e-01 1.01369791e-01
-1.78239867e-01 4.56427544e-01 1.48689044e+00 1.65560484e-01
-3.10634732e-01 -4.01356548e-01 -9.53803718e-01 1.37731624e+00
1.14939380e+00 -3.03686947e-01 -6.88393056e-01 -8.63148212e-01
9.10083950e-01 -7.04693720e-02 -8.62575844e-02 -5.36896825e-01
-2.85889022e-02 -6.51017964e-01 2.65538275e-01 -1.04408756e-01
-3.21122199e-01 -4.52864021e-01 -6.40542924e-01 5.98653138e-01
-5.99778414e-01 4.47819293e-01 -4.82916497e-02 -1.96507543e-01
-3.90517712e-01 -8.69497120e-01 4.29416984e-01 -3.84903774e-02
-3.19689691e-01 -6.76748514e-01 -6.75176382e-01 7.98797190e-01
8.30283940e-01 -5.26011229e-01 -8.53622973e-01 -1.15145013e-01
-1.84629276e-01 1.45894438e-01 4.79923815e-01 5.58564603e-01
3.73115897e-01 -1.21520329e+00 -1.19787323e+00 -1.65379152e-01
7.99922273e-02 -1.55675113e-01 -2.38707233e-02 7.58166373e-01
-8.91097248e-01 2.09748328e-01 -5.06927669e-01 -2.57433355e-01
-1.54833329e+00 5.31166568e-02 -1.80639803e-01 -2.46898517e-01
-4.47504431e-01 6.67059004e-01 -1.59260314e-02 -6.19685352e-01
-2.96641827e-01 -3.87986183e-01 -6.85869396e-01 6.63880587e-01
2.92364776e-01 6.85182214e-01 -3.38912964e-01 -5.51114619e-01
2.81393349e-01 5.83433867e-01 2.84103397e-02 -2.74296999e-01
1.02558613e+00 5.30987084e-02 -4.77183610e-02 6.37750030e-01
6.92489207e-01 1.18609846e+00 -5.93541980e-01 1.87892746e-02
3.36433291e-01 -3.16542983e-01 -4.74885523e-01 -7.14717448e-01
-4.71924156e-01 9.28156555e-01 -1.08485119e-02 4.32316661e-01
4.82997566e-01 -7.57454485e-02 5.06981790e-01 4.36856359e-01
-1.29664600e-01 -1.96495450e+00 3.95052195e-01 1.12780559e+00
8.31251562e-01 -1.10546041e+00 -7.36567229e-02 -2.93328911e-01
-7.02460051e-01 1.46828508e+00 6.81078553e-01 3.29701543e-01
3.65531176e-01 2.34810747e-02 6.38515770e-01 -3.65400344e-01
-6.20666444e-01 2.08508655e-01 3.29994410e-01 2.78359622e-01
1.48810601e+00 2.17554957e-01 -1.36797345e+00 5.91534674e-01
-1.06770742e+00 -6.00213885e-01 9.11124825e-01 9.87866580e-01
-4.23929542e-01 -1.39616501e+00 -5.51379263e-01 2.00649187e-01
-6.79703176e-01 -2.43347734e-01 -9.96738195e-01 9.15273368e-01
8.20472240e-02 1.05464613e+00 1.14127574e-02 2.97318697e-01
6.64340034e-02 3.20500165e-01 7.43778467e-01 -6.66641176e-01
-1.28533149e+00 -5.04937291e-01 5.84913492e-01 4.35702831e-01
-8.65276814e-01 -8.62828434e-01 -1.69898176e+00 -3.47840130e-01
-3.97454321e-01 4.88939583e-01 8.14736485e-01 1.20673251e+00
-1.94488525e-01 7.91832387e-01 1.80667732e-03 1.87578961e-01
-5.33835888e-01 -1.16111350e+00 -1.79930463e-01 4.43159372e-01
-4.87863831e-02 -5.36939681e-01 -1.59620926e-01 -1.84602872e-01] | [11.213569641113281, 9.308149337768555] |
733e110d-8bfb-4e53-9e63-af47c919d383 | cyclic-guidance-for-weakly-supervised-joint | null | null | http://openaccess.thecvf.com/content_CVPR_2019/html/Shen_Cyclic_Guidance_for_Weakly_Supervised_Joint_Detection_and_Segmentation_CVPR_2019_paper.html | http://openaccess.thecvf.com/content_CVPR_2019/papers/Shen_Cyclic_Guidance_for_Weakly_Supervised_Joint_Detection_and_Segmentation_CVPR_2019_paper.pdf | Cyclic Guidance for Weakly Supervised Joint Detection and Segmentation | Weakly supervised learning has attracted growing research attention due to the significant saving in annotation cost for tasks that require intra-image annotations, such as object detection and semantic segmentation. To this end, existing weakly supervised object detection and semantic segmentation approaches follow an iterative label mining and model training pipeline. However, such a self-enforcement pipeline makes both tasks easy to be trapped in local minimums. In this paper, we join weakly supervised object detection and segmentation tasks with a multi-task learning scheme for the first time, which uses their respective failure patterns to complement each other's learning. Such cross-task enforcement helps both tasks to leap out of their respective local minimums. In particular, we present an efficient and effective framework termed Weakly Supervised Joint Detection and Segmentation (WS-JDS). WS-JDS has two branches for the above two tasks, which share the same backbone network. In the learning stage, it uses the same cyclic training paradigm but with a specific loss function such that the two branches benefit each other. Extensive experiments have been conducted on the widely-used Pascal VOC and COCO benchmarks, which demonstrate that our model has achieved competitive performance with the state-of-the-art algorithms.
| [' Liujuan Cao', ' Yongjian Wu', ' Yan Wang', ' Rongrong Ji', 'Yunhang Shen'] | 2019-06-01 | null | null | null | cvpr-2019-6 | ['image-level-supervised-instance-segmentation'] | ['computer-vision'] | [ 4.45572078e-01 1.80094779e-01 -5.07556856e-01 -4.06562209e-01
-9.32714581e-01 -3.29888970e-01 3.17157030e-01 2.73732960e-01
-6.35519922e-01 5.64770341e-01 -4.34304178e-01 -5.81941344e-02
2.29803011e-01 -3.76854807e-01 -6.97004318e-01 -9.58034575e-01
2.22877458e-01 3.96145880e-01 9.74339902e-01 2.83939123e-01
-1.18198991e-02 -1.12801969e-01 -1.37143159e+00 3.61029714e-01
8.76095533e-01 1.13662183e+00 4.08098876e-01 2.87191719e-01
-1.49009928e-01 8.96047175e-01 -2.49895304e-01 -5.42968988e-01
2.29955092e-01 -1.63955912e-01 -9.77012515e-01 3.48689467e-01
2.52317041e-01 -2.27040146e-02 2.04895526e-01 1.16812873e+00
2.99535185e-01 -1.15061879e-01 3.96503568e-01 -1.17922115e+00
-1.62880898e-01 4.87665236e-01 -1.12881303e+00 -2.69559711e-01
-1.61877438e-01 1.23948537e-01 1.05207396e+00 -9.76002157e-01
2.84280270e-01 1.03079629e+00 7.21704960e-01 6.03864133e-01
-1.30017638e+00 -6.57705307e-01 4.72126365e-01 -4.29851152e-02
-1.43735230e+00 -1.69059917e-01 7.45098531e-01 -3.71836811e-01
5.48984051e-01 6.21730909e-02 4.26057994e-01 6.93953514e-01
-3.28484476e-01 1.24179149e+00 1.13065732e+00 -4.43076938e-01
2.13507473e-01 3.07873219e-01 3.67397964e-01 1.01778889e+00
1.12465948e-01 -2.10727051e-01 -3.93788308e-01 -7.00374246e-02
3.43700498e-01 -1.15791321e-01 -5.14670499e-02 -5.36487758e-01
-1.02149534e+00 6.58828914e-01 4.02956337e-01 2.48167068e-01
-2.46552452e-02 7.18737990e-02 4.84582931e-01 -3.45197842e-02
6.74673259e-01 1.26100499e-02 -6.56346560e-01 2.64950633e-01
-9.97443676e-01 3.64024527e-02 7.06797779e-01 8.77969444e-01
8.59729230e-01 -4.32594150e-01 -2.40908861e-01 1.00472212e+00
4.53695178e-01 -1.64262369e-01 3.29251885e-01 -7.05493808e-01
4.12613899e-01 7.99189031e-01 1.00961588e-01 -6.43727183e-01
-2.77698278e-01 -7.59879231e-01 -6.67937160e-01 1.34533346e-01
4.92258489e-01 -2.28284057e-02 -1.04565501e+00 1.79209030e+00
7.04948246e-01 4.70317423e-01 -3.24411765e-02 8.19870174e-01
5.67605555e-01 4.77133244e-01 4.15975511e-01 -2.81574368e-01
1.45657909e+00 -1.53580022e+00 -4.61033046e-01 -4.18966740e-01
6.89693272e-01 -7.59362519e-01 1.09341943e+00 1.96201965e-01
-1.04752588e+00 -7.65974641e-01 -1.02666342e+00 -1.47244871e-01
-1.24321498e-01 3.87405217e-01 5.25215745e-01 3.72479320e-01
-7.07643092e-01 4.11454797e-01 -1.06195092e+00 -8.34216624e-02
7.80595720e-01 4.24822092e-01 -1.35244047e-02 4.45371494e-02
-7.83478379e-01 4.74302948e-01 7.79402733e-01 3.29952061e-01
-8.70152593e-01 -5.28397977e-01 -7.27633059e-01 -1.28106311e-01
9.62574005e-01 -3.84193271e-01 1.40400910e+00 -1.19941163e+00
-1.32995415e+00 1.25210452e+00 -1.91333100e-01 -3.37874502e-01
8.07924747e-01 -3.87661755e-01 4.44694497e-02 -2.18760427e-02
4.53659981e-01 8.32669556e-01 9.56859410e-01 -1.53819227e+00
-1.13565803e+00 -4.31078941e-01 -6.11611195e-02 1.30215421e-01
-5.24699628e-01 1.33217111e-01 -1.20577562e+00 -6.08850062e-01
2.38542810e-01 -8.70690823e-01 -3.72417033e-01 2.34864116e-01
-6.30715251e-01 -7.12184191e-01 1.14806306e+00 -1.66281700e-01
1.24344265e+00 -2.18157339e+00 1.03244305e-01 2.75420323e-02
2.83675641e-01 4.98476118e-01 -9.79308132e-03 -1.68155536e-01
2.08593875e-01 -2.87583526e-02 -6.20504856e-01 -1.09193456e+00
-2.19230846e-01 3.28991950e-01 -3.74856289e-03 5.28695881e-01
3.50724876e-01 9.15753365e-01 -1.09030521e+00 -1.00437915e+00
3.52150947e-02 2.18104869e-02 -3.18827450e-01 3.88888419e-01
-5.19763231e-01 3.98202330e-01 -5.29214144e-01 7.88794219e-01
5.86030781e-01 -5.62702060e-01 2.05026373e-01 -2.28172615e-01
-2.73945838e-01 -8.19346607e-02 -1.14952016e+00 1.82811069e+00
-1.26711503e-01 2.31223125e-02 3.88775140e-01 -1.31756592e+00
6.90477908e-01 1.98426262e-01 4.84805048e-01 -5.06838441e-01
-1.56928282e-02 5.75455010e-01 -2.30971128e-01 -4.41811323e-01
1.32717192e-01 -1.33226156e-01 5.10032428e-03 3.75996381e-01
7.42656440e-02 6.73088804e-02 4.58505422e-01 1.25102088e-01
6.63035810e-01 5.31901479e-01 9.85374153e-02 -2.66705692e-01
8.48208666e-01 4.00325768e-02 1.01113248e+00 8.10474396e-01
-4.02303308e-01 5.36186397e-01 5.14085531e-01 -2.90284365e-01
-5.92598319e-01 -8.45959008e-01 -1.22058593e-01 1.35408378e+00
5.39742172e-01 -2.96292365e-01 -9.21572626e-01 -1.23878706e+00
-1.38233751e-01 1.55476168e-01 -4.96297359e-01 -5.91544062e-02
-6.18449330e-01 -8.57234240e-01 4.63361889e-01 6.04573965e-01
7.22977459e-01 -1.09247351e+00 -5.36881089e-01 2.50259519e-01
-1.47609517e-01 -1.42454112e+00 -4.60306078e-01 6.05543852e-01
-7.40539014e-01 -1.21765959e+00 -9.31665540e-01 -1.37186730e+00
8.21481705e-01 3.04854900e-01 8.83814037e-01 3.03436518e-01
-1.88719243e-01 -8.55182111e-02 -3.45047474e-01 -3.47945094e-01
-9.09617394e-02 3.03117216e-01 -1.54777870e-01 3.86867493e-01
3.07442427e-01 -2.20455721e-01 -5.68853021e-01 5.81507385e-01
-9.47802424e-01 2.18002871e-01 6.51434660e-01 8.59088242e-01
1.00447047e+00 -3.70755419e-02 5.65072119e-01 -1.10126328e+00
1.00558996e-01 -3.81178677e-01 -7.72618532e-01 4.10992563e-01
-5.75821161e-01 -1.81289017e-01 5.55148363e-01 -2.58056849e-01
-1.17811096e+00 6.02835655e-01 -6.33354113e-02 -2.58544564e-01
-1.78193823e-01 3.85452777e-01 -2.62683928e-01 -7.08059371e-02
2.79068738e-01 1.99665830e-01 -2.12701544e-01 -6.03309691e-01
1.94417670e-01 7.67381430e-01 7.16209054e-01 -4.95463520e-01
8.46739233e-01 7.17293262e-01 -3.06912661e-01 -5.20945430e-01
-1.62532508e+00 -8.74675870e-01 -7.49231219e-01 -1.84233457e-01
1.01926064e+00 -9.88135457e-01 -5.94880760e-01 9.52717543e-01
-1.08276057e+00 -4.34030324e-01 -2.28132531e-01 2.97865301e-01
-3.70997220e-01 5.37572145e-01 -5.80047607e-01 -8.95186841e-01
-1.45203039e-01 -1.26386988e+00 1.30149341e+00 3.75190407e-01
1.02789231e-01 -8.71602476e-01 -1.87992945e-01 7.04505622e-01
-1.12186838e-02 2.00881705e-01 7.30488241e-01 -6.07413173e-01
-5.42228222e-01 -7.99636394e-02 -5.78629851e-01 7.00480461e-01
1.21232525e-01 -3.17892134e-01 -1.09258783e+00 -3.63086641e-01
-1.59093980e-02 -8.74684334e-01 1.14791787e+00 8.79834220e-02
1.44151187e+00 1.79797020e-02 -4.32072878e-01 6.29220724e-01
1.28980780e+00 -6.24792613e-02 9.69885141e-02 3.42647433e-01
9.28018570e-01 8.30810010e-01 7.84868479e-01 9.96628106e-02
3.65568519e-01 7.32097149e-01 4.45011467e-01 -5.34676313e-01
-8.17143992e-02 -1.98237449e-01 3.30088407e-01 7.14944661e-01
2.17525497e-01 4.18742262e-02 -8.34937513e-01 5.14868379e-01
-2.33064651e+00 -4.72676605e-01 -3.04367185e-01 1.97759402e+00
1.19654667e+00 6.00971699e-01 1.97459489e-01 1.18053466e-01
7.51552045e-01 2.81713456e-01 -7.17556298e-01 7.88262710e-02
6.06177282e-03 9.25579108e-03 3.19939137e-01 2.87250698e-01
-1.48529005e+00 1.15130329e+00 4.78405428e+00 1.09615707e+00
-9.36961651e-01 3.16681981e-01 8.84815991e-01 1.71214901e-02
2.33904973e-01 9.77076963e-02 -8.80793333e-01 4.82634366e-01
3.45448703e-01 4.23422366e-01 -1.71412036e-01 8.64772916e-01
2.12638512e-01 -3.32937956e-01 -1.17736435e+00 7.21991420e-01
-1.93485588e-01 -9.89959478e-01 -1.47146627e-01 -2.81157732e-01
8.72747779e-01 2.80656032e-02 -1.73575863e-01 2.86642343e-01
1.33387178e-01 -7.08921134e-01 1.00280273e+00 5.02991071e-03
6.73625827e-01 -6.32590532e-01 7.03281820e-01 7.37288117e-01
-1.60072196e+00 -1.92599613e-02 -9.70031917e-02 1.56808496e-01
1.97919413e-01 7.17253625e-01 -2.90899426e-01 5.00864327e-01
7.72485375e-01 7.57566988e-01 -3.99883151e-01 9.94286716e-01
-3.71672779e-01 7.36382186e-01 -2.57854193e-01 1.71188489e-01
5.81137419e-01 -2.50966191e-01 3.07899266e-01 1.32055044e+00
-3.77168745e-01 -1.42070785e-01 9.11003470e-01 8.49394739e-01
-2.82873243e-01 6.58331364e-02 3.64581123e-02 3.28294218e-01
1.92587703e-01 1.37866199e+00 -1.15611076e+00 -4.24624115e-01
-6.00176334e-01 1.00022566e+00 4.73692417e-01 2.08595857e-01
-1.02083278e+00 -3.48350674e-01 2.22454771e-01 4.29848535e-03
3.09819371e-01 -7.19284192e-02 -4.54090208e-01 -1.00098181e+00
3.02302122e-01 -5.90031981e-01 5.19243896e-01 -2.41016343e-01
-1.17940664e+00 3.69643927e-01 -1.63259342e-01 -9.30724740e-01
3.21013987e-01 -4.14971679e-01 -6.05718136e-01 6.19421661e-01
-2.04135585e+00 -1.46828842e+00 -2.77042836e-01 6.57318473e-01
8.16430032e-01 2.24135041e-01 3.77154410e-01 4.59565520e-01
-1.01777267e+00 5.52111447e-01 -1.49038151e-01 2.50044256e-01
5.61168611e-01 -1.36219692e+00 7.66374320e-02 9.26886201e-01
1.37521416e-01 2.30425775e-01 1.66355610e-01 -6.70940518e-01
-8.46329689e-01 -1.42385256e+00 6.95819795e-01 -5.84725216e-02
6.09973013e-01 -5.49047768e-01 -1.07342005e+00 4.35487837e-01
-1.11308850e-01 4.58103240e-01 4.82560337e-01 -3.72116007e-02
-2.79915810e-01 -1.64611682e-01 -8.43268096e-01 3.68353784e-01
1.01505804e+00 -4.28035110e-01 -4.22106832e-01 6.70271456e-01
7.45848060e-01 -3.46980840e-01 -3.72480541e-01 6.71546042e-01
3.50682914e-01 -8.79340053e-01 7.60807753e-01 -3.68461877e-01
3.51426423e-01 -6.62281096e-01 1.70484558e-01 -6.69138014e-01
1.03441980e-02 -7.51761556e-01 -2.74120063e-01 1.46696973e+00
4.78718609e-01 -3.57334107e-01 9.91031051e-01 3.39946151e-01
-2.05728695e-01 -1.23593998e+00 -8.02911580e-01 -8.70094001e-01
-2.95632333e-01 -5.14418781e-01 -4.71095815e-02 7.17172623e-01
-3.93350333e-01 4.15571004e-01 -3.46413493e-01 1.75512120e-01
9.76047039e-01 2.69122988e-01 6.25117064e-01 -1.28682923e+00
-4.10073668e-01 -3.05210143e-01 1.23002172e-01 -1.35136020e+00
1.96593642e-01 -9.08368707e-01 5.77183902e-01 -1.24081326e+00
6.09535992e-01 -8.14123154e-01 -5.26501596e-01 8.23442876e-01
-6.61008239e-01 5.28702796e-01 1.67644113e-01 5.02594829e-01
-1.29164863e+00 5.66791117e-01 1.05596900e+00 -1.07189931e-01
-3.42981517e-01 2.09720269e-01 -6.85250819e-01 1.24566925e+00
6.12475395e-01 -7.68140376e-01 -2.88278252e-01 -4.77250636e-01
-1.50534451e-01 -3.62295747e-01 4.22674924e-01 -8.60649168e-01
4.39217687e-01 6.06622100e-02 3.13162096e-02 -4.80668664e-01
6.87339604e-02 -7.80563176e-01 -4.83175337e-01 6.62598193e-01
-4.02223498e-01 -5.27679443e-01 1.83678456e-02 8.24151337e-01
-4.57056642e-01 -4.01936144e-01 1.23275042e+00 -9.67419371e-02
-8.11389267e-01 3.99851054e-01 1.19411938e-01 2.04040438e-01
1.47487688e+00 -2.78051436e-01 6.31767660e-02 3.03464234e-01
-6.85446858e-01 7.99160779e-01 1.45318106e-01 3.61109167e-01
4.00015265e-01 -8.47182155e-01 -6.69837177e-01 -4.68078349e-03
2.76972860e-01 7.39878893e-01 -7.63718486e-02 1.02710640e+00
-1.26727074e-01 8.43596831e-02 3.44974577e-01 -9.71455097e-01
-1.32203436e+00 2.74996549e-01 2.52839118e-01 -6.46446526e-01
-3.57256472e-01 1.13836992e+00 4.27531958e-01 -3.25903207e-01
6.75026655e-01 -1.07478380e-01 -8.15387592e-02 2.36854434e-01
2.72676855e-01 2.69466221e-01 -1.01665128e-02 -4.46666628e-01
-4.36407834e-01 6.25728190e-01 -3.48304302e-01 2.84988463e-01
1.32975125e+00 -1.68722302e-01 -2.21282408e-01 3.62495959e-01
1.18596280e+00 -1.70303911e-01 -1.63252306e+00 -6.19603932e-01
4.97411996e-01 -1.77489206e-01 1.25608683e-01 -6.76250577e-01
-1.20255673e+00 7.15980232e-01 4.34288800e-01 2.30127290e-01
1.16002226e+00 2.71761835e-01 1.08856940e+00 1.62773877e-01
2.81461120e-01 -1.32816994e+00 4.05048639e-01 4.22459871e-01
3.25057417e-01 -1.46434259e+00 -1.42542377e-01 -7.07969606e-01
-4.67843741e-01 7.84317672e-01 1.02396131e+00 1.20746689e-02
5.65516651e-01 3.41036439e-01 8.95873923e-03 -7.04970732e-02
-4.41099197e-01 -4.89854842e-01 2.44749844e-01 2.33923003e-01
2.57832676e-01 -1.54961124e-01 -5.15811086e-01 6.64990246e-01
6.23524129e-01 9.12685916e-02 -1.34370431e-01 1.11195004e+00
-5.96096456e-01 -1.24243212e+00 -1.88711837e-01 3.11979026e-01
-6.80074751e-01 9.62629169e-03 -3.75839680e-01 6.38924003e-01
5.00429869e-01 9.03007448e-01 -1.34402037e-01 -1.25542298e-01
2.02956885e-01 5.15177473e-02 3.60392630e-02 -9.24939096e-01
-6.75880134e-01 3.90760839e-01 -2.10756689e-01 -6.69751108e-01
-8.60988915e-01 -5.56951880e-01 -1.49053299e+00 3.87483686e-01
-6.27016544e-01 1.47969246e-01 4.39745933e-01 1.14188135e+00
9.03282594e-03 5.38531601e-01 6.66170597e-01 -6.61561191e-01
-6.48541629e-01 -6.39038086e-01 -3.96076024e-01 4.79686081e-01
2.89845258e-01 -5.71985662e-01 -1.22893326e-01 2.78373390e-01] | [9.378549575805664, 1.0193181037902832] |
82933be2-1fac-44db-85e9-181c32cfeaf9 | don-t-guess-what-s-true-choose-what-s-optimal | 2302.10578 | null | https://arxiv.org/abs/2302.10578v1 | https://arxiv.org/pdf/2302.10578v1.pdf | Don't guess what's true: choose what's optimal. A probability transducer for machine-learning classifiers | In fields such as medicine and drug discovery, the ultimate goal of a classification is not to guess a class, but to choose the optimal course of action among a set of possible ones, usually not in one-one correspondence with the set of classes. This decision-theoretic problem requires sensible probabilities for the classes. Probabilities conditional on the features are computationally almost impossible to find in many important cases. The main idea of the present work is to calculate probabilities conditional not on the features, but on the trained classifier's output. This calculation is cheap, needs to be made only once, and provides an output-to-probability "transducer" that can be applied to all future outputs of the classifier. In conjunction with problem-dependent utilities, the probabilities of the transducer allow us to find the optimal choice among the classes or among a set of more general decisions, by means of expected-utility maximization. This idea is demonstrated in a simplified drug-discovery problem with a highly imbalanced dataset. The transducer and utility maximization together always lead to improved results, sometimes close to theoretical maximum, for all sets of problem-dependent utilities. The one-time-only calculation of the transducer also provides, automatically: (i) a quantification of the uncertainty about the transducer itself; (ii) the expected utility of the augmented algorithm (including its uncertainty), which can be used for algorithm selection; (iii) the possibility of using the algorithm in a "generative mode", useful if the training dataset is biased. | ['P. G. L. Porta Mana', 'A. S. Lundervold', 'K. Dyrland'] | 2023-02-21 | null | null | null | null | ['drug-discovery'] | ['medical'] | [ 4.51670676e-01 2.20186248e-01 -1.50527462e-01 -3.25851887e-01
-7.14333832e-01 -5.75873494e-01 5.08193493e-01 5.12851894e-01
-6.09524429e-01 9.61889923e-01 -4.11546052e-01 -5.74710250e-01
-4.90990847e-01 -1.03287578e+00 -3.62870514e-01 -1.08478653e+00
-6.32641688e-02 1.01568234e+00 1.46615461e-01 4.80169319e-02
3.54042411e-01 6.20209157e-01 -1.77608037e+00 1.40814200e-01
8.56594086e-01 1.06202590e+00 1.04501463e-01 5.67615926e-01
-1.58907533e-01 4.88287479e-01 -4.70663399e-01 -4.86282557e-01
1.66544825e-01 -7.68097401e-01 -9.01448429e-01 2.93364406e-01
-4.38319981e-01 9.80727598e-02 5.12492597e-01 1.22943842e+00
1.48768187e-01 1.44746929e-01 1.14443672e+00 -9.76690829e-01
8.70301351e-02 6.17534518e-01 -1.12821780e-01 -2.05481648e-01
3.97767216e-01 1.64621845e-01 1.12179160e+00 -3.54333729e-01
5.06709576e-01 9.50689435e-01 2.28566960e-01 3.25197399e-01
-1.43035400e+00 -3.10895890e-01 -1.60844088e-01 1.13240324e-01
-1.48032343e+00 -2.33118013e-01 4.36546326e-01 -5.36495328e-01
5.19150794e-01 5.99817216e-01 6.98863089e-01 3.75349343e-01
3.53985131e-01 4.98312026e-01 9.86265063e-01 -5.84304988e-01
8.15830529e-01 6.47221148e-01 2.02345978e-02 5.93105495e-01
2.93476552e-01 4.50862162e-02 -6.29568025e-02 -4.28248078e-01
3.86716515e-01 -5.48273548e-02 -2.98698485e-01 -3.84302109e-01
-9.85636353e-01 8.81410122e-01 1.70565639e-02 5.36648691e-01
-6.14169896e-01 -1.44035071e-01 1.78205982e-01 3.36030781e-01
2.22017020e-01 7.85262704e-01 -7.24117458e-01 -3.24564427e-02
-9.17994499e-01 3.55650336e-01 1.05844641e+00 4.78365451e-01
8.18660080e-01 -5.16998768e-01 -1.20150462e-01 4.50422376e-01
1.48716226e-01 3.45166981e-01 4.15513486e-01 -5.31272531e-01
5.02118431e-02 6.30915523e-01 1.96218669e-01 -5.24895072e-01
-4.43768442e-01 -4.48447615e-01 -6.41023993e-01 5.33485711e-01
9.34421480e-01 -2.06704974e-01 -7.29485810e-01 1.73802638e+00
4.00745630e-01 -2.90236294e-01 1.24517344e-02 6.26182675e-01
1.56973287e-01 5.33706903e-01 9.83790383e-02 -8.29600930e-01
1.27378654e+00 -4.86571379e-02 -5.95691741e-01 1.90159380e-01
9.54693913e-01 -6.73507035e-01 7.64509380e-01 7.88299441e-01
-8.41434121e-01 -6.17125668e-02 -9.13885236e-01 4.01830077e-01
-3.06905985e-01 8.56110975e-02 5.95773041e-01 8.87537956e-01
-8.69058073e-01 9.32609022e-01 -5.10587454e-01 -4.86917794e-02
2.97157288e-01 7.84223318e-01 -2.90904641e-01 1.19554371e-01
-1.03544390e+00 1.06242657e+00 5.80478370e-01 -5.49113937e-02
-4.12994653e-01 -2.06323758e-01 -5.28892696e-01 2.74886996e-01
4.64232594e-01 -6.11030936e-01 1.02215409e+00 -1.25406289e+00
-1.45281577e+00 6.93421245e-01 -5.83336800e-02 -2.72828013e-01
6.93550110e-01 5.21282136e-01 -2.29454190e-02 -7.92710781e-02
-1.55415805e-02 3.42944354e-01 6.48854792e-01 -7.47097671e-01
-6.71040237e-01 -4.75028634e-01 -7.71152452e-02 2.25217953e-01
1.37225971e-01 -2.79647410e-01 -1.50169045e-01 -2.52121449e-01
2.46437028e-01 -9.39731717e-01 -5.53284287e-01 -2.76261999e-04
-5.45905709e-01 -5.59240878e-01 1.91021830e-01 -3.19366217e-01
1.12662172e+00 -2.16359115e+00 2.60214895e-01 6.47489250e-01
-1.23106353e-01 -3.72960158e-02 3.29365283e-01 1.48014024e-01
-4.48946357e-01 1.12780035e-01 -4.61927056e-01 7.32168853e-02
-2.26246834e-01 1.47996038e-01 1.06705494e-01 7.09763408e-01
3.30598533e-01 2.96214134e-01 -8.80339563e-01 -6.15150928e-01
2.52871692e-01 4.03243564e-02 -5.75566530e-01 1.80092692e-01
-3.48004311e-01 1.53198034e-01 -8.54119718e-01 1.48445204e-01
4.29099441e-01 -1.53201044e-01 4.51746196e-01 1.13033261e-02
-6.12215400e-02 2.16345429e-01 -1.58504617e+00 1.01413369e+00
-2.99634099e-01 1.38117075e-01 -4.43248719e-01 -1.31595576e+00
1.08001304e+00 4.35737461e-01 6.92651272e-01 -9.45188776e-02
5.11143923e-01 3.45612079e-01 2.90410072e-01 -3.67144227e-01
-1.23566166e-02 -6.66782498e-01 4.01405618e-03 6.54778600e-01
1.23539679e-01 -1.92711130e-01 3.05351406e-01 -2.10279599e-01
9.97011900e-01 -1.51349425e-01 7.64944315e-01 -5.14191806e-01
5.74049175e-01 -1.88578859e-01 4.45807546e-01 5.88132918e-01
1.80019051e-01 4.16840523e-01 9.60867047e-01 -4.11439031e-01
-8.08375776e-01 -8.85023117e-01 -6.00012898e-01 5.34482539e-01
-6.01546951e-02 6.10899776e-02 -6.74736857e-01 -7.61718571e-01
-1.10237613e-01 8.98025632e-01 -7.64920175e-01 -6.84326962e-02
-1.27635244e-02 -9.77733076e-01 -1.06329821e-01 -1.58253253e-01
-5.14528640e-02 -9.91252780e-01 -7.24014997e-01 4.73870724e-01
9.02179033e-02 -4.11789507e-01 -4.15616371e-02 8.13495994e-01
-1.08064425e+00 -1.34436977e+00 -3.30246866e-01 -3.36803198e-01
8.11880529e-01 -3.44687253e-01 8.24688733e-01 8.27782303e-02
-8.95683691e-02 -5.35275694e-03 -2.38089323e-01 -5.66551745e-01
-6.67018354e-01 -1.72706962e-01 8.67232159e-02 2.07536727e-01
2.49958575e-01 -4.45058912e-01 -2.94445097e-01 2.56263226e-01
-8.34416568e-01 -3.45927268e-01 5.99776745e-01 8.76075208e-01
5.81073344e-01 5.48589110e-01 5.41530848e-01 -1.12853074e+00
3.56258005e-01 -4.50437129e-01 -6.76808715e-01 3.29659432e-01
-7.09605157e-01 6.51121140e-01 8.00253749e-01 -4.43985075e-01
-8.51081312e-01 4.25986648e-01 -4.18953776e-01 2.70432234e-02
-1.67485565e-01 4.02793974e-01 -5.80653131e-01 1.08217895e-01
9.40231323e-01 1.69636905e-01 1.23138778e-01 -4.80530709e-01
2.03285500e-01 7.09129512e-01 1.85092747e-01 -5.35443544e-01
2.84320623e-01 2.78973002e-02 3.37256432e-01 -8.28292131e-01
-4.82552826e-01 -2.96682864e-01 -6.09689176e-01 -1.21983826e-01
6.24530315e-01 -4.07342821e-01 -8.50946367e-01 1.15658350e-01
-9.81094480e-01 -3.56434509e-02 -7.14707553e-01 4.79319096e-01
-6.69509947e-01 1.67937025e-01 -3.06354612e-02 -1.15890729e+00
-7.60910520e-03 -1.53701878e+00 7.50831366e-01 1.50977716e-01
-3.70813161e-01 -8.64069045e-01 -2.67660290e-01 1.05539281e-02
-1.60539061e-01 1.12922035e-01 1.18785989e+00 -1.13353860e+00
-4.46363628e-01 -8.40263009e-01 1.37067422e-01 4.02054280e-01
2.09534019e-01 2.15455115e-01 -8.81940007e-01 -1.07609238e-02
2.08176926e-01 6.78981245e-02 6.95408881e-01 6.31451309e-01
1.31264329e+00 -3.94225925e-01 -4.18805063e-01 3.00706536e-01
1.28692949e+00 4.87971842e-01 7.03860939e-01 -1.13910824e-01
2.00731397e-01 6.73675299e-01 6.91562831e-01 5.83607078e-01
-1.40402406e-01 8.47129047e-01 3.92757624e-01 1.46931902e-01
5.04040122e-01 1.07918113e-01 -4.40100348e-03 1.47857949e-01
-1.22177228e-01 -3.22380930e-01 -5.09124279e-01 3.51224035e-01
-1.59619570e+00 -9.73230422e-01 -1.51873738e-01 2.83419752e+00
1.05010092e+00 3.47359300e-01 2.59898067e-01 6.23229563e-01
6.84524059e-01 -3.84863108e-01 -4.52422887e-01 -6.39275134e-01
3.08678031e-01 3.01117450e-01 3.86728466e-01 5.51366627e-01
-9.47367489e-01 1.22675724e-01 6.28565788e+00 1.17956436e+00
-9.74805415e-01 -2.17532828e-01 9.99073148e-01 1.86294928e-01
-5.07757723e-01 1.65670142e-01 -5.19990146e-01 6.76187575e-01
9.07224059e-01 -3.40474308e-01 3.66256088e-01 8.46624553e-01
-1.30146462e-02 -6.69872761e-01 -1.45366228e+00 6.72818244e-01
-4.19616073e-01 -1.13066864e+00 -2.35270217e-01 4.22768861e-01
4.14892435e-01 -5.03788233e-01 -3.32708865e-01 -5.30223288e-02
1.74567118e-01 -9.21356916e-01 5.35763800e-01 4.34879422e-01
8.12651753e-01 -1.11639571e+00 1.02557337e+00 6.92336917e-01
-6.13700628e-01 -1.31803170e-01 -2.19808266e-01 -1.73588708e-01
-2.26118878e-01 1.08164155e+00 -1.00907218e+00 4.35107440e-01
1.24704324e-01 1.63864523e-01 -5.17997518e-02 1.05566120e+00
-1.20911993e-01 4.07891244e-01 -6.56785786e-01 -5.84226608e-01
-1.18584763e-02 -3.61121237e-01 5.34943819e-01 1.20806038e+00
3.30528378e-01 3.84601623e-01 -1.37844374e-02 7.41760790e-01
2.48968661e-01 4.10077304e-01 -5.61062276e-01 3.71297188e-02
3.64523858e-01 1.04085076e+00 -1.14694309e+00 -3.30157280e-01
1.42491773e-01 6.09527588e-01 1.02299273e-01 5.65778166e-02
-4.71312404e-01 -4.03559566e-01 5.50881445e-01 1.27107784e-01
2.51282230e-02 3.06291133e-01 -5.68166375e-01 -7.69473135e-01
-4.25497107e-02 -7.18527377e-01 6.01614237e-01 -2.88719922e-01
-1.05821133e+00 4.57870722e-01 7.87519366e-02 -1.30878592e+00
-5.79118192e-01 -6.70316100e-01 -5.43998182e-01 1.24014485e+00
-8.35540950e-01 -3.42330217e-01 4.62556452e-01 3.89261723e-01
1.85870126e-01 -7.64234141e-02 9.88403440e-01 3.71076763e-02
-3.52355123e-01 4.99715388e-01 1.28286615e-01 -2.65805036e-01
2.93990463e-01 -1.36118269e+00 -4.12958920e-01 5.72330654e-01
4.07033078e-02 3.17426264e-01 1.19295132e+00 -3.75653744e-01
-1.17620015e+00 -7.49999940e-01 1.26085556e+00 -3.66456062e-01
5.41712701e-01 -1.16818063e-02 -8.74259591e-01 1.48690492e-01
-4.75138545e-01 -2.59242713e-01 6.13745928e-01 4.05299842e-01
1.78839117e-01 -1.84541568e-01 -1.44059885e+00 3.83068204e-01
3.79384786e-01 -1.11525908e-01 -4.17680800e-01 5.53415537e-01
2.91220635e-01 -2.70233959e-01 -7.29437828e-01 1.12737000e-01
6.37552142e-01 -9.95057881e-01 7.37931192e-01 -5.80709815e-01
3.29634547e-01 -2.46069148e-01 -3.77859436e-02 -1.20247102e+00
-3.12575877e-01 -3.25124383e-01 4.52388287e-01 8.74474108e-01
8.58874977e-01 -7.35348821e-01 6.82457805e-01 7.62572408e-01
2.98920512e-01 -1.08560324e+00 -1.26317620e+00 -5.15714645e-01
-2.20820695e-01 -6.13697588e-01 4.84246552e-01 9.11360383e-01
1.73077837e-01 2.96541274e-01 -1.60200909e-01 2.23567709e-02
5.95004797e-01 1.64039284e-01 2.85991281e-01 -1.41811109e+00
-7.30947673e-01 -5.55661500e-01 -7.13171422e-01 -6.62571430e-01
-2.40377262e-01 -7.69326091e-01 2.91812927e-01 -1.13209760e+00
1.81131557e-01 -7.37948298e-01 -2.21138299e-01 5.40847778e-01
-2.26914436e-01 -2.23508120e-01 -3.06812786e-02 1.09954113e-02
-2.12923974e-01 1.03265978e-01 1.12078428e+00 8.51747021e-02
-3.70708227e-01 7.28932440e-01 -6.98505044e-01 6.94883347e-01
5.72829068e-01 -7.86243439e-01 -4.12155241e-01 5.32819152e-01
3.99452567e-01 5.72362959e-01 -2.16054991e-02 -7.84057319e-01
-6.25033379e-02 -4.24040020e-01 4.72543180e-01 -2.04514965e-01
1.54834837e-02 -9.72335517e-01 5.30189693e-01 6.57853246e-01
-4.21642989e-01 -4.13526118e-01 -2.40176573e-01 4.00340259e-01
1.12685174e-01 -9.60389793e-01 8.86945307e-01 -5.01461364e-02
-1.15404293e-01 7.14660808e-02 -4.89858568e-01 -3.18932027e-01
1.08441329e+00 -2.86299109e-01 2.56697595e-01 -3.06893378e-01
-1.00588739e+00 1.36690587e-01 3.53326201e-01 -1.83888689e-01
2.79801935e-01 -9.95566487e-01 -6.30427063e-01 1.92990884e-01
2.12969743e-02 2.67304648e-02 3.12289660e-04 6.17239773e-01
-3.06562364e-01 2.29521245e-01 -1.93852410e-02 -5.85283041e-01
-1.13688827e+00 7.10170388e-01 4.38184023e-01 -4.05225068e-01
-1.80271506e-01 7.22290874e-01 1.10381171e-01 6.13514800e-03
6.43391302e-03 -3.76603782e-01 -3.06862473e-01 3.08810264e-01
5.07505596e-01 1.20051481e-01 2.49166802e-01 -2.04346091e-01
-3.49707156e-01 2.06576675e-01 1.32908270e-01 -2.08807841e-01
1.22645199e+00 1.97393045e-01 -1.90360457e-01 5.41823208e-01
1.04500008e+00 -1.79448262e-01 -9.19578731e-01 -1.78936049e-02
3.51712666e-02 -4.33392823e-01 1.75078288e-01 -9.58727419e-01
-7.75816858e-01 7.85031617e-01 5.66150904e-01 5.78420758e-01
1.51770997e+00 -1.07812487e-01 1.00346068e-02 3.34538758e-01
6.68744624e-01 -9.08752620e-01 -5.29457033e-01 -4.21625115e-02
6.94801152e-01 -1.10430098e+00 2.77280807e-01 -3.74998063e-01
-6.66873991e-01 1.38957000e+00 -5.27720787e-02 7.67237172e-02
6.80124342e-01 2.86381900e-01 -2.48192817e-01 -7.80880824e-03
-7.99573302e-01 -3.21761787e-01 3.89614195e-01 4.31424260e-01
4.21987414e-01 4.22044694e-01 -8.15719783e-01 5.15925825e-01
5.68839256e-03 1.56309798e-01 4.67309803e-01 6.51538849e-01
-7.31334567e-01 -1.34546793e+00 -4.43770438e-01 8.81775200e-01
-3.27111095e-01 1.74778923e-01 -1.90374359e-01 5.43647408e-01
4.90155041e-01 8.43853474e-01 8.10074806e-02 -2.80073464e-01
3.35910797e-01 2.17914462e-01 4.89218503e-01 -5.96686065e-01
-3.50875586e-01 1.23229854e-01 3.12905282e-01 -3.38825375e-01
-7.65006393e-02 -8.91246676e-01 -1.18428910e+00 -2.76521951e-01
-6.75849140e-01 6.38018906e-01 8.19498658e-01 1.35817850e+00
-1.66435376e-01 2.74980903e-01 8.73083651e-01 -6.95374310e-01
-7.71268249e-01 -6.93382800e-01 -9.67542052e-01 2.26762831e-01
2.21600771e-01 -6.33649707e-01 -4.33711171e-01 -2.67764460e-02] | [7.798662185668945, 4.5949273109436035] |
e24aaab6-e362-4392-8521-c336d0d118de | 4dcontrast-contrastive-learning-with-dynamic | 2112.02990 | null | https://arxiv.org/abs/2112.02990v2 | https://arxiv.org/pdf/2112.02990v2.pdf | 4DContrast: Contrastive Learning with Dynamic Correspondences for 3D Scene Understanding | We present a new approach to instill 4D dynamic object priors into learned 3D representations by unsupervised pre-training. We observe that dynamic movement of an object through an environment provides important cues about its objectness, and thus propose to imbue learned 3D representations with such dynamic understanding, that can then be effectively transferred to improved performance in downstream 3D semantic scene understanding tasks. We propose a new data augmentation scheme leveraging synthetic 3D shapes moving in static 3D environments, and employ contrastive learning under 3D-4D constraints that encode 4D invariances into the learned 3D representations. Experiments demonstrate that our unsupervised representation learning results in improvement in downstream 3D semantic segmentation, object detection, and instance segmentation tasks, and moreover, notably improves performance in data-scarce scenarios. | ['Angela Dai', 'Matthias Nießner', 'Yujin Chen'] | 2021-12-06 | null | null | null | null | ['3d-instance-segmentation-1', 'unsupervised-pre-training'] | ['computer-vision', 'methodology'] | [ 4.29770142e-01 4.06344861e-01 -3.15515608e-01 -4.99734819e-01
-4.15534735e-01 -7.44286835e-01 7.55222440e-01 7.43849054e-02
-1.57973528e-01 -1.33594498e-01 2.55185604e-01 -3.15201998e-01
-1.30266219e-01 -7.52182782e-01 -1.13394880e+00 -4.02498424e-01
-5.30176274e-02 8.10279250e-01 4.15760487e-01 -1.41037479e-01
2.39863515e-01 1.05273688e+00 -1.47915375e+00 1.47003904e-01
6.26648784e-01 7.26892114e-01 4.81253296e-01 7.45777845e-01
-1.27866209e-01 2.46915966e-01 -1.83200508e-01 2.46680811e-01
7.03076541e-01 7.18155652e-02 -1.04856658e+00 8.09674859e-01
2.73382187e-01 -3.67056191e-01 -3.29898328e-01 6.35070622e-01
1.76561307e-02 4.50869888e-01 9.58554864e-01 -8.62389624e-01
-1.02850270e+00 9.30301771e-02 -6.88495517e-01 2.34021068e-01
3.06040227e-01 4.07711059e-01 8.23832989e-01 -7.94944942e-01
8.89324784e-01 1.74401510e+00 4.32623357e-01 5.65370262e-01
-1.45276403e+00 -2.00904757e-01 8.56921434e-01 -1.29298910e-01
-6.40300095e-01 -3.43093693e-01 1.05960715e+00 -5.41254640e-01
1.06760108e+00 -1.24440327e-01 6.12429857e-01 1.37779927e+00
-2.06154302e-01 1.16441369e+00 9.09190476e-01 -3.82611513e-01
2.28362888e-01 -3.17228943e-01 3.16545367e-01 6.16199136e-01
2.00634554e-01 1.50395572e-01 -4.07872349e-01 3.84009540e-01
1.13033807e+00 2.50189126e-01 2.24476635e-01 -8.86145175e-01
-1.19386125e+00 5.46715796e-01 7.81145871e-01 1.31746642e-02
-2.13687867e-01 3.24213207e-01 2.37292409e-01 1.48986265e-01
7.20618308e-01 4.22611505e-01 -9.11613822e-01 1.16992667e-01
-2.28934288e-01 3.42442542e-01 2.25443512e-01 1.16456532e+00
8.97138894e-01 4.71629985e-02 -2.34243833e-02 6.62899852e-01
3.73029709e-01 9.18808460e-01 3.51811558e-01 -1.27911580e+00
4.55007434e-01 8.16103637e-01 1.90478355e-01 -7.53653884e-01
-5.13878405e-01 -2.91378856e-01 -3.23768914e-01 2.50710279e-01
3.05151314e-01 4.76936847e-01 -1.63973677e+00 1.59871280e+00
5.33351958e-01 3.48382175e-01 8.57002363e-02 9.22061324e-01
6.75739050e-01 4.90929514e-01 1.22476101e-01 3.59587580e-01
1.12076712e+00 -9.14400101e-01 -1.92678526e-01 -5.31912029e-01
5.79385042e-01 -3.57955158e-01 1.23958826e+00 -1.35239199e-01
-9.62047696e-01 -9.38683093e-01 -8.39028478e-01 -3.29684108e-01
-3.82075787e-01 -3.14655453e-01 9.36942816e-01 3.99132878e-01
-8.31444383e-01 4.85101819e-01 -1.39133263e+00 -4.37548816e-01
1.05097795e+00 3.18563670e-01 -4.37714487e-01 -2.20799088e-01
-4.74965215e-01 7.52936721e-01 5.42095959e-01 -1.72884881e-01
-1.28722775e+00 -9.31738973e-01 -1.30657351e+00 -3.31919283e-01
4.29880828e-01 -8.32376122e-01 1.15948665e+00 -6.13893867e-01
-1.49360740e+00 1.35338938e+00 -1.23078056e-01 -3.47349793e-01
3.75578612e-01 -4.38676775e-01 2.51453966e-01 2.24219084e-01
2.23373890e-01 1.06315792e+00 9.76115167e-01 -1.66723144e+00
-1.89779028e-01 -8.12089264e-01 3.64360690e-01 3.84233266e-01
1.30133167e-01 -4.50375944e-01 -5.33875346e-01 -7.60249436e-01
8.08937788e-01 -1.04176664e+00 -6.85662508e-01 2.82384843e-01
-3.78130376e-01 -1.32948905e-01 1.10407722e+00 -3.90446454e-01
-1.25784636e-01 -1.91863477e+00 2.71911144e-01 2.75877304e-02
3.68300569e-03 1.61958486e-01 -4.86305952e-01 -1.04206249e-01
-4.35837917e-02 2.12478966e-01 -3.41957808e-01 -4.83773649e-01
4.13414724e-02 6.67123258e-01 -5.64756155e-01 4.44011331e-01
7.32663155e-01 1.43970358e+00 -1.09206438e+00 6.89401925e-02
5.70126176e-01 4.50222701e-01 -8.66986930e-01 3.80939186e-01
-7.76020229e-01 1.11833298e+00 -1.11521661e+00 6.90597653e-01
7.98882723e-01 -3.78647089e-01 -2.22178295e-01 4.19987515e-02
1.96299255e-01 3.06896001e-01 -7.73956001e-01 2.24899507e+00
-6.16081595e-01 3.64579946e-01 -2.04244956e-01 -1.45409119e+00
1.00273061e+00 -2.47318834e-01 5.70482016e-01 -7.18616605e-01
5.93295358e-02 -1.85031280e-01 -3.07468921e-01 -6.44047856e-01
2.47743428e-01 7.73672713e-04 -1.87736869e-01 4.47745085e-01
1.17715962e-01 -8.45451772e-01 -3.21441919e-01 7.59904012e-02
8.13133180e-01 8.12074661e-01 -7.96632841e-02 -2.67532736e-01
1.42594039e-01 -2.68766508e-02 2.41808653e-01 7.74807215e-01
-1.50878787e-01 6.37921810e-01 2.97668070e-01 -4.57659841e-01
-1.07603347e+00 -1.44345677e+00 -9.84134674e-02 1.14741278e+00
5.47921479e-01 2.90911287e-01 -3.47927123e-01 -1.06781137e+00
3.48143756e-01 7.01415062e-01 -8.61951530e-01 -3.11374962e-01
-7.68261254e-01 -4.59509850e-01 9.51063633e-02 9.98869419e-01
4.38850582e-01 -1.03369629e+00 -7.93319106e-01 -1.67893041e-02
2.69632906e-01 -1.48810065e+00 -7.11440817e-02 3.56216818e-01
-1.36685753e+00 -1.01206446e+00 -5.87638915e-01 -8.23621213e-01
9.70166028e-01 7.28584349e-01 9.14852560e-01 -1.75404608e-01
-3.59330088e-01 9.54172671e-01 -5.72673440e-01 -3.06011856e-01
-5.05781412e-01 -2.91495826e-02 1.36432797e-01 -2.23749340e-01
1.10101424e-01 -6.75161779e-01 -4.79956448e-01 3.32927287e-01
-7.37583160e-01 1.16448253e-01 4.42226768e-01 5.86742103e-01
8.46475244e-01 -3.27421099e-01 4.84875143e-01 -8.89447629e-01
-2.74816245e-01 -2.12899014e-01 -5.43809772e-01 -6.78705126e-02
-2.94365972e-01 4.11207765e-01 1.88050792e-01 -4.38435614e-01
-1.24820518e+00 2.49406949e-01 -1.58728078e-01 -7.41799891e-01
-6.41275644e-01 -1.08047910e-01 -3.19969237e-01 8.04460607e-03
5.65033913e-01 4.05125879e-02 3.64006273e-02 -8.44076812e-01
9.23441648e-01 6.97922185e-02 4.40117627e-01 -9.65097725e-01
1.22606242e+00 9.88730967e-01 3.73049565e-02 -6.05767429e-01
-1.25225639e+00 -4.99858975e-01 -1.23530042e+00 1.10656135e-01
1.10265887e+00 -1.05382907e+00 -4.84940708e-01 4.45391268e-01
-1.01578808e+00 -7.44927049e-01 -6.50144994e-01 2.74892330e-01
-1.05444813e+00 2.22799450e-01 -2.28577793e-01 -5.69055557e-01
1.82094544e-01 -1.15195894e+00 1.70011914e+00 6.19912520e-02
-8.05583298e-02 -1.10762668e+00 -2.25299865e-01 6.16131544e-01
-4.07021679e-03 4.28423584e-01 1.08887732e+00 -6.05899453e-01
-9.15132523e-01 4.62511592e-02 -4.21509534e-01 2.62123913e-01
3.18280190e-01 -2.92165399e-01 -1.11204958e+00 -2.01709777e-01
-1.21794909e-01 -4.68326390e-01 1.03865194e+00 4.83308673e-01
1.70237815e+00 -5.21690864e-03 -5.34820914e-01 8.21584940e-01
9.24570382e-01 -9.23012197e-02 1.27041593e-01 2.85024166e-01
9.07169342e-01 7.86036789e-01 6.94738150e-01 2.92749822e-01
3.08225542e-01 6.11516237e-01 7.52698481e-01 -2.12692022e-01
-3.99177223e-01 -5.80443263e-01 -1.01936068e-02 4.79822665e-01
-1.29267573e-01 -7.31448382e-02 -8.98455799e-01 6.00388706e-01
-1.75265431e+00 -4.64828104e-01 1.72821775e-01 1.67412043e+00
4.60675538e-01 3.26705068e-01 -1.54228732e-01 -1.28146350e-01
4.05730456e-01 3.18488568e-01 -1.25036168e+00 4.83146161e-02
-1.09091885e-02 4.39116210e-01 4.88172978e-01 3.93645555e-01
-1.10540903e+00 1.46024692e+00 6.53459597e+00 3.29094172e-01
-8.66797507e-01 -3.04820971e-03 7.25070238e-01 2.31492504e-01
-6.15424573e-01 -5.59205599e-02 -5.70303380e-01 -6.84680790e-02
3.02327722e-01 2.01173890e-02 9.31770951e-02 9.83652234e-01
4.02787805e-01 2.26050287e-01 -1.36723518e+00 8.39994252e-01
-1.79973960e-01 -1.46650779e+00 5.67602813e-01 8.01524669e-02
9.50750530e-01 1.36101088e-02 4.51076627e-01 3.58925968e-01
5.09597600e-01 -9.33875799e-01 7.78059244e-01 3.00766945e-01
5.48501194e-01 -5.74823439e-01 8.72598067e-02 5.05919158e-01
-1.06502569e+00 -5.49389012e-02 -3.37861329e-01 -5.30998223e-02
1.78810045e-01 2.61233658e-01 -9.39282358e-01 4.81486291e-01
7.49012411e-01 1.30317938e+00 -4.81223196e-01 5.65902352e-01
-3.54174554e-01 4.48722273e-01 -2.07000360e-01 3.43441308e-01
3.65660071e-01 -1.02624446e-01 8.05961072e-01 8.98105323e-01
2.60920618e-02 1.34287015e-01 5.00032127e-01 1.12586308e+00
-1.72175169e-01 -2.99576283e-01 -8.76618743e-01 7.36696348e-02
1.36856407e-01 7.50233412e-01 -1.07959175e+00 -2.40477145e-01
-8.81083682e-02 1.02609015e+00 2.68954396e-01 6.71680272e-01
-5.75187683e-01 3.42093289e-01 1.06827939e+00 7.76753649e-02
6.60665452e-01 -7.59127438e-01 -4.55303967e-01 -1.21116936e+00
-4.66117971e-02 -3.39504302e-01 1.03520624e-01 -7.89629042e-01
-1.26862192e+00 3.28866728e-02 3.19132984e-01 -9.51007724e-01
-1.29027274e-02 -1.06298375e+00 -3.98775041e-01 3.85927886e-01
-1.79509592e+00 -1.49966693e+00 -2.15112507e-01 4.64684010e-01
1.13001132e+00 1.66404396e-01 5.84855139e-01 -3.31538409e-01
-1.52276337e-01 1.89545766e-01 -2.32885092e-01 -2.87854094e-02
2.95048445e-01 -1.23774326e+00 8.86762559e-01 6.52982056e-01
3.64022404e-01 3.93279046e-01 3.20785284e-01 -5.87269723e-01
-1.51732528e+00 -1.42619634e+00 -9.94161665e-02 -1.10117686e+00
4.99031216e-01 -7.71010041e-01 -9.86872494e-01 9.09242630e-01
-3.58876705e-01 3.90509844e-01 3.60596985e-01 2.20362738e-01
-6.17019892e-01 1.98075548e-01 -1.13090110e+00 4.99714881e-01
1.94842875e+00 -5.79557419e-01 -8.54980946e-01 5.82821608e-01
1.47105813e+00 -6.24012053e-01 -7.79857874e-01 7.92961895e-01
1.73299402e-01 -5.07209897e-01 1.53236365e+00 -1.19301426e+00
2.43986741e-01 -7.29487911e-02 -3.50999713e-01 -1.14389229e+00
-1.05513342e-01 -5.79082549e-01 -2.72393316e-01 7.26384819e-01
2.34116212e-01 -3.55844438e-01 1.09387600e+00 3.75607073e-01
-4.43176776e-01 -4.36357826e-01 -8.50319982e-01 -9.21173751e-01
3.14366251e-01 -8.44124377e-01 5.34474254e-01 8.15324187e-01
-7.20863640e-01 5.01324497e-02 1.87076747e-01 5.04804254e-01
6.57614470e-01 4.96819049e-01 1.00901401e+00 -1.29972768e+00
-1.42918870e-01 -4.03964788e-01 -7.09796429e-01 -2.09359026e+00
6.93384588e-01 -1.17827761e+00 1.17451653e-01 -1.46380770e+00
6.78278059e-02 -7.15846837e-01 -2.29788199e-01 5.34327090e-01
-2.08828613e-01 3.99235599e-02 1.72284737e-01 2.19830647e-01
-5.69960475e-01 8.86928320e-01 1.79078138e+00 -3.95412445e-01
-4.21567738e-01 1.29393041e-01 -5.93510211e-01 8.12027156e-01
3.72743040e-01 -1.96239784e-01 -7.55470157e-01 -8.52683902e-01
-3.68519932e-01 -3.54203939e-01 6.10624611e-01 -6.24085069e-01
-4.21365678e-01 -3.60763490e-01 6.49439156e-01 -6.84950471e-01
5.00895143e-01 -7.50411808e-01 -6.80555522e-01 2.51783371e-01
-4.45927292e-01 -3.82934868e-01 4.92618263e-01 1.06501806e+00
2.35797271e-01 2.11460918e-01 7.79591620e-01 -4.39172208e-01
-1.18672287e+00 6.47110164e-01 -1.31125405e-01 2.04911277e-01
1.02696252e+00 -4.97617453e-01 7.88065791e-02 8.33563134e-02
-1.20790327e+00 3.63545686e-01 6.36424780e-01 8.64540219e-01
7.07447529e-01 -1.08056021e+00 -5.19383848e-01 5.32102823e-01
2.61761308e-01 6.64581895e-01 2.31048867e-01 2.26291224e-01
-3.55982006e-01 1.69554830e-01 -2.39701509e-01 -1.36583412e+00
-6.92113578e-01 6.12271667e-01 1.52234048e-01 1.50567651e-01
-1.13827026e+00 9.91447687e-01 7.54071653e-01 -9.61569071e-01
1.75347373e-01 -8.35556686e-01 -1.17738463e-01 -3.08445007e-01
8.79831091e-02 3.31317075e-02 -9.60620940e-02 -5.69072485e-01
-2.65743464e-01 1.04402256e+00 -1.00205578e-01 2.93565765e-02
1.64058864e+00 -2.41150901e-01 4.02105480e-01 4.32176024e-01
1.21794891e+00 -3.64105284e-01 -2.01074553e+00 -3.44899386e-01
8.35798979e-02 -7.33371794e-01 -9.82641354e-02 -5.15506625e-01
-9.34325814e-01 9.81103539e-01 5.08885741e-01 -1.71141908e-01
8.09806585e-01 6.86322212e-01 6.25773430e-01 7.14487851e-01
4.55603689e-01 -7.64205396e-01 7.49881148e-01 7.44004190e-01
9.19623733e-01 -1.47901738e+00 -1.21560708e-01 -6.41009986e-01
-4.60748434e-01 1.02268612e+00 6.97916090e-01 -4.20743674e-01
7.63342440e-01 -5.69324978e-02 4.35178764e-02 -3.10157597e-01
-4.49863017e-01 -5.13697445e-01 4.74502712e-01 1.09683025e+00
-2.28713512e-01 -1.21542916e-01 7.38239706e-01 2.36543745e-01
9.01999250e-02 -5.65474987e-01 1.33439988e-01 8.93256545e-01
-4.18907911e-01 -9.01000977e-01 -7.39775002e-02 -4.07823250e-02
2.34465137e-01 4.64891434e-01 -2.47090071e-01 9.25207317e-01
2.71889180e-01 5.22479057e-01 2.93900788e-01 5.35756089e-02
6.12003505e-01 -3.11865993e-02 7.52451360e-01 -1.06564689e+00
1.45801440e-01 3.27829458e-02 -2.80739427e-01 -7.30429053e-01
-6.71240747e-01 -6.87917292e-01 -1.64414573e+00 4.63042468e-01
-5.65692186e-02 -3.61662567e-01 6.63947344e-01 1.14826894e+00
5.50603092e-01 6.97912931e-01 7.33315229e-01 -1.42974949e+00
-4.45070118e-01 -6.92044377e-01 -1.19638547e-01 6.96520984e-01
5.72618961e-01 -1.13038623e+00 -2.65985429e-01 3.58136833e-01] | [8.25251579284668, -3.1235873699188232] |
1f8db83c-e42b-48de-882f-52407306e4a1 | mcbert-momentum-contrastive-learning-with | 2203.12940 | null | https://arxiv.org/abs/2203.12940v2 | https://arxiv.org/pdf/2203.12940v2.pdf | mcBERT: Momentum Contrastive Learning with BERT for Zero-Shot Slot Filling | Zero-shot slot filling has received considerable attention to cope with the problem of limited available data for the target domain. One of the important factors in zero-shot learning is to make the model learn generalized and reliable representations. For this purpose, we present mcBERT, which stands for momentum contrastive learning with BERT, to develop a robust zero-shot slot filling model. mcBERT uses BERT to initialize the two encoders, the query encoder and key encoder, and is trained by applying momentum contrastive learning. Our experimental results on the SNIPS benchmark show that mcBERT substantially outperforms the previous models, recording a new state-of-the-art. Besides, we also show that each component composing mcBERT contributes to the performance improvement. | ['Jong-Hyeok Lee', 'WonKee Lee', 'Seong-Hwan Heo'] | 2022-03-24 | null | null | null | null | ['zero-shot-slot-filling', 'slot-filling'] | ['natural-language-processing', 'natural-language-processing'] | [-1.5557232e-01 2.6068762e-01 -7.0135331e-01 -1.2167629e-01
-9.6118325e-01 3.4467480e-01 4.4146901e-01 -5.6625992e-02
-3.5994723e-01 9.4791132e-01 1.2667401e-01 -3.2693322e-03
1.2600592e-01 -7.9063487e-01 -6.2364137e-01 -5.2879447e-01
1.3011277e-02 7.1942413e-01 7.3973650e-01 -3.6941969e-01
-6.4280242e-02 -3.4115085e-01 -1.4627993e+00 1.9014755e-01
7.8667301e-01 1.1828918e+00 6.3256341e-01 3.2181337e-01
-4.4320402e-01 1.0683422e+00 -3.9659527e-01 -4.8525313e-01
8.4951743e-02 -5.4354978e-01 -7.8637969e-01 7.2113283e-02
-2.5881770e-01 -4.8720172e-01 -5.4595089e-01 9.8251200e-01
4.6501771e-01 5.4733491e-01 4.8829892e-01 -1.3364123e+00
-5.5910671e-01 9.7098690e-01 -4.0964314e-01 4.1144431e-01
-1.9402134e-01 -1.9147550e-01 1.1083490e+00 -1.2152368e+00
6.2287509e-01 1.1690093e+00 4.6729931e-01 6.9207579e-01
-8.0878848e-01 -5.2540225e-01 8.6529791e-02 6.4660662e-01
-1.3495466e+00 -4.9183479e-01 6.2480015e-01 -1.7882426e-01
1.0401391e+00 8.6024217e-03 4.7352338e-01 7.8961450e-01
-1.0366970e-01 1.2559340e+00 5.6944227e-01 -6.1839181e-01
5.3645808e-01 2.5045452e-01 2.8741708e-01 7.8207856e-01
6.3998871e-02 1.2597343e-01 -7.4928081e-01 -2.0994724e-01
7.1762693e-01 1.5384656e-01 -9.3974564e-03 -6.6345668e-01
-6.4482260e-01 1.1904672e+00 4.2773467e-01 2.8783983e-01
-3.3862644e-01 1.9550693e-01 4.0432876e-01 1.6201894e-01
7.8652871e-01 3.4690508e-01 -3.4503531e-01 -8.0164474e-01
-9.9244487e-01 4.7633182e-02 7.3880196e-01 1.1576581e+00
8.4379768e-01 2.8156137e-01 -6.8157625e-01 1.1274388e+00
2.2654426e-01 8.7923547e-03 7.3786390e-01 -7.2699982e-01
3.2709029e-01 3.6047339e-01 2.4878351e-02 -2.8777233e-01
-3.4031443e-02 -3.5530135e-01 -5.7729721e-01 -1.5200579e-01
-2.4688713e-01 1.0006285e-02 -1.2208971e+00 1.5468204e+00
1.3801259e-01 5.9769100e-01 2.9909036e-01 8.8230664e-01
9.8201585e-01 8.3049577e-01 1.8226892e-01 -3.0939180e-01
1.1307448e+00 -1.5851909e+00 -8.7118125e-01 -4.5480469e-01
3.9442223e-01 -3.8611671e-01 1.0542636e+00 6.9760390e-02
-1.1221200e+00 -3.4512818e-01 -1.3103504e+00 -1.5803190e-01
-2.6428556e-01 2.9938836e-03 8.2733053e-01 3.6945099e-01
-6.3239968e-01 8.3153188e-01 -1.0015095e+00 -1.9087555e-01
5.9996498e-01 -1.8293135e-02 4.0372066e-02 -1.5399690e-01
-1.5831238e+00 8.6781776e-01 7.5537211e-01 -4.3838161e-01
-1.1089211e+00 -4.9059513e-01 -1.0498298e+00 5.2865165e-01
8.8526702e-01 -3.4314364e-01 1.5838586e+00 -3.0707338e-01
-1.5086266e+00 5.1117003e-01 -2.3767050e-01 -9.1283411e-01
5.3134239e-01 -4.2100045e-01 -2.4059714e-01 3.7613388e-02
3.3183712e-01 6.2937701e-01 7.3552728e-01 -9.5224261e-01
-5.8496314e-01 1.3457544e-04 -1.2827753e-02 1.7375351e-01
-3.9523223e-01 -1.8296584e-01 -1.1085062e+00 -5.6383258e-01
3.4710612e-02 -6.0332555e-01 -2.8286490e-01 -1.1801952e-01
-1.4852731e-01 -4.2193344e-01 8.2705861e-01 -5.1502222e-01
1.3838955e+00 -2.3613188e+00 1.5167536e-01 -2.6310384e-01
1.2406575e-01 4.6952930e-01 -2.1909796e-02 4.1708049e-01
3.0613497e-01 -1.8990794e-01 -2.6867193e-01 -6.9146931e-01
3.8041189e-02 5.4855609e-01 -4.5201868e-01 1.3140574e-01
2.6182872e-01 1.1282408e+00 -1.0913209e+00 -5.1967967e-01
2.8747517e-01 4.1066119e-01 -5.9422356e-01 3.1150788e-01
-3.6623871e-01 -1.2574045e-01 -2.5006199e-01 7.5954413e-01
4.1794056e-01 -6.1033136e-01 1.5970005e-01 2.2034222e-01
1.3553187e-01 2.8784093e-01 -1.1155984e+00 1.8978714e+00
-5.0488275e-02 4.5101282e-01 -3.7298039e-01 -1.0851184e+00
9.2071301e-01 3.2255596e-01 4.5424485e-01 -9.8581034e-01
2.6384798e-01 2.4246013e-01 -3.7263584e-01 -2.9168656e-02
8.9633846e-01 -1.8568689e-01 -1.7248290e-02 2.5467652e-01
6.7574960e-01 -7.8808762e-02 5.1648676e-01 5.6753629e-01
8.2046664e-01 -5.6377448e-02 4.6111980e-01 1.1704224e-02
8.3101891e-02 -2.0990150e-01 9.0896517e-01 7.9024130e-01
-3.1822124e-01 7.0180720e-01 6.5507066e-01 -1.6202788e-01
-9.9454546e-01 -9.7370332e-01 -7.1997784e-02 1.3355007e+00
4.8371080e-01 -7.6911211e-01 -4.2654505e-01 -5.3057569e-01
-4.5654390e-02 8.6878258e-01 -5.9724146e-01 -5.9152484e-01
-4.3155145e-02 -6.5576494e-01 1.4568831e-01 6.8015015e-01
6.3953614e-01 -1.0801234e+00 -7.4083382e-01 4.8113459e-01
-2.1674193e-01 -9.5122421e-01 -1.8027397e-01 5.5868745e-01
-8.1891739e-01 -1.0153394e+00 -1.0242618e+00 -6.4175224e-01
1.3171524e-01 6.2680840e-01 1.1693633e+00 3.9096616e-02
-4.3304712e-01 -8.4683977e-02 -8.1816596e-01 -3.7753454e-01
-4.5738062e-03 2.3798360e-01 -2.4216298e-02 -4.6041217e-01
3.7243566e-01 -3.9616403e-01 -2.1251519e-01 3.0615178e-01
-8.4649056e-01 1.6124561e-01 3.1735703e-01 1.2918353e+00
7.6371545e-01 3.3993367e-02 5.2043706e-01 -1.1413914e+00
5.5846816e-01 -7.4906886e-01 -4.7159651e-01 3.6374402e-01
-7.9918039e-01 2.0484978e-01 7.9549700e-02 -3.1286031e-01
-9.3313134e-01 -2.4198046e-01 -7.7658698e-02 -7.8855479e-01
6.4613199e-01 5.8409470e-01 -2.4973897e-02 1.3982937e-01
7.0013618e-01 3.3319354e-01 -2.6594961e-01 -6.3310790e-01
6.1367756e-01 5.6063652e-01 4.1659355e-01 -5.2179551e-01
4.7743043e-01 2.3820165e-01 -5.7753968e-01 -7.3772043e-01
-1.1427116e+00 -8.5779637e-01 -2.8739440e-01 1.5886357e-01
4.4365171e-01 -1.1150858e+00 3.5652574e-02 3.8321313e-01
-7.6004970e-01 -3.5875529e-01 -8.7797266e-01 2.9660261e-01
-8.0350071e-01 2.0871490e-01 -8.2109898e-01 -1.0853937e+00
-5.9226948e-01 -9.9824637e-01 8.1821990e-01 4.8025504e-01
1.8853183e-01 -5.8676493e-01 2.8076583e-01 2.2839988e-02
5.8140492e-01 -3.8969928e-01 7.2471887e-01 -6.3037705e-01
-4.4555417e-01 -1.0141560e-01 -3.2467502e-01 2.8643027e-01
-1.1763406e-01 -4.9936298e-01 -1.1077085e+00 -3.2935172e-01
-7.7332452e-02 -8.9410430e-01 1.4694734e+00 3.7515604e-01
1.1955237e+00 -8.4257923e-02 -3.0939877e-01 6.2440473e-01
1.2718191e+00 2.7370483e-01 8.7773710e-01 3.0936685e-01
3.6890626e-01 -5.5480286e-02 9.9043405e-01 8.6184490e-01
3.7055600e-01 6.7082071e-01 3.5243979e-01 -1.8640492e-02
-5.6505177e-02 -5.9414774e-01 -3.3515755e-02 7.9048043e-01
3.0680839e-04 -8.4924825e-02 -6.3248235e-01 5.3804231e-01
-2.3105013e+00 -9.6561599e-01 5.6445181e-01 1.9587147e+00
9.0236819e-01 4.1043565e-01 7.7211797e-02 7.5025477e-02
6.2178320e-01 5.1057160e-01 -7.6995909e-01 -1.2837615e-02
-1.4619622e-03 4.5716253e-01 5.0134093e-01 2.8017521e-01
-1.2668767e+00 1.4599332e+00 6.5696964e+00 1.1460652e+00
-9.2522973e-01 5.0467598e-01 5.8299637e-01 -2.8188068e-01
-1.2286180e-01 8.3538249e-02 -9.4546324e-01 4.3121874e-01
8.1167746e-01 -4.7651222e-01 4.1811064e-01 1.4148673e+00
-4.3194145e-01 -9.0168804e-02 -6.9123608e-01 1.0024818e+00
2.8899411e-02 -1.5207052e+00 -2.2838281e-01 -1.6313702e-01
8.7219179e-01 2.1254018e-01 -8.5570335e-02 1.2813458e+00
5.2288246e-01 -7.9689485e-01 7.0896727e-01 2.9278797e-01
9.3371302e-01 -8.0743498e-01 7.2972715e-01 5.2053720e-01
-1.3180505e+00 5.0731882e-02 -1.1521671e+00 -5.3363465e-02
3.9094368e-01 5.7148963e-01 -8.7451154e-01 3.9741930e-01
4.2065784e-01 7.7523261e-01 -1.7593303e-01 1.3326148e+00
-2.3360606e-01 5.9410197e-01 -4.7121514e-03 1.7309973e-02
3.1112522e-01 2.4733868e-01 3.4685940e-01 8.3343130e-01
3.5073200e-01 2.0258990e-01 5.4927057e-01 1.0203581e+00
-2.5122777e-01 1.8114846e-02 -1.4366679e-01 -4.9163514e-01
6.9614893e-01 1.0460958e+00 -6.6548532e-01 -6.7658311e-01
-4.6422404e-01 1.1462307e+00 7.3982430e-01 2.6607201e-01
-8.3840871e-01 -4.1198379e-01 5.5519831e-01 -2.0711099e-01
6.8450201e-01 -3.5404039e-03 2.7584540e-02 -1.3827218e+00
-2.1530162e-01 -4.5160791e-01 4.4808298e-01 -4.4027406e-01
-9.7254974e-01 5.2375704e-01 -7.9150371e-02 -1.1743344e+00
-4.5809150e-01 -2.2509031e-01 -5.5217677e-01 8.4627473e-01
-1.7464705e+00 -8.6207408e-01 -1.2536278e-01 4.5924115e-01
1.0196676e+00 -3.5566765e-01 9.5561004e-01 2.1100822e-01
-5.4642594e-01 5.6268394e-01 1.8375458e-01 -9.2500836e-02
4.1548020e-01 -1.0912191e+00 7.4831206e-01 6.2911278e-01
3.8788319e-01 3.8133594e-01 5.1116633e-01 -7.9345500e-01
-1.0286006e+00 -9.7270322e-01 7.9850566e-01 6.6173576e-02
4.7679046e-01 -1.6228041e-01 -1.0642883e+00 5.0760019e-01
-1.4120592e-01 2.6684260e-01 6.4508820e-01 3.3424628e-01
-3.0577454e-01 3.0403140e-01 -9.3306011e-01 2.2558306e-01
8.7796354e-01 -6.1011034e-01 -8.3179045e-01 2.1718881e-01
1.0276428e+00 -7.3279887e-01 -3.2621437e-01 3.5131016e-01
2.6729620e-01 -8.3968455e-01 8.4101450e-01 -5.7950515e-01
4.3389064e-01 2.7895552e-01 -3.3827049e-01 -1.3611246e+00
-5.5678093e-01 -2.7487648e-01 -9.0181202e-01 1.0365922e+00
2.9048207e-01 -2.4931747e-01 1.0299746e+00 6.5094775e-01
-2.1816498e-01 -1.0098273e+00 -1.0711404e+00 -9.6626848e-01
-1.4447203e-01 -4.3816477e-01 5.3729492e-01 7.7872878e-01
3.7406701e-01 4.8030788e-01 -1.0385655e+00 -5.5786222e-01
3.5945240e-01 -1.1405685e-01 5.9746736e-01 -1.3022487e+00
-5.2942675e-01 -1.5310316e-01 -2.6615638e-01 -1.1213932e+00
8.0592334e-03 -8.5737056e-01 2.9405373e-01 -1.7735555e+00
5.9388596e-01 -3.6872038e-01 -6.9426197e-01 6.0836303e-01
-4.9964622e-01 5.8511935e-02 4.6748212e-01 3.5881159e-01
-1.1094857e+00 1.2500377e+00 9.7347075e-01 -2.7753493e-01
-3.6694816e-01 1.1538610e-02 -7.9761952e-01 4.4425267e-01
5.9894955e-01 -4.0599838e-01 -6.5931654e-01 -1.7822762e-01
-1.0035679e-01 5.6057405e-02 -1.2220628e-02 -1.0653517e+00
1.7934440e-01 2.2684755e-02 1.7507394e-01 -6.5164071e-01
7.9187781e-01 -4.8984817e-01 -3.2472020e-01 5.3324395e-01
-2.7515018e-01 -4.9892470e-01 2.0489065e-01 5.9429020e-01
-3.4942320e-01 -4.7956064e-01 7.8872609e-01 -3.0785954e-01
-1.3049438e+00 5.9223258e-01 1.1399607e-01 3.9634740e-01
9.4126356e-01 1.4705198e-01 -3.6971495e-01 -3.0938363e-01
-6.7979622e-01 5.1484871e-01 2.3407364e-01 5.6908500e-01
8.0472082e-01 -1.5601944e+00 -4.1813567e-01 3.6184949e-01
4.1324481e-01 -7.5430743e-02 3.3106524e-01 4.5045894e-01
-9.0869501e-02 4.7520158e-01 -1.0398767e-01 -2.3994203e-01
-8.1919616e-01 4.5528537e-01 1.4149797e-01 -5.0256824e-01
-8.8022506e-01 1.2439666e+00 -6.6227472e-04 -1.5212929e-01
5.9670931e-01 1.0537746e-02 -2.5763485e-01 7.0133328e-02
8.5711902e-01 4.3452695e-01 -3.7270110e-02 -2.5875416e-01
-2.3929414e-01 -2.7892247e-01 -4.4834727e-01 -2.4659915e-01
1.3863449e+00 6.0972620e-02 3.5517633e-01 6.9661754e-01
9.1219515e-01 -6.8022126e-01 -1.4736832e+00 -6.8972665e-01
-9.7942144e-02 -7.1971250e-01 2.3260066e-01 -6.1348784e-01
-9.3840605e-01 9.6515501e-01 6.2453246e-01 6.6440113e-02
7.0098329e-01 1.3137609e-01 9.9263942e-01 4.3200937e-01
8.4401453e-01 -1.3955593e+00 1.9094537e-01 8.3795482e-01
5.5414426e-01 -1.5258305e+00 -1.5368201e-01 -1.8878385e-01
-9.8697084e-01 7.0125049e-01 7.4347562e-01 -1.0573361e-02
6.6742975e-01 7.0178114e-02 -2.9937080e-01 -1.5648589e-01
-9.8395258e-01 -5.5690122e-01 1.8087623e-01 3.6576951e-01
3.6877730e-01 1.7240158e-01 -5.4261565e-01 1.1051258e+00
4.8889581e-02 4.5573536e-01 6.5901294e-02 1.1844298e+00
-1.0309498e+00 -1.1421422e+00 -2.1525709e-02 5.8850724e-01
-2.6647899e-01 -2.3610556e-01 -1.2507825e-01 4.4001386e-01
-1.6021544e-01 6.7841417e-01 -3.9404754e-02 -4.0895328e-01
1.7039709e-01 2.4561474e-01 1.9148821e-01 -9.3240905e-01
-2.5386898e-02 1.6936256e-01 -3.2076295e-02 -6.1009961e-01
2.0907059e-02 -2.5178972e-01 -1.3749713e+00 -1.5738280e-01
-4.7136578e-01 5.4539651e-01 3.6690670e-01 9.2621356e-01
3.0958325e-01 7.8852826e-01 3.6210138e-01 -7.1993399e-01
-7.1347260e-01 -1.1583200e+00 -1.0305467e+00 1.7569013e-01
2.0102618e-02 -1.3141727e+00 -6.9883972e-02 -4.7194758e-01] | [9.994941711425781, 3.055532932281494] |
990543f5-1209-4318-a9b6-a332967a28f1 | back-to-the-future-sequential-alignment-of | 1909.03464 | null | https://arxiv.org/abs/1909.03464v3 | https://arxiv.org/pdf/1909.03464v3.pdf | Back to the Future -- Sequential Alignment of Text Representations | Language evolves over time in many ways relevant to natural language processing tasks. For example, recent occurrences of tokens 'BERT' and 'ELMO' in publications refer to neural network architectures rather than persons. This type of temporal signal is typically overlooked, but is important if one aims to deploy a machine learning model over an extended period of time. In particular, language evolution causes data drift between time-steps in sequential decision-making tasks. Examples of such tasks include prediction of paper acceptance for yearly conferences (regular intervals) or author stance prediction for rumours on Twitter (irregular intervals). Inspired by successes in computer vision, we tackle data drift by sequentially aligning learned representations. We evaluate on three challenging tasks varying in terms of time-scales, linguistic units, and domains. These tasks show our method outperforming several strong baselines, including using all available data. We argue that, due to its low computational expense, sequential alignment is a practical solution to dealing with language evolution. | ['Isabelle Augenstein', 'Wouter Kouw', 'Johannes Bjerva'] | 2019-09-08 | null | null | null | null | ['rumour-detection'] | ['natural-language-processing'] | [ 4.80608381e-02 -1.40085727e-01 -2.48687401e-01 -4.83280092e-01
-3.63205224e-01 -6.58735394e-01 1.28024018e+00 7.65154898e-01
-7.59603977e-01 9.04560864e-01 3.70850980e-01 -2.66710877e-01
5.51896021e-02 -4.90151972e-01 -8.32714736e-01 -4.48788851e-01
-1.82923585e-01 5.94786286e-01 -1.28328381e-03 -3.51939172e-01
3.92452002e-01 3.33685786e-01 -1.24904859e+00 4.02724028e-01
5.52693367e-01 5.42904019e-01 -5.45279980e-02 5.15073359e-01
-3.48706186e-01 8.45484376e-01 -8.57175112e-01 -4.99746859e-01
-1.49952456e-01 -2.73175955e-01 -9.38659966e-01 -1.35788307e-01
2.46261016e-01 4.20776814e-01 -2.71153629e-01 1.04964292e+00
1.38653994e-01 1.25590414e-01 7.55371034e-01 -9.68859851e-01
-7.84493685e-01 1.14250779e+00 -9.05415952e-01 8.95018637e-01
1.46052867e-01 -8.31278339e-02 1.08420932e+00 -4.70807105e-01
9.95901287e-01 1.45898223e+00 7.88390100e-01 3.46960783e-01
-1.29498076e+00 -3.72195840e-01 7.15472102e-01 2.88128495e-01
-9.69899535e-01 -3.61381531e-01 9.25043464e-01 -8.55136812e-01
8.58243167e-01 1.32698610e-01 3.10522735e-01 1.74167013e+00
4.31944847e-01 1.05413544e+00 8.90430748e-01 -4.87343907e-01
9.37065035e-02 2.10131168e-01 3.93739164e-01 3.60765129e-01
2.45655343e-01 -4.59848374e-01 -9.45090950e-01 -2.13723958e-01
1.27507433e-01 -3.43525931e-02 -4.49122339e-02 2.54611790e-01
-1.46316922e+00 6.94280565e-01 3.43553662e-01 9.59948957e-01
-5.59598684e-01 -8.27046186e-02 4.79906708e-01 6.24896467e-01
9.36970472e-01 6.25604510e-01 -6.08773291e-01 -3.97286564e-01
-1.12427199e+00 4.57518697e-01 8.01869273e-01 6.23417199e-01
4.87717479e-01 -4.44765948e-02 -1.54367745e-01 8.73293102e-01
-6.27218783e-02 7.14198053e-02 1.06466627e+00 -4.94055003e-01
4.30046618e-01 4.81233537e-01 2.39862800e-01 -1.03074360e+00
-6.23311400e-01 -6.62728012e-01 -1.00142193e+00 -3.68917227e-01
7.13733017e-01 -8.95386413e-02 -5.00972867e-01 1.81975424e+00
-1.98799632e-02 -1.74981095e-02 -9.06591788e-02 6.12616837e-01
2.57808596e-01 5.96677721e-01 1.57623261e-03 -5.88761091e-01
1.37675643e+00 -6.39209390e-01 -7.95621455e-01 -5.66193223e-01
4.91068751e-01 -6.19603038e-01 9.09241438e-01 5.13832808e-01
-8.21702600e-01 -3.34283799e-01 -6.89803720e-01 8.58948603e-02
-5.88471651e-01 -2.17262983e-01 2.83101350e-01 1.01351045e-01
-7.22431123e-01 9.12244022e-01 -7.76534498e-01 -7.10175812e-01
2.17015177e-01 -5.00575788e-02 5.34880459e-02 3.56400311e-01
-1.42569113e+00 8.14136744e-01 1.62976131e-01 1.04928344e-01
-1.75914362e-01 -6.46588206e-01 -2.81418622e-01 -3.08308661e-01
1.23783939e-01 -5.21484256e-01 1.38795328e+00 -1.27320576e+00
-1.19193423e+00 1.27106774e+00 -5.02502382e-01 -1.07551026e+00
1.10527277e+00 -3.87408078e-01 -6.43225133e-01 -4.47526097e-01
1.76707864e-01 2.08218411e-01 9.58973169e-01 -5.82374930e-01
-7.41123974e-01 -5.75807571e-01 -3.94281358e-01 3.40327970e-03
-6.25249863e-01 2.78387249e-01 -2.92005360e-01 -8.46579254e-01
-1.18212946e-01 -8.95961881e-01 -1.71138585e-01 -6.48229420e-01
-1.29256710e-01 -8.14351916e-01 4.90511268e-01 -8.32821012e-01
1.50385070e+00 -1.84243417e+00 4.24344689e-01 -1.22983776e-01
1.42815307e-01 -2.32252166e-01 1.21063642e-01 2.36147046e-01
1.78598791e-01 3.23252141e-01 -1.11210637e-01 -4.55614746e-01
-9.80369523e-02 8.87077525e-02 -5.39206922e-01 5.15668154e-01
2.07212061e-01 8.79807830e-01 -1.09807789e+00 -1.34118527e-01
-4.80382860e-01 2.35245943e-01 -7.66455606e-02 -2.17621386e-01
-4.88309711e-01 5.00740230e-01 -2.47712985e-01 5.95753253e-01
6.32972270e-02 -4.01762605e-01 1.87443942e-01 3.44874889e-01
-4.13480908e-01 6.46626353e-01 -5.85257113e-01 1.67161846e+00
-2.35843137e-01 1.24378514e+00 -2.39225462e-01 -1.07235491e+00
9.34879184e-01 1.25917509e-01 4.96041983e-01 -8.22109342e-01
1.56137887e-02 1.95082471e-01 -3.25226337e-02 -3.94967616e-01
7.82839060e-01 6.92288354e-02 -3.17080140e-01 5.86794853e-01
-3.07448953e-01 2.54931659e-01 3.18609774e-01 1.84876382e-01
1.25303876e+00 -1.33751303e-01 2.42259592e-01 -3.99634123e-01
3.30142409e-01 4.50051986e-02 7.30677903e-01 9.19790745e-01
-3.16785246e-01 3.68489802e-01 7.30618298e-01 -7.97795177e-01
-1.08001935e+00 -5.69499552e-01 -1.60686359e-01 1.36943758e+00
-3.99119318e-01 -2.74523765e-01 -4.45375055e-01 -5.98415256e-01
2.52988696e-01 8.68613541e-01 -9.11752045e-01 -1.13118134e-01
-9.22327399e-01 -1.01318944e+00 4.49664026e-01 2.42368117e-01
4.66838814e-02 -1.21442199e+00 -6.09910846e-01 6.63757145e-01
-2.56213099e-01 -9.67296839e-01 -2.77667820e-01 2.19160303e-01
-8.52040172e-01 -6.68080568e-01 -8.31839919e-01 -3.34938973e-01
2.92304307e-01 8.01643208e-02 1.43829286e+00 -7.99453706e-02
-1.92765012e-01 3.40001941e-01 -2.41230860e-01 -7.87926316e-01
-6.06105268e-01 5.97258806e-01 5.22647977e-01 2.19524533e-01
6.64080799e-01 -3.99188310e-01 -4.48561668e-01 -5.09035635e-05
-7.74033725e-01 -3.62615377e-01 3.31848800e-01 7.88713157e-01
6.22424968e-02 -5.69972284e-02 6.45661533e-01 -1.04721975e+00
1.04345095e+00 -9.01886761e-01 -5.28838277e-01 2.95512110e-01
-8.01171482e-01 2.20579728e-01 4.46721166e-01 -7.23330855e-01
-8.04324806e-01 -5.75056076e-01 5.80257833e-01 -1.35248572e-01
-8.56745616e-02 7.94500172e-01 4.85364914e-01 7.73824513e-01
9.23786283e-01 6.25521764e-02 -7.84231573e-02 -6.30314529e-01
5.15807644e-02 8.94026101e-01 5.30935407e-01 -6.37832701e-01
4.96088982e-01 4.74741310e-01 -5.22996008e-01 -1.08076715e+00
-9.63905752e-01 -5.67566991e-01 -8.36390376e-01 -1.86507314e-01
3.16557676e-01 -9.70170498e-01 -3.36340606e-01 5.25359809e-01
-1.45127857e+00 -4.94648404e-02 -6.32798597e-02 2.95379370e-01
-2.44967967e-01 1.22667946e-01 -5.94807327e-01 -7.62685180e-01
-2.07445994e-01 -6.22319043e-01 7.56706834e-01 1.96747169e-01
-1.05923998e+00 -1.26386642e+00 2.34635532e-01 2.32881054e-01
4.49144214e-01 3.51021171e-01 7.43070424e-01 -7.98344374e-01
-2.19116434e-01 -1.82941496e-01 1.68587133e-01 -2.05511618e-02
2.17078224e-01 1.96028277e-01 -8.88995290e-01 -3.34691793e-01
-5.67419939e-02 4.64388393e-02 1.16200447e+00 4.03095752e-01
1.00532985e+00 -4.89920765e-01 -4.76278573e-01 -1.46605941e-02
7.79440582e-01 5.65498807e-02 -5.11857718e-02 8.64771307e-01
5.21515429e-01 9.32218373e-01 3.71542454e-01 5.32653213e-01
4.37086701e-01 7.72918701e-01 -9.11061838e-02 2.57083118e-01
2.31098309e-01 -1.38041943e-01 5.81532001e-01 1.09013844e+00
-1.28202215e-01 -2.06614241e-01 -1.30645430e+00 1.03648019e+00
-2.18644404e+00 -1.09356964e+00 -1.92893595e-01 2.20332026e+00
1.04511714e+00 4.47307348e-01 3.73514503e-01 -1.16182588e-01
6.25012934e-01 3.03254873e-01 -8.14783752e-01 -4.13333148e-01
-5.65778553e-01 -4.75146830e-01 5.18501401e-01 3.89263004e-01
-1.10127532e+00 8.95387709e-01 6.10579872e+00 2.13855952e-01
-1.34216356e+00 2.45197311e-01 6.73280418e-01 -3.80285382e-01
-2.13270903e-01 -2.52173007e-01 -9.93885458e-01 7.94964194e-01
1.24668872e+00 -6.15302563e-01 2.32734635e-01 4.68862116e-01
3.07749122e-01 1.49082661e-01 -1.35028470e+00 9.39523697e-01
1.70247376e-01 -1.30613089e+00 -1.18077606e-01 -3.99527512e-02
8.15272570e-01 4.35418278e-01 1.13918908e-01 2.84497172e-01
5.62113523e-01 -8.08174312e-01 1.03699923e+00 8.33431005e-01
1.14541098e-01 -4.44731563e-01 4.87090230e-01 6.65209293e-01
-5.62797070e-01 -1.37684122e-01 -3.15829396e-01 -1.53069481e-01
6.33963272e-02 1.03518319e+00 -1.09618318e+00 3.77628744e-01
9.46724892e-01 1.14357781e+00 -7.95197010e-01 6.37261152e-01
-9.62702706e-02 8.84633958e-01 -2.56178200e-01 -2.19561860e-01
2.40550816e-01 -4.93159555e-02 8.19624543e-01 1.53325009e+00
2.08049089e-01 -5.22420943e-01 -1.47488996e-01 7.50900745e-01
-4.63187754e-01 -1.13921473e-02 -7.56361485e-01 -3.90600145e-01
4.43782777e-01 8.48616421e-01 -5.95133007e-01 -3.29276800e-01
-2.60648936e-01 1.14864028e+00 4.88480240e-01 4.06248957e-01
-6.49908841e-01 3.46831791e-03 7.81710804e-01 3.00100267e-01
8.78711939e-02 -3.76810491e-01 -3.97352368e-01 -1.40375149e+00
2.17239529e-01 -7.87116289e-01 5.84006846e-01 -2.81424046e-01
-1.58641732e+00 7.34033465e-01 -2.65132785e-01 -1.03716302e+00
-6.16159737e-01 -3.76311272e-01 -3.19748312e-01 6.43086851e-01
-1.47200465e+00 -8.77812386e-01 5.58058098e-02 2.21838713e-01
8.86525452e-01 -4.68720138e-01 4.05965149e-01 4.66505848e-02
-6.13224268e-01 3.24898154e-01 4.30949062e-01 1.79336682e-01
1.07152772e+00 -1.32072222e+00 7.80330360e-01 8.75660360e-01
5.95647454e-01 7.06372738e-01 1.15427113e+00 -6.75429165e-01
-1.01548064e+00 -9.52849090e-01 1.58862877e+00 -9.52727377e-01
1.17390907e+00 -3.95082325e-01 -1.32743537e+00 7.66506016e-01
3.78343642e-01 -4.19797361e-01 3.81956011e-01 6.83783591e-01
-5.20314217e-01 -1.46430414e-02 -4.67186630e-01 5.85674703e-01
9.44935143e-01 -5.80889821e-01 -9.72709775e-01 6.34258926e-01
5.82367539e-01 -2.65770942e-01 -6.41089678e-01 -3.17001162e-04
5.11863649e-01 -5.01250565e-01 6.97432160e-01 -1.08378685e+00
4.11240458e-01 -1.08220778e-01 1.55585766e-01 -1.41133285e+00
-4.90463942e-01 -9.62843060e-01 -2.68765092e-01 1.42102301e+00
5.06032765e-01 -4.84198332e-01 5.50074279e-01 4.48905528e-01
5.28890371e-01 -2.94119298e-01 -9.35383856e-01 -8.17065120e-01
4.16770399e-01 -2.49758199e-01 4.43308800e-01 1.46052432e+00
-1.27667800e-01 5.85269392e-01 -4.44190562e-01 -1.22210965e-01
4.63333368e-01 1.43092200e-01 5.85848331e-01 -1.87298357e+00
-7.85154663e-03 -9.78272259e-01 -2.97856331e-01 -7.75449991e-01
4.24615860e-01 -8.56754601e-01 -1.84218019e-01 -1.11030567e+00
-3.31367254e-02 -1.87573150e-01 -7.30290830e-01 2.97728658e-01
-2.12239474e-01 -2.31683120e-01 -2.29939213e-03 6.24255359e-01
-7.07310498e-01 2.14245737e-01 5.21683753e-01 -5.95320344e-01
-3.42372864e-01 1.89407811e-01 -7.44794369e-01 7.56145835e-01
7.92954922e-01 -6.13793671e-01 1.99772611e-01 -5.68788707e-01
7.83499837e-01 -3.54632288e-01 1.96198002e-01 -6.76745832e-01
4.77430403e-01 -1.78905174e-01 2.34936550e-01 -3.94453496e-01
-1.04912646e-01 -4.25689340e-01 -8.98392405e-03 3.58108014e-01
-6.26638830e-01 3.58684748e-01 3.24950591e-02 8.36554229e-01
-3.88713211e-01 -7.33020455e-02 5.90433240e-01 -2.74022996e-01
-5.38760066e-01 -2.09243931e-02 -5.64531446e-01 9.40592587e-02
7.52219796e-01 2.34594457e-02 -3.31591576e-01 -2.63559580e-01
-6.76699817e-01 3.34623873e-01 3.14873904e-01 1.17888498e+00
-8.29233602e-02 -1.02204728e+00 -9.77639973e-01 -9.39305946e-02
9.12473351e-02 -6.84773549e-02 -1.38791233e-01 9.77145791e-01
-2.34176833e-02 4.54789937e-01 4.46077585e-02 -7.99087524e-01
-1.27741456e+00 2.80578077e-01 1.07849032e-01 -5.68051338e-01
-5.74144721e-01 9.26548302e-01 -1.74836203e-01 -2.33285218e-01
4.15274382e-01 -4.21996772e-01 -2.93159992e-01 8.64583254e-01
6.09302580e-01 2.23330721e-01 3.50906760e-01 -3.98715228e-01
-4.48981881e-01 1.14687040e-01 -7.37708747e-01 -1.68981716e-01
1.63394094e+00 -2.52270430e-01 -2.88566619e-01 1.51708829e+00
1.00014818e+00 -3.13561335e-02 -1.01626909e+00 -8.22300971e-01
9.42880929e-01 -1.70391407e-02 -1.36040524e-01 -7.48362482e-01
-4.47197348e-01 7.61453807e-01 2.66094714e-01 5.91024876e-01
5.67328930e-01 1.18102551e-01 3.61210972e-01 3.91100973e-01
3.20111334e-01 -1.39685547e+00 1.34512577e-02 8.93275976e-01
1.09904468e+00 -1.30901098e+00 2.45064199e-02 5.28034091e-01
-5.50677001e-01 1.11481249e+00 3.46002042e-01 2.67565936e-01
4.62028056e-01 -7.95122832e-02 8.58155861e-02 -3.21123824e-02
-1.24267375e+00 2.14896068e-01 2.34031662e-01 -1.61472708e-02
7.68153429e-01 1.75905153e-01 -4.30127859e-01 5.10909021e-01
-3.99369240e-01 -8.70162323e-02 5.68806350e-01 7.91654825e-01
-2.45963365e-01 -1.03254533e+00 -3.62101644e-01 3.51468086e-01
-7.18183815e-01 1.31401196e-01 -8.16908240e-01 4.04761225e-01
-1.55093923e-01 5.83742857e-01 3.79196554e-01 -4.88491617e-02
1.61585301e-01 4.54007298e-01 1.54951259e-01 -3.02218974e-01
-7.52073228e-01 -6.33174032e-02 3.66255231e-02 -1.44013539e-01
-6.05512321e-01 -1.35628462e+00 -8.11586320e-01 -2.85373837e-01
2.62498468e-01 1.97690979e-01 6.39536440e-01 9.86447513e-01
4.05904561e-01 7.12284923e-01 4.63928640e-01 -4.55844402e-01
-5.56709409e-01 -1.27262366e+00 -3.36442888e-01 6.53407097e-01
5.86404026e-01 -3.53849620e-01 -4.50857043e-01 3.36298227e-01] | [10.071425437927246, 8.621700286865234] |
8d2aedbb-0c92-40f9-b7fb-ee34acecdd4e | rapid-model-comparison-by-amortizing-across | null | null | https://openreview.net/forum?id=rylGty24YB | https://openreview.net/pdf?id=rylGty24YB | Rapid Model Comparison by Amortizing Across Models | Comparing the inferences of diverse candidate models is an essential part of model checking and escaping local optima. To enable efficient comparison, we introduce an amortized variational inference framework that can perform fast and reliable posterior estimation across models of the same architecture. Our Any Parameter Encoder (APE) extends the encoder neural network common in amortized inference to take both a data feature vector and a model parameter vector as input. APE thus reduces posterior inference across unseen data and models to a single forward pass. In experiments comparing candidate topic models for synthetic data and product reviews, our Any Parameter Encoder yields comparable posteriors to more expensive methods in far less time, especially when the encoder architecture is designed in model-aware fashion. | ['Michael C. Hughes', 'Lily H. Zhang'] | 2019-10-16 | null | null | null | pproximateinference-aabi-symposium-2019-12 | ['topic-models'] | ['natural-language-processing'] | [ 1.75740346e-02 2.40996331e-01 -3.85438472e-01 -6.11796796e-01
-1.44686151e+00 -7.47035980e-01 8.23246956e-01 -4.60300893e-02
-3.50059450e-01 8.50941718e-01 -1.45476624e-01 -5.15208542e-01
5.71773238e-02 -7.09917545e-01 -1.21056199e+00 -4.55179632e-01
1.97983697e-01 9.72961426e-01 1.22060217e-01 3.76315653e-01
1.35564893e-01 6.28147721e-02 -1.41257691e+00 6.57391846e-02
5.87057531e-01 1.02335227e+00 1.87817335e-01 1.04522526e+00
1.18000522e-01 3.91750634e-01 -5.15706182e-01 -1.04228508e+00
1.01248473e-01 -1.05598211e-01 -6.49880886e-01 1.18644057e-04
6.49252236e-01 -5.61130464e-01 1.38268014e-02 1.15219665e+00
2.41245717e-01 1.28662601e-01 9.54914451e-01 -1.20802665e+00
-4.72177923e-01 8.86674106e-01 -5.35131454e-01 2.82242119e-01
-1.30131677e-01 3.53608578e-01 1.42998827e+00 -1.01824427e+00
4.21454757e-01 1.60029936e+00 5.55836916e-01 2.65219301e-01
-1.89196074e+00 -6.52718782e-01 4.36567247e-01 -3.71237993e-02
-1.51952565e+00 -4.67769235e-01 4.88517702e-01 -3.11589509e-01
1.17921531e+00 1.44039467e-01 5.97826302e-01 1.45847142e+00
5.12290478e-01 9.91634011e-01 3.91301900e-01 -5.36304340e-02
7.32443035e-01 4.03421193e-01 1.98282197e-01 6.82618499e-01
5.51473975e-01 -4.73387390e-02 -5.49196541e-01 -7.20220685e-01
4.31204647e-01 -1.33276388e-01 8.02950114e-02 -2.70096034e-01
-8.50003183e-01 1.26651585e+00 -1.63608104e-01 -5.53991914e-01
-3.33587736e-01 9.32735741e-01 3.66894424e-01 2.84416676e-01
6.16915107e-01 5.04145622e-01 -8.42493892e-01 -3.95580024e-01
-1.16686344e+00 6.75833046e-01 1.05234063e+00 1.07460999e+00
7.16952503e-01 1.15193851e-01 -8.04432705e-02 6.79741085e-01
7.06500709e-01 5.65420866e-01 3.06987464e-01 -1.25742340e+00
3.34960014e-01 2.04648077e-01 4.30599540e-01 -5.21030784e-01
2.37608358e-01 -3.61159205e-01 -4.92274523e-01 -1.20649636e-01
1.14405483e-01 -4.76987869e-01 -1.01100147e+00 1.89246559e+00
4.35608417e-01 2.74018794e-01 -6.09977357e-02 2.22940281e-01
1.61134869e-01 9.34056699e-01 9.54264998e-02 -1.13249123e-01
1.44370544e+00 -7.53272951e-01 -5.52573502e-01 -4.34266150e-01
4.22262400e-01 -5.57617009e-01 9.36022937e-01 6.48613870e-01
-1.41116798e+00 -4.12559152e-01 -1.13483167e+00 -3.14718723e-01
-2.95151770e-01 1.46151751e-01 6.87701464e-01 5.70368230e-01
-9.60249484e-01 5.07474303e-01 -1.23550284e+00 2.03130901e-01
5.59333265e-01 4.50884908e-01 6.36964440e-02 1.16918989e-01
-1.14913583e+00 9.06801999e-01 5.76076865e-01 4.69184332e-02
-1.39000404e+00 -1.06260586e+00 -9.88385975e-01 3.39254171e-01
3.89280796e-01 -9.39014375e-01 1.82081318e+00 -6.32578135e-01
-1.54951513e+00 2.34823212e-01 -5.81104815e-01 -9.03770626e-01
4.28829849e-01 -2.96398550e-01 -2.45056197e-01 -3.69150817e-01
-7.89680928e-02 9.52186525e-01 1.27209902e+00 -9.55435097e-01
-5.43170273e-01 3.13533805e-02 -1.04692116e-01 -1.67748749e-01
1.49030248e-02 -1.94656655e-01 -5.92361987e-01 -3.42413753e-01
-2.34216288e-01 -8.79745781e-01 -1.65461138e-01 2.25960627e-01
-6.56181991e-01 -5.13377249e-01 5.52693129e-01 -5.65391302e-01
1.29701591e+00 -1.80006170e+00 1.84886888e-01 2.09850982e-01
-9.61764529e-03 -1.22113921e-01 1.10830620e-01 2.00352103e-01
2.44113535e-01 3.49814057e-01 -4.99806821e-01 -8.37363541e-01
4.90547329e-01 4.41868335e-01 -5.75978518e-01 1.85840890e-01
5.56665778e-01 9.16255593e-01 -5.41706204e-01 -7.17716873e-01
7.15957284e-02 5.45658171e-01 -8.74414146e-01 1.15514740e-01
-9.01080489e-01 -3.58680040e-01 -3.96543950e-01 5.59089363e-01
7.09249794e-01 -6.22339547e-01 2.65966296e-01 -1.27965854e-02
2.85518169e-01 6.37288630e-01 -1.33093369e+00 1.50856543e+00
-6.16989613e-01 7.05302417e-01 -1.18310876e-01 -5.86453795e-01
6.86107814e-01 2.01162636e-01 -1.23238236e-01 -4.49325331e-02
-4.70786169e-03 1.56778455e-01 -1.61465108e-01 -1.38402969e-01
6.42153263e-01 8.16556043e-04 -2.22016856e-01 6.26081705e-01
2.71757185e-01 -3.72437507e-01 3.33712935e-01 3.44738901e-01
7.05661178e-01 1.78242162e-01 1.29850477e-01 -2.70332098e-01
6.64084926e-02 -2.11193282e-02 5.62385917e-01 1.13572538e+00
3.00095379e-01 4.01826382e-01 7.76430845e-01 -4.51026291e-01
-1.14119875e+00 -1.29545391e+00 -5.70235372e-01 9.76842463e-01
-3.33301783e-01 -6.76622808e-01 -6.43786669e-01 -6.49375677e-01
2.62849808e-01 1.34559500e+00 -7.42377579e-01 -1.58788458e-01
-2.73075044e-01 -9.25630093e-01 4.19727266e-01 7.15918422e-01
7.73648992e-02 -5.61052084e-01 -5.96077144e-01 3.96597117e-01
8.55756849e-02 -7.19930053e-01 -4.30990398e-01 4.70663726e-01
-8.59497666e-01 -6.88619077e-01 -4.02362883e-01 -3.38284910e-01
4.54533279e-01 -4.96284992e-01 1.47376919e+00 -4.29152131e-01
-1.20058566e-01 -4.68999445e-02 4.99787986e-01 -7.49305844e-01
-5.83212495e-01 2.42094651e-01 1.54102072e-02 -2.22581148e-01
6.53202057e-01 -2.54390001e-01 -3.41446012e-01 2.87440754e-02
-7.69346356e-01 1.01239206e-02 3.01634073e-01 9.49334800e-01
8.21649253e-01 2.43337750e-02 2.18851641e-01 -9.57318544e-01
6.84309781e-01 -8.15347135e-01 -1.34165525e+00 5.21494150e-01
-7.78348386e-01 5.60440063e-01 4.15067166e-01 -4.76255566e-01
-1.26301706e+00 -2.02624984e-02 -6.36935234e-02 -5.50178885e-01
1.99177489e-01 4.58444208e-01 1.16090206e-02 7.10340738e-01
5.00552654e-01 -7.16953864e-03 9.47610661e-02 -6.00917637e-01
3.72365057e-01 4.96504188e-01 3.19672734e-01 -6.77262068e-01
6.72101080e-01 1.11795954e-01 -2.24732950e-01 -3.80817503e-01
-9.01431859e-01 7.62450024e-02 -2.91699380e-01 3.52534354e-01
6.85699284e-01 -1.23915863e+00 -7.10377574e-01 5.83251268e-02
-1.35251749e+00 -3.76111746e-01 -3.99248600e-01 4.67607766e-01
-4.60787654e-01 -1.90950140e-01 -5.53296268e-01 -1.00826871e+00
-3.62270653e-01 -1.34342194e+00 1.37049711e+00 1.11112520e-01
-4.60221797e-01 -1.22612441e+00 3.17218095e-01 -2.03950837e-01
3.25015217e-01 -1.61027744e-01 1.12933242e+00 -8.70693266e-01
-7.23961473e-01 -3.65406483e-01 1.29563110e-02 3.91407222e-01
-1.88102067e-01 4.81767714e-01 -1.14737463e+00 -2.40969554e-01
-1.20576330e-01 -5.33788681e-01 1.05507100e+00 6.25257015e-01
1.37285244e+00 -5.92076838e-01 -5.46597421e-01 5.20100594e-01
1.39830136e+00 -1.20871119e-01 1.94051966e-01 -1.29410565e-01
3.10170650e-01 2.30940476e-01 3.66626054e-01 4.77014244e-01
3.10824066e-01 5.19917786e-01 3.32670361e-01 2.65028715e-01
5.39102256e-01 -4.51691240e-01 4.89558458e-01 4.52257484e-01
5.58645725e-01 -4.84222829e-01 -7.17431605e-01 6.62524760e-01
-1.77888834e+00 -8.26478481e-01 3.68400067e-01 2.12805581e+00
1.14223659e+00 3.87247086e-01 -6.69617057e-02 -4.50282305e-01
5.40595114e-01 4.53317352e-02 -1.00041246e+00 -8.29557955e-01
1.65247500e-01 2.87950695e-01 5.01142204e-01 8.84428263e-01
-9.47820246e-01 6.95476949e-01 7.62425184e+00 9.44731414e-01
-5.94107568e-01 1.14645019e-01 8.51455927e-01 -6.82492912e-01
-8.14576030e-01 1.59650892e-01 -1.31173921e+00 7.35310555e-01
1.59800184e+00 -2.65689373e-01 3.77279669e-01 1.24347878e+00
-2.36387998e-02 -1.66642949e-01 -1.54837596e+00 7.17708170e-01
-1.33085102e-01 -1.63824964e+00 1.65415913e-01 1.60102502e-01
7.67459571e-01 3.18487376e-01 2.57741392e-01 4.27625477e-01
1.09771848e+00 -1.09476805e+00 7.88481236e-01 5.06914437e-01
4.84249622e-01 -9.90986109e-01 6.85490668e-01 2.90635616e-01
-6.92126751e-01 5.01521043e-02 -7.63976097e-01 3.55615109e-01
4.41493392e-01 5.34535348e-01 -1.08044541e+00 -4.62726578e-02
6.52597129e-01 4.33032602e-01 -4.43554431e-01 5.85105658e-01
-3.69597301e-02 8.71614516e-01 -7.90514886e-01 -1.36649743e-01
2.73488462e-01 -1.32390261e-01 4.27006245e-01 1.14816678e+00
2.15840116e-01 -6.25154853e-01 -1.58640862e-01 1.52106881e+00
-1.40363231e-01 -4.76341099e-01 -4.91361409e-01 -8.56822655e-02
8.14376712e-01 9.27488446e-01 -3.07347208e-01 -6.29887581e-01
-2.29299188e-01 6.34653628e-01 3.57600719e-01 3.82226795e-01
-1.16512704e+00 -2.14179620e-01 9.06774938e-01 -1.73764065e-01
8.75558257e-01 3.46317962e-02 -4.36031818e-02 -1.10902929e+00
-1.35578290e-01 -7.84794688e-01 3.92283976e-01 -7.25051701e-01
-1.25747609e+00 2.68830836e-01 5.53246200e-01 -5.78894734e-01
-9.23496902e-01 -5.31707942e-01 -6.04476631e-01 1.09180868e+00
-1.39966500e+00 -8.18644166e-01 3.21216673e-01 2.23607361e-01
8.67396533e-01 2.20470950e-02 7.60149658e-01 -2.66143441e-01
-8.77088785e-01 7.96881616e-01 2.44737670e-01 -3.71938497e-01
3.43375951e-01 -1.43536818e+00 1.00556087e+00 7.75531888e-01
2.67948002e-01 9.74638879e-01 9.50757682e-01 -5.30036509e-01
-1.47538733e+00 -1.08510303e+00 7.81157434e-01 -7.39744663e-01
7.28113830e-01 -5.97404659e-01 -9.52532530e-01 1.14482296e+00
3.81348431e-01 -2.55135983e-01 5.58752537e-01 4.24141169e-01
-4.24496621e-01 2.49393214e-03 -1.03919864e+00 5.83595872e-01
3.71853590e-01 -5.92087388e-01 -5.27943015e-01 3.31339419e-01
9.75987136e-01 -3.33318383e-01 -1.14829469e+00 1.13371879e-01
7.41308749e-01 -5.96325517e-01 8.48458171e-01 -8.04607570e-01
4.10063446e-01 -1.47702739e-01 -2.49379560e-01 -1.12300587e+00
-1.16316684e-01 -1.02129674e+00 -8.45121920e-01 1.06216145e+00
1.03673387e+00 -6.62125647e-01 6.22978389e-01 1.09017527e+00
1.77117214e-01 -8.52028191e-01 -8.27550471e-01 -4.43311572e-01
2.56903976e-01 -7.43286490e-01 9.66637313e-01 2.07707211e-01
-5.41374087e-01 4.23666000e-01 -3.32720816e-01 2.07279250e-01
7.32024133e-01 1.26928061e-01 7.19479918e-01 -1.28656673e+00
-7.99549043e-01 -3.91283214e-01 1.61191881e-01 -1.19321632e+00
2.94824153e-01 -6.41291916e-01 1.83171034e-01 -1.08029962e+00
3.37188661e-01 -3.48514915e-02 -1.99936256e-01 3.25335741e-01
-2.66700327e-01 -1.19705901e-01 -2.99187541e-01 -1.00008585e-01
-5.72646797e-01 5.30751467e-01 7.19159663e-01 -2.13966697e-01
5.43717667e-02 1.39780134e-01 -7.41132855e-01 7.92514622e-01
6.17123842e-01 -7.95097291e-01 -7.63491392e-01 -6.27667964e-01
6.22598708e-01 -1.26823097e-01 5.09396493e-01 -4.66365993e-01
2.58324862e-01 -9.97269750e-02 6.06629968e-01 -8.39338779e-01
6.52216434e-01 -5.05644321e-01 3.23549241e-01 2.96824038e-01
-7.67068028e-01 2.38743141e-01 4.12830442e-01 9.04082000e-01
-6.06137179e-02 -4.61362511e-01 7.19387591e-01 -2.59503514e-01
-3.76720101e-01 3.55241418e-01 -3.78479272e-01 1.66038916e-01
6.19160771e-01 -4.18565646e-02 -1.86111525e-01 -3.66813689e-01
-5.79274476e-01 3.82063687e-01 3.85801464e-01 2.64854759e-01
3.84790182e-01 -9.06549156e-01 -6.64457679e-01 3.94293815e-01
-4.20094505e-02 3.83074641e-01 6.35099038e-02 4.92405891e-01
-2.42941618e-01 4.83240455e-01 4.77881819e-01 -8.29904974e-01
-8.70755374e-01 3.85699570e-01 4.67094630e-01 -2.80009329e-01
-3.48126888e-01 1.13494551e+00 1.22047096e-01 -3.21986645e-01
3.22455049e-01 -5.23036540e-01 4.36793983e-01 8.98781291e-04
5.93460500e-01 2.86340714e-01 -1.00903839e-01 1.12549722e-01
-1.85963988e-01 6.85365275e-02 -5.68970740e-01 -5.02081394e-01
1.12006462e+00 -1.02873549e-01 1.87191084e-01 7.86160290e-01
1.33646214e+00 -4.27845478e-01 -1.91767621e+00 -3.01958434e-02
-2.17698943e-02 -2.06827670e-01 3.88346970e-01 -6.60824835e-01
-6.51287436e-01 9.04658079e-01 2.77597874e-01 1.81368455e-01
4.03909773e-01 -7.70275714e-04 6.20353222e-01 6.60629809e-01
1.12497471e-01 -1.24916530e+00 -1.64391667e-01 1.87610984e-01
6.07427597e-01 -1.40829098e+00 1.00497819e-01 1.67682707e-01
-6.45639837e-01 7.91487217e-01 5.35137951e-01 -1.30798295e-01
9.31731761e-01 5.54266572e-01 -4.40221786e-01 -2.41338536e-01
-1.73661208e+00 3.67978036e-01 4.39227462e-01 1.78468958e-01
1.74596325e-01 -1.08730182e-01 3.70378196e-01 4.20521945e-01
-3.31049234e-01 -7.19482154e-02 3.24872047e-01 5.58015466e-01
-3.64335567e-01 -8.98477256e-01 2.20761567e-01 7.77971625e-01
-6.16988003e-01 -2.94451237e-01 5.31179272e-02 9.86337245e-01
-3.22613150e-01 7.64981091e-01 5.85041046e-01 7.20137060e-02
-2.17700720e-01 4.66796577e-01 2.63020903e-01 -5.07187903e-01
-3.46005917e-01 3.25816609e-02 2.58371353e-01 -6.01645112e-01
1.38529360e-01 -7.69854903e-01 -7.97410905e-01 -4.24104840e-01
-7.57786334e-01 2.06967071e-01 8.65886390e-01 9.32480037e-01
6.16742313e-01 4.02308911e-01 4.24392492e-01 -5.66736281e-01
-1.30053198e+00 -1.07227683e+00 -3.57183337e-01 -6.20606206e-02
4.01637018e-01 -6.32479370e-01 -4.72332507e-01 -2.70640831e-02] | [7.006649971008301, 3.9927642345428467] |
c5ea7765-84b9-4c80-9d5e-9b0f500b2839 | attentional-graph-convolutional-network-for | 2301.00145 | null | https://arxiv.org/abs/2301.00145v1 | https://arxiv.org/pdf/2301.00145v1.pdf | Attentional Graph Convolutional Network for Structure-aware Audio-Visual Scene Classification | Audio-Visual scene understanding is a challenging problem due to the unstructured spatial-temporal relations that exist in the audio signals and spatial layouts of different objects and various texture patterns in the visual images. Recently, many studies have focused on abstracting features from convolutional neural networks while the learning of explicit semantically relevant frames of sound signals and visual images has been overlooked. To this end, we present an end-to-end framework, namely attentional graph convolutional network (AGCN), for structure-aware audio-visual scene representation. First, the spectrogram of sound and input image is processed by a backbone network for feature extraction. Then, to build multi-scale hierarchical information of input features, we utilize an attention fusion mechanism to aggregate features from multiple layers of the backbone network. Notably, to well represent the salient regions and contextual information of audio-visual inputs, the salient acoustic graph (SAG) and contextual acoustic graph (CAG), salient visual graph (SVG), and contextual visual graph (CVG) are constructed for the audio-visual scene representation. Finally, the constructed graphs pass through a graph convolutional network for structure-aware audio-visual scene recognition. Extensive experimental results on the audio, visual and audio-visual scene recognition datasets show that promising results have been achieved by the AGCN methods. Visualizing graphs on the spectrograms and images have been presented to show the effectiveness of proposed CAG/SAG and CVG/SVG that could focus on the salient and semantic relevant regions. | ['Yangsheng Xu', 'Tin Lun Lam', 'Junjie Hu', 'Xiaonan Qi', 'Yuhongze Zhou', 'Liguang Zhou'] | 2022-12-31 | null | null | null | null | ['scene-classification', 'scene-recognition'] | ['computer-vision', 'computer-vision'] | [ 3.21293712e-01 -3.11543286e-01 4.29516971e-01 -3.06197912e-01
-5.89734554e-01 -2.95025826e-01 1.89999357e-01 3.32559764e-01
1.22101262e-01 1.97436646e-01 4.61542875e-01 7.92051926e-02
-2.41563335e-01 -4.61884052e-01 -7.34493136e-01 -7.21157134e-01
-2.83525676e-01 -3.72241288e-01 4.61961508e-01 6.64890632e-02
3.87644887e-01 3.87444139e-01 -2.05154490e+00 7.11564004e-01
2.05329761e-01 1.31035244e+00 4.95114565e-01 8.11970532e-01
-3.00115258e-01 8.24063897e-01 -5.04468501e-01 3.47187996e-01
-2.45273829e-01 -4.59043682e-01 -6.38166666e-01 4.50218439e-01
5.11022806e-01 1.41438425e-01 -2.74282932e-01 1.15468395e+00
4.48403507e-01 5.16928136e-01 4.49066460e-01 -1.59037864e+00
-8.51738095e-01 5.01725197e-01 -6.47673726e-01 4.62975651e-01
4.59667712e-01 7.76644573e-02 1.21828210e+00 -1.13251722e+00
3.22349131e-01 1.56978595e+00 7.58641660e-02 1.42835855e-01
-9.30792809e-01 -6.00204885e-01 5.86349249e-01 7.56891012e-01
-1.56696081e+00 -2.36411333e-01 1.38822055e+00 -7.29520977e-01
8.65585983e-01 4.77276474e-01 1.08566046e+00 6.54456973e-01
6.88491240e-02 6.88922882e-01 6.80172563e-01 -4.48001504e-01
1.80564135e-01 -1.53683886e-01 2.18494222e-01 6.97426498e-01
-1.70048952e-01 -1.11868478e-01 -8.55077922e-01 -6.72432687e-03
6.70391977e-01 7.88626075e-02 -4.28084463e-01 -4.59431589e-01
-8.74336720e-01 4.78802919e-01 8.63362968e-01 3.23320359e-01
-4.31426555e-01 3.84586036e-01 5.54601312e-01 -6.04146607e-02
3.57336968e-01 -5.03532439e-02 1.68071628e-01 2.62393594e-01
-6.25652373e-01 -8.53047986e-03 5.73660508e-02 8.30484748e-01
8.61242533e-01 4.36722875e-01 -3.50772232e-01 8.66330385e-01
5.35660625e-01 2.99377143e-01 2.11327404e-01 -6.90292835e-01
5.62976599e-01 8.26184690e-01 -4.13860083e-01 -1.62108529e+00
-3.51741105e-01 -2.77989835e-01 -8.27936828e-01 -3.74843739e-02
-1.11473352e-01 3.34800482e-01 -9.40622509e-01 1.50345027e+00
4.40388113e-01 4.74404216e-01 -2.28328720e-01 1.26616263e+00
1.33720529e+00 8.92795205e-01 2.94542819e-01 -5.94051667e-02
1.50318480e+00 -7.29467928e-01 -7.12816715e-01 -2.53251404e-01
7.49152526e-02 -6.66962326e-01 1.31632936e+00 7.66409338e-02
-9.49611902e-01 -1.07687664e+00 -9.43692744e-01 -1.52993619e-01
-3.62133950e-01 -9.86111350e-03 2.28200495e-01 -2.24506427e-02
-1.05923569e+00 3.94477509e-02 -3.53901684e-01 -2.91696429e-01
4.62799668e-01 -4.43071239e-02 -3.46009105e-01 -8.13460052e-02
-9.79572356e-01 -3.71629782e-02 6.53277338e-01 4.69720274e-01
-1.36580765e+00 -6.61752522e-01 -1.12340784e+00 4.27665442e-01
4.24089134e-01 -4.88725483e-01 8.48438680e-01 -1.10726368e+00
-1.01598823e+00 4.77345347e-01 -1.50343537e-01 -4.87422524e-03
-1.92859098e-01 6.20869175e-02 -3.95295054e-01 5.18647850e-01
-1.77867785e-02 5.29983699e-01 1.13019860e+00 -1.56032646e+00
-6.95561826e-01 -3.48414719e-01 1.11317895e-01 4.26243573e-01
-2.48860300e-01 2.18409434e-01 -5.89619040e-01 -7.64537215e-01
3.63964409e-01 -4.74149674e-01 7.37446472e-02 -1.15337498e-01
-4.96130198e-01 -2.10059568e-01 1.26571822e+00 -6.77722216e-01
1.39271629e+00 -2.62700176e+00 3.01543415e-01 2.35610157e-01
2.74738312e-01 8.93230364e-02 -3.48771244e-01 3.54435295e-01
-2.89499015e-01 5.87380864e-02 -9.13527980e-03 4.03975621e-02
-2.29634255e-01 1.97812226e-02 -4.47829425e-01 1.52907535e-01
1.95398197e-01 9.13635671e-01 -9.48271811e-01 -6.09743059e-01
4.94710416e-01 6.72428668e-01 -7.22067177e-01 3.38132173e-01
-2.46700168e-01 4.97188479e-01 -5.24990857e-01 7.48268783e-01
4.06728446e-01 -2.20164657e-01 2.84085446e-03 -4.85252410e-01
1.20510291e-02 -1.96289886e-02 -1.15461564e+00 1.56860995e+00
-3.01486284e-01 7.28440166e-01 2.33246282e-01 -1.05296254e+00
9.81057823e-01 2.28577122e-01 3.39098811e-01 -8.95136416e-01
2.43953783e-02 -2.03095049e-01 -1.76340699e-01 -8.24233055e-01
3.80016923e-01 6.47448301e-02 -1.38958871e-01 3.34425904e-02
1.50823399e-01 -1.09247684e-01 -1.47095487e-01 3.56827885e-01
4.88503784e-01 -1.57214418e-01 6.40778914e-02 -8.08460936e-02
9.36684132e-01 -3.84842694e-01 4.08720255e-01 3.97342294e-01
-1.92939892e-01 8.36535394e-01 4.54854876e-01 -4.65911955e-01
-8.05041432e-01 -1.05835712e+00 3.79947335e-01 1.40425861e+00
3.32769901e-01 -6.58442378e-01 -5.75629354e-01 -2.70961761e-01
-1.22689180e-01 2.91277647e-01 -7.14114070e-01 -1.82400361e-01
-4.15601015e-01 -7.96611309e-02 2.32053638e-01 6.73018873e-01
4.15631264e-01 -1.39240289e+00 -4.78793234e-01 1.95489794e-01
-2.58653194e-01 -1.13041949e+00 -4.75193471e-01 -1.88230798e-01
-3.37884933e-01 -1.12505543e+00 -3.88347566e-01 -9.47352529e-01
4.83614385e-01 6.61256075e-01 7.57143319e-01 1.21923469e-01
-6.35894597e-01 6.73069537e-01 -4.82122540e-01 -4.39685196e-01
2.10981563e-01 -6.06657624e-01 -1.99500784e-01 7.39783406e-01
-9.70006287e-02 -8.43016446e-01 -6.82508469e-01 1.08975634e-01
-7.90576339e-01 2.61975944e-01 1.36581928e-01 6.47240818e-01
8.82157564e-01 -1.73219256e-02 3.68442357e-01 -4.07190472e-01
5.87686121e-01 -4.29965913e-01 -4.08876836e-01 1.94966242e-01
2.17246786e-01 -3.72429669e-01 6.40484154e-01 -3.80645961e-01
-8.08086097e-01 1.57872811e-01 2.08257154e-01 -1.03132117e+00
-3.57243776e-01 6.59314752e-01 -4.58834827e-01 1.48239896e-01
4.10338640e-01 4.31500465e-01 -3.05307209e-01 -4.15134072e-01
5.12134612e-01 6.15293324e-01 5.91173470e-01 -3.54927748e-01
5.09021819e-01 4.46674883e-01 4.17004600e-02 -1.25941193e+00
-8.02055717e-01 -5.51014543e-01 -5.82281053e-01 -7.40936100e-01
1.15115559e+00 -9.90144134e-01 -5.47569692e-01 2.86980003e-01
-1.08735836e+00 -2.88691502e-02 -1.53960392e-01 4.11124408e-01
-4.38937873e-01 5.95925093e-01 -2.32333720e-01 -1.02182519e+00
-2.01859102e-01 -1.16050279e+00 1.44333386e+00 3.47208112e-01
-3.67082190e-03 -8.91007543e-01 -3.24405670e-01 1.97865382e-01
1.26899764e-01 4.15047526e-01 1.26190746e+00 -2.68913358e-01
-7.28290141e-01 1.78736269e-01 -4.93259609e-01 2.28522643e-01
2.03027010e-01 1.47187948e-01 -1.26293278e+00 -2.39244759e-01
-3.38590801e-01 -2.35429123e-01 7.73243725e-01 6.26766384e-01
1.60703599e+00 -3.32530856e-01 -1.66855156e-01 5.31244814e-01
1.28326011e+00 5.16612232e-01 2.06115797e-01 -1.11965165e-01
1.25415957e+00 7.96421230e-01 6.43353045e-01 4.86907184e-01
2.91351169e-01 7.84959793e-01 8.11600029e-01 -2.89312631e-01
-4.52225178e-01 -4.57876474e-01 3.41297299e-01 1.11957598e+00
3.86534669e-02 2.06885696e-03 -8.17915201e-01 7.33170390e-01
-1.94508946e+00 -9.26812649e-01 -1.76884785e-01 1.74174225e+00
1.71941817e-01 -1.00883901e-01 4.76837717e-03 3.97554368e-01
1.00302005e+00 5.21495402e-01 -3.49321812e-01 -3.49220812e-01
-1.48015410e-01 1.13758128e-02 -4.53694284e-01 3.53226095e-01
-1.12873363e+00 8.80293608e-01 5.23160172e+00 7.88062036e-01
-1.22424030e+00 -6.26871511e-02 4.06060249e-01 -2.01557294e-01
-3.15641969e-01 1.65682703e-01 -2.35532671e-01 2.09110558e-01
5.00957906e-01 -1.65196538e-01 4.00473297e-01 7.92449772e-01
3.61018896e-01 2.21162707e-01 -8.95124972e-01 1.37972414e+00
1.36211097e-01 -1.29917753e+00 6.91742659e-01 -2.77656764e-01
2.99470246e-01 -4.13526833e-01 1.62606776e-01 1.02217562e-01
1.89004634e-02 -8.72405291e-01 1.00766945e+00 5.72547495e-01
7.01561034e-01 -8.94853055e-01 3.23790997e-01 -3.40717728e-03
-2.11890268e+00 -2.36320421e-01 -4.54357654e-01 -7.55759999e-02
1.55487746e-01 2.04814285e-01 -6.84222877e-01 8.95998240e-01
1.08746004e+00 9.85072136e-01 -7.26389527e-01 1.01896453e+00
-1.98006317e-01 6.35184228e-01 -9.84655600e-03 -8.62791669e-03
4.93579954e-01 5.01644984e-02 6.66287124e-01 1.26852834e+00
2.52674788e-01 6.33456707e-02 3.48071337e-01 9.83610332e-01
2.50299096e-01 4.51714456e-01 -8.46319437e-01 -7.27611780e-02
4.55332696e-01 1.37217379e+00 -6.55917227e-01 -2.52042145e-01
-4.15745795e-01 4.22383875e-01 2.36957937e-01 6.65557325e-01
-7.74598897e-01 -6.04737163e-01 5.93735516e-01 9.24773738e-02
4.07734007e-01 -1.69183567e-01 1.25892162e-01 -7.76895165e-01
1.17259562e-01 -5.52693129e-01 7.22412169e-01 -1.32486665e+00
-9.54452872e-01 6.96997106e-01 6.20789938e-02 -1.36604786e+00
1.03878276e-02 -3.33626479e-01 -7.36758530e-01 8.55250895e-01
-1.34338748e+00 -1.23567104e+00 -5.91792166e-01 1.17313361e+00
7.30278194e-01 -1.70122147e-01 2.87544608e-01 1.94061533e-01
-3.61189723e-01 1.81743160e-01 -4.86146986e-01 1.72601506e-01
9.48375911e-02 -8.44255328e-01 6.21033348e-02 9.04705644e-01
6.32913351e-01 3.92386287e-01 2.77335376e-01 -5.66974699e-01
-1.44616055e+00 -1.48969042e+00 4.62196738e-01 4.14729901e-02
6.90130711e-01 -4.81609553e-01 -1.18087673e+00 5.47463655e-01
1.80177003e-01 3.24590862e-01 5.57721794e-01 -1.11096099e-01
-5.20634651e-01 -1.85621038e-01 -4.38776970e-01 5.38738489e-01
1.00288188e+00 -1.09921861e+00 -5.24259031e-01 1.22715920e-01
9.92307723e-01 -2.10559547e-01 -5.25285244e-01 3.37547928e-01
4.16147798e-01 -8.68212342e-01 1.19342816e+00 -6.65074170e-01
1.88711345e-01 -7.75702298e-01 -4.94654059e-01 -1.11644137e+00
-5.40118337e-01 -1.73514277e-01 1.64135367e-01 1.45874107e+00
-2.56465692e-02 -4.06063758e-02 1.72176749e-01 -7.01206103e-02
-3.50861043e-01 -3.77305508e-01 -1.07273996e+00 -3.04204732e-01
-5.89752018e-01 -8.17163825e-01 5.47943294e-01 8.52006555e-01
-2.62230467e-02 5.90830266e-01 -3.22974831e-01 5.15849173e-01
5.03148913e-01 5.15568078e-01 6.72978282e-01 -1.31536961e+00
-1.58124454e-02 -3.81175607e-01 -8.52331221e-01 -6.63194239e-01
1.04767047e-01 -1.05812001e+00 -5.73353544e-02 -1.64532042e+00
1.35221481e-01 5.09664156e-02 -6.75234973e-01 5.40367782e-01
-1.58188656e-01 3.28435540e-01 5.89840889e-01 -3.08282766e-02
-8.73950243e-01 7.56837487e-01 1.38265109e+00 -3.79161537e-01
-2.44368464e-01 -3.91902030e-01 -5.31351388e-01 7.08506465e-01
4.92451221e-01 -2.00527102e-01 -7.14380622e-01 -4.09703314e-01
-3.64412218e-02 2.66405791e-01 9.26454067e-01 -9.59177434e-01
3.11673552e-01 -1.48759902e-01 3.24356914e-01 -8.05895507e-01
4.92073953e-01 -8.01187873e-01 2.63430208e-01 1.11956231e-01
-3.47474933e-01 -2.46810317e-02 4.00869399e-01 8.56103420e-01
-8.05370331e-01 3.34968477e-01 6.63053870e-01 -1.26409307e-01
-1.06843698e+00 3.78143132e-01 -3.21618289e-01 -1.95515621e-02
8.94316316e-01 -4.25510556e-01 -1.42608240e-01 -4.73971546e-01
-1.12144887e+00 8.92775208e-02 -2.05562010e-01 8.11173975e-01
1.20505369e+00 -1.69688499e+00 -5.42114854e-01 3.81892353e-01
3.74380082e-01 2.77669597e-02 1.04830456e+00 6.40063822e-01
-3.33190918e-01 3.87041837e-01 -3.33267123e-01 -1.10426998e+00
-1.59102702e+00 7.26223946e-01 1.71819821e-01 3.20759863e-01
-8.72728109e-01 9.71197665e-01 1.00308597e+00 6.49399608e-02
4.46994096e-01 -6.70221567e-01 -6.03901148e-01 2.28716582e-01
5.67591846e-01 1.22551814e-01 -2.29367882e-01 -1.07312989e+00
-5.23413360e-01 9.66123044e-01 2.87798166e-01 5.15717976e-02
1.27345502e+00 -4.15821910e-01 -1.30051672e-01 6.76612437e-01
1.23835564e+00 -1.28935665e-01 -1.27402949e+00 -4.11250412e-01
-2.25782499e-01 -4.21351194e-01 1.17798056e-02 -2.48436615e-01
-1.41928804e+00 1.57326531e+00 5.64547539e-01 4.01569456e-01
1.36827123e+00 1.57551885e-01 1.84302315e-01 -6.61667064e-02
1.15486562e-01 -7.89665580e-01 5.52662253e-01 4.06701684e-01
1.32329047e+00 -6.31932378e-01 -1.88691750e-01 -4.52438623e-01
-8.35711062e-01 1.12449765e+00 6.32796407e-01 -5.99953905e-02
7.95790553e-01 -6.91642910e-02 8.76620561e-02 -5.00986040e-01
-7.91178584e-01 -4.74750608e-01 8.47959280e-01 6.46905720e-01
1.62624225e-01 3.94963883e-02 4.47498709e-01 7.90979445e-01
-1.72868162e-01 -4.83860850e-01 1.36118352e-01 8.37025404e-01
-6.08887851e-01 -4.41317439e-01 -4.25990313e-01 -8.08065087e-02
-2.21480012e-01 -1.85790986e-01 -6.50794327e-01 6.57900870e-01
3.38307098e-02 1.11642730e+00 2.44434118e-01 -7.37703979e-01
4.60410804e-01 5.54695018e-02 2.42654637e-01 -6.53152108e-01
-5.73219061e-01 5.94516158e-01 -2.77244359e-01 -7.01959431e-01
-4.34435368e-01 -2.68013984e-01 -1.49364269e+00 3.24924171e-01
7.50611210e-03 2.58203119e-01 3.32406491e-01 6.56369746e-01
4.72039312e-01 1.06217802e+00 6.06514990e-01 -1.08483398e+00
4.31975216e-01 -8.08297217e-01 -7.82334566e-01 6.02212906e-01
6.08088791e-01 -8.68604720e-01 -2.91361600e-01 2.17436522e-01] | [14.881149291992188, 4.822327613830566] |
55828482-29ca-41cb-becd-08ee10a91712 | tabgenie-a-toolkit-for-table-to-text | 2302.14169 | null | https://arxiv.org/abs/2302.14169v1 | https://arxiv.org/pdf/2302.14169v1.pdf | TabGenie: A Toolkit for Table-to-Text Generation | Heterogenity of data-to-text generation datasets limits the research on data-to-text generation systems. We present TabGenie - a toolkit which enables researchers to explore, preprocess, and analyze a variety of data-to-text generation datasets through the unified framework of table-to-text generation. In TabGenie, all the inputs are represented as tables with associated metadata. The tables can be explored through the web interface, which also provides an interactive mode for debugging table-to-text generation, facilitates side-by-side comparison of generated system outputs, and allows easy exports for manual analysis. Furthermore, TabGenie is equipped with command line processing tools and Python bindings for unified dataset loading and processing. We release TabGenie as a PyPI package and provide its open-source code and a live demo at https://github.com/kasnerz/tabgenie. | ['Ondřej Dušek', 'Ondřej Plátek', 'Ekaterina Garanina', 'Zdeněk Kasner'] | 2023-02-27 | null | null | null | null | ['table-to-text-generation', 'data-to-text-generation'] | ['natural-language-processing', 'natural-language-processing'] | [-2.16650143e-01 9.97086763e-02 2.34655831e-02 -4.46081132e-01
-1.01632965e+00 -1.06532204e+00 8.53824139e-01 4.46901351e-01
3.43803287e-01 7.20772266e-01 5.12384892e-01 -5.71098924e-01
1.40012115e-01 -1.30168283e+00 -3.98645371e-01 -3.16067398e-01
1.83031037e-01 7.92600632e-01 -8.68670922e-03 -2.47662827e-01
9.47592854e-02 7.73452362e-03 -1.82242334e+00 7.24236786e-01
9.91674244e-01 5.27869999e-01 2.80297548e-01 7.95544207e-01
-5.90082943e-01 2.96050221e-01 -7.18240678e-01 -4.48465675e-01
3.57967615e-01 -5.28922260e-01 -6.23099864e-01 -3.62964481e-01
9.85640883e-02 -1.15612872e-01 9.99822989e-02 6.97906673e-01
7.35227883e-01 -1.89956412e-01 2.50483632e-01 -1.59086931e+00
-4.80260879e-01 9.86133575e-01 -5.39313704e-02 -2.46046960e-01
6.79017842e-01 5.12200177e-01 8.08442116e-01 -1.07285273e+00
1.10646093e+00 1.21462572e+00 3.63643438e-01 3.05763721e-01
-1.22681725e+00 -7.93308020e-01 -5.22581995e-01 -4.26594108e-01
-1.22262561e+00 -6.20904088e-01 1.80859968e-01 -7.41776526e-01
7.91471303e-01 8.28798473e-01 6.45576179e-01 1.06042254e+00
1.58367366e-01 6.29332721e-01 6.84000969e-01 -2.28503153e-01
5.38060330e-02 1.06070004e-01 -4.83313715e-03 4.63566780e-01
4.76763844e-01 -2.36189380e-01 -8.08597565e-01 -3.76356035e-01
7.25856602e-01 -2.41022393e-01 1.23931438e-01 2.79824063e-02
-1.56648886e+00 6.20419323e-01 2.42216941e-02 2.71964278e-02
-1.60212263e-01 7.72512809e-04 5.18346786e-01 2.00091600e-01
5.18124759e-01 2.64921516e-01 -4.03975129e-01 -6.62063956e-01
-6.91432893e-01 8.18852127e-01 9.52521503e-01 1.56914818e+00
7.48463452e-01 1.83014013e-02 -5.24935186e-01 6.76991880e-01
8.16921424e-03 6.42996490e-01 3.51965785e-01 -6.46239340e-01
9.82893407e-01 8.52012336e-01 1.49412617e-01 -6.80247128e-01
-4.45100635e-01 5.73704094e-02 -7.73003876e-01 7.51021057e-02
3.44364733e-01 -5.58950961e-01 -6.77743018e-01 1.07184899e+00
6.55048430e-01 -9.74472344e-01 -9.25686583e-03 5.70651829e-01
1.37571418e+00 6.69188797e-01 -1.44620672e-01 1.69043347e-01
1.40823555e+00 -4.58037257e-01 -9.14299309e-01 1.08038366e-01
9.95940924e-01 -1.07347620e+00 1.39934349e+00 1.06554709e-01
-1.16448462e+00 -4.01980400e-01 -5.94054461e-01 -4.76282954e-01
-1.12586153e+00 1.59467861e-01 5.51908016e-01 4.72321033e-01
-9.94399607e-01 5.65140963e-01 -9.23689544e-01 -7.49869525e-01
4.97990280e-01 -1.61512122e-01 -4.39037442e-01 4.65975076e-01
-1.00552452e+00 4.32034373e-01 8.30764353e-01 -4.34961945e-01
-3.08069944e-01 -1.14207268e+00 -8.96640837e-01 -9.58021954e-02
3.80248040e-01 -1.09475744e+00 1.29894161e+00 -4.79385816e-02
-1.18886113e+00 7.52491415e-01 -2.03612760e-01 1.44965902e-01
6.52950943e-01 -1.09810561e-01 -3.24941367e-01 -4.41821218e-01
4.23116595e-01 5.86213052e-01 1.04203679e-01 -1.13983905e+00
-5.93979776e-01 -4.21289593e-01 -3.84419352e-01 1.96462318e-01
2.69569997e-02 2.73016065e-01 -7.08252609e-01 -8.84835541e-01
-2.35317528e-01 -4.16718245e-01 -8.60732794e-02 -1.22146510e-01
-9.41570938e-01 -2.51526479e-02 7.25320101e-01 -7.47889519e-01
1.78711700e+00 -1.87846482e+00 -3.56132388e-01 1.13716841e-01
7.65459463e-02 -7.63370991e-02 -6.45572646e-03 1.31230009e+00
-2.47788742e-01 7.34914124e-01 -1.34183288e-01 -4.40371156e-01
3.51542383e-01 -8.08984116e-02 -2.91860163e-01 -2.86778301e-01
-5.84658468e-03 9.83120143e-01 -8.15121353e-01 -6.16024971e-01
2.76533425e-01 1.07126199e-01 -5.22899449e-01 2.98241854e-01
-5.22672892e-01 3.74663860e-01 -3.92227262e-01 8.78154576e-01
8.50720406e-01 -2.06484392e-01 3.20241660e-01 -1.39055029e-02
-7.05101013e-01 6.58812761e-01 -1.40199411e+00 1.81447256e+00
-1.31109789e-01 6.13572478e-01 -4.72855531e-02 4.59357351e-03
1.03712368e+00 2.66062975e-01 2.46031880e-01 -3.78856570e-01
-4.41459268e-02 9.36663449e-02 -5.98688066e-01 -4.31721598e-01
1.08651161e+00 5.39129794e-01 -4.68245745e-01 1.01959121e+00
-9.68628228e-02 -4.46618676e-01 9.86996233e-01 6.03963017e-01
7.58756340e-01 4.30378348e-01 4.57678080e-01 -3.14875036e-01
-9.24493894e-02 3.62693518e-01 1.51391461e-01 7.34791934e-01
8.87890875e-01 7.23922014e-01 6.95455015e-01 -3.64481837e-01
-1.33544004e+00 -1.28314412e+00 -4.61166471e-01 1.14202869e+00
-5.60034990e-01 -1.26023436e+00 -7.92620599e-01 -2.52151221e-01
2.33581677e-01 1.27236259e+00 -5.83176315e-01 5.54735780e-01
-1.86640921e-03 -8.33810031e-01 5.95071733e-01 3.59202236e-01
2.46730268e-01 -1.26171303e+00 -5.76957226e-01 1.24351598e-01
-3.66265535e-01 -5.13287067e-01 -4.63107467e-01 2.58931573e-02
-7.66707242e-01 -8.18278313e-01 -4.03227471e-02 -1.78149059e-01
6.52714431e-01 -1.72196310e-02 1.44892597e+00 9.08154473e-02
-5.32240093e-01 4.29971106e-02 -6.01617873e-01 -8.54353666e-01
-7.90348232e-01 5.93492925e-01 -2.73782641e-01 -6.18213475e-01
1.46398619e-02 -5.89124620e-01 -2.73389906e-01 2.64005303e-01
-1.39971340e+00 8.62554848e-01 -8.47471878e-02 3.77987236e-01
5.53397655e-01 -1.48146346e-01 4.01137382e-01 -1.21658170e+00
9.01684463e-01 -6.70088291e-01 -1.01013553e+00 2.94245392e-01
-6.66127563e-01 1.57647833e-01 6.66707814e-01 4.52644467e-01
-9.81394470e-01 -2.02589035e-02 -3.03972065e-01 1.33567497e-01
-1.75459340e-01 8.02296162e-01 -4.79208142e-01 1.03057873e+00
6.86442196e-01 3.27242255e-01 6.45566136e-02 -8.99648130e-01
8.64403725e-01 9.51945126e-01 5.69077551e-01 -5.46613574e-01
1.02612436e+00 2.05704704e-01 -4.73783642e-01 -3.27492446e-01
-3.90810817e-01 -1.19021080e-01 -7.93099642e-01 -1.79016277e-01
6.57731175e-01 -7.58956611e-01 -5.55765450e-01 5.61475217e-01
-9.87303078e-01 -7.29822457e-01 -3.16062689e-01 -3.60587358e-01
-3.84317249e-01 -6.64783940e-02 -1.84751198e-01 -5.02475321e-01
-8.38492692e-01 -7.94161141e-01 1.10422289e+00 1.69349805e-01
-5.66274643e-01 -1.06578720e+00 6.23482503e-02 1.88381791e-01
4.14839834e-01 5.97469091e-01 8.81209075e-01 -6.50609434e-01
-6.36345327e-01 -2.38906860e-01 -2.10669950e-01 -4.63096946e-01
3.61949444e-01 7.79970884e-01 -8.15964222e-01 -1.51701244e-02
-8.60384524e-01 -3.72312479e-02 2.14402854e-01 7.65376985e-02
1.36787260e+00 -6.42574489e-01 -5.56307733e-01 9.45988178e-01
1.27786291e+00 1.32253096e-01 9.15629804e-01 4.80995417e-01
7.09314823e-01 5.70085466e-01 5.93989074e-01 1.14828503e+00
7.48638153e-01 6.44549191e-01 2.47363910e-01 -3.01950812e-01
4.35741358e-02 -5.63030660e-01 5.10421246e-02 7.28616059e-01
2.49206960e-01 -7.44201481e-01 -1.00517976e+00 3.36169958e-01
-1.84488142e+00 -1.14490819e+00 -8.42594922e-01 2.30993390e+00
1.11953974e+00 -2.01363221e-01 3.04853648e-01 -1.42008856e-01
5.76492071e-01 8.41786861e-02 -2.80545294e-01 -2.98878163e-01
-1.45334542e-01 3.44939493e-02 4.01958108e-01 4.17120218e-01
-8.96152914e-01 1.00993240e+00 6.56972456e+00 8.32609475e-01
-7.52166092e-01 -6.30074739e-02 3.85804206e-01 -1.87255785e-01
-6.70875967e-01 1.43139422e-01 -9.93074834e-01 6.45175278e-01
1.01118231e+00 -1.10681629e+00 3.95606726e-01 8.62968445e-01
6.95848167e-01 -2.79329628e-01 -1.13775134e+00 7.54748106e-01
-4.53114897e-01 -1.78546154e+00 4.88316983e-01 1.00212663e-01
4.45185661e-01 -8.10451526e-03 -2.31960461e-01 -5.46344630e-02
8.76560688e-01 -8.28326702e-01 9.76045430e-01 7.00323105e-01
1.31280303e+00 -6.29763544e-01 4.07740980e-01 3.71326245e-02
-1.23698521e+00 4.52555090e-01 -2.02725098e-01 -1.74341109e-02
3.34613055e-01 9.19215381e-01 -1.08653975e+00 1.22407401e+00
8.75308514e-01 7.47779667e-01 -1.00466692e+00 6.80049419e-01
1.13252439e-01 3.22263986e-01 -3.15924197e-01 1.43469468e-01
-5.76992214e-01 -2.98737049e-01 4.93109673e-01 1.42289007e+00
7.64788806e-01 -1.26303032e-01 8.74135494e-02 1.08531368e+00
6.10023625e-02 2.86486745e-01 -7.27194130e-01 -2.80341774e-01
8.38130534e-01 1.50576746e+00 -7.53962219e-01 -6.22941196e-01
1.10002875e-01 5.85514545e-01 -6.05699606e-02 7.48998225e-02
-6.09489322e-01 -1.06556380e+00 7.05516756e-01 5.06557524e-01
-7.70010054e-04 -1.75549626e-01 -9.48935986e-01 -1.15902197e+00
2.07790043e-02 -1.09900784e+00 5.71708798e-01 -1.39785194e+00
-1.00354755e+00 5.84150612e-01 4.51883823e-01 -1.23963785e+00
-6.24656141e-01 -1.88183352e-01 -7.26303518e-01 1.06579721e+00
-4.77781922e-01 -1.05588889e+00 -7.99618542e-01 4.44033146e-01
3.44996005e-01 -2.83862446e-02 9.88557518e-01 2.01788709e-01
-8.11760604e-01 5.70313394e-01 4.25481677e-01 3.11388254e-01
7.98374951e-01 -1.43281841e+00 1.27609813e+00 8.71724427e-01
-2.94893116e-01 8.72783184e-01 7.88844585e-01 -1.07008517e+00
-1.44283140e+00 -1.33932650e+00 1.02605832e+00 -9.61001933e-01
8.58214557e-01 -1.00644267e+00 -8.44783604e-01 9.26712334e-01
5.32215774e-01 -5.42606533e-01 7.27450550e-01 1.06480278e-01
-3.79110128e-02 -2.59246081e-01 -8.11601818e-01 8.02473366e-01
1.01849365e+00 -1.44051671e-01 -7.87309110e-02 5.74958384e-01
5.92031658e-01 -8.67795944e-01 -1.15893006e+00 -2.36171469e-01
6.07237637e-01 -9.79403973e-01 4.42366511e-01 -9.49803814e-02
6.95056140e-01 -5.57383776e-01 1.12600863e-01 -1.19733429e+00
-1.04303062e-02 -9.64672267e-01 4.07220632e-01 1.87410593e+00
8.45100224e-01 -7.45868683e-01 2.36774519e-01 5.33540547e-01
-3.03241134e-01 -2.51602322e-01 -4.05449748e-01 -5.36836386e-01
-1.27006114e-01 -6.47552669e-01 1.33958590e+00 8.95523369e-01
3.76000434e-01 1.30418196e-01 -1.32392809e-01 -2.04950213e-01
3.40195268e-01 1.16734482e-01 1.65902340e+00 -1.15516174e+00
3.64057682e-02 -4.26309019e-01 2.96438187e-01 -5.50707281e-01
-7.73456156e-01 -1.43715990e+00 -1.11394316e-01 -2.13509130e+00
1.88996837e-01 -4.47823048e-01 6.74411297e-01 1.09561002e+00
-1.86285079e-01 1.13237777e-03 4.38131481e-01 4.43932533e-01
-2.10667714e-01 2.39918873e-01 1.09381998e+00 2.99156159e-01
-4.21552807e-01 -2.77289510e-01 -8.54731381e-01 -7.70842098e-03
9.72227037e-01 -8.90594482e-01 -2.57848471e-01 -4.50753540e-01
4.47793126e-01 1.14524186e-01 2.36492202e-01 -1.00453413e+00
-5.75567596e-02 -3.30334514e-01 4.28272843e-01 -1.27033627e+00
-1.40733972e-01 -2.13319138e-01 1.10637629e+00 1.78913191e-01
-2.60459274e-01 6.39069974e-01 4.85734195e-01 -1.63270712e-01
1.93250582e-01 -5.62787196e-03 2.82535046e-01 -2.10381508e-01
2.79435795e-02 2.07782939e-01 -6.12560451e-01 2.12715507e-01
8.06976616e-01 -2.38286495e-01 -9.74033058e-01 -3.38675827e-01
-5.12950480e-01 4.86856371e-01 9.71154690e-01 6.43241465e-01
8.05870295e-02 -1.23758519e+00 -8.23893547e-01 4.94793922e-01
4.32132810e-01 1.56928375e-01 8.20514038e-02 3.90279442e-01
-9.57045019e-01 4.44521457e-01 -2.35754296e-01 -3.84525657e-01
-1.29219568e+00 2.91895539e-01 -2.32837666e-02 -3.73896211e-01
-6.80753052e-01 3.58709216e-01 -8.61951634e-02 -9.11053002e-01
-3.36799592e-01 -4.84702915e-01 2.24847957e-01 2.94036627e-01
7.64089406e-01 4.75803137e-01 2.05598697e-01 -2.26197049e-01
-3.02621782e-01 -1.52401989e-02 2.46920303e-01 -2.89512485e-01
1.38118029e+00 -1.06363900e-01 -3.04323792e-01 6.82093799e-01
6.47365868e-01 2.59349793e-01 -8.98651779e-01 2.66742021e-01
-8.66057500e-02 -5.01463830e-01 -5.06811798e-01 -1.16357672e+00
-8.41794312e-01 5.06166041e-01 1.17854394e-01 5.06466806e-01
9.43534136e-01 -1.05441734e-02 3.88341516e-01 -1.01579539e-03
1.06686734e-01 -9.16635096e-01 -4.65825051e-01 4.54904944e-01
1.26861095e+00 -1.05791283e+00 3.00490260e-01 -3.11838537e-01
-7.82671750e-01 1.23679340e+00 4.93169934e-01 6.38846636e-01
5.31606138e-01 7.66330540e-01 4.64492559e-01 -2.05314294e-01
-1.30787241e+00 -5.73322177e-02 -7.98573066e-03 8.59197140e-01
7.13521659e-01 1.41982570e-01 -3.26909453e-01 7.20183134e-01
-1.10565698e+00 2.80085087e-01 8.52765977e-01 8.94809783e-01
-3.87137942e-02 -1.59944594e+00 -8.25413525e-01 8.42937648e-01
-3.26987922e-01 -4.25182432e-01 -3.36438656e-01 9.10975158e-01
-7.36446306e-02 9.17743981e-01 3.96356106e-01 -3.35967690e-01
2.84765512e-01 2.05844000e-01 -2.62805253e-01 -8.22506726e-01
-9.83478010e-01 1.29929036e-01 4.67500061e-01 -5.88284910e-01
3.03430766e-01 -8.46778691e-01 -1.35516202e+00 -8.38808119e-01
2.29439542e-01 1.88861191e-01 8.45811844e-01 2.76113421e-01
1.01214159e+00 4.21362549e-01 1.68029770e-01 -8.54858577e-01
1.87416673e-01 -1.20597374e+00 -5.32305658e-01 3.33175778e-01
-2.17534497e-01 -1.34160608e-01 -1.33825049e-01 5.61388791e-01] | [11.428180694580078, 8.80521297454834] |
c57eed16-0107-43f4-8c09-1067ac8c5d09 | inferno-inferring-object-centric-3d-scene | null | null | https://openreview.net/forum?id=YVa8X_2I1b | https://openreview.net/pdf?id=YVa8X_2I1b | INFERNO: Inferring Object-Centric 3D Scene Representations without Supervision | We propose INFERNO, a method to infer object-centric representations of visual scenes without relying on annotations. Our method learns to decompose a scene into multiple objects, each object having a structured representation that disentangles its shape, appearance and 3D pose. To impose this structure we rely on recent advances in neural 3D rendering. Each object representation defines a localized neural radiance field that is used to generate 2D views of the scene through a differentiable rendering process. Our model is subsequently trained by minimizing a reconstruction loss between inputs and corresponding rendered scenes. We empirically show that INFERNO discovers objects in a scene without supervision. We also validate the interpretability of the learned representations by manipulating inferred scenes and showing the corresponding effect in the rendered output. Finally, we demonstrate the usefulness of our 3D object representations in a visual reasoning task using the CATER dataset. | ['Aaron Courville', 'Nicolas Ballas', 'Lluis Castrejon'] | 2021-09-29 | null | null | null | null | ['video-object-tracking'] | ['computer-vision'] | [ 4.60795820e-01 5.29449821e-01 1.67468309e-01 -6.56961918e-01
-5.33419847e-01 -8.80155742e-01 8.46048772e-01 -6.30720928e-02
3.94695073e-01 1.92087561e-01 4.16320235e-01 -3.12640786e-01
8.14708415e-03 -8.39542329e-01 -1.27907145e+00 -4.77709174e-01
8.10884312e-02 4.59371477e-01 -2.02187374e-01 2.05526035e-02
2.38266513e-01 8.58271480e-01 -1.76457345e+00 8.50845397e-01
3.26649785e-01 8.35745513e-01 2.05727175e-01 9.43164945e-01
-1.21225808e-02 1.08517003e+00 -7.02334344e-01 -1.55902073e-01
5.75760663e-01 -4.04625207e-01 -7.15376556e-01 5.60647607e-01
9.67048168e-01 -6.49132967e-01 -2.09686980e-01 6.80748582e-01
-1.12117812e-01 1.99564338e-01 8.41711879e-01 -1.21102059e+00
-1.09331644e+00 2.49202579e-01 -4.34990674e-01 -6.32140860e-02
5.33050537e-01 3.50362360e-01 1.24636531e+00 -8.67343426e-01
7.56002367e-01 1.56506419e+00 1.74266189e-01 4.87043500e-01
-1.98226392e+00 -2.07133487e-01 3.44379276e-01 -3.17758799e-01
-9.87942338e-01 -4.80360895e-01 9.61299896e-01 -7.33650446e-01
9.27699149e-01 5.72075367e-01 7.69230008e-01 1.10100687e+00
-3.76574583e-02 6.42180502e-01 1.13468170e+00 -2.61963576e-01
3.79070640e-01 2.17828844e-02 -4.45158929e-02 1.08983707e+00
4.20436189e-02 1.42328039e-01 -7.10724950e-01 -1.18024409e-01
1.19950771e+00 -4.81608026e-02 -2.46831432e-01 -7.63398409e-01
-1.14288092e+00 6.19430900e-01 8.66522908e-01 -4.21403021e-01
-2.23270401e-01 6.55232608e-01 -1.37088532e-02 -1.36456722e-02
5.79758584e-01 7.19545662e-01 -4.99978900e-01 5.76306760e-01
-2.61604041e-01 4.35778648e-01 5.55415094e-01 9.89190698e-01
8.17976236e-01 1.55351371e-01 -1.92143604e-01 3.56250644e-01
4.67640132e-01 4.46184218e-01 -2.32555807e-01 -1.63342893e+00
3.38038623e-01 6.81197822e-01 1.96525916e-01 -1.03827238e+00
-1.46774784e-01 -2.06105322e-01 -4.11220759e-01 8.15168798e-01
1.79852039e-01 2.75451213e-01 -9.29709136e-01 1.78768134e+00
3.32563698e-01 3.03210109e-01 -7.10611269e-02 1.08609140e+00
8.34904432e-01 6.61278844e-01 1.18467502e-01 3.98316830e-01
1.21611643e+00 -8.07022035e-01 -1.35680318e-01 -1.43741339e-01
2.05428287e-01 -3.71692121e-01 1.23731101e+00 4.90083009e-01
-1.29394829e+00 -6.56456709e-01 -1.06008804e+00 -6.59937143e-01
-2.00643048e-01 9.54907760e-02 9.03121889e-01 2.94341803e-01
-1.10776579e+00 5.81377149e-01 -7.19058454e-01 1.91792160e-01
6.42718971e-01 1.68003917e-01 -3.70855153e-01 1.85598969e-01
-4.36906219e-01 6.92068636e-01 2.48799264e-01 -1.04178293e-02
-1.46648955e+00 -9.08178091e-01 -9.84783590e-01 1.96839288e-01
1.59403801e-01 -1.28174293e+00 1.23038936e+00 -1.10049319e+00
-1.29188073e+00 1.36132455e+00 -1.63777515e-01 -1.50661170e-01
3.58554095e-01 -1.35267645e-01 6.53360114e-02 3.52089226e-01
3.72406878e-02 8.92202675e-01 9.01741743e-01 -1.90600801e+00
-2.48156831e-01 -4.73697215e-01 7.41243362e-01 2.16960773e-01
2.83957571e-01 -4.41552997e-01 -1.72797203e-01 -5.45165300e-01
4.51848745e-01 -5.94117939e-01 -2.08490342e-01 5.54523170e-01
-7.22597778e-01 8.18809494e-02 5.88348627e-01 -4.86804903e-01
1.61478862e-01 -2.12706161e+00 5.09864450e-01 1.44270450e-01
5.08563817e-01 -4.64434117e-01 -2.98121065e-01 1.01383261e-01
-2.70854801e-01 3.61399531e-01 -1.35959834e-01 -6.03805125e-01
9.51935053e-02 2.06383929e-01 -8.57933700e-01 4.72684592e-01
4.50439543e-01 8.71509492e-01 -9.38591063e-01 -6.12533838e-02
4.72932428e-01 7.61951387e-01 -9.30892885e-01 6.04904771e-01
-9.24643576e-01 8.74820530e-01 -5.11529803e-01 3.57871681e-01
6.48519456e-01 -4.37203318e-01 2.13840857e-01 -4.05975461e-01
1.01968713e-01 6.07413292e-01 -6.52200341e-01 2.03287697e+00
-6.68427408e-01 7.82522261e-01 -1.18843012e-01 -8.72055471e-01
8.35235238e-01 9.62542072e-02 2.29450449e-01 -5.41681111e-01
-3.22944969e-02 -3.25684279e-01 -5.50174654e-01 -7.41996646e-01
3.44098300e-01 -8.69791433e-02 4.72782850e-02 6.31076574e-01
-6.27256278e-03 -5.37667215e-01 -4.26967174e-01 3.79845053e-01
8.44942212e-01 8.75833571e-01 -4.00001509e-03 -2.57087052e-01
-1.61586702e-01 -9.83854979e-02 -5.38439676e-02 6.27455592e-01
4.73241657e-01 8.40009868e-01 6.82263970e-01 -6.49513721e-01
-1.13429916e+00 -1.62666464e+00 5.92909344e-02 1.12719226e+00
2.54213125e-01 -2.89245516e-01 -4.20022964e-01 -6.64977193e-01
1.56446517e-01 1.13967109e+00 -1.05363381e+00 -1.30272612e-01
-3.77796292e-01 -9.05482993e-02 3.06342542e-01 4.28125441e-01
-4.09059525e-02 -1.05132401e+00 -1.09004080e+00 -3.87071520e-01
-6.56989440e-02 -1.09001601e+00 5.09029403e-02 1.60494253e-01
-9.13906097e-01 -1.23599851e+00 4.61091027e-02 -5.05988240e-01
1.20712328e+00 4.82554585e-01 1.52320135e+00 2.10674137e-01
-4.63728905e-01 5.64263463e-01 1.23980798e-01 -2.20783383e-01
-5.27451456e-01 -3.17598999e-01 -4.28544670e-01 3.23071592e-02
-1.88165173e-01 -8.81413043e-01 -5.87289929e-01 1.47689823e-02
-9.41979647e-01 8.34092140e-01 1.20240346e-01 3.08182955e-01
6.78072512e-01 -4.41095054e-01 5.33896312e-02 -1.13059568e+00
1.61166072e-01 -4.97508019e-01 -8.93635333e-01 1.73672333e-01
7.42305964e-02 5.00087917e-01 4.54492629e-01 -1.90481573e-01
-1.28594971e+00 2.61208832e-01 3.87016833e-01 -6.14540637e-01
-5.12401104e-01 5.41228764e-02 -2.32506841e-01 4.06324714e-01
8.94862235e-01 -1.41460478e-01 -3.71022761e-01 -5.34380496e-01
9.04054403e-01 -4.85420078e-02 7.20777094e-01 -9.71135080e-01
8.34559560e-01 8.95365596e-01 3.09809089e-01 -4.28605795e-01
-1.35188603e+00 7.16631114e-02 -7.38758385e-01 -1.04818158e-01
1.20898354e+00 -1.04402661e+00 -1.17475772e+00 -1.44699290e-01
-1.55390453e+00 -5.35853505e-01 -4.31779534e-01 3.82972695e-02
-8.84103298e-01 -2.17022404e-01 -3.10561687e-01 -8.42833400e-01
2.34174788e-01 -1.12375283e+00 1.61230254e+00 -1.11257955e-01
-3.45806688e-01 -8.98825169e-01 -1.59709960e-01 5.20535469e-01
-7.98448399e-02 7.81209886e-01 1.34553874e+00 -7.09161209e-03
-1.32032883e+00 2.97751129e-01 -4.99468833e-01 -4.61808220e-02
2.24979028e-01 5.86586669e-02 -1.50736463e+00 4.78347652e-02
-8.59481283e-04 -5.75252116e-01 8.49406898e-01 2.07172781e-01
1.92825305e+00 -3.89759958e-01 6.81726588e-03 1.03402376e+00
1.47712243e+00 -1.39591202e-01 3.90922517e-01 6.25651255e-02
1.06570256e+00 7.74978220e-01 2.17154156e-02 3.21661681e-01
2.36094773e-01 4.84571546e-01 9.50405419e-01 -1.96669534e-01
-3.44598621e-01 -7.41237342e-01 1.01074263e-01 1.21045120e-01
-1.79163411e-01 -2.16403693e-01 -5.79102635e-01 1.44800752e-01
-1.66902506e+00 -8.83087337e-01 -1.08454481e-01 1.98447227e+00
6.16395950e-01 2.35349927e-02 -1.95678622e-01 -2.06636369e-01
2.95033514e-01 3.15840632e-01 -8.74760211e-01 -4.85825181e-01
-3.73413749e-02 2.85476685e-01 1.69590414e-01 7.13838875e-01
-6.54195011e-01 7.56559968e-01 6.88592625e+00 -1.18251845e-01
-9.33160126e-01 -2.09179193e-01 8.70932639e-01 -2.85977155e-01
-1.19042802e+00 2.64833629e-01 -2.25318566e-01 -2.34332010e-01
3.89880806e-01 2.31448472e-01 8.17122698e-01 7.43471324e-01
2.22405568e-01 -1.19022327e-02 -1.91882288e+00 8.87287080e-01
2.57521510e-01 -1.69314206e+00 5.82139492e-01 1.41133592e-01
7.76962996e-01 -2.79935420e-01 2.30232418e-01 -3.95979509e-02
7.45737493e-01 -1.44822526e+00 1.17914450e+00 6.47140324e-01
8.91489983e-01 -3.80046815e-01 -4.55384880e-01 2.24376008e-01
-8.07632208e-01 1.29379094e-01 -2.81880379e-01 -2.82041997e-01
-4.68276143e-02 3.07169437e-01 -1.10247242e+00 3.29520077e-01
4.80224162e-01 6.08986616e-01 -4.59133655e-01 4.86007601e-01
-6.79045260e-01 3.47684741e-01 -1.96982529e-02 3.81388545e-01
-6.03558086e-02 -3.66603822e-01 5.16294360e-01 7.39656985e-01
6.24816748e-04 1.92713976e-01 9.64640453e-02 1.79576135e+00
-3.73292804e-01 -4.80127990e-01 -8.88386905e-01 2.12343886e-01
5.85001484e-02 1.09247684e+00 -6.13658071e-01 -3.73865396e-01
-4.70568091e-02 1.15185010e+00 6.02145314e-01 8.04773986e-01
-8.35243881e-01 2.72101790e-01 7.86567569e-01 1.27266496e-01
1.99770465e-01 -4.04397100e-01 -5.41631937e-01 -1.21815252e+00
-9.54563450e-03 -5.60664415e-01 -6.32225573e-02 -1.77760744e+00
-1.12893116e+00 4.45372999e-01 2.00123116e-01 -9.30436492e-01
-1.60072714e-01 -7.84342408e-01 -2.97783524e-01 1.06571651e+00
-1.21473026e+00 -1.16560411e+00 -4.74565595e-01 3.79334241e-01
7.31096685e-01 2.17074275e-01 9.70863998e-01 -3.77365381e-01
-1.13093480e-01 -2.67069843e-02 -5.38349807e-01 -3.93579379e-02
1.23282075e-01 -1.37252605e+00 7.49049425e-01 4.67403710e-01
4.80147719e-01 5.75588465e-01 7.16806829e-01 -3.50831300e-01
-1.63726425e+00 -1.07169223e+00 2.11879179e-01 -1.05959260e+00
3.31400335e-01 -9.05308127e-01 -6.26994193e-01 1.12745202e+00
1.09278463e-01 3.08815718e-01 5.60973585e-01 1.28210932e-01
-1.03129673e+00 1.04257584e-01 -1.10325265e+00 7.83097446e-01
1.45684969e+00 -9.40664589e-01 -5.09784997e-01 3.96673679e-01
1.13158345e+00 -6.02172315e-01 -4.74772125e-01 6.58861324e-02
6.31417036e-01 -1.33616352e+00 1.29239130e+00 -1.15291321e+00
1.03745317e+00 -3.57672781e-01 -5.62782526e-01 -1.27560496e+00
-2.53138959e-01 -1.41426742e-01 -1.93329692e-01 6.70857131e-01
3.97210598e-01 -3.51516932e-01 7.32083023e-01 7.18462110e-01
-7.82563537e-02 -5.94147623e-01 -3.70221078e-01 -1.39771968e-01
-1.83778420e-01 -4.75663543e-01 7.73596704e-01 8.53536189e-01
-6.20816410e-01 4.82887268e-01 -1.40706301e-01 5.67688048e-01
7.71596849e-01 6.72606647e-01 8.92929435e-01 -1.19020128e+00
-8.48545730e-01 -2.52191424e-01 -2.73134887e-01 -1.26586688e+00
4.42951739e-01 -1.09728968e+00 1.13039598e-01 -1.44691586e+00
3.55979860e-01 -4.72593546e-01 -2.99384035e-02 4.52487171e-01
1.45847052e-01 3.66983831e-01 3.32567692e-01 1.77296609e-01
-5.28456151e-01 5.18341780e-01 1.48757076e+00 -2.75975317e-01
5.56778498e-02 -3.15636486e-01 -9.06782568e-01 7.85415113e-01
6.61370337e-01 -3.11802596e-01 -8.66655111e-01 -1.14010501e+00
5.15829742e-01 2.16309994e-01 1.03796232e+00 -3.74502987e-01
-4.71876085e-01 -4.07288134e-01 8.87550235e-01 -3.53986323e-01
7.43487477e-01 -7.31734097e-01 2.75870740e-01 2.88005825e-02
-9.39730287e-01 -1.09221846e-01 7.40444437e-02 5.63656509e-01
3.42716098e-01 7.36876801e-02 3.83005619e-01 -3.33189964e-01
-4.64744687e-01 2.28434429e-01 4.24381644e-02 -8.55101049e-02
6.96258545e-01 -5.99495061e-02 -2.51215905e-01 -3.08269948e-01
-8.27774107e-01 -9.66625884e-02 6.16222978e-01 4.25171107e-01
8.27471495e-01 -1.24865639e+00 -5.42793751e-01 3.86790216e-01
2.21663296e-01 5.26505232e-01 7.53346235e-02 -8.99716541e-02
-6.24159217e-01 -1.32872039e-04 -2.64652580e-01 -9.02675629e-01
-9.96035516e-01 6.55197501e-01 5.03279150e-01 3.64142776e-01
-7.87373126e-01 7.35208392e-01 1.04493868e+00 -5.88692904e-01
1.31094977e-01 -6.61943734e-01 2.12698653e-01 -5.96432388e-01
3.36267769e-01 -5.75635210e-02 -1.65985286e-01 -4.73526269e-01
-3.04760914e-02 4.14152563e-01 3.80921990e-01 -4.36180800e-01
1.38916910e+00 2.57640518e-02 -1.33907214e-01 7.57907271e-01
1.31979787e+00 1.00500412e-01 -1.75477231e+00 2.12882563e-01
-3.97157103e-01 -7.30248868e-01 2.74070278e-02 -7.89626300e-01
-1.05500782e+00 8.58414829e-01 1.50165290e-01 1.30981714e-01
9.21536326e-01 3.58182698e-01 1.60313696e-01 5.14656901e-01
3.33592772e-01 -2.97014683e-01 4.56004381e-01 4.49376628e-02
1.33200252e+00 -1.08539402e+00 5.39636752e-03 -6.70937240e-01
-3.69289458e-01 1.01308250e+00 6.54664457e-01 -4.33613151e-01
4.48619485e-01 1.75163046e-01 -1.10445712e-02 -6.59158528e-01
-8.41969490e-01 1.88003913e-01 4.66217011e-01 6.49890363e-01
3.91628325e-01 2.67654389e-01 8.16516876e-01 5.59592769e-02
-4.35435265e-01 -3.56809467e-01 3.66317928e-01 5.49314916e-01
-2.15337396e-01 -6.80161715e-01 -2.31234789e-01 6.90431446e-02
7.36311227e-02 -1.08114528e-02 -6.66523099e-01 5.14634669e-01
2.37840265e-01 5.04195333e-01 4.23109055e-01 -1.07126854e-01
2.88123757e-01 -2.12358534e-01 1.08387685e+00 -9.21325862e-01
-2.57530808e-01 -1.63047045e-01 1.74235776e-02 -8.34873021e-01
-4.87131178e-01 -3.17375541e-01 -1.37766457e+00 3.70491520e-02
1.43299431e-01 -2.06956923e-01 8.52047801e-01 7.40349114e-01
3.77375931e-01 7.92767167e-01 7.20649064e-01 -1.24380672e+00
-2.29567960e-01 -3.60692710e-01 -3.63429010e-01 8.16966355e-01
7.06636071e-01 -5.90157807e-01 -3.01839113e-01 5.24845481e-01] | [9.079679489135742, -3.0657904148101807] |
4b327c39-1417-4e0b-82dc-8722d3b8689f | a-preliminary-study-on-pattern-reconstruction | 2302.12972 | null | https://arxiv.org/abs/2302.12972v1 | https://arxiv.org/pdf/2302.12972v1.pdf | A Preliminary Study on Pattern Reconstruction for Optimal Storage of Wearable Sensor Data | Efficient querying and retrieval of healthcare data is posing a critical challenge today with numerous connected devices continuously generating petabytes of images, text, and internet of things (IoT) sensor data. One approach to efficiently store the healthcare data is to extract the relevant and representative features and store only those features instead of the continuous streaming data. However, it raises a question as to the amount of information content we can retain from the data and if we can reconstruct the pseudo-original data when needed. By facilitating relevant and representative feature extraction, storage and reconstruction of near original pattern, we aim to address some of the challenges faced by the explosion of the streaming data. We present a preliminary study, where we explored multiple autoencoders for concise feature extraction and reconstruction for human activity recognition (HAR) sensor data. Our Multi-Layer Perceptron (MLP) deep autoencoder achieved a storage reduction of 90.18% compared to the three other implemented autoencoders namely convolutional autoencoder, Long-Short Term Memory (LSTM) autoencoder, and convolutional LSTM autoencoder which achieved storage reductions of 11.18%, 49.99%, and 72.35% respectively. Encoded features from the autoencoders have smaller size and dimensions which help to reduce the storage space. For higher dimensions of the representation, storage reduction was low. But retention of relevant information was high, which was validated by classification performed on the reconstructed data. | ['Farhana Zulkernine', 'Sazia Mahfuz'] | 2023-02-25 | null | null | null | null | ['human-activity-recognition', 'human-activity-recognition'] | ['computer-vision', 'time-series'] | [ 8.85504261e-02 1.11604430e-01 4.73712713e-01 -3.08245510e-01
-3.53421569e-01 2.67176963e-02 2.33424827e-01 6.39692008e-01
-6.59017980e-01 7.29560077e-01 4.92166072e-01 9.83367935e-02
-3.60845983e-01 -1.06408024e+00 -6.21733963e-01 -6.23660624e-01
-3.73662055e-01 2.08440512e-01 1.18623801e-01 -1.22647688e-01
-1.90524699e-03 5.43105006e-01 -2.22799730e+00 7.38584876e-01
2.64225543e-01 1.74223769e+00 5.08298039e-01 8.23238134e-01
-1.85472801e-01 1.11686957e+00 -7.33195186e-01 2.29699966e-02
-3.21059600e-02 -3.67510468e-02 -7.82075346e-01 -1.59624025e-01
-2.48865291e-01 -6.59047127e-01 -4.71521139e-01 5.38429022e-01
7.10543156e-01 1.22413240e-01 2.68067569e-01 -1.01625311e+00
-5.78265905e-01 4.62676674e-01 3.18225086e-01 3.82271558e-01
4.29109544e-01 -2.58527279e-01 3.29775065e-01 -8.40747178e-01
4.06120092e-01 8.10369074e-01 7.82087922e-01 3.90042543e-01
-4.19126540e-01 -2.02109352e-01 -6.16734862e-01 4.46290255e-01
-1.40935218e+00 -4.52515602e-01 4.27317083e-01 -1.43646553e-01
1.76567328e+00 4.00836617e-01 1.05932069e+00 9.17250872e-01
5.92884243e-01 5.34765065e-01 6.36534095e-01 -3.47773522e-01
3.36264789e-01 3.44968736e-01 3.40339132e-02 6.53423488e-01
2.87890851e-01 1.04932509e-01 -5.76618969e-01 -2.58815259e-01
5.29552937e-01 8.75592351e-01 3.66995707e-02 2.54352897e-01
-1.15229583e+00 8.11501741e-01 2.94781893e-01 5.94593763e-01
-1.17186332e+00 5.17880172e-03 8.07820499e-01 4.54265445e-01
4.14813384e-02 -2.38826256e-02 -6.54543042e-01 -4.78138119e-01
-7.77744949e-01 -1.51157469e-01 8.23557079e-01 8.28864694e-01
6.34936392e-01 3.34340066e-01 1.20068580e-01 6.51059330e-01
5.44591658e-02 5.64477384e-01 1.34200776e+00 -7.97269583e-01
3.08479667e-01 8.52994442e-01 -6.49566054e-02 -1.08260024e+00
-5.37276924e-01 -9.87240300e-02 -1.21706724e+00 -1.96873054e-01
-2.13098139e-01 -1.47762895e-01 -8.35207522e-01 9.93997455e-01
2.69252867e-01 -2.37429500e-01 4.74635094e-01 4.94350582e-01
1.05095363e+00 1.02367258e+00 -2.07427144e-02 -9.08142254e-02
1.48514485e+00 -5.99065185e-01 -1.01518404e+00 9.19995084e-02
5.08404016e-01 -4.63867456e-01 6.40585780e-01 4.26812828e-01
-1.16047430e+00 -8.34764361e-01 -1.16375232e+00 -1.33648023e-01
-8.59005213e-01 8.98171514e-02 5.82367659e-01 4.45758879e-01
-1.13864052e+00 8.90473485e-01 -1.00515854e+00 -4.35521692e-01
3.14440221e-01 6.01997137e-01 -5.38905799e-01 4.51277904e-02
-1.11615801e+00 6.54087305e-01 8.12276840e-01 -1.39663413e-01
-5.22934198e-01 -5.69339097e-01 -7.87505686e-01 4.02415276e-01
-3.21054816e-01 -4.77173120e-01 9.01616991e-01 -7.41051257e-01
-1.35944116e+00 2.59883404e-01 7.25061893e-02 -9.85691249e-01
-2.05704987e-01 -1.42919436e-01 -9.93850291e-01 4.90840852e-01
-3.85318846e-01 6.81847751e-01 6.37675703e-01 -2.84532875e-01
-8.10743511e-01 -5.53189814e-01 -5.83258212e-01 -7.33714700e-02
-1.01775157e+00 -5.87259457e-02 1.09690107e-01 -4.61307645e-01
-8.34633969e-03 -6.66863561e-01 -4.39150967e-02 -8.63865837e-02
1.68959484e-01 -1.48827076e-01 1.07605469e+00 -9.62708771e-01
1.22111380e+00 -2.23008633e+00 -4.83605266e-01 -1.69010852e-02
-4.03039977e-02 4.80959028e-01 1.48284972e-01 8.10475886e-01
6.27962574e-02 -3.49137187e-01 1.39064398e-02 -4.00456153e-02
-3.51894736e-01 6.55288637e-01 -3.00454915e-01 1.24588914e-01
1.51179582e-01 9.80630517e-01 -4.81677741e-01 -5.70543885e-01
2.96350926e-01 1.11763728e+00 -4.31715548e-01 2.92866856e-01
2.25343257e-01 -1.16917051e-01 -5.34893513e-01 6.99861407e-01
2.83950001e-01 -4.96804178e-01 -1.54660285e-01 -3.66609901e-01
-1.32791949e-02 2.15837106e-01 -1.14504039e+00 1.51696098e+00
-3.41704637e-01 4.57509607e-01 -3.42689395e-01 -1.06416774e+00
1.11861420e+00 7.16482162e-01 9.43728745e-01 -1.10606515e+00
2.81754673e-01 1.13028437e-01 -4.24235284e-01 -9.41061795e-01
8.01283896e-01 -1.55047581e-01 -7.15798438e-02 3.07608664e-01
1.98122576e-01 5.53558350e-01 -1.01089135e-01 -1.63092807e-01
1.30701125e+00 -5.17626464e-01 3.74739915e-01 -3.27144898e-02
4.01811510e-01 -6.67901859e-02 2.36828789e-01 4.07238543e-01
-2.06866488e-01 5.52767694e-01 -2.54157960e-01 -1.18076503e+00
-1.31555152e+00 -7.80118108e-01 6.42993767e-03 7.42272437e-01
-4.36975241e-01 -5.59144020e-01 -4.33469176e-01 -1.96541473e-02
1.90537050e-02 2.76208013e-01 -5.39439261e-01 -2.94809133e-01
-5.63826799e-01 -5.51517427e-01 5.44226646e-01 8.29935491e-01
6.34236574e-01 -1.51984823e+00 -1.64482558e+00 6.05527103e-01
-4.64479961e-02 -9.16259527e-01 1.59652233e-01 3.33335191e-01
-1.29466712e+00 -7.77368128e-01 -4.08201247e-01 -7.07304001e-01
4.20466512e-01 -1.62321851e-01 9.31060255e-01 5.19785918e-02
-5.36409974e-01 3.60914588e-01 -5.27406514e-01 -7.33596027e-01
-3.38303775e-01 -2.33710453e-01 1.81463584e-01 -1.71642557e-01
5.75742722e-01 -6.13656223e-01 -8.48896801e-01 -2.49680579e-01
-1.31562746e+00 -2.12121695e-01 7.81728327e-01 8.25801015e-01
8.80933583e-01 4.55572098e-01 6.20156825e-01 -4.11433548e-01
7.31651425e-01 -6.66227579e-01 -6.12876043e-02 -2.80451905e-02
-7.01785445e-01 2.60433942e-01 9.71802413e-01 -4.69586104e-01
-6.51406944e-01 2.58400798e-01 -4.19363111e-01 -5.14417887e-01
-2.25725338e-01 5.27421832e-01 3.06364566e-01 2.56650388e-01
5.23005188e-01 7.31123388e-01 2.78582096e-01 -6.04871154e-01
-1.35229588e-01 1.27323747e+00 4.68260944e-01 1.13999486e-01
-1.42981827e-01 5.86761594e-01 -2.15465397e-01 -1.04559112e+00
-1.31300762e-01 -3.57962877e-01 -2.40666434e-01 7.83737078e-02
8.27243745e-01 -1.00707805e+00 -8.54661107e-01 2.41588593e-01
-9.11416054e-01 6.20883517e-02 -9.21277583e-01 5.61829329e-01
-4.27090645e-01 1.39026374e-01 -8.17769170e-01 -8.17774117e-01
-1.06575155e+00 -8.06088805e-01 9.52874780e-01 7.63799399e-02
-2.02311039e-01 -6.26341879e-01 1.76474303e-02 3.08914017e-02
9.62639570e-01 5.05293131e-01 6.40073836e-01 -8.61621439e-01
-1.36920065e-01 -7.72518456e-01 4.38863151e-02 5.46443164e-01
2.33872399e-01 -4.65885013e-01 -8.07915449e-01 -3.13220471e-01
4.81456488e-01 -3.42913657e-01 5.65444529e-01 1.87204018e-01
1.51126206e+00 -8.31004441e-01 -1.61291435e-01 2.76232570e-01
1.51069200e+00 4.72291619e-01 8.50205719e-01 1.18006572e-01
1.49761438e-01 3.53301883e-01 3.41066897e-01 9.75294769e-01
3.53706986e-01 7.76024710e-04 4.04105783e-01 3.24607551e-01
1.99244660e-03 -2.08204001e-01 2.90479183e-01 1.51344132e+00
4.94257919e-02 -5.37504591e-02 -5.92010200e-01 7.26539612e-01
-1.55330014e+00 -9.68956709e-01 2.36276999e-01 1.99796200e+00
6.10170960e-01 -2.42031753e-01 1.27422258e-01 7.31841922e-01
4.03032333e-01 -1.46760166e-01 -5.25187969e-01 -7.03340769e-01
7.44544938e-02 4.47540402e-01 3.77760202e-01 -1.08910032e-01
-9.44879830e-01 4.16587852e-02 6.24277020e+00 3.83732319e-01
-1.28400540e+00 1.84183538e-01 4.42711353e-01 -4.10767972e-01
1.67268693e-01 -6.08339131e-01 -5.83546460e-01 9.32571828e-01
2.11386204e+00 1.49999633e-01 3.11614335e-01 1.18331039e+00
-1.55177280e-01 -6.56551793e-02 -8.50141943e-01 1.35192239e+00
-2.32594237e-02 -1.52448344e+00 1.28924325e-01 1.46289513e-01
4.46119964e-01 3.01514834e-01 -1.62381858e-01 2.57370681e-01
-2.22890720e-01 -1.05955791e+00 3.81150067e-01 9.58922267e-01
5.12685061e-01 -9.54984486e-01 1.06719089e+00 3.95606518e-01
-1.21113217e+00 -6.95607960e-01 -7.28690863e-01 -3.00697207e-01
3.59958448e-02 7.41968155e-01 -8.84929717e-01 2.22367480e-01
1.34614170e+00 4.03148443e-01 -4.54511106e-01 7.65945554e-01
8.19013238e-01 1.79885909e-01 -6.05749011e-01 -4.14705068e-01
3.34816016e-02 2.95193970e-01 2.29561124e-02 1.16255403e+00
7.43828177e-01 3.01362664e-01 -2.20698968e-01 3.27757567e-01
1.16386242e-01 5.47345504e-02 -6.93826258e-01 -3.35299939e-01
5.01306832e-01 9.84287620e-01 -3.75949293e-01 -7.04086423e-01
-3.80526721e-01 1.04647326e+00 5.08837737e-02 -9.08117145e-02
-4.06635821e-01 -6.62351906e-01 3.04138392e-01 1.60541400e-01
7.04737842e-01 -6.12575896e-02 1.26639664e-01 -7.58553803e-01
2.72303164e-01 -6.13699436e-01 7.10875034e-01 -7.51434326e-01
-1.00054622e+00 9.52220559e-01 -2.26962939e-01 -1.24400544e+00
-7.70912826e-01 -4.82869864e-01 -1.57836556e-01 4.40948129e-01
-1.15705538e+00 -8.18272054e-01 -4.96356279e-01 9.71014798e-01
3.97958040e-01 -2.93607324e-01 1.40595901e+00 5.08114636e-01
-1.23325139e-01 3.38092238e-01 3.98105145e-01 -3.08782700e-02
1.40752615e-02 -7.82028079e-01 -8.30139071e-02 2.26138338e-01
-6.89862221e-02 4.68861163e-01 4.39809829e-01 -4.65019852e-01
-2.14145660e+00 -1.21024573e+00 1.13684058e+00 -1.04589008e-01
3.58577460e-01 -2.93487255e-02 -8.99366617e-01 5.60045004e-01
1.81211054e-01 4.17554379e-01 8.82081747e-01 -6.81325197e-01
1.55175894e-01 -5.24328053e-01 -1.60757530e+00 -1.05477475e-01
5.64315796e-01 -5.17648935e-01 -6.09085262e-01 6.56295493e-02
7.59145796e-01 -2.15782784e-02 -1.54012930e+00 3.33096653e-01
6.90938473e-01 -9.60632980e-01 1.07098007e+00 -5.25825262e-01
3.83936167e-01 -2.59828568e-03 -5.87856352e-01 -8.39203238e-01
-1.33574888e-01 -1.06168941e-01 -9.31559086e-01 8.70880365e-01
7.44953081e-02 -6.16786778e-01 8.02466452e-01 7.51154006e-01
-5.97641692e-02 -1.05949235e+00 -1.01462519e+00 -6.43534601e-01
-3.98232758e-01 -2.04333454e-01 8.99216235e-01 5.06762028e-01
1.13760740e-01 -1.05199918e-01 -4.81752783e-01 -9.64551568e-02
2.44044662e-01 1.29883856e-01 2.13444993e-01 -1.27958620e+00
-3.09975803e-01 2.97413141e-01 -9.16671097e-01 -4.84036118e-01
-6.44144535e-01 -5.35650373e-01 -3.92796487e-01 -1.59637094e+00
-9.29181743e-03 -1.71332657e-02 -7.17409194e-01 5.90028763e-01
5.45976102e-01 2.05803722e-01 -5.68558164e-02 1.82582766e-01
-5.03810644e-01 5.64806461e-01 7.39771962e-01 -8.83903578e-02
-2.49000177e-01 -2.26852790e-01 -4.43978816e-01 2.62565702e-01
1.06914461e+00 -3.65950346e-01 -3.73674095e-01 -3.17537010e-01
2.75285333e-01 2.52552986e-01 2.69223154e-01 -1.47396111e+00
6.24794543e-01 2.81999141e-01 1.09033954e+00 -9.83144045e-01
6.36721492e-01 -1.17411673e+00 6.40414476e-01 7.83751369e-01
-1.04191322e-02 5.26537597e-01 2.84321785e-01 4.84592795e-01
-3.89116883e-01 -1.04145288e-01 3.89411002e-01 -3.33252281e-01
-7.20908403e-01 1.77196428e-01 -5.70274651e-01 -5.80441952e-01
8.35884035e-01 -4.59185034e-01 7.97718614e-02 -2.38653630e-01
-8.18447709e-01 -2.86121845e-01 -2.96919961e-02 4.21155334e-01
1.22518611e+00 -1.49915242e+00 -4.29736614e-01 7.09236383e-01
6.16381876e-03 5.25772683e-02 4.94394213e-01 5.17559230e-01
-6.45476639e-01 7.67581940e-01 -5.39366722e-01 -3.72415155e-01
-1.08767319e+00 9.69346166e-01 5.64220101e-02 -1.10374674e-01
-1.15433800e+00 3.42495859e-01 -6.17554128e-01 8.53964686e-02
3.37816536e-01 -6.34974539e-01 -2.95458972e-01 1.78138427e-02
1.09390068e+00 6.20078504e-01 6.36542559e-01 -4.23192650e-01
-3.88164848e-01 2.09718972e-01 -3.13867293e-02 2.97522694e-01
2.00547552e+00 -7.95757100e-02 -8.35965723e-02 3.62764508e-01
1.65669715e+00 -5.93001604e-01 -9.14460182e-01 -7.46061280e-02
-4.20646742e-02 5.01492731e-02 1.94288909e-01 -6.56913579e-01
-1.16189039e+00 7.27344692e-01 1.14579642e+00 5.59785903e-01
1.46169853e+00 -1.11298300e-02 1.43018305e+00 8.08363497e-01
3.46777678e-01 -1.37069404e+00 2.25457586e-02 2.41254628e-01
8.86940956e-01 -9.04124439e-01 -1.20090984e-01 3.47084254e-01
-4.28425997e-01 1.27862203e+00 5.59962774e-03 -3.47520500e-01
1.20535529e+00 4.50432599e-01 -2.33315900e-01 -4.65094835e-01
-9.19496119e-01 1.35458514e-01 6.17955104e-02 5.03854513e-01
1.85573250e-01 8.08202177e-02 -4.47371230e-02 8.59562278e-01
-5.92445612e-01 4.27281588e-01 2.11755440e-01 1.41291332e+00
-5.91391563e-01 -7.62241125e-01 -3.26370358e-01 8.85879636e-01
-6.95787311e-01 1.07257046e-01 9.76523980e-02 4.46891844e-01
5.74761033e-02 9.01063085e-01 4.96834427e-01 -6.45026624e-01
3.79369467e-01 3.72129619e-01 1.22528583e-01 -6.32860884e-02
-6.70895100e-01 -2.17906892e-01 -8.66992921e-02 -7.64467001e-01
-2.40623742e-01 -3.52963567e-01 -1.54145384e+00 -3.29352230e-01
4.46616001e-02 1.43854216e-01 1.17522252e+00 7.33799040e-01
9.99604821e-01 6.38312340e-01 4.40426975e-01 -5.81672311e-01
-6.47627532e-01 -1.11620283e+00 -6.09532416e-01 5.06574035e-01
5.93450129e-01 -8.69907215e-02 -4.40661497e-02 2.89444298e-01] | [13.935821533203125, 3.2961573600769043] |
a67cf8b7-951f-42e1-ba14-381b074d91d6 | generalized-multiple-intent-conditioned-slot | 2305.11023 | null | https://arxiv.org/abs/2305.11023v1 | https://arxiv.org/pdf/2305.11023v1.pdf | Generalized Multiple Intent Conditioned Slot Filling | Natural language understanding includes the tasks of intent detection (identifying a user's objectives) and slot filling (extracting the entities relevant to those objectives). Prior slot filling methods assume that each intent type cannot occur more than once within a message, however this is often not a valid assumption for real-world settings. In this work, we generalize slot filling by removing the constraint of unique intents in a message. We cast this as a JSON generation task and approach it using a language model. We create a pre-training dataset by combining DBpedia and existing slot filling datasets that we convert for JSON generation. We also generate an in-domain dataset using GPT-3. We train T5 models for this task (with and without exemplars in the prompt) and find that both training datasets improve performance, and that the model is able to generalize to intent types not seen during training. | ['David Barber', 'Edward Challis', 'Cristi Cobzarenco', 'Marius Cobzarenco', 'Arthur Wilcke', 'Harshil Shah'] | 2023-05-18 | null | null | null | null | ['intent-detection', 'slot-filling'] | ['natural-language-processing', 'natural-language-processing'] | [ 4.21178132e-01 7.84998715e-01 -1.68886244e-01 -4.29138869e-01
-5.93432903e-01 -6.56939328e-01 9.69741642e-01 6.56101227e-01
-4.60058033e-01 1.06575084e+00 4.99337077e-01 -6.88854158e-01
-4.85405140e-02 -1.01402330e+00 -7.46350229e-01 3.86262089e-01
1.94915198e-02 1.11535013e+00 3.96516055e-01 -5.37735045e-01
4.99700084e-02 -6.08248748e-02 -1.63924468e+00 7.00132668e-01
1.10120857e+00 7.94562459e-01 1.42173469e-01 4.31295276e-01
-7.12898910e-01 9.68598425e-01 -9.43979561e-01 -2.67341971e-01
1.67306557e-01 -2.51089931e-01 -1.18107653e+00 1.23383954e-01
1.98283732e-01 -1.85805768e-01 -7.84164760e-03 3.48384023e-01
1.17193349e-01 2.21391827e-01 3.95252436e-01 -1.52885079e+00
-1.62336364e-01 8.53599310e-01 4.25498486e-02 -1.28721103e-01
8.97520781e-01 -2.08502084e-01 1.16931081e+00 -4.86676067e-01
1.13848674e+00 1.19196439e+00 6.80647314e-01 4.02080983e-01
-1.31943691e+00 -3.03395629e-01 7.44003654e-02 -1.47897512e-01
-1.18589699e+00 -4.89364833e-01 3.07659686e-01 -3.36360753e-01
1.26829660e+00 5.19568503e-01 3.53657871e-01 9.19930995e-01
-3.34426522e-01 8.36149931e-01 1.06338739e+00 -6.46068335e-01
3.03165406e-01 5.54144025e-01 2.90799797e-01 3.96054566e-01
1.04765169e-01 -2.72128731e-01 -5.27964950e-01 -4.38462466e-01
2.44157135e-01 -2.36415520e-01 1.37928918e-01 -1.42452329e-01
-1.19716215e+00 8.22639704e-01 -8.75227302e-02 1.68661803e-01
-4.78732944e-01 -2.37042800e-01 5.48996985e-01 3.71508747e-01
5.11410534e-01 9.43031728e-01 -7.53823817e-01 -3.51016700e-01
-9.43935692e-01 7.81418681e-01 1.44522667e+00 1.27899241e+00
8.60040665e-01 -6.02855742e-01 -5.26477873e-01 9.67627287e-01
1.01130970e-01 1.79742873e-01 2.06260949e-01 -8.50749075e-01
6.16248608e-01 9.66466904e-01 5.91647506e-01 -7.32148647e-01
-6.76693261e-01 -1.81432366e-01 3.81563641e-02 -3.90849352e-01
5.09872913e-01 -4.42164481e-01 -1.04424453e+00 1.85275316e+00
4.55508262e-01 -1.84075329e-02 3.17761987e-01 3.96529466e-01
1.10578871e+00 7.25318968e-01 5.43011606e-01 -2.08669096e-01
1.66410410e+00 -4.83548135e-01 -8.25169265e-01 -7.44382858e-01
9.79256690e-01 -6.69576943e-01 9.52051044e-01 7.41896704e-02
-6.69884801e-01 -2.05915153e-01 -7.68596590e-01 -2.83899099e-01
-6.36671066e-01 -5.67337275e-02 9.74798858e-01 4.44384724e-01
-7.98943579e-01 2.70895272e-01 -4.25828397e-01 -7.93224037e-01
2.59060785e-02 1.05658457e-01 -3.32186460e-01 -1.70913965e-01
-1.64085889e+00 8.59131753e-01 6.99375391e-01 -7.28810549e-01
-5.20265281e-01 -8.54169071e-01 -1.07345700e+00 -7.55042285e-02
8.74636471e-01 -7.93883085e-01 1.58940649e+00 -5.31593025e-01
-7.00561762e-01 9.44353521e-01 -4.14130241e-01 -7.16244698e-01
3.49286079e-01 -2.01197322e-02 -5.01687765e-01 -1.64511174e-01
6.17388904e-01 1.02305305e+00 4.33352113e-01 -1.31308842e+00
-1.00472641e+00 -1.41705424e-01 7.49244750e-01 3.41658086e-01
-4.22293395e-02 1.93240032e-01 -4.51638937e-01 -3.47174585e-01
9.70972106e-02 -8.41042995e-01 -9.53797773e-02 -3.19939047e-01
-7.24523127e-01 -3.55353892e-01 7.39715219e-01 -7.96323061e-01
1.46144462e+00 -1.72931063e+00 -4.92876232e-01 1.70215264e-01
-7.82791227e-02 -5.97307533e-02 -4.56413031e-02 8.63277674e-01
4.08523437e-03 2.68436462e-01 -1.40104458e-01 -3.75394017e-01
4.02529955e-01 6.12680197e-01 -7.23074377e-01 -3.73576730e-01
7.32594207e-02 7.68454432e-01 -7.89485574e-01 -6.09087288e-01
-2.54545957e-01 8.40513110e-02 -8.91916156e-01 1.00530632e-01
-1.17169702e+00 9.23924893e-02 -3.66281360e-01 4.85335410e-01
4.28400606e-01 -3.82733524e-01 5.65231800e-01 -3.04884221e-02
9.45163053e-03 1.00851989e+00 -1.33000791e+00 1.62843812e+00
-6.51383638e-01 4.20992166e-01 -2.63390273e-01 -6.31582916e-01
7.91858435e-01 4.14322078e-01 5.65801442e-01 -8.03079486e-01
-3.65956515e-01 5.01452535e-02 -3.01226258e-01 -5.37207484e-01
1.03791809e+00 -2.66479939e-01 -5.77493787e-01 8.06848645e-01
6.67923540e-02 -1.34695739e-01 8.67351174e-01 5.59022129e-01
1.26814294e+00 -1.50635391e-01 2.70477146e-01 -1.45010158e-01
5.69393039e-02 6.02780998e-01 5.27588189e-01 1.04629374e+00
5.30250788e-01 2.16498986e-01 8.53219509e-01 -2.99232066e-01
-1.01395726e+00 -6.80765748e-01 -2.54138231e-01 1.37509000e+00
6.31066114e-02 -8.55821371e-01 -5.60595930e-01 -1.05041742e+00
2.48788148e-01 1.33795309e+00 -4.26522523e-01 1.81291357e-01
-3.49819571e-01 -5.44826269e-01 5.81367910e-01 1.81479320e-01
3.05387348e-01 -1.07889128e+00 -7.30690002e-01 3.37458014e-01
-6.08535111e-01 -1.44141912e+00 7.00053051e-02 1.94569722e-01
-3.91051650e-01 -1.07509220e+00 5.52843697e-02 -6.16789579e-01
6.26025796e-01 -1.82387680e-01 1.64770961e+00 3.17453295e-01
-1.44038126e-01 4.90753919e-01 -4.61409003e-01 -5.67788601e-01
-5.87045789e-01 2.19588250e-01 -1.75303161e-01 -4.24622923e-01
7.05323637e-01 -3.27792555e-01 -9.35122743e-02 2.69212395e-01
-1.26318538e+00 3.63320500e-01 1.04294851e-01 6.33788168e-01
1.50247470e-01 2.47091562e-01 5.10387123e-01 -1.43411839e+00
8.61779869e-01 -8.64002287e-01 -3.79554123e-01 2.76852906e-01
-5.02324283e-01 1.81095228e-01 3.22268814e-01 -4.14507184e-03
-1.16248930e+00 -9.74918082e-02 -3.26113492e-01 4.31530088e-01
-4.27615106e-01 8.93126369e-01 -1.27758339e-01 5.22932768e-01
7.60586083e-01 1.04180343e-01 -2.18919560e-01 -6.13411605e-01
3.21037889e-01 6.75146937e-01 1.53039262e-01 -1.01813388e+00
6.88782096e-01 2.90334255e-01 -2.95265526e-01 -8.30484390e-01
-1.10024381e+00 -5.07011473e-01 -2.86294371e-01 2.65319824e-01
4.84145164e-01 -9.81124699e-01 -4.18349475e-01 -8.19562003e-02
-1.20145845e+00 -4.92623001e-01 -6.17210746e-01 5.72549962e-02
-4.78460640e-01 2.00439110e-01 -3.98588717e-01 -6.79613888e-01
-2.79470533e-02 -8.29526961e-01 1.10576892e+00 -3.60275619e-03
-7.41538644e-01 -9.56609130e-01 1.85430199e-02 3.22584212e-01
3.66928518e-01 -1.67358890e-02 1.17133760e+00 -1.28873706e+00
-6.06601179e-01 -1.05802014e-01 -2.26602748e-01 -3.00584495e-01
6.72800243e-02 -3.18922013e-01 -8.27067137e-01 1.82265520e-01
-3.39020371e-01 -4.51208979e-01 5.39262176e-01 -1.46589592e-01
7.68303514e-01 -6.66953564e-01 -6.39150083e-01 1.26540020e-01
1.10499549e+00 3.59515518e-01 7.09609926e-01 5.94857156e-01
8.45310241e-02 7.59073257e-01 9.42674756e-01 6.00197315e-01
8.87249470e-01 9.15043592e-01 6.67127892e-02 -9.89782512e-02
1.95008665e-01 -7.27666736e-01 -1.00440353e-01 -2.00046286e-01
6.37785792e-01 -5.30598998e-01 -1.10442305e+00 8.68793964e-01
-1.70291376e+00 -8.76486838e-01 1.12321012e-01 1.95564449e+00
1.27232254e+00 3.17884535e-01 1.04068883e-01 1.21919252e-01
2.48115346e-01 2.62185670e-02 -2.58366346e-01 -2.56892204e-01
5.15596159e-02 2.02105433e-01 2.28320658e-01 7.53729761e-01
-1.10358047e+00 1.11979795e+00 6.43824577e+00 5.23959756e-01
-8.22714686e-01 -9.64452792e-03 4.30775583e-01 2.97524720e-01
-8.69862497e-01 4.80935663e-01 -1.12592340e+00 3.43520999e-01
8.15547585e-01 -3.11093658e-01 2.00104892e-01 6.95615768e-01
-4.78240438e-02 -3.74482036e-01 -1.18924177e+00 2.63187587e-01
-1.83573633e-01 -1.31016397e+00 6.61370531e-02 -5.26065566e-03
3.88752371e-01 -3.07388842e-01 -3.18329871e-01 9.20557439e-01
5.01110137e-01 -9.50796962e-01 6.47207141e-01 2.68083751e-01
7.70365298e-01 -2.38071457e-01 4.07998264e-01 6.44048333e-01
-7.38955081e-01 7.97382463e-03 -7.28754252e-02 -2.04054072e-01
4.16649461e-01 8.20161164e-01 -1.66651475e+00 4.02930290e-01
2.54366457e-01 4.33110029e-01 -7.03577995e-01 9.68361080e-01
-2.70515442e-01 4.51902211e-01 -7.11848140e-01 1.63646061e-02
2.58097798e-02 1.24848008e-01 6.32069945e-01 1.17863345e+00
2.03372836e-01 -4.95470054e-02 4.35691237e-01 9.03402746e-01
3.47051420e-03 -2.88356319e-02 -5.99224985e-01 -2.87230462e-01
7.49923527e-01 1.15304148e+00 -7.05762565e-01 -6.22547686e-01
-4.02772158e-01 7.07004309e-01 -2.03752890e-02 4.06805605e-01
-6.17211163e-01 -4.08231199e-01 7.71931887e-01 4.77066576e-01
2.10956410e-01 -6.32726680e-03 -2.83611625e-01 -1.01193392e+00
8.97570923e-02 -1.04944468e+00 7.03725219e-01 -9.39765871e-01
-8.08633566e-01 4.20376629e-01 3.60452294e-01 -9.01306152e-01
-8.05413723e-01 -2.99203992e-01 -3.21116477e-01 7.95030177e-01
-1.45319963e+00 -1.04689205e+00 -1.77585766e-01 1.96291283e-01
3.78172189e-01 3.42178881e-01 9.89383638e-01 4.01952296e-01
-1.39721781e-01 4.15065557e-01 -6.09023213e-01 1.58388615e-01
6.09579742e-01 -1.27140391e+00 7.43292093e-01 7.14818120e-01
4.37949188e-02 1.06302154e+00 1.21326339e+00 -9.97872233e-01
-1.14920008e+00 -9.87807989e-01 1.61411428e+00 -5.93449056e-01
5.99489272e-01 -6.69105768e-01 -8.96483481e-01 1.11953211e+00
1.69690162e-01 -2.11915925e-01 6.40394926e-01 5.46432734e-01
-4.17273730e-01 3.55269343e-01 -1.22750974e+00 4.29287583e-01
1.23536408e+00 -5.27324557e-01 -9.12780762e-01 6.89999759e-01
1.02079678e+00 -8.01881909e-01 -8.36654305e-01 3.14757794e-01
1.97550893e-01 -6.49524748e-01 8.99746716e-01 -8.79708946e-01
3.39917362e-01 -4.21916664e-01 -1.11116029e-01 -1.20301104e+00
3.02911311e-01 -6.36587262e-01 -1.06523857e-01 1.43294382e+00
9.10121679e-01 -5.71550846e-01 7.44026959e-01 1.12266946e+00
-1.58513725e-01 -3.65117848e-01 -7.51400948e-01 -5.55679798e-01
-3.51445258e-01 -5.93403101e-01 9.29428577e-01 1.07777703e+00
5.57357132e-01 6.44271016e-01 5.13143651e-02 -2.45064739e-02
1.85262203e-01 3.38380933e-01 9.31509435e-01 -1.48285913e+00
-3.57369840e-01 1.19912766e-01 1.49028718e-01 -1.29195249e+00
1.40058219e-01 -8.41165423e-01 5.88754490e-02 -2.10303879e+00
5.45614362e-02 -1.23390472e+00 3.90661120e-01 1.01305079e+00
2.85229050e-02 -2.09824950e-01 4.15511876e-02 -4.93247807e-02
-7.99488127e-01 1.96169361e-01 8.04074466e-01 -6.05709068e-02
-3.38960499e-01 -1.12007558e-01 -1.01081467e+00 3.47658753e-01
6.15146637e-01 -4.97246027e-01 -6.42103374e-01 -2.44263247e-01
6.03943825e-01 2.64158517e-01 2.55670816e-01 -8.58967066e-01
1.12365805e-01 -3.41872901e-01 9.23870131e-03 -4.72016215e-01
3.99749875e-01 -8.65495682e-01 2.66859472e-01 4.16595265e-02
-6.04362726e-01 -2.00939804e-01 5.44599771e-01 9.86406431e-02
-1.47467792e-01 -3.97012353e-01 6.72184229e-02 -3.01001281e-01
-9.72069621e-01 -6.20530397e-02 -3.23720157e-01 5.50819635e-01
7.95529366e-01 -1.13238201e-01 -6.27191007e-01 -5.06894767e-01
-6.87709928e-01 4.04766977e-01 6.20488882e-01 6.10288203e-01
1.66296557e-01 -1.01050746e+00 -3.84533554e-01 3.39406013e-01
4.24340814e-01 7.10382760e-02 2.62329336e-02 5.72830677e-01
-1.74662977e-01 6.99754477e-01 6.30142465e-02 -3.77905458e-01
-1.07264292e+00 3.08481067e-01 1.78276226e-01 -5.12591779e-01
-3.83370966e-01 5.36959827e-01 -3.54345366e-02 -9.96556818e-01
2.48013169e-01 -3.51850331e-01 -4.30334151e-01 2.40559712e-01
5.39490402e-01 -1.49372905e-01 2.23658055e-01 -4.01514739e-01
-3.02393615e-01 -2.50268072e-01 -1.29132926e-01 -2.96428770e-01
1.24967372e+00 -3.72713581e-02 -1.54719993e-01 2.25645155e-01
7.64858007e-01 2.03759462e-01 -7.58017242e-01 -3.22834820e-01
4.12321776e-01 -4.00952339e-01 -3.66602719e-01 -1.21208811e+00
-3.15489262e-01 2.51933038e-01 4.43900079e-02 7.73000538e-01
8.33740711e-01 1.26740009e-01 8.45074356e-01 4.66403574e-01
4.81791645e-01 -1.11218750e+00 -3.32523733e-01 9.34374392e-01
7.41773844e-01 -1.12103653e+00 -1.20483987e-01 -6.57137454e-01
-6.71002984e-01 7.02656269e-01 6.98851883e-01 6.02097034e-01
3.44208211e-01 3.11866105e-01 1.24279194e-01 -4.07875568e-01
-1.11039245e+00 -4.19841558e-01 1.47022009e-01 6.66564703e-01
5.57837963e-01 -2.82721408e-02 -5.26625931e-01 5.55808008e-01
-4.37734872e-01 2.79146403e-01 6.14948034e-01 1.40533376e+00
-4.50913221e-01 -1.41458988e+00 -1.45002395e-01 7.38911331e-01
-3.13810229e-01 -3.71798843e-01 -4.87317502e-01 8.51871789e-01
1.93318442e-01 9.97809470e-01 2.29553536e-01 -3.24891448e-01
4.75046724e-01 4.46644962e-01 9.20295939e-02 -1.20614469e+00
-4.94467676e-01 -3.69683266e-01 1.02370512e+00 -3.86702865e-01
-1.37107596e-01 -6.92314327e-01 -1.51553631e+00 -9.71427932e-02
-4.54102047e-02 6.05857730e-01 6.29727185e-01 1.15107620e+00
6.73150778e-01 4.16289121e-01 7.22170845e-02 -1.00788660e-01
-1.93400934e-01 -9.35238004e-01 -3.27752918e-01 7.34876335e-01
1.61115795e-01 -6.69547200e-01 -1.22956246e-01 1.49811739e-02] | [10.277863502502441, 8.533125877380371] |
49fb1133-1409-4de5-a9e4-0f3c02337cba | low-cost-on-device-partial-domain-adaptation | 2203.00772 | null | https://arxiv.org/abs/2203.00772v1 | https://arxiv.org/pdf/2203.00772v1.pdf | Low-Cost On-device Partial Domain Adaptation (LoCO-PDA): Enabling efficient CNN retraining on edge devices | With the increased deployment of Convolutional Neural Networks (CNNs) on edge devices, the uncertainty of the observed data distribution upon deployment has led researchers to to utilise large and extensive datasets such as ILSVRC'12 to train CNNs. Consequently, it is likely that the observed data distribution upon deployment is a subset of the training data distribution. In such cases, not adapting a network to the observed data distribution can cause performance degradation due to negative transfer and alleviating this is the focus of Partial Domain Adaptation (PDA). Current works targeting PDA do not focus on performing the domain adaptation on an edge device, adapting to a changing target distribution or reducing the cost of deploying the adapted network. This work proposes a novel PDA methodology that targets all of these directions and opens avenues for on-device PDA. LoCO-PDA adapts a deployed network to the observed data distribution by enabling it to be retrained on an edge device. Across subsets of the ILSVRC12 dataset, LoCO-PDA improves classification accuracy by 3.04pp on average while achieving up to 15.1x reduction in retraining memory consumption and 2.07x improvement in inference latency on the NVIDIA Jetson TX2. The work is open-sourced at \emph{link removed for anonymity}. | ['Christos-Savvas Bouganis', 'Aditya Rajagopal'] | 2022-03-01 | null | null | null | null | ['partial-domain-adaptation'] | ['methodology'] | [-1.54739782e-01 6.84228241e-02 -4.66599047e-01 -5.59127033e-01
-2.31931880e-01 -1.03618228e+00 -3.99526618e-02 -1.26898333e-01
-4.54367518e-01 7.91775823e-01 -1.87706485e-01 -9.25135434e-01
-2.16124505e-01 -7.57608712e-01 -9.52103734e-01 -3.51256251e-01
3.14443521e-02 3.80850941e-01 1.93983302e-01 2.26906836e-01
-2.57564485e-01 8.75675797e-01 -1.11232185e+00 1.94546402e-01
4.37200427e-01 1.37065589e+00 -3.34755599e-01 7.95570970e-01
1.00052677e-01 4.43963587e-01 -9.79051232e-01 -2.46836558e-01
2.97970444e-01 2.24803552e-01 -5.55779934e-01 -3.43166858e-01
5.26969731e-01 -6.33177102e-01 -4.34878767e-01 9.26992834e-01
8.46441269e-01 9.71554816e-02 8.59764591e-02 -1.62339652e+00
-2.98381507e-01 5.40279508e-01 -2.45046407e-01 4.92237389e-01
-2.54864782e-01 -1.28960595e-01 3.30632478e-01 -2.70226061e-01
2.80203462e-01 5.97564697e-01 9.22542334e-01 6.20176971e-01
-1.03614557e+00 -1.00927198e+00 1.29691228e-01 6.03857711e-02
-1.59954250e+00 -5.98511994e-01 7.31068909e-01 -8.95152017e-02
9.41628873e-01 3.41481656e-01 1.93754047e-01 1.47616124e+00
-2.82527283e-02 5.79109907e-01 5.90527415e-01 7.49813914e-02
7.15771616e-01 5.24851263e-01 1.27078831e-01 1.23246387e-01
4.66488153e-01 1.41798332e-01 -4.04049665e-01 -3.11202914e-01
6.37167156e-01 -2.21088544e-01 5.70009686e-02 -1.46427721e-01
-4.67860043e-01 4.68949109e-01 4.66672063e-01 -2.17432365e-01
-5.11644125e-01 2.55780011e-01 9.05906081e-01 2.88060844e-01
5.54310262e-01 3.08837742e-02 -9.81686652e-01 -7.30913520e-01
-6.90767109e-01 1.72071494e-02 9.31560695e-01 1.32758808e+00
3.56793851e-01 2.24127188e-01 9.75067243e-02 5.19323528e-01
1.66897941e-02 5.30564189e-01 1.14489578e-01 -8.30338717e-01
6.40487254e-01 4.92394686e-01 1.78006098e-01 -5.99433661e-01
-6.29945695e-01 -6.06185377e-01 -6.23584211e-01 2.22329021e-01
2.84814239e-01 -1.01826763e+00 -7.45122313e-01 1.62694848e+00
5.59999824e-01 3.76839787e-01 2.71595884e-02 7.74895608e-01
6.16348207e-01 3.12561572e-01 1.20327733e-01 5.26401758e-01
8.74860048e-01 -5.55798769e-01 -4.01838601e-01 -8.52862597e-02
6.10023797e-01 -4.66474533e-01 9.32430446e-01 5.28806925e-01
-7.30324507e-01 -5.03497481e-01 -1.24348462e+00 4.19692069e-01
-4.84170109e-01 2.16697678e-01 5.83113432e-01 1.14816689e+00
-1.00337934e+00 5.32427907e-01 -1.02343893e+00 -5.15380621e-01
1.00249469e+00 9.35902953e-01 -1.59784243e-01 -1.42976850e-01
-9.87464845e-01 4.19175774e-01 4.26977903e-01 3.45527790e-02
-4.93404716e-01 -1.08435571e+00 -4.21756119e-01 6.37032613e-02
2.20394194e-01 -5.28164089e-01 1.12084615e+00 -9.64716077e-01
-1.61949968e+00 2.38723636e-01 3.05093855e-01 -7.47392118e-01
3.92167985e-01 -2.46361017e-01 -1.01422441e+00 -1.24120168e-01
-2.49031678e-01 5.14712095e-01 6.56101346e-01 -8.34622204e-01
-8.46052766e-01 -3.12861115e-01 1.60123054e-02 -8.83812364e-03
-7.75827587e-01 -1.49534464e-01 -3.91474396e-01 -2.30937019e-01
-6.35813475e-01 -1.26653421e+00 1.96305603e-01 -5.88622727e-02
-4.05265421e-01 7.54160732e-02 1.49597514e+00 -4.19081360e-01
1.15090263e+00 -2.37115407e+00 -7.80854106e-01 3.30129027e-01
1.15870774e-01 9.18707907e-01 -3.72045711e-02 3.31587791e-02
-7.00239241e-02 1.70566142e-01 4.14948255e-01 -3.78592610e-01
-3.42573971e-02 2.52225816e-01 -2.89503813e-01 5.25639296e-01
1.30936801e-01 4.70310628e-01 -6.28544450e-01 9.49258730e-02
3.74485254e-01 6.91497445e-01 -5.02096057e-01 9.49718896e-03
3.04917637e-02 4.86615241e-01 -5.06851792e-01 5.82220972e-01
1.13862729e+00 3.81687023e-02 9.17313844e-02 -1.91430822e-01
7.99796134e-02 2.60302454e-01 -1.03860962e+00 1.31244373e+00
-6.53152466e-01 8.77472639e-01 1.30730957e-01 -8.53351891e-01
1.05835664e+00 2.63771325e-01 4.24288899e-01 -5.73904037e-01
5.18085301e-01 3.68839093e-02 6.67416083e-04 -3.50566417e-01
3.90576512e-01 3.66353482e-01 -2.35523239e-01 1.61396861e-01
-1.70546144e-01 6.35460198e-01 -3.64477694e-01 1.84274409e-02
1.51278865e+00 -1.11225054e-01 -2.45521486e-01 -1.96800038e-01
-5.14425337e-03 3.49881090e-02 5.18929899e-01 7.90645063e-01
-4.22239244e-01 3.81683409e-01 3.88684988e-01 -6.79228902e-01
-9.70086515e-01 -1.26004660e+00 -2.75251389e-01 1.00981593e+00
-9.21472684e-02 -1.33474022e-01 -7.16303587e-01 -1.03940475e+00
8.62253085e-02 1.01002765e+00 -5.03808081e-01 -4.62964892e-01
-2.42081791e-01 -6.43728971e-01 1.23075068e+00 1.00180399e+00
9.10716414e-01 -6.23190582e-01 -7.56179988e-01 8.52366239e-02
3.16926569e-01 -1.33416867e+00 -3.55043888e-01 3.57215613e-01
-8.21056604e-01 -8.76276731e-01 -1.25649631e-01 -3.22214752e-01
5.83049119e-01 -6.97905123e-02 9.45787549e-01 -2.43035898e-01
1.67892184e-02 5.20276904e-01 -3.32024932e-01 -7.58880377e-01
-1.48184597e-01 6.13539875e-01 2.94976324e-01 -4.83794212e-02
7.79162407e-01 -7.34690487e-01 -6.32285297e-01 2.53304511e-01
-6.28948748e-01 -6.01469100e-01 3.13521892e-01 4.63989794e-01
2.69079864e-01 5.67883730e-01 7.99576044e-01 -1.07508278e+00
6.88751996e-01 -8.66788089e-01 -6.33170009e-01 -3.95758636e-02
-5.71836889e-01 -3.70494843e-01 9.70267355e-01 -8.63956153e-01
-1.02218330e+00 2.38989532e-01 -7.00456649e-02 -8.53763700e-01
-5.64213216e-01 2.37741098e-01 -4.24166381e-01 3.78391482e-02
7.66729891e-01 -3.85053366e-01 -1.09837905e-01 -2.44023576e-01
-6.36432087e-03 1.17763186e+00 6.04841173e-01 -4.90149319e-01
5.09782016e-01 4.31592464e-01 -4.51624468e-02 -7.90666699e-01
-4.85989928e-01 -2.63489872e-01 -3.20641637e-01 -5.79488426e-02
5.22760034e-01 -1.09765220e+00 -9.08022523e-01 4.88248885e-01
-9.01579082e-01 -8.16303313e-01 -1.98029086e-01 5.52223563e-01
1.28118336e-01 -2.74404883e-01 -1.32962674e-01 -6.16750181e-01
-5.98268449e-01 -7.63468385e-01 8.77199531e-01 5.10206580e-01
-5.29923677e-01 -1.27180910e+00 -2.74107993e-01 -3.93689685e-02
7.90362775e-01 1.93202570e-01 4.48276430e-01 -8.95131111e-01
-2.25288630e-01 -7.07777679e-01 -2.78967619e-01 4.72227991e-01
1.94653094e-01 -1.23488763e-03 -1.14118445e+00 -6.49054825e-01
-2.97147691e-01 5.61083704e-02 8.91946256e-02 4.98151243e-01
1.51302040e+00 -5.03547899e-02 -4.98799026e-01 8.29199255e-01
1.22987485e+00 4.76504028e-01 7.45076120e-01 3.42683911e-01
6.69928372e-01 1.67032616e-04 5.16836882e-01 5.70499361e-01
2.75743991e-01 5.94020128e-01 6.91043377e-01 -1.64420158e-01
5.60408458e-02 -2.28306293e-01 1.57566398e-01 -4.55210507e-02
3.19732159e-01 -5.65754831e-01 -8.55320692e-01 6.17915273e-01
-1.67513132e+00 -5.02333641e-01 -1.13076426e-01 2.43181968e+00
2.67033726e-01 3.40354264e-01 4.05801862e-01 -1.77647874e-01
8.58270943e-01 -3.40216041e-01 -1.22575605e+00 -7.97190011e-01
2.82076240e-01 3.89133126e-01 1.13999546e+00 2.69384801e-01
-1.04952765e+00 6.88754976e-01 5.60278130e+00 5.16851485e-01
-1.72749984e+00 3.58977392e-02 7.19679713e-01 -3.63146842e-01
3.26965630e-01 -2.27534279e-01 -1.00010657e+00 8.12335372e-01
1.45822895e+00 7.15171695e-02 4.28751558e-01 1.32993305e+00
1.12411723e-01 -1.63141312e-03 -9.42875743e-01 1.15879309e+00
-1.23761021e-01 -1.13664007e+00 -5.11752725e-01 3.84510487e-01
4.68891174e-01 4.96883959e-01 3.98025513e-01 4.73329902e-01
4.53123957e-01 -8.72702599e-01 4.66368169e-01 1.57210469e-01
1.13319159e+00 -1.20702291e+00 1.05411112e+00 2.59627372e-01
-1.07687664e+00 -1.40898958e-01 -3.11670780e-01 -1.10859551e-01
-8.27007964e-02 5.51923394e-01 -1.39202523e+00 1.76099285e-01
1.15059423e+00 3.73868614e-01 -5.48118889e-01 9.31735337e-01
6.21189810e-02 1.01677048e+00 -7.05184281e-01 -9.32840779e-02
-9.43241343e-02 3.43663901e-01 4.86217290e-01 9.69525933e-01
3.66376221e-01 -2.15691775e-01 -7.64739290e-02 5.68360448e-01
-2.84937203e-01 -5.38097799e-01 -5.58661282e-01 2.32381430e-02
1.02674425e+00 1.30826509e+00 -5.19835651e-01 -2.05249071e-01
-1.46319166e-01 9.31661963e-01 1.93563193e-01 2.72735864e-01
-1.21739769e+00 -5.27047873e-01 8.95593047e-01 1.50016412e-01
4.81834173e-01 -2.79876553e-02 -5.66998720e-01 -5.14499843e-01
1.37508780e-01 -5.03579855e-01 3.63845825e-01 -5.98155022e-01
-1.30250013e+00 7.65098333e-01 3.71650867e-02 -1.15365064e+00
2.97059678e-02 -6.11045361e-01 -7.92493939e-01 8.34719539e-01
-1.06055474e+00 -9.00577486e-01 -4.97701496e-01 7.38334537e-01
2.08846226e-01 -3.94944906e-01 8.82028878e-01 8.05837870e-01
-7.18917906e-01 1.18560743e+00 1.97883040e-01 1.34603500e-01
6.99619055e-01 -1.08823061e+00 5.77201247e-01 6.57410145e-01
-3.89158934e-01 5.57669520e-01 5.40215671e-01 -6.48695827e-01
-1.20436037e+00 -1.74380076e+00 2.37808391e-01 -6.32039726e-01
6.04007125e-01 -5.14525473e-01 -8.65606785e-01 9.08324063e-01
3.58164012e-02 5.47731519e-01 8.49787235e-01 2.33658761e-01
-1.28234774e-01 -5.88782609e-01 -1.47519100e+00 4.83029515e-01
9.55524981e-01 -4.15529191e-01 3.51525903e-01 -2.28341687e-02
5.36362708e-01 -7.67841280e-01 -1.01158857e+00 2.28776887e-01
3.63577604e-01 -5.70635855e-01 6.91018462e-01 -6.65355682e-01
-2.20585018e-01 -2.15745375e-01 -2.40235075e-01 -1.46728551e+00
7.86220878e-02 -6.86818302e-01 -4.03816938e-01 1.56807995e+00
4.00655061e-01 -8.37114692e-01 1.30978167e+00 1.01146448e+00
-2.11104870e-01 -5.77698469e-01 -8.74031305e-01 -7.81505167e-01
3.85886878e-02 -6.86403036e-01 1.04376435e+00 8.15650582e-01
-3.31259906e-01 3.03646624e-01 -2.39781663e-01 7.24863827e-01
3.99495691e-01 -6.90981686e-01 1.02542579e+00 -1.02252400e+00
-2.65670925e-01 5.48998192e-02 -5.25540352e-01 -9.46764350e-01
6.30553439e-02 -6.04099452e-01 -2.69734830e-01 -1.00359511e+00
-5.84613502e-01 -9.01634395e-01 -2.18045548e-01 5.37988842e-01
3.95184308e-01 1.74348176e-01 -8.41114745e-02 -4.19575959e-01
-7.49190509e-01 1.89738035e-01 4.14817035e-01 2.74064709e-02
-6.05166674e-01 4.43396002e-01 -8.08403254e-01 5.02697706e-01
1.37415969e+00 -5.55078566e-01 -7.97931731e-01 -6.36836886e-01
1.46954387e-01 -3.41882169e-01 4.75917518e-01 -1.40792632e+00
3.97972077e-01 3.04993868e-01 6.05891824e-01 -5.32714725e-01
1.37408957e-01 -1.21962702e+00 3.89302641e-01 8.94839466e-02
1.37185147e-02 9.35050994e-02 9.98210013e-01 4.36533511e-01
3.07534873e-01 1.08297370e-01 4.93443191e-01 5.85078955e-01
-6.60844803e-01 3.04341167e-01 -3.53298753e-01 -1.11010112e-01
1.05941153e+00 -2.80366480e-01 -5.12525678e-01 -2.55047709e-01
-5.82785487e-01 1.18881851e-01 3.21580440e-01 5.78682244e-01
1.88688651e-01 -1.20642102e+00 -8.12366307e-02 2.35638335e-01
-8.80900249e-02 3.32635045e-01 4.19934809e-01 5.87350786e-01
-4.48434412e-01 2.59797186e-01 -2.10854396e-01 -5.89581370e-01
-1.37550914e+00 2.87427455e-01 5.27205229e-01 2.18258128e-01
-4.52826917e-01 8.11632514e-01 -3.90960932e-01 -5.81857085e-01
4.59635258e-01 -1.09749056e-01 1.92583293e-01 -4.85221714e-01
5.42456329e-01 5.60422599e-01 4.58257198e-01 -1.88955203e-01
-5.09150445e-01 -6.29552528e-02 -2.57709801e-01 3.87338609e-01
1.22710252e+00 -1.03841923e-01 4.15922672e-01 1.06459543e-01
1.48131514e+00 -3.14581990e-02 -1.72123086e+00 7.28755891e-02
-3.40719938e-01 -3.11501414e-01 3.71682167e-01 -1.01402783e+00
-1.40661001e+00 6.13056123e-01 1.16991079e+00 1.00495763e-01
1.20677221e+00 -1.68055058e-01 8.96343887e-01 3.33576053e-01
3.48799020e-01 -1.17071688e+00 -2.14350566e-01 2.96366990e-01
1.38364479e-01 -1.11948943e+00 -1.83009103e-01 -2.74515718e-01
-5.62355101e-01 9.85339940e-01 9.20712411e-01 -1.69865191e-01
9.45531607e-01 6.74335539e-01 -1.68697592e-02 -1.11391649e-01
-5.27298391e-01 3.92584294e-01 -1.72533885e-01 1.19615257e+00
-8.97796825e-02 2.69386023e-01 5.01273334e-01 7.05735147e-01
-1.33776098e-01 3.00085515e-01 4.53903079e-01 8.62851441e-01
3.04971337e-01 -1.03611374e+00 -2.88091362e-01 6.09162748e-01
-4.91472065e-01 1.83896333e-01 -3.03518385e-01 8.34581494e-01
3.97812873e-01 9.73814964e-01 6.42243505e-01 -6.51340604e-01
5.23453951e-01 -1.95533484e-01 -6.75141392e-03 -3.23103279e-01
-5.10399997e-01 -4.92703915e-01 4.05374855e-01 -3.43621701e-01
9.14342776e-02 -6.39609337e-01 -1.22050834e+00 -7.85645485e-01
-2.29441673e-01 -2.09357366e-01 9.75800157e-01 6.61971509e-01
1.07689154e+00 7.95964420e-01 6.18724108e-01 -5.33388615e-01
-3.92075360e-01 -6.44163072e-01 -5.50202787e-01 3.08903530e-02
1.00531615e-01 -5.84676087e-01 -2.53029376e-01 -3.33927840e-01] | [8.162870407104492, 2.6087539196014404] |
d9415531-29d7-4904-ae47-1b9b1924de58 | autolv-automatic-lecture-video-generator | 2209.08795 | null | https://arxiv.org/abs/2209.08795v1 | https://arxiv.org/pdf/2209.08795v1.pdf | AutoLV: Automatic Lecture Video Generator | We propose an end-to-end lecture video generation system that can generate realistic and complete lecture videos directly from annotated slides, instructor's reference voice and instructor's reference portrait video. Our system is primarily composed of a speech synthesis module with few-shot speaker adaptation and an adversarial learning-based talking-head generation module. It is capable of not only reducing instructors' workload but also changing the language and accent which can help the students follow the lecture more easily and enable a wider dissemination of lecture contents. Our experimental results show that the proposed model outperforms other current approaches in terms of authenticity, naturalness and accuracy. Here is a video demonstration of how our system works, and the outcomes of the evaluation and comparison: https://youtu.be/cY6TYkI0cog. | ['Sanjay Jha', 'Yang song', 'Wenbin Wang'] | 2022-09-19 | null | null | null | null | ['talking-head-generation', 'video-generation'] | ['computer-vision', 'computer-vision'] | [-1.01535045e-01 1.87029153e-01 7.83238411e-02 -3.88661653e-01
-1.05266905e+00 -9.25620615e-01 5.07912934e-01 -4.96388376e-01
9.21654403e-02 8.44646633e-01 4.83328909e-01 -4.31728572e-01
3.17834139e-01 -4.87416804e-01 -6.50841415e-01 -6.65125847e-01
4.91313994e-01 -9.86189023e-02 3.35850894e-01 -3.82931292e-01
2.95653105e-01 2.87916481e-01 -1.47913945e+00 2.51086533e-01
5.36373377e-01 6.35097206e-01 2.44824495e-02 1.25192678e+00
-1.54256880e-01 9.72769618e-01 -1.18526769e+00 -5.26910901e-01
-2.60864254e-02 -7.88216949e-01 -8.41671228e-01 1.60655439e-01
6.54050708e-01 -3.83854747e-01 -7.09005833e-01 8.32966983e-01
1.14175189e+00 7.09809363e-01 9.44893658e-02 -1.33433163e+00
-5.00275671e-01 7.02051580e-01 -1.08240262e-01 3.08707267e-01
1.06055152e+00 3.45357895e-01 3.31076205e-01 -4.79050964e-01
7.47479677e-01 1.17139387e+00 4.85482693e-01 1.10792840e+00
-5.61434031e-01 -1.14463341e+00 4.55815233e-02 1.94274768e-01
-1.35995579e+00 -8.16926241e-01 8.91166270e-01 -2.93960627e-02
1.98802575e-01 6.22766912e-01 4.22369808e-01 1.58940411e+00
1.77979708e-01 6.05968833e-01 7.43779004e-01 -4.44825590e-01
1.49020582e-01 5.60026884e-01 -3.10587853e-01 6.56669021e-01
-6.11578405e-01 -2.59923041e-01 -7.83059299e-01 8.89793187e-02
5.68247914e-01 -4.77785729e-02 -7.96650052e-01 1.75512925e-01
-1.18959188e+00 5.43829441e-01 6.14609243e-03 1.20747142e-01
1.10312022e-01 -6.18660636e-02 3.68468016e-01 2.88992316e-01
6.74599484e-02 2.08867952e-01 1.57130182e-01 -4.85172808e-01
-1.10967410e+00 2.45301634e-01 1.05290508e+00 1.41590571e+00
2.54913326e-02 4.75348383e-01 -2.92153299e-01 5.43731987e-01
2.51368970e-01 5.68250120e-01 1.03092432e+00 -1.28893602e+00
2.96630442e-01 3.92279401e-02 4.53262925e-02 -6.36352062e-01
2.40400419e-01 -1.21589363e-01 -2.54995733e-01 1.24338858e-01
1.76186755e-01 -6.24787211e-01 -6.17109358e-01 1.47574890e+00
4.19294894e-01 7.45898128e-01 3.07918221e-01 7.59906590e-01
1.64574778e+00 9.33066726e-01 -2.13095337e-01 -3.26068729e-01
1.29334247e+00 -1.27129483e+00 -1.23913455e+00 3.86327922e-01
1.31898269e-01 -1.36002338e+00 1.39850271e+00 3.41091275e-01
-1.32301891e+00 -9.26152825e-01 -9.63828444e-01 2.56307404e-02
-9.08138826e-02 -1.37254745e-01 -1.42032534e-01 9.16904390e-01
-1.11957443e+00 3.27002257e-01 -5.20549297e-01 -1.65207610e-01
4.60189618e-02 1.20950639e-01 -2.56262451e-01 1.00084521e-01
-1.11362135e+00 2.99670011e-01 -3.05185974e-01 -2.35408068e-01
-1.08161044e+00 -8.05004358e-01 -8.06217670e-01 1.57390252e-01
2.17988998e-01 -3.97804946e-01 1.93902731e+00 -1.01929331e+00
-2.57284904e+00 5.19499302e-01 -2.20690444e-01 -8.90272111e-02
6.50072753e-01 -1.18871585e-01 -6.76962495e-01 5.27421296e-01
-1.98688030e-01 6.82804108e-01 7.06100166e-01 -1.00725257e+00
-6.22916281e-01 2.05887020e-01 1.87515810e-01 4.67653841e-01
-2.71129817e-01 -1.52207047e-01 -2.24957973e-01 -6.11943424e-01
-4.63477820e-01 -9.03834522e-01 1.71543971e-01 -2.13917032e-01
-5.04896164e-01 6.95223585e-02 1.31925154e+00 -5.83502769e-01
1.27402043e+00 -2.22757721e+00 -2.65064031e-01 -1.16489582e-01
-1.51066467e-01 5.04806995e-01 2.33359337e-02 6.57751799e-01
-1.03009358e-01 2.14647666e-01 3.95527452e-01 -4.11751658e-01
-1.51749954e-01 -2.94160187e-01 -5.48610270e-01 1.94041297e-01
-3.23340684e-01 6.15309060e-01 -8.58388364e-01 -5.07316053e-01
2.47646496e-01 7.65120268e-01 -4.27343577e-01 7.65127420e-01
2.50792533e-01 5.92915356e-01 -2.23047420e-01 4.40586030e-01
4.45825189e-01 2.89125532e-01 -9.50333923e-02 2.34972849e-01
-1.87794283e-01 4.91792411e-01 -1.29102874e+00 1.79805624e+00
-5.23832738e-01 1.09739053e+00 2.55412042e-01 -3.56446624e-01
1.14771843e+00 1.01666772e+00 -6.46148846e-02 -1.25156179e-01
1.01660334e-01 -1.02900468e-01 -2.02618539e-01 -9.08339620e-01
3.77750039e-01 -2.06384718e-01 1.07757491e-03 4.09792930e-01
2.92285591e-01 -5.01937926e-01 -4.51633893e-02 4.90000784e-01
9.52120304e-01 2.22985893e-01 1.33172106e-02 -7.52105862e-02
9.33720231e-01 -5.30884743e-01 3.14745754e-01 4.35242146e-01
-5.81628382e-01 6.97068751e-01 2.72768319e-01 -6.63133413e-02
-6.44273281e-01 -1.18159854e+00 2.33961269e-01 1.10712087e+00
6.93793818e-02 -3.34340662e-01 -1.14551198e+00 -7.05031693e-01
-5.27530730e-01 1.00346172e+00 -2.60527462e-01 -1.66554004e-01
-4.43617761e-01 3.18264633e-01 7.99269974e-01 1.54035345e-01
4.09038991e-01 -1.29071534e+00 -1.76371947e-01 -6.05360046e-03
-4.90685284e-01 -1.06907201e+00 -1.09705830e+00 -4.70130831e-01
-4.84117597e-01 -6.18371129e-01 -6.63768351e-01 -1.14324641e+00
5.83460569e-01 4.97606605e-01 7.73417771e-01 -1.01885416e-01
-7.87522346e-02 7.57052600e-01 -2.93401480e-01 -6.01614654e-01
-8.05860877e-01 7.63333812e-02 6.99036941e-02 5.53225959e-03
1.47669576e-02 -6.82819128e-01 -7.61387527e-01 3.18087280e-01
-8.39336276e-01 -1.48245037e-01 5.88183254e-02 6.57941163e-01
1.01678245e-01 -2.27063343e-01 8.59116316e-01 -9.05878842e-01
8.24567676e-01 -1.46515548e-01 -2.70134389e-01 1.24718830e-01
-2.21835062e-01 -4.72669870e-01 9.06349421e-01 -6.09552503e-01
-1.39910710e+00 -9.70441401e-02 -4.39559609e-01 -6.11128688e-01
-5.43867767e-01 -4.45322841e-02 -3.30218554e-01 -2.70100623e-01
6.67523324e-01 4.06494230e-01 2.62304306e-01 -1.85693242e-02
1.16239935e-01 1.19005120e+00 8.36461604e-01 -1.82043776e-01
9.21187639e-01 -2.19478142e-02 -4.23618078e-01 -9.44501936e-01
-5.82443595e-01 -2.95071989e-01 -4.09751952e-01 -7.07133830e-01
5.47206819e-01 -1.13258851e+00 -1.20542169e+00 1.57235309e-01
-9.64502931e-01 -2.35000744e-01 -2.20168978e-01 5.04658759e-01
-3.82519752e-01 3.34841520e-01 -8.30780745e-01 -6.58007145e-01
-4.41000104e-01 -1.34797299e+00 7.92063534e-01 8.57034743e-01
-3.18326831e-01 -1.17475367e+00 2.10328683e-01 9.04240251e-01
4.56149071e-01 2.19829649e-01 4.73906361e-02 -8.28299880e-01
-4.42178547e-01 -3.79630119e-01 6.96661472e-01 4.13793683e-01
3.06178987e-01 5.39623559e-01 -1.24698913e+00 -3.23637486e-01
2.32326556e-02 -2.40533292e-01 1.28969908e-01 1.17465377e-01
9.04350400e-01 -7.46802032e-01 -2.78055854e-02 5.91575205e-01
9.86559033e-01 4.73222822e-01 7.70012498e-01 4.64973114e-02
5.76214433e-01 5.88583946e-01 6.47477031e-01 1.89314872e-01
2.54737943e-01 5.18220246e-01 2.50407085e-02 1.72069386e-01
-3.06567013e-01 -4.23876494e-01 8.22743475e-01 1.36444712e+00
4.77191657e-02 -5.03590465e-01 -6.06777430e-01 3.45739245e-01
-1.25819969e+00 -1.51008666e+00 2.01600920e-02 2.06713438e+00
1.03865862e+00 1.02852419e-01 1.59583703e-01 8.78067128e-03
9.30126905e-01 1.33046195e-01 -3.88322994e-02 -8.23201418e-01
5.31201959e-01 2.74890870e-01 6.93441108e-02 9.27891374e-01
-7.51074791e-01 7.51129150e-01 6.53637266e+00 6.90403163e-01
-1.54713488e+00 -1.22229204e-01 5.54654419e-01 -4.66897666e-01
-1.98089227e-01 -4.32966232e-01 -8.62952709e-01 6.57349885e-01
1.43986642e+00 -6.83304906e-01 3.44697356e-01 9.27027345e-01
4.32689369e-01 3.42039794e-01 -8.19520593e-01 8.23668242e-01
5.43003857e-01 -1.53452933e+00 8.31828192e-02 -3.73355538e-01
7.34158993e-01 -6.92996681e-01 9.98299643e-02 4.39346939e-01
9.14789289e-02 -8.94379377e-01 7.16807365e-01 1.96384221e-01
8.18371594e-01 -1.02386367e+00 7.07116187e-01 2.67418683e-01
-1.00287163e+00 1.36904821e-01 3.00666727e-02 -1.97811574e-02
1.57129616e-01 -2.69183129e-01 -1.22574425e+00 3.41841131e-01
5.98970532e-01 2.36050591e-01 -2.24406630e-01 9.40861344e-01
-4.52922851e-01 9.15265501e-01 2.35694766e-01 -2.24510059e-01
1.22257762e-01 6.39636219e-02 7.40192831e-01 1.43462706e+00
5.48972309e-01 3.83454561e-01 1.67542502e-01 4.08928692e-01
-4.45970625e-01 1.07742637e-01 -5.50819337e-01 2.08728909e-01
8.89840305e-01 1.51603746e+00 -5.06434023e-01 -4.25602525e-01
-1.56296894e-01 8.96197915e-01 -4.01386797e-01 4.16264504e-01
-1.15925646e+00 -1.11177790e+00 6.11686826e-01 2.42038935e-01
2.55207092e-01 1.40627220e-01 3.29580843e-01 -9.57364380e-01
-1.56442523e-01 -1.18346798e+00 1.01131806e-02 -1.00287068e+00
-7.11684704e-01 8.81573260e-01 -3.73012841e-01 -1.49224699e+00
-3.89273316e-01 -1.74979135e-01 -1.37333775e+00 7.42392719e-01
-1.26407266e+00 -9.12364304e-01 -6.68144763e-01 7.23001599e-01
1.01197553e+00 -3.32618088e-01 9.78332698e-01 3.56008917e-01
-7.03171253e-01 1.03993118e+00 8.04853290e-02 1.35453030e-01
1.29386044e+00 -1.13671386e+00 8.01634118e-02 7.97222197e-01
-4.31376062e-02 4.96112525e-01 1.01895964e+00 -2.27310181e-01
-1.28454566e+00 -9.51768517e-01 5.96704006e-01 -3.94854784e-01
4.12323177e-01 -4.06467199e-01 -8.02442968e-01 6.16690695e-01
7.35713959e-01 -2.17930898e-01 1.17623854e+00 -6.27982795e-01
1.26358151e-01 -3.49009722e-01 -1.29156685e+00 7.77870953e-01
5.05745709e-01 -4.39297736e-01 -4.67597216e-01 2.03270003e-01
9.17851448e-01 -8.90502393e-01 -9.14281309e-01 -1.23193853e-01
5.68747818e-01 -9.32925999e-01 7.16797411e-01 -4.17879492e-01
6.81765735e-01 -2.11643144e-01 3.36265028e-01 -1.23438358e+00
-5.23178093e-03 -1.43514609e+00 1.49289280e-01 1.83343208e+00
2.90628731e-01 -3.01753312e-01 7.15477407e-01 6.04744494e-01
-4.76100802e-01 -4.24023122e-01 -8.67023945e-01 -4.99640197e-01
-1.90011546e-01 3.57161462e-02 5.49471617e-01 1.10351765e+00
3.93931895e-01 4.90720183e-01 -3.51941615e-01 2.46004000e-01
2.98306167e-01 8.07485264e-03 1.08399916e+00 -5.36984146e-01
-1.23311289e-01 -2.78394699e-01 -4.21932369e-01 -7.82914698e-01
2.12527424e-01 -4.88448828e-01 2.14964062e-01 -1.12526572e+00
-1.42350838e-01 3.49674731e-01 -3.24307382e-02 1.49436444e-01
-2.70270646e-01 2.55330116e-01 1.87613308e-01 -2.30916411e-01
-6.97527885e-01 3.20488185e-01 1.49904823e+00 1.54248521e-01
-4.54504877e-01 3.48401546e-01 -6.16745114e-01 6.06859565e-01
8.35553408e-01 -3.46639127e-01 -6.83854580e-01 4.17966545e-02
-7.24999368e-01 3.99220258e-01 4.24281657e-02 -1.18620563e+00
3.56093228e-01 -2.38589302e-01 2.61709064e-01 -6.42514303e-02
2.87726820e-01 -5.86863518e-01 -2.78974287e-02 4.64922279e-01
-7.83845842e-01 7.74044022e-02 3.92104000e-01 3.23095053e-01
-4.07648146e-01 -2.40263700e-01 8.00883234e-01 -5.59988879e-02
-3.27379137e-01 6.70756176e-02 -7.18580902e-01 2.93705286e-03
1.34030414e+00 -2.76150286e-01 -4.53412265e-01 -1.14835000e+00
-5.92212677e-01 2.17786863e-01 3.53134781e-01 7.08046377e-01
8.18848729e-01 -1.32287431e+00 -6.33759558e-01 1.20076343e-01
-2.69504249e-01 -5.32802939e-02 5.83521068e-01 3.66806358e-01
-9.22113299e-01 2.54593819e-01 -2.70893872e-01 -3.28259826e-01
-1.83024573e+00 3.76440942e-01 2.82902062e-01 6.29684702e-02
-4.38292354e-01 1.08083498e+00 7.97428004e-03 -4.24686909e-01
6.77945137e-01 -1.58411525e-02 -2.44952992e-01 -3.12873334e-01
9.75781441e-01 4.29643929e-01 -1.88797295e-01 -5.76044321e-01
-1.66419670e-01 1.29493728e-01 -6.25673085e-02 -3.53282154e-01
9.83427882e-01 -4.10666734e-01 5.61296046e-01 6.09492302e-01
9.66112673e-01 6.64585054e-01 -1.24687803e+00 1.80037454e-01
-5.86119831e-01 -5.93713999e-01 -2.06953168e-01 -8.27607036e-01
-9.47628021e-01 9.97874737e-01 5.99098384e-01 1.23834990e-01
9.85214114e-01 -3.41370463e-01 9.71064210e-01 2.97353268e-01
-1.51783481e-01 -8.65128040e-01 4.23157096e-01 1.53368801e-01
8.01115870e-01 -1.06876028e+00 -1.44095987e-01 -4.40750271e-01
-9.13011730e-01 1.22864282e+00 7.69195855e-01 9.41773579e-02
4.33710098e-01 3.38920504e-01 7.64320493e-01 3.31271410e-01
-1.01883423e+00 4.37802076e-01 2.86774457e-01 5.70046306e-01
8.05594981e-01 -8.45131204e-02 1.30934373e-01 5.33852100e-01
-7.84146965e-01 -1.06764972e-01 1.12196362e+00 9.61180925e-01
-4.90810096e-01 -1.12033796e+00 -5.82015157e-01 -2.93808222e-01
-9.80064869e-01 1.03554785e-01 -4.13886696e-01 6.44404888e-01
-1.55744895e-01 1.18261540e+00 -5.40821068e-02 -4.98520344e-01
4.12301630e-01 4.31359112e-01 1.24747559e-01 -7.13566422e-01
-8.72743785e-01 1.30241394e-01 1.99586600e-02 -3.24879318e-01
-2.13217318e-01 -3.80840540e-01 -1.12492430e+00 -7.75228262e-01
-2.22487167e-01 6.83547616e-01 6.30521595e-01 5.19639075e-01
4.57863539e-01 9.52563822e-01 1.06361055e+00 -7.53023267e-01
-4.89135742e-01 -1.02484632e+00 -3.09239984e-01 3.14523637e-01
5.76525092e-01 -3.74488309e-02 -5.31260788e-01 6.29738152e-01] | [14.766676902770996, 6.532095909118652] |
47217747-77cc-4f51-ad58-baf5b88cd603 | turning-a-blind-eye-explicit-removal-of | 1809.02169 | null | http://arxiv.org/abs/1809.02169v2 | http://arxiv.org/pdf/1809.02169v2.pdf | Turning a Blind Eye: Explicit Removal of Biases and Variation from Deep Neural Network Embeddings | Neural networks achieve the state-of-the-art in image classification tasks.
However, they can encode spurious variations or biases that may be present in
the training data. For example, training an age predictor on a dataset that is
not balanced for gender can lead to gender biased predicitons (e.g. wrongly
predicting that males are older if only elderly males are in the training set).
We present two distinct contributions: 1) An algorithm that can remove multiple
sources of variation from the feature representation of a network. We
demonstrate that this algorithm can be used to remove biases from the feature
representation, and thereby improve classification accuracies, when training
networks on extremely biased datasets. 2) An ancestral origin database of
14,000 images of individuals from East Asia, the Indian subcontinent,
sub-Saharan Africa, and Western Europe. We demonstrate on this dataset, for a
number of facial attribute classification tasks, that we are able to remove
racial biases from the network feature representation. | ['Christoffer Nellaker', 'Andrew Zisserman', 'Mohsan Alvi'] | 2018-09-06 | null | null | null | null | ['facial-attribute-classification'] | ['computer-vision'] | [ 5.71870387e-01 3.83913130e-01 -1.00960672e-01 -9.39039946e-01
9.53274406e-03 -3.18434060e-01 5.91910720e-01 5.19445390e-02
-7.88758516e-01 9.56316233e-01 1.69900686e-01 -1.80394843e-01
-8.72113975e-04 -1.00290692e+00 -6.10934317e-01 -5.82021713e-01
-2.29652971e-01 5.07951558e-01 -4.75085497e-01 -7.40268901e-02
1.62330925e-01 6.01277769e-01 -1.77888978e+00 2.37776667e-01
7.23850012e-01 1.04299831e+00 -7.09434509e-01 3.63727301e-01
1.87894017e-01 5.52601695e-01 -6.23941898e-01 -7.00541675e-01
2.17475682e-01 -4.82071191e-01 -6.36446536e-01 -1.77667961e-01
1.26218319e+00 -2.63384223e-01 -2.63551503e-01 9.56184864e-01
4.92877454e-01 -2.97468752e-01 1.08694565e+00 -1.40426326e+00
-6.22311771e-01 7.43105531e-01 -7.64206469e-01 -5.18971533e-02
-1.72724187e-01 -2.01491550e-01 8.34298551e-01 -7.80356169e-01
6.62733316e-01 1.56206942e+00 8.76072705e-01 8.86574626e-01
-1.50058854e+00 -1.19145358e+00 1.95428431e-01 3.13890055e-02
-1.34453201e+00 -7.65261829e-01 6.72444284e-01 -4.23339427e-01
3.39971781e-01 2.87616909e-01 8.34821224e-01 1.34696615e+00
1.31812632e-01 2.90673643e-01 1.33401334e+00 -2.18338683e-01
4.80474457e-02 7.91842490e-02 1.09944575e-01 7.32596755e-01
4.76177305e-01 3.10868979e-01 -8.63430262e-01 -4.19499278e-01
6.58563912e-01 -2.94086039e-01 5.92859760e-02 -3.37609261e-01
-9.58429277e-01 8.89421821e-01 6.33206666e-01 5.63888177e-02
-3.02926093e-01 1.34184167e-01 2.48825490e-01 4.60916430e-01
6.66111290e-01 5.09416580e-01 -4.30456072e-01 1.15121536e-01
-9.16432321e-01 3.07187736e-01 7.12126791e-01 3.74826670e-01
9.15451109e-01 2.38163948e-01 2.49311537e-01 1.08107340e+00
-2.26064511e-02 6.86342716e-01 4.19181406e-01 -1.02146649e+00
2.04470642e-02 5.73819101e-01 -2.29731560e-01 -1.20230496e+00
-7.40556657e-01 -4.15465444e-01 -1.01171422e+00 4.74123448e-01
8.10974121e-01 -4.93273914e-01 -1.16398954e+00 2.27437949e+00
1.27357274e-01 -4.28607345e-01 -9.79060531e-02 6.98129594e-01
8.13011169e-01 -1.01041883e-01 4.57999408e-01 -1.82518326e-02
1.26156390e+00 -1.07519381e-01 -1.86114162e-01 -7.26994157e-01
3.50771368e-01 -3.86049151e-01 4.56676632e-01 3.03203732e-01
-9.76179481e-01 -4.27412629e-01 -9.16346133e-01 3.45268488e-01
-5.10105371e-01 1.40760124e-01 9.15687561e-01 9.57806230e-01
-9.83773351e-01 8.12018454e-01 -3.63818526e-01 -5.30827999e-01
9.62487936e-01 8.36501896e-01 -7.72318721e-01 -1.86048105e-01
-9.90230799e-01 8.91312242e-01 1.63131312e-01 5.85874505e-02
-4.13537204e-01 -9.96613920e-01 -8.48472416e-01 -1.23738997e-01
-3.82443331e-02 -5.57287276e-01 6.49679720e-01 -1.97426617e+00
-6.32228792e-01 1.26829684e+00 -1.64561048e-01 -2.06226885e-01
5.47472835e-01 5.18868603e-02 -4.77454305e-01 -7.75695145e-02
1.13231027e-02 1.02320719e+00 1.04441273e+00 -1.19509733e+00
-5.40798485e-01 -1.06439757e+00 -4.87857848e-01 -1.02264740e-01
-5.81079960e-01 7.08771199e-02 2.61510760e-01 -6.81321561e-01
2.85614073e-01 -1.11626041e+00 -1.48824409e-01 2.89689958e-01
-2.94399112e-01 2.06757545e-01 4.22100902e-01 -8.00188422e-01
7.02765822e-01 -2.25736046e+00 2.15527877e-01 7.10837543e-01
4.19556648e-01 2.51487251e-02 -1.44561991e-01 -1.79080307e-01
-4.73982632e-01 2.62255102e-01 -2.30903327e-01 -7.80971497e-02
-3.25420290e-01 4.03536975e-01 1.71024203e-01 6.64287329e-01
4.70579416e-01 5.29760838e-01 -5.21466732e-01 -4.65702891e-01
-2.16278225e-01 4.44577515e-01 -5.11110663e-01 -1.51775718e-01
3.50580901e-01 3.59203905e-01 4.08295495e-03 6.83114231e-01
7.44501829e-01 4.16486502e-01 3.11653107e-01 -3.63568030e-02
1.88183799e-01 -1.10603578e-01 -8.70259523e-01 9.01316464e-01
-2.11002648e-01 7.56225288e-01 1.10671729e-01 -8.31089199e-01
1.12691247e+00 -1.12168610e-01 3.91646236e-01 -5.58431447e-01
4.61225003e-01 4.72751230e-01 6.18616223e-01 -8.75493065e-02
3.90675664e-01 -3.73333424e-01 2.28841584e-02 4.32157785e-01
1.92793116e-01 2.04837039e-01 1.01276468e-02 -1.64315149e-01
6.28614664e-01 -3.82258952e-01 7.10784793e-02 -4.31395888e-01
3.15612197e-01 -2.63217896e-01 8.10842335e-01 8.10450912e-01
-3.81792158e-01 7.57358015e-01 1.01792741e+00 -7.69206643e-01
-1.26112926e+00 -1.03701174e+00 -4.20964658e-01 1.35710216e+00
-7.12501585e-01 -1.30205110e-01 -5.86690664e-01 -8.38098764e-01
4.33270365e-01 3.07501495e-01 -1.32184398e+00 -5.28864622e-01
-4.96956795e-01 -1.08784699e+00 9.13059831e-01 6.11645281e-01
4.49221343e-01 -7.21839905e-01 -3.06388408e-01 -3.63568723e-01
2.28708670e-01 -8.74686301e-01 1.32425711e-01 1.24284126e-01
-7.77197242e-01 -1.20211422e+00 -8.90047371e-01 -6.44207776e-01
1.02343738e+00 -4.34302896e-01 1.30709469e+00 3.45608264e-01
-2.05548465e-01 1.91149667e-01 1.39151916e-01 -7.65960991e-01
-4.79798019e-01 3.57130051e-01 2.95588881e-01 1.69293448e-01
7.10080743e-01 -7.44405210e-01 -4.83588457e-01 3.70995790e-01
-3.86839598e-01 -1.48555264e-01 5.67809880e-01 8.26204002e-01
-6.48685545e-02 -3.70712817e-01 9.75004196e-01 -1.36100006e+00
2.75104880e-01 -4.47180629e-01 -7.57895485e-02 1.79184438e-03
-6.74906492e-01 -2.61168592e-02 2.59527117e-01 -4.89982367e-01
-8.35874200e-01 1.53389767e-01 -7.45771006e-02 7.64883161e-02
-2.58856028e-01 2.20870823e-01 1.55753149e-02 -3.28446269e-01
1.00051260e+00 -2.27816314e-01 5.36410749e-01 -2.35848203e-01
-1.90128181e-02 7.94549584e-01 5.72416425e-01 -7.33138740e-01
6.33120596e-01 3.96535724e-01 4.99337554e-01 -9.00583506e-01
-7.06079781e-01 2.11599872e-01 -1.00051844e+00 -4.58428115e-01
5.23436725e-01 -9.01166081e-01 -5.03972650e-01 7.76948690e-01
-5.94859898e-01 -7.05168024e-02 -7.86577314e-02 4.06302184e-01
-1.97505280e-01 -2.41393089e-01 -2.73203254e-01 -7.04016268e-01
-1.74273714e-01 -5.90160728e-01 5.96572578e-01 3.32233429e-01
-6.23270988e-01 -8.63887012e-01 -3.01436305e-01 8.57486110e-03
4.29253399e-01 6.29535258e-01 1.19944251e+00 -7.75308311e-01
2.89524376e-01 -1.60915717e-01 -4.60924923e-01 4.63099509e-01
1.93145424e-01 3.27892035e-01 -1.28512967e+00 -2.14313939e-01
-4.93733674e-01 -5.16703725e-01 1.32744646e+00 3.08639824e-01
1.18761146e+00 -2.17200890e-01 -2.07185328e-01 6.56840980e-01
9.62180316e-01 -9.84099954e-02 6.19707406e-01 1.23889506e-01
7.23334372e-01 1.08978438e+00 1.59904733e-01 2.99602658e-01
2.15440631e-01 1.67430341e-01 2.71482408e-01 -2.31444553e-01
1.57973054e-03 -1.11721441e-01 -4.74564731e-02 -7.12333024e-02
-5.55745184e-01 5.03750384e-01 -9.96894658e-01 5.62189579e-01
-1.55238128e+00 -7.68753648e-01 -1.80268347e-01 2.29928875e+00
8.31111610e-01 6.89470628e-03 4.82822716e-01 2.00603813e-01
7.26560950e-01 2.38532066e-01 -6.59102499e-01 -6.85867786e-01
-4.19159621e-01 4.43734974e-01 6.63099766e-01 2.17092395e-01
-1.12058115e+00 6.71069205e-01 7.05375671e+00 1.46015897e-01
-1.19811022e+00 -4.78519708e-01 1.12964678e+00 -1.86179489e-01
-1.80950359e-01 -4.02031630e-01 -4.09095526e-01 1.14690989e-01
9.67343152e-01 -1.31778046e-01 3.51557970e-01 7.50895143e-01
-3.24847639e-01 -1.83935761e-02 -1.19889224e+00 9.61158574e-01
4.84382391e-01 -7.69574285e-01 -2.51360005e-03 2.60907173e-01
6.38532460e-01 -2.72465944e-01 4.45758104e-01 1.69607103e-01
1.33538023e-01 -1.52068913e+00 6.03585064e-01 3.14262837e-01
9.91562366e-01 -1.21828222e+00 7.93796241e-01 -9.39255208e-02
-4.20939595e-01 -4.19257432e-01 -4.69907373e-01 -3.60220790e-01
-5.77203512e-01 6.41869187e-01 -5.90548754e-01 8.67282674e-02
7.40847945e-01 6.75766885e-01 -1.15827632e+00 6.66618526e-01
-7.75061026e-02 3.29944700e-01 -3.56674969e-01 9.12659913e-02
-8.87866989e-02 7.00419247e-02 6.73147291e-02 8.11430156e-01
2.81488866e-01 -1.03940524e-01 -4.26246792e-01 6.76611483e-01
-3.30381662e-01 3.76649089e-02 -7.19579101e-01 3.78350983e-03
2.27308318e-01 1.21875250e+00 -3.95530552e-01 -1.22465476e-01
-2.60365367e-01 6.17845595e-01 3.58332634e-01 3.10869664e-01
-4.11292344e-01 -2.26130888e-01 9.94009256e-01 1.42346203e-01
-2.61892620e-02 8.12317207e-02 -4.99211460e-01 -8.51487756e-01
-3.94304961e-01 -1.14751673e+00 5.72388709e-01 -6.13578558e-01
-1.45378447e+00 3.85781020e-01 -1.04110181e-01 -5.41119933e-01
-3.35373014e-01 -9.16319907e-01 -4.14079249e-01 9.80832756e-01
-1.23238575e+00 -1.21382964e+00 -4.13813412e-01 4.92762715e-01
-3.31656039e-01 -5.04800558e-01 8.59397292e-01 4.36567605e-01
-4.73002553e-01 7.50274479e-01 -1.95495754e-01 5.53043008e-01
1.05072010e+00 -1.16115630e+00 2.60048926e-01 3.70567769e-01
5.63488863e-02 6.51785195e-01 5.17362356e-01 -5.21612108e-01
-7.56878376e-01 -9.30567145e-01 1.07253838e+00 -3.32027942e-01
3.78015012e-01 -3.91267478e-01 -7.63214290e-01 9.10438240e-01
-1.96512908e-01 -1.50017347e-02 1.16102052e+00 7.46264815e-01
-7.81919837e-01 -1.80899739e-01 -1.35721731e+00 6.57376707e-01
1.20049047e+00 -4.99433607e-01 -3.49693149e-01 -9.23769176e-02
-2.69846618e-01 -2.36909449e-01 -8.69104028e-01 4.11363065e-01
1.34141374e+00 -1.17232811e+00 9.93613660e-01 -1.11633384e+00
8.35238814e-01 3.37237865e-01 -6.03570454e-02 -1.53844690e+00
-2.50400394e-01 -1.85985312e-01 1.86013415e-01 1.32666075e+00
5.24991989e-01 -7.67556131e-01 1.05569887e+00 8.90616596e-01
4.79138196e-01 -6.61237657e-01 -1.00464213e+00 -4.41048115e-01
6.45085275e-01 -9.91935730e-02 7.49658942e-01 9.62613344e-01
-4.59550172e-01 3.27784956e-01 -3.47667247e-01 -2.28827745e-01
8.07536900e-01 -2.09530905e-01 6.91820920e-01 -1.77898312e+00
3.40916395e-01 -5.29227316e-01 -7.48031616e-01 2.60206573e-02
5.42977989e-01 -7.20812380e-01 -2.58932501e-01 -8.53011787e-01
2.56070554e-01 -6.47670805e-01 -2.63586253e-01 6.88704669e-01
-1.41837373e-01 6.79601669e-01 1.53093442e-01 -3.34171504e-02
2.47018337e-01 1.78501591e-01 9.56728816e-01 -1.11142486e-01
2.87118822e-01 5.05830385e-02 -1.11722183e+00 1.04751337e+00
9.10844564e-01 -6.53064728e-01 -1.25522524e-01 -1.81837633e-01
2.79915661e-01 -5.17518997e-01 5.44146895e-01 -9.74036396e-01
-1.91403121e-01 -1.39637217e-01 1.39429402e+00 5.13923503e-02
4.09660101e-01 -7.58528888e-01 -2.76826252e-03 5.81564188e-01
-4.53919709e-01 3.59935984e-02 1.73933640e-01 9.42738131e-02
3.50048505e-02 -9.43719894e-02 1.02514541e+00 -2.88720578e-01
-5.16240239e-01 2.69355208e-01 -2.00047299e-01 2.12685093e-01
5.46135128e-01 -7.47894347e-02 -6.45771027e-01 -3.27477872e-01
-6.54993892e-01 -3.63324992e-02 6.90543056e-01 3.92382741e-01
3.72959286e-01 -1.39313674e+00 -9.90794539e-01 7.04643309e-01
1.54281318e-01 -6.03035986e-01 3.05215474e-02 6.39516592e-01
-2.94362545e-01 -1.49265349e-01 -9.43387747e-01 -3.71328771e-01
-1.64202929e+00 1.25804305e-01 6.03176057e-01 4.21734929e-01
1.60022769e-02 8.47153604e-01 1.17667224e-02 -6.07895672e-01
-6.83410913e-02 2.74668992e-01 -3.18236768e-01 6.14188254e-01
4.50643033e-01 2.32630327e-01 -1.22161828e-01 -9.29770589e-01
-4.45773780e-01 6.46900237e-01 -8.26426223e-02 -9.87659097e-02
1.46462965e+00 2.02684179e-01 -3.75301242e-01 4.62567300e-01
1.06354690e+00 -3.14038135e-02 -7.20271170e-01 6.43338189e-02
-2.37754703e-01 -7.79197514e-01 -7.74338320e-02 -7.92906344e-01
-1.42222703e+00 7.47268200e-01 7.64418542e-01 1.32998332e-01
1.01756084e+00 -1.90006390e-01 1.34499088e-01 3.77971500e-01
1.23720951e-01 -1.22941136e+00 -3.75139624e-01 3.72456938e-01
6.29813135e-01 -1.36509526e+00 3.24973494e-01 -3.26853663e-01
-5.67296207e-01 1.29791582e+00 7.39229679e-01 -1.73961699e-01
5.76135159e-01 -9.58978683e-02 3.29196841e-01 -1.02946900e-01
-5.65119922e-01 -7.42843971e-02 3.54772151e-01 1.01680577e+00
6.26799524e-01 5.93132749e-02 -4.08761173e-01 5.50728738e-01
-6.10350966e-01 -1.86552390e-01 5.51393747e-01 7.43987560e-01
-2.14672133e-01 -1.02195299e+00 -4.04177278e-01 1.03799725e+00
-6.83844626e-01 3.91995274e-02 -1.09478366e+00 9.06991720e-01
4.13072199e-01 4.37349170e-01 5.51035762e-01 -4.15438771e-01
2.28718534e-01 5.33076704e-01 8.14436197e-01 -2.49551848e-01
-7.54092455e-01 -4.91259754e-01 5.58278501e-01 -3.47488821e-01
-4.29385453e-01 -9.23052192e-01 -7.16195703e-01 -6.79493427e-01
1.80124402e-01 -1.21904261e-01 6.80825472e-01 5.28123558e-01
1.23741791e-01 3.02752584e-01 5.04167914e-01 -7.83498585e-01
-2.42682397e-01 -9.73249197e-01 -9.00060177e-01 5.96413374e-01
4.29382503e-01 -7.42856205e-01 -5.60065687e-01 -1.31159499e-01] | [13.100668907165527, 1.2901705503463745] |
68ba9ba8-2a88-4ccc-a70b-e93b6b2bc331 | exploring-simple-siamese-representation | 2011.10566 | null | https://arxiv.org/abs/2011.10566v1 | https://arxiv.org/pdf/2011.10566v1.pdf | Exploring Simple Siamese Representation Learning | Siamese networks have become a common structure in various recent models for unsupervised visual representation learning. These models maximize the similarity between two augmentations of one image, subject to certain conditions for avoiding collapsing solutions. In this paper, we report surprising empirical results that simple Siamese networks can learn meaningful representations even using none of the following: (i) negative sample pairs, (ii) large batches, (iii) momentum encoders. Our experiments show that collapsing solutions do exist for the loss and structure, but a stop-gradient operation plays an essential role in preventing collapsing. We provide a hypothesis on the implication of stop-gradient, and further show proof-of-concept experiments verifying it. Our "SimSiam" method achieves competitive results on ImageNet and downstream tasks. We hope this simple baseline will motivate people to rethink the roles of Siamese architectures for unsupervised representation learning. Code will be made available. | ['Kaiming He', 'Xinlei Chen'] | 2020-11-20 | null | http://openaccess.thecvf.com//content/CVPR2021/html/Chen_Exploring_Simple_Siamese_Representation_Learning_CVPR_2021_paper.html | http://openaccess.thecvf.com//content/CVPR2021/papers/Chen_Exploring_Simple_Siamese_Representation_Learning_CVPR_2021_paper.pdf | cvpr-2021-1 | ['self-supervised-image-classification'] | ['computer-vision'] | [ 2.66314834e-01 3.12375903e-01 -4.98531342e-01 -5.88625669e-01
-4.83073860e-01 -4.25815821e-01 7.80456245e-01 -9.64278262e-03
-6.80370986e-01 6.06747270e-01 4.34948623e-01 -3.08849126e-01
-3.93384881e-02 -2.67457545e-01 -9.11825895e-01 -6.91519082e-01
-3.75918329e-01 4.36942011e-01 1.22767061e-01 -2.63338238e-01
2.74014682e-01 3.83696616e-01 -1.33094776e+00 4.20616597e-01
5.89599133e-01 6.19376540e-01 2.07359269e-01 4.43198383e-01
-4.99915518e-02 1.13040566e+00 -4.19291973e-01 -3.08120519e-01
4.76667702e-01 -6.43839896e-01 -8.50569189e-01 7.45597109e-02
7.54908144e-01 -2.21512780e-01 -3.23979199e-01 1.15122819e+00
2.17311978e-01 4.08299297e-01 9.57846284e-01 -1.50533736e+00
-8.69494438e-01 8.82309794e-01 -7.76269972e-01 4.07939166e-01
-3.64195615e-01 3.64667445e-01 1.46572483e+00 -7.61879921e-01
9.05771792e-01 1.12799263e+00 5.51223576e-01 6.60083830e-01
-1.48179412e+00 -5.56969583e-01 4.52380270e-01 -6.58308119e-02
-1.27801156e+00 -5.98097920e-01 4.93744224e-01 -3.82767707e-01
1.34279466e+00 7.35569224e-02 5.35738409e-01 1.20244515e+00
2.74627395e-02 8.42992723e-01 7.63460219e-01 -3.65465581e-01
7.07108453e-02 2.64842898e-01 1.37204871e-01 8.09019268e-01
3.20275456e-01 2.02404708e-01 -5.15780330e-01 -2.22607423e-02
9.38050807e-01 -4.55962159e-02 -3.06261573e-02 -6.31751955e-01
-1.00348234e+00 1.16600561e+00 1.06827378e+00 2.53377169e-01
-1.47635937e-01 3.98872346e-01 4.36938643e-01 4.85530764e-01
4.01794553e-01 9.23005700e-01 -1.89354748e-01 1.36557177e-01
-8.77762556e-01 1.29308477e-01 4.39596385e-01 8.67720068e-01
7.33126223e-01 3.05609852e-01 -1.31622151e-01 9.19588804e-01
1.52526006e-01 1.81835762e-03 7.08799183e-01 -1.20257246e+00
6.10084474e-01 3.17859411e-01 -1.52983457e-01 -8.39429796e-01
-2.68657982e-01 -5.67774773e-01 -8.02583098e-01 3.46569926e-01
4.75231647e-01 -1.27171502e-01 -1.15949023e+00 2.08936262e+00
-3.63981843e-01 3.43947038e-02 8.94400179e-02 1.03343594e+00
5.12350976e-01 5.42423546e-01 2.91224509e-01 1.36043295e-01
1.01008904e+00 -9.93320525e-01 -3.71387392e-01 -3.74442846e-01
8.64080608e-01 -4.62063640e-01 1.21326971e+00 1.50191039e-01
-1.09406269e+00 -3.25834423e-01 -1.19804978e+00 -3.05724472e-01
-3.04421276e-01 2.51356006e-01 8.68278682e-01 2.69560218e-01
-1.11723459e+00 7.67367899e-01 -8.67998004e-01 -5.88986933e-01
7.37844110e-01 5.31926930e-01 -4.59584266e-01 9.30925757e-02
-7.83056378e-01 9.02669847e-01 2.80605793e-01 2.88247000e-02
-1.02396965e+00 -5.78755438e-01 -8.95602763e-01 2.04334706e-01
6.19864166e-02 -6.24451637e-01 1.07564998e+00 -1.64296019e+00
-9.61183369e-01 1.16566598e+00 -2.07666606e-01 -9.49386299e-01
4.11479414e-01 -3.00229967e-01 -4.15486097e-02 2.20980093e-01
2.21392959e-01 1.26322091e+00 9.77261782e-01 -1.30745280e+00
-1.83729857e-01 -1.74295187e-01 1.23338811e-02 3.41479242e-01
-4.27940786e-01 2.12973028e-01 -2.25651130e-01 -7.33382106e-01
1.11182742e-01 -9.70310807e-01 -5.16816974e-01 1.68305710e-01
-5.79363048e-01 -3.10637921e-01 4.86952692e-01 -3.02257359e-01
8.36298168e-01 -2.35945511e+00 3.75031978e-02 3.83753568e-01
3.84640962e-01 1.43348321e-01 -5.55908799e-01 1.27493888e-01
-4.01252985e-01 3.68568510e-01 -3.89863431e-01 -5.59234560e-01
-5.84122948e-02 2.84234315e-01 -6.00359917e-01 5.09531438e-01
6.70562923e-01 1.00175178e+00 -7.50294387e-01 -3.27021390e-01
-2.37301767e-01 3.12326282e-01 -6.51900351e-01 4.37916182e-02
-2.31780842e-01 -2.07503419e-02 -2.07573082e-02 2.48681128e-01
4.25870806e-01 -5.91917992e-01 1.94138899e-01 -1.11915059e-01
-4.61048633e-03 4.78337705e-01 -8.99397731e-01 1.73426700e+00
7.81089067e-02 8.80606472e-01 -1.40212541e-02 -1.14538670e+00
5.93121946e-01 7.48745650e-02 1.45476311e-01 -6.68980062e-01
1.28648236e-01 1.69678017e-01 4.66530591e-01 -2.38248155e-01
4.75875109e-01 -2.61065453e-01 3.17339718e-01 6.32613301e-01
3.76167148e-01 5.80843762e-02 4.71755117e-01 8.23033273e-01
9.06663775e-01 -2.67420951e-02 -2.59703472e-02 -4.35648978e-01
-6.97007924e-02 7.35216588e-02 4.81621087e-01 1.03398073e+00
-2.45788336e-01 1.05736303e+00 8.76766205e-01 -2.92038530e-01
-1.27687716e+00 -1.22119164e+00 1.83721885e-01 1.15736616e+00
-4.68404964e-02 -5.11004269e-01 -4.45768952e-01 -7.34417140e-01
-1.36653874e-02 5.14043391e-01 -8.76116514e-01 -3.71447772e-01
-5.63751519e-01 -7.30831861e-01 6.83220387e-01 9.59193051e-01
1.77078485e-01 -1.10352588e+00 -3.51707131e-01 -1.14538074e-01
8.11374933e-02 -8.58907521e-01 -5.54470003e-01 8.62161338e-01
-1.11271298e+00 -8.59866142e-01 -7.39157498e-01 -9.50929344e-01
9.97033238e-01 3.83625001e-01 1.05487144e+00 2.60492533e-01
-1.76978633e-01 6.95693642e-02 -4.18970995e-02 -2.26498723e-01
-1.72737807e-01 2.44637638e-01 -1.93898547e-02 -3.41433674e-01
3.52839112e-01 -6.92811906e-01 -5.59799671e-01 9.08399373e-02
-8.97876322e-01 -5.96551336e-02 6.95385039e-01 8.31483006e-01
4.89615262e-01 -4.32035029e-01 4.26930696e-01 -1.02034032e+00
7.39737988e-01 -5.82527399e-01 -4.52593178e-01 -2.80580875e-02
-5.92026472e-01 4.55852926e-01 6.99607551e-01 -5.51803768e-01
-7.80965745e-01 1.53724760e-01 9.48295593e-02 -7.61599898e-01
-2.83314817e-04 4.74790365e-01 1.67387202e-01 8.02753866e-02
1.06637239e+00 9.00164694e-02 3.34687948e-01 -5.07117987e-01
5.22230387e-01 2.84837857e-02 4.40308034e-01 -6.61042333e-01
7.33116627e-01 6.47582114e-01 -2.96451956e-01 -8.37159038e-01
-8.60631108e-01 -3.90760303e-01 -4.03305203e-01 3.18381786e-01
8.16239536e-01 -1.09983635e+00 -2.23975748e-01 -2.87317969e-02
-1.02326775e+00 -5.54512918e-01 -7.03381777e-01 5.54926157e-01
-5.20179033e-01 1.17211506e-01 -8.56870592e-01 -5.12400031e-01
-1.14744483e-02 -9.51694250e-01 5.35339296e-01 3.65290433e-01
-3.87641609e-01 -7.79894173e-01 1.19610116e-01 1.82593405e-01
4.29606348e-01 -3.83180648e-01 8.65244746e-01 -7.89805293e-01
-6.15617335e-01 3.52220163e-02 -5.25397539e-01 5.64665914e-01
-1.44639060e-01 1.65420264e-01 -1.05216086e+00 -3.47374201e-01
-3.83361250e-01 -7.70316124e-01 1.46923733e+00 5.40447950e-01
1.26423693e+00 -3.52594256e-01 -1.04592793e-01 9.28044856e-01
1.31311631e+00 -4.03867736e-02 7.67874718e-01 2.93563396e-01
7.18378007e-01 6.51594102e-01 -3.68432850e-02 5.78994565e-02
1.10112973e-01 3.20990950e-01 3.84169400e-01 -3.62484515e-01
-2.39548862e-01 -4.36604559e-01 4.25669402e-01 6.33221626e-01
-1.37228802e-01 -1.34921983e-01 -6.89214885e-01 8.40565860e-01
-1.77545917e+00 -1.03693616e+00 9.31189805e-02 2.11335993e+00
7.69531012e-01 4.13110614e-01 3.10990304e-01 -3.74163836e-01
5.00380278e-01 4.39081877e-01 -6.63137197e-01 -4.37014401e-01
-4.97499764e-01 3.27270716e-01 6.17751241e-01 5.32097995e-01
-1.17798734e+00 1.15321302e+00 6.77513361e+00 5.79200745e-01
-1.16392159e+00 -4.90519255e-02 6.41383648e-01 -3.07945609e-01
-3.62317741e-01 1.65481299e-01 -6.28842235e-01 1.33612961e-01
7.20398009e-01 5.04071917e-03 2.88562804e-01 8.47500324e-01
4.29668054e-02 -3.86006571e-02 -1.12171555e+00 7.24678338e-01
1.10874504e-01 -1.57612610e+00 1.60563394e-01 4.04542647e-02
9.24640656e-01 7.28504777e-01 2.73708552e-01 2.18386427e-01
6.86785936e-01 -1.25895047e+00 5.92656255e-01 2.95855582e-01
4.91378546e-01 -4.39730853e-01 3.39517951e-01 7.56899789e-02
-7.93470204e-01 -2.30258163e-02 -5.26912391e-01 -1.00185901e-01
-1.00995168e-01 2.33718500e-01 -6.75065041e-01 4.78616208e-02
5.29573977e-01 8.70109737e-01 -7.90220201e-01 1.17347705e+00
-4.56074744e-01 8.41813982e-01 -4.76362556e-01 8.38078409e-02
5.29996693e-01 -1.06950864e-01 5.07588148e-01 1.22757399e+00
-2.83204317e-02 -3.11718166e-01 -4.65307795e-02 1.30221915e+00
-3.90628338e-01 -8.26299563e-03 -8.25911999e-01 -4.49460775e-01
2.09261943e-03 9.16980088e-01 -1.03487337e+00 -3.91688079e-01
-4.25524056e-01 9.99346077e-01 6.60265625e-01 7.45234549e-01
-5.14867485e-01 -5.09073496e-01 7.26407230e-01 1.85756788e-01
4.58488762e-01 -2.21716106e-01 -5.52550852e-01 -1.41297972e+00
-3.71216424e-02 -7.02877641e-01 3.63312066e-01 -8.79530787e-01
-1.44503820e+00 3.75679910e-01 -1.10213481e-01 -9.71360385e-01
-2.08230078e-01 -6.40288413e-01 -8.27697277e-01 6.88839138e-01
-1.37726712e+00 -8.43973339e-01 1.95584565e-01 6.06907487e-01
4.35477883e-01 -2.55627096e-01 6.25472784e-01 1.02575161e-01
-6.99132323e-01 6.91997230e-01 1.31439999e-01 4.56827134e-01
6.40775979e-01 -1.26184320e+00 3.64246577e-01 9.89597917e-01
7.84246862e-01 1.05056059e+00 7.82250583e-01 -4.91542250e-01
-7.54985988e-01 -8.31547618e-01 8.70971978e-01 -2.96533614e-01
6.44383788e-01 -4.98988688e-01 -1.07003081e+00 1.09508979e+00
4.02083248e-01 3.14547345e-02 5.09154379e-01 4.74007875e-01
-8.47265065e-01 1.88329834e-02 -7.87258863e-01 6.72757268e-01
1.09250450e+00 -6.29081666e-01 -5.95249593e-01 3.81369799e-01
6.48801088e-01 7.22567067e-02 -1.97198465e-01 1.12681918e-01
4.54926431e-01 -9.41524923e-01 9.03022051e-01 -1.20531273e+00
8.73112321e-01 -1.53646544e-01 -2.50251442e-02 -1.28724837e+00
-2.93723136e-01 -4.62593973e-01 -8.03885702e-03 8.71329427e-01
8.14060211e-01 -5.60052991e-01 1.11429954e+00 5.45736194e-01
-1.16514958e-01 -6.57291293e-01 -6.19919121e-01 -1.04533875e+00
3.56456727e-01 -2.44967282e-01 3.97004522e-02 1.06344759e+00
-3.21246721e-02 6.39473200e-01 -3.28095973e-01 -1.42377689e-01
6.09752417e-01 -1.79952353e-01 6.01987660e-01 -1.00893700e+00
-2.61589766e-01 -7.12356269e-01 -3.34794104e-01 -1.10112917e+00
1.53689831e-01 -1.11182988e+00 4.00601048e-03 -1.45060384e+00
4.29998398e-01 -4.18225795e-01 -5.65046668e-01 7.65662193e-01
-1.06714619e-02 3.96768540e-01 3.85044754e-01 3.98449361e-01
-7.09578633e-01 5.94578505e-01 1.04706383e+00 -2.89001167e-01
-2.40076512e-01 -2.81381965e-01 -1.06542325e+00 6.82995319e-01
9.23599720e-01 -6.73343241e-01 -7.20564485e-01 -6.52621627e-01
2.83991992e-01 -6.90911829e-01 4.95990515e-01 -8.31311464e-01
1.94733560e-01 -9.34191123e-02 4.93622303e-01 -1.54106587e-01
4.78516400e-01 -5.21879375e-01 -4.59797114e-01 3.72023493e-01
-8.28451097e-01 5.50209507e-02 2.33895138e-01 4.55684215e-01
-2.59837449e-01 -3.05612892e-01 9.42555606e-01 -3.00271243e-01
-7.34287083e-01 1.67747498e-01 -4.33483213e-01 4.01842028e-01
5.67690790e-01 -2.12404177e-01 -4.86980110e-01 -4.72545773e-01
-8.12122583e-01 2.07114160e-01 4.26122099e-01 3.53255481e-01
5.59482992e-01 -1.20545077e+00 -7.41390169e-01 4.17445809e-01
4.44056578e-02 -2.71219462e-01 -6.08766451e-02 8.45133007e-01
-3.96678478e-01 2.35705808e-01 -3.29553902e-01 -5.16604185e-01
-1.03840041e+00 3.23564887e-01 4.05605674e-01 -2.17276588e-01
-6.57865345e-01 1.36851108e+00 5.48210979e-01 -1.98741034e-01
4.14866179e-01 -3.04499656e-01 8.06719139e-02 1.11794211e-01
4.59962964e-01 -1.51949093e-01 -1.98270634e-01 -5.15492857e-01
-2.94921666e-01 6.93903789e-02 -5.93912482e-01 -2.93822378e-01
1.46250463e+00 1.03356987e-01 3.53453569e-02 5.25905371e-01
1.40200865e+00 -2.69865185e-01 -1.39223945e+00 -1.71871468e-01
-3.72121371e-02 -2.52161860e-01 -2.38452926e-01 -5.29428661e-01
-1.18425453e+00 1.03111768e+00 4.50512856e-01 5.15512116e-02
8.26197863e-01 2.89712071e-01 4.09447402e-01 5.66511631e-01
-1.23285964e-01 -1.08728158e+00 2.13475734e-01 5.83638310e-01
8.73420238e-01 -1.30546653e+00 6.52194098e-02 -1.28364181e-02
-8.43571365e-01 7.52278268e-01 8.68766725e-01 -6.47249937e-01
4.86186057e-01 1.49671718e-01 -1.48418928e-02 -3.83713990e-01
-9.74447787e-01 -4.17521954e-01 2.20103934e-01 6.49562657e-01
6.29736602e-01 -1.64774597e-01 -1.65247738e-01 2.79390395e-01
-1.50612995e-01 -2.86037683e-01 2.99032927e-01 8.57769608e-01
-3.69695187e-01 -8.94486547e-01 -6.30400516e-03 7.19456792e-01
-2.67760724e-01 -4.34460610e-01 -6.00862384e-01 9.67976332e-01
-9.61130038e-02 4.90808278e-01 3.25942129e-01 -8.41929018e-02
1.43302456e-01 1.40176594e-01 6.61389470e-01 -7.54504502e-01
-5.90139747e-01 -2.76652854e-02 -2.90416051e-02 -5.94341099e-01
-2.98621476e-01 -5.09355962e-01 -1.31987262e+00 -1.41543880e-01
-2.47404709e-01 1.40761569e-01 3.25063318e-01 8.29758763e-01
4.34515357e-01 2.51475781e-01 2.18454152e-01 -7.72441149e-01
-7.02803969e-01 -9.71593142e-01 -7.59826362e-01 6.12469554e-01
3.65975559e-01 -5.21863043e-01 -6.20058954e-01 2.88274914e-01] | [9.314146041870117, 2.7761447429656982] |
f7a7f622-af02-4b4b-a8a0-9bb26bfd0a91 | pidnet-an-efficient-network-for-dynamic | 2009.00312 | null | https://arxiv.org/abs/2009.00312v1 | https://arxiv.org/pdf/2009.00312v1.pdf | PIDNet: An Efficient Network for Dynamic Pedestrian Intrusion Detection | Vision-based dynamic pedestrian intrusion detection (PID), judging whether pedestrians intrude an area-of-interest (AoI) by a moving camera, is an important task in mobile surveillance. The dynamically changing AoIs and a number of pedestrians in video frames increase the difficulty and computational complexity of determining whether pedestrians intrude the AoI, which makes previous algorithms incapable of this task. In this paper, we propose a novel and efficient multi-task deep neural network, PIDNet, to solve this problem. PIDNet is mainly designed by considering two factors: accurately segmenting the dynamically changing AoIs from a video frame captured by the moving camera and quickly detecting pedestrians from the generated AoI-contained areas. Three efficient network designs are proposed and incorporated into PIDNet to reduce the computational complexity: 1) a special PID task backbone for feature sharing, 2) a feature cropping module for feature cropping, and 3) a lighter detection branch network for feature compression. In addition, considering there are no public datasets and benchmarks in this field, we establish a benchmark dataset to evaluate the proposed network and give the corresponding evaluation metrics for the first time. Experimental results show that PIDNet can achieve 67.1% PID accuracy and 9.6 fps inference speed on the proposed dataset, which serves as a good baseline for the future vision-based dynamic PID study. | ['Jiming Chen', 'Jingchen Sun', 'Jiayuan Fan', 'Tao Chen', 'Shibo He'] | 2020-09-01 | null | null | null | null | ['feature-compression'] | ['computer-vision'] | [ 2.79058442e-02 -8.58448267e-01 7.30479658e-02 -9.74653885e-02
-3.08577474e-02 -2.37635657e-01 2.83543080e-01 -2.60203093e-01
-7.37529933e-01 3.92698318e-01 -3.99190575e-01 -3.17640632e-01
2.93701440e-01 -1.06455600e+00 -6.12638652e-01 -7.94772029e-01
2.10073479e-02 1.28423586e-01 1.05818081e+00 5.02499798e-03
-5.29944152e-03 6.82740211e-01 -1.48101664e+00 1.52050406e-01
5.60538709e-01 1.22814703e+00 1.45212278e-01 8.11758220e-01
2.70388037e-01 5.60875833e-01 -9.34242964e-01 -3.89294595e-01
4.31749284e-01 -1.47652798e-04 -1.11243173e-01 7.10059032e-02
4.84502405e-01 -8.78701091e-01 -9.05583858e-01 1.02314734e+00
5.80793619e-01 2.49566481e-01 3.69024575e-01 -1.67351890e+00
-2.13862494e-01 -7.83504993e-02 -7.33105004e-01 1.03443122e+00
1.56069562e-01 7.45441914e-01 3.33520979e-01 -4.92018193e-01
2.56045252e-01 1.41466987e+00 6.64353192e-01 6.66831374e-01
-3.29566866e-01 -8.15846145e-01 1.73864722e-01 7.93284118e-01
-1.49243915e+00 -3.82668912e-01 8.09419632e-01 -1.84962094e-01
8.79923403e-01 2.89844245e-01 7.04318881e-01 1.23614264e+00
3.12012553e-01 8.00310671e-01 3.86564732e-01 -4.80064489e-02
4.19867039e-03 -1.27413556e-01 3.56587142e-01 8.31248641e-01
7.82863319e-01 4.31037694e-01 -6.13942072e-02 -6.46160468e-02
7.08321691e-01 2.49124601e-01 -2.21866027e-01 1.05218098e-01
-9.91624951e-01 7.32154071e-01 4.68114078e-01 5.69326207e-02
-2.22750962e-01 5.31035252e-02 6.23131871e-01 1.14066795e-01
1.23258894e-02 -2.59917736e-01 -1.48284376e-01 -1.25708804e-01
-5.19398868e-01 3.45552385e-01 5.12996197e-01 9.57583725e-01
4.18596864e-01 2.59149671e-01 -1.34210244e-01 5.90990663e-01
4.41294104e-01 8.44269097e-01 3.70915979e-01 -6.88582122e-01
5.49887359e-01 6.67284071e-01 3.72434668e-02 -1.64097321e+00
-4.54309970e-01 2.66670454e-02 -9.87535119e-01 1.37969300e-01
5.18070221e-01 -4.97690976e-01 -7.92630792e-01 1.34617531e+00
5.14606893e-01 4.39993322e-01 -1.17756389e-01 1.04701006e+00
1.05939090e+00 8.42447042e-01 3.02153286e-02 -7.73631781e-02
1.65105402e+00 -9.59828198e-01 -4.88143772e-01 -2.36369804e-01
1.27660081e-01 -5.56380510e-01 4.78109062e-01 2.67893404e-01
-7.30408192e-01 -9.33973849e-01 -1.19850349e+00 2.89074898e-01
-5.39406776e-01 2.22519174e-01 4.84892428e-01 9.19242024e-01
-7.90024340e-01 -7.21335486e-02 -7.71801233e-01 -4.14071769e-01
5.94879806e-01 2.98289418e-01 2.45605339e-03 -8.13891590e-02
-1.38613272e+00 4.85651463e-01 5.27524769e-01 2.83453196e-01
-1.15823793e+00 -3.36681038e-01 -8.23288918e-01 1.96661353e-02
5.99325240e-01 -7.69434631e-01 9.21203732e-01 -7.71112084e-01
-1.07651877e+00 4.93847728e-01 -1.60150141e-01 -7.30547369e-01
6.44479036e-01 -1.33924633e-01 -6.90450132e-01 4.57794845e-01
5.68735935e-02 7.68117368e-01 1.11855197e+00 -1.03097498e+00
-1.29863346e+00 -3.61928821e-01 3.16567808e-01 1.63868025e-01
-4.17140186e-01 3.87966990e-01 -8.84237289e-01 -5.65932453e-01
-4.60861146e-01 -8.28693748e-01 -3.18659127e-01 4.35787201e-01
-4.28678185e-01 -2.67716110e-01 1.48331630e+00 -6.09380007e-01
1.37424767e+00 -2.06224751e+00 -4.14936066e-01 1.47159956e-02
4.97700214e-01 1.14663959e+00 -4.27380353e-02 -3.80903780e-01
2.82177329e-01 -2.00379655e-01 5.67741916e-02 2.86829770e-02
-4.49253589e-01 1.52969927e-01 -9.26769674e-02 3.46342236e-01
1.69735953e-01 7.52613842e-01 -6.68585241e-01 -4.80863363e-01
5.62756538e-01 4.26621228e-01 -2.71093756e-01 2.64342874e-01
9.69584063e-02 2.60656234e-02 -4.94189054e-01 1.00125694e+00
9.97261643e-01 2.04941526e-01 -6.17687404e-01 -3.21083099e-01
-2.43590906e-01 -3.33317220e-01 -1.29992282e+00 8.85270894e-01
9.39408690e-02 7.05888927e-01 6.55043051e-02 -1.00596786e+00
9.34575319e-01 2.07988545e-01 5.20961404e-01 -6.37299120e-01
3.62384200e-01 -1.19001046e-01 7.09095076e-02 -6.38490498e-01
3.92084628e-01 4.68138516e-01 -2.32435595e-02 3.87223884e-02
-2.59915620e-01 4.88653809e-01 3.76898050e-01 -8.82498473e-02
1.24065077e+00 -5.26285112e-01 2.03233019e-01 1.33322291e-02
9.20787632e-01 9.30173695e-02 9.91212368e-01 8.29083323e-01
-9.77258325e-01 2.80955642e-01 1.79529205e-01 -1.14886308e+00
-8.70089471e-01 -1.02829254e+00 -8.75269249e-02 5.72153628e-01
5.63755155e-01 -1.75674111e-01 -7.93401182e-01 -7.56645620e-01
-2.99659204e-02 3.28063428e-01 -2.30904788e-01 -4.28589255e-01
-8.38235915e-01 -1.02461755e+00 7.60735989e-01 5.55267453e-01
1.30121064e+00 -1.05845213e+00 -1.10408783e+00 2.10303694e-01
-3.31635952e-01 -1.35435367e+00 -5.98631263e-01 -5.09855211e-01
-3.68517935e-01 -1.44015229e+00 -6.28302336e-01 -8.05911362e-01
5.64598620e-01 1.22946477e+00 6.53920352e-01 1.79170638e-01
-6.11998975e-01 4.48490947e-01 -2.13833690e-01 -7.15586722e-01
-9.40410569e-02 -2.28029728e-01 2.70465165e-01 2.48839930e-01
7.47633696e-01 -2.63960660e-01 -7.25973547e-01 6.89201593e-01
-8.07407796e-01 -5.45398667e-02 4.21518624e-01 5.79097331e-01
1.87069714e-01 6.02737486e-01 2.75957793e-01 -1.26330152e-01
4.07964408e-01 -4.12548751e-01 -8.11062753e-01 6.36185408e-02
7.11402074e-02 -7.28058577e-01 7.01668501e-01 -5.51772714e-01
-1.00020206e+00 -1.02149434e-01 -4.64433692e-02 -5.98490179e-01
-5.72482586e-01 -5.08387163e-02 -4.85644251e-01 -2.82959312e-01
5.68495810e-01 2.97279805e-01 1.53951973e-01 3.04004047e-02
-1.70104325e-01 7.63898015e-01 6.91211402e-01 -1.81530550e-01
9.39089715e-01 7.59612501e-01 8.63722786e-02 -1.18425930e+00
-4.34049517e-01 -5.35623431e-01 -4.42601532e-01 -5.09020627e-01
1.03179371e+00 -1.02389956e+00 -1.13400197e+00 9.83519435e-01
-1.40820026e+00 1.72313794e-01 2.67561495e-01 2.82110453e-01
-9.99906883e-02 7.18325138e-01 -5.69345176e-01 -7.16984451e-01
-7.16028512e-01 -1.26023936e+00 6.95604384e-01 7.01434076e-01
3.71998221e-01 -7.08290160e-01 -2.87867576e-01 5.38777769e-01
4.48387742e-01 2.97710419e-01 3.37912202e-01 -3.17399204e-01
-8.54066432e-01 -3.66644144e-01 -7.22234607e-01 2.89466232e-01
1.23104870e-01 3.54844213e-01 -8.29031408e-01 -3.81024748e-01
-7.87353963e-02 1.78029910e-01 1.01651752e+00 5.51494658e-01
1.20648837e+00 -3.09401244e-01 -5.81431329e-01 7.88979828e-01
1.08599424e+00 9.22339201e-01 6.82414114e-01 6.22787952e-01
9.74399090e-01 1.96928635e-01 6.86934412e-01 5.32832026e-01
5.01165688e-01 8.59269738e-01 5.35219550e-01 -2.26373021e-02
-1.83597445e-01 2.00879321e-01 6.97263360e-01 3.26505184e-01
-2.83982940e-02 -4.34418797e-01 -9.49631572e-01 4.03427452e-01
-1.90786922e+00 -1.23069215e+00 -2.19306812e-01 2.07373381e+00
2.73883373e-01 3.45819682e-01 3.95270944e-01 5.79838008e-02
9.69296038e-01 1.60854936e-01 -6.71274364e-01 -6.58230484e-02
-5.75886257e-02 -3.03046405e-01 3.39146614e-01 2.78398097e-01
-1.73358822e+00 8.87926519e-01 4.85393810e+00 8.79431546e-01
-1.18789494e+00 -1.11509807e-01 7.37784624e-01 2.43308991e-02
5.65168619e-01 -5.36116302e-01 -1.45711374e+00 8.95036161e-01
8.51932049e-01 -1.10697508e-01 2.06495896e-01 8.93608153e-01
3.42857003e-01 -1.19795501e-01 -4.61515367e-01 1.17942667e+00
3.10504138e-01 -1.02023590e+00 3.12052667e-01 -1.71525441e-02
2.04883486e-01 -2.26664305e-01 -1.21739227e-02 2.16508180e-01
-3.82514894e-02 -5.63130438e-01 4.00219858e-01 3.94637674e-01
6.02626204e-01 -1.04074407e+00 8.41119409e-01 1.97504982e-01
-1.57529271e+00 -4.10727441e-01 -7.22593665e-01 3.77670601e-02
2.48885110e-01 5.04980266e-01 -5.72275698e-01 4.15123016e-01
9.34364736e-01 7.77999580e-01 -7.92223036e-01 1.51075256e+00
8.16877484e-02 4.62844700e-01 -5.67201614e-01 -1.17352724e-01
3.60778630e-01 -3.26379202e-02 1.01524794e+00 1.23079681e+00
2.89033413e-01 2.03268714e-02 4.34622049e-01 5.32981873e-01
2.52166033e-01 -3.61489326e-01 -8.14684212e-01 4.87529278e-01
4.80688840e-01 1.30854404e+00 -6.85792208e-01 -4.58476603e-01
-6.53828382e-01 8.91034186e-01 -1.54039621e-01 2.40846008e-01
-1.31586564e+00 -4.93642896e-01 1.07778668e+00 -7.38870651e-02
5.40145576e-01 -4.53826904e-01 -9.51642096e-02 -1.13801503e+00
1.54967546e-01 -6.69522345e-01 4.89167511e-01 -4.46598262e-01
-1.06863391e+00 5.92710733e-01 4.32995856e-02 -1.38491130e+00
5.12023605e-02 -9.06070709e-01 -1.01294363e+00 4.10895973e-01
-1.47011626e+00 -1.07033825e+00 -9.69518900e-01 8.46712649e-01
7.61746705e-01 -5.20452440e-01 4.09860104e-01 7.12344766e-01
-1.32035077e+00 8.01941991e-01 -4.12514091e-01 3.98594767e-01
4.57422942e-01 -7.38071859e-01 7.26836860e-01 1.34680367e+00
-2.98633754e-01 2.25985408e-01 2.82948136e-01 -7.89760709e-01
-1.33994472e+00 -1.45142293e+00 4.50414240e-01 -2.83724695e-01
3.64302814e-01 -1.89704493e-01 -7.10191190e-01 3.55230868e-01
-1.41703084e-01 2.15877593e-01 4.21067894e-01 -6.58978462e-01
-1.10609727e-02 -4.29018438e-01 -1.24989080e+00 8.04447174e-01
1.15618420e+00 1.10057279e-01 -2.58160651e-01 1.90777168e-01
6.50801480e-01 -3.57346892e-01 -3.49419087e-01 4.47725266e-01
4.94551539e-01 -1.07130754e+00 1.35842693e+00 -3.57620686e-01
-1.02086961e-01 -6.87511921e-01 3.62102762e-02 -7.36452460e-01
-4.57615316e-01 -4.29901332e-01 -4.26916540e-01 1.14344299e+00
-2.14300141e-01 -6.05079770e-01 7.60416210e-01 5.73060036e-01
-1.04447998e-01 -5.90991855e-01 -1.21332896e+00 -7.27617800e-01
-5.57068110e-01 -4.13784087e-01 4.66627240e-01 4.59361553e-01
-6.54168308e-01 1.91486016e-01 -4.15165365e-01 5.32029867e-01
7.70471871e-01 -4.25392747e-01 1.02723205e+00 -1.17559052e+00
1.60071343e-01 -5.17653644e-01 -9.43739295e-01 -1.05765545e+00
-3.10353845e-01 -1.67711109e-01 -2.18295664e-01 -1.01730406e+00
1.39540091e-01 -2.37994179e-01 -4.31448966e-02 3.47243816e-01
-3.98444057e-01 9.97847021e-02 3.56085569e-01 6.05596490e-02
-7.79405057e-01 4.39881504e-01 9.76294577e-01 -4.49361533e-01
-1.70939192e-01 4.41760480e-01 -4.41437632e-01 9.30446446e-01
1.04158854e+00 -1.27364516e-01 -2.93925226e-01 -3.35819006e-01
-6.08452380e-01 -1.27098978e-01 7.42124081e-01 -1.67045951e+00
5.27904570e-01 -7.57481307e-02 7.03676283e-01 -8.05581748e-01
3.81197363e-01 -9.79087293e-01 -1.59148768e-01 7.91981101e-01
4.43639636e-01 2.95462698e-01 5.42482018e-01 6.03224933e-01
-7.58699421e-03 -2.14572310e-01 8.73556137e-01 -5.51276058e-02
-1.32384229e+00 7.46021628e-01 -8.02756965e-01 -1.42838717e-01
1.45672631e+00 -4.16836351e-01 -5.13737261e-01 -1.54446229e-01
-1.93007588e-01 2.84139067e-01 1.25448763e-01 6.57963574e-01
1.16202092e+00 -1.36314416e+00 -5.94867110e-01 6.01280034e-01
-1.36468694e-01 1.79917470e-01 4.06253040e-01 5.10069072e-01
-7.74205863e-01 6.15700364e-01 -4.98666555e-01 -6.48333430e-01
-1.60299218e+00 6.59284890e-01 4.50772583e-01 -1.80355981e-01
-6.42092109e-01 6.56387627e-01 2.61748642e-01 3.77802327e-02
4.21939015e-01 -1.34886071e-01 -4.38142061e-01 -5.02599403e-02
1.07999945e+00 7.10794151e-01 -2.89005578e-01 -8.31122875e-01
-3.47456813e-01 3.85025501e-01 -4.48101848e-01 3.04696292e-01
8.21072519e-01 -5.71189336e-02 7.49451071e-02 -1.19000964e-01
8.86257291e-01 -5.53515494e-01 -1.43287301e+00 -2.33156800e-01
-4.29562986e-01 -6.41984165e-01 -2.36455295e-02 -4.21624303e-01
-1.32073689e+00 7.88400590e-01 8.27848494e-01 4.69873147e-03
1.16341126e+00 -5.61585903e-01 1.42306721e+00 5.70129693e-01
4.57912683e-01 -8.91163826e-01 2.36607835e-01 6.24563754e-01
5.10563552e-01 -1.20822573e+00 -2.01607689e-01 -4.38805521e-01
-4.87069458e-01 1.30544341e+00 1.33661866e+00 -6.88125566e-02
6.07917011e-01 2.99550354e-01 3.50131770e-03 -4.09708545e-02
-4.43220913e-01 -2.13684902e-01 6.61897063e-02 1.05167079e+00
-4.80676085e-01 -4.58749682e-02 8.68025795e-03 5.77518821e-01
1.36122212e-01 -5.70517443e-02 4.30285007e-01 8.24626029e-01
-8.14105928e-01 -5.09836733e-01 -6.71585381e-01 5.57118893e-01
-1.79486260e-01 2.74925828e-01 -1.95715472e-01 6.64203346e-01
4.42899495e-01 1.24512804e+00 2.49251530e-01 -6.65484488e-01
3.92293423e-01 -4.77836370e-01 -7.58911856e-03 -2.24303324e-02
-3.21709454e-01 -2.60217577e-01 -7.47416466e-02 -5.35268426e-01
-1.68529004e-01 -7.06813693e-01 -9.73312736e-01 -5.27123332e-01
-1.16916098e-01 -3.18935037e-01 2.32707232e-01 7.67578542e-01
3.56610507e-01 5.99263787e-01 7.14655638e-01 -1.00666285e+00
-2.91344255e-01 -7.24115014e-01 -9.21681076e-02 2.32000872e-01
3.72187495e-01 -8.83637249e-01 -1.51441261e-01 -1.05684094e-01] | [8.258193016052246, -0.7755528688430786] |
e42af24e-428a-4d42-8442-db4922921617 | an-efficient-style-virtual-try-on-network | 2105.13183 | null | https://arxiv.org/abs/2105.13183v2 | https://arxiv.org/pdf/2105.13183v2.pdf | An Efficient Style Virtual Try on Network for Clothing Business Industry | With the increasing development of garment manufacturing industry, the method of combining neural network with industry to reduce product redundancy has been paid more and more attention.In order to reduce garment redundancy and achieve personalized customization, more researchers have appeared in the field of virtual trying on.They try to transfer the target clothing to the reference figure, and then stylize the clothes to meet user's requirements for fashion.But the biggest problem of virtual try on is that the shape and motion blocking distort the clothes, causing the patterns and texture on the clothes to be impossible to restore. This paper proposed a new stylized virtual try on network, which can not only retain the authenticity of clothing texture and pattern, but also obtain the undifferentiated stylized try on. The network is divided into three sub-networks, the first is the user image, the front of the target clothing image, the semantic segmentation image and the posture heat map to generate a more detailed human parsing map. Second, UV position map and dense correspondence are used to map patterns and textures to the deformed silhouettes in real time, so that they can be retained in real time, and the rationality of spatial structure can be guaranteed on the basis of improving the authenticity of images. Third,Stylize and adjust the generated virtual try on image. Through the most subtle changes, users can choose the texture, color and style of clothing to improve the user's experience. | ['Neal N. Xiong', 'Yukun Dong', 'Xixi Tao', 'Shanchen Pang'] | 2021-05-27 | null | null | null | null | ['human-parsing'] | ['computer-vision'] | [ 1.17862597e-01 -1.85217455e-01 3.70605802e-03 -2.27632523e-01
5.68759084e-01 -5.17427742e-01 -2.82270640e-01 -5.80547512e-01
-4.88088885e-03 2.20817983e-01 -3.83482464e-02 -1.74605269e-02
5.28634414e-02 -1.07564139e+00 -3.90949339e-01 -5.46974480e-01
5.90053856e-01 1.19887367e-01 4.17287529e-01 -6.17719412e-01
6.75604418e-02 6.31990135e-01 -1.26228356e+00 2.95061827e-01
7.88269579e-01 8.65718961e-01 5.65284133e-01 1.87620938e-01
-3.82980347e-01 7.53291324e-02 -5.18735290e-01 -5.83092213e-01
3.48484755e-01 -6.48262620e-01 -2.68979311e-01 4.58122551e-01
-2.24234816e-02 -5.00863254e-01 -3.01731020e-01 1.43726099e+00
4.01865751e-01 -1.90157574e-02 2.43151948e-01 -7.73377061e-01
-1.29657924e+00 4.56636995e-01 -7.60527909e-01 -3.14978182e-01
2.07061902e-01 1.33763522e-01 2.51660049e-01 -3.79958451e-01
6.17659926e-01 1.49991608e+00 5.80614328e-01 6.53632104e-01
-9.58285749e-01 -6.53683841e-01 2.20414340e-01 7.59368166e-02
-1.34833050e+00 -4.03507128e-02 1.17458248e+00 -1.65112056e-02
-1.76832620e-02 5.35921037e-01 1.12501299e+00 6.66185558e-01
4.28966910e-01 7.83959329e-01 1.06795216e+00 -3.30625117e-01
-2.08537385e-01 2.63739824e-01 -1.99225590e-01 8.46744061e-01
1.01996951e-01 -5.94604434e-03 2.71220297e-01 5.80982387e-01
1.49848628e+00 3.94872516e-01 -4.11017120e-01 -2.32919648e-01
-1.01303589e+00 3.04660529e-01 6.32627964e-01 4.01581317e-01
-2.16158584e-01 -2.77521104e-01 3.17869544e-01 3.13693322e-02
3.93409520e-01 4.98417199e-01 -4.95814234e-01 5.29810227e-02
-6.64862275e-01 -2.66826693e-02 4.29504097e-01 1.00877941e+00
5.97331464e-01 2.15567738e-01 6.07802495e-02 9.82432067e-01
2.94300884e-01 7.22860932e-01 4.37263846e-01 -6.76248312e-01
1.82864457e-01 8.38876426e-01 -7.82963559e-02 -1.54093742e+00
-3.59086424e-01 -1.38808206e-01 -1.15698385e+00 1.79141358e-01
1.88526869e-01 -2.20508322e-01 -1.17065263e+00 1.21693075e+00
3.58690917e-01 -3.36451381e-01 -4.74698514e-01 1.33937097e+00
1.09071088e+00 8.55985701e-01 -1.20806798e-01 -1.31446838e-01
1.43941212e+00 -6.73621476e-01 -1.17007589e+00 3.53570692e-02
-2.85573229e-02 -1.20660853e+00 1.38570833e+00 4.63095695e-01
-1.06134665e+00 -1.05964088e+00 -1.29862189e+00 -1.62711851e-02
-3.11953276e-01 3.83660227e-01 6.09431207e-01 7.11049139e-01
-5.58198273e-01 8.10594201e-01 -4.37042534e-01 -2.43791938e-01
2.68535107e-01 3.18319678e-01 -1.46500722e-01 -6.07609488e-02
-1.31353188e+00 7.47850776e-01 4.12531555e-01 7.97376812e-01
-1.99223652e-01 -2.96739578e-01 -7.65376806e-01 -2.95705885e-01
3.79662305e-01 -5.04151642e-01 8.09908569e-01 -1.29679608e+00
-1.79405916e+00 7.50126898e-01 2.63811409e-01 5.73910475e-01
7.00597167e-01 7.65848383e-02 -8.31328213e-01 -2.01692089e-01
1.23543084e-01 4.05771375e-01 6.24976933e-01 -1.49350119e+00
-6.57910407e-01 -3.81489277e-01 -9.13501009e-02 2.40915149e-01
5.19646034e-02 -1.59480751e-01 -1.14015234e+00 -9.52891767e-01
5.11328995e-01 -7.95116723e-01 -7.11436942e-02 3.84107605e-02
-5.52102566e-01 1.29980832e-01 1.06606066e+00 -1.31533551e+00
1.07423663e+00 -2.24374843e+00 3.74841899e-01 3.81630987e-01
-4.99469861e-02 4.07085180e-01 1.35708824e-02 -1.28091007e-01
4.78623137e-02 2.00168192e-01 -8.73809531e-02 -1.87570509e-02
-2.05277696e-01 5.21837711e-01 1.46615997e-01 2.26180509e-01
-2.08612412e-01 8.10424268e-01 -4.57679331e-01 -5.83732426e-01
3.86347353e-01 2.14300171e-01 -1.65907577e-01 8.92587975e-02
-2.37874851e-01 5.13151526e-01 -6.12930238e-01 8.93340230e-01
1.07897151e+00 3.10569674e-01 1.46288157e-01 -8.49973679e-01
5.78233749e-02 -6.04808629e-01 -1.55798912e+00 1.36461675e+00
-3.31751585e-01 1.92990780e-01 4.00255919e-01 -3.73558670e-01
1.34805810e+00 9.31678154e-03 2.79080927e-01 -1.20277882e+00
4.59465891e-01 -9.02926177e-02 -3.12932700e-01 -8.95312309e-01
6.97113991e-01 -8.42327401e-02 -1.59241572e-01 2.48809561e-01
-6.19798005e-01 -2.20059678e-01 -5.51525839e-02 -2.03542829e-01
-2.57741779e-01 6.52749419e-01 -3.37866455e-01 -1.23548806e-01
2.91175216e-01 1.33873805e-01 7.28171051e-01 -1.14096496e-02
3.75720114e-01 7.03950405e-01 3.39210242e-01 -7.42675602e-01
-1.41961896e+00 -9.03271675e-01 -6.33445308e-02 7.32377529e-01
9.50412154e-01 2.44072735e-01 -1.07503259e+00 -5.08510470e-01
-6.73880801e-02 4.34550166e-01 -5.63102901e-01 -3.02977443e-01
-8.31165612e-01 -5.58943272e-01 1.09062001e-01 4.19031799e-01
1.21344817e+00 -1.34505475e+00 -4.29640003e-02 1.64684221e-01
-1.50693566e-01 -5.89877486e-01 -7.91253567e-01 -4.76098657e-01
-6.75664127e-01 -8.93197119e-01 -8.89362037e-01 -1.08829761e+00
8.17051411e-01 3.26668829e-01 7.03783631e-01 3.29594076e-01
-1.75206244e-01 -1.15777135e-01 -4.05959219e-01 -2.65023500e-01
-4.15708840e-01 -8.32149386e-02 4.59305234e-02 1.39123350e-01
-1.68095529e-02 -3.45080167e-01 -5.59894681e-01 7.91700244e-01
-1.09590435e+00 5.16110182e-01 3.79751682e-01 4.33634877e-01
8.61021101e-01 5.04370630e-01 1.99941218e-01 -7.36717045e-01
8.78099918e-01 6.54289797e-02 -3.95246029e-01 3.91521931e-01
-3.70902866e-01 -4.09879476e-01 7.95949340e-01 -5.93472362e-01
-1.23556721e+00 -2.00728312e-01 -1.94427148e-01 -5.43229580e-01
-8.07142630e-02 1.02079913e-01 -7.85763204e-01 1.24053117e-02
1.83168069e-01 2.16615468e-01 3.86187166e-01 -6.69738889e-01
5.06278634e-01 7.50451982e-01 6.94062412e-01 -2.06371725e-01
6.85373962e-01 1.25328049e-01 -2.38514051e-01 -5.56303501e-01
-4.14735079e-01 3.23637694e-01 -8.51361275e-01 -5.89441299e-01
1.12465262e+00 -3.47548038e-01 -9.26666379e-01 6.40365064e-01
-1.22632170e+00 -2.20007077e-01 -2.62316257e-01 3.45056474e-01
-8.25985670e-02 4.78594393e-01 -1.03407705e+00 -4.19949234e-01
-1.91144794e-01 -1.23117483e+00 7.34599531e-01 8.03536832e-01
-1.25771001e-01 -6.26328170e-01 -5.31313658e-01 2.17930257e-01
1.69738665e-01 2.33688921e-01 1.04580438e+00 3.92462820e-01
-5.77987731e-01 -2.39112645e-01 -4.32335883e-01 5.63510895e-01
5.47784090e-01 2.22498193e-01 -4.79946196e-01 1.68121591e-01
2.57110726e-02 4.96710122e-01 5.39526045e-01 4.48016286e-01
1.09405887e+00 -2.09456220e-01 -3.26462209e-01 8.22921932e-01
1.23427010e+00 9.64317739e-01 9.81450438e-01 2.82868922e-01
1.16242957e+00 8.80184054e-01 8.19234490e-01 -2.04899073e-01
4.88672629e-02 6.23633385e-01 2.24342942e-01 -7.61520922e-01
-2.96137631e-01 -5.89285135e-01 -7.91985542e-02 9.42991912e-01
-5.80080450e-01 -2.73571014e-02 -2.41191715e-01 1.12442002e-01
-1.66368306e+00 -8.37190688e-01 -2.92438984e-01 1.92181623e+00
6.91733778e-01 2.87889857e-02 5.25172912e-02 -1.38609752e-01
1.03756595e+00 -1.55404061e-01 -5.29745281e-01 -6.33671761e-01
-1.56268746e-01 -1.66266963e-01 5.69554150e-01 2.00289696e-01
-7.93473721e-01 9.24550653e-01 5.82908916e+00 1.04747128e+00
-1.40752208e+00 -1.14783265e-01 8.67375553e-01 1.73397690e-01
-4.31126863e-01 -3.97981673e-01 -6.11587405e-01 8.54566514e-01
-2.24765450e-01 2.78574198e-01 7.70471096e-01 7.35566437e-01
4.13264930e-01 -1.03172380e-02 -4.89322156e-01 1.15862310e+00
-6.53076023e-02 -1.17067695e+00 4.21516001e-02 -1.07998386e-01
7.49510527e-01 -9.22585666e-01 2.13134453e-01 7.70943007e-03
2.06715409e-02 -1.01688373e+00 8.97531331e-01 9.46436703e-01
1.09583604e+00 -9.28222477e-01 9.79777694e-01 1.32612750e-01
-1.30398333e+00 3.39197695e-01 -5.97764790e-01 3.49761210e-02
3.36206734e-01 3.39928120e-01 -2.90095776e-01 7.02200472e-01
8.50242317e-01 5.72094262e-01 -4.99153614e-01 6.62824810e-01
-1.83318451e-01 4.49206568e-02 -2.08847359e-01 -3.80310148e-01
4.54861522e-02 -9.02021468e-01 2.07842097e-01 9.70435083e-01
3.36654752e-01 2.93217570e-01 2.05528140e-01 1.16931093e+00
1.28703445e-01 3.58228207e-01 -3.15607488e-01 7.42970780e-02
1.45821810e-01 1.28883195e+00 -9.72470284e-01 -3.00121814e-01
9.56180915e-02 1.42945242e+00 -3.31522077e-01 3.84784728e-01
-8.85445058e-01 -6.20456994e-01 4.30252440e-02 1.98675260e-01
1.28528833e-01 -1.91414624e-01 -5.16680002e-01 -7.28190124e-01
1.20138682e-01 -8.26294959e-01 -9.58622918e-02 -1.06126261e+00
-1.22218812e+00 5.84673405e-01 -1.55269951e-01 -1.20002174e+00
4.57298070e-01 -4.94461566e-01 -6.09593153e-01 1.30362082e+00
-5.81151724e-01 -1.29322863e+00 -4.82230306e-01 4.39437807e-01
4.99688447e-01 4.71597239e-02 5.60108840e-01 5.47269762e-01
-8.69316936e-01 4.09587950e-01 1.88269556e-01 2.08954319e-01
4.99052048e-01 -7.66875923e-01 3.10848147e-01 4.88805860e-01
-3.23760897e-01 4.37884510e-01 6.05301917e-01 -1.13315046e+00
-1.21681035e+00 -1.02004063e+00 3.17315757e-01 -9.59295481e-02
-8.74642581e-02 -3.82094860e-01 -8.00574601e-01 4.66468096e-01
-3.44209932e-02 -5.33233762e-01 8.07042867e-02 -1.88574836e-01
3.17506850e-01 -4.37055081e-01 -1.34112990e+00 9.99209344e-01
1.04873824e+00 -1.45083308e-01 -4.39051151e-01 2.03963652e-01
7.82414198e-01 -6.83366716e-01 -7.63404250e-01 3.39030951e-01
8.20355773e-01 -9.59141076e-01 8.30988884e-01 -1.85205773e-01
5.59504688e-01 -6.55386567e-01 1.58945605e-01 -1.24029338e+00
-6.20340109e-01 -2.87064135e-01 7.67937064e-01 1.46890819e+00
1.78581566e-01 -4.84339416e-01 8.35093856e-01 5.55352926e-01
-1.16907068e-01 -7.33169377e-01 -4.34780627e-01 -4.02521551e-01
-5.48563063e-01 -3.91029231e-02 1.07522774e+00 1.08278549e+00
-4.39802229e-01 1.54735893e-01 -7.43360341e-01 3.93634550e-02
3.19329798e-01 3.22636306e-01 6.16871715e-01 -1.10190296e+00
-1.57597661e-01 -4.07038897e-01 -1.38004214e-01 -1.22081923e+00
-5.27782261e-01 -5.18337607e-01 -3.66580449e-02 -1.80110312e+00
-1.09378941e-01 -5.76779604e-01 1.76135108e-01 2.69392729e-01
-5.32985777e-02 4.57873821e-01 2.80376703e-01 2.34000444e-01
-1.12699054e-01 4.02555048e-01 2.38151670e+00 -1.35402530e-01
-5.94764888e-01 9.79528800e-02 -7.08323121e-01 8.32777858e-01
6.97561026e-01 1.36132091e-01 -3.05631220e-01 -4.52459902e-01
-1.18670627e-01 2.78813958e-01 3.09757233e-01 -5.75079441e-01
-4.16218378e-02 -3.00018877e-01 1.15309167e+00 -5.84635496e-01
1.96790785e-01 -9.99132514e-01 5.65242648e-01 5.16143084e-01
1.31668001e-01 1.02221377e-01 2.62498379e-01 1.71489000e-01
1.32748455e-01 -1.23228863e-01 7.45367765e-01 -4.91557240e-01
-8.44357133e-01 3.25172693e-01 4.52785864e-02 -7.28042901e-01
1.05485368e+00 -7.76695549e-01 9.95205045e-02 -2.09417805e-01
-9.88123894e-01 8.34130049e-02 6.90094173e-01 5.15016079e-01
7.34057605e-01 -1.55195844e+00 -2.71709025e-01 5.12950063e-01
-5.83872318e-01 1.05206579e-01 8.55381966e-01 3.48396599e-01
-1.09978068e+00 -1.03219179e-02 -6.58059180e-01 -3.12844396e-01
-1.36679268e+00 7.85208642e-01 5.51571727e-01 1.64155528e-01
-6.86906576e-01 4.68940735e-01 2.67629534e-01 -5.24717510e-01
-1.27394959e-01 -2.40292236e-01 -3.89598817e-01 -1.30140930e-01
2.53596276e-01 3.34767312e-01 -3.42640340e-01 -5.43002248e-01
1.25698224e-01 1.07966733e+00 4.39518094e-02 1.73069909e-01
1.06362653e+00 -4.33772981e-01 -4.35535222e-01 1.84452295e-01
8.62126827e-01 2.66920358e-01 -1.21294069e+00 2.77115908e-02
-7.93636322e-01 -7.75487840e-01 -1.66907027e-01 -9.22321320e-01
-1.47981596e+00 6.42489791e-01 8.42096090e-01 5.32364845e-01
1.28476691e+00 -1.57629713e-01 1.19709051e+00 -2.56755590e-01
4.30434436e-01 -1.26056480e+00 -1.95262939e-01 -4.94732335e-02
9.41997111e-01 -5.83871484e-01 2.53975317e-02 -7.93725669e-01
-8.98169458e-01 1.32419419e+00 7.33116329e-01 -1.73552290e-01
2.21492141e-01 1.05304316e-01 2.68419325e-01 -1.37021661e-01
5.47494829e-01 1.62079155e-01 4.33468282e-01 7.93778419e-01
3.10557336e-02 2.21416444e-01 -3.11532617e-01 1.04447055e+00
-3.88800383e-01 -5.19373827e-02 6.20413795e-02 4.85939384e-01
-4.37822819e-01 -1.05648971e+00 -7.57951379e-01 3.31286520e-01
-2.21620589e-01 1.98890150e-01 -2.91873604e-01 9.38340664e-01
8.43035460e-01 7.27759004e-01 1.66867584e-01 -7.13794112e-01
9.20504808e-01 -4.92939204e-01 5.91286004e-01 -2.61015385e-01
-3.24124634e-01 4.10828054e-01 -2.77407877e-02 -3.10038209e-01
-1.60790473e-01 -5.13633899e-02 -1.21181989e+00 -8.05074036e-01
-1.77181214e-01 -8.06293730e-03 6.30430877e-01 7.59892821e-01
-2.54094340e-02 9.11326528e-01 6.58409894e-01 -8.52577269e-01
1.14643611e-01 -7.34893620e-01 -1.04014075e+00 6.65722013e-01
-2.74268240e-01 -3.81577402e-01 1.55552626e-01 1.05990261e-01] | [11.577136993408203, -0.9808207750320435] |
5bb7e241-b864-49c7-96e6-a145b359cfdb | using-two-losses-and-two-datasets | 2212.07669 | null | https://arxiv.org/abs/2212.07669v1 | https://arxiv.org/pdf/2212.07669v1.pdf | Using Two Losses and Two Datasets Simultaneously to Improve TempoWiC Accuracy | WSD (Word Sense Disambiguation) is the task of identifying which sense of a word is meant in a sentence or other segment of text. Researchers have worked on this task (e.g. Pustejovsky, 2002) for years but it's still a challenging one even for SOTA (state-of-the-art) LMs (language models). The new dataset, TempoWiC introduced by Loureiro et al. (2022b) focuses on the fact that words change over time. Their best baseline achieves 70.33% macro-F1. In this work, we use two different losses simultaneously to train RoBERTa-based classification models. We also improve our model by using another similar dataset to generalize better. Our best configuration beats their best baseline by 4.23% and reaches 74.56% macroF1. | ['Sauleh Eetemadi', 'Motahhare Mirzaei', 'Mohammad Javad Pirhadi'] | 2022-12-15 | null | null | null | null | ['word-sense-disambiguation'] | ['natural-language-processing'] | [ 3.90979573e-02 -7.47953206e-02 -3.82624865e-01 -2.39580616e-01
-8.15675139e-01 -7.45125353e-01 8.57109606e-01 4.41224009e-01
-9.38375890e-01 9.01581824e-01 2.42322758e-01 -3.63172889e-01
2.21524388e-01 -5.35847008e-01 -3.10842693e-01 -3.96984547e-01
7.18626752e-02 3.15943688e-01 6.06228828e-01 -7.93421507e-01
3.77511799e-01 1.52334943e-01 -1.30900717e+00 3.36746722e-01
7.85460114e-01 5.98453641e-01 5.84494352e-01 4.69192237e-01
-5.94195366e-01 5.47188044e-01 -8.80430996e-01 -3.61970812e-01
-1.33652821e-01 -1.81007773e-01 -1.28767288e+00 -4.54974681e-01
5.52719653e-01 6.67608440e-01 -9.65366364e-02 1.11614978e+00
4.95196581e-01 2.95841068e-01 3.28675330e-01 -9.38151062e-01
-6.21076167e-01 1.08202708e+00 -5.94878733e-01 6.03909612e-01
3.58212918e-01 -1.37231722e-01 1.16649806e+00 -8.42916012e-01
8.19152594e-01 1.38965547e+00 6.25575960e-01 7.42690623e-01
-1.13168967e+00 -8.24613392e-01 4.64648157e-01 3.70015025e-01
-1.27612424e+00 -3.15432459e-01 5.16457200e-01 -1.65104926e-01
1.46566224e+00 3.56950969e-01 2.59322822e-01 1.40138578e+00
2.33028471e-01 1.07381117e+00 1.21228445e+00 -7.45416045e-01
9.82537642e-02 -1.80676252e-01 5.87686121e-01 2.85457343e-01
5.32593690e-02 -4.18778121e-01 -7.44848907e-01 6.38259053e-02
6.67414814e-02 -2.67127395e-01 -7.38515630e-02 3.46896470e-01
-1.25598180e+00 5.96089244e-01 9.00049433e-02 1.10466588e+00
-8.89454558e-02 -9.38298255e-02 4.77659971e-01 5.95789075e-01
7.32925832e-01 9.37972188e-01 -1.03877306e+00 -5.08389473e-01
-1.03449881e+00 4.81802046e-01 7.35812068e-01 7.32963681e-01
4.18550134e-01 -1.74161330e-01 -4.64943737e-01 1.12335503e+00
1.47389576e-01 5.70603728e-01 1.01425588e+00 -5.82738221e-01
5.64312160e-01 5.23371994e-01 1.24945626e-01 -6.72524631e-01
-6.50319219e-01 -6.97122931e-01 -5.99149227e-01 -2.64779896e-01
3.45567971e-01 -2.61023819e-01 -1.15122151e+00 1.88092399e+00
-6.09713830e-02 3.84883344e-01 1.44795924e-02 5.33471167e-01
7.75340736e-01 6.05087817e-01 4.95523155e-01 -1.81250215e-01
1.30558515e+00 -8.99604499e-01 -8.18147659e-01 -6.62949443e-01
6.81959033e-01 -1.06539106e+00 1.37208796e+00 5.34374177e-01
-8.78604114e-01 -3.69270802e-01 -1.10054624e+00 -6.71935305e-02
-8.78516734e-01 4.21852283e-02 4.95367438e-01 4.47452903e-01
-1.10464203e+00 6.42521620e-01 -7.97697008e-01 -7.29522169e-01
-2.80306805e-02 1.26206398e-01 -7.82503709e-02 2.34184399e-01
-1.98390996e+00 1.26592541e+00 6.07957423e-01 -4.40140456e-01
-4.15931135e-01 -8.32230806e-01 -7.20502496e-01 -1.67563826e-01
6.24289155e-01 -4.57931876e-01 1.34607160e+00 -7.09209800e-01
-1.32588756e+00 1.27874243e+00 -4.40837532e-01 -6.74616277e-01
2.58873075e-01 -5.55730164e-01 -8.98274720e-01 -4.48627055e-01
3.58588398e-01 3.49575579e-01 3.87797713e-01 -1.03129518e+00
-9.32106316e-01 -3.34606171e-01 2.36708090e-01 -9.34054032e-02
-2.74822652e-01 2.36846060e-01 -4.32649165e-01 -1.08624816e+00
-3.60666327e-02 -9.89476979e-01 -1.89817414e-01 -7.66296208e-01
-4.63913530e-01 -7.59777665e-01 6.40654325e-01 -5.62539518e-01
1.95351195e+00 -1.99837601e+00 -2.03445733e-01 -1.11887768e-01
1.97717741e-01 7.58203685e-01 -4.44224298e-01 5.39400160e-01
-3.75982940e-01 5.11463940e-01 -3.36932838e-01 -7.23908663e-01
-1.49936173e-02 1.43713295e-01 -2.92098254e-01 -7.44649991e-02
1.51025355e-01 6.91365838e-01 -1.32245553e+00 -2.45111927e-01
-5.61024360e-02 2.80370563e-01 -1.26294732e-01 -3.29405665e-01
-1.83767676e-01 1.08131751e-01 -3.13567847e-01 2.61072546e-01
3.35207164e-01 -1.66923016e-01 3.08394611e-01 5.21431975e-02
-2.20676526e-01 7.59012938e-01 -1.28374457e+00 2.04175425e+00
-7.83825278e-01 7.78097391e-01 -4.29474831e-01 -8.52825105e-01
8.66519928e-01 2.46770978e-01 5.01983047e-01 -7.00830638e-01
-5.56059889e-02 1.58912361e-01 -4.84366827e-02 -4.40526843e-01
9.35384095e-01 -2.53312975e-01 -4.02234316e-01 1.28118455e-01
1.01847894e-01 -1.04874194e-01 6.99274361e-01 2.56329209e-01
1.31525230e+00 -1.87325940e-01 5.82556844e-01 -7.01960742e-01
7.18309581e-01 -4.82437983e-02 7.38611400e-01 7.71779656e-01
-2.73742318e-01 6.17879152e-01 1.33709088e-01 -2.34291226e-01
-7.99589932e-01 -1.04878497e+00 -1.04651198e-01 1.30986500e+00
1.38787150e-01 -7.64382482e-01 -5.96635103e-01 -6.99734807e-01
1.01293683e-01 1.24143469e+00 -6.61831260e-01 -7.54595846e-02
-6.66975796e-01 -5.66330194e-01 6.78933382e-01 4.76198524e-01
3.66251290e-01 -1.05723846e+00 -1.17565848e-01 3.84486139e-01
-3.63048315e-01 -1.01209867e+00 -6.67974770e-01 1.44948110e-01
-4.52372402e-01 -9.30075884e-01 -4.58391964e-01 -6.58048153e-01
-7.60930255e-02 2.65229732e-01 1.60557294e+00 -1.28942907e-01
-8.99345279e-02 8.09827745e-02 -7.20148146e-01 -6.59987688e-01
-3.28182399e-01 6.84975088e-01 3.73843610e-01 -3.16307545e-01
7.92094409e-01 -2.66839266e-01 -1.18989386e-01 4.58672689e-03
-9.10069525e-01 -4.17229861e-01 1.24212824e-01 7.23900259e-01
6.25618517e-01 -5.63630909e-02 6.29445076e-01 -1.18378580e+00
8.99280369e-01 -4.55025971e-01 -3.68956745e-01 5.04123986e-01
-9.39294994e-01 3.53660166e-01 4.45823461e-01 -4.58638370e-01
-1.02480888e+00 -2.72934496e-01 -2.83131301e-01 4.98702899e-02
-1.77513257e-01 4.43759322e-01 -9.56378654e-02 3.47867399e-01
5.91790736e-01 -4.94831912e-02 -6.21107221e-01 -7.60156512e-01
3.50732714e-01 8.40415299e-01 3.44695181e-01 -5.77184319e-01
5.81559062e-01 9.97769758e-02 -3.95613551e-01 -8.71562898e-01
-1.51174319e+00 -9.14570630e-01 -5.96113324e-01 1.50214896e-01
8.29699337e-01 -8.60655963e-01 -4.66457576e-01 4.58184570e-01
-1.19001067e+00 -1.80294141e-01 -9.86595824e-02 4.00043905e-01
-1.29600137e-01 2.33897775e-01 -3.50444824e-01 -7.57771015e-01
-4.06831205e-01 -6.71545267e-01 9.91437554e-01 2.48205319e-01
-7.72784591e-01 -1.29137790e+00 3.52446169e-01 4.05287482e-02
5.81574976e-01 1.50477156e-01 4.82793689e-01 -8.54240000e-01
2.24254981e-01 1.66340843e-01 1.75693274e-01 2.48498306e-01
4.15197521e-01 -9.03335363e-02 -8.96722674e-01 -2.85168976e-01
-6.58329874e-02 -7.03639686e-02 1.29111528e+00 2.43674979e-01
1.08356130e+00 -2.64306311e-02 -5.45560241e-01 1.17590174e-01
1.39274573e+00 2.67459154e-01 3.53530079e-01 6.37115479e-01
4.31370705e-01 2.41773397e-01 7.59178400e-01 2.49301538e-01
5.82852662e-01 6.96290433e-01 -6.16918541e-02 2.09285408e-01
-3.44490021e-01 -1.51437014e-01 5.73502421e-01 1.10060930e+00
9.49754566e-02 -6.08487308e-01 -1.25660288e+00 9.43872809e-01
-1.82151353e+00 -1.01888967e+00 -2.48360246e-01 1.91630149e+00
1.26776004e+00 4.91308063e-01 -9.38048512e-02 1.22677766e-01
6.98468029e-01 5.86638272e-01 -3.55542302e-01 -5.84876239e-01
-5.47662914e-01 4.80991334e-01 4.63397533e-01 8.44588459e-01
-1.34908164e+00 1.50886595e+00 5.42207766e+00 1.21414936e+00
-1.05160820e+00 1.83107644e-01 3.93668383e-01 -8.18295777e-02
-4.37554389e-01 -1.49044231e-01 -9.75212991e-01 6.22656584e-01
9.12787855e-01 -3.35337728e-01 2.38823578e-01 4.04006004e-01
1.30600989e-01 -2.08427325e-01 -7.68952429e-01 1.02337170e+00
3.16033006e-01 -1.01773405e+00 -1.35044530e-01 -6.36011541e-01
8.76851201e-01 4.78713185e-01 -3.21014136e-01 5.52907765e-01
6.60388410e-01 -8.70280802e-01 7.30019867e-01 7.04929650e-01
6.13000691e-01 -6.98857605e-01 7.93468058e-01 3.24826837e-01
-1.17602456e+00 2.73164779e-01 -5.22415452e-02 -1.66034758e-01
1.15184322e-01 1.05689800e+00 -7.43226647e-01 5.57460606e-01
7.18056321e-01 9.37796056e-01 -7.71576941e-01 7.32163250e-01
-4.48129088e-01 9.99681056e-01 -1.93245530e-01 -3.28465551e-01
3.65502656e-01 3.81942838e-01 7.23160744e-01 1.63266385e+00
1.24396525e-01 4.94035743e-02 3.93462002e-01 4.29311991e-01
-1.61326170e-01 3.15641724e-02 -2.85723567e-01 -5.08291610e-02
6.77746952e-01 1.01165664e+00 -4.68398094e-01 -3.90131295e-01
-2.21077979e-01 9.41530466e-01 3.01464826e-01 2.89697856e-01
-4.10959870e-01 -6.57414198e-01 1.20351255e+00 -1.77678406e-01
1.17885545e-01 -3.59222561e-01 -3.84507686e-01 -1.31762958e+00
4.67214209e-04 -5.67808092e-01 4.83436376e-01 -5.25402009e-01
-1.73090684e+00 7.15946496e-01 -2.41174564e-01 -1.02425778e+00
-2.23847076e-01 -6.57838106e-01 -4.21316028e-01 9.61953521e-01
-1.71308577e+00 -9.00634348e-01 9.04175788e-02 3.34634125e-01
7.91632950e-01 -2.23223448e-01 8.56237531e-01 2.40382522e-01
-3.93071502e-01 7.18440533e-01 2.16604248e-01 2.05363557e-01
1.19548404e+00 -1.63409591e+00 9.72705543e-01 8.89556050e-01
3.66262734e-01 7.23964274e-01 9.86939847e-01 -6.39307976e-01
-8.83693695e-01 -9.49464679e-01 1.85142970e+00 -7.35790074e-01
1.11365759e+00 -2.03813493e-01 -8.76960695e-01 4.57353681e-01
4.61191326e-01 -4.35585827e-01 5.84248066e-01 4.93940592e-01
-4.46781695e-01 -1.90266922e-01 -9.08921599e-01 7.21805811e-01
1.20378685e+00 -5.95144987e-01 -1.03915012e+00 1.08720057e-01
1.09035647e+00 -3.78595948e-01 -6.89995825e-01 2.27978244e-01
1.28689930e-01 -5.92375934e-01 8.02646637e-01 -1.04750276e+00
2.05905080e-01 -4.27129537e-01 -5.68858385e-02 -1.76481307e+00
-2.72201598e-01 -7.85577953e-01 1.62270308e-01 1.55278373e+00
7.04754949e-01 -5.25714576e-01 2.76557505e-01 1.79688901e-01
-1.28409669e-01 -5.75611770e-01 -8.19385409e-01 -1.19886172e+00
5.12029469e-01 -8.41465652e-01 6.36944771e-01 1.49835312e+00
2.64205724e-01 6.11038923e-01 3.35305519e-02 -5.79112358e-02
1.37999043e-01 -2.96191033e-03 1.71102196e-01 -1.21276760e+00
5.51705100e-02 -7.77335703e-01 -2.21474081e-01 -8.86810303e-01
5.24257183e-01 -1.08179998e+00 9.87951383e-02 -1.39106858e+00
2.75657009e-02 -3.50390464e-01 -8.30170691e-01 9.14122581e-01
-6.87468648e-01 -1.58425514e-02 4.59558517e-01 -2.50638932e-01
-8.72165680e-01 2.30965614e-01 1.04649270e+00 -3.48027408e-01
-7.74187967e-02 -1.24305315e-01 -6.68696165e-01 5.49323618e-01
1.19794786e+00 -5.67162812e-01 -1.93969443e-01 -6.50290251e-01
3.11448961e-01 -2.93085009e-01 -9.61824432e-02 -8.04247558e-01
2.55889922e-01 -3.48406732e-01 -1.93725586e-01 -3.76711100e-01
-5.89396320e-02 -3.69183123e-01 -3.01152021e-01 4.89236921e-01
-5.10296106e-01 4.70867783e-01 3.28402340e-01 1.88744903e-01
-4.18924928e-01 -3.55243236e-01 6.13105297e-01 -1.15927905e-01
-1.08621120e+00 7.50015900e-02 -1.70972064e-01 7.20076680e-01
7.75234044e-01 3.14904928e-01 -3.31498951e-01 -8.53713304e-02
-7.02604055e-01 4.01363313e-01 1.97944820e-01 1.07192767e+00
1.67295545e-01 -1.25733697e+00 -9.38744664e-01 -2.89854825e-01
3.66775334e-01 -2.73102641e-01 -1.13972723e-01 4.29210067e-01
-7.62444511e-02 5.24153650e-01 2.36318365e-01 -1.89818338e-01
-1.24658465e+00 2.26812050e-01 1.73877701e-01 -6.20857000e-01
-1.90973446e-01 1.11521769e+00 -4.32673126e-01 -4.13531721e-01
-8.39510374e-03 -5.13642669e-01 -4.20450330e-01 3.87674451e-01
7.96320498e-01 4.61805969e-01 2.71602184e-01 -4.25941676e-01
-8.04464936e-01 5.34117699e-01 -2.55793422e-01 -4.73439917e-02
1.33946037e+00 -1.85081571e-01 -1.29357934e-01 9.14066970e-01
1.19571590e+00 3.02896351e-01 -5.67779124e-01 -5.36340058e-01
7.29172587e-01 -1.68742985e-01 3.04412525e-02 -1.25912833e+00
-8.64991009e-01 7.78421521e-01 6.21946394e-01 4.15808588e-01
9.61498857e-01 1.51717708e-01 1.02431667e+00 4.17989880e-01
3.67191195e-01 -1.41398966e+00 -2.56226778e-01 1.43994570e+00
7.54201770e-01 -1.41924322e+00 -1.64308742e-01 -1.06465980e-01
-6.97977781e-01 1.05052018e+00 5.27694762e-01 2.17882488e-02
7.63614953e-01 2.98027605e-01 2.99852759e-01 1.00571709e-02
-8.57142568e-01 -6.73892856e-01 4.80757505e-01 3.56776685e-01
9.15374219e-01 3.57720852e-01 -1.00359595e+00 8.46653581e-01
-5.43765008e-01 -9.25574079e-02 4.23378021e-01 6.75302088e-01
-4.67755109e-01 -1.50293815e+00 3.32689807e-02 3.92581403e-01
-5.80661058e-01 -4.77561176e-01 -5.11955917e-01 6.45990312e-01
2.73290634e-01 1.28073514e+00 -1.53464992e-02 -5.55904746e-01
4.82600361e-01 3.08827609e-01 1.37789533e-01 -9.61032271e-01
-8.04670334e-01 -3.42023224e-01 3.06309551e-01 -5.39231956e-01
-6.05539441e-01 -8.47341359e-01 -1.51004529e+00 -2.25547269e-01
-6.07930720e-02 1.57694504e-01 5.20547926e-01 1.20861173e+00
1.51619837e-01 6.48754895e-01 4.75108504e-01 -1.08500570e-01
-3.67654651e-01 -1.21100676e+00 -6.49505794e-01 6.11013114e-01
4.22643036e-01 -6.07247591e-01 -2.02753171e-01 1.36811122e-01] | [10.252880096435547, 9.058541297912598] |
bc35509c-e107-4a84-a4bf-6a0c6612d1ae | towards-zero-shot-relation-extraction-in-web | 2305.13805 | null | https://arxiv.org/abs/2305.13805v1 | https://arxiv.org/pdf/2305.13805v1.pdf | Towards Zero-shot Relation Extraction in Web Mining: A Multimodal Approach with Relative XML Path | The rapid growth of web pages and the increasing complexity of their structure poses a challenge for web mining models. Web mining models are required to understand the semi-structured web pages, particularly when little is known about the subject or template of a new page. Current methods migrate language models to the web mining by embedding the XML source code into the transformer or encoding the rendered layout with graph neural networks. However, these approaches do not take into account the relationships between text nodes within and across pages. In this paper, we propose a new approach, ReXMiner, for zero-shot relation extraction in web mining. ReXMiner encodes the shortest relative paths in the Document Object Model (DOM) tree which is a more accurate and efficient signal for key-value pair extraction within a web page. It also incorporates the popularity of each text node by counting the occurrence of the same text node across different web pages. We use the contrastive learning to address the issue of sparsity in relation extraction. Extensive experiments on public benchmarks show that our method, ReXMiner, outperforms the state-of-the-art baselines in the task of zero-shot relation extraction in web mining. | ['Jingbo Shang', 'Zilong Wang'] | 2023-05-23 | null | null | null | null | ['relation-extraction'] | ['natural-language-processing'] | [ 2.69171357e-01 2.29125097e-01 -9.28140938e-01 6.99048787e-02
-7.40872920e-01 -4.60978359e-01 5.75360358e-01 5.36731541e-01
2.49280054e-02 5.56534350e-01 3.39486182e-01 -6.86675906e-01
-3.53257835e-01 -1.26556480e+00 -8.98423254e-01 -8.37744549e-02
-6.17647827e-01 5.11156142e-01 5.95384002e-01 -2.44654436e-02
-2.64577985e-01 -1.10294156e-01 -1.28561890e+00 7.45339930e-01
7.49732435e-01 1.30941498e+00 5.00817001e-02 4.75148141e-01
-1.00946128e+00 9.83347535e-01 -3.55313838e-01 -8.72732699e-01
2.73623258e-01 -2.30467528e-01 -1.07352793e+00 8.93583000e-02
6.35835528e-01 -2.08463892e-01 -6.37835383e-01 1.17208922e+00
-3.86088006e-02 -3.99910480e-01 5.11990130e-01 -1.15321636e+00
-4.21642512e-01 1.40225303e+00 -1.16831505e+00 3.02427649e-01
5.03110826e-01 -6.08656406e-01 1.79783428e+00 -6.65503681e-01
1.09049702e+00 9.57020462e-01 5.84642172e-01 -1.69035733e-01
-9.74501669e-01 -4.53761995e-01 3.48672986e-01 5.80585361e-01
-1.22919381e+00 -1.77033275e-01 6.14504158e-01 -2.93911219e-01
1.22077453e+00 4.74613428e-01 3.93184274e-01 1.03767920e+00
1.17859043e-01 7.11030900e-01 5.54233074e-01 -5.79682589e-01
-4.17512245e-02 1.74050983e-02 6.92872107e-01 9.97459352e-01
7.96044171e-01 -3.11690092e-01 -9.13008332e-01 -4.60338324e-01
1.16073638e-01 -1.26034975e-01 -9.69098657e-02 -5.77908993e-01
-6.82850361e-01 6.98178709e-01 2.38292381e-01 1.40908241e-01
-3.23305786e-01 -8.01374093e-02 5.15144289e-01 3.78800720e-01
3.47564757e-01 -2.44338829e-02 -7.89346576e-01 -5.37907556e-02
-7.86013842e-01 1.07568339e-01 1.43728685e+00 1.38142216e+00
7.70556688e-01 -5.36427081e-01 7.70215616e-02 8.30205619e-01
3.89964670e-01 -4.90430072e-02 4.08679873e-01 -3.92122358e-01
1.24757481e+00 1.11016870e+00 -4.85304505e-01 -9.00131226e-01
-2.20317319e-01 -3.66406381e-01 -7.77757168e-01 -2.40820944e-01
1.60178527e-01 2.00953931e-02 -6.28753543e-01 1.09167147e+00
5.84923983e-01 -1.68845803e-01 -6.70847595e-02 2.87051529e-01
1.01238072e+00 5.94726741e-01 7.49088228e-02 -4.61936057e-01
1.66549289e+00 -9.76775467e-01 -8.07067215e-01 -3.98213923e-01
7.70594835e-01 -5.30881941e-01 8.23052466e-01 2.41170064e-01
-7.67591953e-01 -3.44774760e-02 -1.19524932e+00 -4.80795689e-02
-7.52622366e-01 -3.44804347e-01 9.80895221e-01 4.48059678e-01
-3.10670346e-01 6.01242959e-01 -5.17796516e-01 -4.87582862e-01
5.62184751e-01 -1.32924728e-02 -3.15216154e-01 8.56473744e-02
-1.43071270e+00 4.43375468e-01 6.60781443e-01 -4.25949991e-01
-2.34159857e-01 -6.09534264e-01 -1.06942892e+00 3.97207379e-01
1.27559912e+00 -2.76571691e-01 1.19711506e+00 -6.49352133e-01
-7.74854004e-01 5.78467071e-01 -1.61433235e-01 -7.54389286e-01
1.15060173e-01 -2.85050899e-01 -6.52939022e-01 -1.99249119e-01
1.64522335e-01 -2.09115103e-01 6.01674855e-01 -1.07726681e+00
-1.03235173e+00 -4.61391389e-01 -8.36712494e-02 -1.43243998e-01
-7.17341721e-01 4.93094698e-03 -8.76532137e-01 -5.65033019e-01
-4.41405438e-02 -5.09532928e-01 2.43978985e-02 -2.17383802e-01
-9.80551362e-01 -4.56742704e-01 9.40992117e-01 -9.26094830e-01
2.10542059e+00 -1.79864299e+00 -1.98580369e-01 4.79234844e-01
5.19016266e-01 -6.66805431e-02 -2.91254446e-02 8.30295026e-01
2.23749168e-02 4.38752383e-01 -1.11841284e-01 3.27262878e-01
2.50100158e-02 2.45562911e-01 -3.86590898e-01 -5.87696880e-02
-1.53613031e-01 1.07315731e+00 -7.67472982e-01 -9.57449973e-01
-4.37109590e-01 5.22424793e-03 -1.69522047e-01 9.74735916e-02
-6.69156075e-01 -8.71845186e-01 -3.80299658e-01 8.17057610e-01
4.94609326e-01 -8.82923067e-01 1.00879502e+00 -3.23710024e-01
1.70609593e-01 8.48935843e-01 -1.36099708e+00 1.38003492e+00
-2.90792733e-01 4.32536870e-01 -2.29639634e-01 -5.49948096e-01
5.10267675e-01 8.60986188e-02 6.19116068e-01 -9.96078432e-01
-3.47412139e-01 -6.72049150e-02 -6.68874085e-02 -6.61797583e-01
2.71393329e-01 5.76728284e-01 2.22285036e-02 5.60339630e-01
6.71754852e-02 6.31445169e-01 1.00633323e+00 6.19675994e-01
1.52744150e+00 1.35414481e-01 8.85249555e-01 -8.83035548e-03
3.21117163e-01 -3.68201062e-02 3.51134092e-01 6.51405871e-01
6.30497754e-01 -3.10712513e-02 9.23479617e-01 -5.30990899e-01
-8.82381499e-01 -1.01320457e+00 -6.99404404e-02 1.30258191e+00
-7.69043118e-02 -1.35304511e+00 -4.48244005e-01 -1.38838542e+00
3.93980294e-01 4.51286823e-01 -5.49428046e-01 1.42665178e-01
-7.43879139e-01 -8.61832738e-01 1.09459534e-01 4.77616847e-01
1.66640237e-01 -7.61689007e-01 -1.23349629e-01 4.35091257e-01
-4.34071302e-01 -1.38833618e+00 -3.89394253e-01 6.31483793e-01
-6.45689905e-01 -1.52464175e+00 -9.93698929e-03 -9.33456302e-01
3.08733523e-01 2.92826176e-01 1.53117788e+00 -3.91172152e-03
-3.72551292e-01 4.74727713e-02 -6.37854099e-01 -2.23711044e-01
-3.30345124e-01 7.56536186e-01 -5.97486973e-01 -2.31741473e-01
6.38711631e-01 -5.80080986e-01 -2.16175988e-02 1.03460364e-01
-8.37690890e-01 3.51210125e-02 5.81070423e-01 5.97124636e-01
5.68861604e-01 5.43700933e-01 1.25982121e-01 -1.59743202e+00
7.04264939e-01 -8.90163481e-01 -5.85943699e-01 6.20935977e-01
-1.16582298e+00 5.38341641e-01 5.02909958e-01 -4.40086246e-01
-8.41105580e-01 -1.47921965e-01 9.36626643e-02 3.55890512e-01
3.33250761e-01 1.21344197e+00 -4.46089327e-01 3.25508326e-01
5.17573595e-01 3.65452692e-02 -3.80917251e-01 -8.27696323e-01
4.53465432e-01 7.41793096e-01 2.49497727e-01 -3.27485681e-01
1.05053675e+00 2.97255218e-01 4.53749672e-02 -7.91879952e-01
-1.14553523e+00 -6.72309458e-01 -6.91452444e-01 1.99081227e-01
3.74097794e-01 -7.28529811e-01 -3.39607120e-01 -3.78643237e-02
-1.05347431e+00 -1.54413059e-01 -2.00434074e-01 -1.95966825e-01
-2.11199466e-02 7.58509099e-01 -7.10952520e-01 -5.59623957e-01
-5.79174757e-01 -4.71790582e-01 8.27006519e-01 -1.89902157e-01
-4.96510983e-01 -8.41142178e-01 1.98411986e-01 2.76219338e-01
-2.14444339e-01 -3.87789607e-02 1.64251053e+00 -8.79542291e-01
-7.85223126e-01 -2.99772799e-01 -4.41925675e-01 -2.52503037e-01
4.50243682e-01 -9.91965160e-02 -3.34907085e-01 2.23718345e-01
-5.15623689e-01 -1.40615717e-01 9.10287917e-01 -1.90001458e-01
1.30909753e+00 -1.01944935e+00 -5.97279072e-01 7.13522255e-01
1.51464903e+00 -1.35060819e-02 5.16023517e-01 8.55214953e-01
8.09434712e-01 8.02394032e-01 4.26638186e-01 5.93416274e-01
6.74868166e-01 7.13795006e-01 6.13728523e-01 2.52318740e-01
-1.86330825e-01 -6.80230439e-01 2.39472583e-01 8.83497834e-01
2.92954594e-01 -5.06821096e-01 -8.34363282e-01 4.29378301e-01
-1.81656992e+00 -9.46973562e-01 -5.52237391e-01 1.89211333e+00
1.31025517e+00 6.03916049e-01 5.28418273e-02 2.84410715e-01
3.98190349e-01 4.82146561e-01 -2.77206182e-01 -7.08626211e-02
-1.25245929e-01 9.61106345e-02 8.73181880e-01 3.06899607e-01
-1.15951288e+00 7.34188378e-01 5.64107275e+00 7.84339011e-01
-3.94740641e-01 -4.76506613e-02 2.88370907e-01 1.19494863e-01
-3.18333447e-01 1.97346315e-01 -1.41569281e+00 5.84538043e-01
1.04195654e+00 -3.62857223e-01 5.51972091e-01 1.04980302e+00
-3.52608055e-01 -7.61920633e-03 -1.15675032e+00 7.63838172e-01
1.04211628e-01 -1.58045304e+00 4.16703314e-01 3.45312744e-01
4.78098840e-01 -5.21029308e-02 -3.22160870e-01 1.79257959e-01
5.50588429e-01 -7.20380962e-01 4.17866915e-01 1.82105914e-01
7.71581888e-01 -5.57411611e-01 5.29586732e-01 2.44950056e-01
-1.61265588e+00 5.44944927e-02 -1.73268393e-01 2.27369264e-01
-7.79286539e-03 7.64763653e-01 -9.24833000e-01 6.15435600e-01
9.77170646e-01 6.53824151e-01 -8.29199374e-01 7.14177072e-01
-2.95821995e-01 6.96735501e-01 -2.54857928e-01 -3.28085184e-01
3.87702063e-02 -1.59603693e-02 3.36705476e-01 1.30213487e+00
7.90831354e-03 -4.76126254e-01 1.15850657e-01 4.30742741e-01
-5.71703792e-01 4.59433913e-01 -6.16364539e-01 -4.62270319e-01
3.57297152e-01 1.46564579e+00 -4.46702212e-01 -3.51794541e-01
-1.06524742e+00 5.85486352e-01 6.35924697e-01 2.53932804e-01
-4.98088509e-01 -5.02294779e-01 5.08460999e-01 7.94210911e-01
6.79289937e-01 -3.07739466e-01 -2.83005536e-01 -1.08695960e+00
4.63639289e-01 -1.19603074e+00 9.60393488e-01 -3.40652257e-01
-1.24756062e+00 4.53780681e-01 -6.93477807e-04 -8.48885417e-01
-4.36367869e-01 -6.68505132e-01 -2.26136968e-01 5.15710831e-01
-1.51198018e+00 -1.13847435e+00 -1.91843286e-01 2.27813557e-01
7.16169715e-01 -2.51183629e-01 6.03587806e-01 3.19279164e-01
-4.99292254e-01 5.93257189e-01 6.24513775e-02 5.64409792e-01
4.02808130e-01 -1.42198491e+00 1.02034605e+00 8.75718474e-01
8.38803351e-01 6.41940117e-01 2.89820462e-01 -1.22481692e+00
-1.74839997e+00 -1.09840536e+00 1.27096248e+00 -3.97189170e-01
1.29415882e+00 -5.61326504e-01 -1.24431574e+00 7.99943328e-01
1.39236227e-01 -1.23082727e-01 9.25313711e-01 7.06982255e-01
-8.53655398e-01 -3.44649494e-01 -4.16167498e-01 6.26478553e-01
1.35966957e+00 -4.37921107e-01 -4.62815166e-01 3.73661369e-01
9.19904590e-01 4.97738384e-02 -9.11199212e-01 3.12298656e-01
7.17938423e-01 -4.59362566e-01 9.54152167e-01 -8.24758291e-01
6.23273253e-01 1.67412251e-01 -3.80836390e-02 -7.79817760e-01
-3.55962634e-01 -7.18293488e-01 -1.50296462e+00 1.42507660e+00
8.05598497e-01 -9.74848419e-02 1.19122922e+00 1.20206006e-01
6.54105484e-01 -9.47524607e-01 -7.03760564e-01 -8.18071008e-01
-3.71369421e-01 -5.04894257e-01 7.02663422e-01 8.20393145e-01
4.48153585e-01 8.33578229e-01 -3.09937686e-01 -4.99561466e-02
9.12396431e-01 3.19888443e-01 7.23806798e-01 -1.62209225e+00
-3.98425609e-01 -3.58284086e-01 -6.43797070e-02 -1.05430651e+00
-8.59794579e-03 -1.29095542e+00 -4.14768279e-01 -1.91266954e+00
6.10416353e-01 -9.24512595e-02 -1.34150609e-01 4.01847422e-01
-1.05331257e-01 -4.76902306e-01 -2.51393646e-01 2.84103513e-01
-8.26517522e-01 1.08078308e-01 8.03651094e-01 -4.51813191e-01
-3.66471475e-03 1.50579177e-02 -8.95374000e-01 7.02580452e-01
4.16615248e-01 -6.72988653e-01 -3.59252065e-01 -2.04426259e-01
9.11879480e-01 -9.46984664e-02 -8.21872652e-02 -5.95459104e-01
5.05241811e-01 -2.31605276e-01 2.75013268e-01 -8.67956817e-01
-3.02611798e-01 -1.11627507e+00 -9.13947895e-02 2.57779330e-01
-3.85104597e-01 4.15898450e-02 -2.29292437e-01 8.22829604e-01
-8.10935646e-02 -5.04253089e-01 3.18352640e-01 -2.89112419e-01
-8.46041381e-01 6.08594954e-01 -9.77154821e-02 4.31351066e-01
6.43614531e-01 1.00719288e-01 -3.31295848e-01 -1.60538048e-01
-3.52019608e-01 2.20224395e-01 1.39026389e-01 7.73562312e-01
5.76305330e-01 -1.26782584e+00 -3.85135502e-01 1.29929379e-01
4.07057673e-01 -1.48546264e-01 -3.85653287e-01 2.91239649e-01
-3.05152446e-01 3.54877859e-01 3.35325301e-01 2.05036495e-02
-1.62101603e+00 7.88569570e-01 -1.02864057e-01 -9.25648212e-01
-7.63055444e-01 4.21327293e-01 -1.39353350e-01 -9.35597941e-02
5.91176271e-01 -2.80986011e-01 -3.45014274e-01 3.72366965e-01
7.10152686e-01 2.96744466e-01 3.40060949e-01 3.06703895e-02
-1.65257007e-01 2.44755402e-01 -5.95387340e-01 2.59496838e-01
1.34861791e+00 -5.32310866e-02 -5.53534687e-01 5.47910511e-01
1.14079297e+00 1.27386734e-01 -8.76397908e-01 -7.88006783e-01
1.03719449e+00 -2.28399724e-01 -6.34658858e-02 -8.32157135e-01
-7.39494622e-01 2.83379883e-01 -3.99288163e-02 6.32544637e-01
6.39128864e-01 3.76406640e-01 1.16471243e+00 6.64547205e-01
4.66107070e-01 -1.17636311e+00 3.82119119e-02 7.17233717e-01
5.71349978e-01 -1.17137504e+00 4.56666529e-01 -9.59991992e-01
-2.86075473e-01 1.12206721e+00 5.69608271e-01 4.62564319e-01
9.85970438e-01 9.50443566e-01 -2.43803337e-01 -2.65836656e-01
-1.09706914e+00 -4.70028073e-01 6.02535784e-01 5.51184952e-01
5.88977337e-01 -1.60061434e-01 -3.84274572e-01 7.09597170e-01
-2.12635085e-01 -2.11729512e-01 1.94173068e-01 1.27101481e+00
-5.19081533e-01 -1.52599740e+00 1.32775337e-01 1.11597741e+00
-7.16202736e-01 -6.14586473e-01 -6.41440392e-01 8.23418617e-01
-1.23215415e-01 6.10365629e-01 -3.55760120e-02 -5.84008694e-01
1.80373341e-01 2.70618945e-01 2.73481667e-01 -8.93722236e-01
-3.87133509e-01 7.54223540e-02 6.02529645e-01 -4.43538904e-01
7.82256499e-02 -6.09679520e-01 -1.11713707e+00 -1.81003377e-01
-3.20701927e-01 2.39216164e-01 5.69023788e-01 6.20538116e-01
3.14551175e-01 6.32093012e-01 3.04909080e-01 2.84666717e-01
-6.14269733e-01 -9.24585104e-01 -7.09235489e-01 3.40247363e-01
-2.57886667e-02 -3.60825002e-01 -5.13213575e-02 2.52493620e-01] | [9.537734985351562, 8.21106243133545] |
50e0ce7d-31fc-44f7-b72e-22439b9e9b90 | regularized-two-branch-proposal-networks-for | 2008.08257 | null | https://arxiv.org/abs/2008.08257v1 | https://arxiv.org/pdf/2008.08257v1.pdf | Regularized Two-Branch Proposal Networks for Weakly-Supervised Moment Retrieval in Videos | Video moment retrieval aims to localize the target moment in an video according to the given sentence. The weak-supervised setting only provides the video-level sentence annotations during training. Most existing weak-supervised methods apply a MIL-based framework to develop inter-sample confrontment, but ignore the intra-sample confrontment between moments with semantically similar contents. Thus, these methods fail to distinguish the target moment from plausible negative moments. In this paper, we propose a novel Regularized Two-Branch Proposal Network to simultaneously consider the inter-sample and intra-sample confrontments. Concretely, we first devise a language-aware filter to generate an enhanced video stream and a suppressed video stream. We then design the sharable two-branch proposal module to generate positive proposals from the enhanced stream and plausible negative proposals from the suppressed one for sufficient confrontment. Further, we apply the proposal regularization to stabilize the training process and improve model performance. The extensive experiments show the effectiveness of our method. Our code is released at here. | ['Zhijie Lin', 'Jieming Zhu', 'Zhou Zhao', 'Zhu Zhang', 'Xiuqiang He'] | 2020-08-19 | null | null | null | null | ['moment-retrieval'] | ['computer-vision'] | [ 1.44896641e-01 -1.55020609e-01 -5.62560976e-01 -6.16729736e-01
-1.14317930e+00 -3.47022325e-01 7.06368744e-01 -5.28684966e-02
-6.47786558e-01 4.00365561e-01 3.31800967e-01 6.54814467e-02
9.24737528e-02 -2.60025740e-01 -6.26421571e-01 -7.37486601e-01
1.26337647e-01 6.32818416e-02 2.96711415e-01 -3.37148495e-02
2.68127918e-01 1.03774341e-02 -1.25344527e+00 6.17336214e-01
8.39564741e-01 9.93296206e-01 5.22686243e-01 5.02923310e-01
-2.73300111e-01 1.06357348e+00 -4.80546534e-01 -1.74156040e-01
2.32430473e-01 -7.15080321e-01 -4.29880291e-01 5.26073337e-01
5.36319196e-01 -4.64183748e-01 -4.18329149e-01 1.16592669e+00
5.27630389e-01 3.85111898e-01 4.75986838e-01 -1.12417591e+00
-3.16858053e-01 6.35380268e-01 -7.64761150e-01 5.05814373e-01
4.87210065e-01 2.85098195e-01 8.06462824e-01 -1.22456729e+00
7.80128896e-01 1.30577004e+00 1.26311451e-01 6.46130741e-01
-7.88909912e-01 -6.23496950e-01 7.42815733e-01 2.59077132e-01
-1.31209850e+00 -5.94147980e-01 1.35654199e+00 -1.11573555e-01
4.41177040e-01 1.78720132e-01 7.37854064e-01 1.36262488e+00
3.39196362e-02 1.49521089e+00 7.76580215e-01 -2.56779909e-01
1.12859689e-01 1.27933815e-01 -7.94542059e-02 7.78504729e-01
-4.01944488e-01 -3.00818324e-01 -6.61413968e-01 1.24631599e-01
6.41825497e-01 8.81279409e-02 -4.37166750e-01 -2.31750786e-01
-1.40126705e+00 7.29344249e-01 9.87901762e-02 3.60845178e-01
-2.81419307e-01 2.32576067e-03 5.78387618e-01 2.23993570e-01
7.84184754e-01 3.31039019e-02 -3.99544418e-01 -6.82387576e-02
-1.40367699e+00 1.02396138e-01 4.05377686e-01 1.17205155e+00
6.56654775e-01 5.90596125e-02 -6.05925918e-01 7.73597836e-01
3.76894146e-01 2.77138889e-01 5.83265483e-01 -8.52369010e-01
8.88726056e-01 2.15317681e-01 5.26443906e-02 -1.16100419e+00
-1.00652613e-01 -6.26030684e-01 -6.71778619e-01 -5.36549032e-01
1.89862013e-01 -1.15197800e-01 -7.61141479e-01 1.68137658e+00
3.92847002e-01 6.64062381e-01 -1.94520224e-02 1.22002113e+00
8.89554262e-01 1.07151866e+00 6.13791160e-02 -7.35001147e-01
9.86145139e-01 -1.42596149e+00 -8.06850731e-01 -2.66713589e-01
7.46568024e-01 -8.19511771e-01 1.20956755e+00 3.52624476e-01
-1.20106399e+00 -5.52112699e-01 -9.27673459e-01 -1.48413241e-01
2.49631360e-01 4.47381377e-01 2.98115224e-01 1.33942857e-01
-6.17328465e-01 2.56191194e-01 -7.09844172e-01 -7.14115500e-02
2.26105139e-01 2.51304507e-02 -1.79204598e-01 -1.02516606e-01
-1.24749935e+00 4.26305383e-01 4.79758441e-01 3.56555998e-01
-1.01333439e+00 -4.25768882e-01 -1.00248277e+00 -1.23829812e-01
6.26378417e-01 -5.02286315e-01 1.25869274e+00 -1.50272298e+00
-1.41145492e+00 7.55624115e-01 -3.40718031e-01 -4.71978366e-01
7.61461198e-01 -1.45467252e-01 -3.47759336e-01 6.57478571e-01
2.55735636e-01 1.00107753e+00 1.20918202e+00 -1.13768744e+00
-7.71869302e-01 1.16082296e-01 1.81690454e-01 5.46508193e-01
-5.48322082e-01 1.00846082e-01 -1.12378526e+00 -1.08593440e+00
3.00663710e-01 -6.14075899e-01 -2.18590677e-01 3.75451483e-02
-3.85143667e-01 -2.16092050e-01 8.61857533e-01 -3.51547718e-01
1.29911292e+00 -2.17554855e+00 7.47963637e-02 -1.82023481e-01
3.94778363e-02 7.42269605e-02 -4.80806202e-01 8.53297785e-02
-5.85683323e-02 -1.22614190e-01 -1.10337883e-01 -7.46790826e-01
-2.76013136e-01 7.12824464e-02 -4.66791660e-01 6.03996456e-01
4.20048922e-01 6.71329141e-01 -1.20215094e+00 -9.57923055e-01
3.03173244e-01 3.65175396e-01 -5.84281385e-01 3.43137115e-01
-3.45540911e-01 6.14529490e-01 -6.58240438e-01 4.77694124e-01
5.82467139e-01 -1.21698596e-01 1.33507829e-02 -4.49087560e-01
-7.90016502e-02 2.40760654e-01 -9.18696344e-01 2.14207649e+00
-2.34359846e-01 5.14139116e-01 3.68045978e-02 -1.01493502e+00
7.63883770e-01 2.53417432e-01 6.21810615e-01 -6.65648401e-01
1.48223683e-01 -5.23629077e-02 -3.85341555e-01 -7.67982900e-01
5.49773097e-01 -2.02732816e-01 -9.46029741e-03 2.96714455e-01
6.55923486e-02 -4.39551324e-02 3.59405726e-01 4.89573836e-01
6.45558357e-01 3.56101722e-01 8.04541409e-02 -7.03923330e-02
8.77869368e-01 -3.41686517e-01 8.49132657e-01 6.53987527e-01
-5.59609652e-01 8.38065386e-01 7.21166849e-01 -3.01569879e-01
-7.70714402e-01 -7.57741392e-01 2.90778726e-01 1.10473537e+00
5.32372534e-01 -4.46682930e-01 -6.79186940e-01 -1.10876346e+00
-6.60580277e-01 5.71016788e-01 -2.88623124e-01 -2.69143105e-01
-7.13507414e-01 -5.30178368e-01 1.62689865e-01 3.00057381e-01
5.27021945e-01 -9.43822920e-01 -3.65954816e-01 4.82086204e-02
-7.36392498e-01 -1.20322263e+00 -1.02294636e+00 -1.50233567e-01
-8.23726773e-01 -7.49146879e-01 -9.95358407e-01 -1.15059149e+00
8.62942576e-01 4.29260284e-01 9.14317429e-01 1.56692103e-01
2.58528680e-01 2.68574178e-01 -5.16208351e-01 8.93850550e-02
-7.41875991e-02 -9.44288895e-02 1.04296580e-01 2.01072931e-01
1.80227369e-01 -1.86116725e-01 -6.19799912e-01 2.54875034e-01
-9.30712819e-01 4.34867263e-01 5.13119698e-01 7.96001375e-01
8.83543730e-01 1.88406203e-02 5.46226263e-01 -5.57134330e-01
4.61422563e-01 -5.66706300e-01 -2.88584739e-01 2.05145508e-01
-1.44701242e-01 -3.54610127e-03 6.44516528e-01 -7.92719066e-01
-1.06558728e+00 2.30073422e-01 -9.82406437e-02 -8.95885587e-01
4.73759472e-02 5.58397889e-01 -3.20296079e-01 3.35468709e-01
2.21623003e-01 4.55662400e-01 -3.10924113e-01 -2.61866271e-01
3.67895275e-01 4.50214088e-01 5.47687054e-01 -6.12060010e-01
6.79733932e-01 5.93501985e-01 -4.40283746e-01 -7.32051075e-01
-1.29131806e+00 -6.81819379e-01 -3.77623200e-01 -6.20622873e-01
6.67159021e-01 -1.11421168e+00 -3.13425630e-01 2.95655787e-01
-1.25454521e+00 -3.09151709e-01 -1.11173987e-01 6.46059453e-01
-6.56842470e-01 8.03271055e-01 -7.02057004e-01 -8.08690548e-01
-3.65468949e-01 -1.16281211e+00 1.20441246e+00 1.61625966e-01
5.34385480e-02 -7.36313760e-01 -1.58627421e-01 3.62107277e-01
-9.22632664e-02 -1.90620244e-01 3.54364574e-01 -6.10077024e-01
-5.17191529e-01 -1.80895194e-01 -1.98366955e-01 2.92086959e-01
-9.10374615e-03 4.94700558e-02 -6.46090806e-01 -3.54628503e-01
3.61145735e-01 -4.34699148e-01 1.07122207e+00 4.80830222e-01
1.28256595e+00 -4.18018728e-01 -3.70924883e-02 6.15367889e-01
9.42931116e-01 -9.21165720e-02 4.71818686e-01 3.11540544e-01
6.57885551e-01 7.21309781e-01 1.16457343e+00 5.12998343e-01
2.44456127e-01 4.81643975e-01 2.40115806e-01 -3.48607711e-02
1.12512335e-01 -4.32317853e-01 6.86516821e-01 1.15779543e+00
2.79614449e-01 -4.62479621e-01 -4.97916400e-01 5.52812576e-01
-2.18664193e+00 -1.17857695e+00 7.53483474e-02 2.00295806e+00
8.43048096e-01 5.81278145e-01 -3.96055728e-02 -1.41350701e-01
8.64859462e-01 6.45035028e-01 -4.54173476e-01 1.97229773e-01
-2.02935725e-01 -4.02411640e-01 9.87712219e-02 5.57769835e-01
-1.33909941e+00 1.04838550e+00 5.49275970e+00 1.33142495e+00
-1.25611711e+00 8.86734426e-02 9.11902130e-01 -4.45004851e-01
-4.06030715e-01 3.86802256e-02 -7.47376800e-01 6.46021068e-01
7.46543467e-01 -2.47583706e-02 -4.56455536e-02 8.35777462e-01
8.04967284e-01 -1.39717400e-01 -9.95458186e-01 1.14193070e+00
4.78120089e-01 -1.26709390e+00 2.59397745e-01 -5.47211707e-01
6.20141566e-01 -2.37573728e-01 1.07088722e-01 3.79545629e-01
-6.00289285e-01 -3.56856704e-01 1.16970170e+00 6.87376380e-01
6.29466057e-01 -8.57604861e-01 4.77160811e-01 6.33569956e-01
-1.36252809e+00 6.57804608e-02 -3.39344889e-01 6.77594021e-02
5.17860472e-01 6.48327231e-01 -4.86179858e-01 5.29842257e-01
3.84631723e-01 1.16220486e+00 -4.38379318e-01 9.57414567e-01
-4.44636583e-01 5.53896546e-01 -1.78874910e-01 1.13012560e-01
5.08580625e-01 -2.20678777e-01 7.74768174e-01 1.34368491e+00
2.44708091e-01 4.06570770e-02 5.90524197e-01 4.85943466e-01
-1.48649178e-02 2.08771735e-01 -3.05058628e-01 1.01207517e-01
2.63790905e-01 1.29933321e+00 -8.31870675e-01 -7.22757161e-01
-4.96712208e-01 1.15999269e+00 2.08235934e-01 5.55966258e-01
-1.11791790e+00 -1.56112373e-01 8.52510929e-02 -5.52088171e-02
2.22442552e-01 -9.99940410e-02 1.01962402e-01 -1.78081989e+00
1.90326542e-01 -8.44839156e-01 4.42742169e-01 -9.37648773e-01
-1.36162138e+00 4.92047846e-01 8.50140899e-02 -1.72128427e+00
-2.95681268e-01 -5.08936644e-02 -7.90260732e-01 4.32281286e-01
-1.48968792e+00 -1.02922201e+00 -8.04255903e-02 7.22046793e-01
1.26687837e+00 6.33595958e-02 1.25091732e-01 3.70311558e-01
-7.78088808e-01 5.42271137e-01 -1.77681997e-01 2.49087751e-01
9.87605214e-01 -9.10050750e-01 -1.38510495e-01 1.17230797e+00
6.07599914e-02 6.58743978e-01 8.42545211e-01 -7.37908483e-01
-1.12539685e+00 -1.26374578e+00 8.95317495e-01 -7.62683675e-02
6.13278866e-01 -3.43021184e-01 -8.32737088e-01 5.35782933e-01
1.49535581e-01 1.71751246e-01 3.13075602e-01 -4.04569626e-01
-2.49946818e-01 -3.07762921e-01 -8.59589696e-01 7.74332881e-01
7.60777056e-01 -6.14982426e-01 -7.66904533e-01 5.47265410e-01
9.44535255e-01 -4.69114929e-01 -3.07249010e-01 3.36304665e-01
3.53634804e-01 -9.02929008e-01 7.80951381e-01 -4.38340157e-01
7.29003012e-01 -4.17983949e-01 9.20089185e-02 -7.94840276e-01
5.09334132e-02 -8.55041802e-01 -4.16503042e-01 1.26762426e+00
1.97882548e-01 8.03073775e-03 8.62332046e-01 3.92022908e-01
-3.90789628e-01 -8.71994436e-01 -7.68461406e-01 -7.13307917e-01
-2.45281532e-01 -7.29935229e-01 6.20346144e-02 9.38197315e-01
1.90860614e-01 3.71364951e-01 -8.47712636e-01 -1.43994233e-02
6.04933262e-01 1.62572116e-01 5.96086502e-01 -4.34654057e-01
-3.79189461e-01 -2.95576900e-01 -1.74386680e-01 -1.69345438e+00
3.92766684e-01 -6.54698849e-01 4.35432971e-01 -1.20490074e+00
5.76880097e-01 -4.67209183e-02 -4.50552583e-01 1.43588185e-01
-3.24871421e-01 2.50675023e-01 2.96867728e-01 4.61408466e-01
-1.24636889e+00 7.73302853e-01 1.40220082e+00 -2.10962206e-01
-3.11120182e-01 -1.37601361e-01 -3.56710196e-01 8.98562789e-01
5.77816904e-01 -5.06824911e-01 -7.09972203e-01 -4.32010412e-01
3.51300985e-02 1.96470320e-01 1.90095514e-01 -9.04735088e-01
2.52588183e-01 -3.18334639e-01 3.58141452e-01 -9.68222082e-01
3.86099666e-01 -4.42661464e-01 -4.09889251e-01 3.74750465e-01
-7.32146800e-01 -1.08506344e-01 -1.23740748e-01 8.91321361e-01
-3.82793993e-01 -3.27374548e-01 6.67904079e-01 -7.21129104e-02
-7.75077701e-01 5.98992944e-01 -4.12513465e-01 2.42132396e-02
1.00893795e+00 -5.49847633e-02 1.05802398e-02 -6.88788295e-01
-6.27116978e-01 5.21215200e-01 3.64044785e-01 4.56544101e-01
8.63529265e-01 -1.31171513e+00 -5.58919430e-01 1.54586658e-02
7.59693831e-02 1.12465080e-02 4.79196280e-01 1.00723803e+00
-2.04640463e-01 1.39595896e-01 1.76862687e-01 -6.57062292e-01
-1.17803943e+00 7.96331644e-01 1.02038570e-01 -2.15521708e-01
-5.64457238e-01 9.54173386e-01 5.23254633e-01 1.50958952e-02
4.58515286e-01 -3.17519218e-01 -3.45342427e-01 2.04384282e-01
7.05948234e-01 3.98586597e-03 -3.93792123e-01 -8.52600098e-01
-2.71293789e-01 4.28051859e-01 -3.45561802e-01 -2.61043727e-01
1.00148296e+00 -5.14456630e-01 4.01628762e-02 5.96018612e-01
1.43548322e+00 -5.28065711e-02 -1.57913876e+00 -1.80591851e-01
-1.49825960e-01 -5.41390598e-01 7.41207749e-02 -1.93895951e-01
-1.10597599e+00 8.00178707e-01 2.93654293e-01 -1.69033065e-01
1.32890034e+00 -4.37247343e-02 9.70422328e-01 3.28432798e-01
1.08524799e-01 -1.23903954e+00 5.14097810e-01 5.44569016e-01
7.97400057e-01 -1.20824182e+00 1.37989512e-02 -4.44678336e-01
-7.79769361e-01 1.02631438e+00 7.84028709e-01 -1.31470770e-01
3.59355718e-01 -8.95402730e-02 2.92997323e-02 5.86948991e-02
-1.00707078e+00 -3.61433178e-02 3.72545511e-01 1.93859562e-01
4.82151151e-01 -4.69448656e-01 -6.36101186e-01 6.75384641e-01
4.02713537e-01 2.67715901e-02 5.51412880e-01 9.76949036e-01
-4.54344779e-01 -8.41415405e-01 -2.34395757e-01 3.06228697e-01
-5.03921390e-01 -2.12233275e-01 -1.32840931e-01 3.51923674e-01
7.80561268e-02 9.01514649e-01 1.19055085e-01 -2.39628389e-01
4.53403629e-02 -1.39052555e-01 3.12439770e-01 -6.33538067e-01
-4.69627172e-01 8.40263546e-01 -7.24048465e-02 -6.31236196e-01
-7.36803293e-01 -5.54185867e-01 -1.36623645e+00 1.18287221e-01
-4.22706425e-01 2.58680761e-01 1.82845280e-01 1.13953328e+00
1.98860601e-01 2.84399092e-01 8.32286775e-01 -1.15504956e+00
-3.46731097e-01 -8.75805795e-01 -3.73871952e-01 4.35519129e-01
4.38972980e-01 -3.58358324e-01 -5.50036073e-01 2.98366994e-01] | [9.978055953979492, 0.6155602335929871] |
b544afb6-e6a2-441e-9c72-f3fe52b61dc8 | efficient-attention-free-video-shift | 2208.11108 | null | https://arxiv.org/abs/2208.11108v1 | https://arxiv.org/pdf/2208.11108v1.pdf | Efficient Attention-free Video Shift Transformers | This paper tackles the problem of efficient video recognition. In this area, video transformers have recently dominated the efficiency (top-1 accuracy vs FLOPs) spectrum. At the same time, there have been some attempts in the image domain which challenge the necessity of the self-attention operation within the transformer architecture, advocating the use of simpler approaches for token mixing. However, there are no results yet for the case of video recognition, where the self-attention operator has a significantly higher impact (compared to the case of images) on efficiency. To address this gap, in this paper, we make the following contributions: (a) we construct a highly efficient \& accurate attention-free block based on the shift operator, coined Affine-Shift block, specifically designed to approximate as closely as possible the operations in the MHSA block of a Transformer layer. Based on our Affine-Shift block, we construct our Affine-Shift Transformer and show that it already outperforms all existing shift/MLP--based architectures for ImageNet classification. (b) We extend our formulation in the video domain to construct Video Affine-Shift Transformer (VAST), the very first purely attention-free shift-based video transformer. (c) We show that VAST significantly outperforms recent state-of-the-art transformers on the most popular action recognition benchmarks for the case of models with low computational and memory footprint. Code will be made available. | ['Georgios Tzimiropoulos', 'Brais Martinez', 'Adrian Bulat'] | 2022-08-23 | null | null | null | null | ['video-recognition'] | ['computer-vision'] | [ 5.60101151e-01 -6.76622894e-03 -1.53694987e-01 -2.01523498e-01
-6.47579670e-01 -2.91006118e-01 5.70782840e-01 -2.77101696e-01
-6.81996226e-01 3.17489266e-01 2.32292473e-01 -4.03387338e-01
-1.58534855e-01 -5.93316436e-01 -1.14870071e+00 -7.48537362e-01
1.78430393e-01 2.47120902e-01 4.32876408e-01 -1.65961549e-01
3.94343108e-01 4.25064981e-01 -1.86402321e+00 7.90534139e-01
4.14898068e-01 1.47883308e+00 2.66444594e-01 7.55853236e-01
-1.05042897e-01 1.49708164e+00 -4.58716959e-01 -5.91983080e-01
3.30099136e-01 -3.80469143e-01 -1.13182187e+00 -2.71345731e-02
7.95327306e-01 -4.59507376e-01 -5.40209293e-01 7.61194825e-01
4.96015370e-01 2.59305965e-02 3.47856313e-01 -1.27607524e+00
-4.81121957e-01 8.33441436e-01 -3.04940224e-01 5.61187506e-01
1.56925291e-01 6.10094182e-02 1.11085665e+00 -8.52281392e-01
5.22668719e-01 1.10846591e+00 7.68358827e-01 4.95969504e-01
-9.64938164e-01 -5.89792073e-01 3.47079575e-01 8.59387577e-01
-1.28198123e+00 -6.58706427e-01 6.14357471e-01 -2.36226633e-01
1.74451411e+00 3.03646237e-01 8.27609837e-01 1.18273389e+00
7.87493736e-02 1.25087249e+00 8.81352007e-01 -5.35367668e-01
5.66109829e-02 -2.05157280e-01 1.53908044e-01 7.87527263e-01
-1.10395946e-01 -1.27170220e-01 -6.96751773e-01 3.92234296e-01
4.78618979e-01 7.11445836e-03 -3.43986601e-01 -4.46600944e-01
-1.06058383e+00 6.29297495e-01 4.15828496e-01 6.58335745e-01
-4.32730883e-01 5.68890393e-01 7.43382871e-01 4.07513618e-01
3.81113201e-01 1.56980425e-01 -3.52244586e-01 -5.74737787e-01
-1.05027449e+00 1.89355910e-01 7.18421936e-01 8.95153284e-01
3.33616763e-01 3.51816595e-01 -2.25390017e-01 6.34461820e-01
-1.26820076e-02 1.38947546e-01 6.59344733e-01 -8.75045538e-01
5.50217986e-01 5.38015008e-01 -2.70644277e-01 -7.29697943e-01
-1.53739616e-01 -4.93917763e-01 -7.03793406e-01 5.26490845e-02
4.17251766e-01 3.59305680e-01 -7.74595320e-01 1.57424366e+00
-9.27729346e-03 3.91791672e-01 -2.28171721e-02 6.57707214e-01
6.08200550e-01 4.92291808e-01 1.13529965e-01 -8.91300812e-02
1.44612825e+00 -1.23150885e+00 -5.90353847e-01 -3.68888602e-02
6.55369699e-01 -5.55702209e-01 9.89042163e-01 5.90788782e-01
-1.40311170e+00 -6.53725028e-01 -1.12728882e+00 -2.61628687e-01
-4.60109591e-01 1.18458476e-02 6.49630666e-01 7.78546512e-01
-1.37903297e+00 7.04870939e-01 -8.99720848e-01 -4.95327294e-01
5.44387937e-01 5.29523671e-01 -3.53632152e-01 6.53512701e-02
-9.72213686e-01 1.00987434e+00 3.24106038e-01 1.56559274e-01
-9.08271194e-01 -6.94484830e-01 -8.41700971e-01 3.81570488e-01
5.21355391e-01 -6.21220052e-01 1.42565620e+00 -1.53257430e+00
-1.55265737e+00 7.95086682e-01 -1.68890327e-01 -1.14556599e+00
4.73268777e-01 -4.51443285e-01 -1.12685770e-01 2.13473707e-01
-3.01414728e-01 8.79025221e-01 1.03877449e+00 -7.36352503e-01
-9.04922545e-01 -1.05332866e-01 5.42946815e-01 9.37588140e-02
-4.51045990e-01 1.24038070e-01 -5.40332735e-01 -1.01163089e+00
-4.79074448e-01 -8.56914997e-01 2.20959395e-01 -1.72681689e-01
2.67657191e-02 -1.65470630e-01 9.39235806e-01 -5.23939669e-01
1.53185439e+00 -2.31814766e+00 3.27669859e-01 -2.39410460e-01
6.93722218e-02 6.66741550e-01 -1.58289328e-01 1.81622356e-01
-5.36579072e-01 -1.39413215e-02 -2.72508621e-01 -4.49408829e-01
5.28478287e-02 4.10325646e-01 -4.73483413e-01 3.72194707e-01
2.25707948e-01 9.54493403e-01 -6.78595483e-01 -4.17750448e-01
3.95104259e-01 4.66851950e-01 -9.27354157e-01 -7.62110278e-02
7.78799085e-03 -2.08313331e-01 -1.70847669e-01 6.38329506e-01
3.00900996e-01 -1.46258131e-01 1.30191833e-01 -5.77005625e-01
-1.90496206e-01 2.75236368e-01 -9.63915586e-01 1.43826318e+00
-4.09043759e-01 8.16038787e-01 -8.72960389e-02 -1.51595497e+00
3.70727658e-01 4.92683858e-01 7.29636371e-01 -9.00394857e-01
3.03839713e-01 1.69287756e-01 1.53629944e-01 -3.98706168e-01
6.66274548e-01 -5.38760647e-02 2.13388905e-01 2.28751093e-01
4.23624396e-01 3.12951475e-01 3.39985400e-01 1.14814748e-05
1.36845708e+00 3.25248212e-01 2.72774994e-01 -2.63375312e-01
8.26370597e-01 -1.52495146e-01 2.06197008e-01 7.17395425e-01
-2.97230780e-01 4.42970186e-01 5.46124756e-01 -7.40980625e-01
-1.15920866e+00 -6.53347731e-01 5.78968972e-02 1.27348924e+00
-3.12048346e-01 -5.38279355e-01 -1.10923493e+00 -7.07549870e-01
-1.55913725e-01 4.46794689e-01 -6.53809190e-01 -1.13860525e-01
-9.26014543e-01 -6.74213588e-01 8.63431394e-01 8.90189767e-01
8.13040078e-01 -1.07003391e+00 -9.54572737e-01 3.01815033e-01
-2.67772168e-01 -1.20366895e+00 -5.19233227e-01 4.71373409e-01
-8.04611802e-01 -1.02922952e+00 -6.90215766e-01 -7.08572030e-01
2.84009010e-01 9.56295952e-02 9.83144104e-01 1.75490826e-02
2.02189982e-02 6.46010101e-01 -5.31462014e-01 -2.65447587e-01
-8.47468078e-02 2.91542768e-01 -1.94670141e-01 2.08512560e-01
3.37571949e-01 -4.11163181e-01 -4.33568150e-01 2.67948389e-01
-9.95598435e-01 9.96993110e-02 6.69210076e-01 1.00788271e+00
4.69234973e-01 -1.06078908e-01 1.46656424e-01 -6.24209583e-01
2.77359277e-01 -1.76479012e-01 -4.58366275e-01 2.47989461e-01
-4.29491460e-01 3.10951620e-01 6.08522296e-01 -4.09548461e-01
-6.47924244e-01 8.41953680e-02 -3.63053322e-01 -7.99259722e-01
1.02875762e-01 2.83279061e-01 -8.15510470e-03 -4.39532101e-01
2.77884007e-01 4.06258970e-01 -6.99174106e-02 -4.94304061e-01
1.16382666e-01 5.34481168e-01 4.76369351e-01 -4.22250420e-01
2.35894874e-01 4.28624362e-01 -2.35146880e-02 -7.44327128e-01
-5.80398858e-01 -3.41014892e-01 -3.65032792e-01 -3.17476958e-01
8.89956117e-01 -8.15894783e-01 -8.62536192e-01 7.08254099e-01
-1.02061832e+00 -4.46275890e-01 -5.39705276e-01 3.77448410e-01
-8.94546747e-01 3.96592051e-01 -7.79961646e-01 -6.19457245e-01
-3.93438905e-01 -1.55874300e+00 9.66696501e-01 -2.23150775e-01
-5.85134961e-02 -6.60612047e-01 -2.74845988e-01 4.21186596e-01
5.88919580e-01 -8.89172703e-02 7.87896037e-01 -7.27489829e-01
-7.49696076e-01 2.25486010e-01 -2.72119224e-01 6.82630897e-01
-3.43704641e-01 -1.70040503e-01 -1.15479827e+00 -2.18305096e-01
1.09898940e-01 -2.44097799e-01 1.08897305e+00 4.46515113e-01
1.29402494e+00 -2.97690213e-01 -1.35082126e-01 7.18538284e-01
1.39639318e+00 3.21238130e-01 9.79524851e-01 5.04424989e-01
6.34118080e-01 1.84167489e-01 2.89639711e-01 2.86354572e-01
3.51190388e-01 1.02753890e+00 4.48024094e-01 8.67797881e-02
-3.61209393e-01 3.59646417e-02 7.31176138e-01 9.96481419e-01
-4.55657750e-01 -3.94922704e-01 -7.36473083e-01 6.44390583e-01
-1.86122096e+00 -1.20994818e+00 1.52711883e-01 2.10584426e+00
4.13889349e-01 3.69560182e-01 1.98530316e-01 6.43892407e-01
4.58063424e-01 3.08570176e-01 -1.19074591e-01 -7.92189658e-01
-2.01685000e-02 7.04783261e-01 7.89178193e-01 5.61189234e-01
-1.27479565e+00 9.07301426e-01 6.44217348e+00 1.12987936e+00
-1.22176623e+00 2.51753688e-01 4.63465840e-01 -2.59120226e-01
1.29190192e-01 -2.72989392e-01 -8.39489818e-01 4.54649776e-01
1.23898041e+00 1.67120129e-01 5.34323037e-01 9.41548228e-01
-2.01937899e-01 1.04169445e-02 -1.24343097e+00 1.16355693e+00
3.41833413e-01 -1.26159763e+00 2.84430772e-01 -1.32922260e-02
1.37493342e-01 1.63294464e-01 4.73549999e-02 4.81925189e-01
2.10775319e-03 -8.62927914e-01 1.16974580e+00 4.37022567e-01
6.26981676e-01 -5.93191922e-01 8.01578641e-01 -7.95804784e-02
-1.54342866e+00 -4.22057599e-01 -8.95389691e-02 -8.55016410e-02
1.08788230e-01 1.49009869e-01 -4.90963608e-01 5.55936098e-01
9.48316634e-01 8.50778162e-01 -4.44851160e-01 8.80196452e-01
1.86406538e-01 5.76567769e-01 -1.44637421e-01 7.94836059e-02
5.66059530e-01 2.54852057e-01 3.39898080e-01 1.48558569e+00
3.58243972e-01 3.09481602e-02 -2.90659368e-01 2.08120644e-01
-1.27123684e-01 -1.23942330e-01 -4.57441956e-01 -3.96003462e-02
-5.40163293e-02 7.30594337e-01 -5.71865380e-01 -6.02977693e-01
-6.04364038e-01 1.03983974e+00 4.13085669e-01 7.02301711e-02
-1.21952510e+00 -2.72942096e-01 6.01825655e-01 1.43433779e-01
9.72193897e-01 1.88376550e-02 1.10357821e-01 -1.11112249e+00
1.69015765e-01 -1.28346348e+00 4.83436614e-01 -7.99048960e-01
-7.25231290e-01 7.83412099e-01 5.49068637e-02 -1.09309363e+00
-3.15140575e-01 -9.00908172e-01 3.42550725e-02 3.37014109e-01
-1.59404731e+00 -1.02378750e+00 -5.35400324e-02 8.65482926e-01
7.91154146e-01 -1.11568250e-01 7.01771796e-01 8.32343042e-01
-5.27583301e-01 7.15098858e-01 -1.71023533e-01 1.47575825e-01
4.03807968e-01 -9.91098642e-01 3.70441437e-01 8.42578828e-01
2.61370122e-01 2.87232548e-01 4.98684675e-01 -5.69971725e-02
-1.57183266e+00 -8.34083021e-01 9.92418766e-01 -3.84657204e-01
6.55472040e-01 -3.31066132e-01 -6.90755367e-01 9.39789355e-01
3.00212711e-01 1.24167189e-01 3.09823692e-01 -2.46313438e-01
-4.83759969e-01 -4.30885494e-01 -8.94798279e-01 4.68906105e-01
1.27832472e+00 -6.34175837e-01 -6.43717885e-01 -2.90031694e-02
4.87887383e-01 -3.82351160e-01 -8.06885958e-01 4.64218408e-01
8.52249861e-01 -1.13974142e+00 1.07310522e+00 -4.83532846e-01
3.51352066e-01 -2.13837400e-01 -3.84773046e-01 -7.89288223e-01
-3.29650789e-01 -4.60388303e-01 -3.67616564e-01 9.78593469e-01
9.70866755e-02 -5.50056458e-01 8.20640564e-01 2.69420207e-01
-4.82720673e-01 -9.61626351e-01 -1.30010223e+00 -8.21152508e-01
-2.58262992e-01 -7.58713305e-01 3.91084790e-01 6.41104043e-01
8.26441273e-02 8.63290653e-02 -4.91860837e-01 -3.18164617e-01
1.47728145e-01 -1.59250528e-01 5.49367309e-01 -6.18576527e-01
-5.54011941e-01 -7.74467111e-01 -7.20654190e-01 -1.33957648e+00
6.52413815e-02 -7.24028885e-01 -1.05147719e-01 -1.15835440e+00
1.26198828e-01 -5.33198658e-03 -4.12770540e-01 7.05599666e-01
1.77671820e-01 2.77692407e-01 4.36702192e-01 8.38379562e-02
-7.23082721e-01 3.83241653e-01 8.41448307e-01 -2.28611186e-01
4.01089340e-01 -4.00529206e-01 -4.11672711e-01 6.51863813e-01
4.76051301e-01 -3.41495425e-01 -2.98031807e-01 -7.10080445e-01
1.34655496e-03 -2.32861087e-01 4.19183433e-01 -1.50280690e+00
3.17259431e-01 2.92028487e-01 -2.23669354e-02 -3.59089136e-01
4.77542043e-01 -1.03154778e+00 1.35233253e-01 6.95534885e-01
-2.55168378e-01 4.85195458e-01 3.22514504e-01 2.16341794e-01
-4.02099937e-01 -1.59973994e-01 6.98081970e-01 -8.71896744e-02
-9.59790289e-01 1.39746398e-01 -5.62684298e-01 -5.92807680e-02
9.97307837e-01 -4.77291316e-01 -2.78207272e-01 -1.43161967e-01
-5.68593740e-01 -2.90488064e-01 1.66116133e-01 3.29456300e-01
4.83305812e-01 -1.08944952e+00 -4.77705568e-01 1.73702762e-01
-7.77362585e-02 -4.66245800e-01 2.91308999e-01 1.14032423e+00
-6.68855488e-01 7.92219758e-01 -4.51460689e-01 -4.58497941e-01
-1.39217889e+00 8.98064911e-01 5.56378603e-01 -5.49515486e-01
-5.38386345e-01 8.16830099e-01 2.05052435e-01 1.25037462e-01
4.91874874e-01 -8.44604611e-01 -5.99144846e-02 1.32354200e-01
6.07787848e-01 4.13665950e-01 3.91856760e-01 -8.15815806e-01
-4.83462334e-01 6.53622746e-01 -7.69477338e-02 2.04190817e-02
1.36363983e+00 3.06575716e-01 5.19285537e-02 2.29000285e-01
1.14648879e+00 -3.40885520e-01 -1.08155549e+00 -7.83126950e-02
-1.47404790e-01 -2.91956812e-01 1.35151863e-01 -4.72832769e-01
-1.36298645e+00 8.88661563e-01 7.51553655e-01 2.28690922e-01
1.43473518e+00 -3.01911533e-01 7.51300156e-01 3.94340962e-01
3.19637179e-01 -1.04773092e+00 -8.14609602e-02 7.07594752e-01
8.17671061e-01 -7.12120175e-01 -1.36381403e-01 -2.08991781e-01
-4.08629894e-01 1.01283932e+00 3.92026126e-01 -1.48712099e-01
5.66084087e-01 5.79549193e-01 -4.43709075e-01 -9.69859436e-02
-9.21017528e-01 -2.71727890e-01 1.47656530e-01 4.05598432e-01
4.92091268e-01 -3.30520779e-01 -3.10397595e-01 3.80619258e-01
9.16938577e-03 3.34432274e-01 1.44833311e-01 1.08593440e+00
-2.58924276e-01 -1.13582242e+00 -2.05788940e-01 3.62147450e-01
-7.77012527e-01 -4.10939813e-01 -1.63908377e-01 9.48465705e-01
3.54741067e-01 5.62086999e-01 2.03552708e-01 -4.93488699e-01
4.76783246e-01 3.32910448e-01 7.38640785e-01 -2.45946683e-02
-9.79702294e-01 -1.63675427e-01 2.46254787e-01 -9.12840962e-01
-7.97329128e-01 -6.08453989e-01 -7.06741631e-01 -5.32270074e-01
-2.31418625e-01 -4.51211073e-02 4.59152579e-01 1.02265275e+00
4.43603396e-01 7.60955513e-01 4.55814451e-02 -1.01353729e+00
-4.14592683e-01 -8.11697543e-01 -2.86287189e-01 3.60361189e-01
1.89639866e-01 -6.98455095e-01 -1.44618899e-01 2.38667130e-01] | [8.885666847229004, 0.4929862916469574] |
ac6be1ce-9719-4ee1-9bc5-3633867d2735 | context-aware-language-modeling-for-goal | null | null | https://openreview.net/forum?id=qZEAjNzBHv | https://openreview.net/pdf?id=qZEAjNzBHv | Context-Aware Language Modeling for Goal-Oriented Dialogue Systems | Goal-oriented dialogue systems has long faced the trade-off between fluent language generation and task-specific control. While supervised learning with large language models are capable of producing realistic responses, how to steer such responses towards completing a specific task without sacrificing language quality remains an open question. In this work, by viewing a goal-oriented dialogue system as a reinforcement learning (RL) problem, we turn a supervised language model into a dynamics model and a behavioral cloning policy in a partially observable Markov decision process. This view allows RL techniques such as task relabeling and goal-conditioned policy to be naturally adopted as a form of data augmentation and task-specific fintuning of language models. We evaluate our method, Context-Aware Language Models (\method), on a practical flight-booking task using AirDialogue. Empirically, \method outperforms the previous state-of-the-art method by more than 10\% in terms of task success, achieving human-level task performance on this dataset. | ['Anonymous'] | 2021-11-16 | null | null | null | acl-arr-november-2021-11 | ['goal-oriented-dialogue-systems'] | ['natural-language-processing'] | [ 3.40401322e-01 6.71873212e-01 -3.61481369e-01 -5.59912264e-01
-8.31286311e-01 -5.71030319e-01 9.03562844e-01 -7.07592815e-02
-5.53823054e-01 9.77725863e-01 4.56144571e-01 -4.88608181e-01
3.40718254e-02 -4.31133986e-01 -1.79279819e-01 -4.76332664e-01
7.91339502e-02 9.12643731e-01 -4.55395430e-02 -7.32846737e-01
2.50440747e-01 2.42960647e-01 -9.97754753e-01 2.65259564e-01
9.32322800e-01 5.76021433e-01 3.65350902e-01 1.05385590e+00
-2.18796730e-01 1.43221283e+00 -5.29076755e-01 1.31174281e-01
8.90109837e-02 -7.85647929e-01 -1.33032954e+00 4.36915666e-01
-1.81255043e-02 -2.61764109e-01 -2.75348648e-02 6.40258074e-01
4.33589339e-01 6.43078923e-01 6.28840625e-01 -1.10982335e+00
-2.36097053e-01 5.92370272e-01 -7.27552399e-02 -9.53628793e-02
4.15425301e-01 6.02636099e-01 8.61014545e-01 -2.99899727e-01
5.57722688e-01 1.56698501e+00 2.04701032e-02 1.19772470e+00
-1.69263613e+00 -2.62404919e-01 4.35598552e-01 -4.48417813e-01
-8.79823029e-01 -6.05349898e-01 6.86204672e-01 -7.57283986e-01
1.18229091e+00 -2.57630516e-02 4.03967887e-01 1.15797174e+00
3.03958714e-01 8.07672918e-01 1.53587925e+00 -4.26681161e-01
4.24203694e-01 3.30451459e-01 1.96944289e-02 7.23038673e-01
-4.20285344e-01 3.80426347e-01 -6.72064602e-01 -2.71892190e-01
6.83832943e-01 -4.71297860e-01 -1.51152745e-01 -4.55624789e-01
-1.25265610e+00 1.09362054e+00 7.07571208e-02 1.19071342e-01
-6.98235631e-01 1.94146499e-01 4.74076986e-01 6.52676344e-01
5.40221810e-01 1.07130718e+00 -6.92564011e-01 -3.16392452e-01
-5.05499661e-01 6.31365478e-01 1.14245534e+00 7.69236028e-01
6.56777442e-01 4.38622952e-01 -7.13158786e-01 9.67795193e-01
2.27034956e-01 3.35049093e-01 5.43383956e-01 -1.15165222e+00
4.08840358e-01 4.83778387e-01 4.75968063e-01 -4.00972635e-01
-6.03213131e-01 -8.02557915e-03 -5.11020303e-01 3.45875323e-01
5.36674261e-01 -8.55341673e-01 -6.40501261e-01 2.05055213e+00
3.87821317e-01 -3.48484427e-01 4.85783458e-01 9.70477581e-01
2.39967376e-01 7.75177658e-01 5.15255928e-01 -4.73188549e-01
1.24048901e+00 -1.08739972e+00 -9.32242155e-01 -5.35374403e-01
8.55541229e-01 -4.79849428e-01 1.29404974e+00 6.41525924e-01
-1.04232526e+00 -6.11923754e-01 -6.13105774e-01 3.03518862e-01
6.57173470e-02 -2.48057753e-01 7.63961136e-01 4.07333374e-01
-1.07871068e+00 4.83046919e-01 -6.20016515e-01 -4.36513186e-01
-2.98729390e-01 3.21909875e-01 -1.90797701e-01 2.81377256e-01
-1.34165227e+00 1.08011281e+00 5.07556915e-01 -1.42398641e-01
-1.43168044e+00 -3.55103761e-01 -7.64731467e-01 -1.95199206e-01
9.41320181e-01 -8.00758481e-01 1.97292137e+00 -9.85290527e-01
-2.24192977e+00 8.19361746e-01 9.90061313e-02 -6.63888514e-01
4.50130045e-01 -3.92815530e-01 -2.32207790e-01 -1.42366827e-01
-3.01361401e-02 9.16841805e-01 1.03762746e+00 -1.28176057e+00
-8.35970283e-01 -1.66400567e-01 4.04730916e-01 6.71497166e-01
-5.96933328e-02 -3.10983229e-03 2.09279269e-01 -3.36746395e-01
-4.08413440e-01 -1.17215466e+00 -7.15991378e-01 -5.30800581e-01
-3.33607882e-01 -5.72188020e-01 4.06986028e-01 -4.75827843e-01
1.22015607e+00 -1.64136863e+00 4.74898338e-01 -4.19375837e-01
1.89160436e-01 1.68419823e-01 -2.43500754e-01 6.49036527e-01
8.05066824e-02 -4.51419577e-02 -2.49661475e-01 -4.08737302e-01
9.61325914e-02 2.58474976e-01 -3.75734895e-01 7.64066055e-02
2.51950681e-01 7.20425367e-01 -1.20798039e+00 -3.62222373e-01
2.02001736e-01 -3.04371655e-01 -7.52006650e-01 8.95590901e-01
-1.07327092e+00 9.49037135e-01 -6.48191392e-01 2.02217981e-01
-5.40189892e-02 1.02037258e-01 5.08662701e-01 5.64976811e-01
-1.58295408e-01 4.62695420e-01 -8.10831249e-01 1.90018392e+00
-8.96600962e-01 2.10818082e-01 3.46036047e-01 -8.07504237e-01
1.19112170e+00 4.58449513e-01 2.48442978e-01 -6.24701142e-01
-7.11781904e-02 -1.90425768e-01 3.42720002e-01 -6.60345316e-01
7.55100846e-01 -4.89737630e-01 -4.32444781e-01 7.98715174e-01
1.85068026e-01 -6.73528969e-01 9.32307765e-02 1.41055405e-01
8.89449954e-01 3.84534657e-01 3.09804350e-01 -5.09930670e-01
4.73978281e-01 3.54671836e-01 3.60660195e-01 9.18273330e-01
-3.74677032e-01 6.95059374e-02 6.60140455e-01 -4.75046426e-01
-7.92337477e-01 -4.73115355e-01 3.30551654e-01 1.75062692e+00
-3.63769174e-01 -1.68527573e-01 -9.53412712e-01 -8.20825636e-01
-2.75021106e-01 1.29916656e+00 -2.96886683e-01 -3.33568811e-01
-6.84240818e-01 -5.91339231e-01 3.94395471e-01 7.30405301e-02
5.12749791e-01 -1.61029637e+00 -6.85823023e-01 5.99506855e-01
-3.42042089e-01 -1.14508414e+00 -5.53870678e-01 5.45997083e-01
-8.00722897e-01 -7.77368784e-01 -5.44863105e-01 -5.31447232e-01
3.38666350e-01 -7.58994445e-02 1.35976303e+00 -2.25456700e-01
2.94638306e-01 5.97204030e-01 -3.13780040e-01 -4.03366119e-01
-1.06088269e+00 2.51506388e-01 3.14272225e-01 1.12277038e-01
1.98353857e-01 -2.76106656e-01 -3.83268207e-01 1.67475104e-01
-7.86690116e-01 3.24063003e-01 4.76584464e-01 1.12690973e+00
2.07960069e-01 -3.91537994e-01 8.20553422e-01 -1.07538879e+00
1.51558053e+00 -3.19720060e-01 -6.53118849e-01 1.89004689e-01
-9.60369289e-01 5.91318011e-01 8.61647069e-01 -5.95067143e-01
-1.40650403e+00 3.13668579e-01 1.15689598e-02 4.43561971e-02
-4.91912425e-01 3.50256056e-01 1.31826341e-01 2.51972467e-01
9.55060601e-01 5.16454220e-01 2.48548239e-01 -2.23951519e-01
5.77164054e-01 7.37548470e-01 2.04667509e-01 -9.66792941e-01
4.48703498e-01 -8.30398425e-02 -2.17542380e-01 -8.88145328e-01
-9.57920253e-01 -3.66260499e-01 -6.74117804e-01 -3.39744389e-01
9.29564118e-01 -8.88938546e-01 -7.73680747e-01 2.75822520e-01
-1.06851864e+00 -1.26762664e+00 -3.62817556e-01 2.31705666e-01
-1.20924306e+00 4.02151048e-02 -4.48900372e-01 -1.35565996e+00
-2.66306698e-01 -1.04912841e+00 9.31372821e-01 5.53918146e-02
-6.10786855e-01 -1.15195537e+00 4.59657907e-01 5.08520961e-01
4.29461122e-01 1.17745966e-01 9.55278933e-01 -9.65303004e-01
-2.33776927e-01 8.88433233e-02 3.36034834e-01 3.52064073e-01
1.72526911e-02 -5.31847894e-01 -7.94143260e-01 -3.85488689e-01
2.87761360e-01 -1.08135164e+00 3.56402397e-01 1.46342456e-01
6.18966877e-01 -6.06974959e-01 9.69781280e-02 9.49091185e-03
8.66343617e-01 3.96147728e-01 2.36983523e-01 1.50938466e-01
3.22478354e-01 1.05924952e+00 1.09008908e+00 6.61156178e-01
4.78017539e-01 8.37406039e-01 3.28949869e-01 9.39603522e-02
1.13610841e-01 -5.81773937e-01 6.10747516e-01 4.73434150e-01
2.66063362e-01 -3.26758742e-01 -1.08964789e+00 4.49648589e-01
-2.07997513e+00 -7.89465129e-01 1.10001624e-01 2.26888061e+00
1.11816633e+00 1.62324846e-01 4.12855864e-01 -3.87202829e-01
3.47512335e-01 3.50773275e-01 -6.42011464e-01 -8.71900856e-01
4.95101273e-01 -2.40833625e-01 1.15155429e-01 8.55118036e-01
-9.26810145e-01 1.52950037e+00 6.33639288e+00 5.64936459e-01
-1.05529928e+00 5.10338210e-02 6.96119547e-01 1.64764628e-01
4.11403812e-02 7.77664185e-02 -9.49286282e-01 8.18203166e-02
1.32043648e+00 -9.37812701e-02 8.28011930e-01 9.15174127e-01
6.87350512e-01 -3.27947170e-01 -1.23673618e+00 5.62720418e-01
-1.48084968e-01 -7.89810479e-01 -7.67766684e-02 5.06890677e-02
5.15966058e-01 -2.43306160e-01 -1.90563366e-01 9.48562860e-01
1.11851847e+00 -1.14909852e+00 6.40567660e-01 4.76396263e-01
6.08461678e-01 -6.19016945e-01 2.45072365e-01 1.09858501e+00
-5.72120607e-01 -1.11439168e-01 -1.59623995e-01 -3.68833363e-01
2.30153173e-01 -9.56131667e-02 -1.36863172e+00 2.36494258e-01
5.80644459e-02 2.12724909e-01 -1.21193476e-01 1.48751095e-01
-3.89146447e-01 7.33392537e-01 7.53497705e-02 -2.77118057e-01
5.65695882e-01 -2.67896682e-01 5.68162918e-01 1.12198186e+00
-3.95636857e-01 3.56012106e-01 8.64126921e-01 7.96521962e-01
2.31775537e-01 8.22166577e-02 -9.44240093e-01 -4.12978768e-01
1.28975987e-01 1.29892218e+00 -3.88106793e-01 -3.11930031e-01
-2.11212374e-02 9.12165403e-01 5.46319723e-01 3.23954940e-01
-3.88119906e-01 5.98681718e-02 7.87664711e-01 2.28097085e-02
-3.10581148e-01 -3.90664071e-01 -1.28001079e-01 -1.02049243e+00
-5.01626432e-01 -1.37414730e+00 1.92735791e-01 -6.07377648e-01
-8.36442351e-01 8.67136538e-01 1.87246371e-02 -9.77300525e-01
-9.60279226e-01 -4.49873120e-01 -2.26851508e-01 1.03250718e+00
-1.39772606e+00 -8.61466825e-01 2.46479828e-02 6.10520720e-01
1.14491785e+00 -3.92232299e-01 1.16149390e+00 -3.72430116e-01
-4.92115468e-01 7.76221156e-02 -3.78404438e-01 -7.53221959e-02
8.08545470e-01 -1.46337950e+00 2.26460621e-01 5.66066146e-01
-1.52053237e-01 3.73073131e-01 9.67424333e-01 -6.79506242e-01
-1.45028937e+00 -9.42188442e-01 9.25639331e-01 -5.63634396e-01
6.84864700e-01 -5.82002342e-01 -7.24956572e-01 5.27012825e-01
5.25345981e-01 -5.37940085e-01 4.37583148e-01 1.96580321e-01
1.73747405e-01 1.95227697e-01 -9.43925440e-01 9.17879045e-01
8.26489627e-01 -3.80116999e-01 -6.91783309e-01 7.58017838e-01
9.48650658e-01 -4.80426669e-01 -7.12074101e-01 6.62862137e-02
4.00155634e-02 -6.64041996e-01 6.13085330e-01 -1.19269812e+00
3.72909844e-01 9.08991098e-02 3.66158001e-02 -1.66927922e+00
-2.59101897e-01 -1.23361444e+00 2.93963458e-02 1.04022968e+00
4.17599946e-01 -4.58311170e-01 5.48293531e-01 9.18792963e-01
-1.17618859e-01 -5.53907812e-01 -4.95446503e-01 -6.68290675e-01
3.28425080e-01 -2.72827983e-01 1.57363907e-01 7.45489359e-01
3.45236272e-01 9.85712230e-01 -8.59858632e-01 -2.68325180e-01
2.31449530e-01 2.55447403e-02 1.15457964e+00 -9.87541318e-01
-3.97405356e-01 -2.29869470e-01 5.59493363e-01 -1.45570028e+00
6.41597271e-01 -7.26728141e-01 6.45767868e-01 -1.20475757e+00
-1.49970755e-01 -4.69176441e-01 1.14308812e-01 5.75800121e-01
-1.24318644e-01 -7.04219639e-01 3.81608129e-01 8.03643167e-02
-8.20694089e-01 7.46839523e-01 1.50113118e+00 5.30708656e-02
-6.98899865e-01 2.40474761e-01 -7.92862058e-01 5.86889505e-01
9.08331096e-01 -6.00838900e-01 -7.02939332e-01 -1.32052660e-01
-1.91626567e-02 1.12796497e+00 -4.56994176e-02 -6.86945796e-01
2.19748855e-01 -7.36514330e-01 -3.23935837e-01 4.34287712e-02
4.26750034e-01 -5.02324820e-01 -3.23296010e-01 6.72743499e-01
-1.11587572e+00 4.85268347e-02 1.25722706e-01 6.30290627e-01
-1.39911950e-01 -2.13019073e-01 8.25637937e-01 -3.92389476e-01
-8.20401490e-01 1.75542638e-01 -1.08470380e+00 4.42863047e-01
1.03556502e+00 1.17720395e-01 -8.77921376e-03 -7.55452752e-01
-7.92294204e-01 6.44849479e-01 1.28142284e-02 6.39932454e-01
4.16460276e-01 -8.87111127e-01 -9.36280966e-01 6.62809284e-03
1.60202578e-01 -4.56695743e-02 5.94807416e-02 5.14871657e-01
-1.09969169e-01 8.27887475e-01 -1.28300712e-01 -4.97009128e-01
-9.91607368e-01 4.93936718e-01 5.60041070e-01 -1.02417135e+00
-3.43136787e-01 6.06886983e-01 3.98598611e-01 -9.99271691e-01
1.77605048e-01 -1.15801118e-01 -4.66851711e-01 2.75841951e-02
2.80377328e-01 -5.54745495e-02 -2.07019523e-01 -2.47202337e-01
1.28328755e-01 -1.53525814e-01 -7.94664621e-02 -6.33730114e-01
9.09032583e-01 -2.49257997e-01 2.30990127e-01 6.82025492e-01
4.93819207e-01 -5.21538258e-01 -1.57153022e+00 -3.57510984e-01
3.28899652e-01 -5.30894101e-02 4.85187657e-02 -1.29053056e+00
-3.77227485e-01 7.96896100e-01 2.57235527e-01 4.41321492e-01
7.76883185e-01 -3.30285668e-01 4.22076106e-01 7.62197614e-01
5.60273468e-01 -1.21202350e+00 5.05056143e-01 9.25735235e-01
1.10466325e+00 -1.53577840e+00 -5.19935966e-01 7.62624815e-02
-1.43922520e+00 9.44551110e-01 1.00758719e+00 -1.62131190e-01
2.85835773e-01 1.02232277e-01 3.50230187e-01 -1.28957346e-01
-1.49736691e+00 -3.31913173e-01 8.37344453e-02 7.00289369e-01
5.66653252e-01 2.35722154e-01 -1.96725339e-01 4.12370682e-01
-1.68829232e-01 9.99951735e-02 4.58110243e-01 9.32855606e-01
-5.38084924e-01 -1.28944683e+00 -2.78753310e-01 2.11407423e-01
-2.61864483e-01 -1.74050983e-02 -6.99837863e-01 6.62254572e-01
-5.32326579e-01 1.27331221e+00 -2.46058106e-01 -2.97784358e-01
4.73537207e-01 5.36867499e-01 1.84183329e-01 -1.15267801e+00
-9.97880340e-01 1.08493470e-01 5.64831913e-01 -5.29721379e-01
-2.17620373e-01 -5.57057381e-01 -1.13829434e+00 -8.61891210e-02
-1.15354031e-01 3.48568708e-01 3.85175377e-01 9.90795612e-01
2.75971591e-01 6.98465526e-01 8.32702041e-01 -7.17218220e-01
-1.25256836e+00 -1.13009703e+00 -5.03748953e-01 3.44931990e-01
3.24084967e-01 -4.16792721e-01 -7.74747059e-02 5.92338033e-02] | [13.05444622039795, 8.065900802612305] |
2050b5c4-8774-4031-ac2e-11ff36b4c744 | data-aided-active-user-detection-with-a-user | 2205.10780 | null | https://arxiv.org/abs/2205.10780v2 | https://arxiv.org/pdf/2205.10780v2.pdf | Data-aided Active User Detection with a User Activity Extraction Network for Grant-free SCMA Systems | In grant-free sparse code multiple access (GF-SCMA) system, active user detection (AUD) is a major performance bottleneck as it involves complex combinatorial problem, which makes joint design of contention resources for users and AUD at the receiver a crucial but a challenging problem. To this end, we propose autoencoder (AE)-based joint optimization of both preamble generation networks (PGNs) in the encoder side and data-aided AUD in the decoder side. The core architecture of the proposed AE is a novel user activity extraction network (UAEN) in the decoder that extracts a priori user activity information from the SCMA codeword data for the data-aided AUD. An end-to-end training of the proposed AE enables joint optimization of the contention resources, i.e., preamble sequences, each associated with one of the codebooks, and extraction of user activity information from both preamble and SCMA-based data transmission. Furthermore, we propose a self-supervised pre-training scheme for the UAEN prior to the end-to-end training, to ensure the convergence of the UAEN which lies deep inside the AE network. Simulation results demonstrated that the proposed AUD scheme achieved 3 to 5dB gain at a target activity detection error rate of $\bf{{10}^{-3}}$ compared to the state-of-the-art DL-based AUD schemes. | ['Chung G. Kang', 'Ameha T. Abebe', 'Minsig Han'] | 2022-05-22 | null | null | null | null | ['activity-detection'] | ['computer-vision'] | [ 3.62053573e-01 -1.68522671e-01 -4.13385600e-01 -3.64948213e-02
-6.63403928e-01 -2.31031664e-02 6.21426105e-02 -3.02699059e-01
-4.30998981e-01 8.68290961e-01 1.45879285e-05 -9.06466961e-01
-1.61171183e-01 -4.11502510e-01 -1.18874975e-01 -1.12767589e+00
-7.96961486e-01 -1.62178241e-02 -2.31950060e-01 6.74531469e-03
1.17169976e-01 3.39334875e-01 -7.37673223e-01 -3.42628688e-01
6.28916502e-01 1.44215465e+00 5.14021277e-01 9.58235443e-01
1.92868918e-01 9.05520737e-01 -7.32966423e-01 3.95043343e-02
1.53915912e-01 -7.44617879e-01 -2.42789194e-01 1.97852135e-01
-3.53307158e-01 -6.12926483e-01 -1.08869135e+00 7.55440354e-01
8.27815831e-01 -4.57242914e-02 6.50797904e-01 -1.07493949e+00
1.47379756e-01 5.39930165e-01 -7.80421913e-01 5.19403696e-01
-3.36596668e-01 7.70786405e-03 9.66095805e-01 -1.01089633e+00
-2.86646932e-02 6.71880722e-01 2.47117251e-01 2.57206589e-01
-7.76750386e-01 -1.40422034e+00 -4.34177101e-01 3.80937248e-01
-1.90665495e+00 -1.01855111e+00 6.60767972e-01 -6.65614977e-02
7.12553144e-01 1.19064137e-01 4.53281492e-01 5.28308988e-01
1.56304717e-01 7.71672130e-01 6.86516702e-01 -8.66454601e-01
3.20021600e-01 -6.52430207e-02 -1.98173732e-01 1.02726483e+00
1.62704974e-01 2.10136279e-01 -2.75388926e-01 -2.91329026e-01
1.16189551e+00 -5.43413520e-01 -3.86493921e-01 -1.18939281e-02
-8.69515121e-01 7.76130319e-01 2.13907603e-02 2.33760968e-01
-5.45092881e-01 3.99103612e-01 -2.83787757e-01 2.37874836e-01
-1.22008294e-01 6.04071990e-02 1.56463519e-01 -1.80654690e-01
-1.25981426e+00 -1.88393429e-01 1.02279377e+00 1.32764137e+00
8.77101719e-01 6.04002833e-01 -3.28153759e-01 8.46076727e-01
8.96255255e-01 6.76951110e-01 6.41534030e-02 -5.87605417e-01
6.96753025e-01 -1.58135042e-01 -1.67500988e-01 -7.68001795e-01
-3.17240447e-01 -1.34746325e+00 -1.22320592e+00 -1.88500270e-01
2.43599311e-01 -9.16210115e-01 -5.30384481e-01 1.57932568e+00
-1.97177619e-01 6.85113609e-01 3.08295697e-01 8.81741226e-01
-3.16247940e-02 1.00055718e+00 -1.92409202e-01 -6.96766257e-01
1.12595582e+00 -8.31886649e-01 -8.27109158e-01 -4.29382205e-01
2.51414657e-01 -7.82307506e-01 -8.66461396e-02 4.07944530e-01
-1.29260004e+00 -4.06350046e-01 -1.81312799e+00 5.68562448e-01
5.48777282e-01 8.30599010e-01 3.12231570e-01 8.86604071e-01
-6.11002326e-01 -7.92383626e-02 -4.38229680e-01 1.80119962e-01
5.49239874e-01 7.27865636e-01 3.32001448e-01 9.15552489e-03
-1.29121399e+00 3.72877628e-01 4.18642759e-01 1.05260007e-01
-1.18322563e+00 -2.60828942e-01 -7.03027368e-01 5.32400846e-01
6.96871817e-01 1.15269134e-02 1.20789564e+00 -6.78221941e-01
-1.44634879e+00 -5.32355458e-02 -1.63574219e-01 -6.62233889e-01
-3.22308689e-01 1.91792399e-01 -1.03230965e+00 1.51393384e-01
-3.14827114e-01 -8.56005549e-02 8.96112561e-01 -1.08388436e+00
-1.03051233e+00 1.11312211e-01 -2.48518825e-01 2.17344344e-01
-2.25760728e-01 -7.49073476e-02 -8.96916449e-01 -8.77417922e-01
-7.29042962e-02 -6.19764686e-01 -4.79907751e-01 -4.28035915e-01
-3.39008540e-01 1.45115733e-01 7.89779902e-01 -7.51418889e-01
2.13490582e+00 -2.29585052e+00 1.99734837e-01 9.25799131e-01
2.88464785e-01 4.86286938e-01 4.93081361e-02 4.24452633e-01
2.38512605e-01 -2.65881002e-01 -7.33555630e-02 -1.71905801e-01
-2.17834443e-01 1.07106566e-01 -1.01533905e-01 5.77862382e-01
1.57398731e-01 1.83300987e-01 -8.06106508e-01 -3.73764247e-01
9.58531275e-02 1.09551966e-01 -5.31946719e-01 4.82715994e-01
-8.11311696e-03 2.69876480e-01 -5.68178236e-01 5.36338568e-01
8.14476073e-01 -4.72200602e-01 5.36522150e-01 -8.78326073e-02
-2.18250275e-01 3.31461132e-02 -1.40452302e+00 1.41705918e+00
-6.73312843e-01 7.94488072e-01 5.92878938e-01 -9.61948454e-01
5.21676421e-01 7.57719159e-01 4.00034934e-01 -9.01620388e-01
7.24637389e-01 3.44956666e-01 5.11584342e-01 -2.25384552e-02
5.90531945e-01 -1.29738674e-01 -2.81700760e-01 6.78751647e-01
3.90603513e-01 5.32060504e-01 1.36679009e-01 2.91102737e-01
9.82756734e-01 -5.48815787e-01 8.70896220e-01 -2.45858654e-01
1.10630131e+00 -6.38119519e-01 6.55494928e-01 9.16815579e-01
3.00801378e-02 -1.69584513e-01 2.46004567e-01 3.88466418e-01
-1.12567663e+00 -9.41648901e-01 -6.39582500e-02 7.43107021e-01
4.95544851e-01 -2.78269529e-01 -5.88207066e-01 -2.72987366e-01
-4.87546831e-01 7.08202362e-01 2.18917608e-01 -4.57733162e-02
-5.41547537e-01 -7.30836809e-01 9.05009747e-01 7.31216297e-02
9.17495072e-01 -3.52015734e-01 -1.81437090e-01 9.95001435e-01
-1.43166989e-01 -1.37694299e+00 -4.94180918e-01 5.55896819e-01
-2.08996207e-01 -7.57014036e-01 -7.08404541e-01 -6.91299140e-01
3.96401376e-01 2.55290478e-01 4.14690703e-01 2.62367576e-01
5.02611278e-03 -4.10923958e-02 -3.47194821e-01 -3.50797027e-01
-6.98653221e-01 3.03979158e-01 1.71527281e-01 4.10501987e-01
3.16718370e-01 -6.17146254e-01 -4.93275344e-01 4.83396977e-01
-5.63565135e-01 9.92945489e-03 1.19727576e+00 6.72417998e-01
5.51855266e-02 2.38360018e-01 7.08619535e-01 -2.61574835e-01
4.92296785e-01 -7.59425282e-01 -8.38422596e-01 1.65744483e-01
-4.98373479e-01 1.88543335e-01 7.49971926e-01 1.18050659e-02
-8.98772478e-01 -8.18926618e-02 -5.10814190e-01 -4.38002832e-02
3.35344762e-01 6.35263324e-01 -2.79766083e-01 -2.59821624e-01
5.72841108e-01 5.48611164e-01 8.67001563e-02 -1.17525980e-01
-1.85869172e-01 1.60382819e+00 5.13344049e-01 -4.25620347e-01
1.01356280e+00 1.39826283e-01 1.20615505e-01 -1.37005079e+00
-5.75938702e-01 -7.02009320e-01 -2.83822685e-01 -3.04825425e-01
2.97746897e-01 -1.52942443e+00 -5.66515386e-01 3.46161991e-01
-9.35473442e-01 -1.45511404e-01 4.84112024e-01 6.91353142e-01
-3.52244526e-01 6.68969393e-01 -3.43963504e-01 -1.09750807e+00
-4.50228721e-01 -1.19707704e+00 4.63276625e-01 3.29286516e-01
1.13936812e-01 -5.98596632e-01 -4.76895511e-01 9.28737372e-02
7.88410604e-01 -2.60157555e-01 9.57614899e-01 -4.39374477e-01
-9.91398990e-01 -2.12132141e-01 -3.19992334e-01 4.18846250e-01
1.02965079e-01 -6.91911817e-01 -7.10299551e-01 -6.24210596e-01
-4.51324657e-02 -5.75812571e-02 3.90893638e-01 3.82429779e-01
1.14650643e+00 -4.46811557e-01 -5.61059892e-01 5.52878380e-01
1.60929775e+00 5.92302740e-01 1.06843328e+00 -3.13782871e-01
1.48393840e-01 -4.92896289e-01 6.01091564e-01 1.18712962e+00
5.55951744e-02 9.48535383e-01 3.28315407e-01 -5.34308283e-03
9.01297182e-02 3.50041837e-01 4.33273047e-01 1.02219999e+00
5.77525757e-02 -9.39512491e-01 -4.09081936e-01 1.38119310e-01
-1.44723761e+00 -1.04263580e+00 -1.14365071e-01 2.23080373e+00
1.14066529e+00 2.86771148e-01 -1.76219940e-01 4.83159244e-01
5.98454595e-01 1.74609601e-01 -2.57885903e-01 -8.99543390e-02
1.41419649e-01 6.78436279e-01 9.01843905e-01 5.71941316e-01
-1.06310594e+00 6.14149392e-01 4.38869524e+00 1.56591427e+00
-1.11980236e+00 1.83983147e-01 2.12231845e-01 7.00681433e-02
2.44122729e-01 4.30672765e-02 -9.70507383e-01 5.89624763e-01
1.27702379e+00 3.64785567e-02 5.52146792e-01 4.05942917e-01
5.14619827e-01 -4.38250989e-01 -8.29486489e-01 9.54538822e-01
-4.79981899e-02 -1.45685279e+00 -2.95965850e-01 3.60497981e-01
3.44528317e-01 -1.94522277e-01 -3.21190447e-01 4.56107318e-01
9.84324366e-02 -7.30450869e-01 6.05083466e-01 5.94285846e-01
1.29691732e+00 -9.33017731e-01 6.79471254e-01 6.14276409e-01
-1.27060401e+00 -5.54215431e-01 -5.95719293e-02 5.49893826e-02
3.94742191e-01 7.28103042e-01 -9.81683016e-01 6.82492495e-01
-1.71785399e-01 1.83808938e-01 6.21426031e-02 1.24726558e+00
-1.34728238e-01 1.21532834e+00 -3.50834280e-01 -2.00837433e-01
4.57848161e-01 2.35237673e-01 7.02832222e-01 1.39482653e+00
5.79724491e-01 2.93972045e-01 1.85772121e-01 3.25785130e-01
-3.38752866e-01 -8.13813433e-02 1.61629379e-01 1.05189644e-01
1.01589286e+00 1.19075274e+00 1.18550293e-01 -1.95320025e-01
-5.31441867e-01 6.73563778e-01 -9.78750736e-02 5.37052572e-01
-9.09089804e-01 -9.62962389e-01 5.21520674e-01 -5.78240231e-02
4.77328748e-01 -7.55931437e-01 -1.05090395e-01 -8.20183754e-01
-3.34018320e-01 -9.93455112e-01 8.44602659e-02 -2.29845539e-01
-5.88508904e-01 1.13453172e-01 -4.01465595e-01 -1.56226659e+00
-4.44134742e-01 -3.02943289e-01 -4.94030386e-01 1.20129430e+00
-1.79698610e+00 -5.84448278e-01 -2.79148310e-01 7.16074467e-01
3.81450474e-01 -7.77379632e-01 7.75895596e-01 8.81730974e-01
-6.59232676e-01 1.14695287e+00 3.36509883e-01 5.54592788e-01
1.06757395e-01 -6.48568928e-01 -2.35682994e-01 1.16945481e+00
-9.55164731e-02 3.54657084e-01 3.66790861e-01 -3.86129588e-01
-1.51995146e+00 -9.20838535e-01 5.03460586e-01 7.56593883e-01
2.80033082e-01 -4.70952570e-01 -3.98093015e-01 2.57536203e-01
4.18146700e-01 -3.49273473e-01 1.24132180e+00 -4.28772569e-01
5.36911607e-01 1.07021622e-01 -6.44046843e-01 8.06204617e-01
4.57154363e-01 -2.43932158e-01 2.26967484e-01 -2.26002976e-01
4.57570776e-02 -5.15287876e-01 -7.02260852e-01 4.26940061e-02
5.62811732e-01 -5.14829755e-01 7.54525840e-01 -3.19318511e-02
3.23835127e-02 -5.37535667e-01 -4.00428236e-01 -9.06187892e-01
-6.40471458e-01 -9.84392583e-01 -7.27618575e-01 9.26015913e-01
5.41754782e-01 2.31926024e-01 8.66615117e-01 -3.48477274e-01
-3.40534538e-01 -6.41253948e-01 -1.34154534e+00 -5.37612796e-01
-4.90615517e-01 -6.29846096e-01 2.40511298e-01 4.43738431e-01
-3.64057794e-02 7.59896934e-01 -9.83249724e-01 7.75593221e-01
7.22351670e-01 -6.38218760e-01 8.99551392e-01 -9.14919913e-01
-7.95701444e-01 -2.77473301e-01 -1.63524091e-01 -1.90662563e+00
-2.50909179e-01 -8.93633068e-01 4.24523503e-02 -1.06199503e+00
-2.29147255e-01 -8.84610295e-01 -2.58838207e-01 2.00408235e-01
8.00067384e-04 -2.90312886e-01 4.55252565e-02 2.20796883e-01
-4.54182059e-01 5.68798363e-01 9.88644838e-01 1.65651202e-01
-9.78796855e-02 4.79340166e-01 -6.47955418e-01 2.30510205e-01
7.40939379e-01 -2.01688930e-01 -2.68378317e-01 -1.15320906e-01
-1.13825329e-01 8.37784469e-01 9.33578610e-02 -1.63404310e+00
5.22774518e-01 -7.95981288e-02 3.72004062e-01 -6.99276924e-01
4.61035997e-01 -1.22034585e+00 -2.96474904e-01 4.86807346e-01
-1.10601664e-01 -7.91423321e-01 5.95181361e-02 7.88071334e-01
5.27062789e-02 -4.83628541e-01 8.59736741e-01 4.90604073e-01
-9.11037445e-01 4.63631451e-01 -1.20100772e+00 2.97538494e-03
1.01688051e+00 -2.17675328e-01 9.87782329e-02 -6.98004365e-01
-3.48077625e-01 6.93496525e-01 -6.51624918e-01 -5.99652641e-02
5.80851793e-01 -1.18677533e+00 -7.44399607e-01 5.17972410e-01
-1.99947909e-01 -2.46696368e-01 2.33381733e-01 8.97493362e-01
-3.12227130e-01 8.02258074e-01 3.94235887e-02 -4.19698358e-01
-1.19756413e+00 -2.73006141e-01 4.85863268e-01 -4.76916254e-01
1.60108402e-01 6.06924951e-01 -5.60990930e-01 5.18333256e-01
4.28662121e-01 2.94662654e-01 -1.96211636e-01 -3.67484003e-01
4.64042515e-01 3.78156543e-01 -8.24020728e-02 -4.61603582e-01
-6.98253736e-02 -2.94752643e-02 -2.48361573e-01 -1.72448382e-01
8.18709493e-01 -3.71658593e-01 3.10492933e-01 -4.13079202e-01
1.31569672e+00 3.18023324e-01 -1.11051464e+00 -7.69053757e-01
-2.05819309e-01 -3.87788266e-01 5.45265615e-01 -6.40252590e-01
-1.01505065e+00 8.26942205e-01 7.60976553e-01 -1.44557029e-01
1.29782856e+00 -4.60328609e-01 1.01042438e+00 3.99771124e-01
7.22310007e-01 -1.30483651e+00 1.81512445e-01 7.84718394e-01
2.89586663e-01 -6.31409466e-01 2.15945348e-01 -2.06275657e-01
-2.62019992e-01 1.46456861e+00 3.56356710e-01 4.67093885e-01
8.94054353e-01 4.29992199e-01 -3.67680937e-01 -5.50423786e-02
-5.91152012e-01 -1.42053515e-01 1.53979436e-01 2.29886502e-01
2.46466905e-01 -4.95974720e-02 -3.78441185e-01 7.55722404e-01
1.21770926e-01 9.33938175e-02 8.28446388e-01 1.12554860e+00
-9.27837074e-01 -1.28897953e+00 -2.34371424e-01 8.29701185e-01
-6.86675668e-01 -3.29416662e-01 5.57287216e-01 5.64390719e-01
1.62133560e-01 1.17646527e+00 9.33014601e-02 -5.36621034e-01
-8.89957100e-02 -3.21133673e-01 1.92227095e-01 -4.60228115e-01
-3.50187384e-02 4.75837767e-01 5.66826761e-01 -6.35889769e-02
-1.63729280e-01 -4.66380060e-01 -1.62591732e+00 -2.18897253e-01
-7.92719364e-01 3.73617440e-01 5.12315333e-01 1.22842538e+00
2.83241272e-01 6.52203560e-01 1.45578253e+00 -3.77380043e-01
-5.16193092e-01 -7.88580298e-01 -8.15014362e-01 -7.31942952e-01
5.10729373e-01 -3.75936955e-01 -4.52451110e-02 -6.71867505e-02] | [6.299469947814941, 1.4675555229187012] |
e377f564-08ed-457b-a574-723de867ce71 | muffliato-peer-to-peer-privacy-amplification | 2206.05091 | null | https://arxiv.org/abs/2206.05091v2 | https://arxiv.org/pdf/2206.05091v2.pdf | Muffliato: Peer-to-Peer Privacy Amplification for Decentralized Optimization and Averaging | Decentralized optimization is increasingly popular in machine learning for its scalability and efficiency. Intuitively, it should also provide better privacy guarantees, as nodes only observe the messages sent by their neighbors in the network graph. But formalizing and quantifying this gain is challenging: existing results are typically limited to Local Differential Privacy (LDP) guarantees that overlook the advantages of decentralization. In this work, we introduce pairwise network differential privacy, a relaxation of LDP that captures the fact that the privacy leakage from a node $u$ to a node $v$ may depend on their relative position in the graph. We then analyze the combination of local noise injection with (simple or randomized) gossip averaging protocols on fixed and random communication graphs. We also derive a differentially private decentralized optimization algorithm that alternates between local gradient descent steps and gossip averaging. Our results show that our algorithms amplify privacy guarantees as a function of the distance between nodes in the graph, matching the privacy-utility trade-off of the trusted curator, up to factors that explicitly depend on the graph topology. Finally, we illustrate our privacy gains with experiments on synthetic and real-world datasets. | ['Laurent Massoulié', 'Aurélien Bellet', 'Mathieu Even', 'Edwige Cyffers'] | 2022-06-10 | null | null | null | null | ['graph-matching'] | ['graphs'] | [-2.46970102e-01 2.61843055e-01 -1.71468347e-01 -6.38686478e-01
-6.37016654e-01 -1.07222140e+00 2.59768903e-01 5.52014351e-01
-5.73116183e-01 6.09395683e-01 7.33874738e-02 -1.31843984e-01
-1.83075115e-01 -9.56241488e-01 -6.95172966e-01 -1.07264280e+00
-7.71484613e-01 2.47784674e-01 -2.65629172e-01 -6.20057061e-03
-1.24122791e-01 6.92523658e-01 -6.55697823e-01 -2.65671402e-01
3.32644880e-01 7.23946869e-01 -5.97836256e-01 6.34650648e-01
2.73970306e-01 6.00646555e-01 -4.76260632e-01 -6.78813219e-01
9.18901742e-01 -3.45595837e-01 -7.77620614e-01 -2.37280205e-01
2.52612203e-01 -4.70458925e-01 -6.07756436e-01 1.30276763e+00
4.62512583e-01 -1.90382555e-01 1.21942803e-01 -1.86350739e+00
-3.60011429e-01 1.21072972e+00 -7.35365510e-01 -2.33181134e-01
9.41976160e-02 -6.45397827e-02 1.26792002e+00 1.79246098e-01
6.97958171e-01 9.28432345e-01 9.85504210e-01 5.68684220e-01
-1.62881517e+00 -6.62686825e-01 7.95544460e-02 -3.31082553e-01
-1.49903738e+00 -4.14169103e-01 8.84822369e-01 -1.26629129e-01
3.68385077e-01 8.20576847e-01 3.42432201e-01 4.98502910e-01
5.23070544e-02 5.69330335e-01 1.09185684e+00 -5.66478744e-02
5.42544246e-01 3.63682777e-01 1.16315998e-01 5.47209740e-01
4.86609995e-01 -6.05264008e-02 -5.59335470e-01 -7.51170814e-01
4.42507714e-01 1.40834063e-01 -5.77129006e-01 -1.03595352e+00
-7.18295515e-01 8.54680777e-01 7.19594121e-01 2.21717790e-01
-1.93897441e-01 5.34098327e-01 3.70414764e-01 9.84730661e-01
4.06944364e-01 2.83817053e-01 -6.77181244e-01 2.64618605e-01
-6.53816998e-01 2.29159534e-01 1.44279182e+00 9.33552265e-01
1.17446709e+00 -6.20388269e-01 1.00208312e-01 1.69508010e-01
2.63576567e-01 2.78349370e-01 -1.46075383e-01 -1.26874340e+00
3.81595105e-01 4.39404786e-01 2.55102187e-01 -1.62747943e+00
-1.42107978e-01 -2.75885582e-01 -1.01024699e+00 6.33997172e-02
4.58625704e-01 -6.17071390e-01 4.50350530e-02 2.27736306e+00
5.42000234e-01 -2.95082599e-01 3.27895321e-02 8.77051413e-01
6.25368282e-02 3.45872611e-01 -1.07257113e-01 -3.91732424e-01
7.70801306e-01 -6.28998399e-01 -4.52737153e-01 -7.99525529e-02
1.07215106e+00 -9.14819092e-02 4.80411679e-01 -1.84231564e-01
-9.42960680e-01 5.74139535e-01 -8.54370177e-01 -2.31612921e-01
-2.52045065e-01 -4.97760236e-01 8.67633820e-01 1.12312031e+00
-1.56057501e+00 9.62972403e-01 -1.01768291e+00 -3.76675487e-01
4.27929908e-01 6.92591548e-01 -6.56456411e-01 1.28965005e-01
-9.52568769e-01 1.45563260e-01 -8.61892477e-02 -9.73752365e-02
-6.35214806e-01 -8.93890381e-01 -7.57340729e-01 2.72110015e-01
3.05154055e-01 -7.38877058e-01 1.04941189e+00 -1.03673255e+00
-1.12690639e+00 1.01799655e+00 -5.10294102e-02 -8.59188199e-01
1.01781166e+00 3.31038415e-01 2.88688421e-01 -5.47902025e-02
5.79230487e-02 1.04747958e-01 2.96013206e-01 -1.17644143e+00
-5.07644773e-01 -1.06181920e+00 3.05129051e-01 2.74664789e-01
-5.15271842e-01 -1.37560904e-01 -3.11485857e-01 -2.43026853e-01
1.71531424e-01 -1.08431602e+00 -5.20519614e-01 7.26796150e-01
-5.18353701e-01 1.42206967e-01 1.00794756e+00 -3.67557377e-01
9.55296040e-01 -2.31847239e+00 -1.21517301e-01 6.41135752e-01
7.27032959e-01 -2.92690963e-01 -1.63132288e-02 7.77849674e-01
3.09827507e-01 5.01555681e-01 -3.28803301e-01 -8.14851880e-01
3.15920681e-01 1.84622392e-01 4.17646840e-02 1.28301001e+00
-7.67499447e-01 7.97072172e-01 -8.36095572e-01 -1.83110327e-01
-2.96258122e-01 1.64735466e-01 -5.78129172e-01 -6.37206808e-02
4.76318523e-02 3.52986515e-01 -6.89667106e-01 2.91589528e-01
1.15151715e+00 -3.12147081e-01 7.99875557e-01 5.76284267e-02
6.27757087e-02 1.32145777e-01 -1.19667900e+00 1.57513809e+00
-1.09514408e-01 4.50613976e-01 1.15630984e+00 -8.15317273e-01
3.63025278e-01 2.09442630e-01 6.87306762e-01 -1.94508910e-01
1.76198259e-01 3.38479877e-02 -4.11643565e-01 -1.06422491e-02
7.04764277e-02 5.61288074e-02 -2.69320279e-01 9.36197519e-01
-2.68252820e-01 2.48590186e-01 -2.89469600e-01 5.57436109e-01
1.32068872e+00 -7.53635526e-01 1.73168555e-01 -5.39915204e-01
8.29510242e-02 -4.14159924e-01 7.94006288e-01 1.15959561e+00
-4.62707907e-01 1.51441544e-01 1.03637052e+00 -2.18289778e-01
-8.19790423e-01 -7.94690967e-01 2.49250829e-01 9.99379873e-01
3.34381580e-01 -5.91100097e-01 -1.00076318e+00 -8.80360782e-01
4.39596981e-01 2.29436204e-01 -7.08900154e-01 -1.30982129e-02
-1.63775831e-01 -7.42949843e-01 5.31208634e-01 -1.38493134e-02
5.48745990e-01 -1.94853634e-01 -2.22434580e-01 -1.61297590e-01
1.48017276e-02 -8.91336739e-01 -8.85188103e-01 -4.35887836e-02
-7.72922635e-01 -1.04607713e+00 -1.36507124e-01 -4.54192162e-01
9.73235071e-01 5.15315413e-01 6.57442451e-01 2.85966694e-01
5.21620438e-02 7.98034787e-01 3.51632312e-02 -1.27507314e-01
-2.25329801e-01 2.52686888e-01 -5.70928343e-02 3.80651176e-01
1.88519090e-01 -1.06528795e+00 -7.86791265e-01 1.40362754e-01
-8.34311128e-01 -2.75692016e-01 1.80699497e-01 4.32784706e-01
4.97005254e-01 8.92726853e-02 -3.62100564e-02 -1.25478625e+00
7.76721954e-01 -6.33545697e-01 -9.56387699e-01 2.45825171e-01
-8.16692054e-01 2.37193465e-01 5.09634197e-01 -1.20936492e-02
-5.80960274e-01 3.92928123e-01 3.78942460e-01 -8.00341368e-02
3.23473543e-01 3.23494434e-01 -5.48568487e-01 -8.02157223e-01
4.39876944e-01 8.95474777e-02 2.83151150e-01 -4.14243162e-01
6.60633326e-01 4.49540645e-01 7.23409429e-02 -6.59125805e-01
8.01401913e-01 9.31044698e-01 4.62913156e-01 -7.56784797e-01
-1.56072035e-01 -8.86788666e-02 -3.08552921e-01 2.50177592e-01
1.88618585e-01 -7.69715786e-01 -1.37211144e+00 5.79470754e-01
-8.59421730e-01 -1.40140116e-01 -5.05635917e-01 1.16144158e-01
-2.34212399e-01 6.81462944e-01 -6.45998180e-01 -8.53681147e-01
-4.01343465e-01 -8.84520054e-01 5.89972675e-01 -2.70902757e-02
2.95038451e-03 -1.13713968e+00 5.26246652e-02 8.03793222e-03
6.43130243e-01 5.47095716e-01 5.91748595e-01 -8.87244523e-01
-8.96804214e-01 -5.75725257e-01 -1.23685613e-01 8.83794874e-02
1.00611694e-01 -2.90693700e-01 -7.92193949e-01 -7.67874300e-01
2.46662796e-01 -8.47554728e-02 5.16035259e-01 9.33357850e-02
1.03368640e+00 -1.31248856e+00 -5.22538900e-01 1.05655694e+00
1.70102167e+00 -4.01926458e-01 1.52435631e-01 -6.60444200e-02
6.15978301e-01 6.05850101e-01 -2.21799493e-01 7.64019072e-01
6.64713681e-01 2.89420158e-01 5.18641412e-01 3.45393196e-02
5.51387787e-01 -5.77960074e-01 5.67231029e-02 2.77043283e-01
3.31531137e-01 -2.38409624e-01 -4.62470770e-01 5.62526584e-01
-1.83466339e+00 -5.28979838e-01 1.76349692e-02 2.46360660e+00
9.41574931e-01 -4.89383310e-01 1.89949140e-01 -1.85628712e-01
5.66220582e-01 4.33346540e-01 -5.29087424e-01 -3.58582169e-01
-3.04911226e-01 -1.46906465e-01 1.50377810e+00 8.94405067e-01
-8.72470737e-01 6.20262921e-01 6.01460838e+00 3.60073298e-01
-1.05437732e+00 3.34836543e-01 8.71091127e-01 -3.13101113e-01
-7.45261431e-01 3.43948960e-01 -3.66317689e-01 3.38471055e-01
7.72973001e-01 -7.62807488e-01 9.18592751e-01 1.08460224e+00
3.77059057e-02 2.25559369e-01 -1.47750747e+00 7.86284745e-01
-4.32901174e-01 -1.19071937e+00 -2.99143016e-01 5.12888610e-01
7.78079987e-01 1.59624189e-01 5.62685095e-02 -3.78579199e-01
9.79319096e-01 -7.43452966e-01 2.16625720e-01 1.02888696e-01
3.61890912e-01 -8.97434592e-01 4.08341080e-01 4.54183459e-01
-9.29242671e-01 -1.34699434e-01 -2.27819368e-01 7.57819265e-02
-1.91466630e-01 6.58211589e-01 -4.93274003e-01 4.66482252e-01
7.92820394e-01 3.34068805e-01 -1.94297552e-01 6.55753016e-01
-1.43673703e-01 3.67636293e-01 -9.11548555e-01 -4.54701334e-02
9.50774327e-02 -5.08633435e-01 7.11387455e-01 9.50675488e-01
6.62628859e-02 2.47745618e-01 -1.61456671e-02 8.98560643e-01
-8.07933331e-01 1.09347120e-01 -6.84499919e-01 -1.04891576e-01
9.48872745e-01 1.16575587e+00 -1.96742609e-01 1.91722959e-01
-2.17322871e-01 1.23884654e+00 5.14899135e-01 4.09550905e-01
-2.46974051e-01 -4.18704152e-01 1.33424962e+00 1.72780454e-01
1.67171344e-01 -3.17282349e-01 -1.95636615e-01 -1.14262748e+00
3.77862722e-01 -6.23990297e-01 5.70611596e-01 1.80130303e-01
-1.39202249e+00 -2.77393032e-02 -4.63460147e-01 -5.38143516e-01
1.07050493e-01 9.67450663e-02 -5.01492798e-01 7.72236884e-01
-1.11245048e+00 -9.32964087e-01 4.10441943e-02 7.65661478e-01
-7.43867159e-01 5.19315720e-01 8.09422076e-01 2.27865398e-01
-4.14987147e-01 1.21534765e+00 5.34416795e-01 4.10419822e-01
3.45789909e-01 -1.20409632e+00 3.20815861e-01 8.75270367e-01
7.16842934e-02 8.71023178e-01 7.52822757e-01 -3.71253997e-01
-2.09028053e+00 -1.07316923e+00 1.06898689e+00 -1.17874853e-01
6.52944744e-01 -6.47867560e-01 -6.64894283e-01 1.25289190e+00
-2.58962791e-02 4.11262274e-01 7.59415805e-01 1.07045010e-01
-5.24646103e-01 -7.77028739e-01 -2.01123476e+00 6.06808722e-01
1.07566655e+00 -7.69969821e-01 4.87961531e-01 4.14591759e-01
8.37073386e-01 -2.88365453e-01 -9.47254419e-01 3.39459926e-02
4.82974827e-01 -1.06963158e+00 6.30744457e-01 -5.30744255e-01
-4.51347083e-01 -2.20268279e-01 -3.05924624e-01 -1.06956279e+00
-3.70331034e-02 -1.41637135e+00 1.65163592e-01 1.27160013e+00
3.97970945e-01 -1.03214073e+00 1.35878873e+00 1.44635308e+00
9.29805458e-01 -4.24763262e-01 -1.24002540e+00 -4.78224665e-01
1.27200499e-01 -1.31394006e-02 8.46317291e-01 1.36912036e+00
2.09054396e-01 -1.15251355e-01 -5.54590225e-01 6.11205876e-01
1.05506372e+00 -4.47914638e-02 1.04007900e+00 -1.03231609e+00
-3.58826220e-01 -2.02954352e-01 -5.33346772e-01 -1.10380197e+00
1.51882783e-01 -9.09777105e-01 -2.96333820e-01 -9.65326130e-01
1.44619122e-01 -7.38889575e-01 -1.01156294e-01 5.00898838e-01
2.75389999e-01 -2.14018717e-01 1.97284207e-01 1.40165374e-01
-5.89728951e-01 3.80476862e-01 6.26284897e-01 -7.62389079e-02
-3.51057321e-01 3.02605242e-01 -1.19847906e+00 4.19331402e-01
7.42309570e-01 -8.12670171e-01 -4.72443372e-01 -5.35986364e-01
2.62310565e-01 2.52076477e-01 3.32637250e-01 -4.86768901e-01
6.72180414e-01 -1.84327085e-02 -3.44983637e-01 8.72125849e-02
1.31757632e-01 -1.16916168e+00 5.17480910e-01 6.03926420e-01
-6.60777569e-01 -1.21305712e-01 -4.11773682e-01 8.45930696e-01
2.04922214e-01 2.28799462e-01 8.41684461e-01 -1.74003571e-01
1.74562767e-01 7.92747200e-01 -2.31780428e-02 4.23122831e-02
1.15902984e+00 1.84724674e-01 -2.61274755e-01 -9.46337700e-01
-6.00728631e-01 6.55458927e-01 1.00163209e+00 -1.83090925e-01
1.36298850e-01 -9.28047895e-01 -6.72992766e-01 1.56675607e-01
-1.16682433e-01 -1.20600022e-01 9.93996635e-02 9.19443011e-01
-4.44600701e-01 1.01166330e-01 2.76561171e-01 -1.72470465e-01
-1.51691890e+00 5.37134528e-01 6.20809555e-01 -1.52966514e-01
-3.75321567e-01 9.88864362e-01 -4.66365833e-03 -5.16898751e-01
6.66926205e-01 -6.80438653e-02 7.48391628e-01 -3.59709412e-01
3.08736920e-01 4.26664233e-01 -1.55161902e-01 -4.39867616e-01
-4.02527511e-01 1.37931854e-01 -1.94980666e-01 -1.95666760e-01
1.12570179e+00 -4.62514579e-01 -8.26177478e-01 4.35780647e-04
1.78353441e+00 3.76199365e-01 -1.08749282e+00 -5.54470837e-01
-1.04272746e-01 -7.21323371e-01 -1.58514574e-01 -3.25433254e-01
-1.56065524e+00 3.50894094e-01 4.17240769e-01 5.15578389e-01
1.04225183e+00 1.03189148e-01 6.61983848e-01 6.17574990e-01
6.83980048e-01 -7.23934233e-01 -9.62276340e-01 -1.15821108e-01
3.87904376e-01 -1.00278604e+00 1.02296904e-01 -4.11269546e-01
-3.32647383e-01 5.31094551e-01 6.71828687e-02 -1.20158583e-01
1.08201981e+00 3.20357293e-01 -1.50520157e-03 7.73644745e-02
-5.18902183e-01 6.26099706e-01 -5.59350491e-01 5.94934583e-01
6.49515092e-02 4.08737659e-01 -6.12097323e-01 6.76734269e-01
-1.95635974e-01 -3.69135797e-01 4.11157936e-01 1.08156478e+00
-1.77798510e-01 -1.39262140e+00 1.42257437e-01 1.75646961e-01
-7.44693220e-01 7.92514905e-02 -1.01739144e+00 6.96797550e-01
-2.67469734e-01 8.43558431e-01 -1.96999952e-01 -2.68430322e-01
-1.12518258e-01 -4.69193786e-01 1.76583618e-01 -1.64115667e-01
-7.67660618e-01 -4.73789454e-01 -5.16138710e-02 -9.02003288e-01
-2.53103971e-01 -6.36656523e-01 -1.03550267e+00 -1.10410523e+00
-8.44313577e-02 7.23602414e-01 7.93878973e-01 5.11991978e-01
7.55619764e-01 -4.60291147e-01 1.25414705e+00 -1.81731313e-01
-9.15354669e-01 -2.25937843e-01 -1.13396156e+00 2.76605606e-01
7.96187997e-01 4.86592770e-01 -9.75636542e-01 -4.40721154e-01] | [5.925891399383545, 6.450963973999023] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.