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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
18fbdd78-e33e-4d99-a622-64e69112fbed | deep-anomaly-detection-and-search-via | 2208.14834 | null | https://arxiv.org/abs/2208.14834v2 | https://arxiv.org/pdf/2208.14834v2.pdf | Deep Anomaly Detection and Search via Reinforcement Learning | Semi-supervised Anomaly Detection (AD) is a kind of data mining task which aims at learning features from partially-labeled datasets to help detect outliers. In this paper, we classify existing semi-supervised AD methods into two categories: unsupervised-based and supervised-based, and point out that most of them suffer from insufficient exploitation of labeled data and under-exploration of unlabeled data. To tackle these problems, we propose Deep Anomaly Detection and Search (DADS), which applies Reinforcement Learning (RL) to balance exploitation and exploration. During the training process, the agent searches for possible anomalies with hierarchically-structured datasets and uses the searched anomalies to enhance performance, which in essence draws lessons from the idea of ensemble learning. Experimentally, we compare DADS with several state-of-the-art methods in the settings of leveraging labeled known anomalies to detect both other known anomalies and unknown anomalies. Results show that DADS can efficiently and precisely search anomalies from unlabeled data and learn from them, thus achieving good performance. | ['Yang Yu', 'Zongzhang Zhang', 'Feng Mao', 'Dawei Wang', 'Chao Chen'] | 2022-08-31 | null | null | null | null | ['supervised-anomaly-detection', 'semi-supervised-anomaly-detection'] | ['computer-vision', 'computer-vision'] | [-0.1101412 0.07243813 -0.02260375 -0.50242835 -0.6819138 -0.2685731
0.43156213 0.46205205 -0.23773593 0.5035666 0.0149566 -0.26672822
-0.31232718 -0.7334182 -0.4932781 -0.6940973 -0.397802 0.7710317
0.12230595 0.03135561 0.40430695 0.62741816 -1.708849 -0.01310287
1.2057859 1.346499 -0.5813592 0.3816516 -0.39608723 1.1833277
-0.5444676 0.01777821 0.26349965 -0.47482404 -0.6897325 0.27820832
0.10046476 -0.4817987 -0.2046359 0.8591111 0.22293691 0.16705321
0.5758991 -1.7376679 -0.40842947 0.57636297 -0.76812476 0.7194579
0.05791604 0.22104965 0.9813193 -0.84001255 0.02838439 0.99261516
0.5211534 0.63599795 -0.9682312 -0.67969537 0.53063357 0.4163905
-0.8418948 -0.42328924 0.99747926 -0.299113 0.8635906 0.12346167
0.35354078 1.0826823 -0.2410029 1.2014972 0.74473625 -0.45823675
0.7693757 -0.22801591 0.33739766 0.6424708 0.37620893 0.48878205
-0.7311477 -0.63145417 0.19175605 0.48035175 0.12851238 -0.477599
-0.6665984 0.9506008 0.14776704 0.04127766 -0.6002522 -0.28334302
0.61917853 0.63493246 0.6665472 0.6338905 -0.6282812 -0.226166
-0.8250978 0.04784827 0.5191732 0.5458872 0.8916225 0.60371244
0.0904645 0.749326 0.5960475 0.24907903 0.75959533 -0.69625753
0.33666113 1.1312635 -0.01177786 -0.59294903 -0.5907595 -0.2692379
-0.6533463 0.29944998 0.3328868 -0.4880371 -1.0835856 1.5735217
0.43199575 0.52891785 0.2624904 0.64358103 0.36415428 0.26912194
-0.02738716 -0.38784525 0.63896024 -1.0402619 -0.56764936 -0.45721945
0.89583504 -0.00687019 0.7803275 0.6474395 -0.82357484 -0.22720197
-0.924978 0.7762837 -0.38997832 -0.20581308 0.35941347 0.4767602
-0.60685444 0.6753869 -1.3542668 -0.16394427 0.67578554 0.18243612
-0.07428528 0.0329303 -0.9307257 0.4765765 0.5007499 0.01783815
-1.0841784 -0.4784935 -1.0008893 -0.03022006 0.7971044 0.02092092
1.2058657 -1.0281624 -1.0915073 0.6402841 -0.11534163 -0.93788534
0.4683349 -0.75166464 -0.8812861 0.04834669 0.11744586 -0.06048506
1.0413132 -1.1764882 -0.9313101 -0.91360044 -0.5251549 -0.05803127
-0.4436423 -0.05918033 0.07220942 -0.7589988 0.3039682 -0.6641453
-0.47671992 -0.3606749 -0.63353753 -0.43902826 1.3348122 -0.34134835
1.370499 -2.3651466 -0.32862297 0.7369205 0.29418513 0.3898109
0.03565089 0.48755392 -0.3113285 -0.17044465 -0.5892435 -0.45528692
-0.09921592 0.616209 -0.83703727 0.4260077 0.3739418 0.62245697
-1.0199394 -0.20899102 0.0624612 -0.36509585 -0.54419017 0.69466436
-0.4741667 0.788615 -0.6822304 1.0776461 0.5165704 -0.1195678
-0.01113205 0.75392514 0.13151056 -0.01266513 -1.3169194 1.3043761
-0.08408947 0.24415134 -0.261619 -1.5469774 1.1168911 0.17965317
0.7491901 -0.8945686 -0.30531773 0.47383627 -0.16961747 -0.5669266
-0.01737192 0.5508321 0.08609566 0.95440066 -0.03332781 0.52965176
0.21997733 0.10187337 1.4986516 0.10730236 0.2685152 0.07468627
0.52350485 -0.09815006 0.8987839 1.1245195 -0.23638439 0.2590575
0.7094907 -0.94011426 -0.57829 -1.1665804 0.086912 1.3697664
-0.05637297 -0.40666258 -0.41841546 -1.4201628 0.1284095 0.898223
-0.6093194 -0.52463895 -0.4981702 -1.0129285 0.40209544 0.78385866
0.48392528 -1.5236866 -1.039506 0.29617193 0.13687932 -0.63537586
0.107548 0.6277131 -1.0407948 -1.5069796 -0.15448675 -0.37351012
0.91768545 -0.12788725 1.2348303 0.31818724 -0.32018366 0.4375117
-0.6984528 -0.77649444 -0.39458358 -0.1942542 0.43093202 0.3132207
0.7919627 -0.8298301 -0.476989 0.4047719 -1.027219 -0.83917934
0.5304213 1.0403681 0.5401411 0.09678856 0.9076267 -1.0817808
0.43140352 -0.7675261 -0.7964911 0.17403676 -1.177172 0.37508097
0.6755012 -0.05338551 -0.8196288 0.07934935 0.0067527 -0.6921824
-0.70994955 0.31247762 -0.02402738 0.29473305 0.75965196 0.44049326
-0.04773383 -0.6044855 0.01800485 0.57361656 0.4905623 -0.6241782
0.62280315 0.5464798 -0.16686384 -0.47297147 -1.1727796 -0.55788225
-0.58329237 0.07001787 0.35188606 -0.6786295 -0.30731744 0.6400056
-0.45259032 -0.37052232 -0.7249062 0.19097766 -0.48344666 0.42233902
-0.43852594 -1.2008594 -0.31604138 -0.85859835 0.900796 0.202281
-0.11614742 -0.88131255 0.486728 -0.0946888 0.35052592 0.5212175
0.8275138 -1.7835137 -0.14387006 -0.30562395 0.0799633 0.39909118
0.1486686 -0.0643262 -1.0013373 -0.2929792 -0.04033846 -0.6194175
1.0999612 0.04698076 1.6746802 -0.5065894 -0.19465517 0.5234497
1.0084699 0.29970747 0.2569398 0.72058046 0.58013266 0.5116069
0.78423566 1.0410858 0.12176449 0.11116832 0.9425945 -0.04015253
0.8027114 -0.11700547 0.29738936 0.2085368 0.22187035 -0.08426663
-1.216188 0.61360085 -2.3085678 -0.87359977 0.1377978 2.0701191
0.57654023 0.3632683 0.5272211 0.4483116 0.44137076 0.14859697
-1.1759015 -0.40490702 -0.07417304 0.084264 0.07989211 -0.02799254
-1.3724755 0.5910881 6.0308967 0.40659288 -0.8076175 -0.32906502
0.67786473 -0.05367672 0.06667329 -0.14827749 -0.42868504 0.5748425
1.0503329 0.16016835 0.34321535 1.1966907 -0.14019229 -0.07151323
-1.2725964 0.6747551 0.00721491 -0.8256791 0.01475698 -0.03272689
0.83333486 0.3014102 0.02662691 0.43925732 0.70889103 -0.8029948
0.28625762 0.46332368 0.079856 -1.0938389 0.66756386 0.506455
-0.79979277 -0.92997646 -0.09905557 0.03606993 -0.15440238 0.792911
-0.4633458 0.5203514 1.0057333 0.93127567 -0.7606115 1.0008923
-0.2717248 1.0313416 -0.4099965 0.52736294 0.5090221 -0.20173995
0.63323694 0.7460281 0.36567873 -0.35140377 0.5699971 0.6661315
0.09502453 -0.05563357 -0.5987859 -0.02593798 0.51111466 0.94147146
-0.5667432 -0.35221 -0.48141852 0.7190504 0.41650546 0.23387131
-0.54525155 -0.25695953 0.44943437 -0.0543471 0.22962713 0.17713891
-0.25144395 -1.1561179 0.09734272 -1.101176 1.2727228 -0.20677188
-1.669461 0.4900088 -0.17919953 -1.3932956 -0.66360223 -0.43095642
-1.1189655 0.08067609 -1.6489958 -0.78761554 -0.3931686 0.7718863
0.64500767 -0.69064337 0.83203584 -0.06729633 -1.0056558 0.6129271
0.36442316 0.404604 0.7505361 -1.6813977 0.61827576 1.0932118
0.5585352 0.13459219 0.4147699 -0.6427519 -0.8885482 -1.3529873
0.2694706 -0.5077079 0.6180241 0.05882998 -1.369113 0.8698147
-0.10145074 0.5539339 0.82926875 0.20794049 -0.5375898 -0.07175569
-1.2361637 0.3066721 1.0188532 -0.10672284 -0.7237536 0.34088552
0.30822313 -0.32562804 -0.27131438 0.5708056 -0.01440432 -1.3985971
0.6767829 -1.1141882 0.29050133 -0.06500614 -0.03222317 -1.5950525
0.13144271 -0.6213305 -0.9503134 1.0821964 0.29179177 -0.9661956
0.869918 0.56197786 -0.19361483 -0.76577073 -0.8003486 -0.8488314
-0.22559191 -0.53068453 0.83460367 1.04989 0.06080929 -0.20608848
-0.3264814 0.51553446 0.9130654 0.25447866 0.7194928 -1.698004
-0.11452632 -0.23586248 -0.37536532 -0.4761765 0.3314875 -0.5160992
-0.07171587 -0.86002254 -0.30530128 -0.3977929 -0.94711936 0.7709764
-0.29397184 -0.04300759 -0.4723638 0.35988992 -1.1636437 0.5950744
0.26098803 0.15734038 -0.5072916 0.37255347 -0.61940986 1.1478487
0.9881783 -0.5853392 -0.29917923 -0.20702475 0.0827911 -0.21068826
0.30699617 -1.0214431 0.18095624 -0.23147322 0.5767953 -0.72319156
-0.28450212 -0.85306484 -0.7287743 0.5197105 -0.5823124 0.37838656
0.0493708 0.98287237 -0.40425602 -0.08526334 0.6691628 -0.10665191
-0.9845354 0.45583478 -0.38248372 0.5744799 1.2241012 0.1012702
-0.12603933 -0.33448267 -0.88578385 0.8649314 0.07068284 0.40669703
0.7302215 -1.1861999 -0.5906871 0.72703606 0.5915581 0.4405375
0.15807146 0.6957466 -0.14141351 -0.02139211 -0.13687775 -0.7182038
-0.78796697 0.879859 0.26281208 -0.51409495 -0.78422964 0.59499836
-0.19636439 -0.6029144 0.6846178 0.13915719 -0.33465716 0.08726624
0.68466854 0.72205615 0.18221113 -0.09408538 -0.40386572 0.17651556
-0.33930135 0.25147575 1.5361867 0.03584617 0.09349117 0.5027838
0.61731094 -0.12556787 -1.4189775 -0.83845073 0.8412686 -0.35978588
-0.25167453 -0.7819467 -1.2706258 0.8150761 0.7152659 0.43299153
1.3979737 -0.04332308 0.7102294 0.85049325 0.01403207 -1.5404338
0.49742728 0.494904 0.5345681 -1.6170378 -0.16790183 0.29295686
-0.9407397 1.190834 1.1799284 -0.38387886 0.7203014 0.21860954
0.24796824 -0.36344844 -0.8290167 -0.3120941 0.06448009 0.5592278
-0.25232393 -0.43083325 0.48670566 0.6087856 0.26210034 -0.4146802
0.2692511 1.3844256 -0.5966234 -0.8885022 -0.42161834 0.85133153
-0.63915145 0.25272655 -0.45715442 0.42966366 -0.01334238 1.0038912
0.33677658 -0.11066104 0.45941135 0.5613504 -0.31139743 -0.42372537
-0.2666684 -0.06897946 -0.27266666 -0.91914284 -0.24820344 -0.6726547
-1.2871407 0.3098832 -0.31489936 0.53283805 0.16245495 1.2809138
0.5530769 0.5499016 1.1799424 -0.3789803 -1.0229754 -0.7684688
-0.75516623 0.51788145 0.7842249 -0.7336007 -0.60815215 -0.3647192 ] | [7.613340854644775, 2.426100730895996] |
1e8c10b6-b34b-4d26-8d2e-074d4c770a35 | ggadn-guided-generative-adversarial-dehazing | null | null | https://link.springer.com/article/10.1007/s00500-021-06049-w | https://link.springer.com/content/pdf/10.1007/s00500-021-06049-w.pdf | GGADN: Guided generative adversarial dehazing network | Image dehazing has always been a challenging topic in image processing. The development of deep learning methods,
especially the generative adversarial networks (GAN), provides a new way for image dehazing. In recent years, many deep
learning methods based on GAN have been applied to image dehazing. However, GAN has two problems in image dehazing. Firstly, For haze image, haze not only reduces the quality of the image but also blurs the details of the image. For GAN network, it is difficult for the generator to restore the details of the whole image while removing the haze. Secondly, GAN model is defined as a minimax problem, which weakens the loss function. It is difficult to distinguish whether GAN is making progress in the training process. Therefore, we propose a guided generative adversarial dehazing network (GGADN). Different from other generation adversarial networks, GGADN adds a guided module on the generator. The guided module verifies the network of each layer of the generator. At the same time, the details of the map generated by each layer are strengthened. Network training is based on the pre-trained VGG feature model and L1-regularized gradient prior which is developed by new loss function parameters. From the dehazing results of synthetic images and real images, the proposed method is better than the state-of-the-art dehazing methods. | ['Jian Zhang1 · Qinqin Dong2 · Wanjuan Song3'] | 2021-07-13 | null | null | null | journal-2021-7 | ['image-dehazing'] | ['computer-vision'] | [ 3.75751197e-01 -1.97984260e-02 4.41778153e-01 1.02582119e-01
-1.79363415e-01 -1.24636732e-01 4.72050637e-01 -3.71625006e-01
-2.51290321e-01 7.90871799e-01 2.22669199e-01 1.03164479e-01
2.44495228e-01 -1.33747888e+00 -6.93490684e-01 -1.27876544e+00
5.28428912e-01 -2.07782000e-01 4.12254125e-01 -4.04359311e-01
1.38168097e-01 2.16014132e-01 -1.17119730e+00 2.68576145e-01
1.13621414e+00 9.31418777e-01 2.58913279e-01 4.88403320e-01
-1.52852878e-01 8.93401027e-01 -1.09435427e+00 -3.69344383e-01
2.58020461e-01 -8.70807171e-01 -2.87348658e-01 7.24989027e-02
7.26470426e-02 -5.26075423e-01 -6.82429910e-01 1.44176531e+00
4.85597759e-01 2.24574097e-02 6.97056830e-01 -1.36642480e+00
-1.06318259e+00 1.72016934e-01 -6.07879996e-01 1.83218196e-01
-2.61279762e-01 3.30680102e-01 2.04076916e-01 -4.78658110e-01
9.90604609e-02 1.20900035e+00 4.53625053e-01 7.99330235e-01
-7.88090169e-01 -8.90319943e-01 -1.10145926e-01 1.75190389e-01
-1.39510870e+00 -2.42005046e-02 1.02988362e+00 -3.34304184e-01
2.12723583e-01 1.09970622e-01 5.62898159e-01 7.04642534e-01
6.68433368e-01 6.26995265e-01 1.07259166e+00 -2.50557780e-01
2.30002403e-02 5.62081560e-02 -5.88866532e-01 5.92720091e-01
3.01927835e-01 2.69422233e-01 -6.36677817e-02 4.82566416e-01
9.30498660e-01 3.85004491e-01 -4.75544691e-01 1.27420008e-01
-8.54878426e-01 7.79021502e-01 8.30860019e-01 2.79595733e-01
-3.96024942e-01 2.13207439e-01 7.17494413e-02 2.64587075e-01
5.29991448e-01 1.91049471e-01 2.36947075e-01 2.93327361e-01
-8.97222877e-01 1.19284086e-01 2.82487810e-01 5.78250587e-01
9.49488997e-01 5.68702281e-01 -1.90097749e-01 7.23457992e-01
3.56552392e-01 5.28566778e-01 7.01061189e-01 -4.84212190e-01
3.86397153e-01 7.40589738e-01 -1.65660247e-01 -1.30114186e+00
2.82455325e-01 -4.54178482e-01 -1.42978978e+00 7.60937691e-01
8.16070214e-02 -2.62935072e-01 -1.39005685e+00 1.41977537e+00
1.56573638e-01 3.71402979e-01 1.39313817e-01 8.96638036e-01
8.80993366e-01 1.36452091e+00 -9.82103273e-02 -1.88067555e-01
9.36816394e-01 -1.18741810e+00 -9.87679958e-01 -3.48735869e-01
9.73720476e-03 -8.80714476e-01 7.97525525e-01 4.90733922e-01
-1.02227640e+00 -8.20243061e-01 -1.44485950e+00 4.04752046e-02
-4.22941297e-01 -2.45864242e-01 2.57914037e-01 6.75014257e-01
-8.64533186e-01 3.06542873e-01 -6.29687607e-01 8.38924423e-02
6.37578726e-01 1.43405974e-01 -1.59477860e-01 -4.45244163e-01
-1.34377015e+00 9.00289178e-01 6.60087347e-01 3.76496613e-01
-1.21023798e+00 -4.67230588e-01 -8.50714624e-01 6.61455989e-02
1.54753655e-01 -5.50393343e-01 7.21006989e-01 -1.38487422e+00
-1.55179083e+00 5.73756874e-01 2.89871126e-01 -3.19285423e-01
6.53287530e-01 -1.79485187e-01 -6.38708353e-01 -2.04440812e-03
-9.05454159e-02 5.17777324e-01 1.32851231e+00 -1.34401381e+00
-4.32210058e-01 -9.17771682e-02 -1.24717526e-01 2.39250153e-01
-2.07883671e-01 -2.70332456e-01 -2.02955559e-01 -1.11490643e+00
-1.92799240e-01 -4.80069876e-01 -1.47299364e-01 -4.11532223e-02
-3.54255408e-01 1.76944286e-01 1.26310742e+00 -9.97769058e-01
1.25072312e+00 -2.53110361e+00 1.03343695e-01 1.28400669e-01
2.40323693e-01 6.53539121e-01 -4.57681976e-02 3.08719337e-01
-8.58790129e-02 2.47034639e-01 -5.79876304e-01 2.17880514e-02
-3.40144813e-01 3.39486450e-01 -3.65693927e-01 4.10406351e-01
2.32467785e-01 8.52860332e-01 -7.53353775e-01 -3.63110393e-01
4.34613734e-01 7.35619426e-01 -3.54624301e-01 5.79529524e-01
-1.60252795e-01 5.72359264e-01 -1.20999254e-01 1.92543477e-01
1.09118247e+00 1.99144751e-01 -5.86898625e-01 -4.15269211e-02
1.46077881e-02 -1.71659216e-01 -9.96919215e-01 1.29864502e+00
-3.60686362e-01 4.80373949e-01 -4.71368060e-02 -1.09185576e+00
1.04870713e+00 2.90736049e-01 -5.74820414e-02 -7.12400198e-01
1.98013440e-01 1.83157265e-01 -1.18591050e-02 -5.75041056e-01
5.02640903e-02 -4.98808146e-01 2.48874947e-01 2.23038033e-01
-1.00311674e-01 -4.04696912e-01 -1.97917655e-01 -2.10937075e-02
7.58870721e-01 1.60427485e-02 3.19054313e-02 -1.07131852e-02
6.58478022e-01 -2.42960706e-01 5.77993155e-01 2.86263138e-01
1.95085824e-01 9.71760094e-01 2.55271077e-01 -4.67245370e-01
-1.15898168e+00 -9.83209848e-01 1.85413629e-01 3.52830082e-01
5.16312897e-01 5.83120137e-02 -1.08338118e+00 -5.96166492e-01
-4.02233928e-01 6.77910447e-01 -7.24820554e-01 -8.46441746e-01
-5.45391977e-01 -8.30716133e-01 4.48197663e-01 1.18843392e-01
1.47214687e+00 -1.32334113e+00 -8.04397687e-02 2.34151751e-01
-1.17625825e-01 -7.68835425e-01 -6.30140841e-01 -2.72331893e-01
-7.04973102e-01 -8.18117261e-01 -1.08453393e+00 -9.99746501e-01
8.71957242e-01 3.48152459e-01 5.69616854e-01 4.91147071e-01
-8.37423354e-02 -4.25113261e-01 -5.20816863e-01 -7.11211324e-01
-7.49670506e-01 -1.28288716e-01 -4.55506623e-01 2.96313316e-01
-4.25223336e-02 -6.55958951e-01 -8.13787162e-01 3.22637588e-01
-1.51607978e+00 1.21873714e-01 8.48115683e-01 8.45109165e-01
4.17427570e-01 1.10128117e+00 4.30710137e-01 -8.57376754e-01
4.45433468e-01 -4.81692940e-01 -5.90243816e-01 6.40382543e-02
-6.73849344e-01 -1.87942628e-02 9.29444075e-01 -4.41685438e-01
-1.15509284e+00 -4.32009906e-01 -3.87748837e-01 -5.25352538e-01
-4.04530093e-02 3.75055671e-01 -6.36816442e-01 -2.02867731e-01
5.54838479e-01 7.07594216e-01 2.27326617e-01 -3.19412053e-01
1.08613774e-01 5.76597989e-01 7.06838906e-01 6.16062544e-02
1.38682294e+00 3.97641629e-01 -1.07579842e-01 -6.98872745e-01
-5.20160973e-01 1.41069412e-01 -5.53640500e-02 -1.54745042e-01
1.29551470e+00 -7.71961153e-01 -2.33327985e-01 1.17890739e+00
-1.15161026e+00 -3.90379012e-01 -1.50350064e-01 3.01166385e-01
-1.49147704e-01 3.24680656e-01 -4.59564000e-01 -6.77426279e-01
-4.64272588e-01 -1.04674554e+00 6.04114056e-01 6.35186493e-01
5.39620757e-01 -1.04787898e+00 -1.15075290e-01 2.12715626e-01
6.32759750e-01 7.00787067e-01 9.53082919e-01 -3.01735252e-02
-7.71417558e-01 -1.95835561e-01 -9.56418812e-02 1.07668483e+00
4.99023974e-01 -2.84824409e-02 -8.76882076e-01 -3.13763231e-01
4.47160721e-01 9.79046226e-02 1.07375574e+00 3.08500051e-01
1.26074910e+00 -6.74076080e-01 -2.87788156e-02 9.40149844e-01
1.61574948e+00 6.29648387e-01 1.57633460e+00 5.35560191e-01
9.09133494e-01 2.34182790e-01 3.95484030e-01 -1.00454450e-01
-1.21805415e-01 2.74458945e-01 8.42108250e-01 -5.73498785e-01
-4.19908017e-01 -5.18837690e-01 3.94231111e-01 6.36093140e-01
-1.61625057e-01 -4.89255190e-01 -5.23418903e-01 3.47665340e-01
-1.52035034e+00 -9.52403963e-01 -5.27570350e-03 2.09694600e+00
7.56436169e-01 2.60028392e-01 -3.62231582e-01 3.51107627e-01
9.90065157e-01 4.49674606e-01 -4.59095657e-01 -2.84683049e-01
-2.16590717e-01 3.97042930e-01 4.45366681e-01 4.86530900e-01
-1.00428724e+00 8.77195477e-01 5.31144619e+00 1.06236982e+00
-1.38460183e+00 1.35413840e-01 6.86171949e-01 2.80921310e-01
-3.95535409e-01 -1.16051614e-01 -3.76217902e-01 1.12443542e+00
3.72014165e-01 -9.73472074e-02 6.61773801e-01 5.71367145e-01
9.96281952e-02 -5.27364351e-02 -4.46044505e-01 9.76109982e-01
3.04953367e-01 -1.18482566e+00 3.40156913e-01 1.02979079e-01
1.01167524e+00 -4.60066497e-01 3.52137536e-01 1.36771768e-01
2.24029273e-01 -1.33153641e+00 4.31334198e-01 4.96906042e-01
8.01176012e-01 -1.12837696e+00 1.18812025e+00 4.74127352e-01
-8.90120625e-01 5.08612692e-02 -5.74661672e-01 7.17849880e-02
4.53836508e-02 8.74010205e-01 -4.91699487e-01 6.45481169e-01
6.65238261e-01 5.28866827e-01 -5.18513322e-01 9.96776938e-01
-8.52791846e-01 7.18525410e-01 9.36991051e-02 4.44972277e-01
4.07697499e-01 -3.84515762e-01 4.36257243e-01 8.67398202e-01
4.94314462e-01 1.32537767e-01 -1.87009931e-01 9.08152819e-01
-2.21804991e-01 -1.36860371e-01 -7.18947768e-01 -3.22039649e-02
1.42885238e-01 9.83574808e-01 -4.57606137e-01 -2.59531111e-01
-8.21029022e-02 1.19671464e+00 -2.00029120e-01 4.60780770e-01
-9.17172968e-01 -8.35941315e-01 3.35432857e-01 2.13159263e-01
2.31821790e-01 -1.53814303e-02 -2.84904409e-02 -9.51235771e-01
8.77075829e-03 -9.62609768e-01 1.73970163e-01 -9.71619189e-01
-1.25573790e+00 5.95220447e-01 -3.42921197e-01 -1.15966380e+00
8.02262053e-02 -3.86086196e-01 -1.19040906e+00 1.30886436e+00
-1.70400012e+00 -1.01524723e+00 -8.68850887e-01 6.72694027e-01
5.43577969e-01 -2.80766398e-01 5.14437258e-01 4.09024924e-01
-6.12261653e-01 5.25821924e-01 9.50368121e-02 3.19978952e-01
6.51042640e-01 -9.48241770e-01 2.14860946e-01 1.36021316e+00
-3.22824985e-01 2.78608024e-01 6.48790300e-01 -6.56204998e-01
-7.15197384e-01 -1.42117298e+00 3.10961783e-01 8.13099295e-02
1.38201296e-01 -1.77099064e-01 -1.21420312e+00 5.54426193e-01
7.39743412e-01 -3.87742333e-02 2.45798275e-01 -8.70536447e-01
1.93077754e-02 -4.40361679e-01 -1.37852848e+00 3.76719922e-01
5.07001758e-01 -2.79228091e-01 -5.11840045e-01 1.00929230e-01
9.16634083e-01 -3.36544067e-01 -5.46927452e-01 4.70709443e-01
1.01868600e-01 -1.01305544e+00 8.14132154e-01 -1.42847627e-01
7.53024936e-01 -7.27199137e-01 1.61421165e-01 -1.75297058e+00
-3.73121262e-01 -6.04842126e-01 1.04472190e-01 1.35649085e+00
-1.51799889e-02 -8.40915620e-01 7.23003864e-01 8.63850769e-03
-2.31064186e-01 -5.60615122e-01 -5.29302478e-01 -6.98429108e-01
1.92848921e-01 2.78853655e-01 9.03782725e-01 9.84982312e-01
-7.38168836e-01 2.08464622e-01 -6.16302073e-01 3.37384462e-01
6.38046324e-01 -3.81830424e-01 7.16251314e-01 -7.48674810e-01
-1.23941541e-01 -3.22473735e-01 -6.47536218e-01 -8.51387262e-01
-4.18658108e-02 -5.57531416e-01 8.60262737e-02 -1.72891390e+00
-9.03572068e-02 -3.24004740e-01 -1.97166473e-01 4.04996961e-01
-6.32388055e-01 4.35135156e-01 3.59376781e-02 2.63974905e-01
1.25882089e-01 8.26584458e-01 1.81221318e+00 -4.97180700e-01
-2.92495731e-02 1.58661250e-02 -7.15350151e-01 5.54208577e-01
1.18226492e+00 -4.78427559e-01 -6.25762582e-01 -6.05893254e-01
1.03380531e-02 -2.29613841e-01 3.95169377e-01 -1.16957653e+00
2.26506680e-01 -2.41541684e-01 6.21001720e-01 -3.07895720e-01
1.47248834e-01 -8.55228424e-01 3.54419202e-01 7.12887347e-01
1.03415839e-01 -5.35455979e-02 7.94837996e-02 4.08650845e-01
-6.48965836e-01 -3.46163303e-01 1.27645290e+00 -3.25965673e-01
-6.03310645e-01 4.54974830e-01 -2.75100559e-01 9.31180492e-02
1.08016467e+00 -3.45830500e-01 -4.67818886e-01 -4.76168931e-01
-2.47634158e-01 6.86845258e-02 5.25643587e-01 2.27557346e-01
9.11469758e-01 -1.41699219e+00 -8.18079233e-01 4.34220523e-01
-2.40034491e-01 5.65352976e-01 4.56562221e-01 4.56953168e-01
-9.85266984e-01 -2.12488979e-01 -4.62540418e-01 -1.49250895e-01
-9.82873201e-01 8.78971159e-01 4.55472440e-01 -1.45480722e-01
-5.43528676e-01 7.67309129e-01 7.26684391e-01 8.47330764e-02
-7.95177966e-02 -3.70709645e-03 -2.87891269e-01 -3.78606707e-01
7.50375986e-01 2.55888581e-01 -7.24312738e-02 -4.60307091e-01
8.32078904e-02 5.01382351e-01 -8.49388614e-02 1.07783945e-02
1.29901373e+00 8.34879279e-02 -5.68314016e-01 6.51526228e-02
1.12164438e+00 8.03233460e-02 -1.35574639e+00 3.26661542e-02
-8.14700186e-01 -6.41030014e-01 2.93109089e-01 -7.07001686e-01
-1.55460346e+00 1.15871036e+00 6.74913406e-01 3.56519997e-01
1.42695904e+00 -5.76029301e-01 1.01547515e+00 -2.23743305e-01
5.90126030e-02 -7.86042809e-01 3.82920623e-01 1.34050205e-01
9.62151051e-01 -1.02070487e+00 -2.55361106e-02 -4.42614943e-01
-5.26197791e-01 8.45681250e-01 8.38387907e-01 -4.55355108e-01
5.49939811e-01 7.73996338e-02 2.36885175e-01 -1.89064592e-02
-1.24040589e-01 1.17679492e-01 1.16913818e-01 6.05013847e-01
-4.03140485e-02 -2.46469274e-01 -2.53608167e-01 2.65013725e-01
-2.92188168e-01 -7.54848123e-02 5.18841982e-01 6.35982573e-01
-5.91234088e-01 -9.98303652e-01 -5.55975676e-01 2.00462997e-01
-5.18933296e-01 -1.60414606e-01 -1.81291372e-01 6.85441256e-01
5.93756139e-01 9.15944040e-01 -5.05518354e-02 -5.10103345e-01
1.52887613e-01 -2.60455638e-01 3.83325785e-01 -4.96459961e-01
-3.00600916e-01 -7.95102715e-02 -5.48752546e-01 -1.14993744e-01
-2.97901809e-01 -2.27203197e-03 -1.18343806e+00 -5.21077871e-01
-3.07483613e-01 3.48742306e-01 5.23392677e-01 8.38225782e-01
5.11494018e-02 9.19441402e-01 9.41181898e-01 -5.00816405e-01
-1.08764261e-01 -1.03708160e+00 -7.96745241e-01 4.55269635e-01
4.15598601e-01 -4.93565738e-01 -5.67205071e-01 2.25389704e-01] | [10.923036575317383, -3.040510416030884] |
f6560cbc-139a-48f3-863c-9a7ce20cc4ba | explainable-automated-coding-of-clinical | 2010.15728 | null | https://arxiv.org/abs/2010.15728v4 | https://arxiv.org/pdf/2010.15728v4.pdf | Explainable Automated Coding of Clinical Notes using Hierarchical Label-wise Attention Networks and Label Embedding Initialisation | Diagnostic or procedural coding of clinical notes aims to derive a coded summary of disease-related information about patients. Such coding is usually done manually in hospitals but could potentially be automated to improve the efficiency and accuracy of medical coding. Recent studies on deep learning for automated medical coding achieved promising performances. However, the explainability of these models is usually poor, preventing them to be used confidently in supporting clinical practice. Another limitation is that these models mostly assume independence among labels, ignoring the complex correlation among medical codes which can potentially be exploited to improve the performance. We propose a Hierarchical Label-wise Attention Network (HLAN), which aimed to interpret the model by quantifying importance (as attention weights) of words and sentences related to each of the labels. Secondly, we propose to enhance the major deep learning models with a label embedding (LE) initialisation approach, which learns a dense, continuous vector representation and then injects the representation into the final layers and the label-wise attention layers in the models. We evaluated the methods using three settings on the MIMIC-III discharge summaries: full codes, top-50 codes, and the UK NHS COVID-19 shielding codes. Experiments were conducted to compare HLAN and LE initialisation to the state-of-the-art neural network based methods. HLAN achieved the best Micro-level AUC and $F_1$ on the top-50 code prediction and comparable results on the NHS COVID-19 shielding code prediction to other models. By highlighting the most salient words and sentences for each label, HLAN showed more meaningful and comprehensive model interpretation compared to its downgraded baselines and the CNN-based models. LE initialisation consistently boosted most deep learning models for automated medical coding. | ['Honghan Wu', 'William Whiteley', 'Víctor Suárez-Paniagua', 'Hang Dong'] | 2020-10-29 | null | null | null | null | ['medical-code-prediction'] | ['medical'] | [ 3.88454944e-01 6.11286700e-01 -1.88603461e-01 -5.68856955e-01
-9.21884418e-01 -2.16994286e-01 3.16406399e-01 8.17918301e-01
-3.47576708e-01 4.23855782e-01 9.02782798e-01 -6.99263752e-01
-4.05768573e-01 -5.06695509e-01 -3.32242191e-01 -4.99310732e-01
-9.72961932e-02 8.52890074e-01 -4.59943950e-01 1.75654650e-01
-1.53761180e-02 1.54263392e-01 -1.07484174e+00 1.01668978e+00
7.88920939e-01 1.05707216e+00 -5.48342206e-02 7.43955851e-01
-1.81684479e-01 1.52232039e+00 -3.52968931e-01 -6.56324029e-01
-5.52911907e-02 -5.99223673e-01 -9.91941452e-01 -9.93560851e-02
2.15702802e-01 -3.74929696e-01 -1.53552070e-01 8.27498674e-01
7.23746359e-01 -3.78018081e-01 1.11738658e+00 -8.31720352e-01
-8.59787643e-01 7.64634967e-01 -2.74695039e-01 1.36252880e-01
2.69728392e-01 5.68115897e-03 1.30668640e+00 -6.40450835e-01
3.42822433e-01 1.09249985e+00 1.22740757e+00 6.52552247e-01
-1.20649493e+00 -6.37245655e-01 -8.18989202e-02 1.37441773e-02
-1.18005586e+00 -1.32121906e-01 2.91492939e-01 -8.68267000e-01
1.25057411e+00 3.26930791e-01 6.53682411e-01 1.10437119e+00
4.48346704e-01 5.91130257e-01 7.94266999e-01 -4.56320554e-01
1.74832810e-02 2.32393384e-01 3.39186668e-01 8.86234820e-01
2.72806615e-01 -5.37101924e-02 6.66022748e-02 -4.86302167e-01
4.69164819e-01 5.58723271e-01 -3.99020106e-01 -1.15818679e-01
-1.40882182e+00 1.18701470e+00 8.59289706e-01 1.95597917e-01
-6.66650653e-01 9.00547728e-02 7.90534973e-01 -2.05089778e-01
5.67655802e-01 8.01282048e-01 -6.62035882e-01 -8.26699212e-02
-9.81368124e-01 -6.43387949e-03 7.39914238e-01 7.31888533e-01
1.10925309e-01 -2.66596377e-01 -8.87936056e-01 9.65550542e-01
4.10598189e-01 1.99287087e-01 5.87136209e-01 -8.45292211e-01
6.43238187e-01 9.14255857e-01 -1.42110273e-01 -9.86715734e-01
-8.87948334e-01 -6.58934057e-01 -1.25815785e+00 -1.90946594e-01
-4.16543409e-02 -3.10311019e-01 -1.18077958e+00 1.43610656e+00
-2.96993911e-01 1.86911914e-02 1.86641827e-01 5.71868837e-01
1.27611148e+00 4.82442051e-01 5.02522588e-01 -7.64358565e-02
1.63768923e+00 -1.08999813e+00 -9.62007642e-01 -1.77920520e-01
1.34188795e+00 -3.67235243e-01 6.39539182e-01 4.07449380e-02
-9.28154051e-01 -4.69905764e-01 -5.58386326e-01 -1.69222847e-01
-2.35701367e-01 2.27276206e-01 4.50048119e-01 3.81373316e-01
-1.17633402e+00 4.07152325e-01 -7.60164201e-01 -8.83129239e-02
8.78065646e-01 4.29292262e-01 -1.96155995e-01 -1.36272132e-01
-1.40517688e+00 9.96636093e-01 4.01557237e-01 -4.14983183e-02
-7.19880998e-01 -9.92950082e-01 -1.12483692e+00 6.05770111e-01
-2.75079578e-01 -1.16161466e+00 1.33569419e+00 -8.07363510e-01
-8.22307885e-01 1.18129730e+00 -1.59489904e-02 -7.07903385e-01
4.11746562e-01 -3.03445198e-02 -2.24422231e-01 1.57416329e-01
1.40845925e-01 9.95753169e-01 2.62499481e-01 -1.10437012e+00
-5.93715966e-01 8.85127783e-02 -4.04497720e-02 1.18497692e-01
-3.82759988e-01 -1.78341135e-01 -3.09110675e-02 -7.13494003e-01
-3.03287268e-01 -9.50415254e-01 -3.79520029e-01 -1.07676044e-01
-4.54297185e-01 -2.14924544e-01 2.57573407e-02 -8.70868325e-01
1.67473125e+00 -2.19809031e+00 -6.65809214e-02 -3.32964957e-02
6.75292134e-01 4.54527408e-01 3.88655663e-02 4.53392088e-01
-5.54563761e-01 4.06201303e-01 -4.82390493e-01 -5.62197447e-01
-9.64348912e-02 3.32997084e-01 -2.08218679e-01 2.26492360e-01
4.74115700e-01 1.22833598e+00 -1.00762510e+00 -6.43074214e-01
2.08203822e-01 7.09925175e-01 -1.04267240e+00 3.34857941e-01
1.75359488e-01 3.11106741e-01 -1.99217439e-01 4.66544479e-01
2.57942647e-01 -8.61082077e-01 2.14366503e-02 -1.46751955e-01
3.83286864e-01 5.04547656e-01 -3.61204177e-01 1.32734454e+00
-5.85279226e-01 5.72822511e-01 -3.66907746e-01 -1.15071142e+00
6.56262577e-01 7.30494678e-01 6.90983236e-01 -2.86351234e-01
2.81657428e-01 1.13116607e-01 3.56974304e-01 -7.99301088e-01
3.51433293e-03 -4.03421938e-01 -1.15848228e-01 2.20959648e-01
-1.25902385e-01 -2.53211837e-02 -3.13732296e-01 2.42993698e-01
1.23538733e+00 -5.30935168e-01 6.89724028e-01 -2.88431615e-01
4.02390599e-01 -7.53886327e-02 4.14181799e-01 8.59244943e-01
-1.35809511e-01 1.02494371e+00 6.64871812e-01 -7.48042881e-01
-8.98951948e-01 -6.30169213e-01 -5.83272457e-01 8.54624510e-01
-5.51819980e-01 -5.10254979e-01 -6.33391798e-01 -9.49196517e-01
1.12194672e-01 8.94702792e-01 -1.18464243e+00 -3.63270730e-01
-2.73651719e-01 -8.95770371e-01 6.03057027e-01 9.28235710e-01
8.45721439e-02 -1.47287130e+00 -7.20636547e-01 4.20863867e-01
-3.80037397e-01 -8.27711344e-01 -3.40881795e-01 6.18845344e-01
-7.39424407e-01 -9.76082504e-01 -8.54718447e-01 -8.49801421e-01
7.52093017e-01 -3.45257282e-01 1.36773157e+00 3.38890016e-01
-1.76130503e-01 2.55037844e-01 -4.77494985e-01 -7.57864654e-01
-6.49881244e-01 2.45509699e-01 -3.64325702e-01 -4.52764146e-02
6.17972612e-01 4.60192263e-02 -7.56237388e-01 -3.84805739e-01
-9.30224717e-01 3.55279714e-01 6.51963830e-01 1.37685287e+00
2.66803920e-01 -4.32871401e-01 5.11911511e-01 -1.31516325e+00
7.98270226e-01 -5.97836852e-01 2.21314713e-01 7.46899620e-02
-8.34973097e-01 1.73070297e-01 5.86794436e-01 2.71698665e-02
-4.55081612e-01 -5.56898527e-02 -6.29398763e-01 -2.46369317e-01
-1.71235844e-01 8.04869115e-01 4.55668390e-01 3.97921920e-01
6.05519235e-01 -4.61854972e-02 -2.70704366e-02 -2.94191152e-01
-6.27688617e-02 1.09333611e+00 1.34046465e-01 -7.03301206e-02
1.10966086e-01 2.26553649e-01 -1.53660133e-01 -2.21952036e-01
-1.37547696e+00 -5.80794752e-01 -5.54367959e-01 3.05118412e-01
1.33292568e+00 -9.94263411e-01 -6.06267095e-01 1.05560357e-02
-1.33054316e+00 -2.01460421e-01 -3.07971537e-01 5.04998028e-01
-4.97312874e-01 7.96893835e-02 -9.18986857e-01 -5.08078754e-01
-5.44641376e-01 -1.41401899e+00 1.40987241e+00 -2.82740533e-01
-9.39828575e-01 -1.39346230e+00 7.66047835e-02 3.37312371e-01
4.10668343e-01 3.97802293e-01 1.49457574e+00 -1.10802317e+00
2.22012907e-01 -2.80601948e-01 -6.67350769e-01 4.33486581e-01
3.79630536e-01 -2.46242732e-01 -1.02643609e+00 -2.50901788e-01
-7.30467811e-02 -2.24565119e-01 1.11476755e+00 7.69906998e-01
1.57176399e+00 -5.33479869e-01 -4.83943135e-01 6.97509229e-01
1.22519112e+00 3.76963168e-01 4.91172910e-01 2.73547679e-01
8.98620188e-01 5.03377080e-01 3.37390937e-02 5.51521719e-01
4.70782101e-01 4.19981182e-01 5.72895586e-01 -6.30835533e-01
-3.11449971e-02 8.75551924e-02 -6.56338260e-02 1.06692898e+00
6.27024919e-02 -1.59852445e-01 -1.42815697e+00 7.58317411e-01
-1.72624791e+00 -6.11276329e-01 -2.73528665e-01 1.55294752e+00
9.92518067e-01 1.04419947e-01 -3.26599747e-01 1.91433057e-01
5.53719640e-01 -5.08580282e-02 -2.95792282e-01 -8.62138093e-01
2.86117017e-01 2.08531007e-01 4.01980937e-01 3.85807991e-01
-1.16769481e+00 2.80919224e-01 6.36434555e+00 4.19543356e-01
-8.80611658e-01 2.19453216e-01 1.11558235e+00 9.08173341e-03
-2.80903429e-01 -5.74599743e-01 -6.70117438e-01 6.07225358e-01
1.37588680e+00 2.49817878e-01 -1.62243962e-01 7.68583655e-01
1.78828865e-01 4.74782228e-01 -1.43764091e+00 9.83761370e-01
5.86638115e-02 -1.57719588e+00 2.95060217e-01 1.44340277e-01
9.41306293e-01 1.42680734e-01 1.06893815e-01 4.96431440e-01
5.25473297e-01 -1.37706888e+00 4.10114259e-01 5.62623680e-01
1.16544950e+00 -5.45371711e-01 1.52744603e+00 1.77359894e-01
-7.77405143e-01 -4.31871712e-01 -2.25186124e-01 -6.62157238e-02
8.13155323e-02 3.25412869e-01 -1.30820787e+00 3.45732331e-01
6.93129599e-01 1.01263642e+00 -6.37987077e-01 9.22256529e-01
-5.84455952e-02 8.56935918e-01 2.31685802e-01 9.91017073e-02
7.79187500e-01 3.52217466e-01 -1.01216719e-01 1.92845440e+00
3.77944976e-01 3.09272528e-01 7.22001940e-02 7.24475503e-01
-2.70474941e-01 1.46759167e-01 -6.38705790e-01 3.58141190e-03
7.61447027e-02 9.63807523e-01 -6.05183899e-01 -7.89667010e-01
-5.41191876e-01 6.80966377e-01 3.73753548e-01 2.07077786e-01
-9.36263144e-01 -2.15339079e-01 5.28426945e-01 2.32883051e-01
3.97034824e-01 6.04972839e-01 -6.55506492e-01 -9.02203918e-01
-5.13460398e-01 -9.36695039e-01 6.91543043e-01 -7.27822602e-01
-1.35244012e+00 9.27072108e-01 -2.09133506e-01 -1.37731922e+00
-3.45762670e-01 -7.60256350e-01 -6.45058155e-02 8.19501281e-01
-1.67593563e+00 -9.48972702e-01 -3.38910073e-01 2.91493922e-01
5.37724316e-01 -1.87854990e-01 1.42807341e+00 3.56573045e-01
-3.86981189e-01 7.52870917e-01 2.41650313e-01 6.48494363e-01
7.11396992e-01 -1.40917468e+00 1.81158319e-01 2.00201720e-01
-8.82209465e-02 6.20507061e-01 2.81390131e-01 -4.30191249e-01
-3.53418380e-01 -1.56867015e+00 1.34669745e+00 -7.14858711e-01
4.31363881e-01 -4.95177917e-02 -1.02073181e+00 8.09448481e-01
3.10767055e-01 8.56842399e-02 1.19541609e+00 -7.25268722e-02
-2.69922882e-01 2.00272694e-01 -1.01683748e+00 2.35657409e-01
7.53478587e-01 -5.26158154e-01 -8.74908805e-01 5.17849684e-01
1.04254603e+00 -3.15139353e-01 -1.07115650e+00 5.66058815e-01
4.59161431e-01 -8.12465370e-01 1.00376952e+00 -9.61187720e-01
1.17687070e+00 1.88207597e-01 1.16662420e-02 -1.39638364e+00
-9.01893377e-01 3.21649462e-02 4.72392328e-02 8.29928756e-01
7.25180745e-01 -5.59980094e-01 3.77139956e-01 7.47089505e-01
-4.24671531e-01 -1.15355849e+00 -6.95347011e-01 -1.20865323e-01
1.97826207e-01 -2.80161440e-01 5.29857039e-01 1.32961512e+00
1.14040658e-01 3.19815397e-01 -3.21081758e-01 -3.16735618e-02
1.54082835e-01 -1.66799992e-01 -5.78812510e-02 -1.37631702e+00
-1.52198419e-01 -6.43874884e-01 -3.81925792e-01 -4.90072489e-01
1.09055497e-01 -1.39984024e+00 2.36190762e-02 -2.25915432e+00
5.70639431e-01 -4.42362159e-01 -7.96326101e-01 8.64246190e-01
-5.94797254e-01 2.28310570e-01 8.60845372e-02 2.33439490e-01
-4.99961495e-01 1.89147055e-01 1.14710474e+00 -4.53762561e-01
-9.85362660e-03 -1.58596545e-01 -9.72341657e-01 6.88846767e-01
5.85240662e-01 -8.71420681e-01 -2.02599362e-01 -6.21716440e-01
3.24728161e-01 1.84382647e-01 3.05221528e-01 -8.14469039e-01
2.76682600e-02 2.79418290e-01 3.89672011e-01 -3.76289427e-01
-1.04043640e-01 -1.12472117e+00 4.33859713e-02 9.75114167e-01
-1.04072976e+00 1.39894158e-01 3.20264041e-01 4.79840040e-01
-3.28504741e-01 -3.32609028e-01 6.50728047e-01 -2.67814428e-01
-8.25402737e-02 1.66571632e-01 -4.98348176e-01 1.67622432e-01
6.67343378e-01 -1.05969615e-01 -1.42527465e-02 -3.59416604e-01
-8.06752086e-01 1.13862887e-01 -7.59313703e-02 2.54039526e-01
5.72199643e-01 -1.30336714e+00 -1.07891631e+00 2.19340354e-01
4.74572539e-01 1.47547573e-01 2.57208973e-01 1.07050741e+00
-9.21934843e-01 8.88316274e-01 2.75972895e-02 -7.77554214e-01
-1.02937222e+00 8.35626423e-01 3.67824942e-01 -9.36389208e-01
-7.15285838e-01 8.56905401e-01 5.96543133e-01 -4.84449863e-01
3.07205379e-01 -1.00527632e+00 -6.57450855e-01 3.12957972e-01
5.56043386e-01 -7.34248683e-02 2.98429936e-01 -6.11974537e-01
-4.80940521e-01 5.41644812e-01 -2.80444235e-01 4.67143685e-01
1.53818572e+00 8.17440525e-02 -9.07603577e-02 3.76245558e-01
1.55392897e+00 -4.80575740e-01 -8.39695096e-01 -1.90256625e-01
7.78975710e-02 5.12293950e-02 2.07024142e-01 -1.00111783e+00
-1.05011225e+00 1.29301775e+00 5.73865175e-01 2.87046373e-01
1.00014400e+00 -1.23495996e-01 6.90876901e-01 9.39967185e-02
-2.64395416e-01 -4.92531806e-01 -1.02801658e-01 4.28614765e-01
8.93714309e-01 -1.55834174e+00 -2.02137455e-01 -5.25238588e-02
-1.01391637e+00 1.10894370e+00 1.33187264e-01 8.81889015e-02
7.67449081e-01 2.37575993e-01 2.94439852e-01 -4.70637828e-01
-8.11624229e-01 1.10052973e-01 6.66113198e-01 4.08158004e-01
8.69001150e-01 2.59710014e-01 -1.00890525e-01 7.69459426e-01
-4.02460657e-02 3.22614722e-02 4.12161976e-01 5.41511714e-01
-1.00511879e-01 -6.50784791e-01 -1.31310105e-01 1.08627665e+00
-9.75938737e-01 -7.36908317e-01 -1.72960460e-01 5.66673756e-01
2.71385193e-01 7.55734324e-01 8.43071714e-02 -3.35074931e-01
3.45911324e-01 3.80192161e-01 -2.82714635e-01 -1.03699362e+00
-1.15570331e+00 -4.40402888e-02 1.27880484e-01 -3.14893574e-01
-4.85706210e-01 -5.79780936e-01 -1.29305708e+00 6.82192370e-02
-8.36986527e-02 1.63705200e-01 1.30100980e-01 7.77323127e-01
5.73037386e-01 1.32343471e+00 1.76047266e-01 -4.93383169e-01
-4.18643862e-01 -1.18481016e+00 -2.04360053e-01 7.76924849e-01
7.89791107e-01 -5.45347333e-01 -4.08839375e-01 2.54528135e-01] | [8.012855529785156, 6.793702125549316] |
228fb2be-5420-402e-8335-c1fd70ee9dbf | a-3d-cnn-network-with-bert-for-automatic | 2106.14403 | null | https://arxiv.org/abs/2106.14403v3 | https://arxiv.org/pdf/2106.14403v3.pdf | A 3D CNN Network with BERT For Automatic COVID-19 Diagnosis From CT-Scan Images | We present an automatic COVID1-19 diagnosis framework from lung CT-scan slice images. In this framework, the slice images of a CT-scan volume are first proprocessed using segmentation techniques to filter out images of closed lung, and to remove the useless background. Then a resampling method is used to select one or multiple sets of a fixed number of slice images for training and validation. A 3D CNN network with BERT is used to classify this set of selected slice images. In this network, an embedding feature is also extracted. In cases where there are more than one set of slice images in a volume, the features of all sets are extracted and pooled into a global feature vector for the whole CT-scan volume. A simple multiple-layer perceptron (MLP) network is used to further classify the aggregated feature vector. The models are trained and evaluated on the provided training and validation datasets. On the validation dataset, the accuracy is 0.9278 and the F1 score is 0.9261. | ['Jingfeng Liu', 'Weijun Tan'] | 2021-06-28 | null | null | null | null | ['covid-19-detection'] | ['medical'] | [ 4.76240307e-01 1.68292359e-01 -1.65407762e-01 -4.10634369e-01
-8.28300536e-01 -2.01783106e-01 1.31957084e-01 5.13621330e-01
-6.91711009e-01 5.03056765e-01 -3.08673307e-02 -2.20686197e-01
-2.42037266e-01 -8.56776655e-01 -2.72422194e-01 -9.18788612e-01
-2.48941317e-01 5.95081568e-01 7.05630839e-01 5.27289689e-01
-2.50203252e-01 7.95699179e-01 -1.19520330e+00 4.44615155e-01
3.55869949e-01 1.12802815e+00 2.93161452e-01 8.31845820e-01
-1.29279554e-01 9.67883825e-01 -4.02418256e-01 2.32843772e-01
3.06457788e-01 -5.32453299e-01 -8.61861885e-01 3.80218655e-01
1.30008548e-01 -4.30868953e-01 -1.74787104e-01 8.74119222e-01
5.12634575e-01 1.24457255e-01 7.06470788e-01 -6.93626106e-01
3.67446959e-01 4.66371894e-01 -3.90545905e-01 3.48130971e-01
-2.90932432e-02 2.84506511e-02 8.02076638e-01 -1.15652955e+00
6.03412747e-01 6.24673426e-01 4.69553351e-01 3.39366555e-01
-1.00816941e+00 -4.54219371e-01 -3.67350787e-01 -6.13855803e-03
-1.24497306e+00 3.24362397e-01 7.65020370e-01 -5.68647623e-01
6.01660669e-01 4.61427748e-01 1.08959758e+00 4.91309524e-01
7.82340467e-01 4.70539451e-01 9.19521213e-01 -3.56469929e-01
2.00859323e-01 4.04353082e-01 3.22016090e-01 8.36170912e-01
3.29224437e-01 -9.28247124e-02 1.92534700e-02 -1.53811157e-01
6.29994035e-01 4.42900807e-01 -2.41090953e-01 -4.43325639e-01
-1.15752816e+00 9.44746733e-01 8.19556117e-01 5.74407220e-01
-8.15031648e-01 -1.86251074e-01 5.31251311e-01 -2.63246614e-02
2.52413958e-01 2.55680352e-01 -1.07932277e-01 5.13486445e-01
-1.18464541e+00 -8.70590732e-02 7.52530038e-01 2.39359930e-01
5.96417129e-01 -2.37585515e-01 -2.46971682e-01 7.70926833e-01
1.70525461e-01 1.84628889e-01 8.07703555e-01 -3.94994587e-01
2.80575961e-01 9.69698668e-01 -3.67299974e-01 -6.60281718e-01
-7.83422768e-01 -4.85512137e-01 -1.02836084e+00 4.04651999e-01
2.52399564e-01 -1.03015073e-01 -1.13040841e+00 9.19534922e-01
5.81388712e-01 6.40029495e-04 -1.22716203e-01 8.76994431e-01
1.02037501e+00 3.82303923e-01 8.92257839e-02 -2.87888020e-01
1.40974641e+00 -8.33222687e-01 -2.98760623e-01 9.03570652e-02
5.20628870e-01 -5.60093343e-01 6.79391563e-01 2.67125070e-01
-9.36837435e-01 -5.96247733e-01 -1.26611447e+00 3.33000034e-01
-2.31822267e-01 3.04429561e-01 8.89169425e-02 3.22398543e-01
-7.53106952e-01 5.79197645e-01 -1.01903617e+00 -3.38978842e-02
6.10545397e-01 6.03160918e-01 -3.61150444e-01 -1.14858389e-01
-9.21104312e-01 9.25289989e-01 6.47238612e-01 -9.23892558e-02
-9.33675766e-01 -6.68127954e-01 -6.55835450e-01 2.88191121e-02
3.35508287e-01 -5.73997438e-01 9.44986939e-01 -9.19774711e-01
-1.11605370e+00 8.31249535e-01 1.28224179e-01 -4.51712608e-01
6.97250783e-01 2.86797345e-01 -2.36889809e-01 6.09846950e-01
1.14712073e-02 3.91276568e-01 9.25906539e-01 -1.16893744e+00
-6.87345743e-01 -3.65300119e-01 -3.01026940e-01 2.46469170e-01
-9.19446126e-02 6.63740188e-02 -4.89863336e-01 -4.23849434e-01
4.31804508e-01 -9.34029222e-01 -4.88826901e-01 -6.18445016e-02
-5.67844272e-01 1.24721274e-01 7.94174910e-01 -7.59929240e-01
1.09609520e+00 -2.16597605e+00 -9.27237123e-02 6.92218482e-01
6.09405935e-01 7.99390897e-02 1.25786275e-01 -1.07508309e-01
-3.90246928e-01 -2.23769173e-01 -5.15282631e-01 -5.24731614e-02
-7.13735640e-01 8.02024901e-02 3.73808384e-01 4.81431603e-01
1.61468580e-01 6.03366971e-01 -6.81473434e-01 -1.01089847e+00
5.47365010e-01 4.37122285e-01 -2.92980194e-01 3.88133496e-01
6.72838241e-02 5.54029167e-01 -5.80479026e-01 4.17239934e-01
5.93945861e-01 -3.18832338e-01 8.66234452e-02 -2.10711882e-01
1.18447550e-01 2.94791795e-02 -1.26019239e+00 1.25133348e+00
-5.82378745e-01 3.18390191e-01 -2.02068910e-01 -7.41157830e-01
6.88809454e-01 4.96660113e-01 1.13491392e+00 -2.12707996e-01
4.09594983e-01 3.29116970e-01 3.75909656e-01 -7.92874575e-01
-2.60000154e-02 -2.34327137e-01 5.35154194e-02 6.59497321e-01
1.64940581e-01 -4.06755328e-01 1.31027803e-01 -1.09241553e-01
1.15198338e+00 -4.99714792e-01 5.35866141e-01 -8.72683898e-02
8.22236180e-01 1.78911701e-01 1.88713759e-01 5.49638689e-01
-2.90786207e-01 7.89648592e-01 3.37202221e-01 -7.39883304e-01
-9.95192468e-01 -9.75257337e-01 -4.69120383e-01 5.56165159e-01
-1.30655631e-01 -4.26387824e-02 -8.51907194e-01 -1.09364021e+00
-1.54664516e-01 4.54743534e-01 -7.96994328e-01 5.39300591e-03
-6.95739150e-01 -8.15920770e-01 9.13927704e-02 5.08468390e-01
3.06658447e-01 -1.28191757e+00 -1.29574883e+00 4.36929256e-01
4.84877452e-02 -7.50746191e-01 -1.55264303e-01 6.07584834e-01
-1.08466697e+00 -1.28688049e+00 -6.72715485e-01 -8.43619585e-01
8.87457609e-01 -3.08799278e-03 7.80471921e-01 3.18644524e-01
-5.28417826e-01 -1.09488524e-01 -8.51535276e-02 -2.59127140e-01
-6.55000091e-01 -6.85813930e-03 -2.23355740e-01 5.48389815e-02
1.68367088e-01 -1.95276007e-01 -5.30962706e-01 2.30226037e-03
-9.09182787e-01 -1.00932986e-01 6.53113008e-01 1.13261425e+00
9.97459650e-01 2.40345731e-01 1.61514506e-01 -9.95219469e-01
4.24859107e-01 -4.37803060e-01 -5.13144433e-01 1.55702263e-01
-2.11823240e-01 -1.28404871e-01 7.24387050e-01 -3.95110875e-01
-6.87443376e-01 3.94036353e-01 -1.11721754e-01 -6.02699816e-01
-1.75130844e-01 6.23832703e-01 2.05869094e-01 -2.71621849e-02
6.68541312e-01 1.86122239e-01 5.57474270e-02 -2.02741936e-01
-1.25304624e-01 6.44651175e-01 4.12995040e-01 2.58585364e-01
5.40294707e-01 4.30625111e-01 2.16367647e-01 -8.24448049e-01
-6.10503078e-01 -6.56805158e-01 -1.14074266e+00 -4.85015661e-01
1.13267767e+00 -4.63630944e-01 -5.20263687e-02 1.52244315e-01
-7.28059053e-01 -7.24989697e-02 -5.47574759e-01 1.18063879e+00
-2.74580419e-01 1.44933715e-01 -4.41386908e-01 -6.67796075e-01
-5.94528913e-01 -1.59359443e+00 7.26703644e-01 3.44329983e-01
-2.99003094e-01 -7.67844856e-01 1.38460621e-01 9.99633446e-02
1.85759440e-01 3.39973241e-01 1.00061822e+00 -1.34456611e+00
-3.91950756e-01 -7.35335886e-01 2.16171518e-02 5.05432069e-01
3.15183669e-01 3.23540084e-02 -8.67402375e-01 -1.99571341e-01
3.97205532e-01 -8.90536159e-02 9.13747907e-01 6.75430954e-01
1.42443609e+00 -3.75692248e-02 -5.14273942e-01 5.30939221e-01
1.54979086e+00 4.84999299e-01 1.96721386e-02 9.62800458e-02
5.88688850e-01 3.05817783e-01 3.34337831e-01 3.08791101e-01
-1.26104429e-01 1.09200358e-01 5.47441244e-01 -3.17169547e-01
-8.76331031e-02 2.13140994e-01 -1.26081675e-01 6.19574606e-01
-6.00599833e-02 -5.38817374e-03 -9.22296822e-01 5.17394722e-01
-1.08049417e+00 -6.44468069e-01 -1.50391147e-01 2.26808071e+00
4.87882614e-01 3.67769599e-01 1.14206403e-01 4.28626537e-01
7.22762942e-01 1.57673061e-01 -4.06598926e-01 -2.05563098e-01
2.73795903e-01 6.40938461e-01 4.87500638e-01 4.11408931e-01
-1.35072470e+00 3.82503718e-01 5.78783369e+00 5.25736451e-01
-1.48762691e+00 7.06462190e-02 7.23443270e-01 -2.08314598e-01
7.49549717e-02 -3.17177027e-01 -3.61740619e-01 2.39249200e-01
7.72306025e-01 1.58800587e-01 4.08428051e-02 8.57030988e-01
1.73612595e-01 -4.82685685e-01 -8.56303155e-01 6.19286597e-01
1.14818484e-01 -1.04157627e+00 -1.61539927e-01 9.22644231e-03
6.09027624e-01 3.27707648e-01 -1.60011113e-01 1.08475842e-01
-1.76818464e-02 -1.10661995e+00 3.00423801e-01 5.18047631e-01
7.30089784e-01 -7.43095100e-01 1.18968987e+00 4.36429381e-01
-1.02652633e+00 -1.65085569e-01 -1.65756524e-01 6.08757615e-01
3.55974101e-02 7.78889894e-01 -1.41512179e+00 6.62677467e-01
6.90274477e-01 2.96060115e-01 -6.73666000e-01 1.22143686e+00
-8.81150588e-02 6.24825835e-01 -5.06011426e-01 -3.44134122e-02
2.63880432e-01 -5.41305616e-02 4.82247472e-01 1.16118050e+00
2.01135412e-01 6.08916618e-02 3.47034395e-01 5.36461771e-01
3.89718600e-02 4.46314067e-01 -5.09310484e-01 4.00865704e-01
2.02915400e-01 1.58261108e+00 -1.15766072e+00 -7.59606719e-01
-4.49515402e-01 7.90668607e-01 -3.43544073e-02 -4.58694845e-02
-7.36110091e-01 -3.12723428e-01 -2.01306418e-01 1.67044431e-01
4.67855275e-01 2.75999844e-01 -3.66036564e-01 -7.24507093e-01
-1.58439949e-01 -5.54183602e-01 7.21700728e-01 -5.34517705e-01
-9.62577820e-01 1.04244590e+00 -1.82759259e-02 -1.48799777e+00
-3.44489157e-01 -5.47313094e-01 -8.86161506e-01 7.52097189e-01
-1.17564762e+00 -7.82682776e-01 -4.93939549e-01 5.11843503e-01
5.21324694e-01 -1.82892472e-01 8.19872499e-01 5.04671736e-03
-4.41807419e-01 1.15358725e-01 1.22148745e-01 4.91899788e-01
2.79184848e-01 -1.29983532e+00 -3.43977422e-01 6.59107924e-01
-4.85017635e-02 2.35486969e-01 2.13944212e-01 -6.51202202e-01
-7.69412875e-01 -1.42430556e+00 7.91381955e-01 -1.75673589e-01
2.72737324e-01 -1.34665430e-01 -1.09977424e+00 6.00853205e-01
-6.46822006e-02 5.93211591e-01 7.24785447e-01 -7.42917418e-01
2.13477537e-01 5.47668263e-02 -1.35235095e+00 2.22060680e-01
1.33724526e-01 -2.32753977e-01 -7.29837298e-01 2.33658716e-01
4.53670800e-01 -7.12608039e-01 -1.20743537e+00 3.59782338e-01
5.98801255e-01 -8.63323629e-01 9.99955714e-01 -3.74422014e-01
4.55782712e-01 -2.74684072e-01 -3.18388306e-02 -1.27228928e+00
-2.97347754e-01 2.08043560e-01 3.18160862e-01 4.38652694e-01
4.92837608e-01 -4.93151963e-01 6.89518273e-01 2.00652778e-01
8.88555590e-03 -1.05679262e+00 -1.04271066e+00 -3.72655720e-01
6.10347465e-02 -3.12268555e-01 5.28234065e-01 6.71905816e-01
-2.01887205e-01 1.63896531e-01 2.31277049e-01 1.02171235e-01
6.33044958e-01 1.98002636e-01 3.19463760e-01 -1.23715949e+00
-1.64232075e-01 -2.24157438e-01 -4.28144485e-01 -1.35031253e-01
-2.54638821e-01 -1.15435028e+00 -6.86926767e-02 -1.59883165e+00
4.47585315e-01 -2.72857934e-01 -6.57953262e-01 2.96933830e-01
-1.18651509e-01 3.87928903e-01 1.48413554e-01 2.96591967e-01
-1.16563357e-01 3.11272610e-02 1.53096032e+00 -6.87508434e-02
-4.02525246e-01 4.75441933e-01 -3.65592502e-02 7.94411421e-01
9.20420706e-01 -8.74088407e-01 -2.20727652e-01 1.85940042e-01
-5.58473110e-01 4.08557534e-01 4.96281117e-01 -1.28069246e+00
6.21325336e-03 1.13446619e-02 1.05161691e+00 -1.08196664e+00
2.65153766e-01 -1.12089896e+00 2.09323466e-01 1.07147455e+00
-3.31686348e-01 -1.85513809e-01 -5.12641855e-02 2.69762516e-01
-3.35046262e-01 -6.40412986e-01 1.11514521e+00 -3.70072871e-01
-1.59032896e-01 5.21731198e-01 -5.76238513e-01 -3.07356805e-01
1.27096939e+00 -3.76192272e-01 5.67774355e-01 1.27817923e-02
-1.09809887e+00 -5.21011949e-02 1.33694023e-01 -6.95452318e-02
6.96112871e-01 -1.26507747e+00 -7.86645293e-01 5.33491075e-01
8.75973031e-02 3.97676468e-01 3.10585856e-01 1.09788716e+00
-6.46523535e-01 2.51711249e-01 -2.12365881e-01 -1.00744891e+00
-1.46619046e+00 4.92345363e-01 7.76320219e-01 -7.12626755e-01
-8.76859367e-01 6.92915201e-01 7.58468956e-02 -1.63278952e-01
-1.15614042e-01 -8.86792123e-01 -5.73848128e-01 2.61985362e-01
4.47926164e-01 7.06296042e-02 3.56302708e-01 -8.21448982e-01
-4.54310268e-01 5.69138408e-01 -4.19276580e-02 -1.89910159e-01
1.48274148e+00 3.75311106e-01 -3.22256163e-02 3.26959312e-01
1.58181906e+00 3.54175493e-02 -8.61127019e-01 -3.09455365e-01
-2.16599450e-01 -2.20709711e-01 1.27563924e-01 -5.69481015e-01
-1.27208209e+00 8.69475603e-01 8.59160662e-01 1.79830626e-01
1.20486629e+00 7.54761174e-02 3.69966298e-01 -6.92589581e-02
-1.36079058e-01 -8.15627337e-01 1.59191787e-02 3.62290800e-01
7.86188245e-01 -1.17780602e+00 1.70264557e-01 -2.02305317e-01
-7.90032923e-01 1.45126545e+00 3.49441528e-01 -4.14563537e-01
9.33687449e-01 3.02637428e-01 1.16061851e-01 -4.57565159e-01
-4.89770412e-01 3.52204517e-02 3.35028499e-01 3.12997550e-01
3.21426570e-01 1.81344688e-01 -1.80298880e-01 6.27168000e-01
-2.42102996e-01 2.25580901e-01 4.26408589e-01 8.74088526e-01
-6.59208715e-01 -6.06684387e-01 -5.45416594e-01 9.19187009e-01
-5.59871078e-01 1.46000400e-01 -9.98110622e-02 1.05401611e+00
2.73914933e-01 4.90802765e-01 2.91002780e-01 -2.99710125e-01
2.26586595e-01 3.29264939e-01 4.14268047e-01 -7.42780924e-01
-1.01532173e+00 2.57099211e-01 -2.73641318e-01 -3.75820965e-01
-1.46819457e-01 -6.24715924e-01 -1.60085094e+00 3.44943047e-01
-4.97876346e-01 -4.82680229e-03 6.70153737e-01 7.40017831e-01
-3.82638574e-01 8.72559011e-01 8.43432844e-01 -6.35281801e-01
-5.52666664e-01 -9.15140748e-01 -5.92015147e-01 3.25773567e-01
5.67223310e-01 -4.54792440e-01 -4.35562730e-01 7.63840526e-02] | [15.21506118774414, -2.143176555633545] |
a2583b87-14e2-4beb-88e3-f2f443a06ff7 | document-image-classification-with-a-specific | 1601.03295 | null | http://arxiv.org/abs/1601.03295v1 | http://arxiv.org/pdf/1601.03295v1.pdf | Document image classification, with a specific view on applications of patent images | The main focus of this paper is document image classification and retrieval,
where we analyze and compare different parameters for the RunLeght Histogram
(RL) and Fisher Vector (FV) based image representations. We do an exhaustive
experimental study using different document image datasets, including the MARG
benchmarks, two datasets built on customer data and the images from the Patent
Image Classification task of the Clef-IP 2011. The aim of the study is to give
guidelines on how to best choose the parameters such that the same features
perform well on different tasks. As an example of such need, we describe the
Image-based Patent Retrieval task's of Clef-IP 2011, where we used the same
image representation to predict the image type and retrieve relevant patents. | ['Gabriela Csurka'] | 2016-01-13 | null | null | null | null | ['document-image-classification'] | ['computer-vision'] | [ 4.06540513e-01 -5.06818652e-01 -4.76018310e-01 -3.44940603e-01
-8.35595489e-01 -9.44112778e-01 9.45658147e-01 1.74048305e-01
-3.16880822e-01 3.87540936e-01 -3.83523107e-01 -6.04088128e-01
-7.40874350e-01 -6.07390821e-01 -5.48153400e-01 -6.63327515e-01
2.24752814e-01 5.61105907e-01 8.04093704e-02 3.30362707e-01
1.08893776e+00 8.43066275e-01 -1.92847085e+00 6.76275969e-01
4.38336372e-01 1.57181048e+00 4.11769271e-01 7.27667749e-01
-3.65960568e-01 5.78432798e-01 -9.89028096e-01 -3.98318440e-01
3.45537722e-01 -7.37962276e-02 -1.03312445e+00 1.07378080e-01
9.01600957e-01 5.86223081e-02 -2.61595726e-01 1.00770974e+00
5.13129354e-01 -3.25913318e-02 1.52340508e+00 -1.17452955e+00
-1.01578248e+00 8.90165865e-02 -4.30201560e-01 6.03989482e-01
3.38822544e-01 -3.16739142e-01 9.64322507e-01 -6.99471354e-01
1.06304777e+00 1.26041889e+00 1.75508276e-01 -1.23482615e-01
-1.00945258e+00 -3.54667395e-01 -8.07874128e-02 6.07651055e-01
-1.35925901e+00 3.72644700e-02 5.37453890e-01 -7.88567483e-01
9.07992959e-01 3.10980946e-01 2.81730384e-01 1.00043154e+00
4.34666127e-01 9.11251247e-01 1.34698546e+00 -7.56213367e-01
-2.42831856e-02 2.88916379e-01 6.78579926e-01 5.44231176e-01
2.61465430e-01 6.92918003e-02 -2.37864077e-01 -1.73113346e-01
7.72068560e-01 2.17048645e-01 -2.81332225e-01 -4.35323268e-01
-1.04298139e+00 1.05488324e+00 3.20188075e-01 9.79989231e-01
-3.60033482e-01 -2.23614667e-02 2.94160396e-01 6.06228769e-01
3.68114203e-01 5.38941860e-01 -4.85485673e-01 1.53549910e-01
-1.02772486e+00 2.67344505e-01 7.14159071e-01 9.51052368e-01
5.85226774e-01 -4.27683771e-01 -5.79369128e-01 1.00922811e+00
2.17507482e-01 5.46088636e-01 7.62175441e-01 -8.66075933e-01
3.51994693e-01 4.48285282e-01 -3.21712159e-02 -9.87586260e-01
1.57259405e-02 -4.53323066e-01 -3.13852191e-01 7.05011785e-02
3.23129326e-01 4.67901558e-01 -1.57129848e+00 7.02023029e-01
-6.06343925e-01 -3.37407976e-01 4.50616628e-02 4.86959428e-01
1.23178589e+00 9.23257470e-01 -9.97438654e-02 1.84703004e-02
1.57178557e+00 -9.30856526e-01 -6.24443889e-01 2.07432702e-01
7.35942662e-01 -1.19424403e+00 9.12370384e-01 6.41864896e-01
-8.10785472e-01 -8.94337595e-01 -1.04494429e+00 1.41198069e-01
-1.14194572e+00 6.20670497e-01 6.86697066e-01 7.20374048e-01
-7.74622202e-01 5.36319196e-01 7.28519931e-02 -4.42311347e-01
4.86215115e-01 2.05707073e-01 -4.64692801e-01 -4.09112394e-01
-9.15791094e-01 8.42864096e-01 4.25150752e-01 -2.06754386e-01
-6.47255480e-01 -6.01149082e-01 -4.62366939e-01 2.37845108e-01
-6.83521032e-02 -3.35522562e-01 9.83239532e-01 -9.20013607e-01
-9.99850273e-01 1.32791173e+00 2.40911618e-01 -4.18752640e-01
1.75172776e-01 2.16859847e-01 -4.45072681e-01 4.13697422e-01
1.16350159e-01 8.86034846e-01 1.03854108e+00 -1.46253800e+00
-7.45480478e-01 -5.46378255e-01 -4.31454517e-02 -2.66171634e-01
-9.69590843e-02 -3.98940742e-02 -8.56482565e-01 -7.02100337e-01
-3.50457728e-01 -1.00591564e+00 2.78914034e-01 -7.33978331e-01
-3.41588140e-01 -5.34480095e-01 8.96568179e-01 -4.26528752e-01
1.10451758e+00 -2.20535803e+00 -4.91856374e-02 6.26108050e-01
-3.22711349e-01 3.22617888e-01 -5.98731101e-01 2.83186197e-01
-4.88389343e-01 3.21501374e-01 4.69408743e-02 1.90782011e-01
-5.19604348e-02 1.53732393e-02 -5.53754151e-01 2.54119873e-01
-6.27583358e-03 1.06847191e+00 -2.02655375e-01 -4.16620374e-01
3.14048737e-01 3.66039068e-01 -2.11223997e-02 -1.33539617e-01
-2.65700594e-02 2.91644987e-02 -8.36906374e-01 8.48964572e-01
6.36517465e-01 -2.13343143e-01 -2.13766381e-01 -5.33131182e-01
4.51120883e-02 -2.50560492e-01 -7.58596241e-01 1.46173263e+00
-2.72463381e-01 8.91718566e-01 -5.07362068e-01 -1.20591736e+00
8.89639497e-01 7.90792108e-02 5.04107058e-01 -1.17258561e+00
1.09921813e-01 2.98910052e-01 -2.65552461e-01 -5.03810346e-01
4.45516258e-01 4.07364100e-01 1.66906808e-02 1.04959853e-01
1.79973319e-01 -4.74885046e-01 6.02213681e-01 -5.84555119e-02
9.08456743e-01 -1.01946793e-01 6.59255162e-02 -2.83428103e-01
9.14800286e-01 2.50510663e-01 -3.03335875e-01 1.03563547e+00
5.54662310e-02 7.17442989e-01 6.31129742e-01 -5.13981462e-01
-8.27489018e-01 -7.75197268e-01 -6.17249727e-01 7.20850527e-01
-1.11246668e-01 -1.93472043e-01 -6.50212109e-01 -9.64528680e-01
3.73517007e-01 5.31686962e-01 -6.49624825e-01 2.45285437e-01
-1.58914030e-01 -6.18714154e-01 2.60053903e-01 1.40969843e-01
2.83694416e-01 -1.16628480e+00 -3.15086961e-01 -2.95293033e-01
3.00528169e-01 -8.18282425e-01 -3.47583711e-01 4.20290440e-01
-7.28187382e-01 -1.24008918e+00 -1.32976294e+00 -1.08873928e+00
6.93111241e-01 3.80833626e-01 1.21411872e+00 2.65717674e-02
-7.43726015e-01 8.86198938e-01 -3.68499607e-01 -5.61796367e-01
-1.51030809e-01 2.92163134e-01 -7.53797710e-01 -2.81053841e-01
3.98329675e-01 1.78181812e-01 -6.21272981e-01 3.64665776e-01
-1.30767703e+00 -8.03569973e-01 8.58180642e-01 8.60360205e-01
9.27358627e-01 4.83515233e-01 2.66660094e-01 -1.02959061e+00
9.25876856e-01 -1.20512374e-01 -8.43518734e-01 6.61926806e-01
-1.09515870e+00 3.88673842e-02 6.87047839e-02 -2.18524367e-01
-6.72989964e-01 7.37498254e-02 2.39034072e-01 -5.73361993e-01
-2.51004249e-01 4.60466921e-01 7.02755749e-02 -5.02319157e-01
4.15511489e-01 1.49485558e-01 -2.17367128e-01 -5.12076437e-01
4.40584928e-01 1.02934527e+00 2.87884206e-01 -4.60750014e-01
3.42491597e-01 1.04914375e-01 2.10356846e-01 -8.56086612e-01
-6.23169065e-01 -8.87777805e-01 -4.99155581e-01 -1.10448837e-01
9.21596766e-01 -6.20825529e-01 -6.72570705e-01 6.91348761e-02
-1.25502872e+00 1.87513858e-01 -2.08722532e-01 3.98701727e-01
-6.55285716e-01 5.10770082e-01 -4.25976276e-01 -5.96963108e-01
-2.99579412e-01 -1.66928279e+00 1.45098996e+00 2.50979941e-02
2.01981604e-01 -7.88885295e-01 7.30638802e-02 7.20543087e-01
2.85548300e-01 -9.58439782e-02 1.46080041e+00 -8.18753242e-01
-5.73265016e-01 -5.32144368e-01 -4.84674901e-01 6.20861053e-01
-1.50263324e-01 2.57402331e-01 -1.00319672e+00 -1.81314036e-01
-1.59854591e-01 -1.76813573e-01 1.51015604e+00 7.85316586e-01
1.63138652e+00 2.12874383e-01 -4.08754379e-01 4.04131562e-01
1.70469868e+00 7.22056806e-01 1.06576943e+00 5.83738565e-01
3.83742332e-01 7.17225134e-01 8.49138916e-01 -8.84855762e-02
-1.49196357e-01 7.41286337e-01 2.62130678e-01 1.70811355e-01
-6.98750913e-02 1.61639869e-01 2.92834733e-02 4.19934124e-01
-1.93203717e-01 -6.54912531e-01 -7.36005783e-01 1.72225371e-01
-1.67944002e+00 -9.00262833e-01 -1.46019921e-01 2.18968749e+00
3.36212218e-02 -5.42179123e-02 6.36341125e-02 2.37060219e-01
6.58733904e-01 3.10249329e-01 -2.85391044e-02 -5.48632205e-01
-2.10437283e-01 5.39842546e-01 8.90303969e-01 1.82711467e-01
-1.40636313e+00 7.73820579e-01 7.33146000e+00 1.47195172e+00
-1.08274364e+00 -1.92254767e-01 8.81328046e-01 2.85937041e-01
-1.68185309e-01 -1.32612661e-01 -6.65649712e-01 4.11996067e-01
1.11090660e+00 2.03226909e-01 2.36367092e-01 8.65976334e-01
-4.47966605e-01 -2.39359632e-01 -8.08901489e-01 1.29483867e+00
3.27818274e-01 -1.51146972e+00 6.38792753e-01 4.78301585e-01
4.57566023e-01 -1.66422233e-01 4.18247372e-01 2.98983246e-01
-3.68528634e-01 -1.28021479e+00 4.74403888e-01 6.75327063e-01
6.01446748e-01 -8.05471718e-01 8.45813692e-01 -2.22043455e-01
-8.32919419e-01 -2.42744669e-01 -5.31642973e-01 7.20262527e-01
-4.70421702e-01 5.28068006e-01 -4.12536263e-01 8.71283054e-01
7.61158347e-01 8.53771746e-01 -8.63122582e-01 1.07460904e+00
1.83694988e-01 1.58092424e-01 3.39938670e-01 -1.11881793e-02
4.83123183e-01 -3.85930747e-01 2.52658397e-01 1.38104475e+00
6.81273580e-01 -5.92933118e-01 -5.92111833e-02 5.79357505e-01
-6.37683645e-02 3.65460902e-01 -8.79588127e-01 -6.03212953e-01
1.02259226e-01 1.43124306e+00 -9.73595381e-01 -3.60324144e-01
-4.85112697e-01 9.06199336e-01 -3.15956533e-01 5.08164287e-01
-4.21571285e-01 -6.49514496e-01 -1.27583342e-02 1.45503618e-02
8.92721474e-01 1.50814191e-01 8.60186368e-02 -7.82897055e-01
-2.76590198e-01 -9.00010765e-01 4.38633054e-01 -1.13066339e+00
-1.33440447e+00 5.92566669e-01 1.59592465e-01 -1.23897946e+00
-1.09335072e-01 -1.39891851e+00 -3.29668462e-01 7.52241731e-01
-1.82990158e+00 -9.88226712e-01 -3.49012688e-02 4.50891256e-01
4.01968837e-01 -7.42297590e-01 7.70230889e-01 4.35851276e-01
-4.09180641e-01 2.39305779e-01 3.92696023e-01 -1.92146003e-01
7.64205813e-01 -1.40450943e+00 -1.53274581e-01 2.55506355e-02
5.61698735e-01 5.49537838e-01 3.30119044e-01 -3.66979718e-01
-1.35321128e+00 -7.00664699e-01 5.17318368e-01 -5.72680235e-01
5.60422897e-01 2.84115851e-01 -4.77521271e-01 3.12860310e-01
4.17493999e-01 -1.95542023e-01 6.77701831e-01 -3.58295143e-01
-3.12536985e-01 -6.60687387e-02 -1.13739145e+00 -3.32396813e-02
3.70821327e-01 -5.59024036e-01 -5.19396663e-01 7.75090516e-01
1.61309689e-01 6.35917708e-02 -1.23990977e+00 4.13252860e-01
7.10691810e-01 -8.39711308e-01 1.44725537e+00 -5.20718277e-01
4.24126476e-01 -2.47073412e-01 -3.77801210e-01 -1.14909875e+00
-3.74906272e-01 -5.44509999e-02 4.08780098e-01 1.18238461e+00
4.05248344e-01 -4.86692548e-01 5.85217774e-01 -2.40349080e-02
2.57998347e-01 -7.76060998e-01 -7.33984113e-01 -1.06244218e+00
1.26764685e-01 -2.32701287e-01 3.84369493e-01 6.23481929e-01
-6.62644565e-01 2.85140127e-01 1.69370592e-01 -3.53910476e-01
3.14184189e-01 5.70240140e-01 5.77913940e-01 -1.52204633e+00
3.94232757e-02 -6.97494328e-01 -8.81849170e-01 -6.39554620e-01
3.45906436e-01 -1.17526352e+00 -3.14167589e-01 -1.54350042e+00
5.68459153e-01 -5.28558977e-02 -6.00789309e-01 1.01923905e-01
1.48838326e-01 1.65976971e-01 4.23578560e-01 6.59402132e-01
-3.94334048e-01 -1.12911329e-01 1.30898845e+00 -7.84752190e-01
2.70600080e-01 6.49617985e-02 -5.82201123e-01 3.67187232e-01
4.82135713e-01 -3.62283498e-01 -5.50391257e-01 -1.10068664e-01
8.63397419e-02 -1.94921881e-01 2.62374520e-01 -5.83707571e-01
-6.83583468e-02 3.21982354e-01 8.07863653e-01 -1.07003951e+00
1.87726513e-01 -1.06567383e+00 -5.50570302e-02 5.36942422e-01
-4.72026050e-01 2.44046301e-01 -2.40950892e-03 6.12623811e-01
-7.18687594e-01 -8.98174822e-01 4.86510575e-01 -2.13357210e-01
-7.95533597e-01 2.22185701e-01 -2.99183339e-01 -4.10310656e-01
1.09049904e+00 -2.42123619e-01 -6.01992130e-01 -1.93241015e-01
-3.58995825e-01 -6.26610890e-02 3.91254336e-01 5.14773667e-01
4.57057148e-01 -1.23313880e+00 -3.92613113e-01 1.31987274e-01
6.21115267e-01 -7.92198658e-01 6.25187084e-02 4.89868671e-01
-6.48849010e-01 1.30006063e+00 -2.23657683e-01 -6.77641928e-01
-1.47959113e+00 8.86035144e-01 3.34333256e-02 -5.65803766e-01
-1.51620403e-01 4.20108527e-01 2.39545181e-01 -1.98067039e-01
3.07906896e-01 -3.41026574e-01 -7.95280755e-01 4.77849722e-01
3.89782041e-01 1.83662623e-01 4.21939641e-01 -6.12925231e-01
-3.22169125e-01 1.03396380e+00 -3.55718225e-01 -9.67161544e-03
1.31493282e+00 3.36991727e-01 -1.65978521e-01 1.35285616e-01
1.67360163e+00 -2.77154207e-01 -3.42181712e-01 1.90535590e-01
3.12009692e-01 -6.28732800e-01 5.67327797e-01 -8.34842682e-01
-1.25958979e+00 7.97242403e-01 1.29417717e+00 6.72662497e-01
1.06988502e+00 1.20576479e-01 3.74755770e-01 6.31322742e-01
2.17970535e-01 -1.22603762e+00 -1.07392706e-02 3.79197478e-01
1.09277058e+00 -1.06655455e+00 2.94058472e-01 -4.01570737e-01
-6.24206841e-01 1.47517514e+00 -2.17730209e-01 -1.10225081e-01
6.83651984e-01 -4.36794758e-02 -2.23923549e-01 -6.46005213e-01
-2.25184426e-01 -4.45495754e-01 9.03646111e-01 6.15204394e-01
6.41788840e-01 -7.95345083e-02 -4.35915381e-01 1.61052212e-01
1.92031711e-01 7.18001947e-02 8.37055072e-02 1.12722576e+00
-4.94041383e-01 -1.53917098e+00 -5.21628559e-01 8.95958543e-01
-7.80009687e-01 -4.11918201e-02 -8.55209351e-01 9.40335214e-01
-1.00912964e-02 8.14487159e-01 2.71798894e-02 -1.59574896e-01
5.65681636e-01 1.24715500e-01 9.81327951e-01 -3.27419370e-01
-4.83856291e-01 1.20191827e-01 7.44429836e-03 -4.71559554e-01
-8.00310612e-01 -4.11837757e-01 -6.75930798e-01 1.34081274e-01
-4.19709533e-01 2.42899075e-01 1.19278634e+00 7.17849731e-01
4.21837151e-01 6.56755626e-01 6.34150922e-01 -9.03392136e-01
-3.54815304e-01 -8.38556409e-01 -9.99675930e-01 5.33582985e-01
-1.53246358e-01 -7.75128901e-01 -4.73926604e-01 6.39889762e-02] | [10.881843566894531, 0.5427860021591187] |
b856125f-c2ce-4254-b72f-e8f4bf7b1467 | speak2label-using-domain-knowledge-for | 2004.05973 | null | https://arxiv.org/abs/2004.05973v4 | https://arxiv.org/pdf/2004.05973v4.pdf | Speak2Label: Using Domain Knowledge for Creating a Large Scale Driver Gaze Zone Estimation Dataset | Labelling of human behavior analysis data is a complex and time consuming task. In this paper, a fully automatic technique for labelling an image based gaze behavior dataset for driver gaze zone estimation is proposed. Domain knowledge is added to the data recording paradigm and later labels are generated in an automatic manner using Speech To Text conversion (STT). In order to remove the noise in the STT process due to different illumination and ethnicity of subjects in our data, the speech frequency and energy are analysed. The resultant Driver Gaze in the Wild (DGW) dataset contains 586 recordings, captured during different times of the day including evenings. The large scale dataset contains 338 subjects with an age range of 18-63 years. As the data is recorded in different lighting conditions, an illumination robust layer is proposed in the Convolutional Neural Network (CNN). The extensive experiments show the variance in the dataset resembling real-world conditions and the effectiveness of the proposed CNN pipeline. The proposed network is also fine-tuned for the eye gaze prediction task, which shows the discriminativeness of the representation learnt by our network on the proposed DGW dataset. Project Page: https://sites.google.com/view/drivergazeprediction/home | ['Sarthak Gupta', 'Shreya Ghosh', 'Abhinav Dhall', 'Nicu Sebe', 'Garima Sharma'] | 2020-04-13 | null | null | null | null | ['eye-tracking'] | ['computer-vision'] | [ 2.06142008e-01 2.73655832e-01 -4.07129079e-02 -7.32391655e-01
-3.26155633e-01 -2.96855986e-01 4.38257813e-01 -3.08549374e-01
-4.50457186e-01 3.00688654e-01 3.51734340e-01 -2.60883242e-01
-4.44843136e-02 -1.39577150e-01 -4.02490735e-01 -8.16624880e-01
3.77440304e-01 -3.27000439e-01 1.60096094e-01 -3.25343400e-01
4.43511337e-01 3.70360076e-01 -2.49323750e+00 7.81672224e-02
4.94046450e-01 1.16022050e+00 1.03563458e-01 8.62654626e-01
1.62164360e-01 8.93547595e-01 -6.46564007e-01 -1.69937134e-01
3.16949636e-01 -4.11752343e-01 -5.34703016e-01 1.33037239e-01
7.31521666e-01 -6.49087429e-02 -1.99916944e-01 9.41139102e-01
7.53379345e-01 3.14854026e-01 4.23432708e-01 -1.69621015e+00
-3.93374413e-01 -2.13205963e-01 -7.46131480e-01 4.63670522e-01
2.65896261e-01 5.23315430e-01 4.11482394e-01 -5.62936902e-01
3.85009766e-01 1.04881001e+00 4.81839836e-01 7.97852337e-01
-6.08643532e-01 -1.13200116e+00 -1.42693430e-01 7.21964419e-01
-1.39642525e+00 -7.80701041e-01 1.05561328e+00 -5.84768891e-01
7.90427208e-01 2.62424320e-01 4.43896741e-01 1.10460019e+00
2.54206210e-01 4.87695932e-01 1.43581283e+00 -5.91651022e-01
-1.70605838e-01 4.76425052e-01 4.52171832e-01 6.30027890e-01
-4.68458831e-02 3.38801801e-01 -8.71015906e-01 3.15004170e-01
-1.82808906e-01 3.16546448e-02 -2.08012044e-01 -2.04871278e-02
-8.50468099e-01 7.88383126e-01 5.26893973e-01 -1.10290781e-01
-2.07795724e-01 -2.64396016e-02 5.47824204e-01 6.88271299e-02
6.73955858e-01 1.74580086e-02 -2.47572020e-01 -3.14434618e-01
-8.26603651e-01 1.29147962e-01 5.36535919e-01 9.82311964e-01
8.58848751e-01 -2.15334550e-01 -3.16621006e-01 8.68128777e-01
6.31617963e-01 5.56428373e-01 7.11758971e-01 -7.90389299e-01
1.74914703e-01 7.28219271e-01 -4.60100099e-02 -9.17208374e-01
-6.81113839e-01 -1.61435813e-01 -3.67596179e-01 5.04665375e-01
3.13980281e-01 -3.17197949e-01 -1.06552172e+00 1.61568952e+00
5.00726879e-01 1.20032437e-01 -1.21189557e-01 9.75552261e-01
1.05865526e+00 5.31854093e-01 2.05825686e-01 -2.03055859e-01
1.80445540e+00 -9.15051818e-01 -1.20913684e+00 -3.90878260e-01
6.06034517e-01 -7.08408475e-01 9.57876801e-01 2.72352785e-01
-6.71857178e-01 -9.00543213e-01 -1.05014741e+00 -4.33409512e-01
-5.70936322e-01 3.65963131e-01 3.22773784e-01 1.03178716e+00
-1.14522350e+00 -3.09733510e-01 -4.15708125e-01 -6.35385633e-01
5.71326911e-01 5.22308707e-01 -2.66280770e-01 2.48559639e-01
-1.00442755e+00 1.02086151e+00 2.05249906e-01 4.83071148e-01
-6.77065432e-01 -2.95091122e-01 -1.04016614e+00 -2.08877996e-01
2.83069491e-01 -5.55265583e-02 1.36646664e+00 -1.12634730e+00
-1.53543210e+00 1.24734044e+00 -8.63510668e-01 -5.60381711e-01
2.16691136e-01 2.09833514e-02 -8.60846639e-01 -9.50821489e-02
-1.92933619e-01 8.78422678e-01 1.21753275e+00 -7.75499344e-01
-1.03902698e+00 -4.09692526e-01 -5.93666472e-02 8.26389194e-02
-1.36909947e-01 3.40761364e-01 -3.24534118e-01 -4.50662039e-02
-5.48672438e-01 -1.22904897e+00 5.74294746e-01 -2.65452504e-01
-2.00790390e-01 -6.77244484e-01 1.04925787e+00 -7.28629827e-01
1.33766818e+00 -2.62240863e+00 -3.99526626e-01 9.57010239e-02
5.33278048e-01 3.01122576e-01 1.04551785e-01 1.15044221e-01
-2.50333577e-01 -2.81590253e-01 5.02727292e-02 -4.85371947e-01
2.02649593e-01 9.63267405e-03 6.69448301e-02 6.72660351e-01
2.49615848e-01 8.47747207e-01 -4.83273715e-01 -4.83987182e-01
4.16516334e-01 6.03872895e-01 -2.55998373e-01 3.56709003e-01
2.93392569e-01 4.69713867e-01 -2.08840240e-02 3.92335027e-01
7.61848271e-01 2.45964333e-01 -5.22224784e-01 -2.48922333e-01
-4.39585984e-01 2.89588004e-01 -7.03287721e-01 1.49976027e+00
-5.35019338e-01 1.52451932e+00 -1.80026233e-01 -5.79390824e-01
9.78666008e-01 2.27892533e-01 1.49167150e-01 -1.06717420e+00
7.00763583e-01 5.42860553e-02 3.86723131e-01 -1.16474986e+00
5.53181231e-01 2.92471796e-01 1.02506056e-01 3.78473967e-01
2.32508466e-01 3.11244547e-01 1.77741379e-01 -2.28919461e-01
6.80393696e-01 7.69163221e-02 6.91613704e-02 -3.58373791e-01
8.04371119e-01 -2.50831023e-02 3.84923786e-01 7.23491758e-02
-8.71873498e-01 3.45224530e-01 5.04557908e-01 -3.72735590e-01
-8.23928416e-01 -1.85082734e-01 -3.16994160e-01 1.34051394e+00
1.27426639e-01 -2.06400245e-01 -1.18945980e+00 -5.09578407e-01
-3.97535324e-01 7.82169342e-01 -1.11107934e+00 -5.35523713e-01
-4.48581129e-01 -5.87231815e-01 4.36586201e-01 1.84465989e-01
8.23857605e-01 -1.24363661e+00 -1.16029596e+00 -5.48313975e-01
-1.52106658e-01 -1.03646767e+00 -5.71253300e-01 -3.94623913e-02
-3.48537415e-02 -1.19490767e+00 -4.03922588e-01 -7.51659334e-01
4.74007577e-01 2.66447067e-01 7.65374124e-01 -8.06804448e-02
-4.14959490e-01 1.99558944e-01 -1.53564602e-01 -1.18439364e+00
-2.96315014e-01 9.78629813e-02 -1.96337819e-01 5.72491407e-01
1.44353855e+00 1.17196567e-01 -9.70364034e-01 2.29677215e-01
-4.67337340e-01 -1.20232955e-01 3.94779533e-01 5.27498543e-01
1.87744021e-01 1.61257777e-02 3.42696995e-01 -6.34037316e-01
7.91963518e-01 -4.76775944e-01 -6.99954927e-01 -5.25687970e-02
-6.36557341e-01 -1.53979287e-01 -1.51675597e-01 -2.83898771e-01
-1.28383732e+00 1.54321507e-01 -9.22409073e-02 -3.39452952e-01
-8.51339757e-01 -1.37406886e-01 -1.80390608e-02 -7.55442455e-02
8.66425753e-01 -1.42941520e-01 3.18061441e-01 -2.39566639e-01
4.65897620e-02 1.37251353e+00 5.32019317e-01 3.40082884e-01
4.70564276e-01 4.54531908e-01 -2.83244792e-02 -7.14335144e-01
-1.04544890e+00 -7.10473359e-01 -6.28289104e-01 -6.10026419e-01
1.32269776e+00 -8.86173427e-01 -1.30020988e+00 7.53523767e-01
-8.57039750e-01 -1.47232130e-01 -4.83964011e-02 5.56522608e-01
-2.60513633e-01 -3.45377594e-01 1.87716022e-01 -1.24619651e+00
-3.11134607e-01 -1.36417878e+00 1.20432305e+00 5.30448437e-01
-2.17736587e-01 -7.52181590e-01 -4.63379771e-02 7.49775708e-01
4.32791471e-01 3.40830714e-01 3.15672904e-01 -5.33338308e-01
-8.94545093e-02 -2.41436779e-01 -3.27727795e-01 4.58055556e-01
1.65467024e-01 9.94497724e-03 -1.81574237e+00 -1.06493093e-01
2.40467981e-01 -2.05568030e-01 7.11984217e-01 4.85622764e-01
1.11762846e+00 1.11302689e-01 -2.39388764e-01 6.45555735e-01
1.13333642e+00 3.27401400e-01 7.34693944e-01 2.93008596e-01
8.80718648e-01 1.16957033e+00 5.48520029e-01 1.23601981e-01
5.32365620e-01 5.49025834e-01 5.70829988e-01 -2.75362015e-01
-4.68232423e-01 1.16629109e-01 3.51702869e-01 2.19247296e-01
-3.00514419e-02 -4.39458221e-01 -1.10905313e+00 6.07646883e-01
-1.58150673e+00 -9.71973836e-01 -5.30920446e-01 1.99709821e+00
5.79663277e-01 -6.83401376e-02 4.29884702e-01 3.15083623e-01
7.33432293e-01 1.14767710e-02 -5.57144165e-01 -8.82547140e-01
2.78299630e-01 7.00367242e-02 8.44563067e-01 4.24446195e-01
-1.06897068e+00 4.78616357e-01 5.40498638e+00 4.77616698e-01
-1.46826470e+00 3.82731378e-01 4.86636072e-01 -3.50175500e-01
3.51971716e-01 -4.55129176e-01 -1.03598595e+00 7.82110274e-01
1.63542497e+00 9.86956619e-03 3.39708090e-01 5.91720521e-01
6.15538359e-01 -4.47688282e-01 -8.25995505e-01 1.42215586e+00
6.31565630e-01 -6.59569204e-01 -8.96924376e-01 3.54223996e-01
3.34653556e-01 3.03336799e-01 2.94079959e-01 1.75208494e-01
-2.66845614e-01 -1.06701910e+00 8.26172590e-01 6.67570055e-01
8.97037029e-01 -9.02545750e-01 8.30973864e-01 3.03046107e-01
-8.32913518e-01 -5.61217010e-01 -5.92948310e-02 -1.91540606e-02
1.04579898e-02 1.49209246e-01 -1.07536125e+00 -6.69826642e-02
1.16093338e+00 7.49115169e-01 -1.11665022e+00 8.93156648e-01
-1.01504281e-01 8.24059486e-01 -2.39412040e-01 -7.89149329e-02
8.29823241e-02 3.18007283e-02 3.86171520e-01 1.16003561e+00
1.15157776e-01 -1.83186755e-01 -5.43941140e-01 8.24662507e-01
-7.31361508e-02 -2.38439471e-01 -7.03554273e-01 3.08855325e-01
1.85491219e-01 1.35262656e+00 -4.16781366e-01 4.97592576e-02
-3.95586669e-01 4.97768074e-01 2.46829558e-02 4.45118427e-01
-1.01956797e+00 -5.97204745e-01 7.63930917e-01 2.86566734e-01
1.44320801e-01 1.68336764e-01 -1.59998670e-01 -5.00063479e-01
1.40223309e-01 -5.56181967e-01 1.87493861e-01 -1.02924275e+00
-8.48323703e-01 7.06872642e-01 2.18225107e-01 -1.41321254e+00
-3.02211910e-01 -6.66664720e-01 -3.83695304e-01 1.27242339e+00
-1.85241210e+00 -1.00583994e+00 -9.36206639e-01 9.01151717e-01
7.21234560e-01 -4.41890150e-01 5.01059353e-01 4.18164253e-01
-8.64683509e-01 6.77096844e-01 -3.82462114e-01 -4.22729813e-02
8.43342781e-01 -1.05400658e+00 1.66989043e-01 9.68182027e-01
-3.58419091e-01 4.39341664e-01 6.67515993e-01 -1.52284920e-01
-1.01977873e+00 -1.09073555e+00 1.05808616e+00 -6.83029771e-01
3.34040821e-01 -7.27406383e-01 -7.41120338e-01 5.24901748e-01
8.60961795e-01 2.24924795e-02 7.69569159e-01 -1.91221491e-01
-3.14636640e-02 -3.28374296e-01 -1.18035352e+00 1.99348614e-01
7.16291547e-01 -7.95582235e-01 -4.95407254e-01 2.05628678e-01
3.41318876e-01 -5.96936703e-01 -3.79963934e-01 1.64715439e-01
4.81593192e-01 -1.04699349e+00 3.73022914e-01 -1.99826509e-01
3.74988243e-02 -5.67233980e-01 3.67785275e-01 -1.02137840e+00
-6.44188002e-02 -7.51343250e-01 -1.32196874e-03 1.30012095e+00
2.77862519e-01 -9.05900657e-01 2.89107054e-01 7.81644642e-01
-1.89886987e-01 -3.68753135e-01 -6.94474936e-01 -1.79545298e-01
-3.91157717e-01 -5.37790537e-01 7.74976015e-01 4.73815560e-01
-2.89189905e-01 4.30143386e-01 -2.70301223e-01 8.51473212e-02
3.98188353e-01 -4.02730197e-01 7.77505636e-01 -1.44522917e+00
4.25121009e-01 -1.64468035e-01 -7.75027454e-01 -5.80231011e-01
3.40996236e-01 -4.33254957e-01 2.84223050e-01 -1.06247425e+00
-1.87219054e-01 9.28989276e-02 -3.35300922e-01 3.97057176e-01
-5.97866811e-02 5.66253424e-01 -8.50412026e-02 1.47332981e-01
-3.21280211e-01 1.24697603e-01 1.07569218e+00 -1.22343346e-01
-1.63258016e-01 1.21216074e-01 -6.95696592e-01 5.11593461e-01
9.51724112e-01 -3.87785375e-01 -7.98729241e-01 -3.14905524e-01
2.62878895e-01 -5.63065350e-01 5.57728231e-01 -9.20046389e-01
3.37000638e-01 2.52353042e-01 1.17775671e-01 -6.16617620e-01
3.84744287e-01 -1.00626469e+00 -7.09290802e-02 8.25871006e-02
-5.05687654e-01 3.80110480e-02 4.36641902e-01 4.30392385e-01
-2.58977354e-01 -1.79937378e-01 8.81781578e-01 4.37821060e-01
-8.97768617e-01 8.75318423e-02 -2.01121613e-01 -1.81239009e-01
1.21463096e+00 -5.97403646e-01 -6.18282080e-01 -1.50752306e-01
-4.32296783e-01 1.22623205e-01 1.84682086e-01 8.59544814e-01
1.91364586e-01 -1.11665606e+00 -4.93518263e-01 6.46120310e-01
6.89498723e-01 -1.43927827e-01 3.50336701e-01 1.26305616e+00
-3.65034223e-01 4.67851996e-01 -4.00773108e-01 -8.24244380e-01
-1.50963533e+00 5.93577325e-01 6.35961056e-01 6.38827860e-01
-4.05491441e-01 8.86389077e-01 2.40551099e-01 7.40457997e-02
3.16748977e-01 -4.75452960e-01 -9.88955319e-01 2.30718985e-01
7.67658889e-01 4.84170318e-01 4.78393078e-01 -1.43249571e+00
-3.14261705e-01 5.62404752e-01 -3.87271005e-03 1.58117980e-01
1.09836805e+00 -7.00458348e-01 5.51219913e-04 5.14111936e-01
1.52353525e+00 -1.65894344e-01 -1.20262766e+00 1.01636298e-01
-1.74393862e-01 -4.49811757e-01 4.42758769e-01 -5.78065336e-01
-1.22948289e+00 1.04813612e+00 1.48915124e+00 3.79687816e-01
1.45962226e+00 -9.30504650e-02 6.15480959e-01 8.18526521e-02
-2.61557281e-01 -1.22359288e+00 -6.09684348e-01 1.83402002e-01
7.62123227e-01 -1.63925540e+00 -4.31537688e-01 -1.62615925e-01
-9.29875553e-01 8.61846447e-01 5.44814825e-01 1.03391610e-01
9.85439301e-01 -1.23074688e-01 5.51214635e-01 -7.26486444e-01
-6.43327832e-01 -5.72239816e-01 6.47586405e-01 8.10385883e-01
4.50275779e-01 -1.27976760e-01 -8.93601403e-02 -4.07101661e-02
-7.03882813e-01 3.08307018e-02 4.67196614e-01 6.99010134e-01
-1.49901822e-01 -5.97773731e-01 -3.44109386e-01 3.76123011e-01
-5.33434570e-01 3.08574766e-01 -3.77751112e-01 5.42234123e-01
6.77978694e-01 1.50792170e+00 2.33729139e-01 -3.18443894e-01
5.14422476e-01 4.70700055e-01 -5.91658764e-02 -3.67114931e-01
-6.78886056e-01 -6.21572994e-02 3.12362816e-02 -7.38331437e-01
-9.37715590e-01 -6.45328283e-01 -9.39285278e-01 -2.09293723e-01
-2.97482193e-01 -2.82494754e-01 1.00995123e+00 8.52478504e-01
5.43838561e-01 8.59734058e-01 7.54877448e-01 -8.86570275e-01
2.95536488e-01 -1.30922711e+00 -4.66294855e-01 6.68507159e-01
9.67772603e-01 -1.00964737e+00 -5.30218482e-01 6.40763760e-01] | [14.01655101776123, 0.12011481076478958] |
842d8941-538f-4671-b0b1-9f4e911cfe71 | a-multi-media-approach-to-cross-lingual | null | null | https://aclanthology.org/P16-1006 | https://aclanthology.org/P16-1006.pdf | A Multi-media Approach to Cross-lingual Entity Knowledge Transfer | null | ['Shih-Fu Chang', 'Heng Ji', 'Nima Pourdamghani', 'Kevin Knight', 'Xiaoman Pan', 'Di Lu'] | 2016-08-01 | null | null | null | acl-2016-8 | ['cross-lingual-entity-linking'] | ['natural-language-processing'] | [-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01
-8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01
-2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00
-3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01
-9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01
7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01
7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00
-1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01
3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01
9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01
5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00
-5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01
1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00
7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01
-1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02
-1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01
1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00
1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01
3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01
8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01
9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01
-9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01
-1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01
-2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01
-8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00
1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01
8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00
-6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01
-6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01
3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01
4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02
4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01
1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01
9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01
-1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01
-7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00
-1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02
1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01
-1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01
-1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01
-4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01
-1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01
6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01
-1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00
-7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01
-1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01
3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01
-1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01
1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01
3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00
-2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01
2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01
-1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01
4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01
2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00
1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01
-3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00
7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01
3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01
-2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01
7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01
-2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01
6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00
7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00
-3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01
-6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00
3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01
1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01
-5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01
-5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01
-2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01
-1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01
5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01
-2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01
-4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01
5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01
-6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00
1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01
8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02
-6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01
-1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01
1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01
8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03
3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01
8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01
-1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01
6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01
-9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01
-1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00
-5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02
-5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01
1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01
-4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01
1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01
6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01
5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01
1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01
1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01
1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01
1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00
-8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01
-1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01
6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01
4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00
-1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00
-7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00
2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01
-4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02
1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01
3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01
9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01
7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01
6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01
1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01
-1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00
3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01
9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01
-4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01
4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00
2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01
5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01
2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01
-2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01
-6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01
-1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01
-5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01
-1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00
-6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01
-1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01
1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02
1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01
5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01
-4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01
-1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01
6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01
7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01
1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01
1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01
8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01
8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00
-4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01
-5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02
5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01
-2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00
-3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01
6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01
5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00
3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01
-8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00
-8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01
3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01
-3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05
-1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01
8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02
6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01
5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00
1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01
2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02
-1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01
-4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02
9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01
-8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02
7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01
-3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01
-1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01
-1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01
-2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01
1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03
6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01
-4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01
9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01
8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01
1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01
-1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01
8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01
5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01
9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01
-4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01
1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01
-9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00
5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01
-7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00
-6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01
-5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01
-5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01
4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01
-9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00
-8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00
-5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01
-2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00
-3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02
2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01] | [-7.326599597930908, 3.7763895988464355] |
43a1d6ec-37b4-4dce-bfcb-948b21bf8730 | pyvhr-a-python-framework-for-remote | null | null | https://peerj.com/articles/cs-929/ | https://peerj.com/articles/cs-929/ | pyVHR: a Python framework for remote photoplethysmography | Remote photoplethysmography (rPPG) aspires to automatically estimate heart rate (HR) variability from videos in realistic environments. A number of effective methods relying on data-driven, model-based and statistical approaches have emerged in the past two decades. They exhibit increasing ability to estimate the blood volume pulse (BVP) signal upon which BPMs (Beats per Minute) can be estimated. Furthermore, learning-based rPPG methods have been recently proposed. The present pyVHR framework represents a multi-stage pipeline covering the whole process for extracting and analyzing HR fluctuations. It is designed for both theoretical studies and practical applications in contexts where wearable sensors are inconvenient to use. Namely, pyVHR supports either the development, assessment and statistical analysis of novel rPPG methods, either traditional or learning-based, or simply the sound comparison of well-established methods on multiple datasets. It is built up on accelerated Python libraries for video and signal processing as well as equipped with parallel/accelerated ad-hoc procedures paving the way to online processing on a GPU. The whole accelerated process can be safely run in real-time for 30 fps HD videos with an average speedup of around 5. This paper is shaped in the form of a gentle tutorial presentation of the framework. | ['Edoardo Mortara', 'Raffaella Lanzarotti', 'Giuliano Grossi', 'Alessandro D’Amelio\u200b', 'Vittorio Cuculo', 'Donatello Conte', 'Giuseppe Boccignone'] | 2022-04-15 | null | null | null | peerj-computer-science-2022-4 | ['physiological-computing', 'photoplethysmography-ppg-heart-rate', 'photoplethysmography-ppg', 'heart-rate-variability', 'heart-rate-estimation'] | ['computer-vision', 'medical', 'medical', 'medical', 'medical'] | [ 1.77637324e-01 -1.97741807e-01 1.85474604e-01 -3.63389760e-01
-6.34084702e-01 -2.73768067e-01 2.61774927e-01 1.58882305e-01
-4.14381027e-01 6.72384322e-01 -7.97301307e-02 -2.48341933e-01
1.40006423e-01 -4.40621853e-01 -8.39091614e-02 -9.03047025e-01
-1.93824306e-01 9.88535285e-02 1.75995499e-01 1.51924565e-01
3.21918100e-01 5.92772782e-01 -1.71588016e+00 -9.68001708e-02
7.59519279e-01 1.11555636e+00 -6.62340596e-02 1.20414889e+00
9.06045139e-02 5.97296059e-01 -3.52704674e-01 -7.72404671e-02
1.76083311e-01 -7.36593604e-01 -1.14850663e-01 -3.15064132e-01
3.43652725e-01 -3.57177764e-01 -3.41687232e-01 3.18958610e-01
1.13557684e+00 -3.17497477e-02 6.01757430e-02 -9.32311296e-01
2.82770544e-01 2.23719016e-01 -3.02730531e-01 5.22922277e-01
4.13257867e-01 5.11068344e-01 1.61853448e-01 -7.63358653e-01
2.57620215e-01 6.41194522e-01 7.71303535e-01 2.78241336e-01
-1.18008864e+00 -2.28346124e-01 -7.43659198e-01 3.80757779e-01
-1.30837727e+00 -6.18588090e-01 7.48065829e-01 -4.78255451e-01
6.93200290e-01 7.19472110e-01 9.50037062e-01 7.90434539e-01
1.39751703e-01 2.10306644e-01 1.62368309e+00 -4.05115277e-01
1.96274981e-01 4.90327984e-01 4.66112681e-02 3.43874604e-01
4.93190028e-02 1.37425691e-01 -7.87966251e-01 -1.16828389e-01
9.03726757e-01 -3.41494530e-01 -5.37667692e-01 -8.18194821e-02
-1.06503606e+00 4.73955154e-01 -3.33367884e-01 3.69652033e-01
-5.19464612e-01 -9.53008980e-02 8.15153122e-01 1.75459400e-01
3.21639836e-01 1.79959342e-01 -3.32498640e-01 -8.51620197e-01
-1.18479109e+00 3.07032704e-01 1.09829354e+00 6.57319784e-01
3.87029618e-01 1.00415751e-01 -3.85435253e-01 6.11897886e-01
2.30195478e-01 5.20631373e-01 5.81120014e-01 -1.13725519e+00
5.33721820e-02 2.58593470e-01 2.12284788e-01 -1.04708958e+00
-5.77064037e-01 -2.44718753e-02 -7.88896561e-01 2.16356337e-01
5.44308186e-01 -1.59233481e-01 -3.35941792e-01 1.08629775e+00
5.74739397e-01 3.64265114e-01 -2.75108367e-01 1.05012906e+00
1.00230777e+00 4.08355117e-01 2.22948194e-01 -7.12462544e-01
1.48504639e+00 -3.08740020e-01 -8.21374416e-01 3.34796995e-01
2.59539276e-01 -6.56826019e-01 9.57688630e-01 6.94759607e-01
-1.16316736e+00 -8.21170688e-01 -8.28851402e-01 -2.28382610e-02
-1.42710268e-01 6.96103349e-02 4.22376543e-01 1.22304773e+00
-1.01725364e+00 1.15342295e+00 -1.03572857e+00 -2.91324973e-01
1.62437141e-01 1.56719208e-01 -1.20913133e-01 2.09999889e-01
-1.03237867e+00 9.47181404e-01 -2.09139232e-02 6.18884683e-01
-4.34674054e-01 -1.03349626e+00 -7.02385664e-01 -1.24294475e-01
3.34360488e-02 -5.65399408e-01 7.50572264e-01 -4.93792862e-01
-2.17968607e+00 1.04082155e+00 -1.54525697e-01 -5.10776103e-01
9.06006336e-01 -2.81651646e-01 -3.93270522e-01 5.76528847e-01
-6.18467569e-01 -5.94940372e-02 9.25382972e-01 -5.33189893e-01
4.58416715e-03 -4.25625950e-01 -5.90615809e-01 2.12749749e-01
-4.97224331e-02 2.17531085e-01 -3.57139349e-01 -1.49508312e-01
-1.64260477e-01 -7.50276685e-01 2.27258839e-02 7.22194910e-02
-2.60672551e-02 6.45636842e-02 5.67812026e-01 -8.27321947e-01
1.36367893e+00 -1.84819889e+00 -1.34772480e-01 1.27738059e-01
2.91275442e-01 7.09948540e-01 5.57761014e-01 6.11167967e-01
-1.60642376e-03 -2.11287528e-01 -8.60370770e-02 -1.29529223e-01
-6.38636155e-03 -1.37569949e-01 -2.34774932e-01 8.76683712e-01
-2.34486893e-01 6.94097936e-01 -6.38210773e-01 -6.15301073e-01
1.07130349e+00 8.32706511e-01 -4.02231514e-02 6.13722205e-01
4.82084572e-01 8.81454349e-01 4.81227897e-02 4.68216479e-01
7.82726169e-01 2.15641350e-01 2.84653664e-01 -4.09494370e-01
-7.47793019e-01 1.24076664e-01 -1.11899865e+00 1.48995399e+00
-4.05586839e-01 8.26932669e-01 -4.42298241e-02 -9.05964613e-01
1.24656534e+00 5.05975425e-01 6.75367177e-01 -5.00998437e-01
2.71110713e-01 1.69691786e-01 -1.07534647e-01 -1.09654963e+00
1.22055881e-01 -3.86981517e-01 5.07861733e-01 2.79546529e-01
-2.82335952e-02 -2.17003122e-01 1.06225669e-01 -2.75316447e-01
9.27926898e-01 6.10282004e-01 5.21880984e-01 -3.79775435e-01
8.95312428e-01 -3.22781533e-01 3.20304692e-01 6.37746036e-01
-7.88305640e-01 6.39060795e-01 4.55502719e-01 -6.02620304e-01
-1.02897084e+00 -8.75892103e-01 -6.01899564e-01 5.17032981e-01
-9.00212526e-02 -5.24972916e-01 -5.55889726e-01 2.05503926e-01
-1.81058690e-01 3.19844097e-01 -2.96041161e-01 1.57154188e-01
-5.19788504e-01 -8.47963810e-01 4.39564735e-01 3.50465357e-01
4.29152459e-01 -1.32821715e+00 -1.21568573e+00 4.28699374e-01
-1.12267807e-01 -1.19860077e+00 1.73468888e-01 -5.10622561e-02
-1.06865728e+00 -1.00064147e+00 -7.69892931e-01 -2.90296506e-02
7.74637088e-02 -1.49939686e-01 9.56660628e-01 -7.92862624e-02
-1.00043809e+00 7.11525083e-01 -3.01903576e-01 -5.23854792e-01
-2.04803526e-01 -3.66422713e-01 7.28511661e-02 9.40762833e-02
5.37530780e-01 -9.51996624e-01 -1.11572623e+00 2.31473818e-01
-3.59898537e-01 -4.87050526e-02 2.68490285e-01 2.61367530e-01
6.21810794e-01 -6.20388806e-01 5.49969137e-01 -6.29463851e-01
3.07874680e-01 -1.78750947e-01 -7.69164443e-01 -7.92554319e-02
-5.84809124e-01 -4.63748127e-01 7.56782293e-01 -1.37450203e-01
-7.89749980e-01 1.10130988e-01 -2.38042906e-01 -5.25170386e-01
-3.29523116e-01 2.41151124e-01 2.99792081e-01 -2.26256266e-01
7.63575375e-01 4.36818987e-01 3.67681801e-01 -1.94906428e-01
1.07150443e-01 7.37520874e-01 8.30050111e-01 -4.23103869e-01
5.14536798e-01 4.28453714e-01 2.53052682e-01 -1.39441836e+00
-2.77108669e-01 -7.08882511e-01 -8.15113008e-01 -8.31839502e-01
7.71541059e-01 -8.18673551e-01 -1.14432776e+00 7.11695254e-01
-6.65982187e-01 -4.61416960e-01 -2.12483168e-01 7.06700504e-01
-7.62055278e-01 7.27625668e-01 -5.09895086e-01 -1.30903375e+00
-7.69008875e-01 -6.06019497e-01 6.62748873e-01 5.42282820e-01
-3.14680845e-01 -1.06152010e+00 4.60220665e-01 3.99853259e-01
8.04562271e-01 8.52795959e-01 3.05293016e-02 -9.92559120e-02
-2.06526294e-01 -4.47784066e-01 4.72720489e-02 3.96570235e-01
-1.60817206e-01 2.36957133e-01 -1.37110150e+00 -1.81320831e-01
3.59957516e-01 -2.36173570e-01 3.31592321e-01 6.96572244e-01
1.27145243e+00 1.87190697e-01 9.24633518e-02 6.50540948e-01
1.56548059e+00 -1.41652003e-02 1.23383939e+00 9.22960043e-03
4.47086304e-01 6.72699213e-01 5.10581434e-01 9.28943276e-01
1.36377096e-01 7.20803320e-01 1.08144246e-01 -1.35602266e-01
1.80813938e-01 1.75822198e-01 3.13724995e-01 6.73437178e-01
-6.76610053e-01 2.91888207e-01 -7.61994720e-01 1.12652712e-01
-1.46908033e+00 -1.01026559e+00 -8.92210782e-01 2.90505481e+00
6.91588700e-01 -1.05457485e-01 4.49150622e-01 2.59108633e-01
7.76863277e-01 8.33017752e-02 -3.73235494e-01 -6.92283809e-01
2.97296822e-01 5.43503106e-01 3.44467789e-01 2.22286463e-01
-1.00040841e+00 2.51781315e-01 6.33298683e+00 7.77285397e-02
-1.41548312e+00 -1.16353169e-01 5.90459645e-01 -4.00259867e-02
3.56671125e-01 -6.46404922e-02 -6.12405241e-01 6.91931546e-01
1.70005178e+00 -3.93547833e-01 2.84375101e-01 7.35155582e-01
9.47774172e-01 -5.51481605e-01 -7.17228115e-01 1.36055291e+00
7.70646185e-02 -1.16023910e+00 -7.56271183e-01 -1.08811855e-01
-6.03621639e-02 -2.58952808e-02 -4.26384956e-01 1.53833494e-01
-8.62226605e-01 -7.44689703e-01 1.07144199e-01 8.01984787e-01
1.02921689e+00 -4.40646648e-01 7.55899847e-01 1.39394850e-01
-1.08468628e+00 2.48182073e-01 -4.88175362e-01 -1.73955411e-01
2.80501753e-01 9.97446477e-01 -3.91963571e-01 6.15444899e-01
7.12082624e-01 5.70617914e-01 -3.28758717e-01 1.29469895e+00
-1.67301401e-01 7.56474257e-01 -4.73846555e-01 7.61894733e-02
-3.66637707e-01 -4.86695021e-01 4.93128598e-01 1.24925494e+00
3.73997509e-01 3.15042049e-01 -2.80956596e-01 8.06924462e-01
6.37926459e-01 4.03036535e-01 -4.19064969e-01 5.05261607e-02
2.59151191e-01 1.85532212e+00 -7.39973903e-01 -4.26982909e-01
-4.03615296e-01 5.78169584e-01 -1.54743388e-01 -8.68000761e-02
-1.03335857e+00 -7.08728313e-01 3.30138147e-01 4.23215657e-01
-1.50381299e-02 -2.51670599e-01 -3.48412395e-01 -1.01094186e+00
7.59833828e-02 -5.26727021e-01 3.83675337e-01 -7.39542067e-01
-8.95682395e-01 5.25771737e-01 8.08069017e-03 -1.32898498e+00
-2.01664910e-01 -5.07561982e-01 -9.70320046e-01 1.11203170e+00
-1.74018025e+00 -3.85485172e-01 -9.49259818e-01 5.38255751e-01
2.04022020e-01 3.78370285e-01 1.02488101e+00 4.34019923e-01
-7.33632863e-01 1.18833348e-01 -3.01845104e-01 -2.73919433e-01
7.87434757e-01 -1.35874021e+00 -1.31145194e-01 7.44857430e-01
-4.07426476e-01 5.05893230e-01 9.54833746e-01 -1.17250875e-01
-1.74959874e+00 -6.36953592e-01 8.16064179e-01 -3.79803509e-01
5.01774251e-01 -2.57356465e-01 -9.33566034e-01 3.95665504e-02
9.50155482e-02 4.54677880e-01 8.81442547e-01 -2.30776817e-01
1.61885321e-01 -5.38016796e-01 -1.10218751e+00 1.71149969e-01
3.84948552e-01 -3.31138939e-01 -3.30507576e-01 1.44342735e-01
-2.06944138e-01 -8.56520772e-01 -1.53003025e+00 1.53976649e-01
8.38039279e-01 -1.44856596e+00 6.58234358e-01 2.00598361e-03
2.27239624e-01 -2.85581470e-01 4.78792936e-01 -7.36157298e-01
2.16793716e-01 -1.35590971e+00 -3.24878633e-01 1.12843633e+00
-2.57990271e-01 -7.53585279e-01 5.93036652e-01 9.90934372e-01
-1.86995208e-01 -6.44707024e-01 -8.65408003e-01 -5.61874330e-01
-4.71051812e-01 -5.90729713e-01 -1.67938918e-01 5.30059636e-01
4.88662064e-01 -4.47483175e-02 -5.19595325e-01 -7.43313804e-02
7.57408142e-01 1.75137848e-01 9.15563524e-01 -1.08978975e+00
-3.57206345e-01 -1.23166993e-01 -6.27203524e-01 -4.81040776e-01
-4.55046594e-01 -3.97707433e-01 3.95337269e-02 -1.20588064e+00
-7.61541212e-03 -1.30835772e-01 -3.19048315e-01 1.63657114e-01
-2.13556796e-01 4.46448624e-01 2.66935546e-02 2.95412429e-02
-1.70285419e-01 1.10067539e-01 8.83416951e-01 6.81671858e-01
-6.04861856e-01 1.85036302e-01 -2.21376255e-01 2.84806550e-01
8.28400791e-01 -2.85877734e-01 -3.33619356e-01 6.14814878e-01
3.63967866e-02 5.17584503e-01 5.67104459e-01 -1.35485840e+00
1.19698290e-02 2.60565400e-01 4.19146508e-01 -4.63485867e-01
1.89430118e-01 -5.47124743e-01 3.27097178e-01 6.30033612e-01
-8.74130651e-02 -1.50131434e-01 2.01685563e-01 2.15707287e-01
-2.03839675e-01 4.78350148e-02 1.03989363e+00 -2.99425215e-01
-3.93081307e-01 2.83974767e-01 -2.89281338e-01 -3.30142095e-03
1.03858197e+00 -5.01285493e-01 -2.44194657e-01 -3.92992169e-01
-8.12483668e-01 -6.25895932e-02 8.52015987e-02 -8.85782018e-02
4.72063303e-01 -8.64254832e-01 -7.92391658e-01 1.12312175e-01
9.73570999e-03 -5.15662551e-01 9.39178467e-01 1.76407635e+00
-9.92573082e-01 4.28045541e-01 -4.04983044e-01 -7.99911499e-01
-1.39020574e+00 4.68285382e-01 5.31919479e-01 -3.14800553e-02
-1.16285610e+00 3.11379373e-01 -3.51174831e-01 3.64139736e-01
5.30157089e-02 -7.34076127e-02 -4.90565747e-01 4.25286442e-02
8.82799983e-01 7.85547912e-01 1.70564428e-01 -2.25406140e-01
-2.28529230e-01 5.82789004e-01 5.36890924e-01 1.65072247e-01
1.28808284e+00 -4.60623860e-01 -1.17361434e-01 7.99992263e-01
9.83474016e-01 -9.09766555e-02 -1.33559000e+00 3.66161823e-01
-2.31163099e-01 -4.74597722e-01 1.73540249e-01 -6.74190521e-01
-8.60440791e-01 9.77114320e-01 9.39950645e-01 2.92402327e-01
1.36666715e+00 -5.31238914e-01 4.97869968e-01 -2.39078432e-01
1.31614596e-01 -1.11742961e+00 -3.77070636e-01 -2.28212789e-01
7.08664060e-01 -9.05963004e-01 3.71096522e-01 -4.86524045e-01
-6.49588883e-01 1.67269981e+00 3.87796998e-01 -1.44113258e-01
5.89885414e-01 3.90254050e-01 2.63263077e-01 -4.19199727e-02
-6.66647077e-01 -2.33610440e-02 1.23890944e-01 6.91643417e-01
7.63077736e-01 -6.11034296e-02 -9.33634102e-01 2.16312334e-01
3.31282057e-02 4.57792997e-01 8.21172059e-01 6.75191820e-01
-4.07333672e-01 -8.27226877e-01 -2.02489704e-01 4.22417343e-01
-5.69743097e-01 8.77223313e-02 2.19946504e-01 8.26634526e-01
-1.98250771e-01 8.58730495e-01 -8.33882019e-02 -3.10470778e-02
3.32674831e-01 3.76926541e-01 5.88099122e-01 -2.16948047e-01
-6.75700843e-01 1.85541824e-01 -3.07875080e-03 -8.90330970e-01
-7.10438192e-01 -7.74583340e-01 -9.34551835e-01 -3.25337619e-01
6.34038970e-02 5.07052578e-02 9.79392827e-01 7.35094547e-01
3.54004920e-01 1.96019992e-01 8.43348086e-01 -1.03102899e+00
-1.07303828e-01 -9.59426403e-01 -9.47489679e-01 1.32244766e-01
7.18427002e-02 -3.28973800e-01 -6.16340756e-01 1.17752947e-01] | [13.870320320129395, 2.8020215034484863] |
ca213940-193d-43dc-830d-f31d2db09973 | clustering-multilayer-graphs-with-missing | 2103.03235 | null | https://arxiv.org/abs/2103.03235v1 | https://arxiv.org/pdf/2103.03235v1.pdf | Clustering multilayer graphs with missing nodes | Relationship between agents can be conveniently represented by graphs. When these relationships have different modalities, they are better modelled by multilayer graphs where each layer is associated with one modality. Such graphs arise naturally in many contexts including biological and social networks. Clustering is a fundamental problem in network analysis where the goal is to regroup nodes with similar connectivity profiles. In the past decade, various clustering methods have been extended from the unilayer setting to multilayer graphs in order to incorporate the information provided by each layer. While most existing works assume - rather restrictively - that all layers share the same set of nodes, we propose a new framework that allows for layers to be defined on different sets of nodes. In particular, the nodes not recorded in a layer are treated as missing. Within this paradigm, we investigate several generalizations of well-known clustering methods in the complete setting to the incomplete one and prove some consistency results under the Multi-Layer Stochastic Block Model assumption. Our theoretical results are complemented by thorough numerical comparisons between our proposed algorithms on synthetic data, and also on real datasets, thus highlighting the promising behaviour of our methods in various settings. | ['Christophe Biernacki', 'Hemant Tyagi', 'Guillaume Braun'] | 2021-03-04 | null | null | null | null | ['stochastic-block-model'] | ['graphs'] | [ 2.38283977e-01 2.53072590e-01 -2.33855888e-01 -1.10720195e-01
1.90359280e-01 -5.74264407e-01 8.94724905e-01 6.44874394e-01
-3.31161767e-01 7.07120240e-01 1.70658305e-01 -9.36375640e-04
-6.03155255e-01 -8.76193345e-01 -5.37390053e-01 -1.11637700e+00
-3.72536153e-01 7.12755144e-01 3.23877960e-01 -8.15826431e-02
-1.99433565e-01 3.69207650e-01 -1.06618214e+00 -1.36394739e-01
7.50085175e-01 5.75308740e-01 -6.81766123e-02 3.24476868e-01
-1.42856956e-01 7.11610317e-01 -4.32938308e-01 -5.34496665e-01
3.63820404e-01 -1.93121240e-01 -5.54453313e-01 6.44115031e-01
-1.09226301e-01 5.77581823e-01 -3.69892895e-01 1.15954137e+00
1.52131632e-01 -1.13973871e-01 8.24625969e-01 -1.61099648e+00
-1.84806570e-01 9.80976939e-01 -9.00544822e-01 -2.15412378e-01
1.88455656e-01 -3.59626472e-01 9.69366372e-01 -1.57291964e-01
6.18961334e-01 1.12218213e+00 4.28960025e-01 3.76231253e-01
-1.59661889e+00 -3.10655653e-01 4.46717530e-01 5.71025489e-03
-1.45425820e+00 -1.41366050e-01 9.85831797e-01 -7.04569519e-01
6.25985786e-02 3.09929162e-01 6.89692378e-01 7.50682414e-01
-1.01219095e-01 4.32630301e-01 1.33707035e+00 -3.41572613e-01
2.66199321e-01 2.14506015e-01 4.57141876e-01 5.24308264e-01
8.06076646e-01 -5.02066076e-01 -3.92666548e-01 -2.98241556e-01
5.19274533e-01 2.90068805e-01 -4.22401965e-01 -1.08143294e+00
-1.34665656e+00 6.78974509e-01 5.78871429e-01 5.75344980e-01
-3.43134135e-01 -1.77801102e-01 2.73201704e-01 3.75709862e-01
4.28498596e-01 -1.16738662e-01 8.68824050e-02 5.63889623e-01
-6.92186356e-01 -1.11739911e-01 1.14860725e+00 9.66585040e-01
8.48666549e-01 -4.22756821e-01 2.24106386e-01 6.35211945e-01
5.19692302e-01 2.32000975e-03 -6.52723685e-02 -6.72867239e-01
5.43723643e-01 8.84628415e-01 1.58813566e-01 -1.22682595e+00
-6.85505748e-01 -7.59081423e-01 -1.54560399e+00 2.27433327e-03
6.99630857e-01 -1.61306649e-01 -4.02876377e-01 2.03907037e+00
5.32077134e-01 3.23249161e-01 1.83903292e-01 6.19628966e-01
5.24652600e-01 4.04716492e-01 -6.57649115e-02 -5.43202698e-01
1.18764818e+00 -6.96569324e-01 -6.18870378e-01 2.04580054e-01
4.61036175e-01 -2.53627300e-01 3.06738883e-01 4.29346144e-01
-9.41183627e-01 1.34613011e-02 -8.84526312e-01 5.60226977e-01
-2.02886552e-01 -3.61368716e-01 5.03637969e-01 5.47323108e-01
-1.40289891e+00 3.61048847e-01 -5.35047650e-01 -7.25496233e-01
1.39191076e-01 3.80351275e-01 -5.63386738e-01 -9.54442024e-02
-9.75221336e-01 2.97962666e-01 2.75358230e-01 3.84373724e-01
-5.60357451e-01 -1.12494148e-01 -6.93005204e-01 -7.63819590e-02
6.76294744e-01 -9.96242523e-01 5.57280660e-01 -9.31303144e-01
-9.86518323e-01 7.56441116e-01 -5.95137291e-02 -4.71733093e-01
7.78195679e-01 4.50685710e-01 -2.09246159e-01 2.24616334e-01
-6.34859353e-02 1.35548517e-01 6.75126076e-01 -1.62831771e+00
-3.97022039e-01 -4.78235394e-01 5.87990463e-01 2.37374216e-01
-4.11913246e-01 -1.42845109e-01 -4.95777428e-01 -4.31812257e-01
2.06361026e-01 -1.13766479e+00 -4.21638310e-01 -9.75963697e-02
-8.09874833e-01 -2.60264248e-01 5.63171566e-01 -1.38929440e-02
1.11638033e+00 -1.90194702e+00 8.36916447e-01 6.65428460e-01
6.73633456e-01 -6.03139810e-02 8.17102864e-02 8.90226007e-01
1.90905333e-02 2.19945669e-01 -4.41524506e-01 -5.74591100e-01
1.39009669e-01 3.35796684e-01 2.46347055e-01 9.43268239e-01
-3.38823736e-01 4.86527026e-01 -8.14227164e-01 -6.91196740e-01
1.30347580e-01 4.60173756e-01 -3.89706105e-01 -1.12900607e-01
4.56739217e-03 6.65562451e-01 -4.91998523e-01 1.26733810e-01
7.21041799e-01 -5.82381904e-01 7.93428779e-01 -1.10345017e-02
9.50008631e-02 -2.73085594e-01 -1.54929101e+00 1.38624167e+00
-1.90786168e-01 2.46258661e-01 7.59204090e-01 -1.32453561e+00
5.91724277e-01 5.02996027e-01 6.41735613e-01 1.26385942e-01
9.25945863e-02 -1.00426421e-01 1.81256428e-01 -2.16547221e-01
1.17338113e-01 -1.82218477e-01 -1.52757123e-01 4.15647745e-01
-3.30636501e-01 2.61335671e-01 6.32088363e-01 4.97459173e-01
1.08222735e+00 -4.50214088e-01 3.48385632e-01 -4.97730672e-01
6.59626007e-01 -3.24322253e-01 5.74806571e-01 6.80770218e-01
5.65706678e-02 1.99243098e-01 1.02480352e+00 6.25300258e-02
-8.27134550e-01 -9.19959962e-01 -1.65271714e-01 4.82580453e-01
5.25138497e-01 -3.89975458e-01 -7.72049248e-01 -3.52671146e-01
-1.46728950e-02 -2.33039245e-01 -8.05931687e-01 1.12063184e-01
-1.61318421e-01 -1.07010674e+00 3.42929453e-01 2.85579581e-02
4.81605023e-01 -6.42600358e-01 -1.58148959e-01 1.46685928e-01
-5.36576286e-02 -1.17854881e+00 -7.53023028e-02 -1.21415136e-02
-9.63744819e-01 -1.34388852e+00 -9.44364607e-01 -9.70432281e-01
9.29281652e-01 2.41661444e-01 9.10442173e-01 4.41587597e-01
1.91243991e-01 5.23768067e-01 -4.42355692e-01 2.47249618e-01
-3.85077477e-01 2.11581722e-01 2.56397665e-01 7.38029361e-01
-1.40738472e-01 -6.81666732e-01 -5.42800903e-01 4.12615448e-01
-1.23257387e+00 1.39569432e-01 4.54529971e-01 4.41052854e-01
2.82082975e-01 3.99753630e-01 5.01469016e-01 -1.25508249e+00
5.71264386e-01 -8.08562160e-01 -5.59318841e-01 3.63343626e-01
-3.38536799e-01 6.84796125e-02 5.65526843e-01 -2.44682819e-01
-8.26694608e-01 -9.07505155e-02 4.13974106e-01 -2.43821181e-02
-1.21386953e-01 7.97920704e-01 -5.24447858e-01 -8.95902440e-02
-1.34194121e-02 -7.52004236e-02 8.93430859e-02 -4.83076245e-01
3.12189370e-01 4.87834781e-01 1.30401373e-01 -4.55986887e-01
1.04419565e+00 8.14345002e-01 6.12728298e-01 -1.00788319e+00
-3.46118510e-01 -3.16924304e-01 -8.04205596e-01 -4.30493414e-01
6.71099126e-01 -7.12879658e-01 -9.74390686e-01 6.19572639e-01
-7.68492162e-01 -3.27968597e-01 1.17250681e-01 3.77591103e-01
-2.14962006e-01 8.01860273e-01 -7.87655294e-01 -8.95130157e-01
2.37862363e-01 -1.01760602e+00 5.13847709e-01 1.19558677e-01
1.01635531e-01 -1.52439129e+00 2.79658616e-01 1.86024904e-01
1.09370984e-01 4.64750618e-01 1.10010326e+00 -6.49138749e-01
-4.69860733e-01 -3.70860323e-02 -8.70901197e-02 -5.10559268e-02
3.25092822e-01 -6.71402663e-02 -4.38889712e-01 -3.88645023e-01
-1.56324178e-01 2.09758520e-01 9.15235400e-01 3.34657729e-01
6.23349488e-01 -3.64389122e-01 -7.42350459e-01 2.81602919e-01
1.40318656e+00 -8.24313685e-02 2.52604485e-01 1.52959466e-01
7.48317659e-01 1.02676058e+00 -9.81093273e-02 3.42621654e-01
5.56578457e-01 7.54073203e-01 5.83289087e-01 -1.01476677e-01
3.59661907e-01 1.32721543e-01 1.34261578e-01 1.09392488e+00
-1.63388088e-01 -7.50126243e-01 -6.46440864e-01 4.46476161e-01
-2.07021451e+00 -8.96462142e-01 -4.11882102e-01 2.40505266e+00
5.65135896e-01 1.20931730e-01 5.49949169e-01 2.98150778e-01
1.30259526e+00 3.16293716e-01 -3.27587605e-01 2.16020495e-01
-4.30432707e-01 -4.50546354e-01 3.10178697e-01 7.19343305e-01
-1.02242148e+00 3.12187403e-01 5.56126308e+00 4.54920352e-01
-5.77641368e-01 -8.17615166e-02 5.34742951e-01 1.07335322e-01
-3.43372792e-01 8.75019431e-02 -4.15165931e-01 4.08407629e-01
4.98241097e-01 -2.34682083e-01 3.17303866e-01 1.80419356e-01
1.86120376e-01 -2.17195496e-01 -1.06960905e+00 6.34704709e-01
-8.61972049e-02 -9.00593281e-01 4.15168330e-02 5.78909576e-01
9.35343742e-01 -4.82524872e-01 8.79723951e-03 -3.60366195e-01
5.05613327e-01 -8.49849999e-01 3.86929840e-01 6.81809008e-01
4.92974937e-01 -7.03488886e-01 5.66855788e-01 6.02489412e-01
-1.53984261e+00 4.69289236e-02 -1.89748183e-01 -1.71815872e-01
2.89516151e-01 7.84502864e-01 -3.93513739e-01 1.09994590e+00
3.05098742e-01 9.51406598e-01 -4.38202679e-01 1.34067178e+00
8.53916556e-02 5.22637546e-01 -3.98282737e-01 -9.93617624e-02
1.66180760e-01 -8.25197637e-01 6.54590666e-01 9.14985120e-01
6.16592467e-02 -1.84648305e-01 2.03234985e-01 4.29003030e-01
-8.23554397e-02 8.47114995e-02 -7.34844208e-01 9.47801247e-02
3.82844985e-01 1.35192025e+00 -1.21064150e+00 -2.96514183e-01
-6.46902919e-01 7.06345916e-01 4.00103539e-01 5.51950455e-01
-6.06230795e-01 -1.51006788e-01 5.99190295e-01 3.18724960e-01
1.75954342e-01 -2.67618716e-01 1.38920590e-01 -1.17404532e+00
7.18912401e-04 -4.37522799e-01 3.96061361e-01 -2.45206967e-01
-1.63610780e+00 5.49686432e-01 2.78910309e-01 -1.26290810e+00
-6.36211261e-02 -2.45573506e-01 -4.10982341e-01 4.01043653e-01
-1.36033952e+00 -9.41611826e-01 -1.61820203e-01 8.15311015e-01
-9.36868563e-02 1.94809049e-01 4.25567687e-01 3.17812920e-01
-5.86204946e-01 1.66696399e-01 4.35663849e-01 2.10372046e-01
2.48544961e-01 -1.35300469e+00 -3.19235504e-01 6.70659900e-01
1.97378784e-01 7.17284381e-01 8.00317228e-01 -6.86471581e-01
-1.16491449e+00 -1.03143537e+00 8.44551623e-01 -4.39142920e-02
8.01190019e-01 -6.93286896e-01 -8.79436314e-01 6.78037822e-01
4.42333966e-01 -1.44656971e-01 6.28078163e-01 8.66298843e-03
-1.54118031e-01 -1.98577970e-01 -1.08664322e+00 7.87868381e-01
1.20617163e+00 -2.72950560e-01 -1.85444117e-01 2.98525900e-01
3.98619592e-01 3.27869922e-01 -1.02424538e+00 3.77296299e-01
2.67352670e-01 -1.01513326e+00 7.06586182e-01 -3.09253901e-01
-5.40306978e-02 -6.79465950e-01 7.83825964e-02 -1.45560431e+00
-1.42099217e-01 -7.81406105e-01 1.45381019e-01 1.44943106e+00
3.41572016e-01 -9.51627076e-01 6.63956583e-01 2.90875435e-01
3.21482331e-01 -5.13146579e-01 -1.06503046e+00 -7.24361897e-01
2.72270571e-02 -6.98677450e-02 4.71153796e-01 9.85990882e-01
2.96447366e-01 5.05274415e-01 -4.81599122e-01 2.69172966e-01
1.06873441e+00 7.99234882e-02 7.21267819e-01 -1.80194449e+00
-3.14310521e-01 -5.24288058e-01 -6.45069301e-01 -1.04684818e+00
3.21374953e-01 -9.36997056e-01 -1.81811810e-01 -1.77570319e+00
6.21437013e-01 -7.31412113e-01 -1.15442239e-01 -4.35847230e-02
1.42812535e-01 3.75909269e-01 2.01671734e-01 4.01297778e-01
-8.25645983e-01 4.17209089e-01 9.96274948e-01 -1.88498855e-01
-5.42560145e-02 3.20822299e-01 -5.28928697e-01 8.54350090e-01
6.12973511e-01 -2.47471303e-01 -6.39571786e-01 -1.85027122e-01
6.14373863e-01 2.75981247e-01 3.84491980e-01 -1.00202990e+00
5.31243742e-01 8.32907483e-02 -1.63576156e-01 -2.43790731e-01
4.28472757e-01 -1.27409971e+00 7.76436448e-01 3.10977727e-01
-4.21480715e-01 7.74213579e-03 -4.12338495e-01 1.10125196e+00
-4.31574509e-02 -2.55739063e-01 4.98760521e-01 -6.18163273e-02
-1.58635616e-01 2.91324824e-01 -4.80110168e-01 -1.23063035e-01
1.34481239e+00 -1.69817179e-01 -3.58790457e-01 -6.73574865e-01
-1.22152984e+00 4.18745309e-01 8.81399035e-01 -3.39745395e-02
2.38713726e-01 -1.10003245e+00 -6.60487592e-01 -1.39930859e-01
9.50048491e-02 7.24681094e-02 2.38885880e-01 1.34174037e+00
-2.13664189e-01 2.83228129e-01 7.79669657e-02 -7.89170325e-01
-1.35821939e+00 7.56388307e-01 3.22510570e-01 -4.64686096e-01
-5.37200272e-01 3.08723181e-01 6.55387342e-01 -1.66445926e-01
3.45495850e-01 -7.68471807e-02 -5.28965473e-01 3.31409246e-01
9.63432044e-02 3.45730156e-01 -2.93543547e-01 -1.01876271e+00
-3.71581703e-01 6.53759658e-01 3.37063931e-02 -2.83246696e-01
1.38806927e+00 -4.73789364e-01 -5.30174375e-01 8.58633459e-01
7.86534965e-01 2.50061959e-01 -1.04438138e+00 -5.40723562e-01
2.47862816e-01 3.18377577e-02 -4.91193146e-01 -1.79966688e-01
-1.32870293e+00 4.48985487e-01 -7.15582818e-02 9.63562548e-01
9.97772992e-01 4.75373477e-01 1.93245690e-02 2.18490466e-01
6.21794283e-01 -5.99900901e-01 -1.59477144e-01 1.67192668e-01
4.92652208e-01 -7.34271705e-01 -3.30598056e-02 -7.19093978e-01
-3.66311640e-01 8.29137802e-01 1.43420428e-01 -2.65197575e-01
9.73056257e-01 1.49643525e-01 -3.63831580e-01 -2.52803653e-01
-6.92062974e-01 -4.52143878e-01 -1.28095627e-01 5.37363648e-01
2.62909472e-01 1.50458619e-01 -3.89112055e-01 2.06758603e-01
2.04289883e-01 -3.35378438e-01 7.85218775e-01 7.59415448e-01
-1.30830035e-01 -1.38649690e+00 -3.86465698e-01 4.86723483e-01
-3.30784380e-01 9.80829820e-02 -6.42601669e-01 9.49185371e-01
-1.07124723e-01 1.26903176e+00 5.21422327e-02 -2.60253906e-01
7.07428306e-02 -2.30269164e-01 4.67997640e-01 -6.61250949e-01
-5.63352525e-01 2.81304806e-01 -1.49760675e-02 5.82172088e-02
-1.20322740e+00 -9.08068001e-01 -8.78515780e-01 -4.51646864e-01
-2.99514979e-01 5.16111910e-01 2.61457890e-01 8.98996055e-01
2.30620801e-03 4.53254044e-01 6.54586494e-01 -7.36187816e-01
6.03196444e-03 -9.49584424e-01 -1.23728859e+00 3.51405561e-01
5.07129192e-01 -7.78326035e-01 -6.13131642e-01 -5.53893261e-02] | [7.042267799377441, 5.249378204345703] |
4dd1d8d7-11b2-41b9-9079-26996e613d0e | interpretable-stochastic-model-predictive | 2205.07150 | null | https://arxiv.org/abs/2205.07150v1 | https://arxiv.org/pdf/2205.07150v1.pdf | Interpretable Stochastic Model Predictive Control using Distributional Reinforced Estimation for Quadrotor Tracking Systems | This paper presents a novel trajectory tracker for autonomous quadrotor navigation in dynamic and complex environments. The proposed framework integrates a distributional Reinforcement Learning (RL) estimator for unknown aerodynamic effects into a Stochastic Model Predictive Controller (SMPC) for trajectory tracking. Aerodynamic effects derived from drag forces and moment variations are difficult to model directly and accurately. Most current quadrotor tracking systems therefore treat them as simple `disturbances' in conventional control approaches. We propose Quantile-approximation-based Distributional Reinforced-disturbance-estimator, an aerodynamic disturbance estimator, to accurately identify disturbances, i.e., uncertainties between the true and estimated values of aerodynamic effects. Simplified Affine Disturbance Feedback is employed for control parameterization to guarantee convexity, which we then integrate with a SMPC to achieve sufficient and non-conservative control signals. We demonstrate our system to improve the cumulative tracking errors by at least 66% with unknown and diverse aerodynamic forces compared with recent state-of-the-art. Concerning traditional Reinforcement Learning's non-interpretability, we provide convergence and stability guarantees of Distributional RL and SMPC, respectively, with non-zero mean disturbances. | ['David Boyle', 'Qiuchen Qian', "James O'Keeffe", 'Yanran Wang'] | 2022-05-14 | null | null | null | null | ['distributional-reinforcement-learning'] | ['methodology'] | [-3.02340448e-01 -2.48245467e-02 -2.61290431e-01 3.26196879e-01
-6.76950097e-01 -9.27468598e-01 6.47032738e-01 1.95250168e-01
-1.70210496e-01 1.33818817e+00 -3.07997018e-01 -4.89534378e-01
-6.00872278e-01 -4.83432323e-01 -9.74489212e-01 -1.06376493e+00
-1.14434829e-03 2.20160499e-01 -1.87350184e-01 -4.01710838e-01
-8.82190540e-02 5.24640024e-01 -1.54511905e+00 -8.33971143e-01
1.46087706e+00 7.95343935e-01 2.14370504e-01 8.87905359e-01
7.74466276e-01 4.45225626e-01 -6.87029004e-01 1.19792596e-01
4.45890874e-01 -1.38902292e-01 4.13266361e-01 7.78056681e-02
2.10139602e-01 -3.28654259e-01 -1.67438552e-01 1.23976851e+00
5.46925068e-01 6.37396872e-01 1.07350707e+00 -1.08222353e+00
-5.62444210e-01 4.79473583e-02 -2.94197381e-01 -5.59423715e-02
-1.30239412e-01 3.52023959e-01 5.42473376e-01 -6.99531972e-01
2.20210150e-01 1.05048287e+00 6.37187481e-01 5.94417274e-01
-9.28120315e-01 -5.18792748e-01 5.04837334e-01 -2.27649882e-01
-1.07810211e+00 -2.02537522e-01 4.66197431e-01 -8.83339047e-01
5.16612113e-01 2.95034796e-01 6.86001837e-01 8.62869203e-01
7.64295459e-01 3.74049634e-01 6.69758320e-01 1.60516024e-01
2.86364377e-01 2.18539163e-01 -5.37890077e-01 4.80804920e-01
7.32301176e-01 8.67688715e-01 3.81038457e-01 -3.71592492e-02
9.49660599e-01 -1.39794216e-01 -3.85575920e-01 -7.31818080e-01
-9.17365134e-01 8.85675788e-01 2.50483334e-01 -3.52316201e-01
-5.37490427e-01 9.93856415e-02 3.54902655e-01 3.51933807e-01
5.37164569e-01 7.78758645e-01 -6.79679692e-01 -2.41357028e-01
-4.60706294e-01 7.55616903e-01 7.03833461e-01 1.57277119e+00
5.50706200e-02 1.07867420e+00 -2.07509175e-01 5.10966539e-01
4.21076477e-01 1.35545886e+00 2.24707082e-01 -7.67760575e-01
3.07490975e-01 2.07521349e-01 9.13866937e-01 -9.50347602e-01
-3.87328655e-01 -9.50883150e-01 -9.54933822e-01 5.28602958e-01
3.55683804e-01 -1.04557669e+00 -5.32159626e-01 1.65585721e+00
6.38648093e-01 2.43115313e-02 1.84165299e-01 1.03055573e+00
-1.10625401e-01 6.70637250e-01 -2.78321087e-01 -8.31846237e-01
8.06219697e-01 -6.24787331e-01 -9.36174750e-01 2.46505767e-01
3.61319363e-01 -5.83950102e-01 9.17468131e-01 6.02138937e-01
-1.00462651e+00 -5.83427012e-01 -1.15336955e+00 6.10084116e-01
-2.39546239e-01 8.18600655e-01 -1.76468328e-01 5.96194804e-01
-6.33529305e-01 7.89700091e-01 -7.45594501e-01 1.88098475e-01
-2.49179944e-01 2.83621132e-01 2.42763877e-01 8.33143950e-01
-1.13500869e+00 1.13780558e+00 1.14342965e-01 5.58969490e-02
-1.20928562e+00 -1.20601118e+00 -7.39156961e-01 -3.40671450e-01
6.49740338e-01 -8.15664411e-01 1.32318139e+00 -6.72926545e-01
-2.29339814e+00 -2.54604489e-01 1.23177581e-01 -5.97345352e-01
6.75334871e-01 -7.20015287e-01 -9.83504802e-02 -5.96587770e-02
-2.30961919e-01 -1.03914440e-01 1.41358864e+00 -1.27720857e+00
-8.67835104e-01 -2.26208512e-02 1.46579146e-02 4.47855443e-01
-7.75143504e-02 -5.83579898e-01 4.19269919e-01 -7.28989542e-01
-5.13221204e-01 -1.06549096e+00 -3.30516458e-01 -4.28322963e-02
-3.49517226e-01 -1.75809935e-01 8.09273481e-01 -5.30477226e-01
1.25526929e+00 -1.82375550e+00 4.18883115e-01 -1.53044969e-01
-2.48076860e-02 5.51669121e-01 3.33443880e-01 5.02318919e-01
1.51326582e-01 -1.56970143e-01 -2.88335383e-02 -8.54108855e-02
2.97527194e-01 -4.38890755e-02 -8.66134644e-01 8.01454186e-01
4.84035581e-01 4.75595325e-01 -1.26670182e+00 3.43442559e-01
6.56299114e-01 3.08204889e-01 -5.16824961e-01 1.78689554e-01
-5.67148209e-01 6.02743387e-01 -6.13122344e-01 3.79172146e-01
6.03791058e-01 4.06635433e-01 -1.93012148e-01 6.11562990e-02
-7.17961311e-01 -3.04563522e-01 -1.05301988e+00 8.71304095e-01
-7.49925673e-01 4.46557939e-01 7.83232272e-01 -9.75639462e-01
1.20139909e+00 1.64204329e-01 3.85682404e-01 1.08383626e-01
5.25019228e-01 3.08202088e-01 -1.43582612e-01 -3.03050578e-01
4.13691372e-01 -3.61777872e-01 1.04590751e-01 -4.26871091e-01
-1.70347422e-01 -7.77885377e-01 3.06895990e-02 -4.35285360e-01
4.12688315e-01 2.92591453e-01 4.04891789e-01 -7.60098577e-01
1.01917052e+00 8.63831267e-02 8.10760617e-01 3.10528070e-01
-2.50943244e-01 -4.62245531e-02 2.84940392e-01 5.07070869e-02
-1.19323146e+00 -9.94352221e-01 -8.14560875e-02 6.89986587e-01
2.55409151e-01 6.20464273e-02 -3.85011464e-01 -2.78972507e-01
6.30986512e-01 1.02547491e+00 -2.53158540e-01 -4.71826643e-01
-2.54982203e-01 -3.18177581e-01 1.00287177e-01 4.23736304e-01
-7.85525739e-02 -3.40467915e-02 -5.67796230e-01 4.02655840e-01
7.76558518e-01 -7.72297084e-01 -5.25404155e-01 9.48854089e-02
-4.48556334e-01 -1.29383087e+00 -5.39171159e-01 -3.97661775e-01
6.19925320e-01 -1.91702545e-01 4.20001835e-01 -6.57160401e-01
3.66877988e-02 5.30759096e-01 -8.67634863e-02 -9.50026989e-01
-3.20885122e-01 -3.00667882e-01 9.00728941e-01 -8.29482973e-02
-6.51478589e-01 -2.49451801e-01 -4.05282736e-01 4.86753285e-01
-4.07392770e-01 -3.96480709e-01 3.56229156e-01 1.20911765e+00
7.07907557e-01 4.21809494e-01 9.04742301e-01 -5.22603214e-01
9.79095876e-01 -5.28457701e-01 -1.57787073e+00 -1.49941877e-01
-6.79907978e-01 -1.00922421e-01 1.78031945e+00 -8.15000236e-01
-9.34198916e-01 -6.57298639e-02 2.35243827e-01 -1.12962770e+00
1.05171047e-01 2.23646730e-01 8.96105357e-03 -1.10899784e-01
4.64849353e-01 1.85238034e-01 3.82319123e-01 -1.97884087e-02
4.45567667e-01 5.36078393e-01 4.25119191e-01 -6.70853555e-01
1.36664724e+00 4.34461385e-02 5.37219882e-01 -9.17944968e-01
-5.73249519e-01 -4.20878738e-01 -3.47064406e-01 -2.78375655e-01
4.33356047e-01 -1.13347554e+00 -1.11655509e+00 3.76120061e-01
-8.81136060e-01 -3.93218130e-01 -6.03796303e-01 7.62936175e-01
-1.07421148e+00 -2.03381941e-01 -2.41380364e-01 -1.56025088e+00
-3.33168358e-01 -1.07921565e+00 8.40272903e-01 3.84651959e-01
3.67053449e-01 -1.39301002e+00 1.01990782e-01 -5.74155390e-01
5.16522706e-01 9.32678103e-01 5.13365805e-01 -1.23306170e-01
-1.33012921e-01 -1.37772679e-01 3.76395404e-01 6.78634465e-01
2.19961941e-01 3.88628960e-01 -2.35457033e-01 -8.33489656e-01
2.22268105e-01 -1.23347126e-01 2.31347829e-01 8.81034017e-01
4.16163802e-01 -6.95691586e-01 -4.05291408e-01 5.44005454e-01
1.67352855e+00 6.37273669e-01 -3.38959217e-01 8.55690762e-02
6.64830625e-01 3.62393230e-01 1.21006441e+00 1.00829804e+00
1.25234306e-01 3.74707818e-01 7.06371784e-01 2.99245924e-01
5.26492655e-01 -2.62287349e-01 6.83386505e-01 8.23159397e-01
2.17799693e-01 -2.77509600e-01 -4.13765341e-01 7.65538394e-01
-1.84997725e+00 -5.51478684e-01 -2.86954284e-01 2.47852683e+00
6.55448675e-01 2.85457987e-02 1.54708043e-01 -1.55439436e-01
7.97936380e-01 -9.72000286e-02 -9.57661569e-01 -6.38031125e-01
1.46872625e-01 -1.89008236e-01 1.13062382e+00 7.91730106e-01
-1.31470132e+00 4.73839879e-01 5.08637524e+00 9.72125232e-01
-1.23713529e+00 -3.73250544e-01 5.83685795e-03 -1.24550305e-01
-1.29033789e-01 -4.09869075e-01 -1.06421936e+00 5.62007427e-01
1.02386212e+00 -8.40470135e-01 5.79761386e-01 1.17354345e+00
9.09727752e-01 2.61664122e-01 -5.74707150e-01 6.05697811e-01
-2.13284209e-01 -8.37390006e-01 -9.10352245e-02 -2.50818823e-02
1.18290627e+00 -3.14604551e-01 5.67693472e-01 6.73317790e-01
4.19687927e-01 -6.43287122e-01 7.15077341e-01 1.12635922e+00
8.02802742e-01 -1.25527716e+00 6.36745095e-01 6.66315436e-01
-1.16848910e+00 -7.12125063e-01 -7.30733037e-01 -2.25923181e-01
1.87800899e-01 5.98000705e-01 -6.92244709e-01 1.02652609e+00
-2.62986392e-01 1.05704141e+00 2.25576729e-01 1.13458383e+00
-4.04041022e-01 4.62233871e-01 -4.23790425e-01 -7.20309794e-01
4.01595324e-01 -5.99747598e-01 1.42832780e+00 5.64232588e-01
5.98474622e-01 -1.45788282e-01 5.81831694e-01 7.44968593e-01
4.18159246e-01 9.89459231e-02 -8.95762503e-01 3.48655060e-02
6.22083247e-01 1.02025759e+00 2.16309994e-01 -1.86235532e-01
3.64901349e-02 8.01960900e-02 -1.56969726e-01 2.97153860e-01
-1.09188879e+00 -7.25885510e-01 1.21606088e+00 -1.05760105e-01
3.80002022e-01 -5.17326236e-01 2.23433953e-02 -8.62361372e-01
-3.04323453e-02 -7.12208211e-01 -1.75067559e-01 -2.43871853e-01
-1.33671129e+00 3.27573240e-01 4.32304777e-02 -1.94614792e+00
-7.48382449e-01 -7.44858801e-01 -5.48265755e-01 8.29130173e-01
-1.62050962e+00 -8.39471579e-01 2.51346081e-01 1.90628290e-01
6.99972630e-01 -3.59466612e-01 5.38092494e-01 -7.30957016e-02
-6.96979523e-01 3.09266835e-01 1.09036183e+00 -5.57731926e-01
8.05334091e-01 -1.64824820e+00 -1.02614380e-01 7.94480026e-01
-8.37530553e-01 4.71688539e-01 1.20193291e+00 -6.93870425e-01
-1.90451252e+00 -1.74961674e+00 9.39174648e-03 -2.91531384e-01
1.13997018e+00 -9.46590304e-02 -4.53776777e-01 3.16712260e-01
-2.33534444e-02 7.68579021e-02 -2.95948088e-01 -5.21979511e-01
5.32897770e-01 -3.71549070e-01 -1.00973332e+00 5.65455317e-01
5.22553265e-01 8.53968114e-02 -4.08929259e-01 2.52579749e-01
9.37895358e-01 -8.74947608e-01 -1.11863446e+00 8.33057463e-01
4.03178126e-01 -1.91607341e-01 6.88157320e-01 -6.77848399e-01
8.41150135e-02 -6.95498109e-01 1.23832725e-01 -2.21235800e+00
-6.33559003e-02 -1.36947203e+00 -4.86261249e-01 8.84680927e-01
1.41917393e-01 -8.44962001e-01 4.24884051e-01 -2.47411191e-01
-5.34696221e-01 -1.20015097e+00 -9.36204553e-01 -1.33908808e+00
6.42391026e-01 -6.64710402e-02 2.14996845e-01 6.63080752e-01
3.34004089e-02 2.30434537e-01 -4.90311831e-01 9.68457699e-01
6.98511302e-01 -1.77298814e-01 9.61768866e-01 -1.08419096e+00
-8.11554492e-02 -5.38616359e-01 -4.31479560e-03 -1.19060016e+00
3.60385597e-01 -3.31466466e-01 4.09364611e-01 -1.24045062e+00
-1.11979663e+00 -1.84790716e-01 1.24739401e-01 -3.87707829e-01
-3.11632991e-01 -5.42158842e-01 -5.68362437e-02 -2.84516633e-01
-1.94034144e-01 1.30197322e+00 1.54681921e+00 -1.88647538e-01
-4.32944387e-01 7.63502598e-01 -1.99791327e-01 7.12930739e-01
8.65158558e-01 -8.39234702e-03 -1.18420577e+00 -6.95016906e-02
-2.13847280e-01 4.78537172e-01 4.75937538e-02 -1.40073490e+00
3.44783485e-01 -5.00750959e-01 7.98797756e-02 -7.57593453e-01
2.81997882e-02 -8.96620691e-01 -1.08559653e-02 7.02337623e-01
-1.24559239e-01 1.06985733e-01 6.07031465e-01 1.01719046e+00
-1.39892370e-01 2.20013633e-01 1.02651596e+00 4.49380368e-01
-1.15789630e-01 4.66450781e-01 -7.01481283e-01 1.01864770e-01
1.25680327e+00 2.81161726e-01 -3.31198871e-01 -5.86176515e-01
-5.90312481e-01 7.41097152e-01 2.39279464e-01 2.61186272e-01
6.13386750e-01 -1.09619021e+00 -5.87982714e-01 2.51490682e-01
-4.64790881e-01 -3.39375623e-02 2.17471808e-01 8.51453006e-01
-1.34280652e-01 8.37427497e-01 1.94264948e-02 -5.01972377e-01
-7.74956942e-01 7.70720184e-01 6.30697846e-01 -2.54299995e-02
-1.86977580e-01 5.25906622e-01 6.10623248e-02 -5.31855166e-01
1.65743455e-02 -8.97485137e-01 -3.78502868e-02 -1.07428104e-01
1.70935154e-01 4.94942367e-01 -1.99606344e-01 -2.64822245e-01
-9.73645889e-04 7.61474729e-01 4.29457426e-01 1.78904206e-01
9.66098368e-01 -2.79151827e-01 5.37451208e-01 5.36266744e-01
6.55477047e-01 7.20009208e-02 -1.94194138e+00 3.17880183e-01
-2.99216807e-01 -1.34785056e-01 3.69631909e-02 -7.40573585e-01
-6.37011349e-01 4.57932681e-01 2.79569060e-01 3.63297820e-01
9.26913679e-01 -9.17891264e-01 4.83009160e-01 3.51689279e-01
3.88109058e-01 -1.19974792e+00 -4.26416725e-01 1.12461698e+00
1.09526062e+00 -7.54488349e-01 1.02699965e-01 -2.63381809e-01
-7.08701909e-01 1.12771797e+00 9.16976988e-01 -9.10363972e-01
6.34601116e-01 3.00917864e-01 -1.63254917e-01 5.27963698e-01
-1.20163059e+00 -1.54295057e-01 4.44324434e-01 6.98107898e-01
-6.75529242e-02 1.95540190e-01 -6.32263303e-01 5.61095357e-01
-1.57624766e-01 -4.93129224e-01 6.25825763e-01 6.32384062e-01
-6.79000080e-01 -6.75663292e-01 -4.77541357e-01 2.20079809e-01
-1.30713612e-01 3.50397229e-01 2.50689209e-01 1.26975703e+00
5.94991371e-02 9.58414078e-01 2.19500847e-02 -2.48389855e-01
8.89077544e-01 -5.01838803e-01 3.91077906e-01 -6.00506008e-01
-2.61833489e-01 2.55285472e-01 4.37843092e-02 -2.42002502e-01
8.11416209e-02 -6.03254378e-01 -1.09423101e+00 -5.91222234e-02
-7.04739153e-01 4.80106086e-01 5.64602911e-01 7.21209526e-01
2.86740094e-01 8.00581217e-01 1.07079482e+00 -7.33331203e-01
-1.64925098e+00 -9.57363307e-01 -6.80273294e-01 -3.30110639e-01
8.51974964e-01 -1.36661422e+00 -7.86141932e-01 -2.38054439e-01] | [5.026538848876953, 2.349423408508301] |
cbcd6280-7901-4d13-a4a8-c057fb1b0857 | obeying-the-order-introducing-ordered | 2306.16916 | null | https://arxiv.org/abs/2306.16916v1 | https://arxiv.org/pdf/2306.16916v1.pdf | Obeying the Order: Introducing Ordered Transfer Hyperparameter Optimisation | We introduce ordered transfer hyperparameter optimisation (OTHPO), a version of transfer learning for hyperparameter optimisation (HPO) where the tasks follow a sequential order. Unlike for state-of-the-art transfer HPO, the assumption is that each task is most correlated to those immediately before it. This matches many deployed settings, where hyperparameters are retuned as more data is collected; for instance tuning a sequence of movie recommendation systems as more movies and ratings are added. We propose a formal definition, outline the differences to related problems and propose a basic OTHPO method that outperforms state-of-the-art transfer HPO. We empirically show the importance of taking order into account using ten benchmarks. The benchmarks are in the setting of gradually accumulating data, and span XGBoost, random forest, approximate k-nearest neighbor, elastic net, support vector machines and a separate real-world motivated optimisation problem. We open source the benchmarks to foster future research on ordered transfer HPO. | ['Aaron Klein', 'David Salinas', 'François-Xavier Aubet', 'Huibin Shen', 'Sigrid Passano Hellan'] | 2023-06-29 | null | null | null | null | ['transfer-learning', 'movie-recommendation'] | ['miscellaneous', 'miscellaneous'] | [ 9.94723067e-02 9.43047851e-02 -5.89528024e-01 -5.40001810e-01
-9.63882625e-01 -5.15939474e-01 5.19648194e-01 6.87145516e-02
-7.18132973e-01 9.88421202e-01 4.21607256e-01 -2.25978062e-01
-1.04344106e+00 -6.89678907e-01 -9.29820716e-01 -9.39021051e-01
-5.10028660e-01 1.26636302e+00 1.84672356e-01 -4.39237386e-01
4.05113101e-01 1.86541630e-03 -1.35485852e+00 5.84878266e-01
3.16894293e-01 1.02593637e+00 -8.74342620e-02 8.49533856e-01
2.73607522e-01 4.06892568e-01 -5.25544703e-01 -6.13056779e-01
6.75584316e-01 -5.49604371e-02 -1.13867819e+00 -1.30803227e-01
3.88355285e-01 -1.52157441e-01 1.88673928e-01 7.11571693e-01
9.73870933e-01 7.57038832e-01 9.19078350e-01 -1.64599705e+00
-6.84744835e-01 9.64443862e-01 -2.87014395e-01 1.71605662e-01
2.33726233e-01 1.97884843e-01 1.33957529e+00 -4.79031652e-01
4.30989385e-01 1.14534867e+00 8.21712077e-01 3.89372349e-01
-1.17987573e+00 -4.28306401e-01 2.35126510e-01 2.31864810e-01
-5.59953034e-01 -1.45075312e-02 1.65514529e-01 -5.05269468e-01
1.28659225e+00 3.65415990e-01 5.89455426e-01 1.07589948e+00
1.28965810e-01 8.06192935e-01 8.96135092e-01 -5.97982764e-01
4.64821190e-01 3.56811702e-01 6.05389103e-02 1.17710933e-01
-1.96206793e-01 5.28574772e-02 -6.70039058e-01 -5.29753268e-01
5.29074967e-01 -3.16594690e-01 1.11123845e-02 -6.39348149e-01
-1.46541476e+00 9.92390573e-01 4.80974764e-01 -1.31199405e-01
-5.35855353e-01 5.41159362e-02 6.03049755e-01 8.53929162e-01
6.92774653e-01 8.65924716e-01 -1.03218675e+00 -2.49341294e-01
-6.99210882e-01 5.68256199e-01 1.11626852e+00 1.06514478e+00
6.49816930e-01 -5.19754171e-01 -2.90614903e-01 1.12265325e+00
1.09661147e-01 -1.63497254e-01 8.27245474e-01 -1.12974381e+00
5.88096678e-01 3.82950120e-02 3.19616050e-01 -1.73273563e-01
-3.55932087e-01 -4.23098713e-01 -4.12144482e-01 5.93178086e-02
4.05473918e-01 -2.52384633e-01 -6.91090107e-01 1.32507837e+00
5.63345671e-01 1.24207973e-01 -1.82910874e-01 9.55789924e-01
4.58990455e-01 7.55169928e-01 1.71373188e-01 -1.95650771e-01
1.22815299e+00 -1.59492242e+00 -2.86717117e-01 1.52775245e-02
6.02760732e-01 -7.64987588e-01 1.24101555e+00 7.46991038e-01
-1.12223601e+00 -3.97798747e-01 -9.24404204e-01 9.36537907e-02
-4.41751122e-01 -4.99160856e-01 8.69423807e-01 4.76593494e-01
-1.17633510e+00 1.07009828e+00 -4.70646143e-01 -4.70076382e-01
2.22009987e-01 8.35026622e-01 -1.42206579e-01 1.63090035e-01
-1.42967379e+00 7.38275528e-01 5.43780744e-01 -2.42671415e-01
-6.68900490e-01 -1.15346336e+00 -1.34284899e-01 3.56322937e-02
4.05017585e-01 -1.12241352e+00 1.87119436e+00 -8.78327966e-01
-1.88277233e+00 6.81806564e-01 5.73840559e-01 -3.55189383e-01
6.84490621e-01 -6.40982509e-01 -1.14070855e-01 -3.14461470e-01
-1.11502558e-01 5.82070589e-01 8.10805619e-01 -8.82903516e-01
-8.57439041e-01 -2.59801447e-01 1.23817198e-01 6.94402397e-01
-4.76945728e-01 3.34531277e-01 -3.68678868e-01 -6.07604623e-01
-5.19069850e-01 -1.11086082e+00 -3.94495249e-01 -4.91472811e-01
-1.90857157e-01 -5.02129853e-01 1.10288218e-01 -1.66925311e-01
1.21445692e+00 -1.66371822e+00 6.54981852e-01 2.06617057e-01
-4.07332852e-02 6.57245815e-02 -2.58566886e-01 6.49410129e-01
-8.75651091e-02 1.48470536e-01 9.93744507e-02 -2.65172303e-01
3.71763557e-01 1.33237600e-01 -3.15185726e-01 3.27758044e-01
-4.25258756e-01 8.90288413e-01 -1.03896415e+00 -4.97017205e-01
-1.21212646e-01 4.10798863e-02 -6.89581513e-01 1.53742358e-01
-4.67053771e-01 3.71034056e-01 -7.03302264e-01 4.29263264e-01
1.49341196e-01 -4.88025993e-01 -8.28870311e-02 5.40427230e-02
1.23746715e-01 3.22710097e-01 -1.05023968e+00 1.38631892e+00
-5.27026296e-01 2.66478539e-01 -3.36050749e-01 -8.47756624e-01
6.73529863e-01 4.97611225e-01 8.16759169e-01 -3.95183027e-01
-1.62599366e-02 5.17443009e-02 -1.17442474e-01 -4.90743726e-01
5.21731436e-01 1.35285258e-02 -7.62191880e-03 6.90154850e-01
1.91416323e-01 -1.80113733e-01 1.21028386e-02 1.57270245e-02
1.04409564e+00 2.89702475e-01 5.75094044e-01 -2.62389868e-01
1.38225809e-01 1.83551442e-02 2.32574508e-01 1.04373801e+00
-1.18810877e-01 4.81207311e-01 3.78704876e-01 -7.36974716e-01
-1.20884216e+00 -9.33294535e-01 -2.90265977e-01 2.38583350e+00
-3.79619211e-01 -2.86561549e-01 -8.95494878e-01 -7.78861344e-01
2.73747325e-01 5.23340642e-01 -1.04905021e+00 1.39158428e-01
-5.37134707e-01 -1.08368719e+00 3.83056924e-02 4.15836722e-01
8.43482018e-02 -1.59958386e+00 -3.08394283e-01 2.81775385e-01
2.18348904e-03 -6.53891444e-01 -6.05675220e-01 3.03061903e-01
-1.15795195e+00 -6.78403795e-01 -9.14655864e-01 -6.66191339e-01
2.57375747e-01 -3.14641744e-01 1.58872652e+00 -2.07306504e-01
-3.94507572e-02 4.76887703e-01 -5.63755631e-01 -4.88058746e-01
-5.33911884e-02 7.17444003e-01 9.58884954e-02 -1.36088729e-01
4.32032943e-01 -5.86031556e-01 -7.65875101e-01 7.58799434e-01
-6.68837011e-01 -3.61506462e-01 6.25652075e-01 8.90390992e-01
5.96763313e-01 -3.88485223e-01 6.36701584e-01 -1.55311561e+00
9.50677812e-01 -9.81459320e-01 -5.03828704e-01 4.45780605e-01
-1.01626050e+00 -8.89727622e-02 5.18438995e-01 -8.04017663e-01
-7.71377385e-01 -2.12321386e-01 3.25247854e-01 -2.87854820e-01
-2.38494799e-02 2.81592369e-01 7.78393298e-02 2.66880784e-02
9.86151099e-01 -3.48393857e-01 -3.45324278e-01 -4.86388892e-01
4.31777626e-01 7.94104338e-01 2.95975685e-01 -9.57830250e-01
4.71409440e-01 6.57301024e-02 -2.36596137e-01 -3.52937967e-01
-1.20559514e+00 -8.38585615e-01 -7.61808813e-01 -1.98019221e-01
6.55105770e-01 -5.71045876e-01 -8.96886826e-01 2.01718912e-01
-6.58085465e-01 -1.03124535e+00 -5.61856866e-01 5.80097139e-01
-1.21995938e+00 -1.83993340e-01 -6.35062158e-01 -3.51737499e-01
-6.37972951e-01 -1.10081863e+00 1.07303596e+00 5.66366985e-02
-1.87261850e-01 -1.24650264e+00 6.35285378e-01 5.08324921e-01
4.60281134e-01 -8.11391920e-02 8.31771910e-01 -1.06428897e+00
-1.74714506e-01 2.43836995e-02 2.50645876e-01 1.35573328e-01
-2.83378601e-01 -1.74399659e-01 -8.23481202e-01 -5.30698359e-01
-2.17220157e-01 -6.08442724e-01 6.17738605e-01 5.65050602e-01
1.23026025e+00 -4.70528096e-01 -1.98741704e-01 8.64821255e-01
1.07442534e+00 -1.55337816e-02 4.81471092e-01 1.15098131e+00
4.71915185e-01 7.38555431e-01 9.57532167e-01 6.14390254e-01
3.99757206e-01 9.44690406e-01 3.28008533e-01 8.53421241e-02
3.18640769e-01 3.18945199e-02 1.70676544e-01 8.79984379e-01
-3.15608501e-01 -3.36918384e-01 -9.03743923e-01 4.95893806e-01
-2.26950693e+00 -6.79373622e-01 1.33690193e-01 2.47725272e+00
9.59722757e-01 1.33352190e-01 6.64045990e-01 -1.01530939e-01
7.44883716e-01 -1.17800407e-01 -6.40128911e-01 -5.73325157e-01
3.14596832e-01 1.78709731e-01 7.44986475e-01 5.18379629e-01
-1.29508841e+00 9.07572806e-01 6.79051113e+00 7.29991615e-01
-7.17896342e-01 2.90019304e-01 7.29839325e-01 -7.16528237e-01
-4.83977795e-02 -1.77495584e-01 -8.24814677e-01 3.93871635e-01
1.11813545e+00 -6.87307939e-02 9.57741141e-01 8.94324839e-01
-3.23072374e-01 1.45254001e-01 -1.48922598e+00 4.97933060e-01
-1.22699887e-01 -7.16681421e-01 -2.84205616e-01 2.07125962e-01
1.06399405e+00 4.71021712e-01 4.44730252e-01 6.61347330e-01
1.05014110e+00 -9.34652567e-01 3.29938680e-01 3.21928024e-01
4.42004412e-01 -6.42891586e-01 5.57320237e-01 1.40270010e-01
-5.32322407e-01 -4.09132093e-01 -8.14065576e-01 1.45894527e-01
1.47502989e-01 4.71867681e-01 -9.81235623e-01 5.22714257e-01
1.26926219e+00 5.03158569e-01 -5.53517163e-01 1.37809336e+00
-7.58504421e-02 8.88977885e-01 -2.25191504e-01 -2.83422738e-01
4.65163887e-01 -1.41152292e-01 5.63326240e-01 1.10305882e+00
5.49578369e-02 -8.13564751e-03 1.82880566e-01 -5.66551872e-02
-9.28366557e-02 4.98450696e-01 -1.54295191e-01 5.70481420e-01
4.17211115e-01 1.35335481e+00 -5.39718986e-01 -1.88214764e-01
-2.01865658e-01 7.46267676e-01 5.22978544e-01 4.37148482e-01
-8.02385747e-01 -2.15750635e-01 4.29491252e-01 5.08010574e-02
5.34224212e-01 3.79214883e-01 -9.60399732e-02 -8.29005361e-01
-2.71880299e-01 -9.27881837e-01 8.50559056e-01 -6.82469428e-01
-1.86460853e+00 4.86248821e-01 3.95339727e-01 -1.27913177e+00
-3.99624914e-01 -5.87594151e-01 -5.03862798e-01 5.77596784e-01
-1.53104103e+00 -8.64310443e-01 -9.91028100e-02 4.67362642e-01
6.66671753e-01 -2.71692306e-01 7.72065580e-01 2.85530895e-01
-2.87630737e-01 7.23435283e-01 5.74863970e-01 -3.47850114e-01
1.41805243e+00 -1.75042760e+00 5.88304520e-01 -2.97917217e-01
-6.78098649e-02 4.12166536e-01 7.20176876e-01 -4.49976265e-01
-1.03297865e+00 -1.09462798e+00 6.00771666e-01 -7.22992241e-01
8.20052564e-01 -3.72317940e-01 -7.90164292e-01 9.73790824e-01
1.78312704e-01 2.24355757e-01 9.47400570e-01 7.75931835e-01
-2.22358897e-01 -2.85772830e-01 -1.01731002e+00 3.19204330e-01
9.84646201e-01 -7.57503286e-02 -6.18319392e-01 9.21576738e-01
8.33183706e-01 -4.23651040e-01 -1.46624541e+00 3.70275259e-01
8.27142596e-01 -8.43163252e-01 1.20060134e+00 -1.30703282e+00
4.59981143e-01 2.73390889e-01 -7.84006417e-02 -1.83100903e+00
-6.97279513e-01 -1.08822334e+00 -8.81333724e-02 1.06102800e+00
6.71645582e-01 -7.52062857e-01 9.33982790e-01 5.95028102e-01
2.58521680e-02 -1.00298095e+00 -7.97491968e-01 -8.06772470e-01
4.17571068e-01 -9.67297852e-02 1.01606214e+00 1.13751268e+00
1.89972490e-01 4.54077899e-01 -4.78711098e-01 -1.98709384e-01
6.25790298e-01 -1.33278266e-01 8.90551090e-01 -1.37472534e+00
-1.08395565e+00 -4.07137454e-01 -1.24285974e-01 -7.83958256e-01
6.34703832e-03 -7.25032330e-01 5.69603741e-02 -1.14478827e+00
1.17107488e-01 -7.81218946e-01 -8.28737915e-01 5.05470634e-01
-3.74769658e-01 2.81974077e-01 1.02981791e-01 5.01581371e-01
-1.02838326e+00 4.25198108e-01 1.27141523e+00 2.47304231e-01
-3.87790233e-01 5.22040725e-01 -4.63804960e-01 4.25504416e-01
9.66753185e-01 -6.85834587e-01 -4.71619695e-01 -5.04021645e-01
6.68972969e-01 1.18507609e-01 -2.22154483e-01 -5.58933794e-01
4.97170091e-01 -1.83933765e-01 2.75500923e-01 -1.31865174e-01
1.43727794e-01 -6.79883778e-01 2.75072418e-02 3.61145325e-02
-7.86710083e-01 2.98577011e-01 -7.83516094e-02 4.25804764e-01
1.56158656e-01 -4.58883494e-01 5.79296410e-01 -1.55071706e-01
-2.91692436e-01 3.64556819e-01 -3.32843885e-02 2.65253752e-01
1.12675452e+00 6.38951808e-02 -4.85295415e-01 -5.21176457e-01
-8.91763091e-01 5.63959360e-01 1.63634226e-01 4.95323569e-01
5.17846830e-02 -1.20986807e+00 -9.26933825e-01 -2.39461660e-01
1.85915887e-01 -4.78791632e-02 1.23309111e-02 7.47246563e-01
-1.80417672e-01 4.39172685e-01 -3.24720085e-01 -4.05499607e-01
-9.78815377e-01 8.08192432e-01 1.81879655e-01 -6.88112915e-01
-2.81746507e-01 9.52332675e-01 4.27857377e-02 -1.09901452e+00
5.46416342e-01 1.33902999e-02 -3.26376051e-01 8.44517946e-02
2.87760019e-01 8.05128455e-01 4.51205462e-01 1.03345260e-01
2.11012140e-01 5.11292756e-01 -3.54966938e-01 -2.54896522e-01
1.63422561e+00 9.69763994e-02 -8.77751112e-02 5.93658507e-01
1.04261875e+00 -4.59510088e-01 -1.28594971e+00 -5.43496430e-01
1.05139390e-01 -3.46761018e-01 1.51961781e-02 -1.04321849e+00
-7.31101036e-01 5.59608102e-01 2.81865597e-01 3.60539407e-01
1.13701808e+00 1.20843239e-02 4.86870795e-01 6.77492321e-01
3.00075650e-01 -1.21865714e+00 4.96029966e-02 4.08964157e-01
8.74053359e-01 -1.16666782e+00 4.40594912e-01 -1.35913372e-01
-8.26910138e-01 9.87118602e-01 6.91226423e-01 -2.21632704e-01
8.92808259e-01 8.10272768e-02 -8.33183601e-02 -1.81514904e-01
-1.23048806e+00 1.54657230e-01 4.69867110e-01 5.07284760e-01
2.97008276e-01 1.44070908e-01 -3.12282536e-02 4.98932183e-01
-5.29453814e-01 9.60089043e-02 5.35052903e-02 6.88140869e-01
-2.87182987e-01 -1.32403052e+00 -3.14298958e-01 7.27049112e-01
-6.80028379e-01 -2.22710326e-01 -2.91789949e-01 7.71524966e-01
-2.31915101e-01 5.64437807e-01 1.74826637e-01 -2.84534097e-01
3.56219053e-01 7.69479126e-02 5.06449342e-01 -6.26574636e-01
-1.27517509e+00 2.39654277e-02 1.10116623e-01 -5.45840919e-01
-1.87584579e-01 -7.59659290e-01 -5.35100102e-01 -6.10851645e-02
-3.19545388e-01 4.24540013e-01 7.92433262e-01 7.31064677e-01
3.03787947e-01 4.52421188e-01 8.16602707e-01 -1.12454021e+00
-1.11043513e+00 -1.27587295e+00 -4.96712595e-01 3.07054609e-01
2.37810031e-01 -6.06856346e-01 -3.26586634e-01 6.25012219e-02] | [9.22208309173584, 4.069727420806885] |
0fa80836-f665-4ad2-adee-1c1d3cd0ec5f | m2-ctts-end-to-end-multi-scale-multi-modal | 2305.02269 | null | https://arxiv.org/abs/2305.02269v1 | https://arxiv.org/pdf/2305.02269v1.pdf | M2-CTTS: End-to-End Multi-scale Multi-modal Conversational Text-to-Speech Synthesis | Conversational text-to-speech (TTS) aims to synthesize speech with proper prosody of reply based on the historical conversation. However, it is still a challenge to comprehensively model the conversation, and a majority of conversational TTS systems only focus on extracting global information and omit local prosody features, which contain important fine-grained information like keywords and emphasis. Moreover, it is insufficient to only consider the textual features, and acoustic features also contain various prosody information. Hence, we propose M2-CTTS, an end-to-end multi-scale multi-modal conversational text-to-speech system, aiming to comprehensively utilize historical conversation and enhance prosodic expression. More specifically, we design a textual context module and an acoustic context module with both coarse-grained and fine-grained modeling. Experimental results demonstrate that our model mixed with fine-grained context information and additionally considering acoustic features achieves better prosody performance and naturalness in CMOS tests. | ['Jiaen Liang', 'Jianqing Sun', 'JianHua Tao', 'Yingming Gao', 'Ya Li', 'Fengping Wang', 'Yayue Deng', 'Jinlong Xue'] | 2023-05-03 | null | null | null | null | ['text-to-speech-synthesis', 'speech-synthesis'] | ['speech', 'speech'] | [-9.69037041e-02 -1.92649230e-01 5.75937033e-02 -6.68113470e-01
-1.09790754e+00 -3.58803302e-01 5.26779711e-01 -3.33558500e-01
2.36629229e-02 5.56236386e-01 1.01246226e+00 -1.19449370e-01
3.07951421e-01 -3.61734331e-01 4.62182797e-02 -6.21506751e-01
3.47678483e-01 2.82694072e-01 1.78342223e-01 -7.36488581e-01
-2.74334420e-02 1.75717734e-02 -1.32852960e+00 7.24159896e-01
6.79136038e-01 8.63126636e-01 6.46279573e-01 7.51058221e-01
-6.33802235e-01 7.96588957e-01 -9.57046986e-01 -2.20059782e-01
-3.05545628e-01 -6.75410807e-01 -6.30867124e-01 -5.76187074e-02
-1.88670576e-01 -3.23659182e-01 -2.01945588e-01 6.20359719e-01
8.35450590e-01 1.54251099e-01 3.18310380e-01 -9.95934248e-01
-3.75167698e-01 1.22439539e+00 -1.83279887e-01 1.16066329e-01
5.55234969e-01 3.50818187e-01 1.29149413e+00 -8.40573728e-01
1.82679832e-01 1.59186196e+00 4.69980121e-01 7.43407607e-01
-8.04055333e-01 -5.46112895e-01 3.56561959e-01 1.34753749e-01
-1.07875800e+00 -7.50211656e-01 1.16031790e+00 -9.05458033e-02
8.69613707e-01 5.96047640e-01 4.80694324e-01 1.38405490e+00
-2.19187941e-02 9.86209214e-01 1.04691935e+00 -4.26317543e-01
-5.53405136e-02 9.44256186e-02 2.01425955e-01 4.33897488e-02
-1.08378172e+00 -4.73152027e-02 -8.63558054e-01 -6.70080706e-02
2.00284854e-01 -2.11417183e-01 -3.60567451e-01 5.55025458e-01
-1.18725371e+00 6.24158442e-01 -1.13214001e-01 6.73638821e-01
-4.03484732e-01 -2.00074568e-01 6.58655584e-01 2.65906870e-01
2.38889471e-01 6.69810548e-02 -4.96794283e-01 -7.01749384e-01
-9.72132385e-01 1.97060302e-01 9.00060117e-01 1.14571655e+00
6.19414449e-01 4.30283993e-01 -3.43277067e-01 1.46301901e+00
2.17331693e-01 7.08938360e-01 7.59826183e-01 -8.66070628e-01
5.06868899e-01 2.73483425e-01 -1.44741625e-01 -6.06846094e-01
-4.45037127e-01 -2.98473686e-01 -7.01866925e-01 -5.47293127e-01
1.45080816e-02 -4.69537765e-01 -3.54323685e-01 1.97788250e+00
1.79808885e-01 2.21097153e-02 2.75259912e-01 8.93587530e-01
1.12485945e+00 1.05790341e+00 6.10530190e-03 -7.65339613e-01
1.57353008e+00 -1.20889175e+00 -1.34656250e+00 -2.92460501e-01
2.50576884e-01 -1.13822699e+00 1.61936402e+00 1.41806677e-01
-1.33240306e+00 -7.49864101e-01 -9.59883332e-01 -1.79917559e-01
5.72391897e-02 6.40631095e-02 2.39588395e-01 8.40635717e-01
-8.52774382e-01 -9.20557976e-03 -4.72453952e-01 -2.52708755e-02
-5.56162238e-01 9.21531990e-02 3.56905460e-02 3.57120425e-01
-1.51585245e+00 6.26317143e-01 5.68377450e-02 1.79095820e-01
-4.83396322e-01 -5.52742422e-01 -8.24397326e-01 1.79062456e-01
3.71389449e-01 -4.26285475e-01 1.68086541e+00 -6.11981511e-01
-2.25874448e+00 1.85769796e-01 -6.33274198e-01 -2.15768129e-01
1.13309294e-01 1.14584826e-02 -7.08643615e-01 2.75072437e-02
-2.73887038e-01 4.35313702e-01 4.11292076e-01 -1.16667461e+00
-8.15815806e-01 -1.08363137e-01 -2.39183053e-01 4.67752934e-01
-6.01673663e-01 4.12749708e-01 -6.22960150e-01 -7.20431864e-01
-3.69887836e-02 -8.13269913e-01 9.09716263e-02 -7.08370745e-01
-4.11823094e-01 -1.68824404e-01 1.13792551e+00 -7.08446920e-01
1.75878310e+00 -2.11334133e+00 4.24028076e-02 -1.32187471e-01
-1.44823536e-01 7.61241391e-02 -4.72856462e-02 6.34097815e-01
3.50327700e-01 -3.30454186e-02 -1.32375255e-01 -7.80767620e-01
2.63801277e-01 1.55760974e-01 -5.22301078e-01 -2.47150019e-01
1.15901176e-02 9.28954601e-01 -5.64661860e-01 -5.46194255e-01
4.89551336e-01 4.47924793e-01 -5.75892329e-01 4.46979970e-01
-3.11931908e-01 6.43549383e-01 -4.89705086e-01 4.68028486e-01
4.28528100e-01 3.25877190e-01 2.75919795e-01 -3.10095370e-01
-3.93149406e-01 8.63021791e-01 -9.74192500e-01 1.59191716e+00
-1.03000164e+00 3.43108892e-01 5.34026027e-01 -5.12015283e-01
1.05154467e+00 8.14005733e-01 5.14049828e-01 -5.54588377e-01
2.36886248e-01 2.23364264e-01 1.30541936e-01 -4.67544526e-01
8.27258110e-01 -4.97272819e-01 -5.82374811e-01 3.46527189e-01
-6.82076765e-03 -4.40314680e-01 -3.02139789e-01 -7.91054517e-02
8.13635588e-01 -2.73663849e-01 2.21808434e-01 -1.71003193e-01
7.97053337e-01 -3.43698412e-01 6.89129949e-01 2.57907361e-01
-3.58160228e-01 8.10927749e-01 1.98521137e-01 1.55830786e-01
-8.00578415e-01 -8.20120871e-01 1.39537901e-01 1.41027164e+00
1.06571414e-01 -5.41648328e-01 -9.83912587e-01 -1.77128538e-01
-5.92565596e-01 9.23767269e-01 -8.46745893e-02 6.77095428e-02
-8.71219635e-01 -6.17954373e-01 7.80071199e-01 2.16952384e-01
5.44205606e-01 -1.46519995e+00 2.17095241e-01 6.88251913e-01
-8.67291212e-01 -1.38178754e+00 -1.30467975e+00 -1.01070479e-01
-3.73265117e-01 -3.85028064e-01 -5.48323691e-01 -1.04622746e+00
-3.72424990e-01 2.12373674e-01 7.79479206e-01 -2.67515659e-01
2.16136292e-01 5.55738211e-02 -5.40703058e-01 -1.38453007e-01
-9.07643557e-01 9.53820720e-02 -5.56603745e-02 1.99855760e-01
3.43484491e-01 -8.51418853e-01 -3.09801638e-01 6.33214235e-01
-5.17471075e-01 1.87632471e-01 3.68740708e-01 8.90832722e-01
1.93456173e-01 -1.22821160e-01 1.19458747e+00 -5.98620653e-01
9.58844483e-01 -1.82603791e-01 8.59380141e-02 2.05276415e-01
1.09384609e-02 -3.16703975e-01 8.90631139e-01 -4.98009831e-01
-1.67760050e+00 -3.02643150e-01 -9.27468777e-01 -8.12350065e-02
-2.15723261e-01 5.58563590e-01 -8.83154094e-01 5.05218685e-01
3.50193530e-01 5.13045669e-01 -1.79111838e-01 -4.69852656e-01
3.13650042e-01 1.51705992e+00 6.61517560e-01 -9.15582180e-01
2.72015989e-01 -1.33199506e-02 -6.30081832e-01 -1.26290524e+00
-5.25974154e-01 -6.88694000e-01 -2.80458093e-01 -2.65666634e-01
7.47470558e-01 -9.31819022e-01 -1.09355438e+00 6.27228618e-01
-1.25125158e+00 -2.71676332e-01 3.00597455e-02 6.12370372e-01
-8.60225081e-01 5.56895375e-01 -1.07271826e+00 -1.11047244e+00
-5.32974482e-01 -1.37112987e+00 1.16191041e+00 1.19889736e-01
-3.19269657e-01 -7.89755046e-01 -1.65041655e-01 7.73525119e-01
7.17605948e-01 -2.28255436e-01 8.71605814e-01 -5.01735032e-01
-1.93371564e-01 3.26168418e-01 1.44840494e-01 4.66702133e-01
4.26581264e-01 -7.03737587e-02 -1.35268295e+00 1.28035650e-01
3.83246541e-01 -2.03258559e-01 3.86011541e-01 4.25735325e-01
7.88914919e-01 -3.93805653e-01 1.04954012e-01 1.87028825e-01
7.04139650e-01 4.99588311e-01 5.16981006e-01 -3.09715211e-01
5.35288215e-01 9.72456396e-01 8.02091062e-01 5.15314937e-01
7.85204172e-01 9.30603206e-01 -1.61817178e-01 2.52291501e-01
-3.97108406e-01 -1.82345718e-01 7.19043255e-01 2.01398945e+00
3.30651730e-01 -4.15145129e-01 -5.89191735e-01 7.10223615e-01
-1.79092145e+00 -9.46937084e-01 2.62375865e-02 1.73019981e+00
1.39727235e+00 2.60882646e-01 1.34450793e-01 -4.07237485e-02
1.01351249e+00 3.89194995e-01 -2.55016297e-01 -5.75881422e-01
-4.04715389e-01 -1.12421945e-01 -2.84198523e-01 8.39693308e-01
-6.89621985e-01 1.16229737e+00 5.74599028e+00 1.42767882e+00
-1.20956957e+00 1.30915001e-01 4.73299712e-01 -1.28189802e-01
-6.48058712e-01 -1.34671003e-01 -9.47555780e-01 5.15351951e-01
1.38164163e+00 -1.84388265e-01 5.92407525e-01 7.31275260e-01
7.50610650e-01 3.29760432e-01 -1.00770557e+00 9.50353205e-01
-1.62406221e-01 -1.18076193e+00 -1.22465141e-01 -3.23318899e-01
4.83731121e-01 -1.61334142e-01 -9.58693996e-02 7.45212078e-01
5.81518561e-02 -6.41321719e-01 8.75542641e-01 2.80992866e-01
5.27715623e-01 -7.75244355e-01 6.05138958e-01 5.03283739e-01
-1.62041032e+00 7.58870691e-02 1.92471370e-01 5.02052605e-02
7.82444894e-01 2.45441303e-01 -9.55339253e-01 5.19873917e-01
2.50214338e-01 1.92412645e-01 2.60746777e-01 5.50878525e-01
9.00524408e-02 8.92023921e-01 -2.52608746e-01 -3.26253027e-01
2.02172145e-01 5.93933053e-02 6.71907008e-01 1.64735901e+00
3.12484026e-01 3.01934659e-01 4.72366720e-01 4.92247254e-01
2.58687139e-02 4.98837382e-01 -2.48252191e-02 -2.92718019e-02
8.98891687e-01 1.04855645e+00 -1.52716860e-01 -2.27106243e-01
-5.17791927e-01 7.26701498e-01 -2.82368034e-01 2.67858773e-01
-7.71521211e-01 -2.45471492e-01 9.16620374e-01 -2.27873728e-01
4.28557545e-02 -3.70041877e-01 -2.64452100e-01 -1.02489626e+00
5.47237054e-04 -1.12095261e+00 3.75133567e-02 -8.14519763e-01
-1.31450129e+00 9.70043898e-01 -1.77042827e-01 -1.16137493e+00
-6.57690644e-01 -1.06014535e-01 -7.76112437e-01 1.12604976e+00
-1.51737893e+00 -1.52759802e+00 -1.74273074e-01 6.11842811e-01
1.31079757e+00 -1.81315824e-01 9.17723119e-01 4.81328756e-01
-4.53660578e-01 7.83933103e-01 -6.49798885e-02 1.78018734e-02
8.48014712e-01 -1.07413340e+00 3.45096350e-01 3.65254492e-01
-2.39538670e-01 7.66866863e-01 8.84694695e-01 -2.50266850e-01
-1.27083361e+00 -8.87464285e-01 9.73571122e-01 -2.77351923e-02
6.94373965e-01 -5.89488149e-01 -9.24445331e-01 4.25469935e-01
5.50498784e-01 -4.68195826e-01 8.18926811e-01 3.29313904e-01
3.43344286e-02 -2.32498676e-01 -9.45848882e-01 9.81878936e-01
7.77237594e-01 -9.48855281e-01 -7.65906930e-01 4.50822478e-03
1.43512118e+00 -4.22110409e-01 -9.83144343e-01 3.09454203e-01
5.60152650e-01 -9.65234876e-01 6.84947670e-01 -2.87774075e-02
4.34804708e-02 -2.58449882e-01 -6.21168852e-01 -1.33282804e+00
1.44450888e-01 -1.15710759e+00 2.63372838e-01 1.74313676e+00
5.17485738e-01 -4.37402666e-01 5.13546109e-01 3.72129828e-01
-8.11837196e-01 -4.54230398e-01 -8.30810010e-01 -5.54295421e-01
-6.31708577e-02 -7.28093386e-01 8.65628242e-01 7.50448823e-01
6.02489650e-01 7.79627264e-01 -7.60287166e-01 1.66748241e-01
1.05447292e-01 4.12195385e-01 7.99132824e-01 -8.38524759e-01
-4.49174941e-01 -6.28896892e-01 1.90840006e-01 -1.49775541e+00
2.15886667e-01 -2.55191684e-01 4.60645348e-01 -1.12587869e+00
-7.09094852e-02 -3.60188335e-01 -3.84862870e-02 2.18637764e-01
-2.10402563e-01 -1.93468839e-01 2.00286165e-01 3.99497747e-02
-3.23397666e-01 1.15840018e+00 1.33694077e+00 -1.32150754e-01
-6.08602166e-01 3.00897092e-01 -5.22228658e-01 5.98820686e-01
7.64567256e-01 -6.60706982e-02 -4.27490830e-01 -9.66720469e-03
-5.26448309e-01 9.34549332e-01 -1.61326483e-01 -7.82387376e-01
4.53570187e-01 -5.79109907e-01 -4.68157947e-01 -8.26910675e-01
9.96975183e-01 -5.86171091e-01 -5.52217141e-02 -4.28824686e-02
-4.80166107e-01 -3.64358395e-01 2.85906345e-01 3.86805981e-01
-7.58016944e-01 -2.06541289e-02 5.35561264e-01 1.11477666e-01
-5.55562317e-01 1.08674556e-01 -6.41372025e-01 3.72098722e-02
5.22783279e-01 6.35230541e-02 -3.13832670e-01 -7.71800160e-01
-8.95327747e-01 2.88446367e-01 -3.42144421e-03 6.64391279e-01
4.29401606e-01 -1.27562702e+00 -7.53936410e-01 1.36236638e-01
1.04739413e-01 -2.40703076e-01 8.49593282e-01 5.05304337e-01
-3.26787718e-02 7.24784732e-01 6.41052276e-02 -5.39993167e-01
-1.47793949e+00 2.10342124e-01 2.64608592e-01 -3.56498957e-02
-3.31087798e-01 6.32751107e-01 4.81537402e-01 -7.70133078e-01
3.09681475e-01 -5.83504736e-01 -2.84888476e-01 5.78291789e-02
5.68492889e-01 6.90267757e-02 7.26903528e-02 -8.64221811e-01
-2.91735053e-01 3.00814301e-01 9.08961892e-02 -5.83318472e-01
1.06812239e+00 -8.32946301e-01 4.97516654e-02 7.85398126e-01
1.11846960e+00 6.34364843e-01 -1.15087962e+00 -5.01989126e-01
-5.97158633e-02 -7.29207769e-02 2.63469853e-02 -8.72819424e-01
-6.66860461e-01 1.10372353e+00 -1.52755514e-01 3.88803482e-01
1.20056975e+00 -2.78855205e-01 1.45924449e+00 4.20444459e-01
3.53064686e-01 -1.32090974e+00 1.59030274e-01 8.85813296e-01
1.13876057e+00 -9.26934481e-01 -6.77826822e-01 -5.81102729e-01
-1.09061682e+00 1.03949213e+00 6.15679026e-01 5.76522171e-01
7.74720550e-01 6.10518396e-01 4.54525918e-01 3.16871345e-01
-1.02925265e+00 -1.19304679e-01 5.25593460e-02 3.84055972e-01
5.71448326e-01 2.71288037e-01 -2.22984329e-01 1.27666116e+00
-7.66009867e-01 -4.33851093e-01 4.73646015e-01 4.64338243e-01
-7.36741841e-01 -1.44853914e+00 -3.41106057e-01 -1.69759735e-01
-4.48976576e-01 -4.46061283e-01 -1.48832589e-01 3.61865282e-01
-1.73711672e-01 1.62242889e+00 -1.81583211e-01 -6.79175377e-01
4.46197838e-01 2.35027373e-01 -4.48545590e-02 -6.75849140e-01
-9.87064898e-01 7.69375920e-01 5.89372993e-01 -2.00790986e-01
-3.05191994e-01 -6.23065114e-01 -1.45453084e+00 -4.75539714e-01
-4.85911578e-01 4.42280322e-01 8.39740634e-01 1.12015724e+00
1.47368535e-01 8.21779013e-01 1.10202777e+00 -7.57241070e-01
-5.06122351e-01 -1.45474315e+00 -6.86306655e-01 -2.70514134e-02
4.64623243e-01 -1.58553287e-01 -3.94354820e-01 -4.94072661e-02] | [14.737101554870605, 6.749542236328125] |
d5b4ff23-c62e-4726-9c65-71fab6975689 | tode-trans-transparent-object-depth | 2209.08455 | null | https://arxiv.org/abs/2209.08455v1 | https://arxiv.org/pdf/2209.08455v1.pdf | TODE-Trans: Transparent Object Depth Estimation with Transformer | Transparent objects are widely used in industrial automation and daily life. However, robust visual recognition and perception of transparent objects have always been a major challenge. Currently, most commercial-grade depth cameras are still not good at sensing the surfaces of transparent objects due to the refraction and reflection of light. In this work, we present a transformer-based transparent object depth estimation approach from a single RGB-D input. We observe that the global characteristics of the transformer make it easier to extract contextual information to perform depth estimation of transparent areas. In addition, to better enhance the fine-grained features, a feature fusion module (FFM) is designed to assist coherent prediction. Our empirical evidence demonstrates that our model delivers significant improvements in recent popular datasets, e.g., 25% gain on RMSE and 21% gain on REL compared to previous state-of-the-art convolutional-based counterparts in ClearGrasp dataset. Extensive results show that our transformer-based model enables better aggregation of the object's RGB and inaccurate depth information to obtain a better depth representation. Our code and the pre-trained model will be available at https://github.com/yuchendoudou/TODE. | ['Bin Li', 'Zhen Kan', 'Dongxu Li', 'Beihao Xia', 'Shaochen Wang', 'Kang Chen'] | 2022-09-18 | null | null | null | null | ['transparent-objects', 'transparent-object-depth-estimation'] | ['computer-vision', 'computer-vision'] | [ 2.55374372e-01 -7.71505088e-02 1.86709598e-01 -3.96229893e-01
-4.76855040e-01 -2.53485620e-01 3.32119077e-01 -1.56773895e-01
1.66977748e-01 4.97174442e-01 -1.95347562e-01 -5.16323410e-02
1.67040512e-01 -8.29856932e-01 -4.92163658e-01 -9.22065198e-01
2.25706786e-01 2.11832318e-02 4.48928803e-01 3.27123702e-02
2.37453446e-01 4.21740711e-01 -1.53064668e+00 5.08505583e-01
1.00856555e+00 1.72493386e+00 7.47815788e-01 4.51631129e-01
2.71334443e-02 6.76116943e-01 -5.19308150e-01 -2.01881349e-01
5.21330535e-01 7.69252852e-02 -3.64409745e-01 2.83673227e-01
6.51748359e-01 -7.99649954e-01 -3.92898500e-01 8.37146342e-01
3.10092479e-01 -2.02234507e-01 6.05178595e-01 -7.63045192e-01
-6.27862275e-01 -4.13848050e-02 -6.53581500e-01 -6.41977191e-02
3.34100276e-01 4.06113476e-01 7.06865132e-01 -9.63947415e-01
2.24993840e-01 1.02051711e+00 4.37748462e-01 3.15126330e-01
-7.84656882e-01 -5.37275672e-01 2.10022047e-01 4.27544773e-01
-1.03594673e+00 -2.47812554e-01 8.74738812e-01 -3.25902462e-01
8.14680696e-01 2.80771941e-01 7.62552083e-01 9.78443444e-01
4.92275029e-01 6.39941633e-01 1.23728645e+00 -2.57073551e-01
7.36541450e-02 3.09084922e-01 -8.97005200e-02 7.90421546e-01
3.73865843e-01 2.14360476e-01 -6.32473052e-01 4.00837749e-01
1.04548717e+00 3.79816234e-01 -4.12719697e-01 -5.05539715e-01
-1.13291693e+00 3.93102020e-01 8.23129416e-01 -2.70783734e-02
-4.20510054e-01 8.48717988e-02 -9.85515043e-02 1.02583751e-01
6.93265200e-01 1.85400844e-01 -4.15491283e-01 -1.35492533e-01
-4.96253192e-01 -1.31274372e-01 5.11060119e-01 8.62255812e-01
9.95314896e-01 -7.09355110e-03 4.38302793e-02 7.49370873e-01
6.00129664e-01 7.05375850e-01 8.30781907e-02 -1.02239633e+00
6.86292112e-01 8.66003811e-01 2.07200065e-01 -6.86199427e-01
-2.69664586e-01 -4.36429739e-01 -7.11547971e-01 7.80856490e-01
4.59413290e-01 1.83913395e-01 -8.60806704e-01 8.82394493e-01
4.00866836e-01 -1.11505261e-03 1.19711142e-02 1.17996001e+00
8.93515468e-01 5.01722276e-01 -4.38621104e-01 1.81888565e-01
1.28435504e+00 -9.75624740e-01 -4.62894410e-01 -4.79702830e-01
2.71567762e-01 -8.06268334e-01 1.13282192e+00 7.23841429e-01
-8.60984325e-01 -6.00819588e-01 -1.24469519e+00 -2.38749042e-01
-1.84214130e-01 5.10757804e-01 7.56070077e-01 6.39852822e-01
-5.86382806e-01 6.22666955e-01 -9.94048178e-01 -1.64891183e-01
6.49420917e-01 2.42600933e-01 -2.71165580e-01 -3.60981524e-01
-5.57557046e-01 7.58040011e-01 3.38167921e-02 3.95444542e-01
-7.64352441e-01 -7.31518686e-01 -4.77834821e-01 -3.57380390e-01
5.38735032e-01 -5.09079158e-01 1.17971623e+00 -5.06728172e-01
-1.78055954e+00 6.05354905e-01 -1.69409171e-01 -3.26094419e-01
5.87039232e-01 -6.34168506e-01 -3.28903049e-02 4.57356870e-01
-3.70181262e-01 2.41973966e-01 8.40351105e-01 -1.56668305e+00
-6.02749169e-01 -6.46881938e-01 4.09699321e-01 1.89778134e-01
-4.38316077e-01 -4.43575799e-01 -3.66554499e-01 -3.78632665e-01
3.77873957e-01 -8.13382328e-01 -1.73349921e-02 6.46065712e-01
-4.43923622e-01 1.42631993e-01 1.00287223e+00 -5.63552618e-01
6.53394639e-01 -2.03925657e+00 -1.84581310e-01 -1.84753567e-01
6.17600918e-01 3.29100639e-01 2.36656189e-01 3.26634884e-01
2.61043012e-01 -2.93958813e-01 -1.11631401e-01 -5.89257181e-01
-8.68280977e-02 -5.31554930e-02 -1.19432583e-01 5.83521128e-01
1.10110864e-01 7.95181811e-01 -5.40309608e-01 -3.67928892e-01
8.65651131e-01 7.74204254e-01 -2.71689653e-01 2.25001246e-01
-1.39160231e-01 7.56300926e-01 -6.58643663e-01 1.00143421e+00
9.63450193e-01 -3.19474816e-01 -3.18855457e-02 -5.15697241e-01
-2.20601320e-01 3.53625327e-01 -8.93885791e-01 1.57158983e+00
-8.05828989e-01 7.05252349e-01 3.85967665e-03 -5.14680624e-01
1.03490210e+00 1.20757394e-01 5.44383883e-01 -9.25589621e-01
2.53228903e-01 2.66504586e-01 -2.57785857e-01 -3.80185306e-01
2.85645962e-01 1.64874524e-01 4.57621038e-01 3.15983072e-02
-4.79050577e-01 -6.30654871e-01 -2.76027650e-01 -2.09670067e-01
1.00205779e+00 5.24059296e-01 2.09769309e-02 4.18933928e-02
3.06240886e-01 -3.45515728e-01 5.25518179e-01 3.52373928e-01
-1.27963632e-01 9.55005765e-01 -1.13801189e-01 -5.74735284e-01
-7.02704847e-01 -1.25590742e+00 -2.98180848e-01 4.36793506e-01
6.70540690e-01 -1.08724587e-01 -4.83779818e-01 -3.88312548e-01
1.69376582e-01 3.96766454e-01 -3.81898195e-01 -1.67749509e-01
-3.63050699e-01 -4.44940865e-01 -2.87546217e-01 6.52942896e-01
1.11339724e+00 -7.05524325e-01 -1.17298639e+00 1.52890738e-02
-6.22914769e-02 -1.35661054e+00 9.53898504e-02 2.57246614e-01
-1.23979080e+00 -1.06200099e+00 -6.07186019e-01 -2.94661075e-01
4.54272270e-01 7.01865196e-01 9.90615427e-01 -1.48125127e-01
-4.46007043e-01 2.79492229e-01 -4.57261354e-01 -6.48045123e-01
1.25403637e-02 -2.12262049e-01 -1.09194107e-01 -1.04119286e-01
2.12180093e-01 -5.89578688e-01 -1.19922447e+00 4.44158196e-01
-5.13305426e-01 3.31526399e-01 8.17066312e-01 3.77557307e-01
4.39262569e-01 -1.97592437e-01 1.34038636e-02 -5.91190457e-01
-2.67744064e-02 -1.57501519e-01 -8.15029919e-01 9.63522471e-04
-5.73964298e-01 -3.41352016e-01 3.32369983e-01 -2.65185356e-01
-1.23541212e+00 1.54113501e-01 6.11986443e-02 -5.97177982e-01
-1.37031019e-01 2.05950141e-02 -2.04876840e-01 -3.39436769e-01
4.97726917e-01 5.27972318e-02 -1.90072909e-01 -6.28278315e-01
-3.38692814e-02 8.60847712e-01 2.28688210e-01 -2.72754610e-01
8.08813453e-01 7.58053541e-01 5.51929101e-02 -9.98229444e-01
-8.17544281e-01 -5.19237280e-01 -6.61425233e-01 -5.17762244e-01
7.65454829e-01 -1.16420972e+00 -9.24063921e-01 7.23588288e-01
-1.04513335e+00 -5.42528808e-01 -2.25565866e-01 5.57365477e-01
-5.26279628e-01 4.71815795e-01 -6.66742563e-01 -1.00737762e+00
-2.33672813e-01 -1.17197514e+00 1.40458608e+00 3.24587822e-01
1.56721726e-01 -7.36502469e-01 -3.89783740e-01 8.29637170e-01
3.76064390e-01 2.07589030e-01 6.09991670e-01 3.01077247e-01
-1.32584023e+00 -1.13714494e-01 -5.21991551e-01 5.03732026e-01
6.54608428e-01 1.40086301e-02 -1.56353533e+00 -8.91682208e-02
1.61660492e-01 -2.92135328e-01 1.02734363e+00 3.44406545e-01
1.19240689e+00 6.10627942e-02 -3.48710090e-01 6.80970728e-01
1.52836442e+00 9.35091823e-02 7.66361833e-01 3.76844376e-01
9.72787082e-01 5.03896713e-01 1.03528571e+00 7.19429910e-01
3.60200405e-01 7.87085176e-01 1.09168458e+00 -1.88058436e-01
-4.33148921e-01 3.61029714e-01 2.92188972e-01 6.56219959e-01
-5.18277764e-01 -2.44027331e-01 -7.89274454e-01 2.57970572e-01
-1.50467134e+00 -4.95661944e-01 -3.17339599e-01 2.26858044e+00
5.95875382e-01 2.53910035e-01 -1.96953595e-01 2.73022443e-01
2.32546538e-01 8.80700164e-03 -6.09764814e-01 -9.46938321e-02
-3.68245468e-02 -1.12981228e-02 4.97625142e-01 3.66685897e-01
-1.00125003e+00 6.47268891e-01 5.18690395e+00 4.14187431e-01
-1.29864275e+00 -3.25603150e-02 4.84601140e-01 -1.89764410e-01
-1.47428483e-01 -3.15796763e-01 -6.81921601e-01 3.72657180e-01
4.81865078e-01 3.00018996e-01 2.27936193e-01 7.26483822e-01
2.61857629e-01 -5.85655272e-01 -1.10730004e+00 1.24478066e+00
-1.11337639e-01 -9.76216495e-01 -1.88569903e-01 4.34036762e-01
8.43732417e-01 2.51894116e-01 4.02161092e-01 -3.05453122e-01
2.69474112e-03 -6.94138527e-01 8.90110672e-01 6.54676139e-01
6.13844037e-01 -4.25867766e-01 7.49589086e-01 1.20921291e-01
-1.29569340e+00 -2.99266726e-01 -5.41050255e-01 -7.15954527e-02
-1.32333398e-01 1.24446177e+00 -7.44674206e-01 8.96631837e-01
1.01418829e+00 9.28316951e-01 -5.57654440e-01 1.09541845e+00
-1.74827322e-01 3.62926185e-01 -2.54978329e-01 1.09906502e-01
-5.51649183e-02 -3.22956890e-01 3.80492777e-01 7.12607503e-01
4.26366180e-01 -2.21168637e-01 1.31179869e-01 1.09762561e+00
1.18874907e-01 -2.82736272e-01 -5.77705204e-01 2.10787244e-02
1.32812276e-01 1.39736712e+00 -5.29310226e-01 -1.73768505e-01
-6.92060590e-01 1.12028933e+00 2.22703889e-01 3.98871958e-01
-9.36188102e-01 -2.27071181e-01 9.54165876e-01 2.43476301e-01
4.23261672e-01 -5.26462495e-01 -6.10628843e-01 -1.15991879e+00
4.82816249e-01 -5.12445331e-01 -2.89431989e-01 -1.00854027e+00
-1.08878517e+00 5.25327444e-01 -3.51180196e-01 -1.61231887e+00
1.35333896e-01 -1.13506985e+00 -2.74794728e-01 9.67385769e-01
-1.83230114e+00 -1.25795698e+00 -9.47862685e-01 3.32753152e-01
7.08540380e-01 2.06110656e-01 7.08640039e-01 1.25867188e-01
-4.35976207e-01 8.37680399e-02 2.84638733e-01 -2.54578739e-01
6.27031267e-01 -1.18882406e+00 2.08949000e-01 7.17286289e-01
-5.94610013e-02 3.46922576e-01 6.51941299e-01 -5.20703256e-01
-1.72866952e+00 -9.83442783e-01 6.09265305e-02 -6.57894969e-01
3.27812403e-01 -4.21189785e-01 -7.49265075e-01 5.28549612e-01
-1.91328991e-02 1.11746840e-01 1.50110647e-01 -2.09128112e-01
-3.40274036e-01 -5.66841722e-01 -1.11070383e+00 5.27909756e-01
1.09306586e+00 -5.17835438e-01 -2.29363516e-01 2.47767389e-01
5.30285418e-01 -5.47752142e-01 -9.62054670e-01 6.07537568e-01
7.82067478e-01 -1.57173753e+00 1.06525278e+00 5.10778666e-01
6.45804286e-01 -2.85650820e-01 -2.67810762e-01 -1.10717773e+00
-1.00806504e-01 -2.23874271e-01 -3.70510370e-01 9.99184728e-01
1.91065356e-01 -9.30751622e-01 1.03033090e+00 4.90287304e-01
-4.46683109e-01 -9.58058059e-01 -8.39985311e-01 -7.62638688e-01
-3.10626268e-01 -4.70881671e-01 4.04253602e-01 3.64746541e-01
-3.41696352e-01 -5.29466122e-02 -5.54441810e-02 4.36317712e-01
8.23975623e-01 5.07268608e-01 7.31535673e-01 -1.46638119e+00
-2.10230291e-01 -1.22824863e-01 -4.30856794e-01 -1.38017392e+00
-3.76366913e-01 -4.03094739e-01 1.59788117e-01 -1.91687775e+00
1.63617313e-01 -6.56764448e-01 -1.88978881e-01 3.44088167e-01
-4.60980721e-02 5.69526732e-01 1.99795023e-01 1.77863762e-01
-4.56025183e-01 7.02486992e-01 1.72013342e+00 -3.11873257e-01
-6.02896735e-02 2.05830812e-01 -3.48225981e-01 6.03493929e-01
1.04868031e+00 -5.39542995e-02 -3.79837841e-01 -7.86600590e-01
6.33435696e-02 -2.70777851e-01 6.72546625e-01 -1.41194439e+00
-9.95038152e-02 -8.45846087e-02 7.34196901e-01 -6.59406424e-01
9.97761905e-01 -1.08841741e+00 -1.28428563e-01 4.14480150e-01
2.98015922e-01 -5.11378765e-01 2.21611962e-01 5.45627058e-01
-1.88331772e-02 3.28328125e-02 6.93745196e-01 -2.07229123e-01
-7.56306946e-01 3.84582341e-01 -2.80995458e-01 -5.62592030e-01
9.51237500e-01 -7.47597575e-01 -5.52313328e-01 -3.19389105e-01
-2.12028190e-01 -1.00466251e-01 7.31818259e-01 3.20258349e-01
9.57468808e-01 -1.02630901e+00 -3.59626859e-01 1.53729439e-01
3.52962524e-01 3.17768633e-01 3.42618048e-01 9.32045162e-01
-6.93533778e-01 4.20177668e-01 -2.60053903e-01 -8.36168706e-01
-1.33697450e+00 1.99923441e-01 3.88202637e-01 9.35440511e-02
-9.72418487e-01 8.89444172e-01 5.51794827e-01 -5.34317419e-02
1.23471648e-01 -9.72610176e-01 1.47658601e-01 -4.19358581e-01
4.39911753e-01 5.43104708e-01 3.20697725e-01 -3.27642381e-01
-2.95952588e-01 8.93064916e-01 4.86174338e-02 2.54807502e-01
1.35507369e+00 -4.65122432e-01 1.40900403e-01 7.10296273e-01
9.59203124e-01 1.42055377e-01 -1.83371949e+00 -1.01736389e-01
-4.56745088e-01 -8.15510154e-01 3.04448843e-01 -8.13826382e-01
-1.29642725e+00 1.30877984e+00 9.09651518e-01 1.10278986e-01
1.19484556e+00 3.92440781e-02 7.82194436e-01 4.46592122e-01
7.47385025e-01 -8.86384904e-01 3.71869892e-01 2.83031553e-01
8.61782551e-01 -1.44900596e+00 1.83352917e-01 -1.14987350e+00
-4.50632960e-01 1.20551515e+00 7.28417337e-01 1.83834016e-01
4.76940781e-01 2.95319706e-01 3.45401138e-01 -1.44264549e-01
-5.32701731e-01 -1.69489056e-01 2.05373585e-01 7.83757865e-01
3.12973410e-01 1.10144382e-02 5.62491953e-01 2.14566410e-01
-9.82950330e-02 -1.03121266e-01 2.89494038e-01 1.01105344e+00
-6.42753124e-01 -8.25757444e-01 -3.03218812e-01 3.89032722e-01
-2.21594095e-01 -8.89046676e-03 -3.06015521e-01 5.59500337e-01
5.10287769e-02 1.13890362e+00 -3.72063033e-02 -5.32570183e-01
2.69830436e-01 -3.73403102e-01 8.87965083e-01 -6.82593048e-01
-1.78523913e-01 -4.77540791e-02 3.44728641e-02 -9.52238858e-01
-3.93832535e-01 -4.60307658e-01 -1.16863084e+00 -1.01209231e-01
-4.94297892e-01 -5.13631403e-01 9.14503992e-01 7.07285285e-01
2.24384472e-01 7.41687834e-01 7.62567282e-01 -1.24626863e+00
-3.48785400e-01 -1.07280648e+00 -6.91475511e-01 4.00934629e-02
4.77775544e-01 -9.64221895e-01 -3.33084911e-01 -3.78813455e-03] | [7.472599506378174, -1.9088778495788574] |
53dbb001-bb20-4bc6-aee5-caf38c3fe0ce | novelty-controlled-paraphrase-generation-with | 2202.00535 | null | https://arxiv.org/abs/2202.00535v2 | https://arxiv.org/pdf/2202.00535v2.pdf | Novelty Controlled Paraphrase Generation with Retrieval Augmented Conditional Prompt Tuning | Paraphrase generation is a fundamental and long-standing task in natural language processing. In this paper, we concentrate on two contributions to the task: (1) we propose Retrieval Augmented Prompt Tuning (RAPT) as a parameter-efficient method to adapt large pre-trained language models for paraphrase generation; (2) we propose Novelty Conditioned RAPT (NC-RAPT) as a simple model-agnostic method of using specialized prompt tokens for controlled paraphrase generation with varying levels of lexical novelty. By conducting extensive experiments on four datasets, we demonstrate the effectiveness of the proposed approaches for retaining the semantic content of the original text while inducing lexical novelty in the generation. | ['Shuyi Wang', 'Yong Zhuang', 'Jishnu Ray Chowdhury'] | 2022-02-01 | null | null | null | null | ['paraphrase-generation', 'paraphrase-generation'] | ['computer-code', 'natural-language-processing'] | [ 2.33303368e-01 -9.90613326e-02 -3.22597533e-01 -1.65560931e-01
-1.10441148e+00 -6.53572142e-01 9.54668880e-01 3.24219584e-01
-4.99339491e-01 8.20878267e-01 6.85142159e-01 -8.78045857e-02
-8.49682689e-02 -5.35421550e-01 -7.89394975e-01 -1.50611281e-01
3.83712143e-01 5.11298180e-01 1.27636388e-01 -5.52758336e-01
1.08066976e+00 5.05218685e-01 -1.47813785e+00 6.01800203e-01
1.16929591e+00 5.26407242e-01 3.78793180e-01 6.90637410e-01
-5.14210105e-01 6.77647412e-01 -7.49532104e-01 -4.12144005e-01
1.84751689e-01 -7.40279913e-01 -1.04084742e+00 -2.55873084e-01
4.22756076e-01 6.48779124e-02 -1.08732179e-01 7.28270650e-01
7.00432301e-01 5.73323011e-01 9.71511185e-01 -8.87648523e-01
-1.07506454e+00 9.62083459e-01 -4.39489126e-01 5.89797795e-01
6.45313859e-01 1.70899689e-01 9.93179440e-01 -1.50484347e+00
7.08218157e-01 1.31814528e+00 4.22436744e-01 7.49708354e-01
-1.13740695e+00 -4.89708781e-01 4.81813625e-02 1.96852356e-01
-1.34959066e+00 -6.61128223e-01 8.59339654e-01 -1.14069432e-01
1.42401195e+00 2.46613786e-01 4.53993589e-01 1.23646915e+00
2.53180593e-01 9.45526540e-01 1.00270569e+00 -9.13503706e-01
3.90652895e-01 3.51101398e-01 2.16786698e-01 3.25016499e-01
1.45494610e-01 -1.70289055e-02 -8.74310851e-01 -5.34238040e-01
4.48552191e-01 -1.56533733e-01 -2.43088435e-02 -5.96023612e-02
-1.21413004e+00 1.01443279e+00 2.39012018e-01 4.57914293e-01
-4.37208921e-01 1.41494140e-01 5.84075511e-01 6.76939666e-01
6.66802824e-01 1.01199841e+00 -2.58222073e-01 -2.59532481e-01
-1.14972413e+00 6.58055723e-01 6.27348661e-01 1.29690504e+00
5.74962676e-01 4.13885787e-02 -9.76277769e-01 1.31427920e+00
-1.50565162e-01 4.07268167e-01 1.28164566e+00 -7.55604327e-01
5.53510725e-01 4.66952264e-01 3.30214858e-01 -5.55244803e-01
1.79328129e-01 -3.64756316e-01 -6.51614487e-01 -5.20359993e-01
-2.58368015e-01 1.37362584e-01 -7.99864113e-01 1.75285101e+00
-1.11656234e-01 -2.70605199e-02 2.27800876e-01 3.24485093e-01
7.81714559e-01 7.67549932e-01 2.46531099e-01 -3.36442798e-01
1.02568674e+00 -1.38788319e+00 -5.05434334e-01 -3.38991344e-01
3.88201058e-01 -1.12538838e+00 1.61533093e+00 -4.67547290e-02
-1.52411747e+00 -6.70095623e-01 -7.34024048e-01 -1.40673786e-01
-4.35921192e-01 -2.80681625e-02 5.24685323e-01 3.37866873e-01
-1.04340553e+00 6.87102497e-01 -1.45110294e-01 -6.56590521e-01
1.30564705e-01 -1.21531181e-01 -9.76258516e-03 -1.50761396e-01
-1.34422326e+00 9.87887621e-01 5.53188920e-01 -4.28620487e-01
-6.58453524e-01 -9.04179156e-01 -7.39863336e-01 3.12372208e-01
3.07869405e-01 -1.25347483e+00 1.55780554e+00 -7.51804948e-01
-1.65204620e+00 8.09513211e-01 -6.52436018e-01 -3.94999832e-01
2.58227199e-01 -3.98187637e-01 -5.82933351e-02 1.43268034e-01
3.44610363e-01 7.70533800e-01 1.12510335e+00 -1.02964771e+00
-3.07905436e-01 5.36335558e-02 -1.51062682e-01 5.40622294e-01
-4.11138177e-01 2.23295227e-01 -1.76228479e-01 -1.35652661e+00
-3.88985476e-03 -7.06807673e-01 -2.88234591e-01 -3.91062737e-01
-2.95329303e-01 -5.50670147e-01 3.10857207e-01 -5.06498516e-01
1.36688292e+00 -1.96354473e+00 2.52397597e-01 -1.05036840e-01
-1.90963045e-01 3.21885079e-01 -5.06126761e-01 1.12049007e+00
6.85380399e-02 3.14841062e-01 -2.35294893e-01 -4.46671754e-01
7.27535114e-02 1.70283695e-03 -1.00844502e+00 -5.39800823e-01
1.95581466e-01 1.34936857e+00 -1.06200969e+00 -5.23629189e-01
-1.46978229e-01 -2.36172244e-01 -4.70152169e-01 4.18237358e-01
-3.92731577e-01 -1.16670849e-02 -3.80887479e-01 5.87953389e-01
3.81516486e-01 -6.90330788e-02 -4.93111044e-01 2.03897819e-01
9.21252817e-02 4.15438205e-01 -6.78543210e-01 1.95889831e+00
-5.69788218e-01 1.49790168e-01 -5.86136103e-01 -5.33547699e-01
7.96225488e-01 1.64786875e-01 -8.56018588e-02 -6.24113679e-01
-7.38457739e-02 3.22303593e-01 -4.88556445e-01 -4.53720361e-01
1.26693308e+00 -3.53404254e-01 -3.10084343e-01 9.11993444e-01
1.11975238e-01 -6.56648934e-01 4.71816897e-01 6.53504193e-01
9.85161662e-01 -3.00712466e-01 5.52975774e-01 -2.48025030e-01
5.31641543e-01 8.97157565e-02 9.82037932e-02 1.47623694e+00
-2.27800254e-02 4.81095284e-01 4.80744839e-02 -4.78152931e-02
-1.22624266e+00 -1.23098493e+00 2.52141267e-01 1.32979751e+00
-5.29282354e-02 -4.86543894e-01 -5.32110751e-01 -6.67918801e-01
7.12666512e-02 1.18194175e+00 -3.48583072e-01 -4.64295864e-01
-6.76486671e-01 -4.77786541e-01 6.59660339e-01 5.89420557e-01
4.76606131e-01 -1.44231677e+00 -5.24997786e-02 2.00357988e-01
-4.86715972e-01 -7.37681627e-01 -1.04713786e+00 -5.86488172e-02
-1.07112861e+00 -3.86650592e-01 -9.31419015e-01 -9.52742755e-01
6.84966624e-01 7.20708370e-01 1.27206588e+00 -2.42669042e-03
-1.20537736e-01 4.97175246e-01 -5.97114384e-01 -3.26320142e-01
-7.95991182e-01 3.19399834e-01 -1.35323207e-03 -4.15381730e-01
3.54162037e-01 -6.15059555e-01 -3.89537901e-01 -2.82717329e-02
-1.14831388e+00 2.04648972e-02 8.52710545e-01 1.04024255e+00
6.21489108e-01 -4.41153020e-01 1.16986668e+00 -9.72144067e-01
1.72132075e+00 -5.40759027e-01 -2.32757609e-02 5.36032140e-01
-8.44530284e-01 2.97902614e-01 7.88784802e-01 -7.44133353e-01
-1.09313893e+00 -4.05570716e-01 5.24790660e-02 -4.28383738e-01
3.14733759e-02 5.92251003e-01 3.88400137e-01 4.39275131e-02
9.37715590e-01 7.94265389e-01 -2.16190368e-01 -4.54568237e-01
8.06097686e-01 5.96070528e-01 4.39049989e-01 -9.35473323e-01
9.00039673e-01 9.91561264e-02 -3.64398330e-01 -6.97251379e-01
-8.93265784e-01 -6.38248026e-01 -3.91561478e-01 1.28175393e-01
2.48308420e-01 -9.18817163e-01 2.13679716e-01 3.26313615e-01
-1.18507683e+00 -2.23489508e-01 -8.01539183e-01 1.12687886e-01
-6.48196697e-01 8.11621010e-01 -5.99231422e-01 -4.88695771e-01
-1.04471910e+00 -5.36353230e-01 1.10052979e+00 3.67692173e-01
-4.90523249e-01 -8.62378657e-01 5.10814726e-01 2.97342151e-01
7.90036380e-01 -3.05289239e-01 1.19709516e+00 -8.67332399e-01
-5.19096911e-01 -2.73958683e-01 -1.43617868e-01 4.27612185e-01
1.65321156e-01 -2.73109376e-01 -6.67407990e-01 -4.07793641e-01
-1.01686092e-02 -6.95869744e-01 9.94450867e-01 1.72276497e-01
9.27785993e-01 -5.10309756e-01 -7.83777386e-02 3.63586485e-01
1.10944843e+00 -1.27495587e-01 4.35690314e-01 2.62106776e-01
2.58408815e-01 4.27407563e-01 9.01730418e-01 5.24211168e-01
-4.83248569e-02 5.30949712e-01 -4.47046161e-01 3.16355914e-01
-3.16094548e-01 -8.69421542e-01 3.96218240e-01 1.06964481e+00
3.68130773e-01 -3.03658605e-01 -2.46949539e-01 8.42550814e-01
-1.94818485e+00 -1.21991336e+00 3.38744730e-01 2.05922127e+00
1.17192590e+00 9.29150879e-02 -4.16776948e-02 -2.16217130e-01
7.47835577e-01 1.58027932e-01 -5.69160819e-01 -8.10810089e-01
-2.82371700e-01 5.98623931e-01 -2.63785981e-02 4.23921764e-01
-4.97703046e-01 1.37283957e+00 6.96152830e+00 1.32255828e+00
-7.41477072e-01 9.30170640e-02 9.71018597e-02 -3.31673890e-01
-5.66373527e-01 1.66590437e-01 -7.55092025e-01 5.26509762e-01
7.96525598e-01 -1.07483459e+00 5.26740193e-01 1.01344454e+00
3.54072928e-01 -7.51613006e-02 -1.09298778e+00 7.88546920e-01
5.39838850e-01 -1.28121293e+00 9.36707437e-01 -5.63922346e-01
1.03692889e+00 -3.07379186e-01 1.95260108e-01 7.73411155e-01
1.47998437e-01 -7.28268504e-01 6.20798588e-01 4.82538998e-01
5.59601963e-01 -7.29919314e-01 3.58630776e-01 4.65289712e-01
-9.17382240e-01 -1.25159979e-01 -7.26226449e-01 7.49716535e-02
3.23493242e-01 5.29580116e-01 -9.38367426e-01 5.13505340e-01
3.04424763e-01 5.00222921e-01 -9.17203307e-01 1.08352447e+00
-4.55764562e-01 5.53542614e-01 3.50002237e-02 -3.30733925e-01
1.16290137e-01 1.67357132e-01 8.51803660e-01 1.44990230e+00
3.15808296e-01 -7.02605098e-02 -1.49265409e-01 1.01691794e+00
-3.77728403e-01 4.64535058e-01 -6.95295274e-01 -1.59473360e-01
9.08745289e-01 9.47020411e-01 -3.70919496e-01 -6.10987365e-01
1.29529074e-01 1.34239781e+00 4.72843260e-01 3.54866207e-01
-4.97038245e-01 -8.12618256e-01 2.28957519e-01 -1.10648297e-01
3.26550841e-01 -3.19813453e-02 -2.48518959e-01 -1.44965219e+00
2.44189203e-01 -8.29824388e-01 3.96809191e-01 -1.05672717e+00
-1.83294785e+00 5.25204659e-01 4.02630419e-01 -1.10953224e+00
-6.47790492e-01 1.90423615e-02 -9.40517485e-01 1.15582979e+00
-1.64083302e+00 -1.13466465e+00 -1.80751801e-01 4.81125057e-01
1.28995907e+00 -4.07090724e-01 6.61496520e-01 -1.64736882e-01
-1.63543105e-01 8.76041889e-01 2.40391269e-01 -4.98662323e-01
9.90786433e-01 -1.13865888e+00 8.52285564e-01 9.52904761e-01
2.48739332e-01 1.26949346e+00 7.56236851e-01 -7.84468472e-01
-1.00849569e+00 -1.21116567e+00 1.58768320e+00 -5.42784452e-01
5.67230284e-01 -2.20550209e-01 -9.82502580e-01 4.61168468e-01
4.32534397e-01 -8.10485423e-01 6.70613885e-01 -3.92129160e-02
-4.63389128e-01 1.55324236e-01 -1.17781568e+00 9.50955510e-01
1.05648422e+00 -7.14213371e-01 -1.39739370e+00 5.68826139e-01
1.21128404e+00 -1.58914775e-01 -3.58047664e-01 2.61991769e-01
2.33468920e-01 -5.65142035e-01 9.94108379e-01 -7.86880791e-01
5.25778711e-01 1.27604306e-01 -3.00045833e-02 -1.52605700e+00
-5.39027214e-01 -8.49429488e-01 -1.64598718e-01 1.23672771e+00
2.70007074e-01 -4.80594546e-01 6.58982277e-01 4.92598832e-01
-2.88447857e-01 -8.02636862e-01 -7.80799091e-01 -1.04246092e+00
2.89081812e-01 8.38733092e-02 4.91827756e-01 7.36957431e-01
2.09054634e-01 7.29983747e-01 -4.10183907e-01 -4.15324658e-01
3.97339076e-01 1.34853616e-01 7.58237302e-01 -6.88659251e-01
-4.98086244e-01 -5.26271582e-01 2.33743072e-01 -1.34691572e+00
3.07202667e-01 -1.05200946e+00 4.39826734e-02 -1.38881361e+00
6.86398685e-01 -5.85718378e-02 -2.02118352e-01 1.88827157e-01
-5.21515310e-01 5.03725605e-03 1.35848388e-01 5.84750593e-01
-5.20468950e-01 9.35468912e-01 1.05152094e+00 -1.69485271e-01
-5.06128192e-01 2.06830740e-01 -1.01356673e+00 2.35982850e-01
7.76894271e-01 -6.20765984e-01 -8.46705675e-01 -3.78446013e-01
2.20155627e-01 -7.64800385e-02 1.84540242e-01 -5.40933073e-01
2.59057343e-01 -3.73105317e-01 -6.20017834e-02 -7.01763391e-01
2.20811635e-01 -1.90749928e-01 -2.34281361e-01 4.20209438e-01
-1.03485870e+00 5.99186659e-01 3.25244814e-01 7.42069840e-01
-1.37917385e-01 -9.17524397e-01 6.33569837e-01 -4.46568727e-01
-4.87186611e-01 -5.61083704e-02 -5.64867854e-01 4.65673536e-01
6.59526646e-01 -2.00250715e-01 -4.04363126e-01 -3.45667273e-01
-2.79728711e-01 8.24939087e-02 4.83659059e-01 7.94797421e-01
9.58285630e-01 -1.27179456e+00 -8.35481882e-01 1.06504627e-01
5.83761632e-01 -3.93732071e-01 2.71013916e-01 9.81437266e-02
-1.81350380e-01 7.80944169e-01 -3.68939973e-02 -1.28471926e-01
-1.14043701e+00 7.23710835e-01 -1.00877188e-01 -5.40786564e-01
-3.58834416e-01 9.17634785e-01 -3.38725047e-03 -3.80307645e-01
7.55671039e-02 -1.88319251e-01 -1.31966144e-01 -2.95351148e-01
4.73696798e-01 2.49156982e-01 1.58276021e-01 -1.44008011e-01
1.68157220e-01 2.05155984e-01 -8.57255340e-01 -5.82196772e-01
9.76007462e-01 -2.35955268e-01 -1.95930809e-01 5.28014719e-01
8.72931421e-01 1.58432536e-02 -3.20453852e-01 -4.72759217e-01
1.43798113e-01 -5.92344403e-01 -3.19939703e-01 -9.70223963e-01
-2.11121172e-01 3.98899704e-01 1.43725961e-01 -1.26081109e-01
9.79613841e-01 5.97256310e-02 1.16156554e+00 8.56436014e-01
3.73683631e-01 -1.17323923e+00 6.07648790e-01 6.36758924e-01
1.27479434e+00 -7.92795002e-01 -6.76000565e-02 -6.01746663e-02
-8.11531663e-01 7.29543328e-01 6.99833274e-01 -7.21849799e-02
2.26583138e-01 -5.21489680e-01 -1.39237896e-01 -6.90255256e-04
-1.18811297e+00 1.31169423e-01 2.47961968e-01 4.01256859e-01
3.27241689e-01 -4.79990005e-01 -8.51642072e-01 5.07168055e-01
-2.51759708e-01 4.27216776e-02 5.86409867e-01 1.18642855e+00
-6.96083546e-01 -1.30038559e+00 -1.13797393e-02 5.82530499e-01
-2.10501865e-01 -6.56720340e-01 -8.42477977e-01 3.77956331e-01
-4.69779193e-01 9.13773417e-01 -3.99076462e-01 -9.72115993e-02
4.77520406e-01 3.84830236e-01 5.60082972e-01 -1.09121001e+00
-9.21202183e-01 -3.63442570e-01 -1.87367618e-01 -2.59724855e-01
-6.97100442e-03 -4.21556503e-01 -7.69064903e-01 -1.59697592e-01
-3.14932555e-01 5.86808264e-01 4.91992205e-01 8.01984310e-01
7.31751502e-01 1.14509054e-01 7.85648823e-01 -7.48905420e-01
-1.28209782e+00 -1.32043147e+00 -4.55369353e-01 6.88848078e-01
-7.65779316e-02 -4.11215097e-01 -6.55558944e-01 5.74792996e-02] | [11.796853065490723, 9.285239219665527] |
a0e07ad8-9721-41a3-8aa8-9a8100219950 | towards-dynamic-multi-modal-phenotyping-using | 2111.02710 | null | https://arxiv.org/abs/2111.02710v1 | https://arxiv.org/pdf/2111.02710v1.pdf | Towards dynamic multi-modal phenotyping using chest radiographs and physiological data | The healthcare domain is characterized by heterogeneous data modalities, such as imaging and physiological data. In practice, the variety of medical data assists clinicians in decision-making. However, most of the current state-of-the-art deep learning models solely rely upon carefully curated data of a single modality. In this paper, we propose a dynamic training approach to learn modality-specific data representations and to integrate auxiliary features, instead of solely relying on a single modality. Our preliminary experiments results for a patient phenotyping task using physiological data in MIMIC-IV & chest radiographs in the MIMIC- CXR dataset show that our proposed approach achieves the highest area under the receiver operating characteristic curve (AUROC) (0.764 AUROC) compared to the performance of the benchmark method in previous work, which only used physiological data (0.740 AUROC). For a set of five recurring or chronic diseases with periodic acute episodes, including cardiac dysrhythmia, conduction disorders, and congestive heart failure, the AUROC improves from 0.747 to 0.798. This illustrates the benefit of leveraging the chest imaging modality in the phenotyping task and highlights the potential of multi-modal learning in medical applications. | ['Farah E. Shamout', 'Krzysztof J. Geras', 'Nasir Hayat'] | 2021-11-04 | null | null | null | null | ['patient-phenotyping'] | ['medical'] | [ 1.39603794e-01 -1.03925318e-01 -1.30121931e-01 -3.37987572e-01
-1.03289080e+00 -4.29792911e-01 7.57343173e-02 4.19096053e-01
-2.88419038e-01 8.89075398e-01 1.44001335e-01 -6.04213119e-01
-4.31187272e-01 -4.04826522e-01 -3.86600822e-01 -8.73703539e-01
-2.58209914e-01 5.49305201e-01 -3.08946818e-01 3.96601617e-01
-2.87622958e-01 4.39133853e-01 -1.00949681e+00 2.60026485e-01
6.62035823e-01 1.30897498e+00 -9.89718884e-02 7.46778190e-01
2.44431734e-01 7.80510664e-01 -6.42112017e-01 1.20554538e-02
1.87884331e-01 -6.05057478e-01 -5.46645105e-01 3.99473943e-02
1.37279466e-01 -3.08804482e-01 -3.81281465e-01 3.12735081e-01
1.16243613e+00 -2.83193737e-01 5.02304852e-01 -8.22242320e-01
-3.99604648e-01 3.60841870e-01 -3.89154077e-01 4.35404897e-01
2.99253672e-01 3.28113079e-01 7.94919074e-01 -4.40187156e-01
4.02747273e-01 5.42918384e-01 9.62102890e-01 5.36532760e-01
-1.13267648e+00 -4.63987738e-01 -5.10754287e-02 8.01351890e-02
-1.22697926e+00 -5.58818281e-02 4.65027094e-01 -5.79034030e-01
7.71475792e-01 3.11325938e-01 6.72834218e-01 1.14593577e+00
3.74407232e-01 6.29866421e-01 1.11704910e+00 -2.17944190e-01
1.31305233e-01 2.51840893e-02 1.28450990e-01 5.82551897e-01
1.53924182e-01 1.25420764e-01 -3.47529143e-01 -5.36041677e-01
6.31259799e-01 4.32583034e-01 -4.72404510e-01 -2.53882170e-01
-1.47311234e+00 6.18442535e-01 2.12736934e-01 3.85105669e-01
-7.56678104e-01 -8.09649937e-03 5.81189156e-01 4.08220112e-01
6.49218038e-02 5.28507829e-01 -7.78514624e-01 -2.01626942e-01
-8.18034708e-01 6.93165138e-02 6.67436004e-01 4.72977251e-01
1.24963015e-01 -6.75852075e-02 -4.04273570e-01 9.74259973e-01
1.64720193e-01 4.76405144e-01 7.73360550e-01 -6.86632037e-01
2.67311007e-01 7.01097965e-01 -4.33191061e-02 -3.00103813e-01
-8.84374857e-01 -8.11172724e-01 -1.03388214e+00 -2.15835482e-01
3.11156482e-01 -4.43228602e-01 -1.13118970e+00 1.63650405e+00
2.22019389e-01 1.37419865e-01 2.48192444e-01 8.92760575e-01
1.09919858e+00 1.74986139e-01 2.92433858e-01 -2.27174610e-01
1.38312399e+00 -4.96468067e-01 -5.62888265e-01 1.47139251e-01
8.31753790e-01 -3.99952263e-01 1.13021386e+00 5.50693154e-01
-8.31749201e-01 -3.44412029e-01 -1.04875731e+00 3.81149769e-01
-1.59635678e-01 3.01129252e-01 5.53954422e-01 5.64684629e-01
-8.68288577e-01 5.86864412e-01 -8.10743451e-01 -3.10897410e-01
5.26466727e-01 3.86906445e-01 -5.09763479e-01 -3.20761144e-01
-1.37642252e+00 7.82958388e-01 2.28295773e-01 -2.04473376e-01
-1.01463425e+00 -1.19725692e+00 -4.90128607e-01 1.01678185e-01
1.92992061e-01 -1.21752834e+00 9.67463136e-01 -5.52354455e-01
-1.22424316e+00 7.85232365e-01 2.77845293e-01 -4.05947417e-01
7.03274429e-01 -2.40499899e-01 -5.85225821e-01 3.17155272e-01
-1.05131254e-01 2.80429810e-01 4.80170369e-01 -9.59123135e-01
-4.27618802e-01 -4.79625970e-01 -7.59020895e-02 1.67920560e-01
-2.31037453e-01 -1.27229035e-01 -4.00589496e-01 -4.79579031e-01
-6.60073608e-02 -1.01243556e+00 -2.53347546e-01 -6.02720715e-02
-2.98227489e-01 1.24197789e-02 7.07703412e-01 -6.86014056e-01
1.30196941e+00 -2.20462513e+00 9.08122957e-02 6.29299507e-02
5.02037108e-01 2.09208876e-01 3.14500809e-01 2.98841894e-01
-3.38270813e-01 7.33169392e-02 -4.54913616e-01 -1.31255925e-01
-2.94159234e-01 1.76589116e-01 2.93503135e-01 5.63560903e-01
1.74153402e-01 9.67502952e-01 -7.15841055e-01 -4.32418823e-01
4.53972459e-01 5.60328901e-01 -1.44413918e-01 3.60770106e-01
2.03636661e-01 8.67915034e-01 -4.97035801e-01 8.97634745e-01
2.94530600e-01 -9.31505561e-01 2.96712667e-01 1.34688467e-01
2.74650514e-01 5.26721999e-02 -7.13225603e-01 1.88847113e+00
-6.96933389e-01 3.82831186e-01 -1.30561367e-01 -1.16361022e+00
7.50301003e-01 7.63368785e-01 1.12480760e+00 -5.63549757e-01
1.91222101e-01 2.25005552e-01 3.85981083e-01 -6.90476656e-01
-4.70069230e-01 -5.27495384e-01 -8.20429549e-02 4.91819888e-01
2.19661109e-02 1.68725073e-01 -1.81801796e-01 -2.62840781e-02
1.56454432e+00 -3.43376487e-01 3.71459216e-01 -1.96825564e-01
6.42142713e-01 -2.47756794e-01 4.78268296e-01 8.60405207e-01
-3.42838198e-01 9.37392890e-01 4.22567278e-01 -6.38459861e-01
-8.38740587e-01 -1.08740652e+00 -7.07681656e-01 6.31074965e-01
-2.52884805e-01 -2.90616006e-01 -2.78045654e-01 -7.10071385e-01
2.12248698e-01 3.23917985e-01 -6.94771409e-01 -2.03831404e-01
-5.25153935e-01 -1.22660100e+00 7.02039659e-01 9.28070784e-01
3.25014979e-01 -9.32046175e-01 -1.08455503e+00 2.95766145e-01
-4.67092767e-02 -1.01106429e+00 1.56415984e-01 2.67991215e-01
-1.13293624e+00 -1.20571935e+00 -1.07954633e+00 -3.44411373e-01
2.22433135e-01 -4.28826392e-01 1.12035418e+00 9.17650461e-02
-6.23894036e-01 5.42998135e-01 -2.79431522e-01 -4.53994155e-01
-3.93919885e-01 1.70362860e-01 -1.17728047e-01 -6.03158846e-02
9.50331017e-02 -5.42190850e-01 -1.02332079e+00 1.47403523e-01
-7.06972301e-01 -9.97423157e-02 8.68703365e-01 1.10821021e+00
7.23069727e-01 -3.86229843e-01 9.34537649e-01 -1.09333205e+00
5.75492144e-01 -8.27910781e-01 -7.41459802e-02 3.12710583e-01
-8.81664455e-01 -2.35964999e-01 7.35178769e-01 -3.46377283e-01
-7.76043534e-01 1.33933410e-01 -6.62873313e-02 -5.99668205e-01
-4.79679316e-01 7.60300159e-01 1.93770632e-01 2.81667620e-01
6.92027628e-01 2.66934127e-01 -1.75397024e-02 -4.24855828e-01
-1.80582583e-01 6.25401378e-01 5.81024945e-01 -4.63544577e-01
1.96098164e-01 2.72043616e-01 2.41304547e-01 -5.24949074e-01
-6.46740735e-01 -5.70811927e-01 -4.66006070e-01 -1.24736853e-01
7.98562169e-01 -1.03652585e+00 -6.40317321e-01 4.62365001e-01
-5.59327424e-01 -1.59411252e-01 -3.62226486e-01 7.62863338e-01
-4.93850946e-01 3.23375255e-01 -5.59508562e-01 -5.12238801e-01
-6.74293339e-01 -1.05336726e+00 1.14578474e+00 -2.02524271e-02
-3.50585788e-01 -1.14612865e+00 1.67562068e-01 3.07879627e-01
5.99676847e-01 8.54391456e-01 1.21626627e+00 -8.74740660e-01
-1.78892314e-01 -5.55867434e-01 -3.78718488e-02 3.36791903e-01
4.62843388e-01 -4.09869075e-01 -1.11105311e+00 -3.67584348e-01
4.95164432e-02 -3.97937626e-01 7.43307590e-01 6.88554525e-01
1.59291613e+00 2.84030974e-01 -2.43347898e-01 7.42289960e-01
1.30467439e+00 5.41233182e-01 4.06610996e-01 1.15626432e-01
6.36187851e-01 1.24644920e-01 3.01463097e-01 8.09696794e-01
2.50111341e-01 2.37229809e-01 4.10296887e-01 -4.30719614e-01
4.80855852e-02 1.89754903e-01 -2.91863918e-01 4.88544106e-01
-4.53642048e-02 -1.70379400e-01 -1.38313782e+00 5.29162586e-01
-1.74374223e+00 -4.07786995e-01 -1.84303883e-03 2.09716940e+00
7.73269355e-01 -6.65042996e-02 3.53497714e-02 2.74958491e-01
5.92529297e-01 -7.35724121e-02 -9.80002046e-01 -2.11679325e-01
-3.75347994e-02 3.74536216e-01 3.02902102e-01 -5.64318039e-02
-1.18279386e+00 2.24482805e-01 6.96443558e+00 1.08174689e-01
-1.41520238e+00 1.60172895e-01 8.66259277e-01 -2.02975243e-01
4.19961587e-02 -4.69218850e-01 -1.95008829e-01 5.85206926e-01
1.28576195e+00 1.60988346e-02 5.97881004e-02 7.68985868e-01
-3.61617212e-03 -8.45398828e-02 -1.33379066e+00 1.14925420e+00
-2.25621071e-02 -1.28018081e+00 -3.58727574e-01 -1.82296094e-02
5.24304032e-01 1.93477944e-01 1.70510590e-01 3.31534296e-01
-2.22380921e-01 -1.26128638e+00 -3.07863876e-02 5.11955619e-01
1.24198723e+00 -3.28345597e-01 9.82683539e-01 1.81954503e-01
-6.80401266e-01 -1.68997005e-01 1.83258072e-01 1.79765373e-01
2.59780362e-02 7.78495073e-01 -9.42016244e-01 7.34458804e-01
7.81130970e-01 7.90845633e-01 -3.63030285e-01 9.85122144e-01
5.42904809e-02 8.60036910e-01 -3.01269412e-01 3.92991513e-01
-3.05767916e-02 3.37078303e-01 2.41646573e-01 1.09035981e+00
1.89370900e-01 2.96934724e-01 1.04177512e-01 5.57562649e-01
-1.55451119e-01 1.89341336e-01 -5.39181411e-01 3.06679178e-02
2.00065270e-01 1.22052491e+00 -4.16777909e-01 -5.84938645e-01
-6.00209534e-01 4.90981042e-01 -1.64906383e-01 1.86356604e-01
-1.04025257e+00 -5.78781404e-02 1.74786732e-01 1.60613224e-01
1.09995939e-01 3.58668953e-01 -7.30414391e-01 -1.05976987e+00
-2.28915382e-02 -9.71115947e-01 7.93501914e-01 -5.17368197e-01
-1.36267853e+00 6.64430976e-01 -6.26612604e-02 -1.23306727e+00
-3.48916620e-01 -4.88138139e-01 -5.70702136e-01 1.07566500e+00
-1.41432309e+00 -9.60095644e-01 -5.52976668e-01 6.68711960e-01
2.53793776e-01 -3.23495895e-01 1.17530453e+00 4.74081874e-01
-4.68365967e-01 7.38353550e-01 1.44552514e-01 4.28169779e-02
8.19047451e-01 -1.29547000e+00 3.28916684e-02 3.74254823e-01
-4.59874839e-01 6.05580568e-01 2.55461663e-01 -3.74082804e-01
-1.29879081e+00 -9.76521254e-01 4.34159130e-01 -3.87648702e-01
4.66841906e-01 1.96234822e-01 -1.03393745e+00 4.14092839e-01
6.73895702e-02 4.85574812e-01 1.26314008e+00 2.29723364e-01
-1.78669944e-01 -1.71014681e-01 -1.30677855e+00 -2.57351995e-02
6.83527470e-01 -3.92970473e-01 -4.31698889e-01 2.87629753e-01
3.38358015e-01 -5.37833214e-01 -1.56818426e+00 9.63641703e-01
7.73219168e-01 -7.54504383e-01 9.19989467e-01 -9.08509910e-01
4.73178297e-01 1.85815170e-02 -8.76607075e-02 -1.19287503e+00
-2.33760253e-01 -3.52972627e-01 -4.66958225e-01 6.43544197e-01
5.00101864e-01 -7.02251792e-01 6.31442010e-01 6.02414906e-01
-2.21247882e-01 -1.29206598e+00 -8.49121153e-01 -4.12421554e-01
1.71378598e-01 -2.88971275e-01 5.37622988e-01 1.17279088e+00
-1.29698470e-01 2.27691323e-01 -2.89342552e-01 4.02817018e-02
2.60080189e-01 3.38337541e-01 3.70193303e-01 -1.42127812e+00
-5.67712843e-01 -9.78524387e-02 -3.28538537e-01 -2.92434573e-01
-2.12835759e-01 -7.97737479e-01 -3.24726343e-01 -1.64858532e+00
4.69936788e-01 -7.71636009e-01 -1.09335899e+00 5.55975974e-01
-4.73550826e-01 6.66012615e-02 3.46883908e-02 2.10929930e-01
-3.24129969e-01 1.67476639e-01 1.28247178e+00 -4.65799533e-02
-2.57533580e-01 9.98647138e-02 -7.99781024e-01 5.55638790e-01
1.06609416e+00 -4.62092549e-01 -5.47263145e-01 -3.19520831e-01
-1.90417364e-01 5.70312738e-01 2.60222286e-01 -1.10957825e+00
-2.17752963e-01 1.23250380e-01 7.67884970e-01 -4.59722430e-01
2.67284304e-01 -7.27016270e-01 2.62565404e-01 8.25219214e-01
-3.56893897e-01 2.43352711e-01 3.11920851e-01 6.08646810e-01
-2.77243197e-01 3.61311287e-01 8.81253898e-01 -2.86331415e-01
-4.26103622e-01 4.41372961e-01 -3.79741818e-01 2.78228194e-01
1.11236620e+00 4.63717710e-03 -3.57264459e-01 -2.25956038e-01
-1.16216350e+00 1.74248531e-01 6.46198764e-02 3.75190586e-01
5.91400743e-01 -1.10451174e+00 -8.74489903e-01 1.98133990e-01
3.28302443e-01 1.18053826e-02 5.22675455e-01 1.32988346e+00
-5.35503864e-01 4.89316076e-01 -2.28824809e-01 -1.15844619e+00
-1.16344202e+00 4.11806226e-01 5.16526520e-01 -4.29444522e-01
-8.22586894e-01 5.93817651e-01 2.00275093e-01 -2.31561095e-01
2.74516016e-01 -4.07532364e-01 1.03386473e-02 -5.32222614e-02
2.62786001e-01 2.67770201e-01 5.00552535e-01 -2.44346872e-01
-6.87574804e-01 4.46939677e-01 -9.10859481e-02 2.09099159e-01
1.34316051e+00 1.92083016e-01 2.59528905e-01 5.02255797e-01
1.19072413e+00 -4.46212590e-01 -8.31454754e-01 -1.20909095e-01
-2.57516056e-01 -1.82157218e-01 1.24281332e-01 -1.43035817e+00
-1.25911438e+00 9.21793818e-01 1.07040155e+00 6.77213073e-02
1.39414990e+00 6.50629625e-02 6.30829573e-01 2.95638323e-01
1.23740338e-01 -7.26991355e-01 -1.72608923e-02 1.11661941e-01
6.18038774e-01 -1.37197721e+00 1.56058118e-01 1.62837491e-03
-1.06261933e+00 8.98296416e-01 4.57763046e-01 3.81837934e-02
8.41244221e-01 2.71494329e-01 4.67925370e-01 -2.51011193e-01
-9.63669479e-01 1.36084229e-01 4.67144810e-02 4.07963812e-01
6.68478608e-01 3.01694661e-01 -3.39762002e-01 6.92468464e-01
1.96503669e-01 2.02612400e-01 3.08850884e-01 1.15159523e+00
-2.07028724e-03 -1.00484419e+00 -1.31988093e-01 8.18213284e-01
-9.79197204e-01 -9.06023234e-02 -3.67496133e-01 9.10358310e-01
1.07062191e-01 6.88107133e-01 -2.36797303e-01 -3.03492725e-01
3.80956739e-01 4.50500906e-01 3.46325964e-01 -5.60316801e-01
-8.02063465e-01 1.32543743e-01 -6.78584576e-02 -5.28014302e-01
-4.65106934e-01 -6.15916371e-01 -1.15581167e+00 7.27345869e-02
-7.48088658e-02 -1.44054398e-01 5.35466075e-01 7.51133084e-01
5.93926489e-01 1.06056726e+00 5.01795471e-01 -2.06351236e-01
-5.30134320e-01 -1.02319169e+00 -5.30122638e-01 5.22869706e-01
6.95211768e-01 -7.20644772e-01 -1.87618896e-01 -1.19418979e-01] | [15.053646087646484, -1.9701387882232666] |
979faaee-dc17-4387-9766-41488866ac09 | improving-answer-selection-and-answer | null | null | https://aclanthology.org/D19-1604 | https://aclanthology.org/D19-1604.pdf | Improving Answer Selection and Answer Triggering using Hard Negatives | In this paper, we establish the effectiveness of using hard negatives, coupled with a siamese network and a suitable loss function, for the tasks of answer selection and answer triggering. We show that the choice of sampling strategy is key for achieving improved performance on these tasks. Evaluating on recent answer selection datasets - InsuranceQA, SelQA, and an internal QA dataset, we show that using hard negatives with relatively simple model architectures (bag of words and LSTM-CNN) drives significant performance gains. On InsuranceQA, this strategy alone improves over previously reported results by a minimum of 1.6 points in P@1. Using hard negatives with a Transformer encoder provides a further improvement of 2.3 points. Further, we propose to use quadruplet loss for answer triggering, with the aim of producing globally meaningful similarity scores. We show that quadruplet loss function coupled with the selection of hard negatives enables bag-of-words models to improve F1 score by 2.3 points over previous baselines, on SelQA answer triggering dataset. Our results provide key insights into answer selection and answer triggering tasks. | ['Nikhil Rasiwasia', 'Shweta Garg', 'Sawan Kumar', 'Kartik Mehta'] | 2019-11-01 | null | null | null | ijcnlp-2019-11 | ['answer-selection'] | ['natural-language-processing'] | [ 2.27750540e-01 1.87704507e-02 -1.63612366e-01 -5.58721185e-01
-1.69705403e+00 -7.64915228e-01 5.24896622e-01 3.96471322e-01
-7.03874111e-01 7.84839571e-01 2.59569347e-01 -2.33030289e-01
-1.48795277e-01 -7.84507096e-01 -7.92582393e-01 -3.39384824e-01
1.82719290e-01 8.62450898e-01 4.84986812e-01 -7.10471570e-01
4.11090910e-01 1.51889220e-01 -1.50751424e+00 5.88246644e-01
1.08528566e+00 1.24917078e+00 -1.82838634e-01 7.25427985e-01
-2.44338095e-01 1.16999936e+00 -6.86125994e-01 -8.75708580e-01
3.45550537e-01 -5.24771273e-01 -1.15482926e+00 -6.50684178e-01
9.57210064e-01 -4.64046627e-01 -1.47601798e-01 5.42499900e-01
7.84834564e-01 3.29208076e-01 6.21630013e-01 -9.22494948e-01
-5.94012201e-01 4.76172000e-01 -9.23522487e-02 2.12135717e-01
6.14046276e-01 3.33680451e-01 1.78413439e+00 -8.47567499e-01
5.71891427e-01 1.21594656e+00 6.54343009e-01 5.12938023e-01
-1.21953940e+00 -4.86881018e-01 -2.60004967e-01 3.38101029e-01
-9.75189507e-01 -4.86097872e-01 5.23166239e-01 -1.89079400e-02
1.27099442e+00 3.04361612e-01 1.24735266e-01 7.77180791e-01
1.91568583e-01 8.76177371e-01 8.69488478e-01 -3.77033085e-01
2.35324606e-01 6.30109981e-02 2.95739710e-01 8.27575564e-01
-2.45577782e-01 -2.26592049e-01 -7.58847773e-01 -4.30016160e-01
2.39183217e-01 -2.99283445e-01 -1.80234969e-01 -5.25996089e-01
-1.08276486e+00 1.19669342e+00 6.68772340e-01 -3.91900986e-02
-2.48138398e-01 4.53966916e-01 6.39024258e-01 9.36020017e-01
3.55752856e-01 9.89961922e-01 -5.32735348e-01 -2.09798351e-01
-8.53139937e-01 4.84147429e-01 7.65524864e-01 7.47432292e-01
6.21979237e-01 -2.12429091e-01 -7.53057837e-01 9.42649603e-01
-2.99480576e-02 6.89019024e-01 3.08312684e-01 -1.33058226e+00
8.26717317e-01 6.14253163e-01 1.70468897e-01 -8.01976800e-01
-2.12975085e-01 -6.19065583e-01 -2.01721936e-01 -7.40052834e-02
5.01151025e-01 3.81951556e-02 -8.95296693e-01 1.95038140e+00
9.10507590e-02 -4.14057612e-01 -1.22359462e-01 6.36712551e-01
8.15367401e-01 3.97181392e-01 -1.03767566e-01 1.86848029e-01
1.20111322e+00 -1.16672182e+00 -6.19912148e-01 -7.53136203e-02
9.83021319e-01 -6.57719553e-01 1.74235749e+00 3.52509439e-01
-1.56602025e+00 -5.22872746e-01 -1.02986312e+00 -4.16458488e-01
-1.92654818e-01 7.40949288e-02 3.46474558e-01 4.91984844e-01
-1.25568557e+00 6.31563485e-01 -4.85011160e-01 -2.97100306e-01
3.76823694e-01 7.51651049e-01 -1.24826908e-01 -1.02172598e-01
-1.44405484e+00 1.15030527e+00 -1.06381133e-01 -2.52420008e-01
-6.94204330e-01 -7.05963075e-01 -7.79477835e-01 1.35746002e-01
3.35436851e-01 -1.00917089e+00 1.44510376e+00 -7.37266839e-01
-1.39667106e+00 7.89171875e-01 -3.47847164e-01 -9.57011759e-01
5.59285104e-01 -5.32107592e-01 -9.79092121e-02 5.07448554e-01
2.23272592e-01 9.87348199e-01 7.96301603e-01 -7.82101870e-01
-5.04724503e-01 -1.92742690e-01 3.96245271e-01 2.44593799e-01
-3.63120019e-01 7.91975632e-02 -3.40672106e-01 -3.86249036e-01
-1.57065272e-01 -6.67939663e-01 2.93913968e-02 -2.42435247e-01
1.90679519e-03 -5.62959552e-01 2.53154844e-01 -6.82054460e-01
1.24548423e+00 -1.84367704e+00 2.13608727e-01 1.98701814e-01
2.82404959e-01 2.10301012e-01 -5.83864808e-01 3.61855358e-01
2.42589265e-01 -2.30072327e-02 -4.13793921e-01 -3.75829995e-01
1.98095053e-01 4.77802232e-02 -5.20622194e-01 2.15023652e-01
7.43037581e-01 1.22928369e+00 -7.76486337e-01 -3.70038420e-01
-1.25943780e-01 1.85295835e-01 -8.88784170e-01 1.98976889e-01
-4.52757061e-01 -7.20824003e-02 -1.59227893e-01 4.94239748e-01
3.76824945e-01 -4.09197181e-01 -2.21799910e-01 -1.22579455e-01
4.44558293e-01 9.74207699e-01 -6.49377942e-01 1.77532446e+00
-6.76819682e-01 4.34574664e-01 -8.64621904e-03 -5.70469797e-01
9.21092510e-01 1.36388510e-01 1.29367217e-01 -1.21332240e+00
1.17183872e-01 6.51039302e-01 2.66504642e-02 -2.47810006e-01
5.73523223e-01 -1.60162151e-01 -1.33939952e-01 5.59795856e-01
3.58414650e-01 -4.44463968e-01 3.99098724e-01 4.49038267e-01
1.43091714e+00 -4.89867367e-02 -1.73875883e-01 -2.10048392e-01
6.56151354e-01 -4.96790791e-03 2.03322604e-01 8.32819462e-01
-3.23175877e-01 9.05673921e-01 5.15468001e-01 -1.65771589e-01
-7.83124745e-01 -1.23860812e+00 5.77253550e-02 1.39910567e+00
-1.04572415e-01 -3.41624141e-01 -4.07343268e-01 -1.16223502e+00
8.85706991e-02 8.17684114e-01 -6.03964329e-01 -4.76831228e-01
-1.05310154e+00 -4.30224687e-01 6.30059898e-01 5.76319158e-01
4.30168241e-01 -1.01303208e+00 -5.36312819e-01 1.61518440e-01
-3.51908207e-01 -8.78151417e-01 -5.98572731e-01 4.22327965e-01
-8.59971523e-01 -8.39206874e-01 -7.29682863e-01 -6.88478053e-01
2.32686803e-01 6.62525669e-02 1.75307775e+00 1.80787966e-01
2.46989444e-01 2.82455206e-01 -3.02575380e-01 1.13411479e-01
-1.33495092e-01 6.10043168e-01 -3.25745523e-01 -2.77299106e-01
3.62625033e-01 -3.25279713e-01 -6.75345719e-01 1.85434401e-01
-7.79154360e-01 -5.27769804e-01 4.12161529e-01 1.18486619e+00
2.96109021e-01 -5.04432559e-01 1.03140390e+00 -9.55129027e-01
7.89298236e-01 -3.90715301e-01 -1.54626697e-01 3.69371772e-01
-6.15313292e-01 4.91063058e-01 6.42991960e-01 -7.10330009e-02
-9.80669677e-01 -3.68327707e-01 -5.17781198e-01 -1.36829957e-01
3.79624099e-01 1.78353041e-01 1.40937656e-01 -1.61999434e-01
9.23341513e-01 -9.30659249e-02 -3.92080508e-02 -1.99222147e-01
6.09058917e-01 4.18182939e-01 2.37364486e-01 -4.43636388e-01
4.55220044e-01 4.79173630e-01 -2.38502160e-01 6.33275928e-03
-1.02157974e+00 -5.44310808e-01 -3.15008909e-01 3.03689271e-01
7.55115509e-01 -8.19378674e-01 -6.40044391e-01 3.92978303e-02
-9.42006826e-01 -3.15605760e-01 -4.64469463e-01 1.33908987e-01
-6.81482315e-01 1.24498017e-01 -8.40379655e-01 -5.71958661e-01
-6.93590343e-01 -1.16414094e+00 1.13344657e+00 -6.46294206e-02
-5.87264299e-01 -1.00050056e+00 5.99978678e-02 8.70306671e-01
8.56253386e-01 -1.85355231e-01 9.89134908e-01 -9.81736481e-01
-6.52600825e-01 -6.45202547e-02 -2.64328718e-01 5.34782231e-01
-2.22504765e-01 -3.82508278e-01 -1.02766335e+00 -1.39865279e-01
1.32371068e-01 -9.96849120e-01 1.44206357e+00 3.32777388e-02
7.83778131e-01 -6.89319521e-02 1.81849763e-01 2.00980440e-01
1.14607549e+00 6.09276183e-02 4.29190189e-01 2.96373487e-01
4.32404578e-01 6.39490068e-01 5.86412013e-01 6.32393286e-02
5.97875416e-01 7.09822416e-01 4.13490981e-01 5.82137853e-02
-4.10326570e-01 -1.40639022e-01 3.89182657e-01 8.58375907e-01
5.31118691e-01 -1.71380252e-01 -8.69643152e-01 8.60718787e-01
-1.64067852e+00 -6.87647045e-01 -6.95815384e-02 2.40500188e+00
1.21100485e+00 4.60376143e-01 2.43078798e-01 2.85866745e-02
2.32750729e-01 2.24619806e-01 -5.16694248e-01 -6.37088597e-01
-3.53919506e-01 9.77362454e-01 3.62207949e-01 8.73247802e-01
-9.08565760e-01 7.75974989e-01 6.30354404e+00 1.12090540e+00
-9.16820467e-01 2.22065151e-01 6.18601680e-01 -5.26032627e-01
-7.83503771e-01 -1.04818024e-01 -7.92578816e-01 2.78705627e-01
1.14327073e+00 2.99027950e-01 2.53678083e-01 2.78742343e-01
-3.27272832e-01 -7.42357448e-02 -1.18668139e+00 5.57803094e-01
3.27236116e-01 -1.18189335e+00 1.00984409e-01 -4.53206807e-01
7.70808280e-01 1.54813707e-01 2.44486198e-01 6.95798218e-01
3.54243726e-01 -1.22321951e+00 5.97305775e-01 3.08492690e-01
5.51058650e-01 -9.19808805e-01 1.00287163e+00 -1.07286736e-01
-9.37336206e-01 -1.67853981e-01 -1.29078284e-01 -9.78379231e-03
1.54873550e-01 4.57509816e-01 -7.27836549e-01 5.08518338e-01
5.53835392e-01 4.42024738e-01 -7.42785692e-01 9.25738871e-01
-3.41910720e-01 9.00091708e-01 -4.37530786e-01 -2.13231087e-01
3.54489177e-01 1.07991323e-01 2.38369867e-01 9.43759918e-01
4.58792001e-02 -3.22420895e-01 -1.58986434e-01 8.24841142e-01
-4.07389343e-01 1.64040551e-01 -3.14450294e-01 1.42298952e-01
4.89722371e-01 8.68604422e-01 -2.24701807e-01 -2.83775717e-01
-1.44511297e-01 1.03715491e+00 6.35201573e-01 2.70103484e-01
-8.57926428e-01 -8.29133928e-01 3.27075243e-01 -1.68430116e-02
4.82798308e-01 7.40736201e-02 -6.11749768e-01 -9.99409556e-01
3.17415655e-01 -1.24933577e+00 5.90716481e-01 -4.25716490e-01
-1.32003140e+00 5.91518760e-01 -3.83926600e-01 -8.48890483e-01
-3.62198353e-01 -4.62078482e-01 -5.97431719e-01 9.79289770e-01
-1.56736612e+00 -9.39197540e-01 2.23945856e-01 3.12536806e-01
4.98479098e-01 -5.51657528e-02 6.63368940e-01 4.78686690e-01
-2.27597788e-01 8.81776392e-01 -1.60620119e-02 -1.53585821e-01
9.88068044e-01 -1.44703174e+00 4.96809095e-01 3.47988755e-01
2.83284336e-01 7.87342787e-01 6.12022519e-01 -2.72570103e-01
-1.18187582e+00 -7.02437758e-01 1.33706939e+00 -1.01909995e+00
6.97440267e-01 -4.04637069e-01 -7.89955318e-01 3.08640778e-01
5.03555417e-01 -3.39986742e-01 6.09962761e-01 2.57304460e-01
-5.25597572e-01 -2.73343086e-01 -1.33988154e+00 5.00901043e-01
8.72384608e-01 -8.39988589e-01 -8.13927054e-01 3.74438256e-01
1.17987049e+00 -2.69747764e-01 -7.42489934e-01 6.89477146e-01
4.08086300e-01 -9.98903215e-01 1.18095255e+00 -7.53114581e-01
5.83438218e-01 -5.07187506e-04 -3.21342856e-01 -1.08770549e+00
-6.96206838e-02 -4.24844861e-01 -2.71079153e-01 9.31833267e-01
7.14530587e-01 -6.40692651e-01 8.94913495e-01 4.55206811e-01
6.57009855e-02 -1.24802339e+00 -9.57957387e-01 -5.90907156e-01
4.85562563e-01 -2.34661222e-01 4.30797070e-01 4.65869606e-01
-2.54365385e-01 7.31207311e-01 -8.25815275e-03 -4.36331779e-01
3.21394533e-01 1.91408619e-01 5.04367352e-01 -8.84284437e-01
-3.50294352e-01 -6.05827630e-01 -4.28757854e-02 -1.25350142e+00
1.95536375e-01 -8.32488179e-01 -1.19528562e-01 -1.61852562e+00
-3.21095325e-02 -4.48415309e-01 -3.88050467e-01 3.07545632e-01
-4.28368896e-01 3.46334755e-01 2.60813534e-01 -7.43237957e-02
-1.02632344e+00 9.89167452e-01 1.02174664e+00 -1.75409243e-01
1.01373784e-01 -3.26855890e-02 -6.64623082e-01 2.63497800e-01
9.09715235e-01 -3.63333255e-01 -5.85529983e-01 -6.70801699e-01
6.87768459e-01 -1.26509398e-01 2.29448587e-01 -7.99386263e-01
1.85935333e-01 4.00077790e-01 -1.80922523e-02 -6.47417009e-01
6.75609887e-01 -4.02783394e-01 -7.87576854e-01 5.55591583e-01
-8.07613611e-01 2.99755484e-01 2.39626423e-01 3.29818159e-01
-5.06999910e-01 -3.56751531e-01 5.10423243e-01 1.94449443e-02
-2.98204064e-01 -1.97300747e-01 2.18749195e-02 7.96137750e-01
5.24564445e-01 1.06231362e-01 -4.18729782e-01 -6.56532764e-01
-3.83691877e-01 6.84668660e-01 1.66546673e-01 2.24022865e-01
6.44304633e-01 -1.39979053e+00 -8.91598225e-01 1.38410658e-01
5.94602488e-02 -2.69803524e-01 5.30896299e-02 9.87460196e-01
-5.55700302e-01 6.75317705e-01 1.22097790e-01 -4.98781621e-01
-1.26874495e+00 -1.23660406e-02 4.90071565e-01 -7.24085212e-01
6.25704527e-02 1.50717592e+00 -9.77033377e-02 -1.08376539e+00
3.93186390e-01 -4.01236326e-01 -1.23725563e-01 1.17333882e-01
2.23368227e-01 4.49479878e-01 5.27751029e-01 -1.36090934e-01
-3.60198081e-01 3.83484900e-01 -2.80873686e-01 -2.92497247e-01
1.18818247e+00 1.95742007e-02 -7.12971613e-02 2.72483110e-01
1.40992510e+00 2.14659572e-01 -8.90411913e-01 -3.57902706e-01
1.46944463e-01 -2.99631447e-01 -1.87087357e-01 -1.06946278e+00
-8.99214804e-01 1.05183816e+00 5.12300909e-01 1.52562961e-01
1.18188477e+00 3.05127148e-02 1.25619769e+00 6.02324724e-01
3.98227513e-01 -8.80626559e-01 4.22234684e-01 7.85590410e-01
8.88973296e-01 -1.44916677e+00 -4.02822047e-01 -7.52763078e-02
-6.09818995e-01 6.87451422e-01 6.15047216e-01 -2.93616384e-01
2.30251536e-01 -4.08992283e-02 7.58331642e-03 -3.63452852e-01
-1.13464642e+00 -3.15598905e-01 5.43560207e-01 9.68128145e-02
5.98717391e-01 -1.70424864e-01 -3.08874428e-01 2.72899479e-01
-4.36992496e-01 -1.80401787e-01 1.03097305e-01 8.80897999e-01
-6.05665088e-01 -1.35412896e+00 -1.05383858e-01 7.26726294e-01
-5.38874090e-01 -4.43425149e-01 -5.27862906e-01 4.90045458e-01
-1.64738655e-01 1.16091466e+00 3.14798169e-02 -3.03028435e-01
4.00490254e-01 3.06080550e-01 6.06188416e-01 -6.31245732e-01
-1.25381613e+00 -6.31226301e-01 4.95708406e-01 -7.07928896e-01
-1.76233411e-01 -4.82227653e-01 -1.30166280e+00 -2.76961893e-01
-5.23079872e-01 2.54210383e-01 1.62980169e-01 8.97151232e-01
6.22277021e-01 4.26791906e-01 4.23654675e-01 3.24505083e-02
-1.23225534e+00 -8.76436234e-01 -1.19748078e-01 4.68037605e-01
5.14168262e-01 -4.24636424e-01 -5.33627570e-01 -4.67977077e-01] | [11.320474624633789, 8.117355346679688] |
9220d84c-6abb-41f5-970d-48bc989bf534 | simlm-pre-training-with-representation | 2207.02578 | null | https://arxiv.org/abs/2207.02578v2 | https://arxiv.org/pdf/2207.02578v2.pdf | SimLM: Pre-training with Representation Bottleneck for Dense Passage Retrieval | In this paper, we propose SimLM (Similarity matching with Language Model pre-training), a simple yet effective pre-training method for dense passage retrieval. It employs a simple bottleneck architecture that learns to compress the passage information into a dense vector through self-supervised pre-training. We use a replaced language modeling objective, which is inspired by ELECTRA, to improve the sample efficiency and reduce the mismatch of the input distribution between pre-training and fine-tuning. SimLM only requires access to unlabeled corpus, and is more broadly applicable when there are no labeled data or queries. We conduct experiments on several large-scale passage retrieval datasets, and show substantial improvements over strong baselines under various settings. Remarkably, SimLM even outperforms multi-vector approaches such as ColBERTv2 which incurs significantly more storage cost. Our code and model check points are available at https://github.com/microsoft/unilm/tree/master/simlm . | ['Furu Wei', 'Rangan Majumder', 'Daxin Jiang', 'Linjun Yang', 'Binxing Jiao', 'Xiaolong Huang', 'Nan Yang', 'Liang Wang'] | 2022-07-06 | null | null | null | null | ['passage-retrieval'] | ['natural-language-processing'] | [-2.70341575e-01 -5.28387725e-01 -5.42450130e-01 -1.74756750e-01
-1.69681621e+00 -9.07599092e-01 7.70510435e-01 4.71429676e-01
-8.18574488e-01 7.20204771e-01 6.05026841e-01 -5.02095938e-01
8.02042782e-02 -6.66333735e-01 -7.80100524e-01 -3.40451956e-01
1.70049027e-01 8.05892646e-01 2.65471131e-01 -4.54609662e-01
2.07518488e-01 4.94537354e-02 -1.10814261e+00 4.15204674e-01
7.59310067e-01 4.99126464e-01 3.47829521e-01 8.46238554e-01
-3.88386577e-01 7.02542841e-01 -5.03190041e-01 -2.95138806e-01
3.39888215e-01 -3.61824870e-01 -1.16281223e+00 -3.76882136e-01
5.89271247e-01 -4.08884227e-01 -9.24240172e-01 7.19512165e-01
7.75058627e-01 5.84760606e-01 7.67765462e-01 -6.99283302e-01
-7.01521158e-01 7.70783961e-01 -3.67556572e-01 6.20321274e-01
4.62840289e-01 -1.32486671e-02 1.27776170e+00 -1.13263559e+00
4.37613755e-01 1.11041164e+00 3.94086599e-01 4.15412396e-01
-1.18409669e+00 -4.41898763e-01 -2.02029645e-02 2.49071598e-01
-1.54051995e+00 -6.39061272e-01 2.69492745e-01 -1.14886291e-01
1.24053073e+00 3.99591684e-01 2.39155352e-01 9.35113370e-01
-2.64874935e-01 1.36894786e+00 5.70093811e-01 -7.15678513e-01
1.07043758e-01 2.01673746e-01 5.58603942e-01 4.64286119e-01
2.47193668e-02 -2.59593248e-01 -3.31007421e-01 -6.20113730e-01
4.76337612e-01 2.48654351e-01 -4.08069491e-01 -2.23091781e-01
-1.00132477e+00 8.59191298e-01 3.35147291e-01 5.65568149e-01
-8.72437656e-02 -7.14948624e-02 5.80738485e-01 7.20393240e-01
4.12695825e-01 5.23454249e-01 -5.69784403e-01 -2.50918061e-01
-1.24856007e+00 3.18379134e-01 8.30331802e-01 1.17346323e+00
8.03066075e-01 -3.61648858e-01 -6.73536360e-01 1.16439760e+00
1.28525883e-01 6.42156601e-01 7.02518046e-01 -9.04861391e-01
6.07412875e-01 3.44732493e-01 1.64872915e-01 -4.79940653e-01
-2.10180491e-01 -4.98879522e-01 -7.57926643e-01 -7.32437491e-01
1.16639957e-01 6.83364272e-02 -8.69001269e-01 1.64545047e+00
1.63701661e-02 1.78784147e-01 3.32567729e-02 7.87731528e-01
8.45674455e-01 9.32291329e-01 2.73008682e-02 -4.66786735e-02
1.05801451e+00 -1.51038003e+00 -3.27706933e-01 -1.93177864e-01
1.11596954e+00 -1.01191950e+00 1.54188156e+00 -8.77892449e-02
-1.43576491e+00 -4.48970824e-01 -7.61204898e-01 -5.76398492e-01
-3.06719810e-01 1.31072745e-01 6.76317632e-01 2.02929713e-02
-1.35154498e+00 4.65382814e-01 -9.02165413e-01 -4.12837327e-01
2.24163830e-01 1.68088302e-01 -2.32147932e-01 -4.18993056e-01
-1.27777457e+00 5.69831729e-01 3.74633878e-01 -4.61980790e-01
-7.82214105e-01 -8.99480700e-01 -7.74043441e-01 4.14791375e-01
1.84610277e-01 -9.07728851e-01 1.59396827e+00 -2.85946548e-01
-1.22483087e+00 7.23747015e-01 -4.42857027e-01 -4.11415398e-01
2.96443135e-01 -3.60575825e-01 -1.67997807e-01 1.88269109e-01
1.95213687e-02 7.03633308e-01 4.38600719e-01 -9.34127867e-01
-3.29074979e-01 -3.22995521e-03 -1.94924101e-02 3.08195800e-01
-5.99265575e-01 6.23366050e-02 -1.27768672e+00 -6.59969985e-01
-2.72773832e-01 -7.52959669e-01 -3.00697684e-01 -3.10656548e-01
-4.08680886e-01 -6.20460570e-01 2.68937498e-01 -4.46623832e-01
1.60088336e+00 -2.06669807e+00 -1.20239668e-02 2.56295592e-01
1.65274695e-01 5.63514411e-01 -7.77753532e-01 8.76961470e-01
3.52139086e-01 1.46805653e-02 -1.59873724e-01 -8.10342550e-01
2.13690639e-01 -2.34757527e-03 -6.04133725e-01 3.10992807e-01
-3.25591296e-01 1.18552828e+00 -8.77273381e-01 -7.14391410e-01
4.42271009e-02 4.04087096e-01 -8.44835699e-01 5.29382885e-01
-4.37596321e-01 -4.08469625e-02 -4.22796816e-01 4.30817723e-01
5.82866669e-01 -7.16641724e-01 -4.36478034e-02 -5.10310475e-03
2.90201455e-01 8.60427141e-01 -7.73506999e-01 2.30303407e+00
-5.83943665e-01 5.96059561e-01 -1.45595431e-01 -8.29040229e-01
4.79997188e-01 6.54382780e-02 2.58329958e-01 -8.70220304e-01
1.42293144e-02 1.09888799e-01 -6.19239271e-01 -2.91118920e-01
8.88677061e-01 2.77679354e-01 -1.13220677e-01 8.18906903e-01
2.87174493e-01 5.61117288e-03 6.24889553e-01 9.81911242e-01
1.19176126e+00 -1.94262341e-01 3.60590778e-02 -1.65898740e-01
3.85046333e-01 1.38738304e-01 2.63063937e-01 1.25027072e+00
2.08887085e-01 5.92786789e-01 -8.56457055e-02 6.13919459e-02
-1.11578083e+00 -1.25412011e+00 -2.83708959e-03 1.40210235e+00
2.38175422e-01 -8.17948163e-01 -4.75347459e-01 -4.91425633e-01
1.34187192e-01 5.72508276e-01 -1.40848070e-01 -1.99269548e-01
-5.91892719e-01 -4.75879043e-01 5.30116856e-01 5.99709272e-01
1.27398044e-01 -8.39554608e-01 2.89547920e-01 1.62263811e-01
-4.12604809e-01 -9.49162781e-01 -9.87399101e-01 -4.26932052e-02
-8.29763651e-01 -6.75803781e-01 -9.14539039e-01 -1.02407837e+00
6.22716665e-01 7.46729016e-01 1.54784322e+00 5.95772505e-01
-3.62149298e-01 5.41659117e-01 -4.22482103e-01 -2.27315188e-03
-2.48571679e-01 6.08377337e-01 3.84211959e-03 -7.79139996e-01
7.41684914e-01 -4.64603662e-01 -8.51029217e-01 1.72593266e-01
-1.09867537e+00 -1.81741387e-01 4.11123723e-01 9.24686313e-01
6.65690720e-01 -4.08563465e-01 3.63249183e-01 -6.70431733e-01
8.41965020e-01 -6.47026241e-01 -5.08271217e-01 6.41593337e-01
-5.08926094e-01 1.57427236e-01 5.57711065e-01 -3.91199052e-01
-5.69247544e-01 -6.01763487e-01 -3.08205128e-01 -4.90778595e-01
3.42258066e-02 5.46753585e-01 2.52585053e-01 1.21683910e-01
6.44733727e-01 5.91742456e-01 -1.23137638e-01 -8.97644222e-01
5.24055660e-01 9.37586904e-01 4.11793560e-01 -6.62946999e-01
8.41696024e-01 2.43660018e-01 -6.05661273e-01 -6.77210808e-01
-1.05364513e+00 -1.22774363e+00 -3.36777955e-01 3.79931301e-01
2.36695141e-01 -1.31325328e+00 -3.49338114e-01 1.12225674e-01
-9.31073129e-01 -7.62600422e-01 -3.48975033e-01 5.68916619e-01
-3.21432143e-01 6.67702794e-01 -1.15948296e+00 -3.18863541e-01
-9.28932726e-01 -7.03942358e-01 1.02281702e+00 1.50428355e-01
-1.70533553e-01 -1.13525498e+00 5.70203125e-01 4.10567135e-01
7.02273905e-01 -9.22232449e-01 8.20267200e-01 -1.02875948e+00
-6.04402423e-01 -2.66102314e-01 -1.23225771e-01 2.47983143e-01
-8.88830721e-02 -3.47110212e-01 -6.44700050e-01 -9.96447504e-01
-4.76599872e-01 -9.20742631e-01 1.19269812e+00 9.27747637e-02
1.23297322e+00 -4.39466149e-01 -3.74774814e-01 6.13904774e-01
1.46957397e+00 -2.79448956e-01 4.93627757e-01 3.61743659e-01
4.31525499e-01 2.59190500e-02 5.26572168e-01 4.16107267e-01
6.68332160e-01 6.31027400e-01 -3.22088659e-01 -3.67013872e-01
-2.01972842e-01 -5.99804997e-01 2.99097955e-01 1.47018480e+00
4.90834236e-01 -6.21012688e-01 -6.84264123e-01 6.78361595e-01
-1.76146352e+00 -1.06458485e+00 3.58063370e-01 2.29407024e+00
1.12177193e+00 -4.77750488e-02 1.91457242e-01 -3.70910764e-01
2.48279914e-01 1.13255017e-01 -4.61598396e-01 -1.72608662e-02
-2.86024570e-01 3.18111241e-01 4.64399308e-01 9.69104350e-01
-1.06496143e+00 1.21359396e+00 6.52118587e+00 1.33360863e+00
-8.40502203e-01 1.55619889e-01 4.60609615e-01 -5.39232671e-01
-7.68023729e-01 -1.14994586e-01 -9.88757610e-01 3.59238029e-01
9.62711394e-01 -4.49850023e-01 5.15952110e-01 6.04421139e-01
-2.56492496e-02 1.35611162e-01 -1.04473460e+00 1.00791347e+00
1.91671044e-01 -1.61830187e+00 3.87023240e-01 -2.68356502e-01
8.36937308e-01 7.61505306e-01 -5.96583486e-02 8.65418077e-01
4.43278909e-01 -7.37565458e-01 2.34461308e-01 3.25766832e-01
7.25820959e-01 -4.81907398e-01 5.73990703e-01 5.34011662e-01
-1.08549929e+00 1.87083542e-01 -8.14370751e-01 1.79570168e-01
2.16245010e-01 2.52147287e-01 -5.95684588e-01 3.30054879e-01
5.14638603e-01 6.42814934e-01 -6.45728052e-01 1.28851688e+00
-1.97987482e-02 8.73569667e-01 -6.67271137e-01 -9.86148566e-02
3.95059913e-01 -4.27673124e-02 3.66517335e-01 1.71220756e+00
1.34027570e-01 5.93490601e-02 4.84516889e-01 6.30311370e-01
-4.74454820e-01 4.76881355e-01 -4.22449827e-01 -1.13906421e-01
8.82772148e-01 1.02018929e+00 -1.94865301e-01 -7.42435515e-01
-5.28970599e-01 1.14476585e+00 6.61067009e-01 7.39479601e-01
-4.32223916e-01 -5.01462281e-01 5.81321478e-01 8.40704830e-04
4.23132241e-01 -2.00232655e-01 2.44844764e-01 -1.70595753e+00
2.42065545e-03 -8.42717826e-01 6.78753197e-01 -5.20648241e-01
-1.38552332e+00 4.88416225e-01 -5.05724996e-02 -1.14693201e+00
-4.91489977e-01 -1.41491458e-01 -5.80911636e-01 8.93497825e-01
-1.75607073e+00 -8.69130611e-01 4.83620688e-02 8.25442851e-01
7.84097850e-01 -1.53312892e-01 9.16712463e-01 6.22077942e-01
-3.90318990e-01 1.29737127e+00 7.68146992e-01 3.11658412e-01
9.89281535e-01 -1.07032430e+00 5.61596572e-01 6.28206015e-01
4.94750023e-01 1.08592510e+00 4.35479313e-01 -3.04583102e-01
-1.57451689e+00 -9.32894707e-01 1.22847056e+00 -4.52806979e-01
7.27713525e-01 -4.74664837e-01 -1.19817150e+00 7.02807009e-01
3.85206640e-01 -9.62723047e-02 1.06977010e+00 1.79269180e-01
-5.04952967e-01 1.59393594e-01 -7.44261324e-01 5.64415276e-01
8.70965958e-01 -9.36483204e-01 -6.96292818e-01 7.38105595e-01
9.59116518e-01 -3.00237149e-01 -8.92513573e-01 2.01339662e-01
1.97379321e-01 -5.24178505e-01 1.27604210e+00 -9.80914056e-01
1.46312460e-01 -3.89472991e-02 -2.34659344e-01 -1.18792784e+00
-5.91269672e-01 -6.72773719e-01 -4.41927582e-01 1.17967653e+00
6.97099626e-01 -3.94576788e-01 6.84062898e-01 4.85814333e-01
4.23225537e-02 -9.10005450e-01 -5.63471138e-01 -9.29531574e-01
5.43426335e-01 -2.27884471e-01 5.58485925e-01 8.97512674e-01
3.18807721e-01 4.89741713e-01 -1.75853267e-01 -1.17163263e-01
6.35722518e-01 4.37571824e-01 8.62144232e-01 -5.29330909e-01
-6.31183565e-01 -3.74074429e-01 -9.53637809e-03 -1.91672647e+00
3.06427032e-01 -1.29835105e+00 -1.30042225e-01 -1.66620028e+00
7.57055640e-01 -5.38901806e-01 -6.56318605e-01 4.75719750e-01
-4.69693124e-01 1.95534721e-01 2.14853659e-01 5.91130972e-01
-1.28698170e+00 8.25020790e-01 9.24407780e-01 -2.82320917e-01
-1.77810937e-01 -2.07642117e-03 -6.93283558e-01 2.41624638e-01
9.10844982e-01 -4.14547086e-01 -4.62654203e-01 -9.10281420e-01
2.84762681e-02 9.01399478e-02 -5.99671789e-02 -6.75907850e-01
5.46463966e-01 1.53940603e-01 1.22405641e-01 -6.92378163e-01
2.79064476e-01 -3.88545513e-01 -4.61533815e-01 1.96486428e-01
-9.15165901e-01 2.38385439e-01 3.72292489e-01 3.90583515e-01
-2.79295176e-01 -2.34364003e-01 4.64362025e-01 -1.17657334e-01
-5.48194885e-01 5.86259067e-01 -2.68550724e-01 6.01221502e-01
3.48356754e-01 4.38748986e-01 -5.32440066e-01 -6.15801811e-01
-3.26583236e-01 7.45441139e-01 6.15236580e-01 3.77411574e-01
5.14808178e-01 -1.41068089e+00 -8.61645758e-01 4.43744473e-02
2.66707301e-01 -9.04229581e-02 2.87293792e-01 5.47635972e-01
-3.31123978e-01 8.74816597e-01 5.18877447e-01 -4.14505690e-01
-1.22822797e+00 6.76220655e-01 9.20672789e-02 -5.96849322e-01
-6.41714454e-01 1.05858409e+00 -1.10721905e-02 -6.24240816e-01
5.88912487e-01 -1.37223482e-01 3.90725918e-02 -3.72835129e-01
1.07306731e+00 2.61220634e-01 4.35261764e-02 -2.92481482e-01
-2.22355723e-01 4.09920931e-01 -6.39065266e-01 -1.95027053e-01
1.18992293e+00 -3.73108387e-01 -4.92177345e-02 2.98902690e-01
1.87532711e+00 -5.56519330e-02 -7.95142353e-01 -9.93495107e-01
-7.90628940e-02 -4.28036928e-01 2.45077997e-01 -6.25627220e-01
-7.59846985e-01 8.36755693e-01 2.53219426e-01 -2.22690508e-01
8.61489296e-01 3.09739381e-01 1.26565194e+00 1.01218188e+00
4.09593910e-01 -9.41844583e-01 6.82482049e-02 7.54305303e-01
6.16304934e-01 -1.31899929e+00 2.71984190e-02 1.14594072e-01
-3.13575894e-01 7.05753028e-01 5.41028917e-01 -1.04255147e-01
6.85332239e-01 1.88253090e-01 2.03133047e-01 -2.25354210e-02
-1.13399267e+00 -2.06204489e-01 5.24534225e-01 1.29184965e-02
5.13680160e-01 -1.99642032e-01 -2.27434561e-01 2.96773404e-01
-1.85346320e-01 1.26955852e-01 9.08132419e-02 1.06147826e+00
-5.86604118e-01 -1.24002826e+00 -2.34432369e-02 6.55372679e-01
-5.98884225e-01 -6.45612299e-01 -9.64567661e-02 3.73174638e-01
-7.44239271e-01 9.27526712e-01 2.50755131e-01 -2.85466790e-01
1.64632812e-01 -3.46368365e-02 4.53342706e-01 -6.59102559e-01
-5.78139424e-01 1.83953270e-01 -5.48326112e-02 -6.65648818e-01
-5.05770594e-02 -3.75795126e-01 -1.22446811e+00 -5.23956597e-01
-3.87072086e-01 8.13337088e-01 3.26067150e-01 8.09569359e-01
6.44156814e-01 -5.88787869e-02 7.32200325e-01 -6.19962692e-01
-9.28924084e-01 -1.03996801e+00 -3.31464440e-01 4.80459780e-01
4.46735263e-01 -8.80337656e-02 -5.44379592e-01 -2.58222997e-01] | [11.430153846740723, 7.754085540771484] |
34ec19f6-8418-48ec-995c-b7cd9290c199 | chatgpt-may-pass-the-bar-exam-soon-but-has-a | 2304.12202 | null | https://arxiv.org/abs/2304.12202v1 | https://arxiv.org/pdf/2304.12202v1.pdf | ChatGPT may Pass the Bar Exam soon, but has a Long Way to Go for the LexGLUE benchmark | Following the hype around OpenAI's ChatGPT conversational agent, the last straw in the recent development of Large Language Models (LLMs) that demonstrate emergent unprecedented zero-shot capabilities, we audit the latest OpenAI's GPT-3.5 model, `gpt-3.5-turbo', the first available ChatGPT model, in the LexGLUE benchmark in a zero-shot fashion providing examples in a templated instruction-following format. The results indicate that ChatGPT achieves an average micro-F1 score of 47.6% across LexGLUE tasks, surpassing the baseline guessing rates. Notably, the model performs exceptionally well in some datasets, achieving micro-F1 scores of 62.8% and 70.2% in the ECtHR B and LEDGAR datasets, respectively. The code base and model predictions are available for review on https://github.com/coastalcph/zeroshot_lexglue. | ['Ilias Chalkidis'] | 2023-03-09 | null | null | null | null | ['instruction-following'] | ['natural-language-processing'] | [-4.14922595e-01 4.53110695e-01 -3.53798181e-01 -1.51472986e-01
-1.24243724e+00 -5.97428799e-01 7.88831413e-01 -1.44127965e-01
-2.05091029e-01 6.51604414e-01 4.56697106e-01 -1.07992661e+00
3.43620777e-01 -2.60432005e-01 -7.34632492e-01 -3.44178855e-01
-3.69227976e-02 7.51352668e-01 1.44498453e-01 -8.76284420e-01
5.56796432e-01 -5.88318288e-01 -1.19251895e+00 6.13898337e-01
9.81067598e-01 3.84637684e-01 2.73750305e-01 1.04750073e+00
-2.78936625e-01 1.47970212e+00 -6.19092107e-01 -7.00081646e-01
4.02459167e-02 -1.13759339e-01 -1.26525950e+00 -6.51872396e-01
3.58016074e-01 -4.45356846e-01 -5.00829160e-01 4.56022173e-01
4.05761570e-01 6.61718771e-02 1.61634788e-01 -1.21122968e+00
-5.73144972e-01 1.23645353e+00 -3.91641438e-01 6.45820647e-02
6.30630136e-01 6.16766632e-01 1.09007168e+00 -8.25615346e-01
5.24319291e-01 1.10916245e+00 6.61495090e-01 7.05653965e-01
-9.73611295e-01 -5.89152217e-01 -1.44424960e-01 6.17379956e-02
-1.28485930e+00 -5.21197140e-01 2.27019172e-02 -3.32779676e-01
1.73747742e+00 3.19513798e-01 2.93969333e-01 1.57597756e+00
7.53824890e-01 1.21941817e+00 1.32085729e+00 -3.82414013e-01
-7.16005936e-02 -1.08865388e-01 4.42163944e-01 8.80420923e-01
-3.90166342e-01 5.77116422e-02 -8.10982645e-01 -3.18729997e-01
1.33645147e-01 -2.31047824e-01 -1.83219805e-01 2.46619880e-01
-1.44873583e+00 7.75277257e-01 6.87830523e-02 3.17354202e-01
-6.46796376e-02 3.86798292e-01 6.05956018e-01 6.61134601e-01
5.56743562e-01 6.95091665e-01 -7.23500550e-01 -1.23758864e+00
-4.72630590e-01 1.76385388e-01 1.38860464e+00 1.33090961e+00
5.70884943e-01 -9.30609331e-02 -1.80094689e-01 7.89850235e-01
-4.17675264e-02 1.60235286e-01 7.15941966e-01 -1.15305090e+00
8.91141891e-01 1.85389072e-01 1.65760536e-02 -2.68472344e-01
-1.13750972e-01 -2.45416895e-01 -1.00042537e-01 -1.95584252e-01
5.90254009e-01 -3.22378188e-01 -3.56484234e-01 1.52650023e+00
-1.34855226e-01 2.33889520e-01 3.70672613e-01 3.34688574e-01
1.03921068e+00 1.05493891e+00 2.53571391e-01 1.23445757e-01
1.24162841e+00 -1.56143892e+00 -6.52740479e-01 -5.22679925e-01
1.44768250e+00 -9.99256015e-01 1.50603664e+00 5.31390786e-01
-9.90372300e-01 -3.40211213e-01 -8.93575788e-01 -2.84204304e-01
-3.77627760e-01 -2.02723622e-01 7.62007475e-01 6.19514942e-01
-1.28667879e+00 4.96000856e-01 -8.71600449e-01 -3.07933003e-01
-7.41257444e-02 9.89245251e-02 -2.18668699e-01 -1.91788942e-01
-1.18747389e+00 9.87055242e-01 3.21858287e-01 -2.95810342e-01
-1.19545281e+00 -7.53292978e-01 -5.72806537e-01 9.69854966e-02
8.22970629e-01 -1.59553126e-01 2.23453450e+00 -3.00434083e-01
-2.01543975e+00 6.88637614e-01 1.10851591e-02 -5.74312687e-01
5.50357640e-01 -5.35027325e-01 -7.13812485e-02 -3.25318635e-01
-1.23750642e-01 6.26503170e-01 1.36952788e-01 -7.78849900e-01
-4.63342935e-01 2.59512126e-01 2.96327233e-01 2.30276316e-01
2.40643159e-01 1.60768703e-01 -2.85639554e-01 -4.99677509e-02
-5.58354318e-01 -9.73298252e-01 -1.54762670e-01 -8.73855352e-01
-4.30449665e-01 -6.33252442e-01 4.74579602e-01 -6.73552752e-01
1.28667569e+00 -1.87641633e+00 -4.10514027e-02 -4.94273245e-01
4.78247166e-01 3.18756104e-01 -3.51725191e-01 1.14669883e+00
1.54027253e-01 2.55701423e-01 1.40035763e-01 -6.00254953e-01
2.93537080e-01 2.53225386e-01 -2.36394793e-01 -6.40129372e-02
-1.36514321e-01 1.22296572e+00 -1.26651168e+00 -2.32253551e-01
2.78569758e-01 8.52276012e-02 -6.45893753e-01 5.50186276e-01
-6.34949327e-01 1.77365959e-01 -2.58338362e-01 4.99826342e-01
8.70117322e-02 -2.57040083e-01 2.12821037e-01 6.50893629e-01
-4.59953338e-01 1.10403204e+00 -2.86071986e-01 1.97055387e+00
-7.72178590e-01 8.14955473e-01 6.49247691e-02 -3.27366531e-01
7.81679153e-01 5.34241080e-01 1.12283774e-01 -6.29993081e-01
2.59834260e-01 5.06848514e-01 2.73740381e-01 -4.81719494e-01
1.01462126e+00 2.51253128e-01 -5.35130501e-01 8.07455897e-01
3.46412480e-01 -3.90604615e-01 6.39435425e-02 7.64356077e-01
1.31848276e+00 1.80522203e-01 2.72823870e-01 -5.20597219e-01
6.43794239e-02 8.21521953e-02 -1.00317985e-01 1.01965666e+00
-2.05646500e-01 2.81768650e-01 6.81002259e-01 -2.52227157e-01
-7.15264201e-01 -5.40776849e-01 1.43193573e-01 1.65853631e+00
-3.59233260e-01 -1.13041270e+00 -1.03491712e+00 -6.61721647e-01
-3.62515152e-01 1.34906864e+00 -4.01127845e-01 -1.80142641e-01
-6.42388284e-01 -3.09284121e-01 7.18122900e-01 1.54997379e-01
5.69545209e-01 -9.09439385e-01 -3.62203687e-01 3.62635821e-01
-7.09541559e-01 -1.01124513e+00 -7.49064445e-01 4.25802082e-01
-4.23999697e-01 -7.77208567e-01 -3.22336912e-01 -6.43274188e-01
-1.20997094e-01 3.73979717e-01 1.56278920e+00 4.52818871e-01
2.59275764e-01 2.64904797e-01 -6.77374482e-01 -1.75456926e-01
-1.00592244e+00 5.76615036e-01 -2.79713303e-01 -6.89169347e-01
6.49563611e-01 -2.61468381e-01 -3.59763294e-01 5.07011533e-01
-1.59540817e-01 5.24857759e-01 3.93740028e-01 9.59512293e-01
-2.27539659e-01 -7.79978216e-01 5.50352514e-01 -1.07064366e+00
7.71363914e-01 -9.23277676e-01 -3.19717854e-01 2.14547604e-01
-5.70058763e-01 -1.14682995e-01 5.35644412e-01 -4.04021710e-01
-1.06151891e+00 -7.16076791e-01 -5.50795197e-01 -1.73738431e-02
-7.64208138e-02 6.78009927e-01 3.32828969e-01 1.26740988e-02
7.36951590e-01 3.73782754e-01 1.17796483e-02 -3.03662956e-01
4.41152573e-01 1.07993853e+00 4.01015133e-01 -9.63833272e-01
1.49814039e-01 -4.33283806e-01 -6.92932189e-01 -6.54637635e-01
-7.69587040e-01 -5.20281434e-01 -2.50573486e-01 -1.35624096e-01
6.31001711e-01 -1.04641461e+00 -9.81066048e-01 5.60160398e-01
-1.07662427e+00 -1.36158586e+00 6.30135164e-02 2.04727203e-01
-8.06161165e-01 3.43479484e-01 -1.26857984e+00 -6.30207300e-01
-5.54671407e-01 -1.68064547e+00 9.44576263e-01 4.61833961e-02
-6.12987757e-01 -1.05615103e+00 2.06225440e-01 8.63209546e-01
6.20088458e-01 -4.72405910e-01 8.26142550e-01 -1.14029467e+00
-6.33285761e-01 -7.51815811e-02 9.85591561e-02 1.78162649e-01
-2.72560477e-01 -3.34160887e-02 -9.55155671e-01 -2.48475701e-01
-1.69548430e-02 -8.60150695e-01 3.91871035e-01 -1.81098998e-01
7.72659361e-01 -5.95623672e-01 -2.40618095e-01 3.21622550e-01
1.11967671e+00 2.38412574e-01 6.36693120e-01 2.73764163e-01
5.63534677e-01 3.06336224e-01 5.49452245e-01 2.30819732e-01
7.55720377e-01 6.04368925e-01 3.18449706e-01 6.66855395e-01
-7.54774213e-02 -6.48136318e-01 8.08234274e-01 1.61440313e+00
9.79565531e-02 -6.40861154e-01 -1.40281224e+00 4.31900620e-01
-1.88741529e+00 -7.15687335e-01 -4.21855479e-01 1.97763288e+00
1.12103069e+00 2.98668116e-01 -1.09649293e-01 -4.93726730e-01
3.73419374e-02 3.75184566e-01 -3.19197364e-02 -1.03442454e+00
2.96501726e-01 2.31063277e-01 3.29881489e-01 1.01422620e+00
-4.66451019e-01 1.36776280e+00 6.53358221e+00 1.14671087e+00
-7.32148826e-01 3.40630263e-01 7.18861818e-01 1.23098098e-01
-1.68343142e-01 2.66823471e-01 -9.82511163e-01 6.01776481e-01
1.92445469e+00 -4.79359567e-01 6.96484625e-01 8.41844320e-01
-6.45816624e-02 -2.23558143e-01 -9.96638119e-01 6.14937544e-01
-3.96393426e-03 -1.29724252e+00 -3.64602178e-01 1.19246565e-01
7.85513461e-01 7.45683432e-01 6.62635565e-02 1.30984068e+00
1.17517328e+00 -1.08266544e+00 6.22303843e-01 6.91557527e-02
6.14591956e-01 -5.98981977e-01 7.61891186e-01 9.75380003e-01
-7.62895525e-01 4.33888882e-02 -2.07986325e-01 -4.98710960e-01
5.10241017e-02 -2.55158365e-01 -1.40749919e+00 4.39118624e-01
4.75410730e-01 6.03386343e-01 -4.38427091e-01 6.77846551e-01
-2.96595097e-01 1.07623625e+00 -2.24797294e-01 -3.91405970e-01
6.43082380e-01 -1.60545945e-01 5.04606307e-01 1.31042814e+00
2.17088789e-01 3.81381422e-01 2.87284613e-01 6.88282967e-01
-2.33290449e-01 3.11479159e-02 -4.59247619e-01 -1.36896640e-01
6.83625519e-01 1.23605359e+00 -5.74049130e-02 -5.32756746e-01
-6.83282971e-01 7.30024993e-01 5.38732708e-01 8.83623213e-02
-8.65901470e-01 -1.61913782e-01 5.74652016e-01 -8.94485712e-02
-9.42930579e-02 -4.15479392e-01 7.56628588e-02 -1.20338893e+00
-4.10685986e-01 -1.50725842e+00 -1.52037917e-02 -1.01844680e+00
-1.04992998e+00 7.96974242e-01 1.48844257e-01 -8.71242821e-01
-6.25766098e-01 -6.18974328e-01 -1.00729239e+00 7.56911278e-01
-1.07982206e+00 -1.01031542e+00 -3.00903600e-02 3.38471651e-01
1.09776080e+00 -8.94086286e-02 9.73177850e-01 1.18717283e-01
-5.21924376e-01 8.18044782e-01 2.07216680e-01 -7.70430937e-02
7.81989932e-01 -1.46388638e+00 1.05145395e+00 3.53288084e-01
-2.32015345e-02 9.33973253e-01 9.46979523e-01 -5.98412633e-01
-1.68434501e+00 -6.45776808e-01 1.13009953e+00 -1.11830842e+00
1.27995825e+00 -6.80487692e-01 -1.11394119e+00 1.25679791e+00
1.07896245e+00 -6.26248062e-01 6.24016225e-01 3.01048070e-01
-4.59380746e-01 6.42116964e-01 -6.13430321e-01 8.91314328e-01
7.64748812e-01 -6.52127206e-01 -7.63166964e-01 6.09000921e-01
1.10904503e+00 -1.00681150e+00 -1.14026153e+00 -1.24527022e-01
4.47439849e-01 -1.01475394e+00 5.82848430e-01 -6.12878442e-01
6.22650027e-01 4.67935622e-01 -2.85535842e-01 -1.55656886e+00
6.03909791e-02 -1.40664685e+00 -4.90954481e-02 9.32579577e-01
7.32426763e-01 -8.57608378e-01 6.80669665e-01 5.01878679e-01
-8.38867903e-01 -9.88865256e-01 -9.22518671e-01 -7.24329829e-01
6.45248353e-01 -4.41223890e-01 4.76781785e-01 1.00001824e+00
8.08306038e-01 5.90100408e-01 -5.71646273e-01 -5.16164720e-01
2.48074442e-01 -2.20237285e-01 1.22396541e+00 -6.96869850e-01
-7.03607678e-01 -3.81219029e-01 4.08717722e-01 -1.52216244e+00
2.13061363e-01 -9.95285392e-01 2.79903561e-01 -1.18623173e+00
4.07291502e-01 -4.15424317e-01 1.59113809e-01 7.72402942e-01
-1.48643866e-01 -2.21234977e-01 3.02932262e-01 1.32584989e-01
-8.87926757e-01 4.90836799e-01 1.16373837e+00 -4.94457502e-03
-6.16575815e-02 1.08880810e-02 -6.21093512e-01 4.17238027e-01
1.22074950e+00 -2.16184676e-01 -3.61182898e-01 -5.69638729e-01
1.30259737e-01 6.16603971e-01 -1.13733605e-01 -9.38959539e-01
2.27530390e-01 -1.84563518e-01 -7.63446808e-01 -3.68928611e-01
4.53225017e-01 -1.55002728e-01 -2.63888407e-02 4.71044004e-01
-5.72783113e-01 4.01119202e-01 4.79156464e-01 9.19509679e-02
-1.01293370e-01 -3.11897218e-01 2.83976138e-01 -3.54353756e-01
-8.22743595e-01 3.27902772e-02 -9.65295911e-01 2.33037457e-01
6.99676573e-01 -4.77872565e-02 -1.21204877e+00 -7.30857790e-01
-2.25733399e-01 5.43615758e-01 3.34134668e-01 7.15529859e-01
2.29256615e-01 -7.92684078e-01 -7.89704502e-01 -7.73683339e-02
3.53433132e-01 -3.35591406e-01 3.86313558e-01 1.11605752e+00
-5.52568555e-01 7.72097945e-01 1.25185475e-02 -4.76930767e-01
-1.19809401e+00 3.99633437e-01 4.40746725e-01 -6.46563709e-01
-4.32719320e-01 1.01763463e+00 6.74509853e-02 -8.98034275e-01
-1.02030434e-01 -4.78055954e-01 3.37583363e-01 -5.62425196e-01
5.33285975e-01 3.42755079e-01 2.24865749e-01 -4.12171751e-01
1.43092796e-01 -1.68058097e-01 -5.07986069e-01 -1.01784699e-01
9.95716214e-01 -9.86938477e-02 -2.55181454e-02 7.16641068e-01
1.21845067e+00 1.43105045e-01 -9.73986268e-01 -3.10251415e-01
3.67361233e-02 -2.45454147e-01 -1.65059149e-01 -1.10743332e+00
-3.66951585e-01 9.05505717e-01 -9.84476805e-02 1.43042892e-01
1.44886538e-01 7.84808211e-03 8.93216133e-01 5.65287471e-01
9.76779044e-01 -6.23815536e-01 3.32228243e-01 1.29914820e+00
7.76365340e-01 -1.26193857e+00 -3.89346123e-01 -1.64249882e-01
-8.23469937e-01 7.98136592e-01 9.44975913e-01 2.04609543e-01
2.69472897e-01 3.47907752e-01 1.97786272e-01 -2.61916846e-01
-1.78083599e+00 3.05969149e-01 -2.05392480e-01 2.10741162e-01
9.77810502e-01 3.60250801e-01 -2.25947335e-01 5.77369511e-01
-7.13577867e-01 -1.53247356e-01 9.43858027e-01 1.01852405e+00
-5.66304684e-01 -1.12467384e+00 2.04626024e-02 3.57159078e-01
-5.08734405e-01 -5.67060769e-01 -2.67672151e-01 9.75787640e-01
-6.33793414e-01 1.21641970e+00 -9.95875746e-02 -6.98639274e-01
3.98455635e-02 4.71039176e-01 1.89244121e-01 -8.11561465e-01
-1.17410398e+00 -1.39041781e-01 7.45391428e-01 -6.44654691e-01
3.41397911e-01 -3.64098340e-01 -1.00272751e+00 -1.09952927e+00
-4.36528742e-01 6.26500368e-01 3.22867274e-01 8.19534063e-01
4.06743586e-01 2.75859803e-01 2.34924495e-01 -7.55557895e-01
-7.50360310e-01 -1.42716968e+00 -8.93701091e-02 -1.55949742e-01
4.16620411e-02 -2.32266888e-01 -4.40837175e-01 -4.55668569e-01] | [11.837773323059082, 8.313521385192871] |
66c64c55-eda0-44f7-b7ec-276f404a6a4f | o-cnn-octree-based-convolutional-neural | 1712.01537 | null | http://arxiv.org/abs/1712.01537v1 | http://arxiv.org/pdf/1712.01537v1.pdf | O-CNN: Octree-based Convolutional Neural Networks for 3D Shape Analysis | We present O-CNN, an Octree-based Convolutional Neural Network (CNN) for 3D
shape analysis. Built upon the octree representation of 3D shapes, our method
takes the average normal vectors of a 3D model sampled in the finest leaf
octants as input and performs 3D CNN operations on the octants occupied by the
3D shape surface. We design a novel octree data structure to efficiently store
the octant information and CNN features into the graphics memory and execute
the entire O-CNN training and evaluation on the GPU. O-CNN supports various CNN
structures and works for 3D shapes in different representations. By restraining
the computations on the octants occupied by 3D surfaces, the memory and
computational costs of the O-CNN grow quadratically as the depth of the octree
increases, which makes the 3D CNN feasible for high-resolution 3D models. We
compare the performance of the O-CNN with other existing 3D CNN solutions and
demonstrate the efficiency and efficacy of O-CNN in three shape analysis tasks,
including object classification, shape retrieval, and shape segmentation. | ['Chun-Yu Sun', 'Yu-Xiao Guo', 'Peng-Shuai Wang', 'Yang Liu', 'Xin Tong'] | 2017-12-05 | null | null | null | null | ['3d-object-classification'] | ['computer-vision'] | [-3.85348052e-01 -7.77009055e-02 1.22473978e-01 -2.83224404e-01
-1.33017659e-01 -5.93788028e-01 9.72766206e-02 2.81064510e-01
-9.03269276e-02 -1.24132648e-01 -2.50362959e-02 -3.69000673e-01
1.42316461e-01 -1.49849534e+00 -5.73948443e-01 -1.90562978e-01
-3.93093109e-01 8.11147273e-01 5.65026164e-01 -1.13112079e-02
1.75583437e-01 1.66708016e+00 -1.47997761e+00 4.83834773e-01
4.88960266e-01 1.48213685e+00 -3.76715988e-01 6.92968249e-01
-5.21195710e-01 -6.90186992e-02 -3.77695620e-01 -9.16884989e-02
4.29484665e-01 4.65364724e-01 -6.35681033e-01 2.10052922e-01
7.35308647e-01 -6.44015610e-01 -4.25987303e-01 6.89653456e-01
8.06443870e-01 -1.25116453e-01 5.66605866e-01 -1.05446029e+00
-6.59122765e-01 -1.23749323e-01 -5.68427265e-01 4.26851995e-02
8.40110779e-02 -5.34694083e-02 6.22941434e-01 -1.38420546e+00
5.34309804e-01 1.56311309e+00 1.02867031e+00 1.91817522e-01
-9.00051534e-01 -3.50335270e-01 -3.68476920e-02 -6.48435175e-01
-1.40747607e+00 -2.69752145e-02 7.07675278e-01 -2.90894717e-01
1.53081489e+00 3.14232200e-01 1.36560237e+00 3.04952115e-02
1.24712385e-01 9.29181099e-01 4.13396209e-01 -8.06182250e-02
3.72579247e-01 -6.90444112e-01 7.20736012e-02 1.14034915e+00
3.95390779e-01 4.21176590e-02 -2.21759290e-01 -5.83093047e-01
1.62006164e+00 8.30633044e-02 3.67478542e-02 -7.43634224e-01
-8.10620427e-01 6.29595578e-01 8.26716840e-01 -7.26652890e-02
-2.58909255e-01 3.52455467e-01 5.37110507e-01 1.15502067e-01
9.73230362e-01 -2.65058540e-02 -6.69463158e-01 5.96130602e-02
-7.65205681e-01 3.81307065e-01 7.19978034e-01 1.30833900e+00
9.61473525e-01 3.20265442e-01 1.59644093e-02 8.08567822e-01
1.72356829e-01 4.65650588e-01 -2.31571510e-01 -9.30985153e-01
3.31389219e-01 1.38172019e+00 -1.80774361e-01 -1.07813334e+00
-5.26723444e-01 -4.31136131e-01 -1.00703359e+00 3.37568969e-01
1.76361464e-02 8.20190907e-02 -1.19926012e+00 8.64606202e-01
8.29404414e-01 -6.80746362e-02 -4.64467466e-01 7.82033503e-01
1.38799858e+00 5.31903684e-01 -4.08520430e-01 5.03293455e-01
1.17635453e+00 -7.66352177e-01 1.26983270e-01 2.28296563e-01
8.50075185e-01 -6.44212067e-01 9.32345092e-01 -2.05158606e-01
-1.39115083e+00 -5.04205167e-01 -9.70953941e-01 -4.86550391e-01
-3.47391576e-01 2.41695166e-01 8.99308205e-01 5.51899910e-01
-1.35781765e+00 6.40074253e-01 -1.03160119e+00 -6.82824478e-02
1.06866348e+00 6.55299664e-01 -2.41067410e-01 -1.41301557e-01
-4.33053046e-01 2.21497133e-01 2.61010349e-01 3.21172506e-01
-6.10331833e-01 -8.52198541e-01 -9.86889422e-01 2.58446455e-01
-2.20086202e-01 -8.10156167e-01 1.09117806e+00 -8.54783058e-02
-1.14268959e+00 1.25291955e+00 -6.96843192e-02 4.98978086e-02
1.31543860e-01 3.55814137e-02 3.41521412e-01 1.64917093e-02
-4.95167971e-02 6.26804411e-01 6.73700511e-01 -8.56147945e-01
-2.52852052e-01 -9.61085856e-01 1.64507151e-01 2.54157454e-01
4.09890488e-02 -2.39647508e-01 -7.78253853e-01 -4.41360533e-01
8.34159970e-01 -6.40046775e-01 -3.88175637e-01 7.57397115e-01
-2.33807474e-01 -4.29476410e-01 1.07971966e+00 5.59976026e-02
7.36207426e-01 -2.33294940e+00 -3.13510627e-01 4.18136656e-01
6.55525744e-01 2.54056841e-01 -1.37023836e-01 5.93563821e-03
9.99721438e-02 2.71262139e-01 -2.46595457e-01 -2.51400083e-01
-1.67669654e-01 3.71674001e-01 -2.63036694e-02 4.10605580e-01
4.55222189e-01 1.20464456e+00 -6.61492705e-01 -3.25242579e-01
-3.35985497e-02 7.83160865e-01 -8.78272176e-01 1.37971729e-01
-2.13038579e-01 -8.67062286e-02 -9.91950035e-01 8.97819817e-01
1.24502289e+00 -3.39340687e-01 -3.83886248e-01 -2.38284960e-01
-2.83796489e-01 3.98847550e-01 -9.97912884e-01 1.60066235e+00
-1.75976321e-01 3.11054856e-01 1.51841834e-01 -6.59242988e-01
1.30308712e+00 -2.29991991e-02 5.86943567e-01 -4.44741994e-01
8.08093399e-02 3.65306079e-01 -4.98373568e-01 1.95312023e-01
2.72233218e-01 4.15935248e-01 7.12293908e-02 6.56640172e-01
-3.41513395e-01 -8.43544126e-01 -3.29109520e-01 -2.04370871e-01
8.94789994e-01 2.51802262e-02 1.31237194e-01 -3.27773243e-01
1.29515722e-01 2.23844692e-01 3.69288266e-01 5.20177841e-01
9.25711915e-02 4.94882643e-01 5.82787395e-01 -1.40791452e+00
-1.15715873e+00 -1.12387502e+00 -2.77401388e-01 7.93339372e-01
-1.02882400e-01 -3.83124441e-01 -5.72834313e-01 -4.41929638e-01
4.85302597e-01 -1.48632154e-01 -5.48707902e-01 3.43411773e-01
-9.07381654e-01 -4.82940018e-01 5.06134450e-01 8.07686329e-01
6.96534753e-01 -1.04769838e+00 -1.02968156e+00 2.49471217e-01
7.03298151e-01 -9.57420051e-01 -6.49642169e-01 2.19014034e-01
-1.70739770e+00 -9.99302983e-01 -6.50342941e-01 -1.03869712e+00
1.06883621e+00 4.56607401e-01 1.25688624e+00 3.38450015e-01
-5.67142367e-01 3.28789860e-01 -1.25648767e-01 -5.41979253e-01
1.84710905e-01 2.46520191e-01 -3.85592282e-01 -4.60182458e-01
3.62477571e-01 -8.71306717e-01 -5.35108685e-01 1.61996260e-01
-6.44059062e-01 -1.18516963e-02 3.63726430e-02 6.36948705e-01
1.21963274e+00 -1.58515461e-02 -1.10871844e-01 -7.64729559e-01
3.41577858e-01 -2.09912863e-02 -9.43312526e-01 -1.28160000e-01
1.10841140e-01 2.14928808e-03 3.05391252e-01 -1.99468374e-01
-3.64231527e-01 4.13355410e-01 -3.39849889e-01 -7.41818428e-01
-2.80412640e-02 1.45389482e-01 -8.29366297e-02 -3.80693257e-01
3.65053862e-01 1.60288244e-01 -1.90450057e-01 -6.50583625e-01
2.19727889e-01 3.39217871e-01 1.78556949e-01 -7.03775048e-01
6.37416720e-01 7.71920860e-01 3.26227129e-01 -1.16030586e+00
-9.10026252e-01 -1.55209273e-01 -1.00010121e+00 -5.23737967e-02
8.25974047e-01 -8.42441857e-01 -9.37480807e-01 7.46712744e-01
-1.72921813e+00 -3.83393705e-01 -5.38398147e-01 -6.35802969e-02
-6.19277298e-01 1.34714857e-01 -6.28629684e-01 -5.91266513e-01
-6.89292908e-01 -1.01351440e+00 1.57324529e+00 -4.98729572e-02
1.61366537e-01 -7.64294267e-01 -1.00723408e-01 -2.72118866e-01
2.67586291e-01 5.14008641e-01 1.54405081e+00 -2.66804308e-01
-9.66824532e-01 -2.78094590e-01 -7.01792598e-01 -1.79661304e-01
-4.05412205e-02 -6.85664043e-02 -8.04563224e-01 -3.35180134e-01
-2.10678786e-01 -2.98504293e-01 7.03571022e-01 5.53986728e-01
1.77568126e+00 -1.92789480e-01 -2.50435889e-01 1.24315822e+00
1.26635361e+00 -5.83110340e-02 4.41959113e-01 -1.22253895e-01
8.00973892e-01 1.74286291e-01 1.13462880e-01 6.01098061e-01
3.09514225e-01 2.92745978e-01 8.08820546e-01 -4.67660844e-01
1.08402222e-02 -2.42935866e-01 -3.88029009e-01 1.07135832e+00
-2.95675278e-01 7.70549709e-03 -9.90094841e-01 5.56784987e-01
-1.50115049e+00 -2.27109164e-01 -3.14473897e-01 2.01230407e+00
4.55592394e-01 1.71195507e-01 1.63068309e-01 -4.22859826e-04
5.29967666e-01 2.88589746e-01 -8.47915232e-01 -8.45286965e-01
-1.20629989e-01 6.75481319e-01 2.51943201e-01 3.65861148e-01
-9.89601612e-01 9.91737723e-01 6.92001867e+00 5.86878896e-01
-1.10525441e+00 -2.08052233e-01 6.91112816e-01 2.38657203e-02
-4.52670872e-01 -2.69697726e-01 -8.69139135e-01 -3.10286045e-01
-1.89276822e-02 1.47193149e-01 1.60905093e-01 1.17016101e+00
-1.31623492e-01 2.31844589e-01 -1.17824376e+00 1.09342587e+00
-3.05158943e-01 -1.49886835e+00 5.17078280e-01 2.95462936e-01
8.05208504e-01 4.32550162e-01 -2.93376446e-02 -7.67930895e-02
1.38523370e-01 -1.20947444e+00 4.43237364e-01 -4.61459719e-02
1.25306594e+00 -9.02382851e-01 4.27722186e-01 4.36923742e-01
-1.71247697e+00 4.22677070e-01 -8.73126984e-01 6.32261112e-02
-2.49714568e-01 7.95855463e-01 -6.78271711e-01 3.18863928e-01
8.49908233e-01 6.82607114e-01 -2.01327547e-01 1.09199798e+00
3.22561324e-01 2.74554994e-02 -7.53064990e-01 -9.22253728e-02
4.69878405e-01 -1.79221630e-01 5.35008788e-01 1.10448992e+00
4.84158218e-01 6.29591942e-01 3.32497269e-01 1.09122097e+00
-3.01355213e-01 6.98394254e-02 -9.01054800e-01 -3.24414037e-02
6.06294811e-01 8.55177760e-01 -9.99769151e-01 -2.22021863e-01
-2.49200255e-01 7.86794543e-01 6.66629195e-01 1.57519549e-01
-2.69486368e-01 -6.12978816e-01 8.35682750e-01 4.96526390e-01
8.01792681e-01 -6.17727757e-01 -7.01355457e-01 -7.41763413e-01
1.16381072e-01 -3.57054234e-01 2.34363869e-01 -6.84244812e-01
-1.14437461e+00 8.88907313e-01 -2.68643409e-01 -1.07184470e+00
2.60780394e-01 -9.21089470e-01 -5.04887462e-01 9.19230640e-01
-1.52065110e+00 -1.06450295e+00 -4.78152841e-01 5.17731309e-01
2.92237431e-01 1.04720723e-02 1.05700386e+00 6.34490252e-02
-1.72387347e-01 3.68812084e-01 -2.38901347e-01 4.83437687e-01
-2.36200258e-01 -8.87303174e-01 1.32403946e+00 2.10772455e-02
-1.17400847e-01 3.84348840e-01 -3.84051949e-01 -8.57553363e-01
-1.58219838e+00 -1.30479133e+00 7.75138080e-01 -2.33085945e-01
1.10373102e-01 -4.25594181e-01 -8.79604757e-01 5.70163012e-01
-4.95649517e-01 7.41421163e-01 3.91201764e-01 -9.15161744e-02
-5.63385665e-01 1.28659502e-01 -1.17168784e+00 5.01328826e-01
1.62606430e+00 -6.14672780e-01 -2.39101380e-01 7.44534284e-02
7.07812846e-01 -1.03818989e+00 -1.09475386e+00 5.66546142e-01
6.95340097e-01 -9.37671781e-01 1.25065362e+00 -7.87478805e-01
2.88175255e-01 -2.37481147e-01 -1.03293948e-01 -1.01327884e+00
-3.65157455e-01 -3.39908749e-01 -2.58508086e-01 1.72477409e-01
1.47579461e-01 -4.13343608e-01 1.22186720e+00 2.27523401e-01
-3.29284906e-01 -1.57227492e+00 -1.18494964e+00 -4.18677330e-01
7.35136271e-02 -7.21492410e-01 1.20273709e+00 6.99417114e-01
-7.30858922e-01 -1.43413721e-02 5.58535457e-01 3.16999435e-01
7.07735598e-01 5.98109543e-01 7.70440638e-01 -1.76073933e+00
2.99570501e-01 -4.76916790e-01 -6.45146906e-01 -1.89584327e+00
-1.73828732e-02 -1.21759331e+00 -4.37348515e-01 -1.58427227e+00
-1.29607871e-01 -9.41986263e-01 3.19106519e-01 8.11977863e-01
3.73066962e-01 5.26157677e-01 -2.68172957e-02 1.63483620e-02
-1.83939770e-01 6.67741835e-01 1.99769783e+00 -1.74351826e-01
-5.01741290e-01 1.04159176e-01 -2.47844949e-01 1.02299941e+00
4.23553139e-01 -3.20702583e-01 -2.51070291e-01 -1.06678331e+00
3.03586990e-01 4.20729294e-02 2.92347193e-01 -5.06348908e-01
1.97607487e-01 1.88372239e-01 9.31561768e-01 -1.63103521e+00
5.04844844e-01 -6.69673264e-01 -1.80205584e-01 4.01714176e-01
1.71408564e-01 3.83166641e-01 5.96882701e-01 1.26695588e-01
6.82206377e-02 8.76095295e-02 7.19094276e-01 -5.05716681e-01
-8.85161832e-02 1.17149293e+00 -1.17873691e-01 1.07053362e-01
5.03068388e-01 -8.61452222e-01 1.06507726e-01 -2.88068019e-02
-8.35942566e-01 1.63619354e-01 4.70170975e-01 -4.54376172e-03
1.22728324e+00 -1.65603805e+00 -6.01698458e-01 7.91087031e-01
-2.17757940e-01 9.77124214e-01 2.76182760e-02 1.60260171e-01
-1.09794664e+00 4.80488360e-01 -3.38205278e-01 -1.07256663e+00
-1.03446412e+00 3.32626365e-02 6.46952093e-01 -1.35432377e-01
-9.06127453e-01 8.43446732e-01 3.05426776e-01 -9.01534736e-01
3.17393392e-01 -7.45259821e-01 1.11147821e-01 -2.59838998e-01
2.59706974e-01 4.41891462e-01 4.59814727e-01 -3.02607507e-01
-3.45926017e-01 1.16984260e+00 8.16563219e-02 4.73325402e-01
1.72176325e+00 6.17760122e-01 -6.20593190e-01 1.21638976e-01
1.30250716e+00 -1.62188187e-01 -1.09265018e+00 -1.22272424e-01
-2.33647570e-01 -5.36407948e-01 2.17302531e-01 -1.14919581e-01
-1.27281296e+00 1.01569664e+00 2.40950868e-01 2.36502334e-01
9.87216651e-01 8.83373395e-02 9.92520750e-01 5.53906620e-01
4.87034261e-01 -6.49561048e-01 -8.83325003e-03 1.27577591e+00
1.11090147e+00 -6.12432778e-01 2.11201936e-01 -1.08713782e+00
1.77801535e-01 1.53436875e+00 6.72291219e-01 -4.92150158e-01
1.21558630e+00 5.90886712e-01 -2.33613372e-01 -7.99793541e-01
-4.29544240e-01 -1.13278702e-01 4.23851430e-01 6.32153392e-01
3.29856694e-01 1.32770380e-02 3.01836342e-01 2.12450743e-01
-2.96195656e-01 -1.70325279e-01 -1.26502872e-03 9.39060867e-01
-4.09947574e-01 -8.30266654e-01 -1.83730364e-01 5.72827697e-01
4.80815843e-02 4.46893610e-02 -5.76431632e-01 7.27670789e-01
1.81102119e-02 -7.10791424e-02 7.22208917e-01 5.55565953e-03
7.30467916e-01 -1.63961649e-01 6.54310763e-01 -9.36527908e-01
-7.20890462e-01 2.32587885e-02 -1.93054542e-01 -7.35590875e-01
-1.43413037e-01 -2.40979031e-01 -1.44360065e+00 -5.11040270e-01
-1.70561478e-01 -2.76609570e-01 5.37341714e-01 4.31548446e-01
8.54913294e-01 1.25762448e-01 8.78811836e-01 -1.28398633e+00
-3.79379243e-01 -3.66439342e-01 -5.71468353e-01 -2.02389359e-01
3.23576927e-01 -5.01428425e-01 -4.15064432e-02 -5.70899367e-01] | [8.054408073425293, -3.6952645778656006] |
31a1fc97-2e55-4e7c-894c-bc5eaf06db35 | quantum-kernel-mixtures-for-probabilistic | 2305.18204 | null | https://arxiv.org/abs/2305.18204v1 | https://arxiv.org/pdf/2305.18204v1.pdf | Quantum Kernel Mixtures for Probabilistic Deep Learning | This paper presents a novel approach to probabilistic deep learning (PDL), quantum kernel mixtures, derived from the mathematical formalism of quantum density matrices, which provides a simpler yet effective mechanism for representing joint probability distributions of both continuous and discrete random variables. The framework allows for the construction of differentiable models for density estimation, inference, and sampling, enabling integration into end-to-end deep neural models. In doing so, we provide a versatile representation of marginal and joint probability distributions that allows us to develop a differentiable, compositional, and reversible inference procedure that covers a wide range of machine learning tasks, including density estimation, discriminative learning, and generative modeling. We illustrate the broad applicability of the framework with two examples: an image classification model, which can be naturally transformed into a conditional generative model thanks to the reversibility of our inference procedure; and a model for learning with label proportions, which is a weakly supervised classification task, demonstrating the framework's ability to deal with uncertainty in the training samples. | ['Joseph A. Gallego-Mejia', 'Raúl Ramos-Pollán', 'Fabio A. González'] | 2023-05-26 | null | null | null | null | ['probabilistic-deep-learning', 'density-estimation'] | ['computer-vision', 'methodology'] | [ 1.11188136e-01 2.03226238e-01 -1.15177736e-01 -3.09108824e-01
-8.73301864e-01 -5.04134357e-01 1.13176680e+00 -2.04960391e-01
-2.51657665e-01 9.38509285e-01 2.95204632e-02 -2.17106417e-01
-1.65130183e-01 -1.05251491e+00 -7.87374258e-01 -1.07484210e+00
1.10420741e-01 7.22149909e-01 -8.94377157e-02 4.79724765e-01
-8.59437585e-02 6.57951355e-01 -1.45605528e+00 -1.13111839e-01
6.21676505e-01 9.14338350e-01 -2.43806224e-02 5.14093161e-01
-1.20825805e-01 1.08581507e+00 -3.97439957e-01 -5.87007701e-01
-3.64076108e-01 -3.84588659e-01 -7.63248146e-01 -1.70498416e-02
2.13053629e-01 -4.36105132e-01 -4.92485881e-01 1.07298994e+00
8.23260695e-02 2.26055503e-01 1.61339939e+00 -1.18964028e+00
-8.80617738e-01 6.11468256e-01 -1.15181068e-02 -1.92322999e-01
-8.40619951e-02 -1.54842928e-01 1.02737665e+00 -5.72130442e-01
4.32850659e-01 1.36175966e+00 6.74811959e-01 4.84897166e-01
-1.76301956e+00 -5.33969700e-02 -4.53503788e-01 1.18012145e-01
-1.54402280e+00 -2.73136705e-01 4.89712328e-01 -5.94562769e-01
6.81434929e-01 -6.35150000e-02 4.94204015e-01 1.40432048e+00
2.19437033e-01 1.06233430e+00 1.13224852e+00 -5.88412166e-01
7.60366380e-01 4.30783838e-01 5.42667285e-02 7.91972876e-01
-9.09563527e-03 2.59190649e-01 -4.46902514e-01 -4.25695270e-01
9.66700971e-01 5.85763156e-02 -5.45211658e-02 -6.76406026e-01
-8.68139982e-01 1.03878641e+00 1.41092926e-01 6.34426326e-02
-2.55698591e-01 5.35700738e-01 1.09265633e-01 -2.00708136e-01
4.34470862e-01 -1.69713095e-01 -1.30158141e-01 -1.82237402e-01
-9.99479592e-01 3.47062409e-01 1.24649382e+00 8.67374063e-01
9.37540591e-01 5.48105389e-02 -2.88027585e-01 7.20483720e-01
6.94909513e-01 8.41534674e-01 8.66338238e-03 -1.22187412e+00
-3.70585799e-01 -1.50104716e-01 1.93088040e-01 -1.60732418e-01
3.89374495e-02 -3.13003242e-01 -7.89284647e-01 3.06117684e-01
4.78083193e-01 1.71303172e-02 -8.02143157e-01 2.10133004e+00
-4.17421455e-04 2.09939018e-01 1.90769747e-01 3.26194108e-01
3.39864045e-01 6.39874637e-01 2.41635591e-01 -1.16115421e-01
1.03613508e+00 -3.72034043e-01 -5.07025778e-01 2.44560033e-01
4.38923120e-01 -3.95258248e-01 8.92413616e-01 3.06834906e-01
-1.08603573e+00 -2.86035776e-01 -8.32427680e-01 -2.62509316e-01
-4.63396251e-01 8.49463195e-02 9.32762921e-01 7.96886146e-01
-1.24189830e+00 1.05458868e+00 -1.22038686e+00 -1.41693547e-01
6.90625608e-01 2.00533405e-01 -3.09268147e-01 -2.42832437e-01
-1.00753319e+00 1.01053834e+00 4.20566589e-01 -6.40672073e-02
-1.14520323e+00 -3.92908216e-01 -1.10150516e+00 3.45515132e-01
-3.00670475e-01 -9.56112921e-01 1.45647252e+00 -4.41375822e-01
-1.85178685e+00 7.48131692e-01 -2.05110848e-01 -5.88306785e-01
3.53457361e-01 3.63618955e-02 1.98773630e-02 6.75364211e-02
-1.11778170e-01 6.84531331e-01 9.51134324e-01 -7.23692477e-01
-1.83804289e-01 -2.82834172e-01 -1.30637035e-01 -8.30390155e-02
9.03647672e-03 -4.38328564e-01 1.51774622e-02 -1.74545065e-01
-2.04504371e-01 -6.84793711e-01 -1.97427347e-02 -4.96475659e-02
-3.46472561e-01 -3.61743003e-01 3.27135086e-01 -4.01814729e-01
6.82722986e-01 -2.32025886e+00 5.38748264e-01 1.12105124e-01
9.61965416e-03 -1.73280433e-01 1.32000558e-02 4.92253989e-01
1.19692050e-01 -4.62087290e-03 -4.80332822e-01 -8.15848410e-01
7.62763143e-01 5.32366276e-01 -5.93142927e-01 6.85863554e-01
4.97298360e-01 1.15859210e+00 -8.21802676e-01 -3.01016927e-01
3.57168347e-01 8.48733425e-01 -4.10582662e-01 1.06680907e-01
-4.50026155e-01 2.27762491e-01 -1.48342088e-01 3.61282885e-01
6.60798013e-01 -2.98569262e-01 1.82660803e-01 -8.92864391e-02
1.80603296e-01 4.86278892e-01 -1.08338869e+00 1.66713107e+00
-4.92605060e-01 5.56623638e-01 -1.36813149e-01 -9.03914034e-01
6.58222675e-01 2.62499630e-01 -2.81113666e-02 -1.39037549e-01
2.53974915e-01 1.45544857e-01 -4.06914055e-01 -5.72257340e-02
3.45773399e-01 -7.06421494e-01 -1.18367337e-01 7.17865467e-01
8.40139747e-01 -4.26772684e-01 2.22897306e-01 2.95808673e-01
7.77880430e-01 6.40379786e-01 2.07136452e-01 -3.83029282e-01
2.17017263e-01 -5.89061260e-01 -1.08352125e-01 9.89880800e-01
-5.65347001e-02 4.46658045e-01 7.29100645e-01 -1.07745707e-01
-1.04325831e+00 -2.00384378e+00 -6.66965961e-01 8.46032381e-01
-3.50658059e-01 -2.43469164e-01 -8.11248302e-01 -3.01152766e-01
1.43250776e-02 9.96007383e-01 -7.27966130e-01 -4.13167357e-01
1.68977216e-01 -1.08178365e+00 5.33655643e-01 5.24530649e-01
3.05358857e-01 -8.65449309e-01 2.82044727e-02 6.58363029e-02
1.11389272e-01 -9.72163975e-01 1.04728073e-01 4.83396530e-01
-8.97637725e-01 -7.61938632e-01 -6.37210906e-01 -4.46827233e-01
3.06405723e-01 -5.03971338e-01 1.10075998e+00 -7.62910366e-01
-2.90068895e-01 8.44876945e-01 2.29690969e-01 -1.88936800e-01
-9.43693519e-01 1.30097326e-02 8.47463384e-02 -1.19284750e-03
4.68483061e-01 -1.01467299e+00 -1.56286702e-01 -3.14817935e-01
-1.19655788e+00 -1.42989755e-01 5.27883291e-01 8.52505803e-01
5.41728139e-01 -1.52683377e-01 3.07048142e-01 -6.78498507e-01
4.70300436e-01 -5.58840871e-01 -6.90104663e-01 3.47023547e-01
-2.29823992e-01 6.77832484e-01 3.41725707e-01 -3.98699582e-01
-1.13109422e+00 -8.42881501e-02 -3.60809833e-01 -2.78573483e-01
-5.35007834e-01 4.38514352e-01 -1.67620316e-01 1.11955471e-01
6.46213591e-01 6.16672218e-01 3.04489642e-01 -6.04850948e-01
9.68801856e-01 5.99144757e-01 7.53657758e-01 -7.89196193e-01
4.95395273e-01 6.06360376e-01 3.65826219e-01 -7.39600837e-01
-9.54080820e-01 -1.00396136e-02 -7.83792019e-01 2.03364789e-01
9.44065869e-01 -1.03161156e+00 -7.40869462e-01 6.86159134e-01
-1.01986861e+00 -4.02475595e-01 -7.05044091e-01 6.68660879e-01
-1.01064873e+00 4.03615117e-01 -1.11694658e+00 -1.10495734e+00
2.87012872e-03 -1.06716263e+00 1.22316563e+00 2.21094728e-01
-1.09312106e-02 -1.64715421e+00 1.88100740e-01 -1.79931283e-01
3.59071434e-01 6.08218201e-02 1.28521788e+00 -5.10450423e-01
-7.94674516e-01 -2.29141578e-01 -1.28254905e-01 7.60175467e-01
-4.61449437e-02 1.96120948e-01 -1.30191052e+00 6.82215616e-02
7.20789880e-02 -5.51145554e-01 1.07030475e+00 3.23824316e-01
8.75217140e-01 6.29477426e-02 -1.81559518e-01 5.13507605e-01
1.28388035e+00 -4.45312738e-01 9.53173399e-01 -3.61749560e-01
5.97903848e-01 8.88516456e-02 -2.05132067e-01 4.09404069e-01
3.97977948e-01 3.79747093e-01 2.32929885e-01 3.68821114e-01
-4.73037921e-02 -4.15347666e-01 4.90330964e-01 6.32511318e-01
1.99463934e-01 5.07111661e-02 -4.52611059e-01 2.04167321e-01
-1.69262326e+00 -1.18096101e+00 7.75789246e-02 2.35133791e+00
1.02724004e+00 -7.13051856e-02 6.86460212e-02 -5.51257357e-02
5.63811421e-01 -4.02044803e-02 -4.15483922e-01 -2.16010556e-01
2.40560789e-02 7.94573367e-01 1.88461110e-01 5.46839058e-01
-1.19186091e+00 7.90097713e-01 7.82567501e+00 1.27006912e+00
-9.39216316e-01 1.44419864e-01 5.06399095e-01 2.82291532e-01
-5.84173977e-01 2.97489781e-02 -7.35203981e-01 4.62819219e-01
1.40421140e+00 9.87687185e-02 6.70706630e-01 8.72860849e-01
-3.52595687e-01 -2.19874203e-01 -1.44818926e+00 9.31312919e-01
-2.01465070e-01 -1.51859438e+00 1.88141823e-01 2.64274716e-01
4.36868966e-01 1.83200508e-01 2.42830500e-01 6.05192304e-01
7.22467661e-01 -1.08848536e+00 8.29287708e-01 8.35303783e-01
7.95323193e-01 -7.44605780e-01 4.35829997e-01 6.52785778e-01
-3.86039466e-01 2.40234911e-01 -4.82690632e-01 -8.61035660e-02
1.59670129e-01 7.75833488e-01 -7.03049302e-01 3.99008036e-01
1.32019117e-01 6.82567954e-01 -1.46623865e-01 7.40274787e-01
-4.83073056e-01 7.57604122e-01 -5.48578382e-01 -9.84188244e-02
-7.69767687e-02 -6.44235373e-01 2.80041099e-01 1.24506068e+00
2.52592683e-01 -5.73147297e-01 -2.31955528e-01 1.68998110e+00
-1.26905128e-01 -3.29388052e-01 -5.17822504e-01 -3.82479310e-01
3.07521373e-01 1.20623350e+00 -6.34660125e-01 -3.35113496e-01
-1.16468713e-01 9.70455348e-01 7.39360213e-01 5.80011308e-01
-1.05869246e+00 -9.97678638e-02 4.17937011e-01 -3.04275244e-01
5.56542933e-01 -5.26785135e-01 8.45409837e-03 -1.45611238e+00
-3.88497531e-01 -2.72278965e-01 -6.30463660e-02 -6.46221638e-01
-1.53089917e+00 1.81005865e-01 3.62695247e-01 -5.44002771e-01
-6.79574728e-01 -9.08762157e-01 -5.53109586e-01 1.21712101e+00
-1.26414537e+00 -1.12593234e+00 2.62786061e-01 4.41231996e-01
-3.65230262e-01 -1.86807029e-02 1.29044712e+00 9.20597166e-02
-4.07706678e-01 2.66222000e-01 6.65376306e-01 -3.02623846e-02
2.52547830e-01 -1.67452943e+00 2.98394918e-01 6.72605157e-01
4.51117247e-01 7.66570568e-01 5.52284598e-01 -7.09991753e-02
-1.32786930e+00 -7.27901280e-01 5.26461124e-01 -6.19439363e-01
8.52428257e-01 -6.69790506e-01 -7.73089170e-01 7.72702813e-01
-2.04942659e-01 -2.50391383e-02 1.02875209e+00 2.58126527e-01
-6.16845608e-01 1.70451850e-01 -1.27164721e+00 4.57672954e-01
5.96691549e-01 -1.15672469e+00 -5.00170231e-01 5.05880892e-01
3.92568499e-01 -1.65395439e-01 -9.34500754e-01 -1.90232752e-03
5.26021898e-01 -1.07384431e+00 8.17959905e-01 -4.34892356e-01
3.10105383e-01 -9.05213058e-02 -4.30792302e-01 -1.03512466e+00
-3.65688175e-01 -6.30285621e-01 -6.15270972e-01 1.19504750e+00
2.60123342e-01 -6.49403155e-01 4.42737877e-01 7.36784756e-01
9.07035992e-02 -6.13330662e-01 -1.14164233e+00 -4.96469080e-01
4.90380019e-01 -7.24009693e-01 3.67369801e-01 3.67888600e-01
-1.56028256e-01 3.08084071e-01 -2.03889087e-01 -1.04109105e-02
8.82513046e-01 -1.48122795e-02 4.17276233e-01 -1.16891408e+00
-8.16081703e-01 -5.46030998e-01 -6.07940733e-01 -1.26837921e+00
5.83703101e-01 -1.10280240e+00 2.73076072e-02 -1.32701278e+00
3.53162616e-01 -2.54969180e-01 -7.55081400e-02 8.41265693e-02
1.95619628e-01 3.21192890e-01 -1.43546715e-01 2.19311312e-01
-5.40738761e-01 8.77232075e-01 7.38504350e-01 1.15022466e-01
2.87848949e-01 1.77532405e-01 -5.14635503e-01 6.32814944e-01
3.82865131e-01 -3.02973866e-01 -3.69046569e-01 -2.76720058e-03
1.79271847e-01 -4.12109867e-02 7.79825687e-01 -9.12709951e-01
-1.53268114e-01 8.51222798e-02 5.11504233e-01 -1.68640673e-01
5.20205677e-01 -2.59483635e-01 1.32539809e-01 1.16623677e-01
-3.04653078e-01 -8.82121503e-01 6.67596981e-02 6.44401312e-01
-4.26122956e-02 -4.81413871e-01 9.95995104e-01 -5.00703305e-02
-4.34412122e-01 2.96669304e-01 -4.54784155e-01 -4.95084152e-02
6.37743771e-01 2.61653543e-01 -1.86953127e-01 -3.11645806e-01
-1.22010207e+00 -4.34720993e-01 4.78818595e-01 -1.97608531e-01
2.60341942e-01 -1.32337177e+00 -4.48517829e-01 1.87415883e-01
-2.30973572e-01 4.18189131e-02 3.00846010e-01 7.93208182e-01
-4.39283073e-01 2.00230867e-01 3.31837572e-02 -7.57483244e-01
-3.37035567e-01 4.22710419e-01 6.26821816e-01 -2.06859291e-01
-3.23026836e-01 7.74756610e-01 1.32578939e-01 -4.84657288e-01
1.02951758e-01 -5.04245639e-01 2.15572655e-01 -2.22602904e-01
6.60374820e-01 1.87580183e-01 -1.47128046e-01 -3.26767206e-01
-7.68471062e-02 4.89779236e-03 -5.51274698e-03 -4.60969239e-01
1.12013614e+00 -7.62835518e-02 -3.49716902e-01 9.84848022e-01
1.26978838e+00 -3.35241854e-01 -1.52256632e+00 -2.57775187e-01
-2.73405790e-01 1.28274977e-01 2.50931457e-02 -7.52823472e-01
-3.24791521e-01 1.29647553e+00 3.16727787e-01 3.34963679e-01
5.40415943e-01 3.31363261e-01 3.71280134e-01 4.51690823e-01
4.78614450e-01 -5.27053118e-01 -2.73546457e-01 6.26265466e-01
4.16873693e-01 -8.34587872e-01 -2.19220847e-01 -3.98921706e-02
-1.87111512e-01 1.28148258e+00 -2.38057062e-01 -2.23406658e-01
8.68419826e-01 3.74393702e-01 -4.62541670e-01 1.31599039e-01
-5.74548781e-01 -2.37202153e-01 3.13927054e-01 9.32361960e-01
3.90733182e-01 2.13242352e-01 3.28187644e-01 3.19465756e-01
-8.50391686e-02 1.57003641e-01 5.43795824e-01 6.77837431e-01
-3.60073656e-01 -1.29123271e+00 8.30173772e-03 3.16442877e-01
-3.43145400e-01 -2.61866808e-01 8.74200985e-02 6.35908723e-01
-8.34293738e-02 3.90500307e-01 1.82520479e-01 5.25510348e-02
-4.49151456e-01 6.72005713e-01 1.17412615e+00 -6.69554591e-01
4.74692315e-01 -3.27490233e-02 -2.27748677e-01 -1.16368920e-01
-4.47906494e-01 -6.94111645e-01 -1.09669280e+00 -3.74421030e-01
-2.08243623e-01 1.89327925e-01 8.37397218e-01 1.35763431e+00
1.54063180e-01 3.33619297e-01 2.23347440e-01 -1.06935441e+00
-1.21495855e+00 -1.13462591e+00 -1.29767108e+00 3.51336211e-01
2.61872530e-01 -8.23801577e-01 -5.07580519e-01 6.97560236e-02] | [6.977067470550537, 3.942051887512207] |
c87919bd-8faa-465d-b3ea-4c2292c91150 | clac-at-semeval-2016-task-11-exploring | 1709.02843 | null | http://arxiv.org/abs/1709.02843v1 | http://arxiv.org/pdf/1709.02843v1.pdf | CLaC at SemEval-2016 Task 11: Exploring linguistic and psycho-linguistic Features for Complex Word Identification | This paper describes the system deployed by the CLaC-EDLK team to the
"SemEval 2016, Complex Word Identification task". The goal of the task is to
identify if a given word in a given context is "simple" or "complex". Our
system relies on linguistic features and cognitive complexity. We used several
supervised models, however the Random Forest model outperformed the others.
Overall our best configuration achieved a G-score of 68.8% in the task, ranking
our system 21 out of 45. | ['Leila Kosseim', 'Elnaz Davoodi'] | 2017-09-08 | clac-at-semeval-2016-task-11-exploring-1 | https://aclanthology.org/S16-1151 | https://aclanthology.org/S16-1151.pdf | semeval-2016-6 | ['complex-word-identification'] | ['natural-language-processing'] | [-1.98493421e-01 -2.79839840e-02 1.66183144e-01 -2.39717185e-01
-5.98849416e-01 -8.27184141e-01 9.51280355e-01 4.87561584e-01
-1.04568803e+00 4.98246461e-01 3.72040153e-01 -5.34554839e-01
-1.47355452e-01 -3.71141165e-01 -6.82003275e-02 -1.74568385e-01
2.85468809e-02 8.20065022e-01 7.77027160e-02 -5.05107939e-01
6.94093883e-01 1.99347124e-01 -1.46449137e+00 4.38145339e-01
6.63815141e-01 9.06272769e-01 6.21349037e-01 8.52735102e-01
-2.13293239e-01 6.08952761e-01 -6.92595363e-01 -5.41636288e-01
-3.16498503e-02 2.54736722e-01 -1.36684382e+00 -5.50081611e-01
4.28163767e-01 4.84057277e-01 8.70747194e-02 9.11796868e-01
7.15960801e-01 1.26759470e-01 7.55481958e-01 -7.40985930e-01
-1.99326575e-01 1.10593975e+00 3.18905056e-01 6.05574846e-01
8.01464140e-01 2.04580147e-02 1.33481121e+00 -1.38890147e+00
5.90382814e-01 1.48912609e+00 5.88545740e-01 3.52819890e-01
-1.20056498e+00 -8.44491005e-01 1.34399176e-01 3.63577545e-01
-1.63188636e+00 -7.46875346e-01 1.97718740e-01 -7.17454314e-01
1.79740095e+00 1.59213006e-01 5.12225270e-01 1.11885417e+00
9.36886296e-02 5.42956471e-01 1.55535018e+00 -9.33466852e-01
1.04628772e-01 4.55663018e-02 8.86849523e-01 7.73929298e-01
2.88069904e-01 -2.56600250e-02 -8.32505286e-01 -9.45353657e-02
-3.15582842e-01 -7.33853757e-01 4.52734679e-02 6.13501430e-01
-1.17320585e+00 8.76693785e-01 -2.55141854e-02 7.40047753e-01
-2.34405696e-01 -2.32922971e-01 5.25943100e-01 2.85465121e-01
6.28922343e-01 1.13280916e+00 -7.33137488e-01 -4.79794800e-01
-8.07103813e-01 3.60735178e-01 1.03325808e+00 8.33685338e-01
4.36636806e-01 -2.34478742e-01 -5.03377736e-01 1.00683427e+00
2.68004864e-01 5.91742158e-01 7.37398207e-01 -6.33797348e-01
5.86235821e-01 8.02035034e-01 -1.07148774e-01 -6.70978129e-01
-1.11466765e+00 -5.23698866e-01 -3.24671447e-01 7.99232945e-02
4.02994603e-01 -1.68890357e-01 -1.01767004e+00 1.53148842e+00
-4.02957857e-01 -6.20944023e-01 -1.70111850e-01 3.70441288e-01
1.06961441e+00 4.91646588e-01 6.94312215e-01 -4.04730365e-02
1.65322661e+00 -7.87036777e-01 -8.64846468e-01 -6.84115529e-01
8.11874807e-01 -1.24278426e+00 1.31378174e+00 5.71948230e-01
-1.14580369e+00 -6.05691195e-01 -1.19998074e+00 -1.48503289e-01
-1.01611662e+00 3.77796501e-01 4.73940969e-01 9.36805665e-01
-1.08724201e+00 3.53313744e-01 -2.39361554e-01 -4.63955969e-01
-9.76232886e-02 1.96325243e-01 -3.92309934e-01 1.56403296e-02
-1.44459701e+00 1.57535088e+00 7.85962284e-01 -3.01190108e-01
-6.60793841e-01 -4.12380546e-01 -8.36939871e-01 -9.37573686e-02
3.13800275e-01 -5.03195405e-01 1.30350304e+00 -2.77992129e-01
-1.05225003e+00 1.36995149e+00 -3.75341207e-01 -3.61423016e-01
3.05780858e-01 -4.12004054e-01 -7.52558410e-01 -2.76067317e-01
3.49830925e-01 4.31729048e-01 4.08541113e-01 -5.95978618e-01
-8.18483591e-01 -3.97687584e-01 2.23357156e-01 2.48038426e-01
-1.54081523e-01 6.81109250e-01 -2.09091708e-01 -5.76219797e-01
-1.52954310e-01 -1.05906880e+00 2.97463662e-03 -1.05263257e+00
-5.10610282e-01 -9.02822196e-01 2.44243085e-01 -7.90487409e-01
1.90899432e+00 -1.72551179e+00 7.70264864e-02 2.01987401e-01
2.86546201e-01 4.47520614e-01 -6.51818737e-02 5.33503294e-01
-3.07911426e-01 6.15367115e-01 8.36355910e-02 -6.12717867e-01
1.64461762e-01 -6.95539042e-02 3.28987725e-02 7.08033219e-02
1.07911512e-01 7.16145754e-01 -8.60108256e-01 -5.33008695e-01
-1.88407358e-02 7.90278167e-02 -2.28288561e-01 1.34719014e-01
-4.94714975e-02 -9.42896828e-02 1.04472496e-01 5.03084064e-01
3.30775738e-01 2.27307081e-01 1.59964502e-01 -1.70464944e-02
-5.21064103e-01 9.78226304e-01 -1.15687943e+00 1.42311907e+00
-7.17700303e-01 7.14343905e-01 -2.45015547e-01 -5.11084557e-01
6.63536668e-01 2.36441359e-01 -3.63494307e-01 -8.15661371e-01
7.66152516e-02 4.02511239e-01 3.87318343e-01 -5.05145013e-01
5.90945303e-01 1.61539376e-01 -4.82215583e-01 6.80584833e-02
2.99916357e-01 -5.01424223e-02 8.52077186e-01 2.88089097e-01
1.25369179e+00 -1.45567104e-01 7.16869175e-01 -9.81790423e-01
8.38743567e-01 -1.66247830e-01 2.66721815e-01 8.48667324e-01
-2.47921646e-01 2.77934253e-01 4.02039498e-01 -5.11920094e-01
-6.84939086e-01 -9.26963449e-01 -4.75523137e-02 1.60110009e+00
-5.30670047e-01 -1.21774745e+00 -7.50673354e-01 -6.05257630e-01
-3.53576839e-01 1.10100210e+00 -5.44694006e-01 -9.32900980e-02
-5.72774649e-01 -3.50257456e-01 8.04172039e-01 2.29114980e-01
2.53274679e-01 -1.23989046e+00 -3.03032428e-01 2.44143203e-01
-5.39718390e-01 -1.29388046e+00 -5.77004850e-01 3.74828279e-01
1.06028775e-02 -8.95887434e-01 -1.48451760e-01 -9.72317576e-01
3.50718945e-02 -1.47436723e-01 1.72735894e+00 6.69286698e-02
-4.36482698e-01 5.36162108e-02 -6.06753349e-01 -7.35706627e-01
-3.37310493e-01 6.19740307e-01 3.22911203e-01 -5.15155733e-01
5.98402083e-01 6.65764958e-02 -4.50533211e-01 7.37700984e-02
-4.12098467e-01 -1.91583842e-01 4.90813345e-01 5.39291680e-01
1.09094866e-01 -1.32643878e-02 5.03393948e-01 -9.08305466e-01
1.00700426e+00 -2.54967213e-01 -2.78327703e-01 2.30464086e-01
-1.05559707e+00 5.68263941e-02 3.63942802e-01 -2.89216131e-01
-5.71717143e-01 2.64932603e-01 -5.65486670e-01 4.33543563e-01
-1.52814567e-01 6.62713826e-01 -2.99960613e-01 1.67550072e-01
8.61737430e-01 3.01604886e-02 -4.91190284e-01 -6.29025877e-01
9.07135159e-02 7.06985056e-01 3.09584081e-01 -6.67943954e-01
4.69518721e-01 -3.07303131e-01 -2.40935534e-01 -9.12200570e-01
-1.12115335e+00 -7.80906200e-01 -5.32810807e-01 -1.11603208e-01
9.95074570e-01 -9.50485170e-01 -8.09552908e-01 5.02873838e-01
-1.47310412e+00 -2.20222265e-01 2.02763155e-01 2.40670443e-01
-1.32382467e-01 2.50952899e-01 -3.23149294e-01 -9.40360367e-01
-6.97736800e-01 -1.08410478e+00 1.06568110e+00 -1.03618413e-01
-1.05734384e+00 -9.51085746e-01 2.45360490e-02 3.72592330e-01
5.76097429e-01 -3.20484936e-01 1.03455174e+00 -1.08361113e+00
4.41784620e-01 -2.73464501e-01 -1.39780557e-02 3.25792402e-01
-2.81329721e-01 -2.13128328e-01 -8.74593973e-01 -1.10696025e-01
-3.26398969e-01 -3.41729790e-01 7.82132447e-01 -2.46089194e-02
9.41167116e-01 -2.26399675e-01 -2.23565817e-01 -5.27365180e-03
1.07784891e+00 -5.88629954e-03 4.90513533e-01 4.75986421e-01
5.14255524e-01 7.14789033e-01 4.67857510e-01 2.76288778e-01
7.36388326e-01 8.22793841e-01 -8.88584554e-02 4.13404733e-01
-3.49257767e-01 -2.01379150e-01 6.19848073e-01 7.99579918e-01
-5.42374968e-04 -2.81949490e-01 -1.72743762e+00 4.89627689e-01
-1.38178813e+00 -7.75541544e-01 -4.76913154e-01 1.86877859e+00
8.51802051e-01 6.81437969e-01 1.86118364e-01 5.26377857e-01
6.67496562e-01 1.44027974e-02 2.73829013e-01 -7.75370240e-01
-3.60450357e-01 7.84843683e-01 3.52655619e-01 8.41958225e-01
-1.24445403e+00 1.48466396e+00 7.09048080e+00 1.07277143e+00
-7.14510262e-01 1.25250325e-01 3.39495867e-01 1.46079302e-01
7.77571723e-02 -7.74006546e-02 -1.38148832e+00 5.39782166e-01
1.39390957e+00 -4.33752269e-01 5.16656697e-01 5.80969512e-01
-3.59498523e-02 -3.95324528e-01 -1.06230438e+00 8.25654924e-01
2.46012539e-01 -9.38619852e-01 1.66519180e-01 -2.32054457e-01
2.57699728e-01 3.06143671e-01 -1.63836196e-01 9.42318738e-01
4.22796160e-01 -1.32779348e+00 9.10050452e-01 5.30564785e-01
8.26326311e-01 -7.31659174e-01 5.78698397e-01 6.82868719e-01
-1.11383402e+00 1.13860659e-01 7.86605943e-03 -4.52316135e-01
-7.51610696e-02 6.39287889e-01 -8.31904531e-01 1.27000570e-01
6.98319077e-01 1.25573352e-01 -1.23526001e+00 7.71955848e-01
-4.95729595e-01 8.19952071e-01 -1.85179099e-01 -4.76981193e-01
8.74007940e-02 3.11480016e-01 6.30458236e-01 2.04387999e+00
-1.39180943e-02 -3.44125293e-02 5.02543569e-01 3.12773138e-01
7.22697750e-02 5.42972088e-01 -5.56929708e-01 -9.88520235e-02
5.53732574e-01 1.41040468e+00 -5.70180655e-01 -5.17412007e-01
3.00831255e-02 6.60683453e-01 6.50623977e-01 -1.41811386e-01
-2.98423856e-01 -7.91704834e-01 3.19666296e-01 1.39471446e-03
-1.98778711e-04 -5.12000561e-01 -5.87542415e-01 -9.48388338e-01
2.35566422e-02 -8.49129677e-01 5.76854706e-01 -6.85819387e-01
-1.26854956e+00 9.84458208e-01 -9.84777808e-02 -5.61097324e-01
-3.23097080e-01 -1.11960924e+00 -4.05834079e-01 1.21061122e+00
-1.30518174e+00 -9.02698457e-01 -2.47772321e-01 5.10575056e-01
7.34181464e-01 -6.10507548e-01 1.43602085e+00 2.98205316e-01
-3.39996755e-01 7.43428290e-01 -5.08285284e-01 1.43624395e-01
8.10049415e-01 -1.62494409e+00 6.85377479e-01 7.38894582e-01
8.02454129e-02 8.16136360e-01 8.43195736e-01 -6.73109114e-01
-8.84817958e-01 -7.89749801e-01 2.04786158e+00 -7.29736090e-01
9.97664690e-01 -6.99739337e-01 -3.90157193e-01 2.60295719e-01
3.84262949e-01 -3.68355781e-01 7.51358151e-01 5.25025189e-01
-7.50726581e-01 3.23143780e-01 -9.72221136e-01 5.53976953e-01
1.19335246e+00 -9.33829904e-01 -1.00292516e+00 5.73148549e-01
7.74485588e-01 -1.39065817e-01 -8.03891182e-01 3.56649607e-01
5.31474054e-01 -5.50362468e-01 8.45696151e-01 -7.77997375e-01
4.42660898e-01 -1.68778747e-01 -3.51745754e-01 -1.72740233e+00
-8.21402371e-01 -4.78950024e-01 1.70689616e-02 1.20616484e+00
1.04876077e+00 -5.02161741e-01 1.97335780e-01 3.79092783e-01
-1.60637230e-01 -3.31784159e-01 -1.08759475e+00 -9.33755636e-01
3.02271038e-01 -8.77764404e-01 1.48579434e-01 8.39940846e-01
2.47914627e-01 1.22100019e+00 1.60711706e-01 -7.71384835e-02
1.01574324e-01 -5.27672052e-01 1.63103834e-01 -1.43752515e+00
-7.48699009e-02 -8.41879070e-01 -4.31827247e-01 -2.95280308e-01
6.11384869e-01 -1.32873023e+00 -7.44097903e-02 -1.42285216e+00
1.05896905e-01 -1.33838803e-01 -2.48110190e-01 5.29619277e-01
-4.48722661e-01 2.43147552e-01 2.65466601e-01 -1.15925685e-01
-6.97400868e-01 -1.13573290e-01 4.36998606e-01 -2.95935005e-01
2.03689754e-01 -4.33266535e-02 -8.19485188e-01 7.86898255e-01
1.08338356e+00 -2.69887745e-01 1.20953269e-01 -4.86892134e-01
6.92496181e-01 -6.47488058e-01 -1.75401658e-01 -1.07697916e+00
3.16713095e-01 1.97224483e-01 1.12024263e-01 -3.93949270e-01
2.05055356e-01 -5.07366657e-01 -2.65420318e-01 6.20800853e-01
-3.79410446e-01 7.32906580e-01 2.81791598e-01 -1.96303204e-01
2.37059891e-02 -5.41869879e-01 6.17359638e-01 -1.45044312e-01
-7.55207121e-01 -2.16266662e-01 -8.84696186e-01 4.37330395e-01
6.47503316e-01 4.44576383e-01 -4.08736765e-01 2.74596401e-02
-8.52309704e-01 8.11597407e-02 -2.42659405e-01 7.24325180e-01
3.16131085e-01 -1.10257185e+00 -1.04041862e+00 -5.80293797e-02
7.76051164e-01 -5.67284763e-01 -4.37811881e-01 5.21475554e-01
-5.39689422e-01 6.34870887e-01 1.33111328e-01 -1.57116741e-01
-1.42364109e+00 3.65634710e-01 1.93082541e-01 -5.72166741e-01
-1.43635616e-01 8.08464289e-01 -5.71721017e-01 -4.62282211e-01
3.43748331e-01 -6.87832013e-02 -7.46106386e-01 5.83470643e-01
8.88517976e-01 6.23250663e-01 6.13024294e-01 -8.95194292e-01
-5.77566803e-01 4.69884634e-01 -3.68540674e-01 -4.23662424e-01
1.00867260e+00 5.26755191e-02 -3.37715864e-01 7.11152077e-01
1.22508180e+00 1.06331840e-01 2.60565400e-01 -3.35015327e-01
8.23516190e-01 -2.38119457e-02 2.47739449e-01 -1.48844993e+00
-2.96705455e-01 6.42495275e-01 7.04566300e-01 2.66339481e-01
6.79814160e-01 3.65812145e-02 3.97182226e-01 9.74867940e-01
3.43416840e-01 -1.66054070e+00 -3.27135950e-01 1.55907214e+00
1.15687621e+00 -1.25428355e+00 -6.54430762e-02 -3.77011478e-01
-4.63807762e-01 1.03541589e+00 3.95730913e-01 -2.55584240e-01
7.36494005e-01 3.43200624e-01 -3.49532329e-02 -2.86754191e-01
-9.02903914e-01 -7.25885451e-01 6.42536938e-01 3.93273324e-01
9.11586463e-01 5.08453012e-01 -1.10176682e+00 9.04601991e-01
-9.18893039e-01 -4.70083714e-01 2.13133037e-01 6.02297485e-01
-5.07588029e-01 -9.71068740e-01 -1.17258266e-01 3.93696576e-01
-7.55004168e-01 -6.29189312e-01 -8.35445523e-01 6.55782402e-01
3.71774554e-01 1.47110081e+00 -1.76079586e-01 -5.51291764e-01
8.48342180e-01 6.07069910e-01 2.80040443e-01 -9.31513548e-01
-1.08865416e+00 -3.52943212e-01 8.93412709e-01 -4.85489607e-01
-1.41762778e-01 -9.34005439e-01 -1.11409628e+00 -2.70978153e-01
-5.29106818e-02 5.06425947e-02 1.00364697e+00 1.19140017e+00
8.85526277e-03 4.02637452e-01 4.27505702e-01 -6.12201750e-01
-2.90825874e-01 -1.53303337e+00 -1.69570595e-01 1.95155308e-01
2.73559000e-02 -6.75436437e-01 -4.37426835e-01 -5.56243695e-02] | [10.614543914794922, 10.468265533447266] |
ccb409b6-98cc-4824-9fa8-a1e92b2f4858 | data-dependent-regret-guarantees-against | 2303.06526 | null | https://arxiv.org/abs/2303.06526v1 | https://arxiv.org/pdf/2303.06526v1.pdf | Data Dependent Regret Guarantees Against General Comparators for Full or Bandit Feedback | We study the adversarial online learning problem and create a completely online algorithmic framework that has data dependent regret guarantees in both full expert feedback and bandit feedback settings. We study the expected performance of our algorithm against general comparators, which makes it applicable for a wide variety of problem scenarios. Our algorithm works from a universal prediction perspective and the performance measure used is the expected regret against arbitrary comparator sequences, which is the difference between our losses and a competing loss sequence. The competition class can be designed to include fixed arm selections, switching bandits, contextual bandits, periodic bandits or any other competition of interest. The sequences in the competition class are generally determined by the specific application at hand and should be designed accordingly. Our algorithm neither uses nor needs any preliminary information about the loss sequences and is completely online. Its performance bounds are data dependent, where any affine transform of the losses has no effect on the normalized regret. | ['Hakan Gokcesu', 'Kaan Gokcesu'] | 2023-03-12 | null | null | null | null | ['multi-armed-bandits'] | ['miscellaneous'] | [ 2.44527295e-01 1.63055539e-01 -6.73243523e-01 -3.90360028e-01
-9.97451544e-01 -1.01930106e+00 1.05040237e-01 3.30453932e-01
-6.54876828e-01 1.08397305e+00 -8.43604654e-02 -5.94965875e-01
-6.30113363e-01 -7.87737131e-01 -1.06454289e+00 -8.41177046e-01
-1.27742603e-01 6.12481117e-01 2.73542076e-01 -3.29395264e-01
1.18164763e-01 5.68583131e-01 -9.19903398e-01 1.84392199e-01
5.60224354e-01 1.54518199e+00 -2.19069794e-01 9.63683724e-01
3.72777618e-02 7.50190675e-01 -5.49761057e-01 -9.26069260e-01
1.12400448e+00 -5.46774924e-01 -6.87567651e-01 -5.89134358e-03
9.01834816e-02 -6.38894379e-01 -3.32601875e-01 1.04476559e+00
7.14244366e-01 8.53874907e-02 3.83151412e-01 -9.80003536e-01
-6.45694435e-02 9.41168845e-01 -3.52244109e-01 2.88833201e-01
7.09818378e-02 -7.59628648e-03 1.50386059e+00 9.41310450e-02
4.46126461e-01 9.93883967e-01 6.20742202e-01 5.20075202e-01
-1.41061151e+00 -5.84349334e-01 3.38538587e-01 1.35022059e-01
-6.60359561e-01 -2.85867423e-01 6.01649463e-01 -4.20306027e-01
3.04780751e-01 6.71780825e-01 6.46175265e-01 8.14889014e-01
-5.97232208e-02 8.96650910e-01 9.88608718e-01 -7.00812578e-01
7.34294653e-01 3.09536844e-01 -6.99262507e-03 2.89838314e-01
1.36097178e-01 6.66939378e-01 -3.26326579e-01 -4.74177092e-01
6.50888681e-01 6.29136860e-02 -3.86362672e-01 -8.63466024e-01
-5.72382033e-01 1.07300198e+00 3.60466659e-01 -2.06951573e-01
-3.40384364e-01 2.91033000e-01 5.57158947e-01 1.05177057e+00
4.75114882e-01 2.45330557e-01 -7.39448905e-01 -4.40429375e-02
-8.45156133e-01 3.11647624e-01 9.78363693e-01 8.78909528e-01
6.08929753e-01 -2.97648102e-01 -4.42849398e-01 8.62804830e-01
-1.94456562e-01 1.42414406e-01 1.59441814e-01 -9.82124388e-01
9.22261238e-01 3.59263308e-02 8.18296373e-01 -5.24157882e-01
-1.92120522e-01 -5.49210668e-01 -2.47857586e-01 3.50516915e-01
8.04198861e-01 -4.87180799e-01 -5.35616755e-01 1.89836884e+00
4.50297803e-01 -4.68066573e-01 -9.32066664e-02 9.56501424e-01
-7.84791335e-02 2.81941921e-01 -4.55566436e-01 -5.39344311e-01
5.22899985e-01 -8.17490220e-01 -3.78887892e-01 -2.11505458e-01
5.53231061e-01 -4.56620574e-01 6.40483856e-01 4.91390735e-01
-1.17998052e+00 9.04336870e-02 -1.09283578e+00 2.96071231e-01
-1.37357622e-01 -3.54483515e-01 5.50528884e-01 1.15482998e+00
-7.38607883e-01 8.31010818e-01 -7.95848608e-01 9.94418189e-02
3.87944818e-01 5.48031390e-01 4.40113991e-02 1.39278606e-01
-1.21637774e+00 7.09296644e-01 5.42051017e-01 -2.02408619e-02
-6.84116244e-01 -7.95502067e-01 -4.60913002e-01 9.34909731e-02
5.99692106e-01 -6.17734432e-01 1.57778513e+00 -1.82140732e+00
-1.59937394e+00 7.01797068e-01 2.91051239e-01 -1.18068540e+00
1.60136962e+00 -5.13758324e-02 1.34177461e-01 -2.04750717e-01
-3.39923382e-01 7.68710673e-02 7.03961015e-01 -7.23321617e-01
-8.29890966e-01 -2.55545259e-01 6.93409383e-01 3.71026278e-01
-1.68705836e-01 -1.16020679e-01 -8.61455202e-02 -6.43636286e-01
-4.82951134e-01 -9.95680511e-01 -4.35698628e-01 3.03329378e-01
-2.77596056e-01 2.60481805e-01 1.31695375e-01 -4.05819654e-01
1.04380274e+00 -2.14555597e+00 -2.54139245e-01 4.55479801e-01
-4.92196441e-01 -8.29692632e-02 7.14632273e-02 7.27569401e-01
-6.63330331e-02 4.33305651e-02 -8.53659958e-02 8.34305361e-02
2.82744616e-01 4.60671261e-02 -6.02010131e-01 7.07160234e-01
-3.87834936e-01 4.53155071e-01 -7.82399416e-01 -1.22413203e-01
3.66364866e-02 -3.41084331e-01 -8.72185946e-01 2.32699916e-01
-5.51926374e-01 1.02992773e-01 -6.44206226e-01 1.32339656e-01
5.02381325e-01 -4.60912511e-02 1.86832860e-01 2.83971518e-01
5.36574498e-02 1.20497696e-01 -1.29371917e+00 1.23330569e+00
-5.75149953e-01 4.51985508e-01 1.94745615e-01 -1.14911664e+00
6.72452629e-01 1.44721329e-01 4.07795787e-01 -3.38471711e-01
1.41918901e-02 2.54082143e-01 1.40349925e-01 -3.71170156e-02
1.31596029e-01 -5.68777382e-01 5.02851419e-03 4.44772512e-01
-1.89161271e-01 4.10854489e-01 5.04640564e-02 -8.11683685e-02
1.28888392e+00 -1.23564728e-01 4.41524267e-01 -1.73875198e-01
1.20328017e-01 3.54359206e-03 7.59175241e-01 1.34437311e+00
-2.30368227e-01 4.31494713e-01 8.07574868e-01 -4.60979640e-01
-1.08415508e+00 -9.29192781e-01 -9.37384441e-02 1.49619830e+00
2.07484141e-01 3.44832331e-01 -1.92787811e-01 -7.42159605e-01
6.03507400e-01 6.22156203e-01 -7.59458125e-01 -1.20005839e-01
-1.19595453e-01 -7.13945806e-01 5.13281375e-02 2.74702460e-01
2.99093783e-01 -7.35833406e-01 -5.67036331e-01 4.42131162e-01
1.53060541e-01 -6.14871979e-01 -8.28385234e-01 3.45513344e-01
-7.97537744e-01 -1.15366709e+00 -5.24961472e-01 -2.58104831e-01
4.57207769e-01 -1.83462098e-01 1.00010002e+00 -4.53060508e-01
-1.79200441e-01 5.87333798e-01 -3.00999075e-01 -7.44674087e-01
-3.70524645e-01 -2.05610409e-01 -2.30988175e-01 2.90653676e-01
-1.84352934e-01 -3.48379999e-01 -8.51710141e-01 2.10029438e-01
-7.39395201e-01 -2.64122128e-01 3.48469585e-01 1.11123252e+00
4.61631477e-01 -1.90736726e-02 5.52133918e-01 -1.42936468e+00
5.04603803e-01 -2.91741610e-01 -1.10036206e+00 4.86709058e-01
-3.95914733e-01 1.95739791e-01 8.59998107e-01 -4.40015852e-01
-8.05764139e-01 1.10434063e-01 1.97946459e-01 -2.56329030e-01
4.57156628e-01 2.63174593e-01 -3.02719355e-01 -2.14707449e-01
6.65842116e-01 6.71378672e-02 -1.30032584e-01 -4.14847761e-01
2.08927363e-01 5.04474938e-01 2.03908369e-01 -6.54356360e-01
4.65947181e-01 3.13938916e-01 2.27137897e-02 -2.15434179e-01
-1.10887539e+00 -3.70413780e-01 -1.34604886e-01 -1.15352258e-01
6.98038489e-02 -6.63518727e-01 -1.04478431e+00 4.53759357e-02
-7.89215744e-01 -5.96036494e-01 -6.07619345e-01 4.03116435e-01
-1.13090098e+00 4.14639600e-02 -5.21865308e-01 -1.40230680e+00
-3.32988322e-01 -7.52277195e-01 4.24036473e-01 -6.72703702e-03
1.52744085e-01 -1.14908707e+00 -7.06084073e-02 2.42573321e-01
2.42265686e-01 4.95382041e-01 7.47280657e-01 -9.80969250e-01
-6.31721973e-01 -3.65710825e-01 6.13438822e-02 6.07351005e-01
-1.28183363e-03 -2.61885226e-01 -6.01741314e-01 -6.41335309e-01
-4.39711064e-02 -5.65634370e-01 7.94274390e-01 4.75575268e-01
1.37567449e+00 -9.34942961e-01 -5.09936884e-02 5.41302383e-01
1.68397379e+00 4.17496622e-01 3.58248979e-01 7.77669549e-01
-2.31420286e-02 5.43027639e-01 7.47730315e-01 7.93771088e-01
-2.83729523e-01 5.93592525e-01 7.77018428e-01 1.77911982e-01
6.92591727e-01 -1.63110390e-01 2.92047292e-01 -1.57259315e-01
3.54268372e-01 -1.02488287e-01 -5.82821846e-01 6.26942217e-01
-2.07713532e+00 -1.15435910e+00 6.49822772e-01 2.97425318e+00
1.36119795e+00 4.41862375e-01 4.61392611e-01 1.08662643e-01
9.70921874e-01 4.38136421e-02 -1.20510149e+00 -8.77225637e-01
1.63605601e-01 1.90643266e-01 1.37499774e+00 5.75965583e-01
-1.12322903e+00 2.70363033e-01 6.34946012e+00 1.03636789e+00
-9.37556088e-01 4.53767143e-02 8.92590582e-01 -5.89560032e-01
-1.25849262e-01 -2.34414055e-03 -5.63441813e-01 6.11205041e-01
8.40185225e-01 -5.84325671e-01 7.41800845e-01 1.02200758e+00
1.27366662e-01 7.93922320e-02 -1.49407661e+00 7.10153461e-01
-7.57723451e-01 -1.30592895e+00 -2.79383510e-01 6.38965368e-02
7.43181705e-01 3.15141827e-02 2.80741025e-02 1.59077108e-01
9.58202124e-01 -6.54828072e-01 8.67046595e-01 4.59114522e-01
6.85113132e-01 -1.12481546e+00 7.01891422e-01 6.30699575e-01
-2.27098539e-01 -5.13800979e-01 -2.79150099e-01 -1.35361567e-01
-1.60493717e-01 5.59476972e-01 -7.29066670e-01 5.62156320e-01
4.23303097e-01 2.12967277e-01 2.14464441e-01 1.50690460e+00
2.89546728e-01 8.19057643e-01 -6.03626907e-01 -1.80484727e-01
2.51729041e-01 -2.45964646e-01 4.76616442e-01 1.02886879e+00
3.85350958e-02 -4.67444165e-03 2.77307332e-01 2.11234242e-01
-1.59237355e-01 3.28068793e-01 -2.41738707e-01 2.00536817e-01
5.02679527e-01 7.34941006e-01 -1.63454413e-01 7.63837025e-02
-3.30966353e-01 6.56484544e-01 1.82255924e-01 1.67804375e-01
-6.21841609e-01 -5.31785131e-01 8.11515033e-01 5.73190600e-02
8.13249886e-01 4.66660529e-01 -1.97012633e-01 -7.96029389e-01
2.22269639e-01 -6.59884632e-01 7.67605901e-01 -1.80231377e-01
-1.60792565e+00 -1.56784579e-01 -1.83664724e-01 -1.21310103e+00
-3.41527224e-01 -5.53396165e-01 -3.99093211e-01 7.27518141e-01
-1.33626580e+00 -4.51642722e-01 3.54983270e-01 5.69148600e-01
1.76177233e-01 -6.14943616e-02 5.67774534e-01 1.11214660e-01
-2.94282794e-01 1.07135785e+00 1.12499332e+00 1.33071244e-01
5.49788296e-01 -1.35755146e+00 -1.93925127e-01 5.38902104e-01
-2.07289249e-01 1.71498790e-01 1.07545066e+00 -1.81127638e-01
-1.12654376e+00 -1.18758440e+00 8.39849040e-02 -3.54871787e-02
1.03967810e+00 -4.29476678e-01 -1.67457759e-01 7.11450696e-01
-3.88002098e-01 3.02418381e-01 6.98810697e-01 1.79129645e-01
-3.53499711e-01 -8.38743985e-01 -1.52967024e+00 4.45332497e-01
9.81065929e-01 -2.97754288e-01 -1.73544437e-02 6.24427319e-01
6.65274858e-01 -7.42568135e-01 -7.14646697e-01 2.79506773e-01
8.24900866e-01 -1.03826761e+00 6.60568893e-01 -1.03220391e+00
3.25085253e-01 3.98319393e-01 -1.63659498e-01 -1.20131135e+00
-4.30307277e-02 -9.48767722e-01 -4.51829694e-02 9.92593229e-01
6.31863475e-01 -1.02081287e+00 1.01105428e+00 8.34613919e-01
3.50496024e-01 -8.34252298e-01 -1.18328857e+00 -1.01186967e+00
2.04671994e-01 -2.51009792e-01 4.66164827e-01 5.50805569e-01
-1.07274823e-01 -1.90930709e-01 -5.97485125e-01 -8.02575722e-02
6.81576729e-01 7.66809225e-01 7.29849637e-01 -9.33689535e-01
-1.08944261e+00 -4.37762648e-01 -3.09075087e-01 -1.19226086e+00
-2.03006029e-01 -6.43974900e-01 -1.74275130e-01 -8.79127622e-01
6.20516837e-02 -6.56805933e-01 -5.61338246e-01 4.35671061e-01
2.58942425e-01 -3.43572766e-01 2.37732917e-01 -5.21860681e-02
-5.51915824e-01 3.17291319e-01 1.06250811e+00 -2.05562413e-01
-2.59780556e-01 9.04115558e-01 -8.59545648e-01 3.06644410e-01
7.75181592e-01 -4.26254511e-01 -4.15036172e-01 -1.18579432e-01
3.28328550e-01 6.71164155e-01 1.16198830e-01 -4.02124465e-01
4.66853678e-02 -5.75938523e-01 1.87386632e-01 -7.03385770e-02
6.72652051e-02 -9.71359670e-01 1.88059628e-01 5.27841926e-01
-9.41466153e-01 -4.79009032e-01 -1.49308201e-02 1.13012278e+00
1.36859208e-01 -4.53906238e-01 9.03277278e-01 -1.19973443e-01
5.01218848e-02 4.37755078e-01 1.07660018e-01 2.20135435e-01
1.12370420e+00 -1.03936888e-01 -4.59753983e-02 -8.59852254e-01
-8.70494425e-01 4.77849603e-01 4.62638974e-01 -1.20777242e-01
-1.02429919e-01 -1.17901683e+00 -6.64098799e-01 -1.87361538e-01
-1.46260001e-02 -2.35414252e-01 -4.27118093e-02 3.98884863e-01
-4.04221326e-01 3.86161178e-01 6.05438016e-02 -3.22770290e-02
-1.01745057e+00 6.75845146e-01 6.41755521e-01 -5.44266760e-01
-3.20077896e-01 7.37519741e-01 1.03257068e-01 -1.31876513e-01
5.50129473e-01 9.44610387e-02 3.10083508e-01 -5.17707430e-02
3.82154793e-01 3.76305908e-01 2.57719517e-01 2.07846090e-01
-9.35063791e-03 -1.80603608e-01 -3.06615472e-01 -1.65578306e-01
1.43434107e+00 -1.39399290e-01 2.40223110e-01 5.86555302e-01
1.12231183e+00 -4.56273481e-02 -1.63590539e+00 -4.34677631e-01
7.95520097e-02 -6.92970395e-01 -1.06927007e-01 -1.04003251e+00
-1.13671029e+00 3.87218297e-01 6.39395416e-01 6.38158202e-01
1.36135864e+00 -3.30972463e-01 5.48600435e-01 4.08379972e-01
8.40615571e-01 -1.16706967e+00 -5.44400036e-01 2.81839341e-01
8.28860879e-01 -1.11314285e+00 -3.51500511e-02 6.95507182e-03
-3.95925134e-01 1.03208184e+00 8.73080716e-02 -4.32844490e-01
5.44240773e-01 4.69396189e-02 -2.30730191e-01 6.47716403e-01
-9.55026686e-01 -1.57182872e-01 -3.18827741e-02 2.62577087e-01
-3.15932035e-02 1.99863285e-01 -8.00559223e-01 6.50572181e-01
-4.90515947e-01 -9.67125073e-02 4.39749092e-01 7.83842981e-01
-4.24490452e-01 -1.34397244e+00 -4.37143207e-01 8.24011385e-01
-9.68497872e-01 1.45860687e-01 -3.50740016e-01 5.48470318e-01
-3.27572316e-01 7.41092086e-01 -7.53013268e-02 -5.91081660e-03
3.22210342e-01 -1.11147352e-01 5.78875601e-01 -3.18231165e-01
-7.10911095e-01 -1.70632854e-01 2.55382895e-01 -2.89343208e-01
-1.99152723e-01 -7.55371451e-01 -6.38013065e-01 -5.21259129e-01
-6.70382380e-01 2.99991488e-01 4.34093624e-01 8.04148316e-01
-3.78798991e-02 -6.63437545e-02 1.20435035e+00 -3.65330011e-01
-1.45067370e+00 -5.23715258e-01 -7.89950013e-01 2.16486499e-01
7.88427234e-01 -3.38912874e-01 -4.75320846e-01 -2.63938993e-01] | [4.576440811157227, 3.3414766788482666] |
94f78718-f120-448e-a192-3e0415216a0b | a-reliable-self-adaptive-face-identification | 2109.01212 | null | https://arxiv.org/abs/2109.01212v1 | https://arxiv.org/pdf/2109.01212v1.pdf | A Reliable, Self-Adaptive Face Identification Framework via Lyapunov Optimization | Realtime face identification (FID) from a video feed is highly computation-intensive, and may exhaust computation resources if performed on a device with a limited amount of resources (e.g., a mobile device). In general, FID performs better when images are sampled at a higher rate, minimizing false negatives. However, performing it at an overwhelmingly high rate exposes the system to the risk of a queue overflow that hampers the system's reliability. This paper proposes a novel, queue-aware FID framework that adapts the sampling rate to maximize the FID performance while avoiding a queue overflow by implementing the Lyapunov optimization. A preliminary evaluation via a trace-based simulation confirms the effectiveness of the framework. | ['Jae young Bang', 'Joongheon Kim', 'Dohyeon Kim'] | 2021-09-02 | null | null | null | null | ['face-identification'] | ['computer-vision'] | [-4.59753862e-03 -5.22857249e-01 -1.59053370e-01 -1.40628278e-01
-5.04755437e-01 -4.02840912e-01 -3.85393091e-02 -4.93506454e-02
-5.09398222e-01 4.92547840e-01 -6.55374050e-01 -5.48799872e-01
2.84949709e-02 -5.89576483e-01 -4.39594656e-01 -6.60886407e-01
-2.27481872e-01 1.38324365e-01 2.30067119e-01 2.48660088e-01
1.81044593e-01 6.19064629e-01 -1.84307480e+00 -2.28346691e-01
5.24418354e-01 1.20483267e+00 7.84344599e-02 6.74974024e-01
1.42931104e-01 5.18660843e-01 -9.28450048e-01 -3.82674068e-01
4.88417178e-01 -2.37552002e-01 -2.36488238e-01 1.74715459e-01
2.01981604e-01 -8.54430497e-01 -7.36165345e-02 1.14467692e+00
4.88105893e-01 -9.16340426e-02 1.26441583e-01 -1.49727833e+00
2.97439247e-01 1.04367934e-01 -5.99558055e-01 4.95081544e-01
2.82709301e-01 2.69923568e-01 4.70758557e-01 -8.41853857e-01
3.44598949e-01 1.04395771e+00 5.55130780e-01 2.45111048e-01
-1.33181524e+00 -8.40523481e-01 -1.23805292e-01 -5.95594421e-02
-1.90237153e+00 -7.72113085e-01 4.45968509e-01 -1.36626899e-01
6.65053427e-01 2.19370723e-01 4.56828445e-01 5.24483204e-01
4.56068069e-01 3.01319093e-01 7.89826035e-01 -2.68353254e-01
5.21932006e-01 -3.35276090e-02 9.08048153e-02 5.53708017e-01
7.75495410e-01 -3.68706733e-02 -6.48021221e-01 -4.69142467e-01
7.38384783e-01 -2.05891460e-01 -1.47335693e-01 -5.29961474e-03
-6.79026961e-01 5.85248649e-01 -2.97279716e-01 -2.28250343e-02
-6.55405879e-01 1.46205962e-01 5.57319641e-01 4.35681224e-01
2.48858333e-01 -2.83699706e-02 -1.02701269e-01 -4.55736816e-01
-1.31289434e+00 -4.88774776e-02 9.02154028e-01 8.83233249e-01
7.06298411e-01 1.69590443e-01 -6.78056479e-02 3.02256614e-01
3.54608774e-01 7.05841303e-01 3.77570391e-02 -9.99301314e-01
2.28225634e-01 2.20015571e-01 2.97602952e-01 -9.78970170e-01
-2.22992167e-01 -3.60714495e-01 -6.87166870e-01 4.54618961e-01
6.44667983e-01 -3.95348519e-01 -2.85812378e-01 1.67609751e+00
5.34758270e-01 3.97704661e-01 -2.55684108e-01 1.16868782e+00
-5.99525087e-02 6.04858637e-01 1.65316500e-02 -8.63900423e-01
1.37295282e+00 -2.85290986e-01 -8.31346095e-01 1.42745614e-01
2.31437206e-01 -9.44550812e-01 1.04133248e+00 5.01513720e-01
-1.02959323e+00 -3.20228696e-01 -1.16779006e+00 5.42992890e-01
3.80054414e-01 2.04329610e-01 4.18028355e-01 1.26379728e+00
-1.07676232e+00 4.80771840e-01 -9.65435863e-01 -1.57444969e-01
1.65071025e-01 7.26646662e-01 4.50067893e-02 2.06416994e-01
-7.89082408e-01 3.44507545e-01 -1.03529707e-01 -3.12292650e-02
-5.33938527e-01 -6.48081303e-01 -4.48239684e-01 1.07907645e-01
4.52460974e-01 -3.33414108e-01 1.24355781e+00 -9.57058370e-01
-1.70057094e+00 4.64809179e-01 -3.28291208e-01 -4.82191354e-01
7.25718796e-01 -1.07650898e-01 -2.10972920e-01 5.02929628e-01
-9.64988321e-02 1.89603031e-01 1.28740644e+00 -6.24547541e-01
-4.18546557e-01 -3.45740736e-01 1.33261561e-01 3.52423489e-02
-7.65525222e-01 2.25907892e-01 -5.59902251e-01 -3.13539386e-01
-2.90667683e-01 -1.11038041e+00 -2.18173638e-01 4.80219014e-02
-4.60290238e-02 1.30469978e-01 9.00944412e-01 -2.20055968e-01
1.45658612e+00 -2.29660702e+00 -6.32037401e-01 3.24631751e-01
-3.44820805e-02 5.23510993e-01 2.09427863e-01 2.81739771e-01
2.69635916e-01 3.93407717e-02 2.93672353e-01 -2.82845020e-01
-5.65608680e-01 1.42497020e-02 9.56730619e-02 7.41236091e-01
9.33226719e-02 6.54568151e-02 -7.84569025e-01 -5.84674001e-01
1.47472128e-01 3.14243436e-01 -7.89241791e-01 2.95527667e-01
1.40548676e-01 2.49887958e-01 -4.53786254e-01 7.24598169e-01
8.56883883e-01 -5.70062101e-01 5.39252996e-01 -2.41179138e-01
-2.84267128e-01 6.85689924e-03 -1.56928754e+00 8.45905125e-01
-5.02609730e-01 6.10966980e-01 3.26143444e-01 -4.86937940e-01
6.55594110e-01 2.86951602e-01 7.14022756e-01 -5.45806050e-01
3.83678555e-01 1.36456788e-01 -1.66982338e-01 -2.30027586e-01
6.71311378e-01 2.29681537e-01 1.77892283e-01 7.63907909e-01
-5.32586277e-01 5.91577411e-01 2.12816551e-01 2.28010099e-02
1.19176328e+00 -4.20529246e-01 2.75111318e-01 -3.67500037e-01
5.17450452e-01 -2.15603143e-01 6.24449849e-01 5.69306374e-01
-4.66875941e-01 7.93238450e-03 3.94683957e-01 -1.72979906e-01
-9.20010030e-01 -9.57713723e-01 -2.90879250e-01 6.67930543e-01
3.82646173e-01 -6.16726995e-01 -8.99353683e-01 -3.07360679e-01
6.20743595e-02 2.00949803e-01 -8.66998956e-02 -2.20935985e-01
-4.72690552e-01 -7.03002632e-01 6.18464112e-01 1.62072301e-01
4.85716969e-01 -3.66214633e-01 -1.41386700e+00 2.87732691e-01
-4.14369889e-02 -1.28369677e+00 -7.74496138e-01 -3.15966308e-01
-7.97028184e-01 -1.00088716e+00 -4.05915201e-01 -3.87263358e-01
9.21526492e-01 7.66801953e-01 8.34199369e-01 4.20660347e-01
-3.75122309e-01 4.65185940e-01 -3.03488746e-02 -1.53120190e-01
-2.39398852e-01 -1.36485696e-01 5.14900684e-01 2.78757334e-01
2.20993623e-01 -1.94673762e-01 -5.68639398e-01 3.78881812e-01
-7.24458694e-01 -2.61851907e-01 8.09323937e-02 9.11004782e-01
5.53304970e-01 3.73042941e-01 7.93574750e-01 -3.44231993e-01
7.67748833e-01 -3.89715195e-01 -1.26172042e+00 -7.50212520e-02
-8.60913992e-01 -1.02485053e-01 8.28359663e-01 -8.35499823e-01
-7.78283894e-01 1.48520201e-01 1.95193291e-01 -7.24198043e-01
4.58799362e-01 1.16359174e-01 -1.14784911e-01 -2.82508224e-01
2.33963594e-01 -6.22626357e-02 4.63924766e-01 -1.28605977e-01
-1.43075749e-01 8.47426832e-01 1.48933545e-01 -2.03473493e-01
4.81438309e-01 3.30758303e-01 2.01920167e-01 -1.17049932e+00
-1.26352668e-01 -2.86998510e-01 5.26782172e-03 -6.31744027e-01
1.35641932e-01 -9.22022223e-01 -1.53186798e+00 5.24029493e-01
-7.96879113e-01 -1.28525853e-01 1.37376755e-01 5.98641753e-01
-2.82296628e-01 3.41527313e-01 -5.22824109e-01 -1.32147348e+00
-5.78661859e-01 -1.22049129e+00 8.26041043e-01 6.13646865e-01
-1.20492280e-01 -3.86154920e-01 -3.87575716e-01 -2.69191321e-02
3.34405869e-01 -3.94700170e-02 2.42556676e-01 -1.98778063e-01
-8.99402857e-01 -2.73217142e-01 -2.63154119e-01 1.34567201e-01
5.06672412e-02 4.65744227e-01 -8.18697333e-01 -7.90889621e-01
9.14480686e-02 1.25257462e-01 4.77750152e-02 2.81185687e-01
9.98014212e-01 -6.09225571e-01 -2.37407997e-01 4.86955881e-01
1.63809049e+00 4.34448868e-01 2.55093634e-01 4.33383770e-02
2.35032290e-03 8.51165429e-02 9.96618748e-01 1.01013720e+00
2.06087887e-01 9.16494429e-01 4.53835934e-01 3.05460751e-01
2.71767288e-01 1.53196558e-01 6.28658295e-01 4.73379016e-01
1.73979238e-01 -2.78737247e-01 -7.53538013e-01 2.39357799e-01
-1.58862054e+00 -7.10839391e-01 1.32153258e-01 2.95629930e+00
7.57338524e-01 2.25837901e-01 4.93409485e-01 3.93908083e-01
9.29004252e-01 -2.74524063e-01 -5.53325474e-01 -3.85270327e-01
3.14847857e-01 7.31252655e-02 8.23998272e-01 4.01732266e-01
-6.48265660e-01 6.87308431e-01 6.68208551e+00 7.10394025e-01
-1.47709990e+00 -1.05566354e-02 5.61899424e-01 -4.02173460e-01
2.75945961e-01 -5.54126427e-02 -9.02579546e-01 9.69878316e-01
1.42699540e+00 -6.04056180e-01 5.55055380e-01 9.27102447e-01
5.44994533e-01 -5.49071312e-01 -9.17427957e-01 1.27434015e+00
-8.79571661e-02 -1.23405361e+00 -3.47456992e-01 2.89142311e-01
-7.47713866e-03 -2.77741164e-01 -5.45569584e-02 -1.43239602e-01
-2.50245422e-01 -4.88903612e-01 6.31689429e-01 2.94530541e-01
9.71857905e-01 -9.97611105e-01 6.15479290e-01 3.63509804e-01
-1.25214922e+00 3.04641128e-02 -3.06985080e-01 -1.41331732e-01
2.31103212e-01 8.06212723e-01 -1.09979093e+00 4.37202714e-02
5.56063294e-01 -1.53678715e-01 -4.35456693e-01 1.03404140e+00
3.54455829e-01 6.21469259e-01 -6.45646393e-01 -3.76741469e-01
-2.23348126e-01 -2.87174523e-01 7.31509507e-01 1.04983115e+00
3.65628183e-01 1.83805197e-01 3.05529207e-01 3.87848049e-01
3.70567776e-02 7.54395723e-02 -5.51378071e-01 1.28235454e-02
9.95453537e-01 1.29097974e+00 -8.20685387e-01 -3.72524202e-01
-2.95284927e-01 8.65209341e-01 -1.32052926e-02 8.21921453e-02
-9.85401511e-01 -3.17544848e-01 8.52229714e-01 1.87563330e-01
-2.91685294e-02 -5.99232495e-01 -1.32926226e-01 -8.07060659e-01
2.27619171e-01 -6.18727803e-01 2.73431152e-01 -3.04997265e-01
-7.52898395e-01 5.01739442e-01 -1.45742267e-01 -1.40231788e+00
-5.34268737e-01 -1.67980000e-01 -2.50160575e-01 6.80648446e-01
-1.14488709e+00 -1.00149721e-01 -2.59826273e-01 6.42044306e-01
3.37445617e-01 1.65952556e-02 5.57111084e-01 7.36119449e-01
-8.70194554e-01 1.21982217e+00 4.53387722e-02 -2.47201398e-01
4.73768979e-01 -5.92779279e-01 3.70138958e-02 1.00910401e+00
-2.58622110e-01 7.00995266e-01 8.84770393e-01 -5.38946927e-01
-1.99655771e+00 -8.79117012e-01 5.94169259e-01 2.66754478e-01
3.47872645e-01 -2.33222321e-01 -7.74437308e-01 6.83128741e-03
-2.84912676e-01 1.98818371e-01 6.37510121e-01 -3.64810109e-01
7.65975863e-02 -4.46677744e-01 -1.49999869e+00 7.88670480e-01
6.48057818e-01 -4.50066417e-01 2.89472938e-01 2.51171768e-01
2.78910935e-01 -3.44803184e-01 -8.78474116e-01 1.11946426e-01
7.47531950e-01 -9.25302684e-01 7.87541330e-01 -5.65035157e-02
-4.76974159e-01 -4.15484756e-01 2.14974303e-02 -7.82297254e-01
9.38714817e-02 -1.18483305e+00 -3.91292691e-01 1.27428925e+00
-1.51693225e-01 -6.76356733e-01 7.26382911e-01 7.98658371e-01
5.94728947e-01 -6.21605217e-01 -1.21778226e+00 -1.27164149e+00
-7.98394024e-01 -4.10457462e-01 6.29220843e-01 3.27112108e-01
-1.44555988e-02 2.78582990e-01 -3.29811692e-01 5.33701777e-01
9.47291553e-01 -8.54464546e-02 8.69500518e-01 -8.86743009e-01
-1.23847630e-02 -1.15110621e-01 -4.08522248e-01 -9.42701876e-01
-8.19017142e-02 -2.81953543e-01 2.58421991e-02 -5.18851638e-01
-2.65023708e-02 -6.33822680e-01 -1.49831325e-02 1.96750835e-01
-5.18782549e-02 3.28585953e-01 3.34187269e-01 4.11968738e-01
-5.95284641e-01 1.39295906e-01 6.85864747e-01 3.74092579e-01
-2.36518458e-01 2.38279164e-01 -1.71959713e-01 6.00163758e-01
8.28465164e-01 -5.75284183e-01 -3.67461532e-01 -2.37803534e-03
3.77107337e-02 5.12853146e-01 2.10920289e-01 -1.11368179e+00
4.54647958e-01 -1.63475841e-01 1.51816174e-01 -4.38525468e-01
4.37417507e-01 -1.02011323e+00 3.72065991e-01 8.28988135e-01
8.44715461e-02 3.53697777e-01 1.71013281e-01 4.79974240e-01
8.35014954e-02 -1.36850134e-01 1.07445538e+00 4.79494482e-01
-2.46375933e-01 3.34864020e-01 -8.01503301e-01 -2.18635485e-01
1.19804823e+00 -3.54003787e-01 -1.05493441e-01 -3.46928507e-01
-1.59976199e-01 9.12895575e-02 6.70655787e-01 2.85252780e-01
4.32583421e-01 -1.19125044e+00 -2.34160706e-01 5.30059576e-01
-2.52060473e-01 -6.58179879e-01 1.00081332e-01 9.25437331e-01
-4.61882055e-01 1.77763715e-01 -7.26624057e-02 -7.77676940e-01
-1.75070870e+00 4.94869202e-01 1.86667442e-01 -1.50380224e-01
-5.92149019e-01 3.39847803e-01 -3.60166520e-01 5.65125048e-01
2.22116157e-01 1.08409271e-01 1.97573736e-01 1.26119986e-01
8.95908594e-01 7.54755318e-01 1.70421883e-01 -6.12948954e-01
-6.07325971e-01 2.62992233e-01 1.18609399e-01 -2.38056019e-01
7.19635367e-01 -6.05106294e-01 6.33975565e-02 1.60795227e-01
9.47658956e-01 7.15116709e-02 -1.45412076e+00 -8.07687640e-02
-5.48978290e-03 -1.07389116e+00 2.20289186e-01 -3.21598239e-02
-1.25357401e+00 4.83791590e-01 8.33189726e-01 1.53228328e-01
1.37312436e+00 -7.15839565e-01 9.15205419e-01 2.37338111e-01
9.63481963e-01 -1.19616437e+00 2.18902215e-01 3.80218893e-01
2.13575482e-01 -9.58565474e-01 1.39404079e-02 -4.16413486e-01
-3.53620201e-01 1.19748795e+00 5.41343927e-01 1.05110994e-02
8.20331097e-01 8.17075729e-01 -1.28350973e-01 2.65273780e-01
-7.27713645e-01 8.65377709e-02 -2.95488060e-01 3.10111254e-01
1.23317987e-01 2.13127300e-01 -6.44168317e-01 5.15590727e-01
1.57361582e-01 3.34836304e-01 8.90557766e-01 1.04336452e+00
-4.13377494e-01 -9.39490259e-01 -6.60744488e-01 4.70705897e-01
-9.10912156e-01 1.17025614e-01 2.50106305e-01 4.21602219e-01
-1.77782252e-01 1.34048486e+00 4.16147709e-01 -2.27286920e-01
6.41640276e-02 -1.09677143e-01 2.18112096e-01 -3.10540646e-01
-5.06194711e-01 2.04922900e-01 1.21764503e-02 -8.46810102e-01
-1.91131681e-01 -7.73422956e-01 -1.38728881e+00 -8.40589285e-01
-4.35548484e-01 -6.35470599e-02 7.77984619e-01 6.36547625e-01
6.10610902e-01 2.22228438e-01 1.12249422e+00 -5.09966195e-01
-7.73827970e-01 -3.41784924e-01 -6.83803737e-01 -1.24955319e-01
2.76794136e-01 -6.63649559e-01 -2.69524902e-01 -8.37372616e-02] | [8.459314346313477, -0.3199966549873352] |
31418f06-5d8c-4931-964c-88cddc04aacb | real-a-representative-error-driven-approach | 2307.00968 | null | https://arxiv.org/abs/2307.00968v2 | https://arxiv.org/pdf/2307.00968v2.pdf | REAL: A Representative Error-Driven Approach for Active Learning | Given a limited labeling budget, active learning (AL) aims to sample the most informative instances from an unlabeled pool to acquire labels for subsequent model training. To achieve this, AL typically measures the informativeness of unlabeled instances based on uncertainty and diversity. However, it does not consider erroneous instances with their neighborhood error density, which have great potential to improve the model performance. To address this limitation, we propose $REAL$, a novel approach to select data instances with $\underline{R}$epresentative $\underline{E}$rrors for $\underline{A}$ctive $\underline{L}$earning. It identifies minority predictions as \emph{pseudo errors} within a cluster and allocates an adaptive sampling budget for the cluster based on estimated error density. Extensive experiments on five text classification datasets demonstrate that $REAL$ consistently outperforms all best-performing baselines regarding accuracy and F1-macro scores across a wide range of hyperparameter settings. Our analysis also shows that $REAL$ selects the most representative pseudo errors that match the distribution of ground-truth errors along the decision boundary. Our code is publicly available at https://github.com/withchencheng/ECML_PKDD_23_Real. | ['Xiaoyong Du', 'Yueguo Chen', 'Lizi Liao', 'Yong Wang', 'Cheng Chen'] | 2023-07-03 | null | null | null | null | ['active-learning', 'text-classification', 'active-learning'] | ['methodology', 'natural-language-processing', 'natural-language-processing'] | [-4.11618203e-02 3.91002238e-01 -7.36994982e-01 -7.00725436e-01
-1.42939985e+00 -4.43121135e-01 1.95182085e-01 4.43985820e-01
-3.81632537e-01 9.73095179e-01 -3.67046952e-01 -2.82220066e-01
-2.37714589e-01 -6.34404361e-01 -6.51722312e-01 -7.81309664e-01
2.49571249e-01 7.81452298e-01 -9.16175917e-03 3.94405603e-01
3.57324123e-01 8.94434080e-02 -1.35894501e+00 3.63352388e-01
1.33243394e+00 1.12393975e+00 2.01472305e-02 2.22442865e-01
-1.89783737e-01 8.46313596e-01 -7.57430673e-01 -5.93739986e-01
1.47577077e-01 -4.53508466e-01 -6.69734001e-01 4.96864915e-02
2.32212871e-01 -1.80901006e-01 9.52915177e-02 1.01762235e+00
4.48206872e-01 2.20987901e-01 9.19882238e-01 -1.21196258e+00
-4.42468971e-01 1.05407512e+00 -5.87842643e-01 3.72010142e-01
-1.46875286e-03 1.06963091e-01 1.26760173e+00 -1.42506135e+00
4.60907549e-01 8.30796361e-01 6.19147718e-01 5.21963537e-01
-1.31022716e+00 -1.03760743e+00 4.63483393e-01 1.89783156e-01
-1.76370728e+00 -4.49079931e-01 6.08218551e-01 -3.43582660e-01
6.29741251e-01 4.65721935e-01 2.04321742e-01 8.78531814e-01
-2.69309551e-01 1.31065178e+00 9.43263173e-01 -5.78148901e-01
5.77467978e-01 5.45763850e-01 7.13572860e-01 5.95097423e-01
2.59426624e-01 -4.90607545e-02 -5.60082853e-01 -4.12986189e-01
1.60689175e-01 -2.24746577e-02 -4.27333415e-01 2.54670456e-02
-6.99335754e-01 8.79349649e-01 3.92099082e-01 -4.00312543e-02
-2.38734186e-01 -2.10900471e-01 -8.70047212e-02 -1.05869628e-01
7.88825154e-01 6.36934280e-01 -6.49066091e-01 9.88253672e-03
-9.03577864e-01 4.67948895e-03 5.09592474e-01 1.01048303e+00
8.43314648e-01 -6.42001405e-02 -3.31223667e-01 1.34611976e+00
4.26544189e-01 3.46832752e-01 4.10221666e-01 -9.90323722e-01
6.85994685e-01 1.05252671e+00 2.00930908e-01 -4.14798498e-01
-2.44743168e-01 -5.50681889e-01 -4.96616602e-01 1.17110886e-01
5.26904225e-01 -3.68151993e-01 -9.44464982e-01 1.47851217e+00
2.56990075e-01 -1.61052719e-02 -3.02060962e-01 6.26721382e-01
7.70751357e-01 5.47362626e-01 3.63543987e-01 -3.78677458e-01
9.07697260e-01 -7.30088353e-01 -4.39357460e-01 -3.86604249e-01
7.62703419e-01 -5.60945451e-01 1.17464817e+00 4.48551506e-01
-9.76154804e-01 -2.44385540e-01 -1.00095177e+00 3.68593961e-01
-1.29036859e-01 4.38504636e-01 4.32664722e-01 7.56077051e-01
-6.22576892e-01 4.89831746e-01 -6.89264178e-01 2.13505343e-01
8.02231848e-01 3.09780478e-01 6.65308861e-03 -1.59054592e-01
-1.02112818e+00 4.92389441e-01 6.04586184e-01 3.95291038e-02
-7.11784124e-01 -7.98421264e-01 -6.04765594e-01 3.92650180e-02
5.64296007e-01 -8.29551592e-02 1.20583165e+00 -9.47125971e-01
-9.63091791e-01 7.90334761e-01 -2.11467102e-01 -4.02718633e-01
5.69595695e-01 -6.81491941e-02 -4.13683027e-01 -1.38818085e-01
1.92858800e-02 7.55541861e-01 4.65233624e-01 -1.39237785e+00
-7.57050991e-01 -4.44670290e-01 -2.92213798e-01 3.39028835e-01
-3.38503897e-01 -9.30606350e-02 -3.92343491e-01 -6.15068734e-01
4.75342423e-01 -8.88218284e-01 -9.33413208e-02 -2.90607244e-01
-7.50597060e-01 -5.30927777e-01 5.16376197e-01 -2.02711076e-01
1.65626919e+00 -1.99629462e+00 -4.03212458e-01 5.19479454e-01
3.08471948e-01 3.72883588e-01 3.43484074e-01 7.74089769e-02
-5.51449321e-02 4.47413743e-01 -3.18199217e-01 -3.55965525e-01
7.28158802e-02 -1.35688111e-01 -2.52751350e-01 2.48106807e-01
2.40960702e-01 4.70263630e-01 -6.58486545e-01 -5.22635043e-01
-2.68779285e-02 9.88223776e-02 -4.87015128e-01 2.23030746e-01
-4.78365451e-01 2.82248765e-01 -5.21886289e-01 7.84164786e-01
5.38801134e-01 -7.78756618e-01 1.52814746e-01 3.94426227e-01
2.41403580e-01 2.81140178e-01 -1.30676889e+00 8.97608340e-01
-3.65620740e-02 2.26331249e-01 -1.49654403e-01 -1.08101678e+00
1.22305524e+00 1.88347787e-01 3.97633642e-01 -3.83307606e-01
2.32618570e-01 4.40538228e-01 -1.19486086e-01 -1.16824426e-01
5.00193954e-01 2.31159136e-01 5.26887700e-02 6.21165514e-01
-1.67512193e-01 2.11014912e-01 2.46736243e-01 2.67239720e-01
9.83827412e-01 -7.83521831e-02 3.33217561e-01 -1.83073446e-01
2.49688059e-01 4.96988138e-03 9.37743187e-01 9.98897910e-01
-4.39439207e-01 7.21391618e-01 6.14113152e-01 -2.28222698e-01
-8.53300393e-01 -1.10591841e+00 -5.19687653e-01 1.16812694e+00
-6.88554198e-02 -2.85752207e-01 -6.26130283e-01 -1.09705818e+00
-1.10266954e-01 1.25668108e+00 -6.16498291e-01 -1.22741580e-01
-4.84114707e-01 -1.20237875e+00 3.52504253e-01 3.63286048e-01
4.17414188e-01 -1.04123974e+00 -1.36346027e-01 -6.90516382e-02
-2.59951293e-01 -6.64560139e-01 -3.30254287e-01 5.66521347e-01
-6.72203839e-01 -1.23733532e+00 -5.39673448e-01 -5.34056127e-01
1.04142368e+00 -2.20894992e-01 1.32160258e+00 1.58341989e-01
-1.27026156e-01 -7.76160508e-02 -7.09363043e-01 -7.28039384e-01
-4.05546993e-01 1.54220775e-01 -3.60746890e-01 -2.09735155e-01
9.05530512e-01 1.00537635e-01 -6.58997357e-01 5.56906939e-01
-6.17393136e-01 -2.35431641e-01 3.23838204e-01 1.00343430e+00
1.07702148e+00 1.94771290e-02 7.87268102e-01 -1.42741680e+00
3.91948968e-01 -7.88304687e-01 -5.08794904e-01 4.55493599e-01
-1.05556905e+00 -1.09487660e-01 6.42113924e-01 -3.23712140e-01
-9.21828508e-01 -3.84190455e-02 -1.25148386e-01 -3.40478808e-01
-1.82259217e-01 3.95955235e-01 -6.59347847e-02 7.89372444e-01
1.04745054e+00 3.40399519e-02 -3.05597961e-01 -2.74070889e-01
-1.22557513e-01 9.68316138e-01 -1.02710053e-01 -5.20789146e-01
1.74556136e-01 1.43553555e-01 -5.16455770e-01 -3.85776371e-01
-1.23993325e+00 -3.95015806e-01 -3.29317033e-01 -2.30731487e-01
2.12913379e-01 -9.01418984e-01 -5.25903225e-01 2.77453959e-01
-4.78089005e-01 -5.90715766e-01 -5.42802155e-01 6.29822195e-01
-3.22303683e-01 1.10641986e-01 -3.42863590e-01 -1.10291135e+00
-3.75252575e-01 -1.36593604e+00 8.12140465e-01 3.74848962e-01
-6.78985775e-01 -7.02456892e-01 -3.00232291e-01 6.44992352e-01
-7.11813718e-02 -1.24708667e-01 1.00337791e+00 -1.15242267e+00
-5.42260706e-01 -3.81620109e-01 -1.27086146e-02 4.15992230e-01
-1.84455663e-01 9.22426432e-02 -1.11313510e+00 -3.19353163e-01
-3.42007101e-01 -5.45402646e-01 9.67898190e-01 6.52919054e-01
1.62062633e+00 -2.69338012e-01 -5.01782954e-01 2.78395414e-01
1.11589098e+00 4.87007380e-01 5.69289565e-01 4.54406813e-02
2.11080477e-01 3.89067620e-01 9.14089024e-01 6.97102249e-01
1.91085607e-01 4.95152742e-01 3.51387620e-01 1.84501603e-01
8.93424749e-02 -2.27960423e-01 1.57652870e-01 3.03295821e-01
3.75428408e-01 -5.64921737e-01 -1.20167017e+00 3.86503905e-01
-1.60965133e+00 -8.68050337e-01 7.55301341e-02 2.46406007e+00
1.11690295e+00 4.63465422e-01 7.99773932e-02 3.11738729e-01
1.13681459e+00 -2.08272144e-01 -9.58326280e-01 -1.18140332e-01
-1.68222114e-01 1.55218124e-01 3.38285863e-01 5.42221665e-01
-1.03139544e+00 7.41176724e-01 5.57129717e+00 1.22865701e+00
-8.36955309e-01 1.12523595e-02 1.49016178e+00 -3.24030250e-01
-3.45182002e-01 -1.47602320e-01 -1.37052703e+00 8.05465698e-01
9.20307279e-01 4.95812520e-02 -4.47747111e-02 1.08077049e+00
1.23742476e-01 -3.38845968e-01 -9.81341302e-01 7.24962175e-01
-4.17049713e-02 -1.28908181e+00 -1.49356395e-01 -5.16132638e-02
8.95999372e-01 6.02449337e-03 2.79207706e-01 5.71617961e-01
7.34358788e-01 -1.02754080e+00 7.46080816e-01 3.07352543e-01
8.40116322e-01 -8.14833760e-01 6.87964559e-01 7.89897919e-01
-7.59438157e-01 -3.48801374e-01 -2.73363411e-01 2.61900842e-01
-2.26188704e-01 1.04696345e+00 -1.28530002e+00 7.14436322e-02
7.85939217e-01 3.83418858e-01 -6.29661322e-01 8.31874371e-01
-2.08556235e-01 1.05450273e+00 -2.78034419e-01 -2.90059030e-01
-1.64922222e-01 -9.94671583e-02 2.78545499e-01 1.08000207e+00
3.33982468e-01 2.80013829e-01 1.66875958e-01 9.30019736e-01
-3.58754605e-01 2.78535187e-01 -1.52598232e-01 5.89622259e-02
1.19489741e+00 8.60913157e-01 -9.15956080e-01 -4.14941967e-01
2.44256668e-03 3.83956641e-01 5.79344690e-01 3.05744171e-01
-8.82663071e-01 -3.76830876e-01 1.16839841e-01 9.04842839e-02
1.10839486e-01 3.84260774e-01 -6.73715770e-01 -1.02846014e+00
-1.24891112e-02 -7.47016609e-01 8.30547690e-01 -6.06440187e-01
-1.44171822e+00 4.85945046e-01 -6.67642280e-02 -1.37963986e+00
-1.80187434e-01 -5.20998061e-01 -2.34310463e-01 8.81480753e-01
-1.04936695e+00 -4.83507603e-01 -3.22402656e-01 2.68281579e-01
6.94554567e-01 -2.66751111e-01 7.84027398e-01 1.26884893e-01
-1.05153692e+00 1.10576296e+00 4.01900440e-01 2.94728816e-01
6.95172608e-01 -1.35863519e+00 -9.71101597e-03 6.81674063e-01
7.50202537e-02 5.71766257e-01 5.20058811e-01 -6.88186646e-01
-5.95282853e-01 -1.19498587e+00 8.29625666e-01 -4.86060739e-01
1.09363675e-01 -1.46504641e-01 -8.92508149e-01 9.22653556e-01
-2.75969118e-01 1.42482966e-01 9.92929041e-01 2.52873331e-01
-3.75857353e-01 -6.99231029e-02 -1.40046072e+00 5.00698268e-01
7.17511475e-01 -8.31373259e-02 -1.52868092e-01 4.97476757e-01
4.03631717e-01 -3.73945206e-01 -9.29430187e-01 6.14029169e-01
2.34043464e-01 -1.14861798e+00 7.12502718e-01 -3.05015802e-01
4.04655039e-01 4.73518148e-02 -1.42755508e-01 -1.19565630e+00
-1.19427361e-01 -2.19435394e-01 -1.93010181e-01 1.35501790e+00
1.27141476e+00 -6.80799544e-01 1.18518782e+00 9.55583036e-01
-1.97471678e-01 -1.09001851e+00 -7.50820637e-01 -4.36869651e-01
2.77337879e-01 -5.74579418e-01 5.79782844e-01 1.16544545e+00
6.38575554e-02 1.17622413e-01 1.28092980e-02 -9.86212939e-02
6.42950594e-01 5.22749461e-02 3.89462024e-01 -1.22985792e+00
-3.09181839e-01 -4.33727503e-01 1.26121730e-01 -1.08005667e+00
1.70749485e-01 -9.36216712e-01 1.82573438e-01 -1.16732907e+00
3.59104395e-01 -1.21534741e+00 -4.52975601e-01 6.12093031e-01
-6.36832952e-01 2.62431055e-01 -1.65732190e-01 4.59194392e-01
-8.35871518e-01 3.20027560e-01 8.21835160e-01 -3.09936348e-02
-2.30154574e-01 3.68595779e-01 -8.13045323e-01 7.58713841e-01
9.75316942e-01 -5.01866043e-01 -4.89612460e-01 -1.37456298e-01
3.04270536e-01 1.90824755e-02 -1.51479140e-01 -6.98976576e-01
2.46641710e-02 -2.77631849e-01 6.17642164e-01 -7.15714455e-01
2.60729194e-01 -5.37274361e-01 4.33817916e-02 2.51910359e-01
-9.35360909e-01 -1.44051880e-01 -1.22516990e-01 5.09085298e-01
-2.42066775e-02 -5.65213084e-01 1.04832482e+00 -8.99589136e-02
-4.93523687e-01 2.85890162e-01 -6.60066009e-02 4.50174332e-01
1.10494220e+00 -2.49089271e-01 -5.96267164e-01 -5.83803356e-02
-8.80854070e-01 2.50927448e-01 3.14405888e-01 1.09082505e-01
4.17672634e-01 -9.37742651e-01 -7.47764945e-01 1.24684706e-01
4.70947355e-01 2.51903236e-01 2.70281643e-01 6.90351069e-01
-3.78962249e-01 3.43182087e-01 4.83378768e-01 -7.16534853e-01
-1.09230769e+00 1.01157114e-01 4.67219710e-01 -3.53156626e-01
-1.86760068e-01 1.29346275e+00 -1.64070368e-01 -6.43714964e-01
4.34943467e-01 1.15306661e-01 -1.44778922e-01 1.48271397e-02
4.06479150e-01 7.06792533e-01 3.22111249e-01 -3.69431168e-01
-1.51040256e-01 -1.54891208e-01 -3.57599646e-01 1.85949758e-01
9.84813929e-01 -1.51958376e-01 2.43582636e-01 5.93900740e-01
9.40689862e-01 -1.49633139e-01 -1.40037930e+00 -4.15759981e-01
1.78302020e-01 -5.54066122e-01 -1.32044647e-02 -1.23445296e+00
-9.34205949e-01 5.28548062e-01 5.16173482e-01 1.85825691e-01
8.67160439e-01 6.10014088e-02 1.97117925e-01 2.29237437e-01
2.88360417e-01 -1.44735718e+00 1.54334381e-01 3.05660754e-01
4.99417096e-01 -1.56875420e+00 2.04788253e-01 -5.20881772e-01
-8.89406025e-01 6.06112003e-01 9.02317047e-01 2.07332805e-01
7.91528106e-01 1.63881674e-01 2.62222975e-01 -1.08449906e-01
-9.05646384e-01 1.79816470e-01 1.19731180e-01 1.58276841e-01
6.57627046e-01 3.61868531e-01 -4.38922256e-01 6.27882481e-01
-1.29591852e-01 -1.86636269e-01 3.32812071e-01 8.18974197e-01
-6.20987475e-01 -9.77273405e-01 -3.61578912e-01 1.04217744e+00
-6.02375746e-01 -6.26480803e-02 -3.44456106e-01 6.62890673e-01
3.06687087e-01 1.16865110e+00 1.20301522e-01 -1.85680017e-01
1.05067037e-01 4.43285108e-01 9.93445292e-02 -8.13329816e-01
-5.59226811e-01 1.67772472e-02 1.24678493e-01 -1.19802989e-01
-2.61844788e-02 -7.98759639e-01 -1.54133356e+00 -2.22990930e-01
-6.96443498e-01 4.09489810e-01 3.68543327e-01 7.65316010e-01
3.71730983e-01 1.50825828e-01 7.56211400e-01 -1.03187405e-01
-8.26735020e-01 -1.08478367e+00 -8.71849418e-01 4.65029091e-01
-1.62066922e-01 -6.20906174e-01 -6.56335890e-01 -2.35941276e-01] | [9.465177536010742, 3.817034959793091] |
17bd20b4-3f9d-49ee-8b67-b62c5c33c8fd | structure-based-approach-can-identify-driver | 2303.04888 | null | https://arxiv.org/abs/2303.04888v1 | https://arxiv.org/pdf/2303.04888v1.pdf | Structure-based approach can identify driver nodes in ensembles of biologically-inspired Boolean networks | Because the attractors of biological networks reflect stable behaviors (e.g., cell phenotypes), identifying control interventions that can drive a system towards its attractors (attractor control) is of particular relevance when controlling biological systems. Driving a network's feedback vertex set (FVS) by node-state override into a state consistent with a target attractor is proven to force every system trajectory to the target attractor, but in biological networks, the FVS is typically larger than can be realistically manipulated. External control of a subset of a biological network's FVS was proposed as a strategy to drive the network to its attractors utilizing fewer interventions; however, the effectiveness of this strategy was only demonstrated on a small set of Boolean models of biological networks. Here, we extend this analysis to ensembles of biologically-inspired Boolean networks. On these models, we use three structural metrics -- PRINCE propagation, modified PRINCE propagation, and CheiRank -- to rank FVS subsets by their predicted attractor control strength. We validate the accuracy of these rankings using three dynamical measures: To Control, Away Control, and logical domain of influence. We also calculate the propagation metrics on effective graphs, which incorporate each Boolean model's functional information into edge weights. While this additional information increases the predicting power of structural metrics, we find that the increase with respect to the unweighted network is limited. The propagation metrics in conjunction with the FVS can be used to identify realizable driver node sets by emulating the dynamics that are prevalent in biological networks. This approach only uses the network's structure, and the driver sets are shown to be robust to the specific dynamical model. | ['Réka Albert', 'Jorge Gómez Tejeda Zañudo', 'Eli Newby'] | 2023-03-08 | null | null | null | null | ['feedback-vertex-set-fvs'] | ['graphs'] | [ 4.26112890e-01 3.55387956e-01 -1.64244071e-01 1.18812717e-01
4.07188237e-01 -8.66192877e-01 9.43848908e-01 3.53449434e-01
5.13731502e-02 9.37602282e-01 2.99510080e-02 -3.70874405e-01
-6.54138505e-01 -1.11511981e+00 -5.26205897e-01 -9.02463019e-01
-5.09750545e-01 2.30797514e-01 6.04886651e-01 -7.09884107e-01
3.55286062e-01 1.04086137e+00 -1.28147721e+00 2.14671716e-02
5.63868463e-01 4.51014042e-01 -1.97549000e-01 7.91706264e-01
1.24663807e-01 5.28553545e-01 -5.66246688e-01 2.62534738e-01
1.15603525e-02 -6.45258009e-01 -6.30999565e-01 -4.68689501e-01
-2.03350589e-01 5.10201991e-01 -4.30383623e-01 7.07717359e-01
1.74107730e-01 -1.98564932e-01 9.27180886e-01 -1.38272333e+00
-2.29143396e-01 4.90165025e-01 3.48977605e-03 2.46212229e-01
3.11028659e-01 4.89045143e-01 9.56819952e-01 -1.59709111e-01
1.12798274e+00 1.23419976e+00 6.10614061e-01 4.79903013e-01
-1.91150463e+00 -4.12029743e-01 -6.54096752e-02 -4.90962684e-01
-1.24889565e+00 -5.10037124e-01 7.84753084e-01 -5.24092913e-01
8.42630625e-01 5.45027852e-01 1.21966016e+00 7.17230558e-01
8.63339186e-01 -8.19536299e-02 1.24592113e+00 -1.84476629e-01
4.46972489e-01 -9.69989598e-02 1.56347245e-01 7.73431897e-01
5.22336066e-01 2.60018945e-01 -3.13320041e-01 -3.60848188e-01
7.09807515e-01 -3.92796397e-01 -5.89942873e-01 -2.23859072e-01
-1.24815154e+00 5.59173405e-01 6.06678128e-01 6.55953109e-01
-1.18318886e-01 6.05510235e-01 1.22289330e-01 6.89487100e-01
3.34263086e-01 1.18223536e+00 -4.06904638e-01 1.40431106e-01
-6.77110136e-01 1.79889649e-01 1.21850502e+00 4.87775177e-01
7.62084246e-01 -1.07460462e-01 -1.81610465e-01 2.63639152e-01
2.14958966e-01 4.88843530e-01 -1.72943145e-01 -8.15762520e-01
-4.04388160e-01 1.22106397e+00 -1.15100592e-01 -1.51817048e+00
-8.39423537e-01 -4.40805435e-01 -1.07185638e+00 2.04279155e-01
7.03781247e-01 8.63239449e-03 -7.55854487e-01 1.87789047e+00
3.26803029e-01 -7.84494877e-02 -1.12278603e-01 6.42518461e-01
3.76337528e-01 7.35771954e-01 -2.37477973e-01 -4.63180363e-01
8.42602432e-01 -1.80162653e-01 -4.79286134e-01 9.53098983e-02
6.72612906e-01 -2.41598770e-01 6.13405764e-01 2.19382904e-02
-8.43600571e-01 4.04036902e-02 -1.17604232e+00 6.36800647e-01
-7.07399249e-01 -5.20796895e-01 5.72296143e-01 4.88719612e-01
-1.46174681e+00 9.09303486e-01 -8.24253261e-01 -6.34092271e-01
1.47884682e-01 6.73021853e-01 -3.63233089e-01 2.87884355e-01
-1.47119379e+00 1.05052578e+00 8.72463360e-02 -5.69133423e-02
-9.09861207e-01 -8.06493819e-01 -4.29707050e-01 7.83579871e-02
1.12431325e-01 -8.04791510e-01 1.82740405e-01 -7.68113017e-01
-1.38712084e+00 4.75691110e-01 -2.52675172e-02 -1.90634146e-01
2.82709807e-01 1.07146943e+00 -3.00125271e-01 1.06833868e-01
-2.81744927e-01 7.73499429e-01 5.44476569e-01 -1.22142279e+00
2.27984712e-01 -1.26099095e-01 2.75714725e-01 -3.09785962e-01
-5.61310112e-01 -2.89841145e-01 1.63609479e-02 -1.68272212e-01
7.94924721e-02 -1.25338101e+00 -4.05015707e-01 1.14325486e-01
-7.38640666e-01 1.03772813e-02 6.29820347e-01 1.54104024e-01
1.46771872e+00 -1.71813715e+00 6.54042184e-01 9.68166709e-01
5.05593359e-01 -6.56864047e-03 -2.84031123e-01 9.48303699e-01
-3.65764042e-03 6.78014517e-01 -2.89284199e-01 6.52567506e-01
-4.29687083e-01 9.80706885e-02 5.66523522e-02 7.03993201e-01
5.44652164e-01 8.34181726e-01 -1.03463984e+00 -4.02720422e-01
-1.13499209e-01 6.08032525e-01 -5.80774426e-01 -2.22949326e-01
-1.51550680e-01 4.38773155e-01 -4.96062458e-01 3.57477218e-01
1.33983135e-01 -3.78413647e-01 5.03286779e-01 -9.51461196e-02
-1.74754158e-01 4.76032980e-02 -8.59623313e-01 7.15259135e-01
-1.25308663e-01 6.77432060e-01 2.77437195e-02 -6.50535941e-01
1.08578014e+00 1.30025148e-01 5.56777835e-01 -9.08266455e-02
2.97114290e-02 2.23818719e-01 7.90825129e-01 -9.93024558e-03
-2.69101169e-02 -4.01474386e-02 -1.75330900e-02 4.37996089e-01
1.98877207e-03 -3.54652077e-01 5.99637568e-01 4.10195261e-01
1.88207877e+00 -3.93081903e-01 2.39607722e-01 -1.00685954e+00
6.36235893e-01 2.48499945e-01 3.37362856e-01 7.14410365e-01
-1.94311306e-01 1.94327697e-01 1.28865445e+00 -2.21299306e-01
-9.73005295e-01 -9.61653829e-01 -3.49656910e-01 4.85656291e-01
2.66450137e-01 -5.33095777e-01 -6.46014035e-01 -2.67726302e-01
4.09169436e-01 2.66797096e-01 -9.79882300e-01 -7.25768983e-01
-3.79637897e-01 -8.16457629e-01 8.62565458e-01 -1.35856181e-01
-6.70840442e-02 -8.38501990e-01 -5.34083307e-01 1.57340646e-01
3.19779038e-01 -3.13257158e-01 -2.68979013e-01 3.29274297e-01
-8.54654014e-01 -1.38758373e+00 -2.58261830e-01 -2.72370368e-01
1.09520233e+00 -1.46248162e-01 1.05420125e+00 5.21153450e-01
-3.30580592e-01 4.20164943e-01 -6.71714172e-02 -1.05331577e-01
-9.76284385e-01 1.99149773e-01 3.09849620e-01 -2.97720820e-01
-3.13484639e-01 -7.81646192e-01 -6.37067080e-01 6.45132124e-01
-1.04258406e+00 -1.25251591e-01 2.85014629e-01 8.65358412e-01
3.62147033e-01 1.23322636e-01 7.08438039e-01 -7.61230826e-01
8.88719916e-01 -5.35233974e-01 -4.98966455e-01 2.29225829e-01
-8.15717518e-01 3.62540632e-01 6.96432531e-01 -5.47628939e-01
-4.13201511e-01 1.79921016e-02 5.07238150e-01 -1.15280464e-01
2.93350846e-01 6.08858883e-01 6.81790784e-02 -4.35088068e-01
6.75372481e-01 1.11014485e-01 4.89681482e-01 2.29069412e-01
9.82562825e-02 1.36415780e-01 -1.75183594e-01 -3.76501590e-01
7.17313886e-01 4.83168781e-01 8.29190969e-01 -8.51212561e-01
5.16855828e-02 1.02129020e-01 -5.41624904e-01 -8.31459880e-01
4.57831383e-01 -7.36901909e-02 -1.19935918e+00 2.33990416e-01
-9.92038906e-01 -1.96531594e-01 -2.18280718e-01 -5.91108277e-02
-3.99936110e-01 -2.90650845e-01 -5.64192712e-01 -8.69333208e-01
-1.83754712e-02 -1.07029247e+00 6.32376790e-01 9.29615228e-04
-6.40451610e-01 -1.26386726e+00 4.05137658e-01 -5.60965121e-01
7.81527460e-01 8.14675510e-01 1.27885675e+00 -6.96561456e-01
-5.46845853e-01 -2.29904398e-01 1.57749459e-01 -2.19793200e-01
2.85122842e-01 6.76649809e-01 -4.99164999e-01 -4.06854838e-01
-3.87545437e-01 1.39792994e-01 9.03685451e-01 4.57701653e-01
4.14547473e-01 -2.45644942e-01 -9.03649867e-01 1.91975072e-01
1.39998364e+00 2.77216375e-01 6.22723281e-01 7.81820714e-02
2.74151802e-01 7.91798532e-01 1.02168936e-02 -2.22245939e-02
-2.01700047e-01 4.27370787e-01 3.93520564e-01 4.79327515e-03
-1.37052119e-01 -1.41479522e-01 5.08329272e-01 8.67437959e-01
-2.73455322e-01 -4.82599169e-01 -1.11991501e+00 3.30715120e-01
-1.52253497e+00 -9.40947771e-01 -3.05945247e-01 2.03492212e+00
8.62558663e-01 5.45548260e-01 -2.52396856e-02 2.16642305e-01
8.39234829e-01 2.00819910e-01 -6.10883474e-01 -4.76677269e-01
-3.12932312e-01 -1.50969522e-02 5.65261483e-01 6.13807976e-01
-4.63514060e-01 5.90805054e-01 7.28163290e+00 4.85691905e-01
-1.21061516e+00 -3.23296309e-01 8.81375492e-01 -8.30995888e-02
-6.61443710e-01 3.33085746e-01 -5.92593670e-01 4.03465092e-01
1.19951892e+00 -3.60069931e-01 6.21207714e-01 1.99542135e-01
5.23481190e-01 -1.37785494e-01 -1.24364042e+00 3.52268144e-02
-4.94878203e-01 -1.63808215e+00 4.19585675e-04 4.81372237e-01
6.44341171e-01 -1.81826189e-01 -2.05424920e-01 -1.34430096e-01
4.33286846e-01 -1.18628395e+00 1.68299943e-01 1.02451468e+00
9.35555458e-01 -5.80341756e-01 4.68766332e-01 1.20766409e-01
-1.02795815e+00 3.41971102e-03 -1.27137199e-01 -1.61651015e-01
2.13915464e-02 9.21765566e-01 -8.27879250e-01 7.63599426e-02
2.30799139e-01 6.49305344e-01 -6.09292030e-01 6.10356271e-01
2.70315915e-01 6.96816683e-01 -5.32054722e-01 -6.73758864e-01
8.46585110e-02 -4.12673503e-01 1.06635118e+00 9.93162990e-01
7.97967538e-02 2.26645619e-02 -2.91956097e-01 1.32736528e+00
7.57609531e-02 -1.90722793e-01 -1.12423956e+00 -6.05380237e-01
6.33243799e-01 1.41537189e+00 -1.30061591e+00 5.45219555e-02
3.48759025e-01 3.24747175e-01 1.50751457e-01 2.86963612e-01
-6.12611890e-01 -5.99618852e-01 9.49514508e-01 5.47339439e-01
-2.27265030e-01 -7.37467185e-02 -1.94023877e-01 -6.02749348e-01
-5.70408165e-01 -6.75439477e-01 -2.52211392e-01 -5.76044977e-01
-1.05606008e+00 4.36218977e-01 5.41467369e-02 -9.40143049e-01
1.65927410e-02 -4.54894513e-01 -7.76107669e-01 6.30100846e-01
-9.22659814e-01 -6.59421146e-01 3.82899540e-04 2.25334659e-01
-3.47010553e-01 1.85566321e-02 6.55537128e-01 -2.11805314e-01
-5.62494516e-01 1.92792490e-01 7.17156976e-02 -2.72023141e-01
3.75162661e-01 -1.18312669e+00 1.04795896e-01 4.88586962e-01
-3.52817535e-01 9.03646767e-01 1.03727245e+00 -9.21704471e-01
-1.84973681e+00 -9.70696092e-01 7.22155750e-01 -4.75567371e-01
9.74713564e-01 -4.39724445e-01 -6.58043206e-01 2.20849216e-01
-4.21653986e-02 9.32846293e-02 2.37065196e-01 -4.61782962e-02
-3.51989381e-02 -2.96404809e-01 -1.18349445e+00 9.44631517e-01
1.36605442e+00 -1.71737209e-01 6.59338161e-02 2.68356383e-01
6.43456221e-01 9.29313079e-02 -1.21391702e+00 4.41207439e-01
7.09695876e-01 -8.66834462e-01 7.41460919e-01 -7.93335855e-01
4.77473110e-01 -4.77448314e-01 3.19185108e-01 -1.59142566e+00
-6.37119114e-01 -7.26829648e-01 1.39936164e-01 1.12900341e+00
7.45901406e-01 -1.06758475e+00 4.06897962e-01 2.50316441e-01
1.35000169e-01 -1.12065446e+00 -9.85548198e-01 -6.89867020e-01
8.31717700e-02 2.07607836e-01 4.77018625e-01 9.48960066e-01
4.91321594e-01 2.80509919e-01 1.39583051e-01 -1.51485085e-01
2.41211340e-01 -2.64460862e-01 5.20441473e-01 -1.46448398e+00
-6.35909811e-02 -7.87863374e-01 -8.70572269e-01 -3.14605027e-01
1.23088837e-01 -1.18318272e+00 -1.90375507e-01 -1.28066719e+00
-3.60546485e-02 -7.11717129e-01 -3.67899120e-01 4.63220894e-01
3.35943073e-01 1.05310157e-01 8.89575407e-02 2.79457390e-01
-7.60066882e-02 1.20141432e-01 1.48394120e+00 -3.14785987e-01
-4.24680591e-01 -3.99556100e-01 -6.12254500e-01 4.48103279e-01
7.20800698e-01 -6.11619949e-01 -4.83645886e-01 4.69335973e-01
6.89449608e-01 2.35319212e-01 2.15671837e-01 -9.42630410e-01
3.24996769e-01 -2.65537858e-01 3.63599449e-01 1.58744022e-01
2.13831097e-01 -6.30669653e-01 7.37940907e-01 1.15940726e+00
-6.35159791e-01 1.22524217e-01 1.26814753e-01 9.13796484e-01
1.27640605e-01 1.75770387e-01 7.62610734e-01 1.21557064e-01
1.18153475e-01 -5.52794896e-02 -8.53420079e-01 -6.99043870e-02
1.16108787e+00 -3.24075997e-01 -7.24323869e-01 -4.90273237e-01
-6.29030287e-01 2.41803363e-01 7.69146264e-01 1.04150288e-01
4.46827441e-01 -1.16640437e+00 -5.50875783e-01 1.90029353e-01
-5.56182303e-02 -6.56724215e-01 -2.94358879e-01 1.23021924e+00
-6.14581108e-01 5.18077970e-01 -2.98332274e-01 -7.96265721e-01
-1.10532463e+00 3.73848945e-01 7.55577862e-01 -3.46978694e-01
-7.82749504e-02 5.22151947e-01 -1.31854266e-01 -3.83771420e-01
-4.53310817e-01 -3.96129280e-01 -2.28964925e-01 6.55769631e-02
-7.90739059e-02 4.61602598e-01 -1.50696710e-01 -4.54429626e-01
-6.08797133e-01 4.31818128e-01 3.58450294e-01 8.36851150e-02
1.22754705e+00 2.37633005e-01 -8.30627739e-01 6.27490282e-01
9.64436829e-01 1.34365574e-01 -7.82115877e-01 5.52823603e-01
2.39248667e-02 -2.04395786e-01 2.51919199e-02 -6.51650250e-01
-1.05345523e+00 2.65992254e-01 1.27510682e-01 9.41794872e-01
9.84645128e-01 4.00725566e-02 1.33136690e-01 4.13114548e-01
3.67708683e-01 -6.56633735e-01 -1.22290410e-01 4.65227574e-01
9.22646224e-01 -6.26120746e-01 6.42224550e-02 -5.32947540e-01
6.52298927e-02 1.19578338e+00 4.77637827e-01 -3.77053708e-01
1.08158970e+00 5.50066352e-01 -4.74134505e-01 -4.05436456e-01
-1.24772465e+00 1.11393154e-01 9.43535343e-02 3.13654065e-01
4.21592534e-01 1.19187310e-02 -7.03423619e-01 -3.31813097e-02
-7.10761058e-04 1.29554868e-02 8.83221507e-01 7.53531039e-01
-4.92009848e-01 -8.85525882e-01 -2.07987726e-01 9.11734879e-01
7.57448971e-02 -5.56137823e-02 -1.01868379e+00 1.02722561e+00
-2.22826585e-01 1.01756299e+00 2.48957708e-01 -6.72100723e-01
1.16893612e-01 -6.73674718e-02 4.63226497e-01 -2.99793541e-01
-8.65077555e-01 -3.29018056e-01 2.33636051e-01 -4.99199986e-01
-4.87196624e-01 -5.26610970e-01 -1.17773175e+00 -7.83061028e-01
-5.62577903e-01 8.66445154e-02 3.93371940e-01 6.09897256e-01
6.88933551e-01 6.60916984e-01 6.16244555e-01 -6.26598358e-01
-1.70810610e-01 -7.02259302e-01 -5.38764179e-01 1.55717075e-01
2.19458401e-01 -7.40655601e-01 -9.04357910e-01 -2.18804121e-01] | [6.307715892791748, 4.61605978012085] |
6964483f-f1e3-4a86-9257-8d765a5b27b7 | boundary-unlearning-rapid-forgetting-of-deep | null | null | http://openaccess.thecvf.com//content/CVPR2023/html/Chen_Boundary_Unlearning_Rapid_Forgetting_of_Deep_Networks_via_Shifting_the_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Chen_Boundary_Unlearning_Rapid_Forgetting_of_Deep_Networks_via_Shifting_the_CVPR_2023_paper.pdf | Boundary Unlearning: Rapid Forgetting of Deep Networks via Shifting the Decision Boundary | The practical needs of the "right to be forgotten" and poisoned data removal call for efficient machine unlearning techniques, which enable machine learning models to unlearn, or to forget a fraction of training data and its lineage. Recent studies on machine unlearning for deep neural networks (DNNs) attempt to destroy the influence of the forgetting data by scrubbing the model parameters. However, it is prohibitively expensive due to the large dimension of the parameter space. In this paper, we refocus our attention from the parameter space to the decision space of the DNN model, and propose Boundary Unlearning, a rapid yet effective way to unlearn an entire class from a trained DNN model. The key idea is to shift the decision boundary of the original DNN model to imitate the decision behavior of the model retrained from scratch. We develop two novel boundary shift methods, namely Boundary Shrink and Boundary Expanding, both of which can rapidly achieve the utility and privacy guarantees. We extensively evaluate Boundary Unlearning on CIFAR-10 and Vggface2 datasets, and the results show that Boundary Unlearning can effectively forget the forgetting class on image classification and face recognition tasks, with an expected speed-up of 17x and 19x, respectively, compared with retraining from the scratch. | ['Chen Wang', 'Kai Peng', 'Gaoyang Liu', 'Weizhuo Gao', 'Min Chen'] | 2023-01-01 | null | null | null | cvpr-2023-1 | ['face-recognition'] | ['computer-vision'] | [ 2.41865978e-01 3.10924083e-01 -3.08943748e-01 -3.26207370e-01
-1.71004191e-01 -7.58138180e-01 8.07772875e-02 -2.73549557e-01
-8.14481676e-01 1.25751722e+00 -1.67555928e-01 -6.29637599e-01
-4.85563837e-02 -6.17962778e-01 -8.67877007e-01 -1.06341410e+00
4.01003540e-01 5.42207301e-01 -1.57307863e-01 4.27276105e-01
-1.25481352e-01 5.97580373e-01 -1.26177120e+00 -1.54258892e-01
8.30017328e-01 9.59858775e-01 -2.62939394e-01 3.56198490e-01
4.18343432e-02 5.46597242e-01 -4.40493077e-01 -5.80704272e-01
5.29380679e-01 -2.39125520e-01 -6.55224741e-01 -9.32092890e-02
5.75563252e-01 -7.79769838e-01 -7.98267126e-01 1.35181081e+00
4.40076917e-01 5.11365868e-02 3.74325335e-01 -1.38889241e+00
-7.62345970e-01 6.88737988e-01 -5.18294573e-01 2.19419435e-01
-6.34570062e-01 -6.49486333e-02 7.60961533e-01 -7.92193949e-01
5.00329077e-01 9.82690454e-01 4.21266079e-01 1.32764244e+00
-1.16878223e+00 -1.39735055e+00 1.78077832e-01 1.25806913e-01
-1.56408000e+00 -6.38410389e-01 3.90994072e-01 -2.23073706e-01
5.45102298e-01 2.34838933e-01 3.93350661e-01 1.21509826e+00
1.79296359e-01 7.48291731e-01 8.37640345e-01 -1.47984833e-01
5.06183207e-01 2.41135210e-01 2.02417895e-01 7.51731873e-01
8.00047696e-01 3.49420339e-01 -3.97728145e-01 -3.86055529e-01
8.96981657e-01 5.98720491e-01 -6.16749942e-01 -6.31791294e-01
-6.81992412e-01 8.18742633e-01 2.64185458e-01 -4.50372249e-02
1.98490219e-03 2.58392572e-01 2.11345851e-01 5.10650933e-01
6.52416229e-01 1.22479714e-01 -8.38185728e-01 1.18528917e-01
-8.08471680e-01 -6.57333359e-02 9.24517810e-01 9.86912072e-01
9.16662455e-01 -1.61971208e-02 -1.65115550e-01 4.86941904e-01
-4.18854989e-02 5.54244280e-01 5.81865847e-01 -6.93045259e-01
1.48797244e-01 4.43176657e-01 -4.71861511e-02 -6.34654105e-01
-2.95319851e-03 -5.91816247e-01 -1.21211886e+00 8.63107890e-02
3.58311176e-01 -5.25026679e-01 -1.27317345e+00 1.94230306e+00
5.15302777e-01 5.64988256e-01 -7.30015039e-02 7.29186773e-01
3.66232604e-01 3.15428644e-01 -1.49606079e-01 -2.97088176e-01
9.76189196e-01 -9.40118253e-01 -7.11809814e-01 -3.07437062e-01
7.96806157e-01 4.92399931e-03 8.86947274e-01 5.32864571e-01
-6.87499762e-01 -1.58374965e-01 -1.13361275e+00 -1.07510410e-01
-1.61578313e-01 -5.43651506e-02 6.73420310e-01 5.89723408e-01
-1.04420662e+00 6.47563994e-01 -8.66888821e-01 5.92593700e-02
1.27646255e+00 6.94737554e-01 -7.41774559e-01 -4.38350469e-01
-1.08684242e+00 4.24984664e-01 5.69101691e-01 9.06768888e-02
-1.25475454e+00 -1.02932107e+00 -5.37584722e-01 1.84695169e-01
6.28991604e-01 -8.44719172e-01 1.00657856e+00 -8.08258653e-01
-1.20247853e+00 8.20725560e-01 -6.67993352e-02 -8.55499566e-01
8.73847008e-01 -3.55611175e-01 -3.03029925e-01 -2.71535575e-01
-3.45054537e-01 6.26429677e-01 1.30862522e+00 -9.66475904e-01
-8.32452953e-01 -7.56462276e-01 -3.12068909e-01 -1.70276731e-01
-8.05429578e-01 -7.04224646e-01 -3.71855199e-01 -7.42509365e-01
5.13380058e-02 -8.77340555e-01 -2.89920300e-01 1.82208017e-01
-3.50612283e-01 5.50029278e-02 1.13443172e+00 -5.53634763e-01
1.34075201e+00 -2.50906205e+00 -2.83923112e-02 1.24732472e-01
8.48806500e-01 7.24664211e-01 -2.47675613e-01 -3.66972089e-01
-1.32690351e-02 3.03066611e-01 -2.93098092e-01 -5.78738630e-01
-1.40502498e-01 8.25353622e-01 -7.47838259e-01 6.45091891e-01
-4.30345178e-01 8.35342467e-01 -8.25279236e-01 -1.07461624e-02
-3.27376038e-01 5.10645151e-01 -9.33483243e-01 -7.78787285e-02
-1.36469409e-01 3.47772211e-01 -2.44399503e-01 4.13344949e-01
1.11978543e+00 -9.49389189e-02 -2.39374787e-02 1.49659336e-01
5.11056364e-01 -5.43181188e-02 -8.35744023e-01 1.30816400e+00
-7.68886656e-02 4.46269155e-01 2.01859310e-01 -6.38144732e-01
8.52725208e-01 2.22601518e-01 8.03378597e-02 -4.55434740e-01
2.46897146e-01 1.33151963e-01 -2.06637099e-01 -2.03136459e-01
1.95541263e-01 -3.83491188e-01 2.58602083e-01 7.09966779e-01
2.38384515e-01 6.11327648e-01 -4.94646341e-01 -6.45663217e-02
1.15851748e+00 -5.63016176e-01 7.76719376e-02 -2.57275045e-01
1.83297932e-01 -6.59132600e-01 1.07349336e+00 8.68479311e-01
-3.72753710e-01 3.95704120e-01 4.56907421e-01 -7.76826024e-01
-7.34667420e-01 -1.07640111e+00 -6.62092566e-02 9.04024601e-01
-1.91783607e-01 -6.06899746e-02 -9.72748816e-01 -1.44985056e+00
4.74669188e-01 7.87060142e-01 -9.67618227e-01 -7.35922873e-01
-4.08116430e-01 -6.50049806e-01 5.46782970e-01 4.74398553e-01
7.33637214e-01 -6.17203832e-01 -2.79394299e-01 -8.95064101e-02
1.77085400e-01 -8.43204677e-01 -7.60402977e-01 3.07446897e-01
-1.14672625e+00 -9.49067175e-01 -5.90364635e-01 -5.19127071e-01
1.13649607e+00 2.02109143e-01 5.68550348e-01 3.33087862e-01
-2.52122730e-01 6.25826940e-02 8.90737399e-02 -5.24696350e-01
-1.56519055e-01 3.25513452e-01 5.55276334e-01 2.82118618e-01
4.75365698e-01 -7.97724485e-01 -6.02626801e-01 3.20182115e-01
-1.10603082e+00 -2.01922894e-01 5.17849445e-01 1.04770696e+00
6.54152632e-01 3.46318036e-01 6.65142059e-01 -1.45977783e+00
3.34934533e-01 -4.56121504e-01 -6.85524344e-01 2.04978168e-01
-9.55460846e-01 2.49306977e-01 4.48420674e-01 -7.73606062e-01
-6.94881380e-01 -4.97720465e-02 1.16523631e-01 -1.10921228e+00
1.88093990e-01 1.64050125e-02 -5.37729621e-01 -1.69893876e-01
3.73166859e-01 1.04474694e-01 -3.75099592e-02 -8.06169093e-01
4.08827871e-01 5.86076319e-01 6.56023979e-01 -2.79246509e-01
8.45195651e-01 6.67124033e-01 -1.73107192e-01 -4.92330819e-01
-1.05276477e+00 -3.48622464e-02 -5.14642060e-01 2.16995791e-01
4.75167751e-01 -8.26208591e-01 -6.64189279e-01 6.00601912e-01
-9.76344168e-01 -3.61555815e-01 -5.73138595e-01 3.72438014e-01
-1.04616517e-02 1.62947744e-01 -4.06612039e-01 -5.54394543e-01
-5.94023645e-01 -6.35643244e-01 3.77739072e-01 4.38983828e-01
1.72978833e-01 -1.04534328e+00 -1.43603042e-01 2.35557377e-01
4.58655149e-01 2.28814147e-02 1.17722237e+00 -1.15580618e+00
-6.00622177e-01 -3.75438452e-01 -2.40891442e-01 7.77981222e-01
2.37094194e-01 -6.05047703e-01 -1.19104993e+00 -9.15915251e-01
4.76002187e-01 -1.03983609e-02 1.32277358e+00 1.19086832e-01
1.47708344e+00 -9.85418975e-01 -6.56788886e-01 1.12766159e+00
1.23291659e+00 1.91955656e-01 5.14153659e-01 -9.47508737e-02
7.52426207e-01 3.69291827e-02 2.15882719e-01 7.67902255e-01
1.56777818e-02 -1.91605270e-01 6.92455888e-01 9.44776759e-02
1.31111905e-01 -6.89383626e-01 1.97458103e-01 6.21709108e-01
3.77172917e-01 -2.03727886e-01 -6.63909256e-01 4.15130436e-01
-1.81343377e+00 -4.02759552e-01 6.50779366e-01 2.55581856e+00
1.05413997e+00 2.16193274e-01 -4.64837462e-01 -6.38901219e-02
6.51164412e-01 -7.14430064e-02 -1.26177788e+00 -1.56894952e-01
3.67405079e-02 2.89741635e-01 9.75659192e-01 6.24231875e-01
-1.01994264e+00 1.14908242e+00 5.71228123e+00 9.55149710e-01
-1.12670803e+00 3.35864514e-01 9.78273630e-01 -5.20077050e-01
-3.28926980e-01 1.72678947e-01 -1.13938904e+00 2.36057729e-01
8.26210976e-01 -2.52879411e-01 8.67314398e-01 1.00067699e+00
-4.06117260e-01 2.96123713e-01 -1.46671760e+00 1.18098676e+00
5.94239384e-02 -1.22749329e+00 4.15171057e-01 3.49059850e-01
6.26480758e-01 8.18440169e-02 2.71856010e-01 4.60265994e-01
4.28217500e-01 -1.11603594e+00 3.19041163e-01 6.03016496e-01
9.66659486e-01 -1.08674407e+00 6.13012969e-01 5.50425768e-01
-4.37309206e-01 -2.79594839e-01 -7.76323915e-01 3.76714133e-02
-4.62031543e-01 7.53607035e-01 -9.76135850e-01 3.72887813e-02
8.14718068e-01 4.08395141e-01 -5.23807526e-01 8.38300824e-01
-3.12891752e-01 7.58032858e-01 -3.28544021e-01 7.77554885e-02
-1.19726956e-01 -1.41945034e-01 5.59590578e-01 6.50937021e-01
3.55221123e-01 1.05821505e-01 -2.96240360e-01 7.78891325e-01
-8.00810993e-01 -3.31120759e-01 -6.62033677e-01 -1.83070496e-01
6.79383099e-01 9.06114101e-01 -3.45284343e-01 -1.06797136e-01
1.36117414e-01 1.13360536e+00 5.00618815e-01 4.97158378e-01
-8.92101049e-01 -4.51813877e-01 9.63385165e-01 1.42409429e-01
5.38777947e-01 -2.03245599e-02 -2.19487697e-01 -1.29306936e+00
-1.54290756e-03 -7.13515699e-01 5.74629009e-01 -2.80524403e-01
-1.12984478e+00 5.57714820e-01 -4.45149601e-01 -7.28621066e-01
1.97961524e-01 -3.78738552e-01 -1.96319535e-01 5.70178449e-01
-1.39525914e+00 -6.89031839e-01 5.89506775e-02 6.51474118e-01
3.97340171e-02 -2.28580981e-01 7.09635198e-01 4.84455705e-01
-7.67082512e-01 1.25296962e+00 4.87097353e-01 3.82936388e-01
6.40267730e-01 -7.96514869e-01 1.33118957e-01 7.68290043e-01
1.91113815e-01 7.09418774e-01 4.98086810e-01 -4.71963078e-01
-1.28821123e+00 -1.43393457e+00 8.83542836e-01 -4.17997897e-01
3.43449146e-01 -8.02487373e-01 -1.11526191e+00 1.07001412e+00
-1.77461863e-01 4.36091721e-01 7.40535259e-01 -6.80431426e-02
-5.95141709e-01 -5.12334406e-01 -1.43116319e+00 6.30550683e-01
1.13097572e+00 -3.97563905e-01 -3.56157333e-01 -8.14675465e-02
1.10242867e+00 -1.94277510e-01 -3.84340465e-01 4.24489766e-01
4.71324384e-01 -5.88666797e-01 7.01902211e-01 -1.06096005e+00
-3.41758251e-01 -6.78181052e-02 -8.22650716e-02 -1.24294221e+00
-4.92369123e-02 -1.20286989e+00 -6.00721955e-01 1.39799833e+00
2.87781447e-01 -1.05687928e+00 1.26310408e+00 8.10066521e-01
3.41763109e-01 -6.66864395e-01 -1.43305266e+00 -5.66639245e-01
2.58022845e-01 -1.65602595e-01 9.50661063e-01 1.08690631e+00
-4.85619366e-01 3.47927034e-01 -5.65093637e-01 1.48383915e-01
7.08057046e-01 -1.26232371e-01 6.95687652e-01 -1.31318140e+00
-1.85299829e-01 -1.38897926e-01 -2.05723658e-01 -1.13294077e+00
2.56298095e-01 -1.01256108e+00 -2.59877056e-01 -7.73851812e-01
2.37745568e-01 -4.87516493e-01 -7.81839669e-01 8.94437790e-01
2.56908555e-02 -1.67476624e-01 -5.66636138e-02 4.24485952e-02
-4.67325002e-01 8.13200414e-01 1.42519927e+00 -2.27862269e-01
-9.95685682e-02 3.69823724e-01 -1.06571555e+00 7.25724697e-01
7.31832147e-01 -9.63355601e-01 -7.12491691e-01 -6.69872522e-01
1.71099052e-01 -2.39559039e-01 3.43622863e-01 -9.09451246e-01
4.66939420e-01 2.04464905e-02 3.39215010e-01 -3.91751915e-01
9.18331929e-03 -1.10510707e+00 2.42107306e-02 5.91785073e-01
-2.76522934e-01 -1.55105636e-01 2.81928182e-01 1.12189317e+00
2.02750549e-01 -1.32295728e-01 1.00480759e+00 3.66107635e-02
-3.06123257e-01 1.05315864e+00 5.40634021e-02 2.38517746e-01
1.10314131e+00 7.41330907e-02 -3.89753163e-01 -1.22419521e-01
-8.10504317e-01 3.40380669e-01 3.33323747e-01 3.67016405e-01
8.99389267e-01 -1.26505053e+00 -4.17289913e-01 8.97370279e-01
-2.75780737e-01 3.67959261e-01 3.73982817e-01 5.47891796e-01
-1.92535251e-01 4.17086542e-01 -3.83331403e-02 -2.65285075e-01
-1.05961335e+00 9.91844356e-01 4.06666130e-01 -2.05926329e-01
-5.18728614e-01 1.20158422e+00 5.45672774e-01 -2.88872749e-01
7.05804706e-01 -4.95537013e-01 3.83741319e-01 1.86821818e-02
7.59238005e-01 2.36332074e-01 8.44995230e-02 -7.19272047e-02
-2.77419180e-01 -7.42055178e-02 -8.35616708e-01 2.08798409e-01
1.22063076e+00 -1.04946069e-01 -2.83385932e-01 3.13459113e-02
1.42771006e+00 -2.26691797e-01 -1.60742593e+00 -6.62728250e-01
-1.08890012e-01 -5.39730668e-01 2.97915876e-01 -7.39656627e-01
-1.45011091e+00 1.03109932e+00 6.56694949e-01 -2.36927822e-01
1.17221129e+00 -1.29345208e-01 1.13527250e+00 6.26676798e-01
5.41188300e-01 -1.09312201e+00 -2.23358035e-01 4.90125209e-01
5.22278070e-01 -9.39446509e-01 -1.87577546e-01 2.18496948e-01
-2.40893304e-01 8.04899871e-01 7.40760744e-01 1.12894677e-01
1.22575784e+00 3.00786823e-01 -2.38434896e-01 1.96338788e-01
-9.97301757e-01 5.24471760e-01 -2.25569084e-01 4.51420009e-01
-5.89177847e-01 4.07406166e-02 2.27240145e-01 1.25210238e+00
5.81787415e-02 4.30801630e-01 2.75305241e-01 9.62202072e-01
-4.75891471e-01 -9.97649789e-01 2.07210314e-02 5.91868401e-01
-5.53479314e-01 -2.07604200e-01 -2.61276364e-01 7.11972952e-01
2.44144961e-01 4.36249197e-01 8.93606395e-02 -6.99070632e-01
1.22599855e-01 8.83009806e-02 2.06786543e-01 -4.72331405e-01
-1.10279173e-01 -3.10310036e-01 -4.33376133e-01 -4.15314168e-01
3.08641940e-01 -4.58578408e-01 -1.15528035e+00 -6.78886592e-01
-5.39072812e-01 2.93658495e-01 3.26577634e-01 9.02203679e-01
7.23100066e-01 1.44739300e-01 7.25541949e-01 -2.04708099e-01
-1.11437321e+00 -7.10488737e-01 -8.80515099e-01 -4.22211923e-02
7.77209759e-01 -5.10707200e-01 -7.96773076e-01 -1.87334239e-01] | [9.547991752624512, 3.664921760559082] |
e75cb865-718c-4e6d-b0e3-e51fb33fea76 | on-field-player-workload-exposure-and-knee | 1809.08016 | null | https://arxiv.org/abs/1809.08016v3 | https://arxiv.org/pdf/1809.08016v3.pdf | On-field player workload exposure and knee injury risk monitoring via deep learning | In sports analytics, an understanding of accurate on-field 3D knee joint moments (KJM) could provide an early warning system for athlete workload exposure and knee injury risk. Traditionally, this analysis has relied on captive laboratory force plates and associated downstream biomechanical modeling, and many researchers have approached the problem of portability by extrapolating models built on linear statistics. An alternative approach would be to capitalize on recent advances in deep learning. In this study, using the pre-trained CaffeNet convolutional neural network (CNN) model, multivariate regression of marker-based motion capture to 3D KJM for three sports-related movement types were compared. The strongest overall mean correlation to source modeling of 0.8895 was achieved over the initial 33 % of stance phase for sidestepping. The accuracy of these mean predictions of the three critical KJM associated with anterior cruciate ligament (ACL) injury demonstrate the feasibility of on-field knee injury assessment using deep learning in lieu of laboratory embedded force plates. This multidisciplinary research approach significantly advances machine representation of real-world physical models with practical application for both community and professional level athletes. | ['Ajmal Mian', 'William R. Johnson', 'Jacqueline A. Alderson', 'David G. Lloyd'] | 2018-09-21 | null | null | null | null | ['sports-analytics'] | ['computer-vision'] | [-5.80687404e-01 -3.39810401e-01 -6.23165369e-01 2.07755610e-01
-8.50279093e-01 -1.17424922e-02 1.72287188e-02 2.91642904e-01
-8.07930410e-01 4.55148041e-01 4.47365403e-01 -2.28836760e-01
-2.64178574e-01 -5.07664800e-01 -9.66041327e-01 8.52442458e-02
-8.59187603e-01 3.78164023e-01 1.72106415e-01 -5.43955207e-01
3.14579129e-01 6.10122681e-01 -1.13774061e+00 3.23808849e-01
3.25132340e-01 9.38078642e-01 9.14409198e-03 8.53432894e-01
5.77055633e-01 8.55374157e-01 -6.48920774e-01 -2.39283457e-01
3.06761831e-01 5.28189875e-02 -4.79589522e-01 -1.99589685e-01
5.76768339e-01 -5.80161989e-01 -6.44071102e-01 2.30493605e-01
9.52026784e-01 2.65021741e-01 3.94807279e-01 -1.05377495e+00
-5.49708962e-01 2.34751701e-01 -5.51303625e-01 8.85683775e-01
4.19134438e-01 8.42354655e-01 5.69856286e-01 -9.55972493e-01
6.35618687e-01 7.47651935e-01 1.39268386e+00 1.48883790e-01
-1.33606863e+00 -6.03996396e-01 -7.01283813e-01 5.10079980e-01
-1.15554309e+00 9.78874117e-02 8.43056500e-01 -9.44520593e-01
1.22672164e+00 1.50593221e-02 1.49903643e+00 1.08713758e+00
1.31250417e+00 6.32131040e-01 9.65556622e-01 -2.28037946e-02
6.10944703e-02 -8.24294150e-01 5.75486049e-02 3.06351036e-01
5.50686061e-01 3.87917161e-01 -9.73075747e-01 7.10800439e-02
1.08457994e+00 1.77747961e-02 1.20287418e-01 -2.39446998e-01
-1.17977166e+00 6.01855338e-01 5.75314999e-01 -2.24502280e-01
-4.10995513e-01 7.08078563e-01 7.77466238e-01 -5.73449954e-02
5.99568129e-01 6.06574297e-01 -2.93236703e-01 -7.32302964e-01
-1.24489212e+00 9.13294435e-01 3.25203598e-01 6.70233294e-02
1.39320105e-01 4.51016247e-01 1.51360691e-01 7.79334188e-01
1.58081189e-01 2.39487126e-01 6.01866961e-01 -1.03115404e+00
3.66148949e-01 4.69053179e-01 -1.79460913e-01 -1.09310615e+00
-8.36306870e-01 -8.57416987e-01 -3.25498968e-01 7.08783746e-01
6.86825216e-01 -1.90205306e-01 -8.58685970e-01 1.04119945e+00
6.91959411e-02 -4.41458300e-02 -4.40225869e-01 1.61907113e+00
4.88808900e-01 -1.73021168e-01 3.44435066e-01 7.11468399e-01
1.42248297e+00 -5.27841747e-01 -2.34499201e-01 -6.49161398e-01
8.95774186e-01 -6.73452318e-01 1.13975739e+00 3.86524081e-01
-1.26822972e+00 -6.46635413e-01 -1.32184196e+00 -3.52733672e-01
-1.79522950e-02 2.15350091e-01 7.54027247e-01 2.67984986e-01
-6.90614045e-01 1.02326167e+00 -1.35785127e+00 8.38819966e-02
3.60795647e-01 4.47457939e-01 -7.46155202e-01 1.27995774e-01
-1.33193827e+00 1.45003331e+00 -2.86427215e-02 6.03319526e-01
-9.33500350e-01 -1.29013968e+00 -7.22675323e-01 -7.50242621e-02
1.19433291e-01 -8.33705246e-01 9.67490077e-01 -3.69898379e-01
-1.17898166e+00 9.70092654e-01 6.27427459e-01 -7.41472960e-01
7.59015858e-01 -1.11521661e+00 -4.10065241e-02 2.04385221e-01
2.22481892e-01 2.45653614e-01 6.89690337e-02 -5.58278680e-01
1.70764238e-01 -4.08365518e-01 -2.06015110e-01 5.53572364e-03
2.49192063e-02 2.40592547e-02 8.46303254e-02 -7.40048528e-01
4.73039523e-02 -1.07593119e+00 -3.61854166e-01 3.41918558e-01
-1.40184686e-01 1.52104646e-01 2.31076449e-01 -1.06044126e+00
9.31442261e-01 -1.80385256e+00 -1.59201726e-01 6.62083104e-02
4.94659424e-01 1.82669327e-01 3.94151360e-01 5.36548376e-01
-2.93168902e-01 -1.30366892e-01 2.23658442e-01 2.65996814e-01
-4.31521833e-01 -7.35277832e-02 1.25350669e-01 7.70843089e-01
3.81503552e-01 1.16203034e+00 -8.65180969e-01 -5.90738118e-01
3.47200930e-01 3.84682596e-01 -8.32682669e-01 1.52754977e-01
3.01423341e-01 2.80824929e-01 5.34872524e-02 6.44577444e-01
2.52907455e-01 5.41257858e-02 1.95317343e-02 -3.03642154e-01
3.70562077e-02 1.22734264e-01 -8.91694546e-01 2.09153223e+00
-3.61402512e-01 9.35346067e-01 -1.08666994e-01 -8.66861105e-01
9.66300249e-01 1.87936619e-01 1.06173670e+00 -4.67578828e-01
3.27503651e-01 5.35816193e-01 6.80009365e-01 -7.77071476e-01
3.90463144e-01 -3.95651579e-01 -2.80648172e-01 1.67038620e-01
3.69004935e-01 1.33271188e-01 8.62985328e-02 -9.66423154e-02
1.07626402e+00 7.52939045e-01 -1.53921545e-01 -4.84353274e-01
-2.41940126e-01 5.68449914e-01 4.32833850e-01 4.56161767e-01
-5.79476953e-01 7.42705762e-01 5.37689447e-01 -8.37066889e-01
-1.40944064e+00 -1.45089102e+00 9.11349952e-02 7.02061296e-01
-2.88805664e-01 -2.56245106e-01 -6.09237015e-01 4.07062650e-01
5.95280349e-01 3.53559479e-02 -6.17143095e-01 -7.41417706e-01
-1.09566462e+00 -6.25920773e-01 9.42905426e-01 1.30774558e+00
7.13461190e-02 -7.38274038e-01 -9.57305253e-01 5.38845599e-01
-1.03965446e-01 -8.61446798e-01 -5.66519909e-02 1.92376897e-01
-1.14878035e+00 -1.22439229e+00 -8.49398971e-01 -6.09944880e-01
-1.75352991e-01 -2.40241408e-01 1.20187044e+00 1.46439642e-01
-8.09975028e-01 2.18731195e-01 4.46012020e-02 -4.33387786e-01
-1.33767307e-01 9.88911092e-02 5.18939018e-01 -5.53429842e-01
5.30582130e-01 -6.15218461e-01 -1.08484948e+00 -1.05364405e-01
-5.09720027e-01 4.72926162e-03 8.85806680e-01 8.04986477e-01
6.31261230e-01 -8.67671788e-01 5.12573600e-01 4.93297428e-02
9.14620221e-01 -7.02638686e-01 5.45059815e-02 -3.11169595e-01
-4.87653673e-01 -3.64784956e-01 3.21767122e-01 -6.52059019e-01
-5.49324274e-01 -1.32885724e-01 -3.09219688e-01 -8.37656081e-01
2.99164187e-02 9.67366397e-01 5.43693602e-01 1.68775693e-02
1.19467878e+00 -1.41523182e-01 7.94627547e-01 -4.84373719e-01
-1.03796646e-01 1.77212358e-01 1.08350122e+00 -9.59171414e-01
2.17526883e-01 3.32239538e-01 6.32044315e-01 -6.41621828e-01
-5.91734529e-01 -3.76947939e-01 -8.32777917e-01 -9.80126798e-01
8.81463170e-01 -1.03204250e+00 -1.13924158e+00 4.04405206e-01
-7.08047807e-01 -5.16511798e-01 -2.86212325e-01 1.34363294e+00
-9.35878992e-01 4.33965951e-01 -1.09175694e+00 -5.77784657e-01
-3.92102271e-01 -1.12352169e+00 1.10999823e+00 -2.28319123e-01
-7.33084977e-01 -7.65731812e-01 3.46482486e-01 6.53786421e-01
3.63575369e-01 9.31576431e-01 5.84030986e-01 -5.39980114e-01
-9.28819850e-02 -8.59125674e-01 3.66413563e-01 4.52495843e-01
-3.07098567e-01 -1.64091706e-01 -3.86937022e-01 -2.28373781e-02
-1.69457629e-01 -6.39878631e-01 4.07188535e-01 6.19926274e-01
8.98950398e-01 3.60629916e-01 1.27653688e-01 5.18091142e-01
1.40481520e+00 -5.32299578e-01 6.07633114e-01 8.98308516e-01
7.29381979e-01 4.88611817e-01 6.44144893e-01 2.41822347e-01
2.21857689e-02 6.63077116e-01 3.23545575e-01 -1.39342561e-01
-4.76037115e-01 -2.81354576e-01 1.56339288e-01 7.97975719e-01
-7.40208507e-01 5.16138911e-01 -1.40632939e+00 7.05760062e-01
-1.51068962e+00 -8.14252019e-01 -4.99954432e-01 2.02818298e+00
7.63216734e-01 6.25164747e-01 2.52642602e-01 1.39576718e-01
3.56980413e-01 -2.39734262e-01 -5.61915100e-01 -4.57485825e-01
2.02832613e-02 4.86031562e-01 7.90171981e-01 -2.10916083e-02
-6.72310770e-01 4.31276083e-01 6.78618908e+00 4.79983926e-01
-1.23553741e+00 1.17904074e-01 4.45136100e-01 -7.61068404e-01
4.31157798e-01 -5.64698838e-02 -3.56419444e-01 4.50357974e-01
1.49184263e+00 -1.91085041e-01 -1.15656525e-01 9.70660329e-01
6.38868570e-01 -3.87581080e-01 -9.64855909e-01 5.78362703e-01
-1.83084980e-01 -1.59380770e+00 -5.03827870e-01 1.82766959e-01
3.18121672e-01 3.32859665e-01 -7.69962296e-02 2.23053545e-01
-7.93915763e-02 -9.59228754e-01 8.45389247e-01 1.08551216e+00
8.20930243e-01 -5.61082721e-01 8.06563258e-01 1.90851688e-01
-1.02195978e+00 2.74142083e-02 -1.40807480e-01 -6.58086300e-01
3.28391582e-01 4.18764710e-01 -8.36426914e-01 2.25128978e-01
8.69703531e-01 4.76629317e-01 -3.78959924e-01 1.13924539e+00
1.49010122e-01 8.12726200e-01 -3.85847718e-01 2.57285208e-01
5.39867878e-02 3.47470909e-01 5.78110158e-01 1.00092542e+00
8.88349786e-02 -3.89586419e-01 3.12299639e-01 6.54453635e-01
2.52922952e-01 8.56602713e-02 -5.32087326e-01 1.15512483e-01
-1.40340075e-01 9.81300950e-01 -4.58030522e-01 -2.86592431e-02
-3.23932588e-01 1.67874441e-01 3.12439561e-01 2.70182500e-03
-7.28776872e-01 -2.78523088e-01 9.72523332e-01 9.65884268e-01
-3.94512296e-01 -9.00907695e-01 -8.14398885e-01 -1.00029182e+00
2.87939280e-01 -5.66764891e-01 1.94162697e-01 -1.17678058e+00
-1.28818095e+00 -1.41615003e-01 1.61953583e-01 -1.47119188e+00
-4.76652086e-01 -9.49330866e-01 -5.72945118e-01 1.11892200e+00
-9.32759762e-01 -1.04682899e+00 -1.46567792e-01 3.28621149e-01
5.24669588e-01 1.29456550e-01 3.40317428e-01 3.88582677e-01
-2.43064657e-01 2.28980258e-01 -2.21709132e-01 6.08397365e-01
7.03607500e-01 -1.02189314e+00 2.40070149e-01 7.28205562e-01
-4.76250291e-01 7.37604737e-01 9.14894104e-01 -1.20216358e+00
-1.60915220e+00 -6.85384035e-01 4.75737393e-01 -5.78223109e-01
1.07595718e+00 -3.98177952e-02 -8.86379480e-01 7.50833452e-01
-2.45641455e-01 2.66344756e-01 9.89981890e-01 5.51416613e-02
1.39458984e-01 -8.27103332e-02 -4.32100207e-01 7.59158075e-01
8.87816727e-01 -3.50180537e-01 -9.62405264e-01 4.55018610e-01
1.12399474e-01 -9.66930807e-01 -1.69785941e+00 5.85933149e-01
1.11088836e+00 -6.24308884e-01 1.23652363e+00 -1.28649032e+00
1.23396719e+00 7.06895366e-02 2.82446176e-01 -1.05283725e+00
-2.43670881e-01 -3.17610018e-02 -3.36095691e-01 2.38947928e-01
-1.69165075e-01 -2.76974831e-02 1.24463010e+00 8.61595213e-01
-8.54506016e-01 -1.37430286e+00 -1.17683101e+00 -9.16035533e-01
5.81549823e-01 -9.61919963e-01 -2.45186035e-03 4.19219673e-01
1.22337900e-01 -2.30061099e-01 -6.56949162e-01 -1.64232254e-01
5.49399674e-01 -3.18688780e-01 9.56186771e-01 -1.09866822e+00
-2.82602906e-01 -4.31526363e-01 -1.31025326e+00 -4.30028945e-01
-1.98008552e-01 -8.70917559e-01 -9.01694521e-02 -1.50362265e+00
-1.46366149e-01 1.46531481e-02 -3.98076087e-01 1.86669081e-01
-9.39880311e-02 6.77920878e-01 2.15381086e-01 5.34187138e-01
1.70234218e-01 1.10058077e-01 1.48212957e+00 5.38008995e-02
-2.23944924e-04 -1.72578722e-01 -1.14093564e-01 5.78442991e-01
8.18362415e-01 -5.05855560e-01 1.04120083e-01 -4.94622529e-01
3.24777097e-01 3.02059174e-01 1.06021929e+00 -1.60347223e+00
2.06043459e-02 -2.19579890e-01 9.66956019e-01 -4.97806013e-01
4.32143509e-01 -3.65669250e-01 3.45592499e-01 9.44500625e-01
-3.48535478e-01 2.65105695e-01 3.90595108e-01 2.82629371e-01
-5.12408651e-02 1.82262689e-01 4.09072429e-01 -2.70266473e-01
-6.48593485e-01 2.17872098e-01 -7.62440622e-01 2.25940898e-01
7.82325745e-01 -7.50162482e-01 -2.37281218e-01 -1.14141658e-01
-1.36301112e+00 -8.33817050e-02 1.01361267e-01 5.04687250e-01
7.22155929e-01 -1.35853863e+00 -7.81400740e-01 6.57553896e-02
-1.77910533e-02 -3.42716724e-01 5.95312178e-01 1.28053355e+00
-1.29162562e+00 3.76310945e-01 -1.04438806e+00 -8.93674672e-01
-5.70531666e-01 2.08261162e-01 5.21844447e-01 -2.95918167e-01
-8.10820699e-01 4.98280019e-01 -6.83208883e-01 9.45369452e-02
-3.42598557e-01 -2.78672248e-01 7.70862177e-02 -3.71345192e-01
-8.07482228e-02 7.41641879e-01 3.15830350e-01 -6.94669545e-01
-2.66373098e-01 2.38790721e-01 2.96221405e-01 -5.00607910e-03
1.40184760e+00 3.51509422e-01 2.59598225e-01 7.46265411e-01
1.14360797e+00 -1.40519172e-01 -1.52553320e+00 1.92846626e-01
-6.93023298e-03 -5.45986116e-01 6.33470416e-02 -6.41908705e-01
-1.09399927e+00 1.16277182e+00 9.31111157e-01 -4.09259140e-01
6.17080629e-01 -5.70677631e-02 1.16516662e+00 1.21326491e-01
2.56262511e-01 -1.31579757e+00 2.53004432e-01 -3.50818262e-02
9.57606733e-01 -9.69241321e-01 4.72568780e-01 -7.39348587e-04
-4.16406035e-01 1.54237258e+00 1.00623477e+00 -1.07453954e+00
7.18731105e-01 3.56672227e-01 2.57161915e-01 -5.66160381e-01
-4.02261406e-01 2.36554310e-01 4.61631209e-01 4.79285747e-01
7.51389325e-01 1.20263591e-01 -8.59474063e-01 7.30030477e-01
-4.64065343e-01 6.29020333e-01 5.72249174e-01 1.38982081e+00
-3.40864778e-01 -5.58828533e-01 -5.03472805e-01 9.62525845e-01
-1.06475437e+00 9.07552540e-02 3.54670510e-02 1.24084079e+00
3.44701670e-03 3.85364741e-01 2.77621180e-01 -5.84009945e-01
6.34434640e-01 1.30279511e-01 3.45754683e-01 -5.02109468e-01
-9.91778433e-01 1.03529848e-01 1.88203320e-01 -6.35500073e-01
-1.52390838e-01 -5.83662212e-01 -9.93194461e-01 -5.19658089e-01
-1.70707375e-01 -2.58804828e-01 7.09793627e-01 7.68793046e-01
4.01128054e-01 7.53911197e-01 -1.24702714e-01 -1.34798205e+00
-4.70825285e-01 -1.19735491e+00 -6.84990168e-01 5.44697821e-01
-4.09075394e-02 -1.31917751e+00 5.55277839e-02 8.17651078e-02] | [6.980124473571777, -0.025964412838220596] |
69886ae4-dddd-45d8-a0ab-4a4f12e03c6b | an-analysis-of-classification-approaches-for | 2301.13507 | null | https://arxiv.org/abs/2301.13507v1 | https://arxiv.org/pdf/2301.13507v1.pdf | An Analysis of Classification Approaches for Hit Song Prediction using Engineered Metadata Features with Lyrics and Audio Features | Hit song prediction, one of the emerging fields in music information retrieval (MIR), remains a considerable challenge. Being able to understand what makes a given song a hit is clearly beneficial to the whole music industry. Previous approaches to hit song prediction have focused on using audio features of a record. This study aims to improve the prediction result of the top 10 hits among Billboard Hot 100 songs using more alternative metadata, including song audio features provided by Spotify, song lyrics, and novel metadata-based features (title topic, popularity continuity and genre class). Five machine learning approaches are applied, including: k-nearest neighbours, Naive Bayes, Random Forest, Logistic Regression and Multilayer Perceptron. Our results show that Random Forest (RF) and Logistic Regression (LR) with all features (including novel features, song audio features and lyrics features) outperforms other models, achieving 89.1% and 87.2% accuracy, and 0.91 and 0.93 AUC, respectively. Our findings also demonstrate the utility of our novel music metadata features, which contributed most to the models' discriminative performance. | ['Valerie J. Gillet', 'Frank Hopfgartner', 'David Cameron', 'Morgan Harvey', 'Mengyisong Zhao'] | 2023-01-31 | null | null | null | null | ['music-information-retrieval'] | ['music'] | [ 9.72622707e-02 -5.70984304e-01 -5.95105469e-01 -1.24166653e-01
-9.79125857e-01 -7.43958652e-01 4.45995301e-01 2.90702969e-01
-3.92457128e-01 7.24452555e-01 6.65747583e-01 2.38852262e-01
-8.10879827e-01 -6.80692255e-01 -3.16900134e-01 -4.46665943e-01
-3.61612290e-01 4.35734600e-01 1.91448689e-01 1.00258783e-01
9.04456675e-01 2.15384036e-01 -1.97997189e+00 6.43613994e-01
2.31373906e-01 1.30906868e+00 9.13777947e-02 7.24637032e-01
1.37470722e-01 8.91695142e-01 -6.79718673e-01 -2.13306695e-01
1.12187088e-01 -1.25440255e-01 -7.41292775e-01 -8.51580143e-01
2.11929291e-01 -1.49506152e-01 -7.12146759e-02 3.79862577e-01
6.98092282e-01 -2.81986524e-03 7.93647945e-01 -1.09070110e+00
-1.26667336e-01 1.31259656e+00 -5.89280784e-01 3.76923680e-01
8.69449556e-01 -3.71987194e-01 1.55829322e+00 -6.70298457e-01
5.39546788e-01 7.14105785e-01 1.09638286e+00 -4.21824120e-02
-1.15812254e+00 -1.19859242e+00 -3.57374638e-01 7.66212463e-01
-1.68859112e+00 -2.04295814e-01 6.82085812e-01 -9.47789311e-01
8.24866414e-01 7.96768606e-01 8.82029712e-01 6.69000387e-01
1.74790174e-01 6.29056096e-01 1.21483326e+00 -6.75628364e-01
4.86848578e-02 3.57482493e-01 2.81306077e-03 -1.61352009e-01
-2.96201169e-01 1.73820153e-01 -1.31676495e+00 -6.69083834e-01
4.32269096e-01 3.94178182e-02 -2.15212494e-01 2.70554632e-01
-1.36655879e+00 7.40377665e-01 1.87985584e-01 5.34481704e-01
-4.35799122e-01 -1.04183853e-01 2.84671992e-01 3.48265320e-01
3.49575788e-01 9.22120035e-01 -7.06146538e-01 -6.83077514e-01
-1.53855073e+00 4.41545069e-01 8.42972994e-01 2.94819593e-01
4.22857851e-01 -1.77164823e-01 -3.79197657e-01 1.25464928e+00
2.54710227e-01 2.98506826e-01 8.02994132e-01 -6.79639161e-01
-8.47676024e-02 3.93353373e-01 -1.22513883e-01 -1.08389521e+00
-3.84471238e-01 -8.79802465e-01 -4.60365951e-01 -2.67425209e-01
7.88259432e-02 4.48662311e-01 -1.99665681e-01 1.43712091e+00
-1.09249987e-01 4.53890234e-01 -3.83763701e-01 7.12769151e-01
8.27267468e-01 6.38909578e-01 -9.10235718e-02 -4.52058494e-01
8.78567457e-01 -7.21143305e-01 -2.93843091e-01 4.97445881e-01
1.95428491e-01 -1.42972779e+00 1.05027282e+00 9.88366723e-01
-9.24515605e-01 -7.17097759e-01 -8.72525096e-01 5.92673182e-01
-1.41784087e-01 -1.03954021e-02 5.60047507e-01 3.98256838e-01
-5.73508084e-01 9.67083573e-01 -1.92722589e-01 -2.49568559e-02
-2.76336423e-03 6.18416727e-01 3.38969752e-02 3.38095397e-01
-1.05659664e+00 3.52125198e-01 2.97782570e-01 -6.86826646e-01
-6.43053830e-01 -1.01467431e+00 -4.08926830e-02 7.61200562e-02
-6.30818233e-02 -3.36490929e-01 1.41215706e+00 -7.13914990e-01
-1.24897623e+00 5.85583091e-01 1.56866729e-01 -3.98704052e-01
3.38101722e-02 -3.80255908e-01 -5.64084172e-01 -1.40159532e-01
2.47281700e-01 4.12529826e-01 5.34001112e-01 -7.23617733e-01
-9.36378479e-01 -3.71784925e-01 -3.77652884e-01 1.41036943e-01
-5.70772171e-01 4.63609815e-01 -1.56775653e-01 -8.65180910e-01
1.25152748e-02 -9.35397148e-01 1.40625551e-01 -8.72516572e-01
-4.48568404e-01 -3.91733974e-01 1.21837214e-01 -8.54818106e-01
2.14697218e+00 -2.09220362e+00 -4.83026635e-03 3.85238051e-01
-2.36982584e-01 -1.76194310e-01 3.01183254e-01 7.31397092e-01
1.14648104e-01 -1.80824921e-02 4.40540761e-01 4.83772486e-01
-6.68786168e-02 -3.62771153e-01 -4.89877880e-01 -1.21832512e-01
-6.77977026e-01 5.45412362e-01 -5.19474089e-01 -5.04779220e-01
1.31378666e-01 3.52936596e-01 -6.58163667e-01 1.43411368e-01
4.20962423e-02 2.63920665e-01 -8.90367702e-02 8.91376793e-01
4.28359732e-02 3.23927775e-02 -1.64370298e-01 -5.35386764e-02
-3.57800633e-01 6.68267846e-01 -1.34872651e+00 1.52887166e+00
-1.83358088e-01 5.56550801e-01 -5.94908297e-01 -2.44180769e-01
1.16487598e+00 4.13664609e-01 7.11164355e-01 -5.83082080e-01
-3.20666909e-01 3.63964319e-01 -6.36798963e-02 -3.42065871e-01
4.71540630e-01 -1.94752678e-01 -1.87563226e-01 3.56354535e-01
-7.50968978e-02 3.02313447e-01 -1.01872303e-01 -3.78050506e-02
1.04574621e+00 3.98860089e-02 5.50376058e-01 -5.61555251e-02
2.59758830e-01 8.13595504e-02 5.84310710e-01 9.52379048e-01
1.83057770e-01 7.30565429e-01 4.02373374e-02 -2.23621905e-01
-8.71418893e-01 -9.60177124e-01 -3.08200896e-01 1.52857089e+00
-3.35324258e-01 -8.90410066e-01 -3.19086969e-01 -1.28838688e-01
1.24608286e-01 5.87395251e-01 -2.26671353e-01 1.38040651e-02
-2.27786034e-01 -5.73460937e-01 7.58121014e-01 2.95696944e-01
1.19282296e-02 -1.25498760e+00 -2.52501875e-01 4.59764808e-01
-3.14193577e-01 -3.25604051e-01 -3.52172136e-01 2.93026626e-01
-8.72835934e-01 -8.20559740e-01 -4.88867730e-01 -6.03010058e-01
-5.35592318e-01 7.46506751e-02 1.13463712e+00 -1.36288688e-01
-5.01950681e-01 2.71067560e-01 -5.72945714e-01 -4.64127839e-01
-5.89751359e-03 4.39064562e-01 3.75595272e-01 -1.55445471e-01
4.36078042e-01 -9.71337378e-01 -6.00944400e-01 4.59251195e-01
-4.84988838e-01 1.44236870e-02 6.24260128e-01 5.89798808e-01
8.15564930e-01 7.97005519e-02 7.04197645e-01 -6.76975191e-01
5.13425231e-01 -6.05396569e-01 1.43488571e-01 1.03194565e-02
-9.73068237e-01 -5.95902324e-01 1.79900214e-01 -6.60689235e-01
-4.08131212e-01 7.34762028e-02 -1.68179020e-01 -6.95777088e-02
-4.07487035e-01 7.05154419e-01 1.92674607e-01 4.87552546e-02
8.99806201e-01 2.76311725e-01 -5.82446635e-01 -1.19660771e+00
-7.76303634e-02 1.31473899e+00 4.72697109e-01 -3.72505605e-01
6.05178297e-01 3.51904631e-02 -2.11634532e-01 -7.43853688e-01
-8.95594478e-01 -1.19588792e+00 -8.44199538e-01 -6.87076867e-01
4.14561570e-01 -9.47615087e-01 -6.15164340e-01 1.59753814e-01
-5.37522316e-01 3.00002992e-01 -3.13636243e-01 1.09740794e+00
-6.24211133e-01 3.55579033e-02 -6.26406491e-01 -1.02709293e+00
-5.80357075e-01 -5.36663473e-01 8.43550324e-01 2.65128911e-01
-8.60829353e-01 -2.23018438e-01 7.53689528e-01 7.15161860e-01
3.94296080e-01 1.25544205e-01 1.04248273e+00 -1.07861567e+00
-3.37975621e-01 -2.47343943e-01 1.01279147e-01 -3.06383427e-02
-2.45392267e-02 3.27353925e-02 -1.21936417e+00 -7.94975013e-02
-4.62805420e-01 -2.97934562e-01 8.87532890e-01 5.38712740e-01
1.13706434e+00 -4.24635679e-01 -2.55295724e-01 3.89630020e-01
1.34901845e+00 4.69695061e-01 3.71578038e-01 9.17642117e-01
2.29939759e-01 3.28109026e-01 9.49320138e-01 7.40289986e-01
5.40230349e-02 1.10684073e+00 1.53726488e-01 3.47861558e-01
-4.93783951e-02 -5.75019956e-01 3.73522401e-01 1.25827241e+00
-4.47489589e-01 2.75738239e-01 -8.31951976e-01 3.39149207e-01
-1.88161826e+00 -1.23766625e+00 -1.13425806e-01 2.55864048e+00
9.09135818e-01 1.57288104e-01 5.86104691e-01 9.52560544e-01
5.26459932e-01 -8.52485597e-02 -2.09698617e-01 -3.11999768e-01
-7.82692432e-02 2.72386581e-01 1.84238672e-01 -3.15124989e-02
-1.15499091e+00 7.46646583e-01 6.75415134e+00 1.27937114e+00
-9.63282347e-01 -6.67552277e-02 2.49572366e-01 -5.77852309e-01
-2.77641956e-02 6.36016726e-02 -8.19836795e-01 4.17010784e-01
1.08532596e+00 -5.30359223e-02 6.71117723e-01 8.78369093e-01
2.78083712e-01 8.54600407e-03 -9.12610769e-01 1.03535891e+00
2.37005204e-01 -1.36495268e+00 1.54450119e-01 2.07249790e-01
6.04081511e-01 1.19361714e-01 -7.92494393e-04 4.12009478e-01
-1.96467280e-01 -9.23230946e-01 7.28013694e-01 9.99773562e-01
7.32858717e-01 -8.95419657e-01 6.16873801e-01 4.34532017e-01
-1.08303452e+00 -1.76319957e-01 -9.35990140e-02 -6.44317508e-01
-9.31566730e-02 5.91884732e-01 -1.22535443e+00 3.98323774e-01
1.32972956e+00 9.09810185e-01 -7.17137277e-01 1.74719846e+00
3.30859423e-01 1.28315759e+00 -3.22629184e-01 -1.89337373e-01
-2.48313189e-01 2.54834592e-01 6.28963828e-01 1.30365014e+00
4.86482531e-01 -3.12529027e-01 6.60900921e-02 2.71039039e-01
3.10484499e-01 8.08132768e-01 -2.72353381e-01 4.42785509e-02
5.79413533e-01 1.02503538e+00 -5.51788151e-01 1.38155952e-01
-2.13853747e-01 3.84766638e-01 -1.93145275e-01 -2.29952350e-01
-4.71875489e-01 -3.82757068e-01 4.35963839e-01 5.83800733e-01
1.01718463e-01 1.85893804e-01 -3.33241522e-01 -7.83382475e-01
-2.81009912e-01 -9.41906869e-01 7.36502051e-01 -9.96523559e-01
-1.25738502e+00 4.56110179e-01 -1.86481863e-01 -1.65204263e+00
-4.20991153e-01 -1.64901257e-01 -4.60732728e-01 3.98495764e-01
-1.09808087e+00 -1.09576297e+00 7.97955468e-02 4.93584633e-01
5.23613453e-01 -7.28646278e-01 1.17865539e+00 6.85888469e-01
-5.99643402e-02 6.06968164e-01 4.99188751e-01 -1.23895206e-01
1.04908013e+00 -1.19829309e+00 -4.05827552e-01 -2.17770338e-01
1.07680357e+00 3.47326815e-01 4.86354679e-01 -5.50308764e-01
-1.00377738e+00 -5.99788249e-01 1.20306671e+00 -3.16382825e-01
7.24216282e-01 1.87338427e-01 -4.74119276e-01 2.44401336e-01
-1.05485670e-01 -9.89863634e-01 1.73015785e+00 7.77240992e-01
-3.68690014e-01 -4.39234585e-01 -6.43157482e-01 1.50485083e-01
6.64008260e-01 -4.20009017e-01 -5.79005301e-01 1.37222156e-01
4.14811611e-01 1.89000115e-01 -1.47259998e+00 6.94190919e-01
1.41502726e+00 -9.40712452e-01 1.15484226e+00 -6.21069849e-01
3.96172374e-01 -3.64147693e-01 -6.53987408e-01 -7.89267898e-01
-6.43368185e-01 -3.51151884e-01 -1.04115158e-01 1.48542726e+00
7.19372451e-01 1.11405075e-01 7.31162310e-01 -4.49821055e-02
1.86127305e-01 -7.61374235e-01 -1.04890525e+00 -6.96232080e-01
-2.24690512e-01 -9.52067673e-01 7.36746848e-01 9.57347631e-01
3.66029710e-01 5.07592738e-01 -9.05616283e-01 -2.81099290e-01
2.91123033e-01 6.19302392e-01 8.69854927e-01 -1.84632051e+00
-8.78476560e-01 -6.10359907e-01 -6.90507352e-01 -5.91039658e-01
-4.08215225e-01 -1.05690420e+00 -4.52852577e-01 -1.12999845e+00
8.17792475e-01 -6.72205806e-01 -1.11893380e+00 4.63596374e-01
2.46276885e-01 8.95546675e-01 3.19113076e-01 9.81495976e-01
-7.08608449e-01 4.22937889e-03 6.58218384e-01 -1.78411752e-01
-4.67166573e-01 8.12212467e-01 -5.69540083e-01 8.54389548e-01
8.92659664e-01 -8.19840670e-01 1.70444012e-01 1.35714367e-01
6.48760915e-01 1.09220147e-01 1.27028421e-01 -1.25451040e+00
3.56727451e-01 -2.40645334e-01 5.73766232e-01 -9.56701994e-01
4.56560731e-01 -4.10972595e-01 6.28661990e-01 4.22948748e-01
-6.43618226e-01 -2.86013901e-01 -8.51487592e-02 4.77576375e-01
-5.16703129e-01 -3.84558141e-01 2.16651574e-01 1.71844363e-01
-4.18855965e-01 -8.82435888e-02 -3.56694072e-01 -3.30294311e-01
6.43564820e-01 -3.33131135e-01 2.02469140e-01 -5.13940513e-01
-1.01645327e+00 -4.18592036e-01 1.15714110e-01 6.83458328e-01
3.55040818e-01 -1.41045713e+00 -7.36200094e-01 5.42321661e-03
2.86931962e-01 -9.26891565e-01 3.38092856e-02 6.89924121e-01
-2.57587880e-01 6.63892210e-01 -1.64767504e-01 -6.53638065e-01
-1.98332119e+00 2.49336198e-01 -3.10052514e-01 -2.84021497e-01
-3.95456433e-01 8.56919289e-01 -4.68055964e-01 -2.19255611e-01
7.53956735e-01 7.85160959e-02 -6.91769481e-01 2.23537207e-01
5.81866264e-01 7.91485131e-01 -5.11845648e-02 -6.91560209e-01
-2.50332981e-01 9.55703318e-01 -1.49554098e-02 -2.17706352e-01
1.43224669e+00 1.98719054e-01 -1.24278270e-01 1.09923077e+00
6.50608420e-01 3.75084370e-01 -2.14857399e-01 -3.06437194e-01
3.84035349e-01 -6.77292585e-01 2.14537218e-01 -1.36931348e+00
-5.53136349e-01 6.28589988e-01 7.56890416e-01 4.85072434e-01
1.21409345e+00 9.44613814e-02 6.60459518e-01 1.54407099e-01
6.48185551e-01 -1.05174553e+00 -2.07032606e-01 1.30330637e-01
7.61809587e-01 -7.74475276e-01 1.84336260e-01 -1.01894736e-01
-4.18746233e-01 1.02757406e+00 2.16649517e-01 7.44903535e-02
1.00557220e+00 1.00755692e-02 -7.34266788e-02 2.44109482e-01
-9.14884150e-01 2.93171629e-02 7.30540037e-01 2.11045548e-01
7.61037230e-01 4.01377350e-01 -6.47046804e-01 1.36846757e+00
-9.11004722e-01 1.34389311e-01 4.65626307e-02 5.04536331e-01
-8.10172141e-01 -1.15008140e+00 -4.94977266e-01 1.01384854e+00
-9.90384221e-01 -2.04458728e-01 -7.81877637e-01 4.68147516e-01
5.18615603e-01 9.87460494e-01 -2.21681893e-01 -1.02295732e+00
1.11364380e-01 2.29228452e-01 2.54505217e-01 -5.49604475e-01
-1.14645886e+00 5.87749958e-01 1.04354043e-02 -1.95469111e-01
-3.57969701e-01 -8.20288658e-01 -7.30361640e-01 -3.31798464e-01
-7.16722727e-01 6.73281491e-01 9.24264729e-01 6.06842220e-01
2.37217456e-01 2.93719262e-01 8.65404248e-01 -6.72569215e-01
-3.09952378e-01 -1.30278087e+00 -9.96643424e-01 3.41854766e-02
-1.74222022e-01 -4.80963498e-01 -3.09131384e-01 1.26004532e-01] | [15.929658889770508, 5.187254428863525] |
09b0061e-965e-4d67-ae27-26efa2baab9a | sample-based-uncertainty-quantification-with | 2209.08418 | null | https://arxiv.org/abs/2209.08418v2 | https://arxiv.org/pdf/2209.08418v2.pdf | Sample-based Uncertainty Quantification with a Single Deterministic Neural Network | Development of an accurate, flexible, and numerically efficient uncertainty quantification (UQ) method is one of fundamental challenges in machine learning. Previously, a UQ method called DISCO Nets has been proposed (Bouchacourt et al., 2016), which trains a neural network by minimizing the energy score. In this method, a random noise vector in $\mathbb{R}^{10\text{--}100}$ is concatenated with the original input vector in order to produce a diverse ensemble forecast despite using a single neural network. While this method has shown promising performance on a hand pose estimation task in computer vision, it remained unexplored whether this method works as nicely for regression on tabular data, and how it competes with more recent advanced UQ methods such as NGBoost. In this paper, we propose an improved neural architecture of DISCO Nets that admits faster and more stable training while only using a compact noise vector of dimension $\sim \mathcal{O}(1)$. We benchmark this approach on miscellaneous real-world tabular datasets and confirm that it is competitive with or even superior to standard UQ baselines. Moreover we observe that it exhibits better point forecast performance than a neural network of the same size trained with the conventional mean squared error. As another advantage of the proposed method, we show that local feature importance computation methods such as SHAP can be easily applied to any subregion of the predictive distribution. A new elementary proof for the validity of using the energy score to learn predictive distributions is also provided. | ['Chetan Gupta', 'Takuya Kanazawa'] | 2022-09-17 | null | null | null | null | ['miscellaneous'] | ['miscellaneous'] | [ 5.57019003e-02 1.21119745e-01 -3.33511233e-02 -3.37802380e-01
-9.22995269e-01 -6.01352632e-01 4.74340200e-01 1.44033328e-01
-3.54637206e-01 1.07431710e+00 -1.90713242e-01 -3.01401436e-01
-5.63042760e-01 -7.73802876e-01 -9.90639150e-01 -9.93146837e-01
5.66014014e-02 8.77697289e-01 -5.59041984e-02 -1.58230484e-01
1.24395497e-01 6.68750763e-01 -1.60326052e+00 -1.54792443e-01
8.79261732e-01 1.92018819e+00 -8.63602236e-02 2.06576467e-01
1.16092242e-01 6.74800813e-01 -5.68509877e-01 -6.44674957e-01
4.40648854e-01 -4.91366237e-02 -6.45401120e-01 -3.52793008e-01
8.15774500e-01 -5.92307150e-02 -1.50746927e-01 1.00223267e+00
7.55878091e-01 5.94398260e-01 1.08769953e+00 -8.84100199e-01
-3.97561222e-01 6.48312330e-01 -2.62626976e-01 -6.39251247e-02
-6.86065778e-02 -1.79729089e-01 1.15841579e+00 -9.79808569e-01
5.69363236e-01 9.62234497e-01 8.89621735e-01 3.25440675e-01
-1.34109735e+00 -7.12881625e-01 -1.08255163e-01 1.95186406e-01
-1.33608556e+00 -5.03780395e-02 7.92990863e-01 -4.40550059e-01
7.68344343e-01 3.37338328e-01 3.54429871e-01 1.09946871e+00
4.71401900e-01 7.98013985e-01 1.07042265e+00 -3.75459671e-01
4.29717064e-01 1.27538472e-01 6.56896308e-02 7.26802349e-01
4.40111682e-02 2.19461262e-01 -4.97156471e-01 -2.00177923e-01
5.53541362e-01 -1.86345950e-01 -2.96658397e-01 -6.03075981e-01
-9.55925345e-01 1.03732920e+00 6.58685207e-01 3.36194515e-01
-3.15267801e-01 4.26795065e-01 4.13992167e-01 4.41034660e-02
7.63965130e-01 6.73853517e-01 -6.21118367e-01 -1.17089979e-01
-1.13117242e+00 4.40831333e-01 6.93279803e-01 6.35755479e-01
5.13076663e-01 2.86336094e-01 -3.84403884e-01 9.41743672e-01
-1.06188864e-01 4.81716841e-01 1.98961571e-01 -1.16772914e+00
3.73656362e-01 4.80190396e-01 3.97886857e-02 -1.06059241e+00
-5.76041341e-01 -8.54335070e-01 -1.24409056e+00 4.73328114e-01
6.20402992e-01 -2.28910744e-01 -6.97332084e-01 1.67300439e+00
1.78424269e-01 -2.85243481e-01 -2.49094620e-01 9.11221921e-01
5.86274862e-01 5.45774043e-01 -3.34243029e-01 -2.01869130e-01
9.44276214e-01 -7.09728777e-01 -4.01418179e-01 2.16864347e-01
3.21577847e-01 -5.43810904e-01 6.38427615e-01 8.43274415e-01
-9.85627174e-01 -5.61283469e-01 -9.49746668e-01 1.32639423e-01
-4.19638216e-01 3.34903032e-01 7.77904093e-01 7.21873820e-01
-8.98750901e-01 1.22984338e+00 -5.41145325e-01 8.35230127e-02
5.13363779e-01 5.41563690e-01 -2.43465975e-01 1.95827052e-01
-1.22383928e+00 9.29483652e-01 6.82350755e-01 3.29756558e-01
-6.97313190e-01 -5.67087471e-01 -6.64305508e-01 1.21918105e-01
6.98008120e-01 -5.03360450e-01 1.12480247e+00 -7.61358500e-01
-1.73775923e+00 1.96010098e-01 1.51067361e-01 -6.44689798e-01
6.70540512e-01 -3.09042901e-01 5.36589287e-02 -2.11851031e-01
-2.19412670e-01 6.35864317e-01 9.75929976e-01 -1.06375217e+00
-4.42742854e-01 -6.20212913e-01 -1.37433305e-01 -2.31424913e-01
-1.43655226e-01 -2.51978517e-01 -1.90322354e-01 -8.53587866e-01
2.23199263e-01 -1.07936394e+00 -2.50301212e-01 -1.83152318e-01
-4.19241637e-01 -5.28855801e-01 2.66314417e-01 -6.22999728e-01
1.13557506e+00 -1.63414955e+00 3.72811794e-01 6.44022763e-01
1.31763652e-01 2.00721458e-01 2.06070393e-01 2.17593983e-01
-1.84725165e-01 -5.63814156e-02 -5.17632604e-01 -2.46193707e-01
2.21869946e-01 4.97576371e-02 -3.26372474e-01 5.21687865e-01
-2.89740358e-02 9.90337372e-01 -5.77571988e-01 -2.47193053e-01
1.55647710e-01 5.45698583e-01 -4.11146402e-01 -1.55083343e-01
-4.95232195e-01 3.29906136e-01 -1.91902459e-01 6.37601852e-01
7.15905547e-01 -1.98520243e-01 3.87788638e-02 -3.00599098e-01
-5.78029715e-02 -2.01922078e-02 -1.44900346e+00 1.45293880e+00
-3.20514381e-01 5.80654740e-01 -1.79693118e-01 -1.19099402e+00
1.07802999e+00 2.12271884e-01 5.64607918e-01 -5.32160997e-01
4.73022848e-01 5.44023097e-01 -1.29300728e-01 8.55997428e-02
4.62767571e-01 -2.17919067e-01 -3.81331705e-02 2.49215826e-01
3.50277573e-01 -1.40485376e-01 1.04390636e-01 -2.67250359e-01
8.05667877e-01 4.10195947e-01 2.12983981e-01 -3.48648369e-01
6.08254492e-01 -1.81689501e-01 3.99792969e-01 9.66470480e-01
-7.44779110e-02 6.98918462e-01 5.74560285e-01 -5.78805268e-01
-1.09268606e+00 -9.38834071e-01 -4.95655000e-01 1.10877299e+00
-2.60744900e-01 -3.22658688e-01 -7.07173109e-01 -8.35887909e-01
2.16698304e-01 6.95181072e-01 -7.79486418e-01 8.62323940e-02
-3.86996359e-01 -7.29686737e-01 4.83622402e-01 8.04748833e-01
3.85262161e-01 -1.03070176e+00 -3.79851729e-01 2.45782763e-01
-5.01592010e-02 -6.51114106e-01 -1.35577902e-01 6.13750696e-01
-8.13890994e-01 -7.42421210e-01 -8.20818007e-01 -1.53897092e-01
2.66743720e-01 -5.38131833e-01 1.19835627e+00 -3.90969664e-01
7.92491511e-02 1.03840873e-01 -1.71079114e-01 -5.65025449e-01
-1.19327188e-01 3.25177670e-01 2.76373953e-01 -3.26258987e-02
1.16300829e-01 -4.52059209e-01 -4.53169793e-01 3.93300086e-01
-8.04698288e-01 -4.04835969e-01 5.04982293e-01 1.25837457e+00
7.67811716e-01 1.63687021e-01 5.24039209e-01 -6.31730855e-01
6.12934291e-01 -2.54024476e-01 -9.65510130e-01 2.14109689e-01
-9.07513380e-01 4.41755861e-01 8.26064587e-01 -1.31383389e-01
-8.96183014e-01 1.43484488e-01 -3.93875539e-01 -6.00563586e-01
3.15042995e-02 6.02115929e-01 1.10043839e-01 -1.75693661e-01
7.46075690e-01 1.06244013e-01 9.72841159e-02 -5.36334395e-01
2.92660862e-01 4.96609896e-01 4.73357975e-01 -7.48167515e-01
7.20328450e-01 3.13756615e-01 4.58984882e-01 -4.10261542e-01
-9.64290082e-01 -2.72627592e-01 -6.48450077e-01 -2.35305220e-01
6.86297655e-01 -4.66412008e-01 -1.15474057e+00 3.60065758e-01
-1.16203976e+00 -1.21198438e-01 -3.54932219e-01 5.27604401e-01
-7.44543195e-01 2.24351898e-01 -3.69929045e-01 -1.21175873e+00
-4.75663841e-01 -1.04763520e+00 1.07034123e+00 -1.14155198e-02
1.03880120e-02 -8.64209116e-01 2.10051499e-02 3.57591718e-01
4.33039755e-01 2.93993354e-01 7.98849165e-01 -7.26205170e-01
-4.87454712e-01 -2.74476886e-01 -2.66148239e-01 6.45552278e-01
-2.07672834e-01 -5.98304756e-02 -9.51017559e-01 -2.20739096e-01
1.68342087e-02 -5.36504030e-01 1.23308623e+00 6.23784840e-01
1.58752668e+00 -1.77912578e-01 -1.06552601e-01 5.32124341e-01
1.39331293e+00 2.24220186e-01 4.94546831e-01 3.37138921e-01
5.56128442e-01 6.15809500e-01 5.39095819e-01 3.63989115e-01
4.38806824e-02 8.25905085e-01 5.79071224e-01 2.82513171e-01
3.32655311e-01 3.41714285e-02 1.00027561e-01 5.27333438e-01
-6.37664914e-01 -2.36749783e-01 -9.17892575e-01 2.24528193e-01
-1.98912859e+00 -8.92276466e-01 1.56869590e-02 2.40220261e+00
5.97628653e-01 2.52007395e-01 5.06049320e-02 3.21084172e-01
4.26175594e-01 1.05946958e-01 -4.95036453e-01 -4.01201665e-01
-2.06854358e-01 6.26773834e-01 6.18325174e-01 3.06587368e-01
-1.29907727e+00 5.76637328e-01 5.61020279e+00 1.46508670e+00
-1.14903212e+00 1.39944792e-01 8.58264208e-01 -1.48815915e-01
-1.39477327e-01 -3.71908754e-01 -7.92031467e-01 4.23278332e-01
8.31540883e-01 4.57868785e-01 4.29855645e-01 1.09871531e+00
-1.69089139e-01 -1.76726744e-01 -9.41130757e-01 1.03814816e+00
2.45038271e-01 -1.44854510e+00 -1.60335511e-01 6.09012283e-02
8.06199551e-01 6.25679567e-02 2.53368676e-01 4.07198131e-01
1.80878639e-01 -1.46411991e+00 7.99561024e-01 5.76622903e-01
7.16756940e-01 -1.01231730e+00 1.14952481e+00 4.27972347e-01
-1.01118505e+00 -2.66223639e-01 -5.69609284e-01 8.12270865e-02
-1.70183424e-02 8.65052581e-01 -6.64755404e-01 9.89729822e-01
9.03242230e-01 3.85958791e-01 -4.75159317e-01 8.76249433e-01
-2.49557607e-02 5.33672094e-01 -5.45023441e-01 -1.85455889e-01
3.41974288e-01 -2.80528098e-01 6.74871147e-01 8.00551116e-01
4.96653497e-01 -2.37226576e-01 -1.76078767e-01 1.00444090e+00
-1.39343739e-01 2.43822396e-01 -5.95782399e-01 1.26759931e-01
4.22571562e-02 1.11473548e+00 -6.89946830e-01 -6.04398325e-02
9.43176169e-03 6.20409667e-01 4.91880715e-01 9.99307334e-02
-6.88038588e-01 -2.61392117e-01 1.59163207e-01 -1.58089668e-01
5.43274939e-01 -1.10083528e-01 -4.39199090e-01 -9.88290548e-01
2.62895584e-01 -6.98621690e-01 3.46304089e-01 -5.57294488e-01
-1.55437207e+00 8.88747215e-01 3.75716649e-02 -1.29945958e+00
-5.21950662e-01 -1.15767825e+00 -1.01687558e-01 9.02135253e-01
-1.09537649e+00 -1.02221608e+00 -4.41784337e-02 4.26345736e-01
3.09563249e-01 -4.76795375e-01 7.56991148e-01 4.40193750e-02
-5.83735943e-01 6.12447023e-01 6.71757996e-01 -6.09529465e-02
5.10793090e-01 -1.54356050e+00 3.11467480e-02 4.03167814e-01
2.17097744e-01 3.86013895e-01 8.27043116e-01 -5.97943842e-01
-1.08570278e+00 -9.80583131e-01 6.45919919e-01 -7.19053745e-01
5.39573193e-01 -3.28610867e-01 -7.81711340e-01 3.99915338e-01
9.85533744e-03 1.26218542e-01 3.11614662e-01 2.38543972e-01
-2.62245804e-01 -3.36697489e-01 -1.05339956e+00 1.96896330e-01
7.89934754e-01 -2.49616534e-01 -3.89114678e-01 2.39939868e-01
3.07872355e-01 -5.84088624e-01 -1.12232852e+00 7.80177176e-01
8.54612172e-01 -1.20884788e+00 9.00293589e-01 -4.88509089e-01
6.01525009e-01 -8.51864275e-03 -4.59591687e-01 -1.42414939e+00
-6.88582584e-02 -4.05237705e-01 -3.18213314e-01 9.95880306e-01
4.23561424e-01 -8.11953127e-01 9.10969019e-01 3.59444976e-01
-6.17389195e-02 -1.33686018e+00 -1.33687520e+00 -1.03000867e+00
5.49743235e-01 -7.03982174e-01 4.47531462e-01 5.35017014e-01
-3.70406717e-01 1.96006164e-01 -4.91736650e-01 -2.56832033e-01
7.03650475e-01 7.00914934e-02 4.80562210e-01 -1.67780983e+00
-3.79208088e-01 -6.06277227e-01 -3.36632818e-01 -7.95772493e-01
3.84478927e-01 -8.83700192e-01 3.46694402e-02 -1.28364897e+00
-1.27632441e-02 -5.73721647e-01 -5.11962712e-01 2.67967999e-01
6.86794072e-02 3.32697004e-01 2.63912022e-01 1.39841124e-01
-3.50560904e-01 7.78559029e-01 1.08070469e+00 -2.98731178e-01
-4.15720977e-02 2.69047141e-01 -2.29033142e-01 7.22070754e-01
6.44775987e-01 -3.94884050e-01 -1.95254788e-01 8.73463154e-02
4.74638551e-01 2.34508306e-01 3.49423379e-01 -1.05409551e+00
1.22389942e-01 1.22270539e-01 6.29715919e-01 -8.30565631e-01
4.57762063e-01 -9.98232841e-01 1.82320133e-01 1.20391183e-01
-1.31813824e-01 6.90137735e-03 2.07049951e-01 5.52401543e-01
-2.71674395e-01 -5.60634851e-01 6.60628974e-01 -5.75333647e-02
-3.35851431e-01 2.41715997e-01 6.28467510e-03 -3.11498553e-01
8.36200774e-01 -2.19132885e-01 -1.98833674e-01 -3.80873084e-01
-6.03520155e-01 -1.82153713e-02 8.04272294e-02 1.06185287e-01
1.95293024e-01 -1.49735463e+00 -5.83229005e-01 -1.23956621e-01
-3.25563811e-02 5.77610023e-02 3.15092146e-01 9.74450052e-01
-3.22470665e-01 8.74565005e-01 -2.27526203e-01 -8.46953034e-01
-8.04604292e-01 3.63852322e-01 5.47219932e-01 -5.29937863e-01
-2.58373082e-01 1.04021084e+00 -1.79078251e-01 -7.32610464e-01
4.03483510e-01 -2.67154157e-01 -1.28773734e-01 2.93713927e-01
9.96887162e-02 6.77410364e-01 4.86433893e-01 -5.89689076e-01
-3.98838669e-01 5.07565439e-01 2.02652097e-01 -4.08372469e-02
1.45768881e+00 2.66328037e-01 -1.33377686e-01 5.79914093e-01
1.03571498e+00 -2.74443929e-03 -1.22983897e+00 -7.77744278e-02
-1.83451213e-02 -2.00798556e-01 2.28781737e-02 -1.03362095e+00
-1.20400333e+00 7.93087006e-01 7.79220104e-01 2.53259689e-01
1.11614096e+00 -6.63978830e-02 4.77188051e-01 5.86831629e-01
5.24003685e-01 -1.32308233e+00 1.45176938e-03 6.89768612e-01
1.14958262e+00 -1.37022221e+00 3.95633429e-02 -5.05234860e-02
-4.87801582e-01 1.22287989e+00 4.76820767e-01 -3.40706259e-02
6.21425152e-01 -5.82588091e-03 -2.33275294e-01 -1.48729980e-01
-4.81091052e-01 -2.52902508e-01 7.57307529e-01 2.67450094e-01
4.21025902e-01 9.60673839e-02 -3.71695876e-01 7.14368761e-01
-4.68411595e-01 4.42410447e-02 9.46157947e-02 5.96604347e-01
-1.65074512e-01 -9.67682898e-01 -5.36364198e-01 7.92894423e-01
-4.87055868e-01 -6.09784387e-03 -1.59583807e-01 8.20739567e-01
3.37933511e-01 5.91073096e-01 -1.07182346e-01 -5.21997809e-01
2.42688626e-01 4.26849663e-01 5.32078385e-01 -1.64472118e-01
-6.50475442e-01 -4.83934246e-02 -3.01805492e-02 -5.70375025e-01
-3.34795654e-01 -6.41110778e-01 -7.66900122e-01 -2.76559711e-01
-5.06444931e-01 1.84255391e-01 8.67438376e-01 1.07202554e+00
5.01676835e-02 5.15979290e-01 3.28008175e-01 -1.08392406e+00
-8.83168638e-01 -1.12400246e+00 -7.56294370e-01 6.31999373e-02
2.13821262e-01 -1.22972250e+00 -4.08668280e-01 -4.88893867e-01] | [7.848121643066406, 3.8958232402801514] |
8d2df339-1be6-4090-8037-923387146923 | categorical-feature-compression-via | 1904.13389 | null | http://arxiv.org/abs/1904.13389v1 | http://arxiv.org/pdf/1904.13389v1.pdf | Categorical Feature Compression via Submodular Optimization | In the era of big data, learning from categorical features with very large
vocabularies (e.g., 28 million for the Criteo click prediction dataset) has
become a practical challenge for machine learning researchers and
practitioners. We design a highly-scalable vocabulary compression algorithm
that seeks to maximize the mutual information between the compressed
categorical feature and the target binary labels and we furthermore show that
its solution is guaranteed to be within a $1-1/e \approx 63\%$ factor of the
global optimal solution. To achieve this, we introduce a novel
re-parametrization of the mutual information objective, which we prove is
submodular, and design a data structure to query the submodular function in
amortized $O(\log n )$ time (where $n$ is the input vocabulary size). Our
complete algorithm is shown to operate in $O(n \log n )$ time. Additionally, we
design a distributed implementation in which the query data structure is
decomposed across $O(k)$ machines such that each machine only requires $O(\frac
n k)$ space, while still preserving the approximation guarantee and using only
logarithmic rounds of computation. We also provide analysis of simple
alternative heuristic compression methods to demonstrate they cannot achieve
any approximation guarantee. Using the large-scale Criteo learning task, we
demonstrate better performance in retaining mutual information and also verify
competitive learning performance compared to other baseline methods. | ['Mohammadhossein Bateni', 'Afshin Rostamizadeh', 'Vahab S. Mirrokni', 'Hossein Esfandiari', 'Lin Chen', 'Thomas Fu'] | 2019-04-30 | null | null | null | null | ['feature-compression'] | ['computer-vision'] | [-7.15022348e-03 1.15252055e-01 -6.38414741e-01 -5.08449495e-01
-1.24857759e+00 -6.80544972e-01 -3.07336479e-01 5.53224027e-01
-7.54748404e-01 5.60914993e-01 -1.72614843e-01 -4.13192332e-01
-4.49051946e-01 -9.12132800e-01 -1.06550109e+00 -6.48104012e-01
-4.87816662e-01 9.53216851e-01 3.14907879e-02 6.67020380e-02
1.51387647e-01 7.81527311e-02 -1.82651019e+00 1.98729746e-02
5.12671530e-01 1.39854038e+00 -7.93033540e-02 8.45824182e-01
-9.84781533e-02 4.60093737e-01 -2.12439060e-01 -5.01166284e-01
7.06954718e-01 -3.30713764e-02 -9.19082999e-01 -9.56135467e-02
9.22519505e-01 -3.21416914e-01 -5.63550413e-01 1.14758396e+00
1.40691131e-01 -4.31109443e-02 4.79167730e-01 -1.21433485e+00
-4.21086013e-01 1.05582714e+00 -3.95365894e-01 1.25281662e-01
4.07658480e-02 -2.59317577e-01 1.80649149e+00 -4.60744202e-01
6.01077020e-01 1.03814900e+00 3.61490488e-01 2.87608504e-01
-1.50895679e+00 -1.03975892e+00 -1.07578887e-02 6.20347224e-02
-1.57319224e+00 -9.08276066e-02 3.27676684e-01 -1.12056278e-01
1.04248822e+00 5.66784680e-01 4.08521622e-01 1.36471584e-01
-8.58918279e-02 8.04855108e-01 8.41341794e-01 -3.04279745e-01
2.52568215e-01 1.67342976e-01 4.09127444e-01 1.02646613e+00
5.84189355e-01 -1.37051195e-01 -5.46206653e-01 -3.41173917e-01
1.96881220e-01 2.72656262e-01 -9.88216028e-02 -5.83966613e-01
-8.34091961e-01 1.19284153e+00 5.64647555e-01 1.89364925e-01
2.26636946e-01 7.75823891e-01 3.14470470e-01 9.80736732e-01
2.70254284e-01 2.86703169e-01 -8.38510990e-01 -1.13907889e-01
-7.81695068e-01 4.62211430e-01 1.20146060e+00 1.38821459e+00
9.84757066e-01 -4.18841422e-01 3.63429248e-01 6.69209659e-01
9.96278524e-02 8.92402709e-01 2.50506341e-01 -1.12127328e+00
9.01877105e-01 5.23945272e-01 6.74125506e-04 -9.59142685e-01
-3.80871028e-01 -1.64052919e-01 -7.84599423e-01 -3.95783246e-01
3.33952099e-01 1.94435403e-01 -5.97768307e-01 2.03253055e+00
3.57624829e-01 -5.04444122e-01 -2.92311370e-01 5.90727806e-01
3.10825467e-01 5.52962184e-01 -1.77356467e-01 -5.00500619e-01
1.41726577e+00 -6.94172025e-01 -4.73970532e-01 -7.27595165e-02
1.40528381e+00 -4.71283257e-01 1.07050204e+00 4.12740022e-01
-1.22980404e+00 -1.55254994e-02 -1.02179992e+00 -4.14285123e-01
-4.15632635e-01 -5.33687949e-01 1.02815473e+00 7.55660117e-01
-9.05067086e-01 3.15953791e-01 -7.64782846e-01 6.07303716e-02
4.44976509e-01 7.39854753e-01 -4.59324747e-01 -4.02119398e-01
-8.92566383e-01 2.65974820e-01 3.57164502e-01 -6.16038799e-01
-6.78349853e-01 -7.07412720e-01 -8.15626204e-01 2.75854319e-01
5.65994978e-01 -4.44098562e-01 1.11405313e+00 -3.65291655e-01
-5.81606627e-01 7.80308723e-01 -3.00248533e-01 -6.72657728e-01
1.45371869e-01 -2.28674170e-02 1.75668314e-01 3.00439835e-01
-1.62399113e-01 6.61720157e-01 4.55501825e-01 -6.75402045e-01
-1.04255819e+00 -7.61509776e-01 1.81251317e-01 1.33618951e-01
-9.54260707e-01 -2.04766855e-01 -5.11415422e-01 -4.49469566e-01
3.99955243e-01 -9.61746812e-01 -2.98307598e-01 3.18310618e-01
-7.18265679e-03 -6.06996596e-01 4.53484952e-01 -1.92815661e-01
1.51607096e+00 -2.16490316e+00 1.17320821e-01 5.95314741e-01
8.30214441e-01 -1.30986601e-01 -7.66855031e-02 3.06518912e-01
4.55208629e-01 4.05281544e-01 -3.40074390e-01 -5.01800179e-01
2.45290995e-01 4.11954820e-01 -3.51730794e-01 5.91243207e-01
-6.78436935e-01 8.88610423e-01 -4.84028071e-01 -5.82506537e-01
-3.12005609e-01 5.62169310e-03 -9.57243323e-01 -7.88643211e-02
-6.01203799e-01 -6.14518821e-01 -3.85885090e-01 6.04176819e-01
6.23937190e-01 -8.64259064e-01 4.43772048e-01 2.25323886e-01
4.98406202e-01 2.85040468e-01 -1.31378961e+00 1.52088022e+00
-5.12078941e-01 3.76183838e-01 2.87309110e-01 -1.14264810e+00
6.20085120e-01 1.77625250e-02 7.22171903e-01 -6.07372820e-01
1.70189932e-01 5.49351692e-01 -4.38389480e-01 -2.25143224e-01
3.42645526e-01 -2.90427029e-01 -4.85391259e-01 8.73280525e-01
-1.12679891e-01 -1.10973187e-01 4.44100089e-02 3.06947201e-01
1.46477854e+00 -9.26478446e-01 1.05809525e-01 -1.15440942e-01
2.48901118e-02 1.27641121e-02 3.22584927e-01 9.02862251e-01
-1.68726444e-01 2.10246854e-02 6.51604950e-01 -6.07831180e-01
-1.21207535e+00 -8.55673015e-01 -2.45517120e-01 1.50587118e+00
2.47161910e-01 -6.68619812e-01 -6.14262581e-01 -6.78707004e-01
4.92574215e-01 1.97804600e-01 -6.65778160e-01 3.23289186e-02
-4.67094749e-01 -5.17576456e-01 3.58779818e-01 5.62814534e-01
3.20612013e-01 -4.07966733e-01 -3.81402373e-01 -2.70254910e-02
2.82656886e-02 -1.04570496e+00 -8.93051326e-01 6.57259345e-01
-1.15657258e+00 -1.03815830e+00 -7.04112127e-02 -1.19180763e+00
5.40506601e-01 4.84397382e-01 1.02193475e+00 2.98432708e-01
-3.40208560e-01 2.74402767e-01 -1.55041665e-01 -1.51681855e-01
5.14619946e-02 3.59228373e-01 -5.65511025e-02 -5.38358450e-01
1.02938771e+00 -5.98395109e-01 -6.90403640e-01 2.27599263e-01
-1.17750406e+00 -2.35416040e-01 4.64207053e-01 8.61669898e-01
1.00814295e+00 1.31192565e-01 2.55687773e-01 -1.07969797e+00
2.27746174e-01 -5.29124916e-01 -1.27927506e+00 2.21403882e-01
-1.22443604e+00 3.65769535e-01 7.79365420e-01 -3.86096209e-01
7.49497488e-02 1.05673835e-01 1.36212751e-01 -5.48027277e-01
4.47756618e-01 5.34929097e-01 3.89330983e-02 -7.52196237e-02
5.49428284e-01 2.71470577e-01 2.54079163e-01 -4.73607719e-01
4.73123193e-01 9.06135440e-01 1.10651530e-01 -4.59903926e-01
8.96093369e-01 5.33345163e-01 1.76675692e-01 -5.67534149e-01
-8.91703427e-01 -7.65712917e-01 -1.81908116e-01 5.03277063e-01
3.64891917e-01 -9.87855136e-01 -1.44509816e+00 -1.01165205e-01
-4.88041580e-01 -3.87697488e-01 -3.51662606e-01 3.41471434e-01
-7.00474262e-01 2.40897685e-01 -5.36244392e-01 -6.82205856e-01
-5.10909617e-01 -8.98454666e-01 1.06542003e+00 -2.71960318e-01
2.22003996e-01 -7.84811676e-01 -1.99337322e-02 6.01906121e-01
3.42515141e-01 -2.07713515e-01 1.32967460e+00 -9.07190442e-01
-1.07641637e+00 -5.11639595e-01 -4.23500925e-01 2.46783078e-01
-3.05772185e-01 -8.20393205e-01 -5.10711372e-01 -6.25468671e-01
-1.02731675e-01 -8.02438140e-01 1.03976405e+00 1.06484905e-01
1.58654106e+00 -7.67475486e-01 -3.10433656e-01 9.85170722e-01
1.75441241e+00 -2.65001357e-01 2.11252004e-01 -9.66538414e-02
5.85531831e-01 2.82081902e-01 6.37161553e-01 7.14145482e-01
5.37464678e-01 4.77733821e-01 3.44239831e-01 4.21873361e-01
1.37741238e-01 -3.35726202e-01 3.21731903e-02 9.39225852e-01
4.46399301e-01 -1.20386221e-01 -6.90829098e-01 7.32976079e-01
-1.57548261e+00 -6.59955263e-01 2.83806443e-01 2.54259038e+00
1.27734661e+00 2.35407680e-01 1.47968695e-01 2.31781945e-01
2.00465485e-01 1.59230772e-02 -6.56555951e-01 -5.80900073e-01
-1.04636811e-01 6.08945847e-01 1.28587151e+00 7.68268347e-01
-1.04157913e+00 1.07207155e+00 6.32269907e+00 1.11714673e+00
-8.68882716e-01 1.56153247e-01 6.02015674e-01 -7.45453715e-01
-6.90909743e-01 -1.05997711e-01 -9.90407705e-01 2.83079863e-01
1.16977632e+00 -3.31303775e-01 1.00093865e+00 1.27077699e+00
-5.54064155e-01 3.22402790e-02 -1.34571612e+00 1.34064531e+00
5.17748576e-03 -1.31964469e+00 -9.39932689e-02 6.63121998e-01
7.93009818e-01 1.55708909e-01 2.21890375e-01 3.60371321e-01
5.58710933e-01 -1.29547215e+00 2.80568421e-01 -2.75589913e-01
1.20649230e+00 -9.32404101e-01 5.17363369e-01 4.96253848e-01
-1.41354072e+00 -6.17948055e-01 -8.26999485e-01 1.38172835e-01
-1.94584459e-01 4.73843575e-01 -7.17182875e-01 -1.34921774e-01
8.54774535e-01 2.91681826e-01 -4.38013375e-01 7.03976035e-01
3.84420246e-01 5.49865365e-01 -1.04262137e+00 -5.51658571e-01
1.19553886e-01 2.00559512e-01 1.66460425e-01 9.23676908e-01
1.38650119e-01 2.90841788e-01 2.57067710e-01 2.48806149e-01
-8.87651563e-01 4.44152415e-01 -7.88714767e-01 -2.27554172e-01
8.24758828e-01 8.37117970e-01 -3.54636639e-01 -3.31208378e-01
-4.31002766e-01 6.22888327e-01 7.10875869e-01 -9.30245966e-02
-7.32573092e-01 -6.68043256e-01 7.25353539e-01 1.04860850e-01
8.99218857e-01 -2.97127843e-01 -5.55363178e-01 -1.02193034e+00
4.14446741e-01 -9.14929807e-01 8.56080890e-01 4.96038608e-02
-1.19918251e+00 2.31165290e-01 2.99043618e-02 -6.89807475e-01
-1.93705654e-03 -6.58478260e-01 4.19825971e-01 4.96990830e-01
-1.35299647e+00 -7.77715564e-01 -9.93038639e-02 6.76070988e-01
1.27339527e-01 1.29599078e-02 8.44790816e-01 3.02424699e-01
3.81177217e-02 1.39702141e+00 5.57815731e-01 -1.06966406e-01
5.01032054e-01 -1.44489753e+00 3.01685277e-02 2.62804836e-01
2.91078269e-01 5.70453405e-01 6.21983886e-01 -3.83939296e-01
-1.94758809e+00 -9.96213138e-01 1.33689368e+00 -3.25985640e-01
7.09940612e-01 -7.11021364e-01 -6.39527798e-01 9.06002343e-01
-4.49653596e-01 3.39557648e-01 1.07682180e+00 3.77839178e-01
-1.08040786e+00 -6.50128365e-01 -1.31557679e+00 3.95446867e-01
1.33582628e+00 -6.82746649e-01 -2.17367172e-01 6.74099445e-01
1.26511431e+00 -2.63237864e-01 -1.26576042e+00 2.83014596e-01
7.40986168e-01 -4.39169675e-01 8.23616743e-01 -9.50613081e-01
1.12473704e-01 2.61609167e-01 -8.69614124e-01 -2.78376281e-01
-3.34626175e-02 -6.97970212e-01 -6.55777156e-01 4.77275491e-01
7.10874081e-01 -7.36399353e-01 1.14262891e+00 8.64434958e-01
5.46754479e-01 -1.13250196e+00 -1.28649712e+00 -7.45639563e-01
5.09440541e-01 -6.02267206e-01 5.94916523e-01 6.11690283e-01
2.27308676e-01 2.34644458e-01 -2.87782013e-01 3.15288641e-02
7.76700377e-01 5.19031942e-01 9.75078166e-01 -1.07995617e+00
-5.23679018e-01 -3.61540228e-01 -4.70433056e-01 -1.58398283e+00
-3.25955637e-02 -1.19979537e+00 -1.85199857e-01 -9.44042861e-01
6.87267363e-01 -9.97774422e-01 -3.75482291e-01 5.67107141e-01
8.90305117e-02 2.47875467e-01 1.56239927e-01 2.80989647e-01
-1.08114612e+00 3.75467718e-01 9.26321208e-01 -1.84837312e-01
5.69178462e-02 -1.47058412e-01 -9.70071971e-01 1.32989064e-01
5.08552849e-01 -8.81959677e-01 -3.58992875e-01 -5.41172087e-01
6.96743071e-01 1.56937912e-01 -1.06029384e-01 -7.59430826e-01
4.39526647e-01 -1.04783639e-01 -1.92313388e-01 -5.84180295e-01
1.64481848e-01 -1.03360248e+00 -3.57829452e-01 6.43502831e-01
-9.03239489e-01 1.84689105e-01 -1.74694419e-01 8.95584702e-01
-1.92242600e-02 -9.79649574e-02 8.04891706e-01 1.74997464e-01
1.12004213e-01 8.24339807e-01 1.25104129e-01 4.78721231e-01
1.09158647e+00 2.27710977e-01 -3.01844090e-01 -4.47823018e-01
-2.96394259e-01 6.20941997e-01 5.95367014e-01 -3.44948797e-03
5.23082435e-01 -1.19073629e+00 -4.47555393e-01 9.75102335e-02
2.86005080e-01 2.31058478e-01 1.36273965e-01 6.91151500e-01
-5.96804798e-01 8.93334627e-01 6.03653491e-01 -5.21062732e-01
-1.43807411e+00 1.03959620e+00 -1.74280629e-02 -5.82221210e-01
-4.17384863e-01 1.19055700e+00 -6.86170980e-02 -3.22240174e-01
9.41504478e-01 -4.23331499e-01 4.59139675e-01 -1.32330716e-01
6.92111492e-01 2.72236645e-01 -7.57182110e-03 -3.28365639e-02
-1.63545236e-01 4.30277258e-01 -4.07347292e-01 -2.10142344e-01
1.39872682e+00 -1.38016999e-01 -2.12273836e-01 2.39250839e-01
2.03964543e+00 -2.44494211e-02 -7.99469709e-01 -7.92445362e-01
1.76283550e-02 -5.68952620e-01 -9.81321335e-02 -3.22620034e-01
-1.13042223e+00 7.28568196e-01 7.06381083e-01 3.31488609e-01
1.16895759e+00 5.07399082e-01 1.33609712e+00 1.11891270e+00
7.86046624e-01 -1.13930261e+00 -1.69690736e-02 2.70420372e-01
3.34204704e-01 -1.22441101e+00 3.70855704e-02 -1.80871591e-01
-9.53076184e-02 8.55066359e-01 3.08762103e-01 -1.29318267e-01
1.13878536e+00 2.24551931e-01 -4.00565654e-01 -3.36731255e-01
-1.27346015e+00 -3.59139293e-02 7.14818295e-03 4.93137650e-02
6.70183226e-02 2.69681931e-01 -4.33814496e-01 6.06073976e-01
-5.78647017e-01 -9.97807980e-02 -8.14332962e-02 9.39895272e-01
-9.74642932e-01 -1.18743241e+00 2.13142447e-02 9.48855639e-01
-6.97260976e-01 -2.84166098e-01 -2.74108332e-02 5.69129348e-01
-2.63105601e-01 8.30757558e-01 1.99770436e-01 -5.48802137e-01
-2.19470754e-01 1.40792951e-01 3.91326129e-01 -4.49341238e-01
-3.52176905e-01 -3.37150656e-02 -3.33559573e-01 -7.33096063e-01
4.22199853e-02 -4.16476190e-01 -1.34188592e+00 -7.79693305e-01
-4.37470585e-01 2.78126270e-01 8.45606029e-01 7.19361067e-01
3.82535726e-01 -3.74901533e-01 9.35241163e-01 5.30709662e-02
-1.19820154e+00 -7.64888227e-01 -6.70063913e-01 3.18919420e-01
3.03750426e-01 -2.66065121e-01 -7.04853952e-01 -1.55923128e-01] | [6.780900955200195, 4.815514087677002] |
629d25f3-fd03-4b15-8306-89b279afec36 | deep-learning-for-background-replacement-in | null | null | https://www.sciltp.com/journals/ijndi/article/view/256 | https://www.sciltp.com/journals/ijndi/article/view/256/128 | Deep learning for Background Replacement in Video Conferencing | Background replacement is one of the most used features in video conferencing applications by many people, perhaps mainly for privacy protection, but also for other purposes such as branding, marketing and promoting professionalism. However, the existing applications in video conference tools have serious limitations. Most applications tend to generate strong artefacts (while there is a slight change in the perspective of the background), or require green screens to avoid such artefacts, which results in an unnatural background or even exposes the original background to other users in the video conference. In this work, we aim to study the relationship between the foreground and background in real-time videos. Three different methods are presented and evaluated, including the baseline U-Net, the lightweight U-Net MobileNet, and the U-Net MobileNet&ConvLSTM models. The above models are trained on public datasets for image segmentation. Experimental results show that both the lightweight U-Net MobileNet and the U-Net MobileNet& ConvLSTM models achieve superior performance as compared to the baseline U-Net model. | ['Yongmin Li', 'Kiran Shahi'] | 2023-06-12 | null | null | null | international-journal-of-network-dynamics-and | ['video-background-subtraction', 'marketing'] | ['computer-vision', 'miscellaneous'] | [ 5.14317393e-01 -5.55446520e-02 -2.80529279e-02 -1.07481763e-01
-2.59131312e-01 -2.50126183e-01 4.03720021e-01 -3.55447203e-01
-4.74305838e-01 7.74879932e-01 -1.08759247e-01 -4.86891568e-01
3.46174210e-01 -5.51924348e-01 -7.13298261e-01 -8.79649460e-01
2.11012706e-01 -2.14160204e-01 7.34980524e-01 6.49726167e-02
1.86156780e-01 1.69606462e-01 -1.48435390e+00 5.49952388e-01
8.72379839e-01 1.02373695e+00 4.20505911e-01 6.78924561e-01
-3.35440725e-01 9.76754069e-01 -8.34377110e-01 -5.32613397e-01
2.32403770e-01 -3.66423070e-01 -4.66837972e-01 2.73778856e-01
7.77083695e-01 -5.96175373e-01 -6.85297608e-01 1.50840712e+00
6.13868475e-01 3.60733062e-01 1.06480703e-01 -1.41307890e+00
-2.06672877e-01 5.02028465e-01 -8.54319155e-01 3.80561978e-01
3.89329582e-01 -2.00211294e-02 9.32508633e-02 -3.00970763e-01
5.88712513e-01 1.52805483e+00 7.32840121e-01 7.82916069e-01
-7.06244051e-01 -9.09239888e-01 6.17764175e-01 1.04007393e-01
-1.05757165e+00 -6.01487279e-01 5.40294588e-01 -2.77230203e-01
5.05365252e-01 5.27581990e-01 7.21150637e-01 1.35536349e+00
3.58645707e-01 1.14158106e+00 9.60174501e-01 -3.10306519e-01
-6.32427260e-02 3.70293438e-01 5.96942268e-02 5.89123607e-01
3.76823060e-02 -2.26270974e-01 -3.73900086e-01 -6.17398247e-02
9.95852053e-01 1.80517897e-01 -6.60945654e-01 -2.20712781e-01
-1.03385532e+00 2.48791710e-01 1.72480121e-01 4.75723743e-01
-2.22946778e-01 2.42833257e-01 4.43818718e-01 1.29257709e-01
6.77922487e-01 -2.19630554e-01 -3.64135236e-01 -4.51599956e-01
-1.01957309e+00 1.20052986e-01 5.24256647e-01 1.12138355e+00
4.84469801e-01 -4.34430502e-02 -3.04318249e-01 7.99798548e-01
2.23913401e-01 1.22120738e-01 4.20824438e-01 -9.20981705e-01
4.49260890e-01 3.82547945e-01 6.20881729e-02 -1.10386539e+00
3.09841204e-02 -7.95458704e-02 -7.37950623e-01 1.71765462e-01
3.97828072e-01 -4.92019683e-01 -1.14990568e+00 1.47893262e+00
3.05090338e-01 7.97728360e-01 -2.22483233e-01 6.60884440e-01
1.25144351e+00 7.86621869e-01 -1.27693545e-02 -2.11777300e-01
1.05492020e+00 -1.21189296e+00 -1.25825083e+00 -3.00717562e-01
4.88291651e-01 -9.91299093e-01 9.05241370e-01 4.80954349e-01
-1.12841094e+00 -5.43571830e-01 -8.61518204e-01 2.00841561e-01
-2.85350502e-01 5.40843904e-02 6.80898488e-01 1.12523973e+00
-1.24551928e+00 5.77508330e-01 -5.15902996e-01 -6.72404408e-01
4.14952815e-01 3.43048155e-01 -8.48563686e-02 -1.17712349e-01
-1.13940012e+00 7.01840699e-01 3.60438764e-01 5.41197002e-01
-7.02986538e-01 -4.14917201e-01 -8.96826386e-01 -1.13528296e-01
5.29237688e-01 -2.86681503e-01 1.32710850e+00 -1.46728957e+00
-1.46101534e+00 9.26659644e-01 -1.17784321e-01 -2.46502414e-01
1.22190523e+00 -4.18211669e-01 -4.79544312e-01 -1.66698899e-02
-2.78727226e-02 7.32414603e-01 7.37199843e-01 -1.24432564e+00
-1.03157961e+00 -1.10698737e-01 2.77337044e-01 1.10032462e-01
-1.37300849e-01 1.97530329e-01 -1.04900551e+00 -6.35064125e-01
-5.58576211e-02 -8.60131383e-01 -1.30235910e-01 8.47390220e-02
-3.92003626e-01 1.89532697e-01 1.62596273e+00 -8.99103582e-01
1.25256264e+00 -2.24997997e+00 -4.27785426e-01 4.42440696e-02
1.46632522e-01 6.72137678e-01 6.76064119e-02 -1.99692756e-01
1.66475490e-01 4.40574922e-02 1.70289073e-03 -5.18469177e-02
-3.55242908e-01 1.07256345e-01 1.60566270e-01 3.65111321e-01
-3.74791354e-01 6.90827727e-01 -9.67257679e-01 -7.29102492e-01
5.19947767e-01 5.20379543e-01 -3.11540008e-01 -6.70013428e-02
-5.53005980e-03 4.17806953e-01 -1.55113503e-01 7.04917312e-01
9.80251372e-01 3.16523612e-01 2.30740651e-01 -7.65495673e-02
-7.04069017e-03 -1.00509927e-01 -1.21949744e+00 1.61081612e+00
-1.69660911e-01 1.13665926e+00 5.39348722e-01 -6.13181710e-01
5.39733291e-01 5.10105968e-01 3.81611466e-01 -6.95154130e-01
3.53357762e-01 2.47565508e-01 -3.27137150e-02 -7.60102332e-01
7.87878096e-01 1.72718279e-02 4.48541075e-01 -4.56628054e-02
-2.62046248e-01 2.41216451e-01 1.84778869e-01 3.24082822e-01
8.66130769e-01 3.44690651e-01 -2.98827469e-01 -1.97842792e-01
4.55553353e-01 -6.41664445e-01 9.30020571e-01 8.24397445e-01
-6.00186050e-01 5.99450350e-01 5.38932025e-01 -3.65039766e-01
-4.08965588e-01 -6.41281009e-01 1.02055058e-01 9.61202264e-01
7.39928365e-01 2.08898503e-02 -1.27549386e+00 -7.79428661e-01
-4.06600922e-01 6.94645166e-01 -3.25895965e-01 -1.37131721e-01
-4.68157858e-01 -5.41246355e-01 5.19278228e-01 3.45998824e-01
1.11198783e+00 -1.28925073e+00 -4.96420413e-01 2.30254486e-01
-6.54156685e-01 -1.32516646e+00 -6.87717497e-01 -2.28045851e-01
-8.06946754e-01 -9.04860854e-01 -9.72129047e-01 -9.40573692e-01
6.00410163e-01 5.88910639e-01 8.10202539e-01 3.02430809e-01
-4.77533266e-02 2.93305099e-01 -1.42350346e-01 -5.90267599e-01
-2.07502723e-01 -2.02411443e-01 -1.20238364e-01 3.41763377e-01
5.23135364e-01 -1.59686685e-01 -5.35377145e-01 5.19035220e-01
-1.03658044e+00 1.91020504e-01 2.19195992e-01 2.69280046e-01
3.66846658e-02 3.27271909e-01 2.85334885e-01 -1.04298615e+00
6.12410843e-01 -2.32938100e-02 -3.29100221e-01 3.59460801e-01
6.36442453e-02 -6.96290731e-01 -1.06368482e-01 -7.34636128e-01
-1.33236527e+00 -3.90292436e-01 -1.05688557e-01 -5.86189568e-01
-1.93096548e-01 -1.00386306e-03 -7.67750978e-01 -3.10482562e-01
1.13746382e-01 -1.04970329e-01 -2.63891160e-01 -2.72025287e-01
-7.36475736e-02 9.36309397e-01 4.30385828e-01 -4.02789384e-01
3.10003072e-01 5.25931597e-01 -5.28499305e-01 -1.31308615e+00
-5.74888945e-01 -4.44254220e-01 -4.14615870e-01 -6.95169210e-01
9.43863332e-01 -6.85322702e-01 -3.97253096e-01 9.42664444e-01
-1.42099953e+00 -4.59739625e-01 1.64990678e-01 3.96351963e-01
-1.78234920e-01 5.71675718e-01 -7.54103065e-01 -9.79572237e-01
5.95944449e-02 -1.38657439e+00 8.66557539e-01 6.39834821e-01
-3.96747217e-02 -1.08418071e+00 -5.23946404e-01 4.58862513e-01
1.93303511e-01 4.35284525e-01 6.82406425e-01 -2.24879131e-01
-7.42057860e-01 -2.53131837e-01 -3.31935853e-01 5.04592836e-01
4.04412806e-01 3.31808180e-01 -1.37000084e+00 -1.27066225e-01
-4.95749675e-02 3.22718978e-01 1.00158262e+00 8.79121184e-01
1.23697293e+00 -1.97887346e-01 -6.36547327e-01 6.66434586e-01
8.34271312e-01 9.13014948e-01 1.24708831e+00 5.56939721e-01
8.01274657e-01 6.03807509e-01 7.62303174e-01 -7.66499788e-02
-1.96528926e-01 3.01160306e-01 4.76011723e-01 -5.03393710e-01
1.38140917e-01 -1.08038396e-01 4.22053188e-01 4.76436138e-01
-2.89543450e-01 -4.59293962e-01 -6.09959543e-01 4.82006818e-01
-2.14618707e+00 -1.00886905e+00 -4.09238666e-01 2.20721078e+00
3.66397291e-01 4.13063914e-01 -8.33979100e-02 -2.52685733e-02
1.41647995e+00 2.28090987e-01 -1.93615332e-01 -5.13742149e-01
-1.88516863e-02 -1.38145059e-01 5.53657115e-01 2.89156049e-01
-1.34696746e+00 1.04431391e+00 5.86611891e+00 1.00228071e+00
-1.37838089e+00 1.33577451e-01 1.01105464e+00 -1.38409734e-01
-1.05655072e-02 -1.26023531e-01 -5.29519439e-01 5.98126650e-01
3.76163781e-01 2.11827174e-01 7.89752305e-02 7.47090697e-01
4.47962761e-01 -2.77329654e-01 -8.74376476e-01 1.31867695e+00
4.00749873e-03 -1.24800110e+00 -6.18023500e-02 3.03700175e-02
7.17012346e-01 -3.23737741e-01 -3.80166620e-03 3.93189222e-01
-1.10587612e-01 -7.46953249e-01 8.17964196e-01 2.42353603e-01
6.42594397e-01 -7.44801521e-01 9.21001256e-01 2.13542446e-01
-8.61092389e-01 2.88100630e-01 -3.58734995e-01 -5.42344823e-02
2.51193047e-01 3.67637455e-01 -3.70509773e-01 2.85342008e-01
1.11111176e+00 6.08360291e-01 -2.51299411e-01 1.20707703e+00
2.11436197e-01 3.11427742e-01 -6.95126206e-02 4.42153811e-02
4.75158900e-01 -4.44093108e-01 6.20875657e-01 1.47166467e+00
2.56849408e-01 -1.75646529e-01 1.46385357e-01 4.73655403e-01
-3.31117332e-01 -9.42054912e-02 -7.46633947e-01 -3.48716713e-02
-6.94663450e-02 1.35877573e+00 -1.08143127e+00 -5.18498182e-01
-5.54234385e-01 1.15822649e+00 -6.33396327e-01 7.38506556e-01
-1.04147935e+00 -5.44412792e-01 6.47673965e-01 3.96528095e-01
9.53280404e-02 1.64410487e-01 1.77130952e-01 -1.16994536e+00
6.54964522e-02 -1.03707111e+00 1.52320936e-01 -8.39805245e-01
-7.51312554e-01 5.55063665e-01 -1.79780141e-01 -1.04656529e+00
3.22870731e-01 -5.16275287e-01 -9.33370411e-01 5.93474805e-01
-1.31960523e+00 -1.00052094e+00 -6.57504201e-01 4.89608794e-01
7.85853148e-01 -7.98238739e-02 1.23551615e-01 7.84369946e-01
-7.64585674e-01 4.50736582e-01 5.46035096e-02 2.88412839e-01
7.88334489e-01 -8.68137658e-01 3.86233419e-01 1.06325912e+00
-1.61272258e-01 6.85547590e-01 7.01089144e-01 -8.84432256e-01
-8.11267197e-01 -1.10867941e+00 4.99802142e-01 -1.83441624e-01
2.84590989e-01 -4.58322555e-01 -8.46761823e-01 8.00785363e-01
4.44530338e-01 -1.31555453e-01 4.21923101e-01 -2.46864393e-01
4.46226448e-01 -8.33196715e-02 -1.46305943e+00 9.48341608e-01
1.01171827e+00 -2.46361658e-01 -1.25505298e-01 1.29320592e-01
5.79223156e-01 -6.08782351e-01 -2.70516038e-01 2.88816750e-01
8.11529338e-01 -1.34581578e+00 7.10432947e-01 -3.56021374e-01
3.32899719e-01 -1.70673266e-01 -1.80626456e-02 -8.30023885e-01
2.93218177e-02 -1.14424908e+00 8.81492421e-02 1.56343865e+00
1.59334272e-01 -5.46723604e-01 1.00807393e+00 9.53993559e-01
-2.50367913e-02 -4.46424723e-01 -9.25484478e-01 -4.75034565e-01
-5.61683655e-01 -4.72216159e-01 4.06814516e-01 1.07290196e+00
-6.07752800e-01 2.91834697e-02 -5.62370241e-01 -2.52329614e-02
4.89073217e-01 -5.02533257e-01 9.31166291e-01 -9.13992763e-01
1.06565930e-01 -3.88244927e-01 -5.88321865e-01 -1.25406229e+00
-2.45050043e-02 -1.97047129e-01 4.97223400e-02 -1.64637375e+00
1.60737053e-01 -2.96905994e-01 -5.02449095e-01 2.01705955e-02
-2.60505915e-01 2.32715145e-01 1.13262206e-01 -2.54489392e-01
-6.81020439e-01 2.22968444e-01 1.37146878e+00 -3.00328523e-01
-4.98966575e-01 2.88662404e-01 -5.89697480e-01 1.20446897e+00
5.86606264e-01 -1.94497809e-01 -3.69477421e-01 -4.71894294e-01
-1.68316320e-01 5.79932285e-03 4.63051319e-01 -8.18140090e-01
7.87817687e-02 -2.02774107e-01 4.76442605e-01 -7.19373882e-01
2.31385827e-01 -1.04063296e+00 -4.37827706e-02 4.15544212e-01
6.26203269e-02 -2.03872919e-01 3.87332618e-01 4.62076306e-01
-2.63246745e-01 -3.16850275e-01 7.95638025e-01 -4.86949861e-01
-9.99731183e-01 3.13661784e-01 -6.43459380e-01 -2.72626102e-01
1.04498053e+00 -8.68232250e-01 -2.82084912e-01 -7.66998231e-01
-5.78273356e-01 2.83533484e-01 4.06537414e-01 7.72383332e-01
5.58488190e-01 -1.18171227e+00 -2.08475143e-01 1.04209475e-01
-3.34932923e-01 1.52874067e-01 4.45299029e-01 1.10843670e+00
-9.49860632e-01 2.19542816e-01 -1.67923689e-01 -6.60892069e-01
-1.75506258e+00 2.48055756e-01 6.08878613e-01 1.20518677e-01
-7.31427073e-01 1.04286230e+00 8.20198298e-01 -3.93628120e-01
7.25248098e-01 -5.76857030e-01 -1.24014415e-01 -5.99595793e-02
6.86092854e-01 6.98752701e-01 -2.81835590e-02 -4.96061653e-01
-2.57999539e-01 1.81328252e-01 -2.61649936e-01 1.08324215e-01
8.74271154e-01 -2.71008521e-01 -2.19890565e-01 3.72093678e-01
8.84793282e-01 -5.43357171e-02 -1.39338923e+00 -5.75106069e-02
-1.62594467e-02 -9.39598501e-01 1.30641311e-01 -4.00405675e-01
-1.35653746e+00 9.99914289e-01 8.33189487e-01 2.18468770e-01
1.09563267e+00 -5.00803888e-01 1.03952241e+00 2.23604366e-01
4.14516687e-01 -1.36769807e+00 -2.04220396e-02 1.92651808e-01
4.95523691e-01 -1.18948293e+00 -2.08010063e-01 -5.94338655e-01
-7.66645551e-01 9.12667453e-01 9.67418730e-01 3.60023797e-01
4.26416785e-01 2.36395314e-01 7.09354699e-01 1.14328280e-01
-2.28287742e-01 5.85776605e-02 4.11213264e-02 7.46984422e-01
5.50639749e-01 -3.26593846e-01 -1.42590720e-02 3.22311908e-01
1.82291746e-01 1.27688702e-02 5.89824140e-01 1.18833697e+00
-4.16109771e-01 -8.68672371e-01 -5.98983526e-01 5.14537752e-01
-9.05925214e-01 7.04215020e-02 -5.94325662e-01 8.92332435e-01
4.37819600e-01 1.12183177e+00 4.95010242e-02 -2.24655509e-01
2.65398890e-01 -6.22007437e-02 5.64620256e-01 -2.87816137e-01
-6.76808655e-01 6.18573606e-01 2.21861899e-01 -4.76567656e-01
-6.46068633e-01 -4.97582018e-01 -8.93004954e-01 -5.16231120e-01
-6.32722914e-01 1.67477317e-02 5.95337451e-01 8.97914410e-01
-1.35815963e-01 8.80387843e-01 5.18727265e-02 -1.04259956e+00
3.51839811e-01 -1.02071810e+00 -7.51006842e-01 2.31843293e-01
2.64355838e-01 -6.44654095e-01 -2.15200797e-01 1.46807462e-01] | [9.197328567504883, -0.5853133201599121] |
2cffb199-510f-4db3-8e62-0ab40f609183 | r3det-refined-single-stage-detector-with | 1908.05612 | null | https://arxiv.org/abs/1908.05612v6 | https://arxiv.org/pdf/1908.05612v6.pdf | R3Det: Refined Single-Stage Detector with Feature Refinement for Rotating Object | Rotation detection is a challenging task due to the difficulties of locating the multi-angle objects and separating them effectively from the background. Though considerable progress has been made, for practical settings, there still exist challenges for rotating objects with large aspect ratio, dense distribution and category extremely imbalance. In this paper, we propose an end-to-end refined single-stage rotation detector for fast and accurate object detection by using a progressive regression approach from coarse to fine granularity. Considering the shortcoming of feature misalignment in existing refined single-stage detector, we design a feature refinement module to improve detection performance by getting more accurate features. The key idea of feature refinement module is to re-encode the position information of the current refined bounding box to the corresponding feature points through pixel-wise feature interpolation to realize feature reconstruction and alignment. For more accurate rotation estimation, an approximate SkewIoU loss is proposed to solve the problem that the calculation of SkewIoU is not derivable. Experiments on three popular remote sensing public datasets DOTA, HRSC2016, UCAS-AOD as well as one scene text dataset ICDAR2015 show the effectiveness of our approach. Tensorflow and Pytorch version codes are available at https://github.com/Thinklab-SJTU/R3Det_Tensorflow and https://github.com/SJTU-Thinklab-Det/r3det-on-mmdetection, and R3Det is also integrated in our open source rotation detection benchmark: https://github.com/yangxue0827/RotationDetection. | ['Tao He', 'Ziming Feng', 'Xue Yang', 'Junchi Yan'] | 2019-08-15 | null | null | null | null | ['object-detection-in-aerial-images'] | ['computer-vision'] | [-3.27038944e-01 -5.69218516e-01 8.80947039e-02 -2.59007722e-01
-7.37142205e-01 -3.18386972e-01 2.72653729e-01 -2.28294283e-01
-3.09573054e-01 1.80755660e-01 1.28631681e-01 -2.85438061e-01
-1.39930546e-01 -7.33480036e-01 -4.37090933e-01 -9.45338428e-01
-4.42279764e-02 2.14115173e-01 1.11878783e-01 -7.58576989e-02
1.50085256e-01 8.18060756e-01 -1.30765295e+00 -1.89464480e-01
8.39354694e-01 8.81000817e-01 3.72004777e-01 6.42889678e-01
4.58123118e-01 4.25921530e-01 -1.65686205e-01 9.84164551e-02
7.49122977e-01 3.87905352e-02 -5.12798786e-01 1.45021960e-01
4.05330807e-01 -7.07009733e-01 -3.86811614e-01 1.09819913e+00
7.16691732e-01 9.40156057e-02 5.91020584e-01 -8.99486899e-01
-5.59686303e-01 3.23027730e-01 -1.39477575e+00 5.05233228e-01
3.61768156e-02 2.90963978e-01 8.22793603e-01 -1.33630431e+00
3.07474285e-01 1.23869455e+00 6.50890052e-01 -5.28390147e-02
-8.73465836e-01 -9.64542627e-01 2.70732313e-01 3.20686102e-01
-1.90391183e+00 -4.36426580e-01 5.10744989e-01 -4.25579220e-01
5.71129739e-01 4.48549241e-01 3.77234399e-01 4.96026903e-01
-2.03151241e-01 6.28689885e-01 7.54513800e-01 1.60794687e-02
-2.21663445e-01 -1.87232226e-01 1.62838586e-02 5.96335053e-01
5.19641280e-01 3.67058031e-02 2.08488163e-02 -6.65940298e-03
1.10447311e+00 4.11638290e-01 -4.42754924e-01 -1.20468676e-01
-1.45914245e+00 6.73091948e-01 9.61901844e-01 -2.47666258e-02
-3.48109692e-01 -4.69193757e-02 1.88147470e-01 1.24766774e-01
6.31051958e-01 2.47927383e-01 -4.94268507e-01 3.27867001e-01
-7.11237073e-01 3.38824660e-01 1.32474974e-01 8.91160965e-01
8.44720483e-01 -2.03983895e-02 -2.64665157e-01 7.83392549e-01
5.32546461e-01 1.09668112e+00 1.46982208e-01 -4.58162665e-01
5.45263588e-01 5.08583486e-01 3.62179756e-01 -1.26219106e+00
-6.54664040e-01 -7.40733087e-01 -1.08673120e+00 -1.29246265e-01
2.61652291e-01 -1.20162234e-01 -9.01330650e-01 1.05551815e+00
9.56507504e-01 2.46693745e-01 -2.20715851e-01 1.40887451e+00
8.58624279e-01 8.47208738e-01 -3.06893885e-01 -9.14463475e-02
1.65603876e+00 -8.27615798e-01 -1.99097663e-01 -2.32978463e-01
6.22479320e-01 -1.13332617e+00 7.73865104e-01 2.41586819e-01
-4.35456544e-01 -6.73995018e-01 -9.72044230e-01 -1.31224617e-01
7.10772499e-02 7.91510642e-01 5.71917713e-01 1.41137064e-01
-3.96387666e-01 1.95140824e-01 -8.79347742e-01 -1.72323704e-01
5.84863544e-01 1.25808328e-01 -2.13364333e-01 -1.91266134e-01
-7.80081511e-01 7.33897030e-01 1.34792358e-01 6.62684739e-01
-6.55991375e-01 -7.28877962e-01 -8.20495307e-01 -2.49463812e-01
3.53252620e-01 -6.17455661e-01 1.17008090e+00 -4.42556947e-01
-1.00713944e+00 5.49391627e-01 -6.48100525e-02 -1.23633631e-01
7.96613395e-01 -5.56197762e-01 -2.94605017e-01 1.06140983e-03
2.48752609e-01 3.89156908e-01 9.34352279e-01 -9.48842764e-01
-7.99142301e-01 -7.35545397e-01 -3.43019694e-01 4.58134025e-01
-3.85961384e-02 1.17901519e-01 -5.80875456e-01 -6.44778371e-01
5.77124476e-01 -1.03059721e+00 -4.56773460e-01 1.06732637e-01
-5.90725720e-01 6.69413656e-02 9.95600760e-01 -6.37161613e-01
1.07452250e+00 -2.33846474e+00 -2.68269002e-01 1.57059133e-01
2.63430059e-01 3.03263366e-01 -2.52879281e-02 2.54027937e-02
-1.04842976e-01 -2.03077585e-01 -4.24952991e-02 -1.14944749e-01
-3.22108567e-01 -2.82505602e-01 -3.86025727e-01 1.25994408e+00
2.77098536e-01 5.21863699e-01 -8.05176914e-01 -2.57638961e-01
5.71643174e-01 7.30586946e-01 -3.29609692e-01 2.81983707e-02
3.22535217e-01 6.03876948e-01 -8.16810310e-01 1.02381456e+00
1.40009665e+00 -2.37531215e-01 -3.48581344e-01 -6.23946428e-01
-5.15415013e-01 -2.08745222e-03 -1.82299995e+00 1.13417804e+00
-4.03132945e-01 5.09631276e-01 -9.93858185e-03 -6.87619269e-01
1.00128686e+00 -1.31827444e-01 4.47349519e-01 -5.98315537e-01
2.92340368e-01 1.75851718e-01 -2.95995101e-02 -4.06992435e-01
6.96322322e-01 2.73557425e-01 1.08004317e-01 -2.71480735e-02
-6.22879803e-01 -9.63771809e-03 3.08988430e-02 -1.03369281e-01
7.76252091e-01 1.37633711e-01 4.36704993e-01 -7.39479885e-02
5.10901093e-01 2.04622336e-02 6.18960440e-01 5.40609181e-01
-1.69877708e-01 1.07501292e+00 -4.22232673e-02 -5.58692634e-01
-9.20917392e-01 -8.13637674e-01 -7.27372468e-01 8.33790958e-01
3.83607447e-01 -1.80315748e-01 -1.67413548e-01 -1.87206164e-01
1.91484123e-01 2.54018694e-01 -4.75473166e-01 2.14861304e-01
-6.64938509e-01 -1.33382607e+00 2.90867418e-01 5.44148028e-01
6.52908564e-01 -5.91798067e-01 -6.89274788e-01 1.07770599e-01
-2.38591865e-01 -1.14230692e+00 -4.24597651e-01 -1.50847659e-01
-8.13798666e-01 -1.01495206e+00 -7.94227362e-01 -3.78143698e-01
9.97336030e-01 1.12207747e+00 6.86158240e-01 1.68460220e-01
-7.49050319e-01 -3.10258031e-01 -5.49102664e-01 -2.79002070e-01
3.03771257e-01 7.42201656e-02 1.97334550e-02 -1.71667757e-03
1.09364927e-01 -2.44017646e-01 -1.23395967e+00 7.79437840e-01
-6.49389982e-01 3.17928195e-01 7.82219350e-01 5.58440328e-01
6.11084998e-01 -1.17407732e-01 6.51223734e-02 -4.02345628e-01
-1.16142891e-01 -4.84077245e-01 -1.12712097e+00 -1.40840784e-01
-3.75357538e-01 -9.16812792e-02 3.20346355e-01 -2.39390254e-01
-8.25935423e-01 3.13427180e-01 -9.29549336e-02 -4.68690336e-01
-1.03043662e-02 2.58714974e-01 -1.07661693e-03 8.54736567e-02
7.27172017e-01 2.45411679e-01 -2.93013364e-01 -6.46928251e-01
3.70432496e-01 8.36341083e-01 3.93156201e-01 -2.31601894e-01
1.34023190e+00 8.26455474e-01 -8.96676555e-02 -9.93510723e-01
-7.37184048e-01 -9.87260640e-01 -3.45242649e-01 -3.37337516e-02
5.80260754e-01 -1.51328182e+00 -6.01997912e-01 6.70147955e-01
-9.12311852e-01 -3.81231643e-02 3.06395050e-02 6.60923958e-01
1.57827362e-01 4.07127023e-01 -3.55777919e-01 -7.41525948e-01
-7.48400807e-01 -1.25968444e+00 1.18659401e+00 4.29716319e-01
3.68897766e-01 -2.90510297e-01 9.08338130e-02 3.17083776e-01
4.46478844e-01 2.00881258e-01 1.73308849e-02 -2.05206975e-01
-7.59585321e-01 -3.20125729e-01 -7.22526252e-01 1.94666088e-01
8.00009072e-02 4.24718946e-01 -8.59587967e-01 -4.90824312e-01
-2.33203620e-01 4.52441834e-02 1.04518139e+00 5.05419374e-01
1.14319193e+00 -1.44790530e-01 -4.60717499e-01 1.08411264e+00
1.58362198e+00 -2.56803811e-01 6.13170981e-01 4.08688396e-01
1.06318414e+00 7.07063824e-02 1.21910155e+00 7.63625085e-01
6.84972763e-01 8.17513883e-01 6.36048555e-01 -2.97798842e-01
-1.73789844e-01 2.24824637e-01 4.71859217e-01 5.06119967e-01
-4.83769357e-01 9.23328921e-02 -9.28609550e-01 4.91148680e-01
-1.69611967e+00 -9.88735437e-01 -6.67908669e-01 2.31722665e+00
5.65789700e-01 -1.36094436e-01 -7.57869557e-02 -4.72497009e-02
7.54222214e-01 3.55437964e-01 -5.10532856e-01 4.18877155e-01
-3.94492932e-02 -1.42693639e-01 1.02824318e+00 5.10531425e-01
-1.46900320e+00 8.33190441e-01 3.55965042e+00 9.33978915e-01
-1.40250850e+00 6.87523037e-02 6.17087424e-01 -7.54470099e-03
2.06660479e-01 1.40710443e-01 -1.30677354e+00 3.25165063e-01
1.67996272e-01 1.57295704e-01 1.82573631e-01 1.02114880e+00
5.34551740e-01 -2.40195692e-01 -3.82092685e-01 1.04006302e+00
-2.22446293e-01 -1.00891209e+00 -3.37359101e-01 3.26579325e-02
5.98516822e-01 6.71318531e-01 2.60525919e-03 1.78499326e-01
6.66866899e-02 -7.12977588e-01 7.27289855e-01 1.32692307e-01
8.33456635e-01 -6.81765616e-01 6.13993049e-01 5.48136085e-02
-1.66968858e+00 -5.81476651e-02 -7.44215071e-01 1.62086878e-02
-5.50334565e-02 1.03441978e+00 -1.01540697e+00 8.48453999e-01
9.00210798e-01 8.25884342e-01 -7.38969028e-01 1.35535610e+00
-4.00925726e-01 2.71982700e-01 -5.79093695e-01 2.75900602e-01
3.87514085e-02 -2.44859666e-01 6.73989356e-01 1.23973477e+00
5.46806991e-01 1.10474102e-01 2.75579453e-01 5.32400131e-01
1.83614343e-01 2.27766529e-01 -2.30187982e-01 6.24315560e-01
6.88284695e-01 1.85548151e+00 -8.50676358e-01 -9.90060791e-02
-3.70712906e-01 5.55107713e-01 2.34512705e-02 2.73820132e-01
-1.23336625e+00 -3.53101671e-01 8.05833161e-01 3.47441047e-01
5.34927130e-01 -3.03656250e-01 -9.82286632e-02 -1.50818062e+00
1.62817255e-01 -7.71911085e-01 2.74168909e-01 -6.40572131e-01
-8.97266388e-01 4.72565979e-01 1.68357696e-02 -1.74364805e+00
3.20575058e-01 -4.29494441e-01 -6.91266418e-01 7.49948919e-01
-1.73095632e+00 -1.15052211e+00 -9.93522584e-01 4.22224402e-01
5.70929348e-01 2.11154550e-01 3.23975146e-01 6.34387374e-01
-9.68684196e-01 3.75265121e-01 3.09191674e-01 4.08065736e-01
8.41830313e-01 -7.99211264e-01 5.15720308e-01 1.12285089e+00
-9.57573652e-02 4.17761713e-01 7.75576711e-01 -3.94544572e-01
-1.59810805e+00 -1.43803692e+00 3.49168599e-01 -1.98953509e-01
5.17709136e-01 -2.31836542e-01 -7.97123909e-01 5.35707712e-01
-3.94691169e-01 5.58574021e-01 1.27966419e-01 -8.61531571e-02
-3.30910087e-01 -4.25397575e-01 -8.67473066e-01 3.50272179e-01
9.65729773e-01 -4.47414108e-02 -1.60077617e-01 5.73869228e-01
4.11931902e-01 -7.73891211e-01 -7.09143937e-01 6.90058708e-01
6.15113318e-01 -7.15117931e-01 1.19133079e+00 1.54241309e-01
2.89893895e-01 -1.07332969e+00 -1.46613866e-01 -9.53482270e-01
-3.60304087e-01 -2.92671651e-01 1.04201190e-01 1.04190445e+00
2.07516149e-01 -7.57886648e-01 3.64469618e-01 -3.55099700e-02
-1.09340727e-01 -5.70424438e-01 -8.18493068e-01 -6.80650413e-01
-3.39915544e-01 -4.10055697e-01 5.29167712e-01 1.00711012e+00
-6.88793659e-01 2.17851982e-01 -3.41779560e-01 7.42503047e-01
7.29800582e-01 4.54574645e-01 1.25963497e+00 -9.53293085e-01
-1.76338479e-01 -2.78718412e-01 -3.76063257e-01 -1.29610455e+00
-5.31458676e-01 -6.42794311e-01 1.95173454e-02 -1.47368443e+00
3.07811856e-01 -7.44271636e-01 -1.04761578e-01 5.01911163e-01
-2.91965663e-01 4.23000216e-01 1.30872667e-01 5.25961578e-01
-5.11663258e-01 5.50096929e-01 1.28511715e+00 4.21040505e-02
-2.08508283e-01 1.34802222e-01 -6.74880445e-01 7.12950230e-01
8.14321637e-01 -6.79574370e-01 -4.81859073e-02 -6.83423162e-01
3.03368896e-01 -1.36421114e-01 7.40852594e-01 -1.10124695e+00
1.41312703e-01 -1.74796566e-01 8.20186198e-01 -1.21412611e+00
7.82805234e-02 -6.36601865e-01 1.87991798e-01 5.72924078e-01
3.20122808e-01 5.39999977e-02 1.56678259e-02 2.66614705e-01
3.53785604e-02 1.51477292e-01 1.04837668e+00 2.37636670e-01
-6.33594155e-01 5.92685699e-01 7.93576017e-02 -9.77364555e-02
9.88255501e-01 -5.58762178e-02 -5.90728641e-01 8.47489089e-02
-4.45187598e-01 4.86936659e-01 4.60596830e-01 4.52646941e-01
5.27643979e-01 -1.03444529e+00 -1.21289134e+00 1.86875820e-01
2.96737671e-01 6.77985191e-01 5.10297179e-01 1.36130142e+00
-9.88988876e-01 1.91499040e-01 2.99321748e-02 -8.03053260e-01
-1.22420800e+00 1.62629694e-01 3.28841358e-01 -1.18176952e-01
-8.15643132e-01 8.76737535e-01 3.21037591e-01 -3.05796146e-01
-6.79935515e-02 -5.95903993e-01 -1.08891368e-01 1.73772633e-01
8.33695829e-01 5.62093437e-01 9.03145894e-02 -8.12078953e-01
-7.82175958e-01 8.03621173e-01 -2.62471348e-01 2.97042698e-01
1.38993120e+00 -1.60332307e-01 4.75169905e-02 -2.73011357e-01
1.02255702e+00 1.83749408e-01 -1.29697800e+00 -3.06725174e-01
-5.13565421e-01 -9.30726230e-01 2.27102950e-01 -2.26388693e-01
-1.37147665e+00 6.04980171e-01 9.65769112e-01 -1.65279686e-01
1.09226859e+00 -3.18206623e-02 4.89781410e-01 3.35499495e-01
4.10851017e-02 -7.65169382e-01 -1.06360413e-01 6.48644745e-01
1.00031328e+00 -1.53350699e+00 7.87470102e-01 -4.35271829e-01
-5.50698161e-01 1.08024061e+00 6.84565425e-01 -3.44525665e-01
5.54107666e-01 1.10272743e-01 2.00161159e-01 -2.08294868e-01
-1.95389211e-01 -3.61373574e-01 2.15921655e-01 2.65490294e-01
2.52135038e-01 2.60281593e-01 -2.60367841e-01 1.44063115e-01
-8.22947696e-02 -3.60246688e-01 3.64258856e-01 7.68053412e-01
-4.56728339e-01 -5.20408571e-01 -8.50440979e-01 3.51506859e-01
-4.18245554e-01 -1.17033623e-01 2.10827395e-01 7.34164774e-01
3.74883823e-02 6.40592158e-01 9.09579471e-02 -3.54776740e-01
3.79507959e-01 -8.46847475e-01 2.45043740e-01 -4.22324777e-01
-2.55621344e-01 3.43182802e-01 -9.62706059e-02 -5.24326742e-01
-3.20503205e-01 -9.69366729e-01 -1.39205003e+00 -2.43618980e-01
-7.24299073e-01 -1.55913800e-01 7.84434438e-01 6.02198958e-01
3.66003096e-01 2.80055195e-01 1.11670685e+00 -1.11967039e+00
-8.74751151e-01 -1.07610261e+00 -4.18963730e-01 4.43615578e-02
4.93238717e-01 -6.43712521e-01 -5.02298653e-01 -1.73242345e-01] | [8.792088508605957, -0.7884263396263123] |
b29d5260-c9bb-41ae-947c-f0d005171672 | one2set-generating-diverse-keyphrases-as-a | 2105.11134 | null | https://arxiv.org/abs/2105.11134v1 | https://arxiv.org/pdf/2105.11134v1.pdf | One2Set: Generating Diverse Keyphrases as a Set | Recently, the sequence-to-sequence models have made remarkable progress on the task of keyphrase generation (KG) by concatenating multiple keyphrases in a predefined order as a target sequence during training. However, the keyphrases are inherently an unordered set rather than an ordered sequence. Imposing a predefined order will introduce wrong bias during training, which can highly penalize shifts in the order between keyphrases. In this work, we propose a new training paradigm One2Set without predefining an order to concatenate the keyphrases. To fit this paradigm, we propose a novel model that utilizes a fixed set of learned control codes as conditions to generate a set of keyphrases in parallel. To solve the problem that there is no correspondence between each prediction and target during training, we propose a $K$-step target assignment mechanism via bipartite matching, which greatly increases the diversity and reduces the duplication ratio of generated keyphrases. The experimental results on multiple benchmarks demonstrate that our approach significantly outperforms the state-of-the-art methods. | ['Qi Zhang', 'Yige Xu', 'Yichao Luo', 'Tao Gui', 'Jiacheng Ye'] | 2021-05-24 | null | https://aclanthology.org/2021.acl-long.354 | https://aclanthology.org/2021.acl-long.354.pdf | acl-2021-5 | ['keyphrase-generation'] | ['natural-language-processing'] | [ 5.69706202e-01 -3.60943705e-01 -4.52556223e-01 -7.96029121e-02
-5.80543280e-01 -8.68557215e-01 4.54835862e-01 2.25522608e-01
-4.79646176e-01 9.65762675e-01 1.65736660e-01 -4.47413534e-01
-2.70163547e-02 -9.77324903e-01 -9.56601381e-01 -6.48451746e-01
2.18730256e-01 3.38703245e-01 4.79718387e-01 -4.67634201e-01
7.91571558e-01 2.87397057e-01 -1.65860367e+00 5.43985605e-01
9.01342571e-01 7.41297305e-01 5.64244509e-01 5.17193913e-01
-4.14321482e-01 7.11895168e-01 -6.17991686e-01 -5.58988929e-01
6.85926080e-01 -7.10142374e-01 -8.32548141e-01 -1.22624144e-01
6.14382803e-01 -3.41849416e-01 -4.98828977e-01 1.09988892e+00
3.77839029e-01 -4.25457731e-02 4.45842385e-01 -1.37724173e+00
-4.69817102e-01 9.77624238e-01 -5.17295539e-01 2.10719675e-01
3.52128625e-01 -4.73443270e-02 1.43986690e+00 -7.58543074e-01
6.51668370e-01 7.64666975e-01 2.61389107e-01 4.00007695e-01
-1.12894607e+00 -1.08328772e+00 2.17963755e-01 3.90802503e-01
-1.45984495e+00 -9.25388709e-02 7.68086672e-01 -2.20766380e-01
5.74987829e-01 5.75839937e-01 9.09226000e-01 5.75653434e-01
2.93485761e-01 9.02088821e-01 9.83612418e-01 -5.29153943e-01
-6.82300255e-02 8.23434740e-02 1.96223840e-01 6.79996908e-01
3.44684035e-01 1.93197072e-01 -7.15672791e-01 -5.99684656e-01
6.27336323e-01 1.50684357e-01 -3.98440003e-01 -4.30577517e-01
-1.62025237e+00 8.35632026e-01 2.46406406e-01 2.36957699e-01
-1.51260346e-01 9.09970179e-02 3.70882958e-01 5.99910676e-01
2.90809348e-02 6.67257309e-01 -6.24457777e-01 3.03955283e-02
-1.00910246e+00 7.09793985e-01 8.78216922e-01 9.16993737e-01
1.16055036e+00 -4.33686137e-01 -4.73216385e-01 6.24119222e-01
-1.66149944e-01 4.31152672e-01 4.74170059e-01 -2.61950463e-01
7.00117826e-01 8.00833642e-01 1.56409979e-01 -1.23940516e+00
8.01717639e-02 -2.36240074e-01 -8.32827806e-01 -4.84749466e-01
1.34324938e-01 -6.14844449e-02 -9.67290521e-01 1.62993670e+00
3.09573084e-01 2.44025186e-01 -2.28527054e-01 6.49129808e-01
4.84800607e-01 9.35901463e-01 -4.60226119e-01 -5.65927625e-01
9.74155843e-01 -1.08935559e+00 -5.10415018e-01 8.09257776e-02
5.97746491e-01 -1.08656526e+00 9.95180070e-01 2.61651039e-01
-7.98340142e-01 -5.23589790e-01 -1.06079006e+00 4.21445638e-01
-2.54587322e-01 3.33648026e-02 5.10581851e-01 3.86412591e-01
-8.54835153e-01 6.13020658e-01 -4.62477989e-02 -7.12296665e-02
-3.83927763e-01 3.60077560e-01 -1.34190902e-01 6.17987849e-02
-1.55060661e+00 5.45729101e-01 1.07531857e+00 -1.84751436e-01
-7.78224409e-01 -9.52812135e-01 -4.34674859e-01 1.44975714e-03
7.51433671e-01 -5.06437778e-01 1.17092931e+00 -8.29209387e-01
-1.23224366e+00 6.54051304e-01 -4.16603126e-02 -3.35775465e-01
3.30362260e-01 -2.15433970e-01 -4.47393239e-01 9.02512744e-02
1.18133821e-01 6.57538056e-01 1.09720087e+00 -1.15519762e+00
-1.09668434e+00 2.31699675e-01 2.12246552e-02 3.29707384e-01
-4.27447200e-01 1.62058383e-01 -6.26859903e-01 -1.11665308e+00
3.22619826e-02 -1.05456972e+00 -4.57901001e-01 -5.91477334e-01
-6.05775952e-01 -1.06571384e-01 4.77039158e-01 -3.38641852e-01
1.97312534e+00 -1.81047940e+00 1.97554767e-01 6.61538303e-01
1.65595815e-01 2.65382051e-01 -1.55184641e-01 1.07520700e+00
-2.95103371e-01 1.62414119e-01 -3.26741010e-01 5.32602310e-01
-2.12740690e-01 8.15080106e-02 -9.26874161e-01 -3.24902795e-02
-1.31508544e-01 7.16973007e-01 -1.09264743e+00 -5.76492429e-01
-3.46618861e-01 -3.27769250e-01 -7.61985600e-01 5.70645869e-01
-4.87166286e-01 3.16441730e-02 -4.20640796e-01 1.83677435e-01
5.74620008e-01 -3.19447100e-01 3.54009718e-01 -3.48446548e-01
-1.25890926e-01 3.79334927e-01 -1.38006055e+00 1.37868953e+00
-8.69666692e-03 3.14062461e-02 -4.63287503e-01 -9.96246874e-01
1.20491993e+00 1.37869224e-01 6.46891892e-01 -2.36443147e-01
-1.04119323e-01 3.42332870e-01 1.21866181e-01 1.07034361e-02
9.07863796e-01 5.41147999e-02 -2.67673343e-01 8.31671596e-01
-2.74255604e-01 -1.23587176e-01 5.91789186e-01 5.85537612e-01
1.09336996e+00 -1.78552300e-01 3.40369672e-01 -2.34852403e-01
7.54124463e-01 1.96688473e-01 6.61533356e-01 1.06803787e+00
3.64698857e-01 4.07174140e-01 4.74048287e-01 -6.60869896e-01
-1.22784424e+00 -6.85843706e-01 2.58897781e-01 1.09054601e+00
2.85814822e-01 -9.71014202e-01 -5.28004408e-01 -9.22705889e-01
2.98320651e-01 3.16849738e-01 -4.63831395e-01 -3.65274817e-01
-8.01061511e-01 -7.09160864e-01 6.04512691e-01 2.18856968e-02
3.91264737e-01 -9.94108558e-01 -3.16323698e-01 4.24175173e-01
-3.10736716e-01 -8.06059480e-01 -1.25742865e+00 2.42606699e-01
-4.97618198e-01 -1.03569877e+00 -8.86547387e-01 -1.06390178e+00
9.89800572e-01 4.67726499e-01 1.13440776e+00 2.99438417e-01
-8.80501568e-02 1.07946545e-01 -7.35374153e-01 -7.15500191e-02
-5.46295106e-01 4.52418089e-01 6.46970198e-02 1.52594700e-01
1.86893195e-01 -4.51450437e-01 -6.05949938e-01 4.57377076e-01
-1.17845476e+00 4.76954639e-01 9.21052933e-01 1.15703523e+00
5.44041812e-01 8.39576274e-02 3.32094342e-01 -8.95609438e-01
9.25550222e-01 -3.26650590e-01 -5.82090735e-01 5.99235237e-01
-9.64168251e-01 4.33735192e-01 1.00648212e+00 -6.78958893e-01
-5.92578173e-01 -3.14382836e-02 3.32102358e-01 -3.11572522e-01
3.45119268e-01 5.66304922e-01 1.15058616e-01 -2.10446194e-01
4.41521198e-01 9.76826191e-01 -1.07869342e-01 -4.34846282e-01
4.02412653e-01 5.16898572e-01 3.26327562e-01 -8.26616347e-01
1.26338482e+00 -3.02063208e-02 -5.80576472e-02 -2.97484934e-01
-6.30799294e-01 -6.80382192e-01 -4.52267438e-01 1.86886732e-02
2.43491009e-01 -7.43797481e-01 -6.03747845e-01 4.35067147e-01
-1.14434469e+00 2.37628892e-02 -7.79709732e-03 9.85635072e-02
-2.94909149e-01 7.73818851e-01 -4.26935792e-01 -3.20978820e-01
-7.26569891e-01 -1.11012983e+00 6.98237658e-01 1.04747929e-01
-1.88399553e-01 -2.42408201e-01 1.28545806e-01 -2.09586412e-01
2.99889054e-02 -1.18711263e-01 1.15653658e+00 -8.81896794e-01
-9.11325693e-01 -1.13371022e-01 -1.02842480e-01 1.48533002e-01
2.39093214e-01 -9.38164219e-02 1.01200901e-01 -4.77028191e-01
-4.63208258e-01 -3.61632288e-01 8.33215773e-01 -3.71923804e-01
1.20005620e+00 -7.03324735e-01 -5.69919825e-01 4.87032145e-01
1.31459391e+00 4.63628352e-01 4.82928246e-01 5.03869414e-01
7.01279581e-01 4.86510962e-01 8.60852182e-01 5.08638799e-01
3.45246255e-01 6.90435708e-01 5.99165820e-02 1.13395631e-01
3.18309069e-01 -8.50922883e-01 2.96216398e-01 1.10047567e+00
3.97106349e-01 -1.15369171e-01 -7.52290547e-01 6.20617867e-01
-1.91590583e+00 -1.04551983e+00 1.10727929e-01 2.29267383e+00
1.53723967e+00 1.89688340e-01 -6.48102537e-03 2.63827890e-01
8.77358794e-01 3.89209569e-01 -2.78876334e-01 -8.59672651e-02
3.20336707e-02 3.18946600e-01 8.36993635e-01 3.57247353e-01
-8.68681788e-01 1.18200791e+00 6.52378845e+00 1.14475405e+00
-1.18081129e+00 -5.30130386e-01 4.60215896e-01 2.27801234e-01
-7.48014331e-01 2.89506465e-01 -1.33748484e+00 6.99994683e-01
5.31904101e-01 -4.90880996e-01 5.76570928e-01 6.77946627e-01
-1.71551570e-01 -1.07583767e-02 -9.63993728e-01 9.04013157e-01
1.39570190e-02 -1.73052573e+00 6.61551654e-01 -2.67159760e-01
8.73077452e-01 -4.62421030e-01 -1.84018433e-01 8.15147981e-02
4.55992490e-01 -5.13327777e-01 6.70452774e-01 4.35304999e-01
7.05706239e-01 -8.60948980e-01 2.22374633e-01 4.27472115e-01
-1.39169002e+00 -7.77166476e-03 -3.94530803e-01 1.37737617e-01
-1.58947945e-01 6.05906367e-01 -1.05351365e+00 6.82985604e-01
5.63865781e-01 4.21615869e-01 -3.76735985e-01 1.05373526e+00
-3.74826878e-01 5.66581488e-01 -2.20674589e-01 -5.55446327e-01
2.03376964e-01 -2.75604546e-01 4.93388414e-01 1.16649449e+00
2.71654040e-01 1.46644697e-01 6.99897230e-01 3.87411147e-01
-2.16081436e-03 5.55356443e-01 -4.60478634e-01 -2.28125975e-01
7.65515268e-01 1.23075044e+00 -7.47229636e-01 -6.10217452e-01
-1.58560887e-01 9.86941576e-01 1.50131956e-01 2.87907481e-01
-7.85723746e-01 -8.02322447e-01 5.47022343e-01 7.81411380e-02
5.00329792e-01 -1.75041616e-01 7.96612948e-02 -1.27133524e+00
7.16687813e-02 -1.37794757e+00 3.67396802e-01 -3.77074450e-01
-1.12175667e+00 5.16622841e-01 1.82415262e-01 -1.40784359e+00
-1.89794153e-01 -1.85754448e-01 -4.32297498e-01 8.51897776e-01
-1.38042903e+00 -8.65947783e-01 1.13980053e-02 6.43227696e-01
3.39441836e-01 -1.73182130e-01 5.93461394e-01 1.98655948e-01
-2.43668646e-01 7.62739062e-01 2.26649612e-01 1.76970512e-01
1.09942889e+00 -1.06978202e+00 6.40721321e-01 9.04695749e-01
2.60262609e-01 9.49441969e-01 6.98235393e-01 -9.20964837e-01
-1.35853779e+00 -8.73903990e-01 1.00224221e+00 6.84175757e-05
7.17848718e-01 -3.84195566e-01 -7.21141398e-01 3.92724484e-01
1.69804558e-01 -3.11769903e-01 7.94346154e-01 -2.96684980e-01
-5.40764749e-01 -4.53058094e-01 -5.21060407e-01 9.32400048e-01
1.04595590e+00 -3.69798601e-01 -6.97567403e-01 2.58194834e-01
1.16013706e+00 -4.99290615e-01 -5.84867477e-01 6.48188174e-01
6.59262002e-01 -6.56689167e-01 8.40187669e-01 -6.15429103e-01
5.10071993e-01 -7.11754084e-01 -5.79753444e-02 -1.33029890e+00
-4.37885910e-01 -1.08127177e+00 -2.11693347e-01 9.47202802e-01
4.62644547e-01 -5.93314826e-01 9.66498017e-01 9.98905972e-02
7.07236677e-02 -8.79425168e-01 -4.10851419e-01 -7.59016216e-01
-2.58514583e-01 8.05813000e-02 9.39332426e-01 1.04868257e+00
1.11129589e-01 4.33993906e-01 -9.22464252e-01 -1.37950003e-01
4.45323586e-01 6.19042695e-01 1.14882052e+00 -9.69645023e-01
-4.99858648e-01 -2.93110877e-01 7.46971965e-02 -1.54424667e+00
-1.40170246e-01 -1.07009554e+00 5.56265488e-02 -1.13932371e+00
2.52239078e-01 -6.02114499e-01 -3.87481451e-01 6.43760920e-01
-6.84887707e-01 -1.73580557e-01 3.40254426e-01 5.29601395e-01
-3.00462484e-01 4.44321275e-01 1.35350108e+00 -2.22882435e-01
-2.64627099e-01 -1.51149973e-01 -7.87066698e-01 1.44003570e-01
7.29345918e-01 -6.50375187e-01 -6.22100651e-01 3.09591144e-02
5.07680178e-01 9.84278992e-02 -1.19189456e-01 -7.36824930e-01
4.52958107e-01 -6.46012843e-01 1.55179024e-01 -9.48161602e-01
-1.98412046e-01 -8.63642752e-01 4.05297875e-01 7.09687233e-01
-5.54580092e-01 3.60389888e-01 1.26505733e-01 5.25689840e-01
-1.52406901e-01 -3.64266425e-01 4.01438951e-01 -2.77657062e-01
-8.52749109e-01 5.16751647e-01 -2.18761459e-01 -3.39164920e-02
9.89123285e-01 -1.03305735e-01 -1.89710677e-01 -2.17028648e-01
-6.88840002e-02 2.53212422e-01 3.75724971e-01 5.17930567e-01
6.24846220e-01 -1.39880371e+00 -6.75081551e-01 3.32560241e-01
3.25596899e-01 -1.94331393e-01 -2.01815307e-01 3.78984451e-01
-4.94638681e-01 5.16392589e-01 -1.74738958e-01 -2.09158435e-01
-1.34503186e+00 6.75081670e-01 -5.34158126e-02 -8.08611393e-01
-2.55677521e-01 8.90848935e-01 6.83164597e-02 -5.73073328e-01
1.88960835e-01 -1.73314661e-01 -5.50532453e-02 1.06221035e-01
5.38573503e-01 -8.46028626e-02 9.90616158e-03 -2.41376355e-01
-5.44673093e-02 5.42408109e-01 -8.19319189e-01 2.10931152e-03
9.08821404e-01 1.89768121e-01 -5.20857990e-01 9.40034613e-02
1.12055337e+00 1.88880786e-01 -7.74457932e-01 -5.34654260e-01
1.66725606e-01 -6.67850494e-01 -4.54238743e-01 -5.88187397e-01
-8.03202689e-01 2.03937098e-01 1.31287768e-01 2.02523276e-01
1.14019918e+00 -3.51980567e-01 1.24585485e+00 6.67197108e-01
5.65164626e-01 -1.06673551e+00 3.65842193e-01 6.79495215e-01
6.85786486e-01 -7.83681989e-01 6.24189749e-02 -5.39468646e-01
-4.90547627e-01 1.13724065e+00 7.38619804e-01 -3.41101624e-02
4.88338172e-01 2.25193892e-02 -1.41687542e-01 2.15119436e-01
-8.82981539e-01 -1.90241992e-01 4.34841007e-01 8.65543485e-02
2.61532217e-01 -5.08036539e-02 -1.01567066e+00 3.16590458e-01
-4.19372708e-01 4.08310182e-02 4.22010273e-01 1.08559310e+00
-7.34013736e-01 -1.72770929e+00 -5.30571640e-01 6.64927006e-01
-4.02788311e-01 -6.45082831e-01 -5.24684131e-01 3.88264686e-01
9.09556225e-02 5.63796818e-01 -1.65445581e-01 -9.06231225e-01
1.69082642e-01 5.84185943e-02 1.89721107e-01 -6.90487325e-01
-8.56763244e-01 -2.25718059e-02 -1.19836658e-01 -1.53281599e-01
-9.64703932e-02 -2.10458577e-01 -9.26497340e-01 -4.13243800e-01
-4.48832452e-01 8.10765564e-01 6.63099438e-02 6.44647598e-01
4.25673097e-01 3.02431107e-01 1.19701314e+00 -3.43412876e-01
-8.95087779e-01 -6.84733033e-01 -3.61424237e-01 3.94482493e-01
1.12251952e-01 -2.56157458e-01 -1.44742280e-01 2.78470665e-01] | [12.298727989196777, 8.906655311584473] |
39e140fe-0efb-4bbf-9d4f-f238e9b56fff | online-bag-of-visual-words-generation-for | 2012.11552 | null | https://arxiv.org/abs/2012.11552v2 | https://arxiv.org/pdf/2012.11552v2.pdf | OBoW: Online Bag-of-Visual-Words Generation for Self-Supervised Learning | Learning image representations without human supervision is an important and active research field. Several recent approaches have successfully leveraged the idea of making such a representation invariant under different types of perturbations, especially via contrastive-based instance discrimination training. Although effective visual representations should indeed exhibit such invariances, there are other important characteristics, such as encoding contextual reasoning skills, for which alternative reconstruction-based approaches might be better suited. With this in mind, we propose a teacher-student scheme to learn representations by training a convolutional net to reconstruct a bag-of-visual-words (BoW) representation of an image, given as input a perturbed version of that same image. Our strategy performs an online training of both the teacher network (whose role is to generate the BoW targets) and the student network (whose role is to learn representations), along with an online update of the visual-words vocabulary (used for the BoW targets). This idea effectively enables fully online BoW-guided unsupervised learning. Extensive experiments demonstrate the interest of our BoW-based strategy which surpasses previous state-of-the-art methods (including contrastive-based ones) in several applications. For instance, in downstream tasks such Pascal object detection, Pascal classification and Places205 classification, our method improves over all prior unsupervised approaches, thus establishing new state-of-the-art results that are also significantly better even than those of supervised pre-training. We provide the implementation code at https://github.com/valeoai/obow. | ['Patrick Pérez', 'Matthieu Cord', 'Nikos Komodakis', 'Gilles Puy', 'Andrei Bursuc', 'Spyros Gidaris'] | 2020-12-21 | obow-online-bag-of-visual-words-generation | http://openaccess.thecvf.com//content/CVPR2021/html/Gidaris_OBoW_Online_Bag-of-Visual-Words_Generation_for_Self-Supervised_Learning_CVPR_2021_paper.html | http://openaccess.thecvf.com//content/CVPR2021/papers/Gidaris_OBoW_Online_Bag-of-Visual-Words_Generation_for_Self-Supervised_Learning_CVPR_2021_paper.pdf | cvpr-2021-1 | ['self-supervised-image-classification'] | ['computer-vision'] | [ 3.95549834e-01 1.80655748e-01 -6.97833523e-02 -4.16809678e-01
-5.84991872e-01 -5.21283388e-01 8.43932927e-01 4.70934987e-01
-5.10747313e-01 3.99345011e-01 5.04198037e-02 -3.79240483e-01
-8.70110560e-03 -9.58948672e-01 -1.13891792e+00 -9.52942073e-01
2.37102322e-02 4.01286781e-01 4.42031085e-01 -4.49796438e-01
3.39931548e-01 5.91437936e-01 -2.03633976e+00 4.86309379e-01
5.85081220e-01 1.02026260e+00 1.56498551e-01 7.52719223e-01
-7.33783245e-02 1.08318031e+00 -3.81732225e-01 -2.68075079e-01
2.59158820e-01 -2.29498953e-01 -8.84728014e-01 8.20599794e-02
6.09944701e-01 -1.54355019e-01 -3.96128803e-01 1.02725458e+00
3.53032827e-01 6.30003512e-01 8.12147737e-01 -1.08572745e+00
-9.32884276e-01 2.98284531e-01 -4.03352439e-01 2.90390074e-01
2.40512818e-01 2.79003233e-01 1.06334174e+00 -7.30094075e-01
5.16418695e-01 1.10326719e+00 3.82345080e-01 6.63715363e-01
-1.44442832e+00 -5.29475987e-01 2.94901520e-01 3.66617620e-01
-1.17329872e+00 -4.22143877e-01 6.84775114e-01 -6.05624557e-01
8.94464552e-01 2.29621515e-01 5.00221074e-01 1.26204371e+00
-1.17842771e-01 9.47267175e-01 1.12760353e+00 -7.39175141e-01
3.13745677e-01 2.65010864e-01 1.71896219e-01 6.75879419e-01
6.91741407e-02 1.92150474e-01 -2.88464010e-01 5.03171720e-02
5.92233181e-01 3.57135564e-01 -2.70055294e-01 -9.06254590e-01
-9.92477179e-01 9.24439847e-01 8.64282548e-01 4.03031468e-01
-2.42640719e-01 2.67297566e-01 5.25569081e-01 4.17261988e-01
3.22856367e-01 4.25645739e-01 -1.65796787e-01 1.67116597e-01
-7.33380795e-01 2.30301976e-01 3.14125359e-01 8.61861765e-01
9.30078745e-01 4.78540026e-02 -3.13657224e-01 7.32912600e-01
1.17042802e-01 9.27768946e-02 7.30796516e-01 -5.74054301e-01
1.48172066e-01 5.90682209e-01 -1.38379544e-01 -1.01301146e+00
-1.66254550e-01 -4.48708296e-01 -9.63417888e-01 4.59510088e-01
4.42410976e-01 5.00019729e-01 -1.24099374e+00 1.76024699e+00
3.92563760e-01 2.41997331e-01 1.32606342e-01 7.28820622e-01
6.73758626e-01 6.41974330e-01 1.54046327e-01 2.01796919e-01
1.40695047e+00 -1.10442781e+00 -3.10637057e-01 -2.40143389e-01
5.78919530e-01 -4.09876376e-01 1.19430614e+00 4.62743521e-01
-8.83869350e-01 -8.92231345e-01 -1.04397845e+00 -1.78662986e-01
-7.97601104e-01 3.05875912e-02 5.13785005e-01 4.79493260e-01
-1.08954430e+00 6.22851253e-01 -6.78327858e-01 -5.26519775e-01
7.21329510e-01 1.36940882e-01 -6.17697656e-01 -2.11989865e-01
-9.73883808e-01 8.11264455e-01 5.09576797e-01 -3.04360092e-02
-1.37218606e+00 -6.03137374e-01 -1.01440382e+00 1.62633270e-01
3.31179053e-01 -4.77422625e-01 1.13461220e+00 -1.40384710e+00
-1.46459138e+00 1.12831545e+00 2.47838899e-01 -6.26185179e-01
4.94704574e-01 -1.79728512e-02 -1.50606092e-02 3.61197412e-01
-5.38188703e-02 8.37768734e-01 1.24578464e+00 -1.33785021e+00
-4.07478452e-01 -2.69092947e-01 4.71941531e-01 -3.79808508e-02
-5.21757782e-01 -1.64256603e-01 -2.68426746e-01 -6.68878973e-01
-9.09814462e-02 -7.75514364e-01 -2.60586470e-01 2.88608700e-01
-1.75927341e-01 -2.67863274e-01 4.96417284e-01 -4.80686605e-01
7.08319426e-01 -2.50888228e+00 2.44318292e-01 1.21477880e-01
1.11462630e-01 3.89532119e-01 -2.92196810e-01 5.03858089e-01
-4.81120259e-01 -1.24570169e-01 -2.83554435e-01 -4.35646772e-01
3.34901772e-02 1.80496439e-01 -6.73758805e-01 6.48190200e-01
4.48786110e-01 7.94502556e-01 -1.02620780e+00 -1.85116053e-01
3.63498390e-01 5.08098543e-01 -6.29624546e-01 5.16955733e-01
-2.59404242e-01 5.98883033e-01 -2.10557103e-01 3.31147969e-01
5.31061888e-01 -1.18996449e-01 1.09093077e-01 1.62743051e-02
-8.82921834e-03 2.28520289e-01 -1.06327212e+00 1.68648183e+00
-5.48892200e-01 6.44807696e-01 -2.01104566e-01 -1.85457695e+00
9.01863515e-01 1.85521588e-01 1.29119515e-01 -9.33177114e-01
4.40524407e-02 -9.28718690e-03 -2.10250020e-01 -4.74207878e-01
3.55184466e-01 -1.10651121e-01 2.09885649e-02 2.45567933e-01
5.25368571e-01 -7.90019035e-02 1.10871442e-01 2.68431246e-01
1.05671585e+00 3.09075981e-01 4.36933994e-01 -3.31301332e-01
6.60390913e-01 -5.96953072e-02 1.38281360e-01 8.00311744e-01
-1.33771136e-01 6.82363033e-01 4.33475256e-01 -5.50588727e-01
-1.00919354e+00 -9.14234400e-01 -1.43972591e-01 1.38910902e+00
1.82606559e-02 -2.64078379e-01 -6.34533167e-01 -6.65006459e-01
1.77368335e-02 6.50106668e-01 -1.01813626e+00 -4.74179417e-01
-3.83723170e-01 -1.45498708e-01 5.36379337e-01 7.60730207e-01
2.43128419e-01 -1.27321529e+00 -9.11562800e-01 -6.97594583e-02
2.28839621e-01 -9.68738794e-01 1.79821685e-01 5.79998612e-01
-7.84019470e-01 -1.01526523e+00 -7.84512162e-01 -7.49224782e-01
9.23180878e-01 4.82345641e-01 1.00074339e+00 3.37094754e-01
-5.18361747e-01 6.62083447e-01 -5.28510988e-01 -4.03193593e-01
-3.13954234e-01 -6.75968006e-02 -1.32241786e-01 2.29819104e-01
9.52568725e-02 -7.48764992e-01 -5.77323020e-01 9.09636468e-02
-1.40187490e+00 -2.40348838e-02 4.47301000e-01 1.10706198e+00
6.16208255e-01 -4.10357825e-02 1.84392646e-01 -9.47326660e-01
2.19690800e-01 -4.12831306e-01 -6.66464031e-01 8.85452852e-02
-2.97762454e-01 1.58267155e-01 8.93279910e-01 -5.47046602e-01
-8.23371112e-01 1.07908413e-01 -1.39074340e-01 -7.38940716e-01
-5.38423181e-01 2.36132532e-01 -3.63929197e-02 6.19639307e-02
9.21973407e-01 4.42215413e-01 -1.10947721e-01 -4.02582973e-01
5.90363324e-01 5.86902797e-01 6.25339806e-01 -7.31208920e-01
1.01417220e+00 6.10349596e-01 -2.26424307e-01 -7.30464101e-01
-7.47656941e-01 -7.41549015e-01 -7.78059006e-01 -7.05993846e-02
8.52940023e-01 -1.06593311e+00 -5.82281470e-01 2.39779651e-01
-1.00516260e+00 -4.80133057e-01 -5.24594188e-01 1.98130205e-01
-8.02074790e-01 2.60568470e-01 -3.34625602e-01 -8.43274176e-01
-4.71638665e-02 -1.08761156e+00 1.02865744e+00 1.92369118e-01
-3.45311724e-02 -9.41085875e-01 1.41804218e-01 3.60665292e-01
4.22991782e-01 2.69338429e-01 1.01949060e+00 -8.20552468e-01
-4.70609218e-01 -1.73787028e-01 -1.82865456e-01 5.36739349e-01
-2.17206612e-01 -7.88883958e-03 -1.30451584e+00 -5.74776828e-01
-2.71364033e-01 -7.92885125e-01 1.17060602e+00 -3.53791043e-02
1.44240010e+00 -1.54662371e-01 -1.61661148e-01 7.57897794e-01
1.54591787e+00 -1.35565907e-01 9.09615099e-01 5.94328880e-01
5.23021936e-01 7.30060577e-01 5.41115642e-01 2.55598366e-01
1.41503379e-01 7.52806842e-01 8.07880104e-01 -2.65822887e-01
-4.25101146e-02 -3.29867393e-01 4.75793779e-01 4.15337801e-01
-2.29967564e-01 -1.21772997e-01 -7.85268307e-01 7.95416653e-01
-1.82053268e+00 -9.99166429e-01 2.70679206e-01 2.27462578e+00
7.87280679e-01 9.93364081e-02 1.53406799e-01 3.08292747e-01
4.92395729e-01 3.87066126e-01 -4.13667530e-01 -5.62838852e-01
1.91479996e-01 5.67968428e-01 1.71872869e-01 2.29018286e-01
-1.26319790e+00 9.83119249e-01 5.29846621e+00 7.36345530e-01
-1.24043286e+00 1.98748171e-01 5.41311085e-01 2.21136019e-01
-2.34302491e-01 2.26070452e-02 -5.30158818e-01 9.91702825e-02
8.17779601e-01 2.41282523e-01 4.21161920e-01 1.14526010e+00
-2.94530511e-01 -4.06517871e-02 -1.21524084e+00 8.12010348e-01
3.09176713e-01 -1.26651895e+00 1.99918032e-01 -1.13016233e-01
6.22132063e-01 -5.83285540e-02 1.50769085e-01 6.21583343e-01
2.18512192e-01 -1.02340424e+00 8.81038547e-01 4.62585151e-01
7.63355613e-01 -6.11116052e-01 5.13033271e-01 4.02951419e-01
-9.88435626e-01 -1.16629988e-01 -6.99091852e-01 -1.58589661e-01
-4.86989856e-01 4.82949108e-01 -5.74429393e-01 5.69947541e-01
8.86326432e-01 6.77712142e-01 -7.93107569e-01 8.26607168e-01
-5.34171462e-01 5.49070179e-01 1.17326900e-01 9.31230560e-02
3.72225106e-01 1.09735705e-01 3.22266847e-01 1.36767471e+00
4.11828533e-02 -4.91994061e-02 1.15774520e-01 8.54627430e-01
-1.13576353e-01 1.98069453e-01 -9.15954411e-01 -3.29738557e-02
5.02610467e-02 1.30471730e+00 -8.28476310e-01 -5.75449646e-01
-2.94830799e-01 1.04249191e+00 8.19241107e-01 3.32987249e-01
-7.95148075e-01 -4.38881487e-01 6.48671448e-01 1.00213848e-01
7.77614415e-01 -1.19852148e-01 2.34168395e-01 -1.18039870e+00
-3.15657631e-02 -1.07440221e+00 4.75091755e-01 -8.67740333e-01
-1.13722897e+00 5.30817986e-01 6.61628321e-02 -1.40250528e+00
-2.50044584e-01 -9.95349765e-01 -7.36879230e-01 4.08418059e-01
-1.74043453e+00 -1.03060269e+00 -5.36326587e-01 8.19739342e-01
6.30054355e-01 -1.94207340e-01 1.05193937e+00 5.88419512e-02
-4.02537555e-01 6.09621882e-01 1.76131740e-01 3.01702052e-01
6.94745421e-01 -1.30619800e+00 1.26677394e-01 8.88724744e-01
5.59874237e-01 6.82042480e-01 6.43520951e-01 -8.86955298e-03
-1.35205758e+00 -1.03067434e+00 3.94105196e-01 -4.08982754e-01
6.29368663e-01 -6.52156353e-01 -9.24708426e-01 8.25251758e-01
2.61085898e-01 4.85501885e-01 6.11803830e-01 7.80249201e-03
-9.11762714e-01 -1.56579286e-01 -8.80436659e-01 6.14856422e-01
1.00455177e+00 -7.28118122e-01 -7.15366066e-01 3.51794451e-01
5.63473105e-01 -3.12247127e-01 -5.68563044e-01 2.29353234e-01
4.92577374e-01 -1.21279991e+00 1.07860041e+00 -8.55492890e-01
6.23285055e-01 -2.95146406e-01 -3.07828039e-01 -1.35567153e+00
-4.10741150e-01 -3.43726128e-01 1.78105216e-02 1.08340251e+00
1.11985207e-01 -7.25464165e-01 4.56616670e-01 1.25977829e-01
-1.19208835e-01 -7.19480634e-01 -8.23727548e-01 -7.91654944e-01
1.12122983e-01 -4.22090381e-01 2.85363883e-01 1.04050756e+00
-1.50943249e-01 2.87905961e-01 -1.26980633e-01 1.40950263e-01
3.56539309e-01 1.72815248e-01 1.08917475e+00 -1.14058423e+00
-4.23867136e-01 -3.36188644e-01 -9.02942538e-01 -9.50137198e-01
3.98424655e-01 -1.02144146e+00 1.82239100e-01 -1.29975152e+00
1.43522203e-01 -2.77113169e-01 -6.35730386e-01 7.61413276e-01
-1.06301606e-01 3.71571183e-01 3.33490759e-01 1.20001912e-01
-6.96146369e-01 5.36220670e-01 9.38887000e-01 -4.99669403e-01
1.64047405e-01 -1.31659731e-01 -7.53164887e-01 5.34999669e-01
6.74431920e-01 -5.59839189e-01 -4.82106209e-01 -3.72648150e-01
6.06290065e-02 -4.34619278e-01 7.24826157e-01 -1.10422504e+00
1.93511099e-01 -6.32815957e-02 4.66230482e-01 -8.18075463e-02
2.06469476e-01 -8.39170992e-01 -2.49160171e-01 5.54275453e-01
-5.23633480e-01 -1.11722238e-01 3.73949289e-01 7.16734052e-01
-4.14554805e-01 -4.25240427e-01 9.31823730e-01 -2.38669142e-01
-8.20354998e-01 1.74894825e-01 -3.81436914e-01 -5.41690551e-02
1.02466154e+00 -2.42641926e-01 -3.62758964e-01 -3.25490326e-01
-5.53018749e-01 -1.29804775e-01 4.83133495e-01 3.62711817e-01
5.73810220e-01 -1.09278286e+00 -6.19805694e-01 4.82466340e-01
5.66675961e-01 1.57560542e-01 1.99570566e-01 8.41590583e-01
-4.88444626e-01 2.52884537e-01 -5.11983931e-01 -7.65930653e-01
-1.04403925e+00 9.94899929e-01 1.96962997e-01 -3.03076237e-01
-8.01603258e-01 9.19477940e-01 5.61843157e-01 -4.54912484e-01
4.07999754e-01 -3.52380931e-01 -3.39945495e-01 1.34436071e-01
7.15083063e-01 -9.26155131e-03 2.61841387e-01 -5.54458678e-01
-3.26101601e-01 5.21722794e-01 -2.06632510e-01 1.07428879e-01
1.50862396e+00 3.33919913e-01 -3.01443413e-03 4.34457570e-01
1.29430056e+00 -2.99476702e-02 -1.19649911e+00 -2.29704067e-01
-7.96859339e-02 -4.39534724e-01 2.23680399e-02 -3.39463264e-01
-9.85089540e-01 1.32882512e+00 6.86922312e-01 2.92350054e-01
1.21618831e+00 1.48258638e-03 2.28542656e-01 6.40269518e-01
3.58425617e-01 -8.90185952e-01 4.52844739e-01 3.50015342e-01
1.01554334e+00 -1.29843330e+00 1.08156092e-02 -1.17286682e-01
-6.04107141e-01 1.18453586e+00 5.50121784e-01 -5.49990416e-01
4.39732283e-01 -1.00692399e-01 1.32800728e-01 -1.23240858e-01
-8.06141019e-01 -5.81919312e-01 3.13061357e-01 7.96157360e-01
2.87564188e-01 -7.59192407e-02 1.42129481e-01 2.59691238e-01
4.30944450e-02 -3.76968026e-01 2.42446169e-01 1.00277352e+00
-3.92256826e-01 -9.26525235e-01 -3.27714533e-01 7.68714398e-02
-2.00597540e-01 -1.20922796e-01 -2.91642368e-01 8.41914237e-01
7.06449226e-02 5.70622563e-01 2.50485778e-01 -2.48203710e-01
4.43325907e-01 1.29750445e-01 5.52801013e-01 -8.77647460e-01
-6.60839796e-01 -2.61504263e-01 -2.25915298e-01 -7.35405743e-01
-5.35600781e-01 -4.22659636e-01 -1.10874677e+00 7.91755989e-02
-1.95473358e-01 -1.01123971e-03 5.16784787e-01 8.64071846e-01
1.70356259e-01 6.23090982e-01 7.13161051e-01 -1.21869457e+00
-6.17801785e-01 -8.76174510e-01 -3.98545682e-01 8.24607849e-01
3.79787028e-01 -9.20487761e-01 -4.78268147e-01 2.30708912e-01] | [9.724201202392578, 2.223015546798706] |
5fde8627-3785-4deb-aad7-9fd97ff91bea | s-clip-semi-supervised-vision-language-pre | 2305.14095 | null | https://arxiv.org/abs/2305.14095v1 | https://arxiv.org/pdf/2305.14095v1.pdf | S-CLIP: Semi-supervised Vision-Language Pre-training using Few Specialist Captions | Vision-language models, such as contrastive language-image pre-training (CLIP), have demonstrated impressive results in natural image domains. However, these models often struggle when applied to specialized domains like remote sensing, and adapting to such domains is challenging due to the limited number of image-text pairs available for training. To address this, we propose S-CLIP, a semi-supervised learning method for training CLIP that utilizes additional unpaired images. S-CLIP employs two pseudo-labeling strategies specifically designed for contrastive learning and the language modality. The caption-level pseudo-label is given by a combination of captions of paired images, obtained by solving an optimal transport problem between unpaired and paired images. The keyword-level pseudo-label is given by a keyword in the caption of the nearest paired image, trained through partial label learning that assumes a candidate set of labels for supervision instead of the exact one. By combining these objectives, S-CLIP significantly enhances the training of CLIP using only a few image-text pairs, as demonstrated in various specialist domains, including remote sensing, fashion, scientific figures, and comics. For instance, S-CLIP improves CLIP by 10% for zero-shot classification and 4% for image-text retrieval on the remote sensing benchmark, matching the performance of supervised CLIP while using three times fewer image-text pairs. | ['Jinwoo Shin', 'Kyungmin Lee', 'Minkyu Kim', 'Sangwoo Mo'] | 2023-05-23 | null | null | null | null | ['partial-label-learning', 'pseudo-label'] | ['methodology', 'miscellaneous'] | [ 7.92626381e-01 -1.26083001e-01 -3.02630037e-01 -4.33766127e-01
-1.31099939e+00 -6.84859216e-01 6.78620279e-01 2.56136596e-01
-6.43191099e-01 5.83027542e-01 -1.80563852e-01 -3.84699374e-01
5.05569205e-02 -4.38176692e-01 -1.03012311e+00 -5.66005290e-01
1.49219364e-01 4.25259441e-01 -9.73293558e-02 1.65650458e-03
5.11953458e-02 2.92189479e-01 -1.70792568e+00 3.14067125e-01
8.54816735e-01 1.09563565e+00 6.16838515e-01 3.94426644e-01
-2.23068178e-01 7.57486522e-01 -1.35482192e-01 -9.41353366e-02
5.34017086e-01 -5.09476781e-01 -7.50831544e-01 4.04187411e-01
8.53510559e-01 -3.55497837e-01 -1.60518080e-01 1.02056932e+00
3.14127475e-01 1.11544259e-01 6.47029817e-01 -1.27421498e+00
-8.64723682e-01 3.42130750e-01 -6.33052289e-01 -1.97252464e-02
1.66902155e-01 1.29076153e-01 1.13641918e+00 -1.27636480e+00
5.70101321e-01 1.07018936e+00 4.82722670e-01 4.80531573e-01
-1.35408628e+00 -6.25107110e-01 1.46087870e-01 1.27267316e-01
-1.73316610e+00 -3.43467563e-01 6.76898539e-01 -6.44038558e-01
6.01862311e-01 1.95801675e-01 3.77573222e-01 7.50051558e-01
-4.08823818e-01 7.74075806e-01 1.44068241e+00 -6.36714637e-01
1.91495091e-01 3.50891501e-01 -6.06237948e-02 6.35056376e-01
-3.63057405e-01 1.19589023e-01 -5.92327237e-01 7.61011988e-02
4.88097131e-01 7.49118906e-03 -3.57175291e-01 -2.41148517e-01
-1.34956300e+00 7.93591321e-01 6.83041334e-01 4.19451781e-02
-1.81971520e-01 -2.43267819e-01 1.63094342e-01 1.79253519e-01
6.62507296e-01 4.89876390e-01 -3.89053941e-01 5.29765427e-01
-1.37215137e+00 -1.48767782e-02 6.23293579e-01 9.81527448e-01
1.27736819e+00 -1.73426777e-01 -3.83934230e-01 8.99991930e-01
1.90045998e-01 1.05006945e+00 5.18334508e-02 -9.14164543e-01
4.93800819e-01 3.49281549e-01 7.94790760e-02 -7.83057511e-01
-8.87649730e-02 -2.65159875e-01 -9.53144312e-01 4.09295745e-02
3.13200504e-01 6.92599043e-02 -1.04441845e+00 1.63870585e+00
1.11353457e-01 3.55092227e-01 9.97714251e-02 9.67368364e-01
7.81940460e-01 9.57882524e-01 1.74859524e-01 -2.65387475e-01
1.19719064e+00 -1.29834318e+00 -3.40852231e-01 -6.05546415e-01
4.88064915e-01 -7.40605295e-01 1.30166662e+00 -2.67822873e-02
-6.78586662e-01 -6.94918871e-01 -8.68070900e-01 -1.98820964e-01
-5.44725001e-01 3.29796135e-01 8.64943042e-02 2.75327325e-01
-9.97967064e-01 2.68002898e-01 -3.16250175e-01 -3.14198434e-01
4.83236402e-01 3.91427763e-02 -4.60009128e-01 -4.79515284e-01
-1.03116119e+00 7.12063789e-01 2.62780339e-01 6.26201332e-02
-1.10655606e+00 -9.23002064e-01 -1.04587066e+00 1.64761677e-01
4.27390903e-01 -3.11919481e-01 9.49144065e-01 -1.20423186e+00
-1.19572043e+00 1.41327059e+00 -1.12201743e-01 -2.84064651e-01
5.25378108e-01 -3.32177170e-02 -1.90514758e-01 4.90995198e-01
4.83401656e-01 1.43831766e+00 1.05301142e+00 -1.56108916e+00
-4.04052675e-01 -2.33334824e-01 2.92755030e-02 4.53007251e-01
-2.25145742e-01 -8.65243673e-02 -6.46896541e-01 -5.46770692e-01
-1.16733581e-01 -9.76220369e-01 -3.11419427e-01 6.07751489e-01
-2.18381390e-01 2.97625903e-02 9.42697763e-01 -7.25888848e-01
4.42453772e-01 -2.39099693e+00 -7.87201151e-02 6.44158795e-02
-1.01570383e-01 3.71539176e-01 -6.44967616e-01 3.02248955e-01
-1.10961705e-01 -5.16494215e-02 -6.11261427e-01 -6.21977329e-01
-2.25765780e-01 3.25519085e-01 -4.78574663e-01 3.81936938e-01
5.16704798e-01 8.65274131e-01 -1.09699678e+00 -7.45828271e-01
4.01019216e-01 4.00737256e-01 -2.11642072e-01 4.07595038e-01
-5.57116449e-01 7.15232968e-01 -1.67581365e-01 6.53624475e-01
7.72141695e-01 -5.88972151e-01 -6.67473627e-03 -2.01289788e-01
-1.01869829e-01 -3.02689582e-01 -8.49568665e-01 1.77268898e+00
-5.08004904e-01 7.34680533e-01 1.03049569e-01 -1.17000937e+00
9.69001472e-01 1.10166430e-01 5.03528178e-01 -9.29818630e-01
-2.14082286e-01 2.86885172e-01 -5.67990005e-01 -6.83849335e-01
3.04048866e-01 -1.12864092e-01 -3.76359299e-02 4.21873838e-01
-8.98052976e-02 -5.19138634e-01 3.31594616e-01 2.49539539e-01
6.67698801e-01 2.49411821e-01 -3.36957164e-02 7.72729889e-02
6.60822749e-01 2.17926994e-01 2.12430552e-01 1.04311323e+00
-8.03995207e-02 8.89476180e-01 -1.18940435e-02 -1.52000099e-01
-1.17989373e+00 -1.11371756e+00 -2.88346976e-01 1.25204766e+00
2.70708442e-01 -5.92903197e-02 -5.99822342e-01 -7.10598946e-01
-9.41441879e-02 5.83413422e-01 -4.38220203e-01 1.10499576e-01
-1.84795454e-01 -2.67085224e-01 3.43409270e-01 2.47761011e-01
7.01647997e-01 -1.13483262e+00 -3.57946604e-01 -1.31629840e-01
-3.89548391e-01 -1.67947483e+00 -7.08625019e-01 3.16573799e-01
-5.21223903e-01 -9.45627570e-01 -9.19069469e-01 -1.08727646e+00
1.01487923e+00 6.68896735e-01 9.57015276e-01 1.54204682e-01
-3.37369621e-01 5.41122794e-01 -4.75076348e-01 -1.34649053e-01
-3.99149984e-01 -1.53553635e-01 -2.44764149e-01 2.90794581e-01
4.52294499e-02 -3.18302661e-01 -4.58575606e-01 2.02015921e-01
-1.17146492e+00 1.77320525e-01 9.31575716e-01 1.13753057e+00
8.19142401e-01 -3.36752087e-01 3.18481117e-01 -7.37683952e-01
-3.53079028e-02 -3.98309261e-01 -6.27637982e-01 6.25260592e-01
-5.08567810e-01 -2.36464664e-02 6.93407893e-01 -5.02590120e-01
-8.50547493e-01 5.14220655e-01 1.89796776e-01 -6.79715157e-01
-2.03848317e-01 8.20648730e-01 9.97502133e-02 -2.11125270e-01
7.00333595e-01 5.86898327e-01 2.92190313e-01 -2.69542634e-01
5.63893199e-01 9.30789173e-01 8.71838868e-01 -6.02701247e-01
9.32805717e-01 5.20627201e-01 -5.15426137e-02 -1.05785728e+00
-1.41543984e+00 -8.52853119e-01 -5.94431102e-01 -3.21278304e-01
9.12463486e-01 -1.32675731e+00 -2.44843364e-01 4.12668526e-01
-9.95470166e-01 -3.59529406e-01 -2.85441309e-01 4.99017894e-01
-4.20054913e-01 5.34256339e-01 -2.08820090e-01 -6.09444141e-01
-3.65555018e-01 -1.08163202e+00 1.51589286e+00 7.89480954e-02
4.25747186e-01 -8.26121509e-01 -2.99874306e-01 6.48701966e-01
2.67941236e-01 -1.02552183e-01 9.11836565e-01 -4.06269699e-01
-6.56668603e-01 -3.41791175e-02 -7.52953708e-01 8.54909241e-01
5.80823570e-02 -2.97592580e-01 -1.06960428e+00 -4.11762536e-01
-3.05686086e-01 -8.85724306e-01 9.08696115e-01 1.41835108e-01
1.18544090e+00 -1.18430287e-01 -2.14853510e-01 5.13293564e-01
1.59668529e+00 -1.88550517e-01 4.58001882e-01 1.79718673e-01
7.86498666e-01 7.40081191e-01 7.33110011e-01 2.25495577e-01
3.42862993e-01 6.41706526e-01 3.96070659e-01 -4.45387840e-01
-2.30033621e-01 -3.92083406e-01 2.76602000e-01 5.29436290e-01
5.10381043e-01 -8.78372863e-02 -9.67769027e-01 6.95287883e-01
-1.78097486e+00 -8.63925755e-01 1.13360912e-01 2.13231301e+00
1.06416535e+00 -2.60917723e-01 -2.42735595e-01 -1.03345260e-01
7.45026827e-01 2.06400916e-01 -7.42635310e-01 3.74269694e-01
-3.81587178e-01 1.49769098e-01 5.57085276e-01 4.50249195e-01
-1.23520064e+00 1.00800931e+00 6.08182240e+00 7.69219756e-01
-1.35843265e+00 1.77207083e-01 7.75996864e-01 2.42192969e-01
-2.84138024e-01 1.96145505e-01 -6.54375911e-01 4.10229087e-01
5.87180793e-01 1.13505195e-03 4.56839234e-01 6.78479791e-01
3.07058305e-01 -3.42593819e-01 -1.24605370e+00 1.22522688e+00
4.88440603e-01 -1.44365013e+00 3.52994829e-01 -2.02716038e-01
1.04311550e+00 3.56407076e-01 2.58943856e-01 2.30592728e-01
-2.01299578e-01 -9.43499446e-01 8.12813580e-01 3.67062598e-01
1.06937695e+00 -2.22960651e-01 5.33344746e-01 4.50435966e-01
-9.48647261e-01 -4.10966352e-02 -2.24633098e-01 6.16776571e-02
1.26766771e-01 5.03501117e-01 -6.93624139e-01 5.92691183e-01
7.07016468e-01 8.64349604e-01 -7.87389874e-01 1.01077807e+00
-1.79473251e-01 6.16598666e-01 -3.96402419e-01 2.68641025e-01
4.02591288e-01 -4.06424195e-01 3.27064484e-01 1.09633994e+00
2.38869622e-01 9.10639856e-03 6.33521199e-01 8.95915091e-01
-2.19447464e-01 2.61896372e-01 -6.95998132e-01 -1.16217516e-01
5.25532246e-01 1.33589888e+00 -5.96567869e-01 -3.22660983e-01
-4.26688135e-01 9.74957824e-01 2.20827058e-01 4.66605455e-01
-6.77734792e-01 -1.81351513e-01 3.75448950e-02 8.56090337e-02
3.00388604e-01 -2.40655690e-01 -1.36018053e-01 -9.63090181e-01
1.04464322e-01 -8.02170992e-01 2.60260224e-01 -1.25210261e+00
-1.32058167e+00 5.35267651e-01 1.06726564e-01 -1.45978999e+00
3.65583822e-02 -5.63395381e-01 -3.85863125e-01 7.62059748e-01
-2.00599384e+00 -1.56608653e+00 -6.04113400e-01 6.37768030e-01
5.20555437e-01 -1.23983905e-01 8.21468592e-01 4.56190616e-01
-2.04774573e-01 4.94534642e-01 2.50455827e-01 2.14608505e-01
9.01106894e-01 -9.31801081e-01 -3.50054651e-02 7.92809069e-01
4.12280053e-01 7.97994360e-02 4.30085331e-01 -2.17087850e-01
-1.11402416e+00 -1.45757937e+00 9.74895358e-01 -7.19980523e-02
5.36617517e-01 -3.27686876e-01 -8.69899690e-01 4.91431445e-01
1.23848736e-01 4.12776381e-01 4.90491778e-01 -4.01279300e-01
-6.18289471e-01 -2.45378807e-01 -9.98772621e-01 4.75236595e-01
7.05356359e-01 -1.03284740e+00 -3.08552921e-01 8.72495353e-01
8.12700152e-01 -2.63929725e-01 -5.39459705e-01 2.69894421e-01
1.59884393e-01 -4.47922051e-01 9.42686439e-01 -1.30651847e-01
6.84185445e-01 -5.72030723e-01 -3.88426453e-01 -1.01299834e+00
7.46558085e-02 -2.84931421e-01 5.46631873e-01 1.14592171e+00
4.57846999e-01 -1.25087589e-01 5.94730675e-01 1.78224728e-01
-8.48343149e-02 -3.30401808e-01 -7.21245110e-01 -9.36941147e-01
-7.94830769e-02 -2.69234598e-01 8.24777931e-02 1.15951455e+00
-5.08138597e-01 5.88602185e-01 -5.83824039e-01 3.41251642e-01
8.44444215e-01 2.33808205e-01 6.86449170e-01 -9.45925534e-01
-2.70150363e-01 -1.22307979e-01 -6.07435890e-02 -1.20057678e+00
6.12263203e-01 -1.21907282e+00 5.81261396e-01 -1.46368134e+00
4.03280973e-01 -6.79269791e-01 -6.74219150e-03 8.21661592e-01
-2.17006192e-01 6.77046120e-01 3.96363586e-01 3.95796061e-01
-7.54746079e-01 5.81471086e-01 1.25531888e+00 -8.14701438e-01
-8.44022110e-02 -2.17236072e-01 -4.23902094e-01 3.48948687e-01
4.54712182e-01 -5.89202523e-01 -3.49490702e-01 -5.58796644e-01
8.48018378e-02 8.29270855e-02 5.69286168e-01 -9.88808572e-01
3.69762301e-01 -2.45873079e-01 3.06654088e-02 -6.09413505e-01
3.75951678e-01 -8.55137408e-01 4.67045791e-02 1.82975888e-01
-5.39356053e-01 -3.19744200e-01 7.42849857e-02 6.47841692e-01
-4.23224598e-01 -3.95622611e-01 8.74475181e-01 -1.23545945e-01
-9.36138690e-01 2.49598697e-01 -3.41957621e-03 1.37674928e-01
9.78038430e-01 -7.73121342e-02 -3.36876273e-01 -4.62434381e-01
-6.01607084e-01 5.25448203e-01 2.61450589e-01 3.95657569e-01
6.68546796e-01 -1.10294306e+00 -9.98766005e-01 2.62647092e-01
6.13304436e-01 2.72031367e-01 2.92226225e-01 7.65358210e-01
-4.50334489e-01 1.76758960e-01 -1.94143370e-01 -9.92093861e-01
-1.15672016e+00 5.31464279e-01 2.14520797e-01 -1.28064334e-01
-4.63402241e-01 6.69524729e-01 2.21822992e-01 -5.61040878e-01
3.21028560e-01 -1.51419416e-02 -7.11243879e-03 3.94533947e-02
5.16035616e-01 -6.07285798e-02 -7.47524500e-02 -8.50100517e-01
-2.56042778e-01 8.15801799e-01 -1.78892955e-01 -2.54236251e-01
1.30607677e+00 -2.74062663e-01 -2.70174891e-01 3.96294445e-01
1.41456282e+00 -2.87692755e-01 -1.54330552e+00 -7.49232829e-01
-1.16412744e-01 -3.76252443e-01 1.63079754e-01 -8.69337916e-01
-9.58954871e-01 8.90401602e-01 7.04327762e-01 -1.77779600e-01
1.13712442e+00 1.36723295e-01 5.52356958e-01 7.14812160e-01
2.51181036e-01 -9.12698627e-01 3.71570110e-01 6.18928730e-01
8.03218603e-01 -1.70300293e+00 -2.05383971e-02 -3.04558218e-01
-7.14779854e-01 7.60331392e-01 4.56748307e-01 2.12644711e-01
7.01783121e-01 -1.16216443e-01 3.61703694e-01 1.67080499e-02
-3.93064201e-01 -4.84045088e-01 4.11621898e-01 6.95914090e-01
-5.45383431e-02 -1.15251288e-01 5.85886054e-02 -4.52577360e-02
2.08964422e-01 3.73973846e-02 3.12579840e-01 9.14036512e-01
-2.57348865e-01 -9.55460310e-01 -3.65643233e-01 2.76146919e-01
-6.43447191e-02 -4.55440909e-01 -2.28338793e-01 3.34578872e-01
1.08578250e-01 1.05915368e+00 2.83028543e-01 -1.54222503e-01
6.22316450e-02 -2.66330894e-02 4.20000345e-01 -8.37820172e-01
-4.09678280e-01 -5.55259958e-02 -2.30445161e-01 -1.70429036e-01
-9.90564406e-01 -4.27652985e-01 -1.04250324e+00 1.69540629e-01
-3.09623927e-01 2.27850363e-01 7.34540045e-01 1.19536042e+00
2.83935934e-01 -2.74692681e-02 9.46095526e-01 -1.10739636e+00
-5.93889832e-01 -7.78676689e-01 -6.54184341e-01 6.61491334e-01
4.44926828e-01 -4.46170270e-01 -5.74043870e-01 5.68141401e-01] | [10.741477966308594, 1.3167554140090942] |
8eb341ab-d2b4-4532-951c-07c69ea5d0a2 | exploring-modality-agnostic-representations | 2106.01149 | null | https://arxiv.org/abs/2106.01149v1 | https://arxiv.org/pdf/2106.01149v1.pdf | Exploring modality-agnostic representations for music classification | Music information is often conveyed or recorded across multiple data modalities including but not limited to audio, images, text and scores. However, music information retrieval research has almost exclusively focused on single modality recognition, requiring development of separate models for each modality. Some multi-modal works require multiple coexisting modalities given to the model as inputs, constraining the use of these models to the few cases where data from all modalities are available. To the best of our knowledge, no existing model has the ability to take inputs from varying modalities, e.g. images or sounds, and classify them into unified music categories. We explore the use of cross-modal retrieval as a pretext task to learn modality-agnostic representations, which can then be used as inputs to classifiers that are independent of modality. We select instrument classification as an example task for our study as both visual and audio components provide relevant semantic information. We train music instrument classifiers that can take both images or sounds as input, and perform comparably to sound-only or image-only classifiers. Furthermore, we explore the case when there is limited labeled data for a given modality, and the impact in performance by using labeled data from other modalities. We are able to achieve almost 70% of best performing system in a zero-shot setting. We provide a detailed analysis of experimental results to understand the potential and limitations of the approach, and discuss future steps towards modality-agnostic classifiers. | ['Juan P. Bello', 'Magdalena Fuentes', 'Ho-Hsiang Wu'] | 2021-06-02 | null | null | null | null | ['music-classification', 'music-information-retrieval'] | ['music', 'music'] | [ 6.54603302e-01 -4.81629163e-01 -2.75299609e-01 -1.40672937e-01
-1.16808093e+00 -1.06205821e+00 8.07155967e-01 8.53980705e-02
-5.60793400e-01 4.05155331e-01 2.97631979e-01 3.47008556e-02
-3.76737386e-01 -4.65049356e-01 -4.64533091e-01 -4.85104948e-01
1.52291059e-01 3.32662046e-01 2.08826184e-01 -7.66870603e-02
3.25322360e-01 4.57409382e-01 -2.22942162e+00 7.92796731e-01
1.88935220e-01 1.09965777e+00 1.93804368e-01 9.32698190e-01
-2.77436316e-01 6.50423169e-01 -7.16799140e-01 -9.89043638e-02
1.65430903e-01 -6.07518852e-01 -9.48849559e-01 3.90059389e-02
7.83863604e-01 -7.27367252e-02 -1.91787392e-01 7.01583147e-01
8.45300734e-01 1.81434602e-01 8.13064158e-01 -1.23418319e+00
-5.43911934e-01 6.74692690e-01 -2.94918925e-01 1.55699775e-01
9.27670360e-01 5.58822267e-02 1.09289038e+00 -7.16073215e-01
6.58509910e-01 1.07723773e+00 3.96824867e-01 6.46251321e-01
-1.34783959e+00 -5.99755824e-01 -4.59554382e-02 3.77559781e-01
-1.34297383e+00 -7.38374710e-01 8.46937716e-01 -3.43731523e-01
7.93836057e-01 4.36982989e-01 4.72156137e-01 1.22636127e+00
-3.78771603e-01 1.01516247e+00 1.38031328e+00 -8.56824696e-01
2.53170859e-02 2.70404071e-01 -8.00553337e-02 1.38588533e-01
-1.75868303e-01 1.03441142e-01 -1.10718107e+00 -1.22918338e-01
6.52795255e-01 -2.47548401e-01 -3.08347315e-01 -3.16650271e-01
-1.47717154e+00 4.60859418e-01 2.21307531e-01 6.48478031e-01
-2.15833321e-01 1.03857644e-01 4.66324419e-01 6.97131276e-01
9.21517462e-02 6.25498533e-01 -3.33350003e-01 -4.02901649e-01
-1.24005353e+00 6.90784752e-02 8.11379850e-01 5.90366244e-01
4.49006021e-01 -1.22993499e-01 -1.46627069e-01 1.17026091e+00
6.04460835e-02 5.04807532e-01 5.52369475e-01 -9.24731612e-01
2.74384230e-01 3.10946167e-01 -1.53256178e-01 -6.91008687e-01
-4.53209728e-01 -1.98122472e-01 -2.49809846e-01 1.56244874e-01
5.94122469e-01 2.21981108e-01 -9.53458905e-01 1.86350000e+00
-4.12716903e-02 1.24468915e-01 1.28057703e-01 1.13905883e+00
1.17308068e+00 2.51291543e-01 1.60491809e-01 -1.46919996e-01
1.46889544e+00 -6.93664551e-01 -4.74953294e-01 1.01949088e-02
3.46646309e-01 -1.18445897e+00 1.26342690e+00 6.10531926e-01
-9.63492632e-01 -5.85118890e-01 -9.56619859e-01 -2.00837385e-02
-6.44357860e-01 1.45920038e-01 5.33034503e-01 6.01236284e-01
-8.21916461e-01 4.79476243e-01 -3.72395962e-01 -7.79061496e-01
2.16836378e-01 3.28517556e-01 -5.20199418e-01 -1.58833757e-01
-1.09157658e+00 8.57428312e-01 3.02267075e-01 -3.91514093e-01
-7.46091902e-01 -5.28600991e-01 -5.79951346e-01 -1.06228016e-01
3.25434625e-01 -7.27395475e-01 1.33926570e+00 -1.48499477e+00
-1.43596971e+00 1.05474710e+00 -4.72289696e-02 -5.69442578e-04
1.80242419e-01 3.93167138e-02 -6.47178173e-01 5.22356868e-01
-6.49279803e-02 7.99058020e-01 8.31997931e-01 -1.44870532e+00
-5.89847624e-01 -2.35107601e-01 3.47676784e-01 4.80242342e-01
-5.80678821e-01 2.42232859e-01 -6.39344096e-01 -6.86007082e-01
1.34457439e-01 -1.10740244e+00 5.39464056e-01 -4.74523045e-02
-3.46875489e-01 6.83554560e-02 7.05062866e-01 -3.00794929e-01
1.04057479e+00 -2.26607037e+00 3.47342312e-01 1.32337391e-01
-4.51915920e-01 -1.93037838e-01 -5.17098844e-01 6.45940006e-01
-1.43204257e-01 8.62326473e-03 -1.47360310e-01 -3.15721035e-01
1.93029359e-01 1.10401914e-01 -1.44183531e-01 1.60097882e-01
-9.84762609e-03 6.34297311e-01 -6.73395395e-01 -5.34788549e-01
2.04866230e-01 4.73373413e-01 -2.60013998e-01 1.10027604e-02
-1.58067644e-02 4.93305206e-01 -1.31755218e-01 9.12294805e-01
2.02078536e-01 -8.98186024e-03 -1.04424551e-01 -3.64512354e-01
1.21890888e-01 2.17960745e-01 -1.46909976e+00 2.14801431e+00
-6.32176399e-01 7.45467663e-01 1.77368838e-02 -8.89272511e-01
5.99277675e-01 7.05901384e-01 5.58423102e-01 -7.42185891e-01
5.87456524e-02 3.30607891e-01 1.56436175e-01 -5.74414492e-01
4.08605278e-01 -5.72745264e-01 -3.03820014e-01 7.07786083e-01
3.78838956e-01 -2.39622414e-01 2.49470055e-01 2.22061574e-02
1.02655351e+00 1.14653803e-01 -1.69066265e-02 2.73662359e-01
3.90301704e-01 1.08084969e-01 -5.75994141e-02 1.01166511e+00
-6.34639040e-02 9.74050343e-01 -9.63102560e-03 1.67308122e-01
-8.20335805e-01 -1.12845623e+00 -3.99952561e-01 1.72024739e+00
1.38389081e-01 -2.93017536e-01 -1.43528327e-01 -4.20686126e-01
2.08009537e-02 4.31230754e-01 -3.93989265e-01 -1.09511487e-01
2.13457309e-02 -4.75847304e-01 9.73886669e-01 4.24523205e-01
1.40696272e-01 -1.31241751e+00 -6.25724792e-01 -1.03238516e-01
-1.67637959e-01 -9.39267814e-01 -7.16559589e-02 3.68324846e-01
-7.87404120e-01 -1.07725251e+00 -9.27205205e-01 -6.83471262e-01
6.81648403e-02 1.70174971e-01 1.08818650e+00 -1.55398935e-01
-2.65491992e-01 1.23844981e+00 -4.67013836e-01 -4.75454271e-01
-2.91664481e-01 1.46074727e-01 1.47204027e-01 9.81625542e-02
3.59899193e-01 -5.63738167e-01 -5.16906679e-01 2.81986922e-01
-1.12918174e+00 -1.37750760e-01 6.13766134e-01 7.23692775e-01
3.97256285e-01 -1.78162336e-01 7.11382985e-01 -6.33961082e-01
6.54593468e-01 -3.99360895e-01 2.23301023e-01 2.12117627e-01
-2.37665981e-01 -2.32748166e-01 2.82655060e-01 -9.06740725e-01
-9.01793301e-01 2.38827288e-01 -5.12355764e-04 -6.64440215e-01
-6.11037970e-01 7.90321767e-01 -9.45301354e-02 -2.27258936e-01
8.16529989e-01 1.18277580e-01 -8.90439784e-04 -6.73705816e-01
5.00530124e-01 9.50371146e-01 6.33405924e-01 -6.69296086e-01
5.14763176e-01 4.10101086e-01 -1.75788812e-02 -9.78649318e-01
-6.83502734e-01 -7.27488339e-01 -6.73949957e-01 -4.65171039e-01
5.27068794e-01 -9.09241796e-01 -5.22036731e-01 3.16314250e-01
-6.35919213e-01 2.51116138e-02 -4.58776236e-01 7.40708768e-01
-7.65266478e-01 1.97177172e-01 -5.97153604e-01 -8.51807058e-01
4.22747321e-02 -8.89809489e-01 1.23707819e+00 1.92720175e-01
-5.23013949e-01 -8.36237729e-01 7.49811232e-02 5.09974599e-01
3.71006727e-01 -6.92215115e-02 8.83252800e-01 -9.00593162e-01
-3.53223622e-01 -4.18533325e-01 -3.29699330e-02 1.51045725e-01
1.22534290e-01 -1.75367013e-01 -1.62895536e+00 -2.22651765e-01
-2.92176455e-01 -7.99889088e-01 1.01121521e+00 2.09778637e-01
9.10899222e-01 2.72055537e-01 -1.48372546e-01 2.18188822e-01
1.26488233e+00 8.16344842e-02 4.61391449e-01 5.20223618e-01
6.17166519e-01 7.32457697e-01 5.45873106e-01 1.59736514e-01
1.43587425e-01 9.52395320e-01 3.08890283e-01 -3.05520650e-02
-4.19361532e-01 -9.01603997e-02 2.90469408e-01 6.77072823e-01
-2.12669875e-02 -2.10954785e-01 -7.14034021e-01 6.35505378e-01
-1.58474314e+00 -1.18157113e+00 2.26924121e-01 2.34502053e+00
8.69144917e-01 -1.67778537e-01 4.07783866e-01 6.44198716e-01
5.59509277e-01 3.56832780e-02 -5.17090857e-01 -2.43010268e-01
-3.17900509e-01 3.80644441e-01 3.88159752e-02 6.16455749e-02
-1.29749846e+00 5.86520791e-01 6.92112207e+00 8.23613167e-01
-1.39998472e+00 2.82694888e-03 -3.19861062e-02 -6.97394848e-01
-2.00397789e-01 -3.06919347e-02 -1.19081914e-01 2.19248518e-01
1.00226235e+00 2.43055806e-01 6.03355348e-01 4.18001890e-01
-9.83573645e-02 -2.45745495e-01 -1.30816317e+00 1.30087698e+00
3.76408815e-01 -8.82811368e-01 6.40165284e-02 -1.94483086e-01
4.61142987e-01 -7.48262405e-02 8.07126984e-02 3.31172884e-01
-3.54930341e-01 -1.04088867e+00 9.44658458e-01 7.19893754e-01
8.72022271e-01 -5.80529749e-01 4.78333145e-01 1.66742712e-01
-1.08804095e+00 -1.93306863e-01 1.74374729e-01 -6.61743954e-02
-3.87552455e-02 -1.24737985e-01 -5.01924336e-01 6.49889112e-01
7.34229386e-01 7.30623662e-01 -7.21192777e-01 1.24133551e+00
1.12006642e-01 5.87419510e-01 -5.97029746e-01 1.55720562e-01
-2.09782287e-01 3.17857891e-01 6.17946506e-01 1.25740778e+00
4.69328314e-01 -1.31638467e-01 3.03122938e-01 3.94922018e-01
1.16168775e-01 3.24449122e-01 -7.19384372e-01 -2.87429422e-01
5.64282179e-01 1.11845911e+00 -7.82844782e-01 -1.91720441e-01
-6.01696193e-01 9.07088518e-01 -1.55419841e-01 4.25134391e-01
-3.19540679e-01 -3.13582897e-01 3.76030266e-01 -3.35244797e-02
1.27155423e-01 1.26007944e-01 -3.83411139e-01 -1.16279137e+00
-7.64846727e-02 -1.01157093e+00 7.42917955e-01 -1.13574362e+00
-1.53305292e+00 2.52992302e-01 1.29017651e-01 -1.55709994e+00
-3.70360851e-01 -6.14384294e-01 -3.65696937e-01 8.86421800e-01
-1.34357250e+00 -1.42566204e+00 -1.95933759e-01 8.71189654e-01
2.75073856e-01 -3.70950878e-01 1.19838345e+00 5.15745103e-01
-7.49550387e-02 5.76344192e-01 -5.50460629e-02 6.90515637e-02
1.25562429e+00 -1.06767440e+00 -6.36612773e-01 3.63340497e-01
8.27497840e-01 5.80057144e-01 6.70755863e-01 -1.69082269e-01
-1.63751733e+00 -4.02375191e-01 5.63266933e-01 -5.73565602e-01
6.89494014e-01 -4.21056263e-02 -7.67715931e-01 2.99886733e-01
2.08728388e-01 -2.47075006e-01 1.20076549e+00 5.51486075e-01
-5.90678096e-01 9.27109867e-02 -8.01271379e-01 3.50765795e-01
8.90318334e-01 -1.23013902e+00 -7.12321758e-01 -1.93018671e-02
1.54460534e-01 -2.44871706e-01 -1.01233602e+00 4.33410645e-01
9.64331508e-01 -7.82763243e-01 1.02180481e+00 -6.01604640e-01
3.69655192e-01 -4.19990599e-01 -5.18572927e-01 -1.19183302e+00
7.07641095e-02 7.87595063e-02 -1.34069761e-02 1.52077520e+00
5.09410620e-01 -2.05239490e-01 4.18930411e-01 1.38572961e-01
6.05497733e-02 -2.09307760e-01 -9.76195574e-01 -7.20332801e-01
-9.14394110e-02 -8.99105310e-01 3.85163039e-01 1.16181481e+00
1.65742531e-01 4.88448530e-01 -3.44232351e-01 -6.18091449e-02
1.12398520e-01 6.16090119e-01 7.46464968e-01 -1.33405423e+00
-5.91302991e-01 -6.32732987e-01 -6.42802179e-01 -6.41686141e-01
1.25395656e-01 -1.16330278e+00 -1.93812549e-02 -1.50610495e+00
4.80076462e-01 -3.01875025e-01 -6.30859613e-01 6.24167025e-01
1.57733947e-01 9.92456555e-01 6.24226391e-01 4.46157008e-01
-5.28336287e-01 1.65104136e-01 1.16030133e+00 -3.49733829e-01
-2.31298342e-01 -9.02242512e-02 -7.48472095e-01 5.99149942e-01
5.32184958e-01 -2.76849568e-01 -4.92951810e-01 -2.47661769e-01
1.82814687e-01 1.01847291e-01 5.46757698e-01 -1.14907897e+00
2.02702269e-01 -2.58555394e-02 4.93900239e-01 -3.09825748e-01
9.10714447e-01 -9.27329540e-01 2.62417406e-01 -1.93174422e-01
-6.71301663e-01 -4.01888192e-01 3.79942387e-01 4.68009174e-01
-5.35110950e-01 -4.53917772e-01 4.55719739e-01 -1.11793309e-01
-8.97439182e-01 -1.48305058e-01 -3.73006076e-01 -9.65819433e-02
6.07363403e-01 -3.90385449e-01 -2.82369554e-01 -6.11644268e-01
-1.20820892e+00 -1.82342008e-01 6.05973482e-01 7.74497986e-01
3.63277614e-01 -1.51691854e+00 -4.57838386e-01 -2.04883665e-02
5.99016368e-01 -7.11304665e-01 2.80797273e-01 7.66024768e-01
2.64725834e-02 3.05824846e-01 -3.11347157e-01 -8.92883301e-01
-1.65695417e+00 6.01281703e-01 1.37807220e-01 3.26186240e-01
-2.75259525e-01 5.16846776e-01 -1.49561509e-01 -4.43588883e-01
4.28836584e-01 2.13758111e-01 -3.84822607e-01 6.57212913e-01
4.24712151e-01 -1.08387005e-02 1.30723253e-01 -8.99153888e-01
-3.63720477e-01 7.69764662e-01 2.99558043e-01 -6.91585422e-01
1.03485107e+00 -1.08439818e-01 1.21017456e-01 1.28395879e+00
1.08219910e+00 -8.71922150e-02 -6.64672494e-01 -2.37514675e-01
-4.10280786e-02 -4.37926203e-01 7.24738762e-02 -1.20880723e+00
-9.06921804e-01 1.01081431e+00 1.03362954e+00 3.76228034e-01
1.54365301e+00 2.61381835e-01 2.35419512e-01 2.79212862e-01
3.17697048e-01 -1.04175627e+00 1.98084876e-01 3.87636006e-01
7.69494116e-01 -1.19840574e+00 -1.16427362e-01 -4.15805504e-02
-6.50254428e-01 1.16028440e+00 2.53759950e-01 2.06488118e-01
4.62057352e-01 2.89405901e-02 4.18737352e-01 -1.53198123e-01
-6.60493195e-01 -7.34050393e-01 8.15031886e-01 5.82786798e-01
7.94948459e-01 -5.95077872e-02 -1.20031998e-01 5.26637495e-01
-2.34019980e-01 1.24589831e-01 2.35648021e-01 1.24345219e+00
-2.61603922e-01 -1.29816997e+00 -7.22126782e-01 5.91784656e-01
-5.98991513e-01 -9.35597531e-03 -7.53896296e-01 7.28329957e-01
1.52990922e-01 1.17512274e+00 5.19831777e-02 -4.37920243e-01
4.76326019e-01 6.79923058e-01 8.32646012e-01 -6.48131669e-01
-7.26655126e-01 3.73711497e-01 1.92852482e-01 -2.30258763e-01
-9.85165715e-01 -7.90998876e-01 -8.16027105e-01 -2.30675545e-02
-2.90461779e-01 -8.24785680e-02 7.56321132e-01 9.63108480e-01
1.73871785e-01 6.78022683e-01 2.95263678e-01 -1.23947465e+00
-1.42030209e-01 -1.05390859e+00 -6.71158552e-01 7.79513836e-01
3.32548559e-01 -7.83940971e-01 -3.93930078e-01 2.30450749e-01] | [15.498393058776855, 5.097670555114746] |
b4cd2cf5-2679-4b23-89e9-12e6a4d107ea | voice-command-generation-using-progressive | 1903.07395 | null | http://arxiv.org/abs/1903.07395v1 | http://arxiv.org/pdf/1903.07395v1.pdf | Voice command generation using Progressive Wavegans | Generative Adversarial Networks (GANs) have become exceedingly popular in a
wide range of data-driven research fields, due in part to their success in
image generation. Their ability to generate new samples, often from only a
small amount of input data, makes them an exciting research tool in areas with
limited data resources. One less-explored application of GANs is the synthesis
of speech and audio samples. Herein, we propose a set of extensions to the
WaveGAN paradigm, a recently proposed approach for sound generation using GANs.
The aim of these extensions - preprocessing, Audio-to-Audio generation, skip
connections and progressive structures - is to improve the human likeness of
synthetic speech samples. Scores from listening tests with 30 volunteers
demonstrated a moderate improvement (Cohen's d coefficient of 0.65) in human
likeness using the proposed extensions compared to the original WaveGAN
approach. | ['Björn Schuller', 'NIcholas Cummins', 'Thomas Wiest', 'Simone Hantke', 'Alice Baird', 'Judith Dineley'] | 2019-03-13 | null | null | null | null | ['audio-generation'] | ['audio'] | [ 4.62349504e-01 4.10569817e-01 2.61735171e-01 -1.46342874e-01
-8.73741984e-01 -4.11565244e-01 7.95304120e-01 -4.56295073e-01
-1.12070227e-02 1.16697001e+00 5.77376604e-01 1.95973497e-02
3.10052246e-01 -8.48787129e-01 -5.90617180e-01 -7.57771432e-01
1.69153184e-01 2.09920615e-01 7.13780820e-02 -2.63584822e-01
5.06678596e-02 3.58186722e-01 -1.54060411e+00 3.07378203e-01
8.16668272e-01 8.59685123e-01 -1.96729768e-02 8.80642176e-01
2.28727534e-02 6.68162763e-01 -9.34499145e-01 -6.08158648e-01
3.30950677e-01 -1.05181038e+00 -4.85557437e-01 -2.84596551e-02
2.48826742e-01 -1.86128199e-01 -2.67238140e-01 8.22728872e-01
1.02488422e+00 1.62188828e-01 5.34998894e-01 -1.30543303e+00
-6.88768208e-01 5.14209867e-01 -3.61823738e-01 -1.69324987e-02
4.30777341e-01 3.31265599e-01 8.23709846e-01 -6.39715672e-01
5.05790353e-01 1.09406900e+00 6.70951843e-01 1.05530596e+00
-1.29211986e+00 -9.95090246e-01 -6.37111843e-01 -3.70497368e-02
-1.18057382e+00 -5.76230288e-01 9.83539760e-01 -2.37837806e-01
6.27880991e-01 3.94403994e-01 7.67998874e-01 1.37391090e+00
1.82958290e-01 5.73110163e-01 1.43807745e+00 -4.89998877e-01
4.08852607e-01 2.34429926e-01 -9.58951294e-01 1.54240549e-01
-3.06494087e-02 3.94002825e-01 -7.15280712e-01 1.15964733e-01
8.81019294e-01 -4.64585215e-01 -1.84438437e-01 5.80887273e-02
-1.18020022e+00 9.80921745e-01 4.38798368e-01 3.86844814e-01
-6.27743840e-01 3.45243037e-01 3.27766001e-01 3.92760038e-01
5.76268196e-01 5.52692652e-01 1.21007085e-01 -4.09943998e-01
-1.12595129e+00 4.71605778e-01 4.87546355e-01 8.32878590e-01
1.57776475e-01 8.95554364e-01 -3.61587495e-01 7.37067223e-01
1.81946486e-01 5.93180418e-01 8.66098046e-01 -7.62758970e-01
2.50576586e-01 9.22361389e-02 -1.09217748e-01 -8.57331336e-01
-5.14876619e-02 -6.00980282e-01 -1.09846365e+00 4.18318003e-01
1.97100118e-01 -4.94113773e-01 -9.87511992e-01 1.88641047e+00
2.64067233e-01 3.94291162e-01 2.42071584e-01 6.26482546e-01
9.44540083e-01 8.82331014e-01 -1.60812102e-02 -6.89131171e-02
9.91752744e-01 -1.01113403e+00 -7.30318546e-01 -8.76617879e-02
-1.86140597e-01 -1.04786730e+00 1.13023484e+00 5.04876137e-01
-1.31210911e+00 -8.97457898e-01 -9.77813303e-01 3.38209361e-01
-1.91167399e-01 -1.62658468e-01 5.46489418e-01 1.16015267e+00
-1.25617886e+00 3.65021646e-01 -3.94581020e-01 -2.55594462e-01
5.17740190e-01 3.64006668e-01 -3.21620256e-01 2.07969204e-01
-1.14481819e+00 5.37498593e-01 1.93623126e-01 -1.47763476e-01
-1.05819702e+00 -6.97952926e-01 -5.48851371e-01 -1.14128239e-01
8.45085755e-02 -8.77193511e-01 1.16524684e+00 -9.40425575e-01
-1.89239132e+00 5.97575247e-01 1.89560026e-01 -7.91911364e-01
6.18621290e-01 -1.29341513e-01 -6.22751653e-01 1.45083800e-01
-1.09907225e-01 1.17124212e+00 1.30128574e+00 -1.12995374e+00
-4.30779785e-01 5.23305833e-02 -2.37906605e-01 1.56352594e-01
-2.14094996e-01 -1.79650977e-01 6.24087974e-02 -1.25764120e+00
-2.96407610e-01 -9.39336181e-01 -1.86000735e-01 -2.68000394e-01
-4.85536128e-01 1.37814894e-01 8.71473491e-01 -8.15867484e-01
7.97231913e-01 -1.93278837e+00 2.17464548e-02 9.95634217e-03
1.47605911e-02 5.18241644e-01 -2.07905412e-01 5.96397936e-01
-1.55095696e-01 1.69932768e-01 -2.71782339e-01 -2.65682727e-01
-9.30540189e-02 3.02305017e-02 -4.26677048e-01 2.10833531e-02
2.57667422e-01 8.83539319e-01 -7.15183079e-01 -1.06575660e-01
2.52851605e-01 6.41234994e-01 -5.94205976e-01 4.19785529e-01
-2.24746540e-01 9.40938771e-01 -1.23122320e-01 3.77784699e-01
4.82484430e-01 4.60198164e-01 -4.60257083e-01 1.34647544e-02
1.97530225e-01 6.14193417e-02 -1.03329217e+00 1.56194353e+00
-5.71410954e-01 7.89663970e-01 -2.74899393e-01 -7.06985354e-01
1.06556964e+00 8.52954268e-01 2.58723855e-01 -5.74406445e-01
6.54824451e-02 2.01056972e-01 3.49144131e-01 -4.82673943e-01
3.03356528e-01 -5.59468508e-01 1.32091150e-01 4.72031832e-01
2.11529762e-01 -6.86464787e-01 1.28342956e-01 -9.22293514e-02
9.43289995e-01 -5.49207404e-02 3.04366052e-01 1.17359951e-01
5.20533144e-01 -3.25727284e-01 3.03886294e-01 4.75293338e-01
8.67193416e-02 1.06347764e+00 2.57850647e-01 -1.65206641e-01
-1.43051755e+00 -1.07535219e+00 2.93683767e-01 4.69131202e-01
-5.11093020e-01 -5.67039028e-02 -1.20896876e+00 -4.88877803e-01
-4.29644912e-01 6.60157084e-01 -5.38262904e-01 -3.13770771e-01
-4.19452816e-01 -4.51381147e-01 9.53360796e-01 4.40945119e-01
8.39788258e-01 -1.71036911e+00 -5.71765304e-01 3.32228780e-01
-1.85641021e-01 -1.02071595e+00 -4.23655927e-01 -2.20839545e-01
-8.39257181e-01 -4.55372632e-01 -1.32505023e+00 -6.49663806e-01
5.16009867e-01 -1.62549958e-01 9.89086807e-01 -2.99028367e-01
-4.25312370e-01 2.75651783e-01 -6.33423865e-01 -8.09148073e-01
-9.15236294e-01 -1.49193155e-02 1.06157601e-01 1.54452071e-01
-1.32365152e-01 -8.58649373e-01 -7.17321992e-01 1.67107344e-01
-1.23137033e+00 1.48569718e-01 7.76318789e-01 9.33130205e-01
4.66923803e-01 1.63303077e-01 1.12457621e+00 -9.26274478e-01
1.11112261e+00 -3.92720342e-01 -2.81538218e-01 -1.94676191e-01
-4.07179594e-01 -1.55991778e-01 7.97297478e-01 -5.48423946e-01
-1.06348264e+00 -1.78451434e-01 -4.97158021e-01 -3.12336892e-01
-3.70542467e-01 2.65915543e-01 -2.33777821e-01 -1.48135111e-01
7.10285246e-01 3.74519616e-01 1.37754738e-01 -2.48767510e-01
3.40642691e-01 7.58687794e-01 7.62288511e-01 -3.84175420e-01
1.12009060e+00 1.84788302e-01 3.14099006e-02 -9.44976389e-01
-2.35475361e-01 4.59350832e-03 -2.14347348e-01 -1.90721482e-01
9.27815259e-01 -6.84025586e-01 -3.36431295e-01 6.92137301e-01
-9.83946800e-01 -3.66214484e-01 -7.79154360e-01 5.73705256e-01
-7.24197447e-01 -1.36019751e-01 -2.58523673e-01 -8.46249044e-01
-7.59021282e-01 -1.02544880e+00 7.71736205e-01 5.24876416e-01
-4.22513127e-01 -9.24701869e-01 1.59895301e-01 3.73804450e-01
8.22312832e-01 6.03221893e-01 8.11600089e-01 -5.24522007e-01
-3.14375460e-01 -3.52373123e-01 1.48696765e-01 7.40784049e-01
3.79050583e-01 -1.78008050e-01 -1.15205097e+00 -3.33243459e-01
5.83292358e-02 -3.68617147e-01 3.95100206e-01 4.89991099e-01
8.31989050e-01 -3.88526440e-01 2.31544256e-01 4.39012676e-01
1.17418790e+00 6.81943238e-01 9.66219425e-01 -2.76571009e-02
3.26207966e-01 3.57942283e-01 3.21383417e-01 4.32855189e-01
-3.82639408e-01 7.29791224e-01 4.67658192e-01 -2.70637929e-01
-7.84534276e-01 -5.27781963e-01 3.23829383e-01 7.76137173e-01
-2.79808193e-01 -4.90411848e-01 -4.77601200e-01 5.65265834e-01
-1.05496609e+00 -1.15953422e+00 1.09248072e-01 2.18144608e+00
7.78468549e-01 6.58344850e-02 4.42779660e-01 5.45038581e-01
7.35628605e-01 6.40062392e-02 -5.65030038e-01 -5.85143924e-01
-8.98137316e-02 9.49949324e-01 7.08644018e-02 1.69409260e-01
-6.87013090e-01 8.62928867e-01 6.74747133e+00 8.92066896e-01
-1.37646544e+00 5.80778383e-02 6.97813034e-01 1.12905771e-01
-3.50727499e-01 -3.51544052e-01 -3.55310559e-01 6.29495859e-01
1.14540946e+00 -2.48994142e-01 4.64654028e-01 6.71370924e-01
3.67350161e-01 1.18727600e-02 -7.42490351e-01 8.60644042e-01
1.84687525e-01 -1.37586105e+00 2.29965076e-01 9.16122645e-02
1.06050003e+00 -4.25485611e-01 5.37709951e-01 4.68709096e-02
9.71530527e-02 -1.33441293e+00 7.40308225e-01 3.64059776e-01
1.12266624e+00 -9.45271671e-01 8.03338766e-01 1.46809995e-01
-8.43203425e-01 1.41576797e-01 -1.28207952e-01 -2.08798479e-02
4.19259876e-01 5.38893998e-01 -1.18214917e+00 2.82731593e-01
5.33707500e-01 5.14071099e-02 -3.23410839e-01 1.09357548e+00
-4.76163894e-01 1.06337667e+00 -1.54501805e-02 1.01555260e-02
1.42260924e-01 -2.78676953e-02 7.03865588e-01 9.28056240e-01
7.27819681e-01 -7.50795081e-02 -5.02114594e-01 1.07248437e+00
-2.11679101e-01 8.44993293e-02 -8.49521756e-01 -2.96519428e-01
4.59168494e-01 1.17496049e+00 -6.48208678e-01 -7.09368214e-02
-1.13606744e-01 9.74143445e-01 -5.07893980e-01 1.85252801e-01
-7.87876070e-01 -7.46401966e-01 3.35316956e-01 2.69574434e-01
2.55415469e-01 -1.42409593e-01 -2.72123784e-01 -4.57308471e-01
3.17206187e-03 -1.35784090e+00 -1.70725230e-02 -1.05741072e+00
-1.06843305e+00 9.59605038e-01 -1.01618633e-01 -1.52983153e+00
-7.09089279e-01 -6.35590553e-02 -8.86150181e-01 1.05131578e+00
-1.04279006e+00 -1.37201238e+00 -5.32601058e-01 6.17343903e-01
8.27541411e-01 -7.38004148e-01 1.03109503e+00 2.88179785e-01
2.91726179e-03 8.02822888e-01 -1.22050896e-01 -1.22956663e-01
6.53085709e-01 -1.05120552e+00 7.79121518e-01 9.80617225e-01
4.63014066e-01 4.20879245e-01 1.05530441e+00 -5.33434451e-01
-1.05650043e+00 -1.07499039e+00 4.96954292e-01 1.39345154e-01
1.24978594e-01 -3.00183266e-01 -5.07720947e-01 3.05912584e-01
7.65851736e-01 -2.05866680e-01 7.35567749e-01 -6.34917974e-01
1.01405466e-02 -2.27600291e-01 -1.66691720e+00 6.15423620e-01
6.06584489e-01 -2.00923607e-01 -2.22996548e-01 -9.57576260e-02
5.98227441e-01 -3.80719990e-01 -7.95632660e-01 3.15961212e-01
4.71246570e-01 -1.12798309e+00 9.06663239e-01 -2.48336241e-01
6.58757925e-01 -2.48417854e-01 2.81613629e-05 -1.83166540e+00
4.85779643e-02 -1.20678365e+00 4.53994386e-02 1.54133034e+00
2.72513628e-01 -5.50677419e-01 9.79008675e-01 1.42352313e-01
-2.23141313e-01 -5.08648753e-01 -9.20073092e-01 -7.00266719e-01
-2.56280154e-02 -4.04761076e-01 7.48368502e-01 5.32658100e-01
-3.59166861e-01 4.00767207e-01 -8.83690715e-01 -2.76031882e-01
4.71689999e-01 -3.27033997e-01 1.16849184e+00 -9.47969735e-01
-4.32748109e-01 -1.07451148e-01 -6.84242785e-01 -6.58474267e-01
-1.79910138e-01 -6.73503458e-01 3.10278544e-03 -1.45393908e+00
-1.42283469e-01 -3.38620186e-01 1.15453720e-01 2.83762217e-01
8.98479298e-02 8.82985950e-01 2.49910831e-01 -1.40379414e-01
2.99203724e-01 5.99936247e-01 1.55034208e+00 -7.56274760e-02
-1.55675113e-01 4.02858794e-01 -7.80436397e-01 4.37895715e-01
9.93496537e-01 -4.44710940e-01 -8.17482054e-01 -1.68097556e-01
4.02085222e-02 1.51904836e-01 2.33486712e-01 -1.59607792e+00
-7.98511803e-02 4.84523661e-02 4.23552006e-01 -2.77054012e-01
5.10004640e-01 -6.96064711e-01 7.46110976e-01 5.29993474e-01
-3.61890227e-01 8.71408805e-02 2.55125552e-01 4.06823695e-01
-5.51380932e-01 -1.85170114e-01 9.94806051e-01 -2.56519437e-01
-3.41112226e-01 1.30909428e-01 -4.24253315e-01 1.06283121e-01
1.16398442e+00 -3.92459631e-01 -3.78311798e-02 -1.05073965e+00
-5.73983729e-01 -7.03934968e-01 1.16165631e-01 4.70510304e-01
6.94132805e-01 -1.57555962e+00 -8.09165895e-01 3.99566740e-01
-3.84147555e-01 -8.32473859e-03 3.04009259e-01 6.49634242e-01
-5.69824398e-01 3.38173985e-01 -7.42665470e-01 -2.22535521e-01
-1.29900610e+00 1.84232131e-01 3.03485002e-02 -1.58837363e-01
-5.69022596e-01 9.85472262e-01 5.44702597e-02 -8.00626129e-02
1.70079470e-02 3.93333696e-02 -7.65200928e-02 -1.78637803e-01
5.03102124e-01 5.49893796e-01 1.38433695e-01 -6.17352307e-01
9.81512144e-02 3.92601132e-01 1.63703769e-01 -6.09422803e-01
1.37303090e+00 3.02088559e-01 1.87351614e-01 3.31696242e-01
9.10677969e-01 2.77699888e-01 -1.01054907e+00 8.71583000e-02
-4.94199187e-01 -4.66587037e-01 -1.50110006e-01 -9.20968533e-01
-1.36450040e+00 1.12201035e+00 8.59067202e-01 4.04852360e-01
1.32281733e+00 -1.95093364e-01 1.14598334e+00 -2.38225415e-01
2.29033723e-01 -8.30781877e-01 4.35749173e-01 7.53227100e-02
1.07955337e+00 -8.09014082e-01 -2.69948363e-01 -1.26197115e-01
-7.45509744e-01 9.43590105e-01 4.44356829e-01 -1.06500171e-01
2.69254684e-01 3.87228012e-01 1.74890876e-01 2.01184601e-01
-5.08428991e-01 5.84933795e-02 4.03555423e-01 1.21930254e+00
6.66379392e-01 -8.79945979e-02 -3.22407335e-01 3.68858248e-01
-6.90046072e-01 4.60276790e-02 5.16293466e-01 5.97443044e-01
-1.90471113e-01 -1.36757469e+00 -4.92537528e-01 4.58853006e-01
-7.72740364e-01 -6.20948412e-02 -3.90431285e-01 7.48650670e-01
1.73330173e-01 1.12487507e+00 -5.31306826e-02 -5.31845570e-01
1.57996476e-01 -6.54283017e-02 3.14965189e-01 -4.83306319e-01
-5.86378574e-01 8.41861442e-02 -1.17811821e-01 -2.08787158e-01
-5.61909020e-01 -5.74156940e-01 -9.71610308e-01 -2.04435676e-01
-9.76688862e-02 2.74580091e-01 9.13244843e-01 6.23627663e-01
3.78867894e-01 8.51704180e-01 7.84167230e-01 -9.07300711e-01
-3.47712755e-01 -1.24495447e+00 -5.29385805e-01 4.72364962e-01
6.80176616e-02 -2.80563354e-01 -2.53960133e-01 3.96030843e-01] | [15.495232582092285, 5.999670505523682] |
ddc450da-ce1e-4d86-96b8-1551a09595b8 | reducing-labelled-data-requirement-for | 2102.12764 | null | https://arxiv.org/abs/2102.12764v1 | https://arxiv.org/pdf/2102.12764v1.pdf | Reducing Labelled Data Requirement for Pneumonia Segmentation using Image Augmentations | Deep learning semantic segmentation algorithms can localise abnormalities or opacities from chest radiographs. However, the task of collecting and annotating training data is expensive and requires expertise which remains a bottleneck for algorithm performance. We investigate the effect of image augmentations on reducing the requirement of labelled data in the semantic segmentation of chest X-rays for pneumonia detection. We train fully convolutional network models on subsets of different sizes from the total training data. We apply a different image augmentation while training each model and compare it to the baseline trained on the entire dataset without augmentations. We find that rotate and mixup are the best augmentations amongst rotate, mixup, translate, gamma and horizontal flip, wherein they reduce the labelled data requirement by 70% while performing comparably to the baseline in terms of AUC and mean IoU in our experiments. | ['Amit Kharat', 'Aniruddha Pant', 'Viraj Kulkarni', 'Rohit Lokwani', 'Jitesh Seth'] | 2021-02-25 | null | null | null | null | ['pneumonia-detection'] | ['medical'] | [ 6.13510668e-01 1.49553061e-01 -2.04225853e-02 -5.17822683e-01
-1.08725214e+00 -8.44174147e-01 1.99820518e-01 3.67809057e-01
-8.12708437e-01 3.29433829e-01 4.19695042e-02 -6.75597847e-01
1.60161316e-01 -4.67266530e-01 -8.31944823e-01 -6.02884769e-01
2.51979113e-01 7.15144277e-01 5.41656017e-01 3.39530617e-01
1.94508880e-02 2.40340158e-01 -1.01809585e+00 5.95834494e-01
6.89518750e-01 8.97801757e-01 1.00232817e-01 1.17000973e+00
-9.74344164e-02 9.82774675e-01 -4.76315677e-01 -3.35327864e-01
3.98072332e-01 -5.81314743e-01 -1.18917441e+00 1.92676365e-01
5.36630332e-01 -6.27285361e-01 -5.51057234e-02 5.97386539e-01
7.31708646e-01 -2.16409057e-01 7.84051836e-01 -7.26375937e-01
-2.27190733e-01 3.01623762e-01 -7.26633906e-01 8.87370408e-01
4.45582867e-02 2.01852922e-03 5.95606983e-01 -4.89397019e-01
4.88164574e-01 6.90876484e-01 1.16863096e+00 5.83358884e-01
-8.50120306e-01 -2.51804322e-01 -2.18375504e-01 -2.50733942e-01
-9.45361137e-01 -1.66210294e-01 8.21489021e-02 -5.11578798e-01
1.01549375e+00 5.54519773e-01 3.36309016e-01 8.66741240e-01
-5.03161252e-02 6.99922502e-01 9.28926945e-01 -5.63612401e-01
1.39291927e-01 -4.05688845e-02 4.11483139e-01 8.34315658e-01
3.96429598e-01 -4.01978225e-01 1.29533887e-01 -2.85055637e-01
7.90870547e-01 1.19238742e-01 -5.04026860e-02 -3.30332905e-01
-1.13442862e+00 8.22839975e-01 5.58233917e-01 2.63701856e-01
-4.18371558e-01 3.07618618e-01 5.59743226e-01 -1.93224803e-01
4.38457280e-01 5.88944197e-01 -7.05062866e-01 4.44743447e-02
-1.02206242e+00 1.08456232e-01 4.42362726e-01 5.38218319e-01
1.08301550e-01 -4.14898068e-01 -3.59048694e-01 1.05309451e+00
9.37334634e-03 4.65166986e-01 5.51383317e-01 -1.00675809e+00
6.95222318e-01 7.95190513e-01 -1.14359669e-01 -2.38915592e-01
-9.51972842e-01 -4.83969867e-01 -4.64751333e-01 -1.14415571e-01
6.20317698e-01 -5.82621515e-01 -1.82025075e+00 1.27225530e+00
2.58870691e-01 -1.04552835e-01 -1.71960771e-01 9.50960994e-01
9.93055165e-01 3.36766511e-01 5.51252723e-01 2.64131248e-01
1.65877616e+00 -1.00530159e+00 -2.70223826e-01 -4.62827295e-01
1.08183932e+00 -8.42686713e-01 1.01700449e+00 8.74071196e-02
-1.14482272e+00 -3.66570234e-01 -9.48963463e-01 -1.27866477e-01
-2.53478676e-01 2.10126474e-01 4.52334046e-01 8.64600241e-01
-1.18827796e+00 4.52452481e-01 -1.20630074e+00 -4.02023315e-01
9.84040678e-01 6.11534894e-01 -1.09186016e-01 -1.61535516e-01
-7.64191389e-01 8.16748619e-01 5.52577019e-01 -3.55972886e-01
-5.59540451e-01 -9.06749964e-01 -6.33135080e-01 -1.00493938e-01
2.50810027e-01 -9.67670918e-01 1.52181017e+00 -9.48652327e-01
-8.91577780e-01 1.25725067e+00 1.93212017e-01 -6.19814277e-01
6.86916232e-01 -2.45877296e-01 1.94908604e-01 4.36486483e-01
9.56795886e-02 1.18509614e+00 6.67205572e-01 -1.06206632e+00
-8.52117300e-01 -1.77023306e-01 2.35216469e-02 3.59993070e-01
-4.62000035e-02 2.71796554e-01 -6.03071928e-01 -7.32113063e-01
1.56051561e-01 -1.17576504e+00 -4.98632133e-01 -1.99966401e-01
-2.81334549e-01 1.19189257e-02 8.82967293e-01 -9.83939528e-01
9.19424653e-01 -1.89806616e+00 -2.34100163e-01 1.17403097e-01
1.97150126e-01 4.10177529e-01 1.07417464e-01 -2.15849370e-01
-3.60594064e-01 4.95300889e-01 -7.01544821e-01 -2.58232296e-01
-3.92435998e-01 4.61680800e-01 2.65434533e-01 2.84933299e-01
3.05260003e-01 1.22654784e+00 -5.64730406e-01 -7.47869253e-01
3.50817561e-01 4.90915269e-01 -7.62730598e-01 1.82823092e-01
-1.27576575e-01 5.53086698e-01 -4.59567219e-01 5.68798065e-01
5.07884204e-01 -7.50894785e-01 4.98066172e-02 1.53266296e-01
3.92199844e-01 3.72065276e-01 -8.07625949e-01 1.62914705e+00
-3.88336629e-01 4.22115386e-01 -7.41419867e-02 -9.97587562e-01
2.27106512e-01 5.74047506e-01 6.69929385e-01 -6.07984602e-01
3.47068369e-01 3.75436991e-01 2.04906732e-01 -6.99131429e-01
-3.58039439e-02 -2.97442049e-01 8.60371962e-02 7.36889482e-01
-1.26548156e-01 -6.43754378e-02 1.95026651e-01 8.54521617e-02
1.47512364e+00 -1.11661740e-01 1.14831135e-01 -3.10212225e-01
4.33392793e-01 2.40294129e-01 -4.82357293e-02 8.30953658e-01
-2.55898476e-01 1.12066031e+00 5.20008266e-01 -5.79937875e-01
-1.35739887e+00 -8.89846325e-01 -2.98131049e-01 1.19025683e+00
-2.19504654e-01 8.24933052e-02 -1.27070916e+00 -1.34841228e+00
-3.84788364e-01 3.68864834e-01 -6.37061894e-01 2.78854877e-01
-9.49515522e-01 -1.32947433e+00 6.84400022e-01 1.05855048e+00
5.70964098e-01 -1.21608853e+00 -1.26708913e+00 -5.46695329e-02
-3.97215694e-01 -1.30023706e+00 -2.92258322e-01 4.65017259e-01
-1.13027239e+00 -1.36142099e+00 -1.13552034e+00 -8.19758952e-01
9.67961848e-01 -1.03704341e-01 1.27696145e+00 4.87671345e-01
-6.38039947e-01 4.41830397e-01 -3.82281631e-01 -7.46225297e-01
-4.12516624e-01 5.10309219e-01 -7.58963764e-01 -5.88590741e-01
2.42810875e-01 6.35306388e-02 -1.01975965e+00 7.54493475e-02
-1.07086408e+00 6.29281327e-02 5.98055243e-01 6.44152701e-01
5.50858438e-01 -5.61964177e-02 2.80130178e-01 -1.25956881e+00
4.58462834e-01 -4.60688531e-01 -9.82332304e-02 -4.43736613e-02
-2.93525875e-01 -6.06555752e-02 3.46537441e-01 -2.44673677e-02
-9.30589139e-01 3.66439909e-01 -3.56227636e-01 -1.41366899e-01
-6.10185981e-01 6.81824684e-02 5.81285417e-01 -1.25962496e-01
7.85337448e-01 -2.40428209e-01 -1.85515478e-01 -4.65080202e-01
1.88040391e-01 5.99638224e-01 4.50640202e-01 -1.44547537e-01
4.47969913e-01 6.24282539e-01 -7.37827048e-02 -4.45457011e-01
-1.03707397e+00 -7.96929538e-01 -9.36650634e-01 1.27050862e-01
1.44508791e+00 -6.81540191e-01 2.22613722e-01 3.48964065e-01
-1.03406858e+00 -4.21741962e-01 -4.70688403e-01 3.84177119e-01
-2.91559428e-01 2.98252672e-01 -6.32340550e-01 -2.19608530e-01
-6.16050482e-01 -1.37275970e+00 1.23505402e+00 1.34632766e-01
-3.78542542e-01 -1.03546202e+00 -7.62901008e-02 7.97652483e-01
3.72450769e-01 5.71404994e-01 1.13449299e+00 -9.96810734e-01
-2.38176972e-01 -2.71709651e-01 -6.37720823e-01 4.70696270e-01
1.82124481e-01 -3.69475394e-01 -9.96224046e-01 -1.39290899e-01
1.91426471e-01 -3.21271390e-01 1.00658882e+00 9.09171402e-01
1.52851903e+00 -1.31802158e-02 -3.99015874e-01 6.15988553e-01
1.35034299e+00 3.88638765e-01 4.17324007e-01 5.71259797e-01
8.18959653e-01 4.33648944e-01 1.61838174e-01 -1.01623591e-02
1.94273442e-01 1.16645873e-01 3.74502391e-01 -7.54279137e-01
-2.97301978e-01 3.73029709e-01 -3.25480878e-01 4.13275242e-01
-7.72113279e-02 -3.69851857e-01 -1.52592874e+00 6.53435051e-01
-1.40064156e+00 -2.10195571e-01 -3.21339190e-01 1.79545689e+00
6.81118906e-01 2.31109872e-01 3.03130239e-01 2.84702718e-01
6.91977143e-01 1.67600694e-03 -3.17692369e-01 -4.13462579e-01
3.74926120e-01 7.52509415e-01 9.22076762e-01 3.10564339e-01
-1.62527955e+00 5.65034330e-01 7.31113911e+00 4.10036236e-01
-1.07946157e+00 4.19988394e-01 1.16009343e+00 -2.43963271e-01
1.90830693e-01 -3.84099901e-01 -2.82056302e-01 4.32308257e-01
1.14582491e+00 8.34602833e-01 4.96838614e-02 7.65350819e-01
-9.51472893e-02 -2.83941090e-01 -1.03592849e+00 5.63711286e-01
7.24845454e-02 -1.08686221e+00 -1.39156997e-01 -1.82218134e-01
9.90601778e-01 5.96231043e-01 -2.55914163e-02 1.71646699e-02
1.59678727e-01 -1.18228519e+00 2.40063921e-01 -1.14061777e-02
7.57412732e-01 -5.82806647e-01 1.21910822e+00 4.53537628e-02
-7.40743816e-01 2.66727749e-02 4.59194779e-02 2.90191114e-01
6.54317066e-02 1.37250438e-01 -1.28238559e+00 3.43121827e-01
8.92286658e-01 1.39866546e-01 -1.03023255e+00 1.10605502e+00
-5.53022102e-02 9.53411341e-01 -5.95326185e-01 3.14540416e-01
5.74814558e-01 2.29149505e-01 1.72804713e-01 1.49202335e+00
-2.38361191e-02 3.66333961e-01 1.55335665e-01 4.26361233e-01
-2.44608894e-01 8.90741721e-02 -1.69248596e-01 8.35674629e-02
-1.74233820e-02 9.40404534e-01 -1.48440337e+00 -7.04792321e-01
-4.16991264e-01 8.87639403e-01 -1.51502907e-01 1.79550543e-01
-1.11917830e+00 -6.67453781e-02 -6.34490252e-02 3.09735805e-01
4.86881554e-01 2.83037275e-01 -8.04492354e-01 -5.60719192e-01
6.34497404e-02 -8.94968748e-01 8.32137585e-01 -7.32351720e-01
-8.35119486e-01 4.77198839e-01 3.95575427e-02 -6.31078482e-01
-1.85500413e-01 -6.39464080e-01 -5.63118160e-01 6.94473624e-01
-1.55921721e+00 -1.03804624e+00 -6.36067390e-01 3.38081658e-01
7.50470340e-01 1.84540957e-01 7.41969645e-01 4.89986211e-01
-5.03681779e-01 5.14171362e-01 -1.77416846e-01 3.34797263e-01
3.48593593e-01 -1.53581738e+00 5.53620279e-01 7.52338648e-01
6.71038106e-02 3.04139614e-01 3.80773842e-01 -4.46853727e-01
-6.25094712e-01 -1.24018645e+00 4.37180400e-01 -9.24905360e-01
2.62203783e-01 5.10889739e-02 -9.92201924e-01 7.58203268e-01
2.51398981e-01 1.44419715e-01 7.47146904e-01 -2.54471749e-01
-7.05419555e-02 5.41215479e-01 -1.20990574e+00 1.97090149e-01
7.48374045e-01 -3.24960381e-01 -6.53322339e-01 7.30498672e-01
7.24245667e-01 -8.01867247e-01 -9.20721352e-01 7.32117712e-01
3.08091402e-01 -9.55063343e-01 1.23187947e+00 -6.39492095e-01
6.36320949e-01 -7.18068480e-02 2.37131909e-01 -7.70284116e-01
-1.57243788e-01 8.76827836e-02 2.99526930e-01 7.34364510e-01
6.94252074e-01 -5.56216121e-01 1.28093600e+00 5.04913628e-01
-2.38330036e-01 -1.05294037e+00 -7.02727139e-01 -2.75504082e-01
3.74743909e-01 -4.07270044e-01 3.31465393e-01 8.70170534e-01
-6.89718366e-01 9.52235982e-02 3.12633723e-01 -1.17232231e-02
3.45313609e-01 -3.12355518e-01 3.54139835e-01 -8.75959575e-01
-2.03889385e-01 -4.43234116e-01 -2.00058013e-01 -5.07864833e-01
-4.25431937e-01 -1.06081927e+00 -1.45627022e-01 -2.06200743e+00
5.17676175e-01 -5.86617112e-01 -4.73322809e-01 7.55205333e-01
-6.15915060e-01 6.52461648e-01 6.44055232e-02 4.01763827e-01
-5.14177501e-01 -4.38311368e-01 1.20982385e+00 4.76308651e-02
-1.36629194e-01 5.36114648e-02 -5.85756958e-01 1.08284378e+00
1.06118429e+00 -6.59292638e-01 -4.82426256e-01 -7.99652398e-01
-1.00815915e-01 -2.01798081e-01 4.15829688e-01 -1.10264599e+00
-2.52601773e-01 3.04827034e-01 7.19399571e-01 -6.73718572e-01
-5.12723289e-02 -8.10603440e-01 -2.09106445e-01 7.52157629e-01
-4.03344810e-01 1.86083809e-01 5.57421684e-01 2.92928696e-01
1.38209060e-01 -4.60820496e-01 1.00108397e+00 -4.83384281e-01
-4.10279512e-01 -9.74876457e-04 -2.74977118e-01 4.31609929e-01
1.02053857e+00 -3.23884130e-01 -2.48561859e-01 1.57853708e-01
-8.40651810e-01 1.16124682e-01 4.09686327e-01 2.16979891e-01
1.65512621e-01 -7.35499859e-01 -5.70213974e-01 1.46368518e-01
-1.29337311e-01 5.62109351e-01 1.88254103e-01 1.05703211e+00
-1.40228367e+00 7.35413849e-01 -3.04559036e-03 -1.02344811e+00
-1.47198188e+00 4.11282420e-01 5.90978861e-01 -6.46560669e-01
-7.11711764e-01 1.18253303e+00 3.96404386e-01 -4.48495090e-01
2.84684360e-01 -7.01591253e-01 -1.77945748e-01 -1.92364767e-01
2.18879282e-01 3.23502213e-01 4.24162686e-01 -2.70315647e-01
-4.99339581e-01 6.39196277e-01 -3.23869526e-01 8.41933787e-02
1.11338747e+00 1.73693269e-01 1.21897243e-01 -2.89707705e-02
1.29162765e+00 -5.31667888e-01 -1.08224940e+00 7.97083303e-02
1.82274625e-01 -1.10662699e-01 9.11568180e-02 -1.06862485e+00
-1.17892158e+00 8.43932569e-01 9.86673415e-01 2.10137025e-01
1.12943029e+00 2.67144948e-01 9.27743137e-01 1.34148270e-01
-1.94179058e-01 -9.66928363e-01 6.63619563e-02 1.38736993e-01
3.87457758e-01 -1.47775519e+00 1.77769378e-01 -4.93851900e-01
-7.86181509e-01 8.10863376e-01 4.83115375e-01 -2.78044522e-01
5.59608817e-01 3.92774671e-01 4.64124858e-01 -5.53243756e-01
-2.81848729e-01 -1.21463947e-01 4.65482920e-01 5.15371263e-01
6.67352796e-01 7.63587467e-03 -3.07768077e-01 2.01144919e-01
-1.32755697e-01 -4.06813547e-02 2.15832189e-01 1.27909184e+00
-4.80776131e-01 -9.23224926e-01 -5.01960695e-01 9.00848746e-01
-1.17335594e+00 -1.18233547e-01 -5.21386802e-01 9.06018853e-01
4.57743108e-01 5.95999658e-01 1.59223691e-01 1.04341596e-01
4.02281702e-01 3.27309936e-01 3.65911394e-01 -8.76062274e-01
-9.01634097e-01 1.42273381e-01 1.98193282e-01 -3.07632208e-01
-5.70097327e-01 -6.13663733e-01 -1.61546969e+00 2.49375492e-01
-3.99214149e-01 -1.98472872e-01 6.75964832e-01 1.21217871e+00
8.16847757e-02 1.18808329e+00 1.06288135e-01 -4.81387764e-01
-3.61099422e-01 -9.36291277e-01 1.23895630e-01 5.87421298e-01
3.84984225e-01 -3.64605606e-01 -2.07279280e-01 4.55669731e-01] | [15.052009582519531, -2.0510826110839844] |
d4bb122a-c859-4fd2-af9e-035dbc48c0c4 | learning-deep-bilinear-transformation-for | 1911.03621 | null | https://arxiv.org/abs/1911.03621v1 | https://arxiv.org/pdf/1911.03621v1.pdf | Learning Deep Bilinear Transformation for Fine-grained Image Representation | Bilinear feature transformation has shown the state-of-the-art performance in learning fine-grained image representations. However, the computational cost to learn pairwise interactions between deep feature channels is prohibitively expensive, which restricts this powerful transformation to be used in deep neural networks. In this paper, we propose a deep bilinear transformation (DBT) block, which can be deeply stacked in convolutional neural networks to learn fine-grained image representations. The DBT block can uniformly divide input channels into several semantic groups. As bilinear transformation can be represented by calculating pairwise interactions within each group, the computational cost can be heavily relieved. The output of each block is further obtained by aggregating intra-group bilinear features, with residuals from the entire input features. We found that the proposed network achieves new state-of-the-art in several fine-grained image recognition benchmarks, including CUB-Bird, Stanford-Car, and FGVC-Aircraft. | ['Zheng-Jun Zha', 'Heliang Zheng', 'Jiebo Luo', 'Jianlong Fu'] | 2019-11-09 | learning-deep-bilinear-transformation-for-1 | http://papers.nips.cc/paper/8680-learning-deep-bilinear-transformation-for-fine-grained-image-representation | http://papers.nips.cc/paper/8680-learning-deep-bilinear-transformation-for-fine-grained-image-representation.pdf | neurips-2019-12 | ['fine-grained-image-recognition'] | ['computer-vision'] | [-5.62282205e-02 -2.97908455e-01 1.23128414e-01 -6.73822582e-01
-7.43043482e-01 -6.49896204e-01 5.15529215e-01 -1.75202608e-01
-3.22379500e-01 5.31712234e-01 2.48084068e-01 -1.47232682e-01
-1.65590897e-01 -1.03514981e+00 -1.06515408e+00 -8.62965345e-01
7.96993300e-02 -6.25888780e-02 1.08722053e-01 -1.66109130e-01
1.03955962e-01 6.97929621e-01 -1.43621767e+00 8.16445827e-01
5.65999985e-01 1.58234394e+00 -1.43975750e-01 4.44019586e-01
-5.08198142e-02 8.80909920e-01 -3.48850489e-01 -2.43666276e-01
4.93268877e-01 2.80167963e-02 -7.52846718e-01 -1.93673208e-01
8.64531040e-01 -5.17789841e-01 -6.26925528e-01 8.25250924e-01
2.94951022e-01 1.11356832e-01 6.16907179e-01 -1.23148811e+00
-1.01183736e+00 3.79038543e-01 -4.23201084e-01 -7.08934367e-02
-2.33113289e-01 4.15448770e-02 1.16776371e+00 -1.16939080e+00
9.81034636e-02 1.40556109e+00 7.71255910e-01 2.31849864e-01
-1.13953876e+00 -1.05088556e+00 2.52319723e-01 1.97291836e-01
-1.54513681e+00 -1.01873362e-02 4.68132615e-01 -5.28280497e-01
1.02043021e+00 2.48271003e-01 4.28716034e-01 6.08785450e-01
1.70844987e-01 5.66084087e-01 1.24523413e+00 -8.10041502e-02
-1.44126147e-01 -5.34419239e-01 2.98833042e-01 1.05689752e+00
9.60644409e-02 2.74239928e-01 -3.80780190e-01 -1.71100404e-02
9.89740670e-01 3.28850120e-01 1.03998788e-01 -2.79220343e-01
-1.25002635e+00 9.96377647e-01 1.31376660e+00 1.60625771e-01
-2.04105496e-01 6.22346699e-01 2.70064056e-01 3.46538275e-01
2.94127285e-01 3.39952916e-01 -3.32897753e-01 9.48692486e-02
-7.21712649e-01 2.90910721e-01 5.34047782e-01 6.81592762e-01
1.19346035e+00 -1.01061083e-01 -5.10571718e-01 8.07579219e-01
2.05720231e-01 1.29885063e-01 2.59559542e-01 -8.09494972e-01
4.36418802e-01 7.98686683e-01 -2.09621504e-01 -1.15605247e+00
-2.35915780e-01 -4.20543700e-01 -1.26428831e+00 2.37595156e-01
3.61856788e-01 9.43737850e-02 -1.17867732e+00 1.51704586e+00
5.28099351e-02 1.38691097e-01 -3.03547293e-01 9.21751738e-01
9.45069790e-01 8.24748337e-01 7.65101761e-02 4.32603002e-01
1.13859236e+00 -1.43061101e+00 -4.07342911e-02 3.99081223e-02
4.92873430e-01 -6.81437314e-01 1.06118095e+00 2.66420215e-01
-8.14228058e-01 -9.63983238e-01 -1.17378330e+00 -6.90343142e-01
-5.05017221e-01 8.30619484e-02 9.49865818e-01 2.96170026e-01
-1.05030096e+00 5.75438201e-01 -7.91167915e-01 1.79180861e-01
6.99692309e-01 5.35261095e-01 -6.09632075e-01 -2.90661216e-01
-1.14443207e+00 4.22953397e-01 3.43811363e-01 2.86961943e-01
-1.03628492e+00 -9.65938747e-01 -8.61009121e-01 4.27354097e-01
-2.82303125e-01 -7.38686383e-01 9.04733956e-01 -6.36853456e-01
-1.21930039e+00 6.64798021e-01 -1.32563367e-01 -2.18673229e-01
3.56879234e-01 -8.63666162e-02 -6.52793571e-02 -2.14048326e-01
1.03989087e-01 9.32264984e-01 1.06383717e+00 -9.89390194e-01
-6.78149223e-01 -3.48746181e-01 3.32917839e-01 -1.16493009e-01
-4.15805191e-01 -8.07267353e-02 -5.36319733e-01 -1.01576531e+00
3.70361805e-02 -1.02595592e+00 -3.70132655e-01 1.14256017e-01
-4.18676853e-01 -4.22605276e-01 8.32920849e-01 -4.51997161e-01
1.06062329e+00 -2.43550324e+00 3.16812806e-02 2.61541277e-01
4.99097019e-01 4.59814183e-02 -3.90366167e-01 1.44482687e-01
-1.47588775e-01 1.22378767e-01 1.04675749e-02 -1.08293556e-01
4.32685226e-01 1.68386102e-01 -4.68545437e-01 2.52882689e-01
3.21561933e-01 1.25402570e+00 -4.45574433e-01 -3.45303416e-02
3.08445603e-01 6.08097732e-01 -8.50823402e-01 2.49799967e-01
5.31775318e-02 9.80436429e-02 -5.50989807e-01 3.91898990e-01
9.06412542e-01 -4.76623088e-01 -2.68115610e-01 -7.39855111e-01
-7.17315748e-02 8.99185911e-02 -8.00017536e-01 1.43340111e+00
-6.11671448e-01 4.86741573e-01 -3.63876611e-01 -9.52026606e-01
1.02879524e+00 -2.24306226e-01 1.52188510e-01 -7.78817177e-01
5.95195703e-02 5.69266193e-02 -3.51989791e-02 2.07938686e-01
4.54255104e-01 -3.86163853e-02 -5.58010876e-01 3.80257636e-01
8.50985423e-02 -1.04219414e-01 -7.92877302e-02 -1.43869609e-01
9.06114876e-01 -1.90124154e-01 1.25911415e-01 -3.27197462e-01
2.99461097e-01 -2.65487522e-01 4.09731537e-01 7.34598696e-01
5.55596650e-02 6.82729185e-01 2.90359110e-01 -1.08600605e+00
-9.13900316e-01 -1.27046883e+00 9.30927396e-02 1.62630558e+00
9.50665250e-02 -3.42844844e-01 -8.83733451e-01 -7.07449436e-01
3.77809197e-01 -2.10311506e-02 -9.30336535e-01 -3.63604814e-01
-6.39276564e-01 -5.07311404e-01 7.68705308e-01 8.25800240e-01
8.37572813e-01 -9.30131614e-01 -1.65520251e-01 1.37884110e-01
2.72445753e-02 -9.44433868e-01 -7.48872578e-01 3.83512348e-01
-6.33226514e-01 -8.67969096e-01 -6.25823438e-01 -9.56359684e-01
8.06271136e-01 3.75671655e-01 8.70208919e-01 -3.30441147e-02
-3.75969291e-01 -3.12649578e-01 -2.04612449e-01 8.99530947e-03
2.54790425e-01 1.06990248e-01 -5.76050878e-02 2.24189803e-01
2.95555711e-01 -3.62507731e-01 -7.34978974e-01 4.03486133e-01
-8.92896891e-01 7.01530054e-02 6.83492601e-01 1.32114255e+00
8.70462418e-01 1.59750685e-01 2.38785103e-01 -7.70497322e-01
5.06921768e-01 -1.34993553e-01 -5.14230788e-01 2.22756803e-01
-4.62854505e-02 3.09803128e-01 9.43844497e-01 -1.41152158e-01
-9.49854195e-01 1.74770970e-02 -3.08814317e-01 -5.44322729e-01
-1.91714957e-01 5.37782490e-01 -1.27365604e-01 -6.10696077e-01
4.71005201e-01 6.65927231e-02 -6.11272514e-01 -5.36841929e-01
5.96602499e-01 6.33266509e-01 6.44240558e-01 -7.80804157e-01
8.31174552e-01 3.80396366e-01 -8.91383272e-03 -5.84136784e-01
-1.19574797e+00 -1.89118490e-01 -5.99777758e-01 4.07771587e-01
8.57320964e-01 -1.07750332e+00 -9.09457624e-01 8.70253265e-01
-9.85754669e-01 -3.90339047e-01 -7.91263208e-02 2.39568278e-01
-3.10995668e-01 2.80817840e-02 -6.74934447e-01 1.58020318e-01
-3.06325644e-01 -1.31859839e+00 1.22686994e+00 2.79979497e-01
3.67490500e-02 -7.67248571e-01 -2.47709811e-01 5.34647286e-01
5.27016461e-01 2.57676691e-01 1.02481544e+00 -9.90872681e-02
-7.88417161e-01 -1.41694531e-01 -9.24443722e-01 5.57914078e-01
3.40124846e-01 -1.57661140e-01 -1.09734988e+00 -5.70075095e-01
-5.00541568e-01 -6.43026173e-01 1.38929904e+00 1.74374297e-01
1.99626541e+00 -3.04810911e-01 -8.28608051e-02 1.22276294e+00
1.24251413e+00 -1.22221820e-01 6.19979620e-01 3.09171557e-01
1.23280120e+00 2.45042861e-01 4.14462060e-01 4.17900652e-01
5.46011329e-01 5.69333494e-01 3.63214731e-01 -4.49209452e-01
-2.83068269e-01 -3.92825246e-01 7.89990425e-02 7.68716395e-01
-1.78634048e-01 3.21237519e-02 -6.94579840e-01 3.36279392e-01
-1.75444567e+00 -6.11844003e-01 3.22832227e-01 1.58595145e+00
7.41302788e-01 1.23753734e-02 -2.87525654e-01 -7.81222209e-02
3.85756254e-01 4.79015708e-01 -5.62002897e-01 -5.63111961e-01
-1.00516357e-01 7.17731297e-01 6.95699930e-01 3.00353616e-01
-1.41289747e+00 1.15058100e+00 6.05485678e+00 1.04246092e+00
-1.50610983e+00 -1.01042846e-02 9.82238531e-01 1.13994785e-01
-2.94840842e-01 -4.29747343e-01 -8.12996864e-01 3.23681384e-01
5.00097275e-01 1.29119411e-01 4.58573192e-01 7.27885246e-01
-1.91750571e-01 4.62412834e-01 -1.17921424e+00 1.21716738e+00
-1.94224641e-01 -1.57980168e+00 5.64753473e-01 -3.11415214e-02
1.20113277e+00 2.34248713e-01 3.94089550e-01 4.20336455e-01
5.37266433e-01 -1.60852420e+00 5.82764447e-01 3.78211975e-01
1.19186497e+00 -9.27508533e-01 5.45022964e-01 4.77267094e-02
-1.51312494e+00 -9.52545926e-02 -7.13631332e-01 -2.16799781e-01
-3.57548445e-01 6.18579686e-01 -4.67608124e-01 3.79185587e-01
1.05466020e+00 7.18803227e-01 -5.26774764e-01 6.85900033e-01
-8.39581564e-02 4.12279129e-01 -1.71253398e-01 1.63713157e-01
8.02071571e-01 2.79226173e-02 -3.58377695e-01 1.33301127e+00
4.18381691e-01 3.30129005e-02 3.70280415e-01 7.54108489e-01
-5.66653907e-01 -3.38679314e-01 -2.75163054e-01 -2.18560562e-01
1.92790434e-01 1.34734666e+00 -3.34295183e-01 -1.94306731e-01
-5.11908770e-01 1.05066133e+00 7.61158466e-01 2.88879544e-01
-8.01653445e-01 -6.31746948e-01 1.15998733e+00 -1.65042982e-01
7.71934271e-01 -2.48359501e-01 -4.62425321e-01 -1.09775651e+00
-1.71237011e-02 -8.81907463e-01 3.88064235e-01 -4.38542515e-01
-1.39514959e+00 7.05855191e-01 -4.10546541e-01 -1.03927469e+00
5.78279458e-02 -9.64054167e-01 -4.11600679e-01 9.50353801e-01
-1.52580833e+00 -1.53866804e+00 -7.39246666e-01 9.76429462e-01
1.85984328e-01 -2.18006372e-02 8.44901025e-01 5.28594255e-01
-1.71444654e-01 1.04614067e+00 1.13265000e-01 4.08765793e-01
6.12178206e-01 -1.24155200e+00 8.04514706e-01 5.06792963e-01
2.61177886e-02 7.03666389e-01 -5.12649631e-03 -1.50116116e-01
-1.25224388e+00 -1.71619570e+00 7.70488679e-01 -8.87041911e-02
6.36297047e-01 -5.56241214e-01 -8.55098307e-01 7.13304698e-01
-1.44281313e-01 8.98048699e-01 5.66672206e-01 1.98982079e-02
-1.14383662e+00 -4.78954822e-01 -9.70268250e-01 4.15207267e-01
1.13262689e+00 -9.56077874e-01 -4.85069424e-01 7.68580958e-02
8.73103380e-01 -4.18719888e-01 -1.11833251e+00 5.58164775e-01
8.94608200e-01 -8.20794106e-01 1.22515428e+00 -6.98740184e-01
4.53407139e-01 -5.31716585e-01 -4.28577095e-01 -1.58601201e+00
-9.29518163e-01 -1.93666920e-01 1.27406344e-01 8.18105400e-01
1.93776980e-01 -7.26246417e-01 8.12520027e-01 3.57022166e-01
-2.08632603e-01 -8.74774456e-01 -7.38305151e-01 -5.17298520e-01
4.42122608e-01 -1.47045165e-01 8.66653502e-01 8.32183897e-01
-5.36668956e-01 1.02027774e-01 -2.93033451e-01 6.95767105e-02
6.32962704e-01 4.42579627e-01 7.10350215e-01 -1.08875561e+00
-3.24692398e-01 -3.70354801e-01 -5.89283049e-01 -1.25315511e+00
1.79699764e-01 -8.75083447e-01 9.23867226e-02 -1.37624133e+00
3.25291395e-01 -8.01538169e-01 -7.91023076e-01 8.54995430e-01
-1.61682591e-01 1.01696002e+00 3.93572897e-01 -5.72067387e-02
-5.50781012e-01 5.36052883e-01 1.67395020e+00 -5.24193466e-01
3.35901022e-01 -4.22929049e-01 -9.76876557e-01 6.23578370e-01
6.72474563e-01 -1.59765288e-01 -1.61782637e-01 -7.60457814e-01
9.45878401e-02 -3.25589657e-01 4.66890126e-01 -9.34485734e-01
2.71811813e-01 -1.26990154e-01 8.07648659e-01 -4.01575685e-01
3.76259327e-01 -6.28384054e-01 3.12051862e-01 3.81628960e-01
-3.01716477e-01 -1.20228209e-01 3.35598588e-01 5.84239028e-02
-7.70896137e-01 4.69284326e-01 9.16283190e-01 -1.23914443e-01
-9.44357097e-01 8.40663433e-01 1.79320216e-01 -1.61566198e-01
8.90401065e-01 -2.25219324e-01 -5.25161564e-01 3.33457664e-02
-5.60664117e-01 -7.70989060e-03 2.79624134e-01 5.37242174e-01
5.96028149e-01 -1.81139719e+00 -6.23447657e-01 6.30055547e-01
2.09111974e-01 2.28513584e-01 5.64576328e-01 3.88829708e-01
-5.55935621e-01 6.81263745e-01 -6.58625960e-01 -4.96375769e-01
-1.08129394e+00 3.48477721e-01 2.56589204e-01 -5.49599946e-01
-2.53728509e-01 1.25805950e+00 1.10485852e+00 -6.06199980e-01
1.05946101e-01 -6.95179224e-01 6.39752075e-02 -1.91336200e-01
5.83642781e-01 2.12320447e-01 2.61269242e-01 -6.00316763e-01
-3.32272679e-01 8.88778985e-01 -3.93622011e-01 5.60936034e-01
1.36174917e+00 1.48896843e-01 -3.78732294e-01 -4.56505865e-02
1.86960375e+00 -2.36684844e-01 -1.46007860e+00 -2.30253220e-01
-4.16749984e-01 -4.72698122e-01 1.70139864e-01 -7.03414023e-01
-1.22258878e+00 1.05604744e+00 4.93387222e-01 -4.51836884e-02
1.30416238e+00 -3.53161953e-02 9.87627685e-01 6.91796005e-01
5.46635687e-01 -5.93500614e-01 8.32010210e-02 9.62144673e-01
1.02640295e+00 -1.17667592e+00 -3.48248720e-01 -4.88087177e-01
-2.43568912e-01 1.09646022e+00 6.32109225e-01 -6.20553792e-01
9.60405588e-01 3.59359175e-01 4.00659777e-02 -4.36208658e-02
-5.58238208e-01 9.36925411e-02 6.58930838e-01 3.16974491e-01
4.91031289e-01 5.90654194e-01 1.28476754e-01 7.02866554e-01
-3.96649390e-01 -9.36511159e-03 -1.53049693e-01 6.11169994e-01
-3.73398066e-01 -9.56041038e-01 -1.59046113e-01 7.21458256e-01
-3.78881603e-01 -4.17690635e-01 -2.81420529e-01 5.07177591e-01
4.62614447e-01 5.47251225e-01 3.09508294e-01 -6.10675097e-01
3.76622915e-01 -5.04504025e-01 6.00881755e-01 -4.79456902e-01
-7.18062818e-01 -3.17690164e-01 -1.35280803e-01 -1.01991999e+00
-1.34135231e-01 -7.58711025e-02 -1.13039052e+00 -5.27329981e-01
2.15708300e-01 -2.64897048e-02 2.24221453e-01 9.24220502e-01
4.45430487e-01 6.90938890e-01 6.91844761e-01 -1.06956804e+00
-4.57182765e-01 -7.76015699e-01 -5.26594639e-01 4.50641483e-01
6.21313989e-01 -7.74803102e-01 -1.34029076e-01 -8.24225768e-02] | [9.563851356506348, 2.0379958152770996] |
539135a8-f185-4a9d-b5db-46dd21b6b102 | advancements-in-noncontact-multiparameter | null | null | https://ieeexplore.ieee.org/document/5599853 | https://affect.media.mit.edu/pdfs/11.Poh-etal-TBME.pdf | Advancements in Noncontact, Multiparameter Physiological Measurements Using a Webcam | We present a simple, low-cost method for measuring multiple physiological parameters using a basic webcam. By applying independent component analysis on the color channels in video recordings, we extracted the blood volume pulse from the facial regions. Heart rate (HR), respiratory rate, and HR variability (HRV, an index for cardiac autonomic activity) were subsequently quantified and compared to corresponding measurements using Food and Drug Administration-approved sensors. High degrees of agreement were achieved between the measurements across all physiological parameters. This technology has significant potential for advancing personal health care and telemedicine. | ['Rosalind W. Picard', 'Daniel J. McDuff', 'Ming-Zher Poh'] | 2010-10-14 | null | null | null | ieee-transactions-on-biomedical-engineering-4 | ['photoplethysmography-ppg-heart-rate'] | ['medical'] | [ 1.41276687e-01 -2.98506826e-01 -1.62898302e-01 -4.04684663e-01
-2.87690014e-02 -4.22731996e-01 -3.32734197e-01 2.57630795e-02
-3.91946375e-01 8.06981623e-01 7.53800943e-02 2.23593995e-01
3.53108048e-01 -2.92583972e-01 2.99145937e-01 -6.84466481e-01
-3.52657109e-01 -3.95943701e-01 -4.74225551e-01 1.95562705e-01
1.04729697e-01 5.53545177e-01 -1.07573986e+00 -1.56609163e-01
4.65974450e-01 1.43253005e+00 -8.10533583e-01 1.00616932e+00
3.55963916e-01 4.34895784e-01 -7.90771782e-01 5.08098267e-02
1.40664458e-01 -1.03668702e+00 -1.29808784e-01 -1.71194822e-01
1.70872927e-01 -4.26652879e-01 -1.82883218e-01 6.00560844e-01
9.54273283e-01 2.19186917e-02 1.21983117e-03 -1.10590577e+00
-4.38201994e-01 2.00145990e-02 -4.30784792e-01 6.29816234e-01
6.59151614e-01 1.41051352e-01 8.61608237e-02 -6.44570529e-01
1.98034078e-01 8.88531208e-01 8.88896346e-01 7.61917651e-01
-1.40780199e+00 -6.43402636e-01 -5.17494917e-01 -4.68897400e-03
-1.52906251e+00 -8.86787772e-01 7.76258171e-01 -3.71702045e-01
5.92544019e-01 5.74439645e-01 1.08728933e+00 8.16432595e-01
6.65859818e-01 -3.86239529e-01 1.64928472e+00 -1.00198425e-01
1.52339771e-01 6.11232519e-01 7.92729557e-02 6.07753038e-01
4.30910826e-01 1.40006453e-01 -6.72326028e-01 -3.21211338e-01
1.06543517e+00 1.00187518e-01 -6.63094997e-01 1.80292383e-01
-9.99059916e-01 2.98234999e-01 -2.31917322e-01 3.48600477e-01
-6.33631051e-01 -1.73932966e-02 2.45072931e-01 6.92278817e-02
3.08033168e-01 2.97275305e-01 -3.19357634e-01 -5.76793373e-01
-7.39542067e-01 -6.63773537e-01 1.11985719e+00 4.78167832e-01
3.48093137e-02 1.41170740e-01 -3.09665591e-01 5.08925974e-01
4.65242594e-01 9.64083850e-01 5.40647745e-01 -1.51906705e+00
-4.86114591e-01 2.79913247e-01 3.13402236e-01 -8.93868804e-01
-8.72054636e-01 2.20509499e-01 -6.99197173e-01 1.41919345e-01
3.61144960e-01 -5.99113166e-01 -3.74740183e-01 1.17569613e+00
5.96888006e-01 1.22736871e-01 -1.10022649e-01 1.15424967e+00
9.33347702e-01 5.74668534e-02 4.67153847e-01 -1.00129402e+00
1.51902831e+00 -1.34574682e-01 -1.18309331e+00 3.71273756e-02
-4.16538149e-01 -4.07748342e-01 5.66235185e-01 2.78346658e-01
-1.02332783e+00 -7.69706190e-01 -9.03576672e-01 3.75452429e-01
-1.89087279e-02 -1.26991689e-01 3.49299431e-01 1.33542407e+00
-9.75454688e-01 1.01742947e+00 -8.90798807e-01 -4.61045384e-01
3.30365688e-01 -1.36557177e-01 -3.36678088e-01 3.60736191e-01
-1.02041888e+00 8.65991414e-01 -3.03144366e-01 6.09308302e-01
-1.95414111e-01 -7.58804142e-01 -7.12348759e-01 -3.72818500e-01
-4.21151906e-01 -6.52512968e-01 9.10960555e-01 -5.80597401e-01
-2.33984089e+00 9.03488040e-01 -4.73247468e-01 1.63037658e-01
1.79800913e-01 -3.25440317e-01 -9.49426353e-01 9.35306609e-01
-4.10799742e-01 1.04615681e-01 1.00775206e+00 -4.28612292e-01
2.82280087e-01 -5.95562339e-01 -8.11387479e-01 9.25046206e-02
-2.08477184e-01 3.39372456e-01 3.05557549e-02 2.32336089e-01
4.50491533e-02 -8.43452096e-01 -8.69711637e-02 4.43292171e-01
1.23489894e-01 5.96017897e-01 3.11592698e-01 -8.50214660e-01
1.05462945e+00 -2.05665183e+00 -2.72012115e-01 4.43929493e-01
4.04722124e-01 2.59660572e-01 2.65415519e-01 -3.16434503e-02
-2.88071092e-02 2.77339250e-01 1.72891572e-01 2.44044885e-01
-4.57480848e-01 -1.10197842e-01 2.46840790e-01 1.18699419e+00
-2.58229822e-01 1.11276472e+00 -6.79700315e-01 -6.64606035e-01
8.03072274e-01 9.38818276e-01 1.30854845e-01 3.76314491e-01
8.25142860e-01 1.10257125e+00 9.63627920e-02 9.47325289e-01
4.60826516e-01 1.78115159e-01 2.09554628e-01 -4.01305079e-01
-1.90722153e-01 -1.17301926e-01 -6.37341917e-01 1.54051173e+00
-1.31221488e-02 6.36199653e-01 2.32005447e-01 -2.96445638e-01
1.37609971e+00 7.21199691e-01 1.05663896e+00 -8.75780523e-01
6.24273121e-01 -2.85570055e-01 -3.00493110e-02 -1.04324055e+00
-4.74608719e-01 -7.42922485e-01 1.79662347e-01 2.09127977e-01
-9.99418497e-02 3.05969603e-02 -2.57628798e-01 -3.46469909e-01
7.96774507e-01 3.64175253e-02 5.91168165e-01 -3.05487216e-01
3.68402898e-01 -4.89067554e-01 3.19116801e-01 3.00264001e-01
-1.25741673e+00 5.99729300e-01 2.46725515e-01 -5.31793535e-01
-2.37654567e-01 -9.94424760e-01 -4.69731301e-01 3.68105263e-01
-8.72115418e-02 -2.25864291e-01 -4.48599726e-01 2.04048499e-01
3.07728469e-01 2.11096242e-01 -7.57254124e-01 -2.38127306e-01
-7.28639737e-02 -8.09941232e-01 5.62207401e-01 6.09406590e-01
8.29128325e-02 -1.05518472e+00 -1.30250621e+00 3.84777516e-01
-1.86491638e-01 -1.29100180e+00 -1.46569923e-01 -1.33217886e-01
-1.13388920e+00 -1.16848254e+00 -6.46970093e-01 8.88354853e-02
2.13643163e-01 -1.75715275e-02 8.42915237e-01 -4.71180141e-01
-1.19683361e+00 1.13571858e+00 -1.08984280e-02 -8.69476974e-01
1.11768909e-01 -7.17522502e-01 3.07202220e-01 1.63539410e-01
7.77257442e-01 -7.63072550e-01 -1.24634767e+00 1.15987122e-01
1.33358359e-01 -7.38052368e-01 -9.05999020e-02 -3.37143093e-01
7.70570934e-01 -7.71219492e-01 3.69639039e-01 -3.15350235e-01
7.64711916e-01 -2.86860585e-01 -3.01114976e-01 -3.47721130e-01
-7.93772221e-01 -6.83866084e-01 3.01054031e-01 -4.92072701e-01
-7.47572184e-01 -6.36753067e-02 2.43286312e-01 -4.14260775e-01
-4.65759546e-01 -4.71852422e-02 4.40914184e-01 -2.85270005e-01
9.52842653e-01 -1.51402161e-01 7.56679416e-01 1.17897518e-01
2.08916932e-01 6.00567460e-01 9.72846270e-01 8.30685720e-02
1.70406520e-01 5.02564311e-01 2.90701807e-01 -1.26143634e+00
-4.95945066e-01 -5.65090835e-01 -8.05977225e-01 -9.33645308e-01
9.79159892e-01 -9.57641304e-01 -1.35498834e+00 2.83659399e-01
-6.17094934e-01 -3.71632800e-02 -3.79303217e-01 1.08955979e+00
-1.92250371e-01 2.35850215e-01 -6.56742871e-01 -1.29784226e+00
-9.37272847e-01 -1.49097547e-01 3.22672546e-01 6.59235299e-01
-9.55556929e-01 -1.01871765e+00 5.75140357e-01 3.66492689e-01
9.26474571e-01 1.03356206e+00 1.24376185e-01 -1.07279263e-01
5.45492291e-01 -6.73696220e-01 7.07882121e-02 1.94602892e-01
6.25600219e-01 3.73942852e-01 -1.39920688e+00 1.37314975e-01
5.13356388e-01 -8.85777250e-02 2.81345807e-02 8.66982102e-01
1.08179748e+00 -2.29031350e-02 -1.12288646e-01 5.33091784e-01
1.25601900e+00 6.78621292e-01 9.63504672e-01 -3.21625441e-01
4.34813708e-01 4.82506365e-01 9.74824950e-02 7.31078565e-01
-7.15103149e-02 3.35678346e-02 -1.98947210e-02 -3.36112440e-01
2.40060389e-01 3.60322118e-01 4.77987170e-01 6.09243989e-01
-6.66931152e-01 4.22886819e-01 -4.14302379e-01 -1.59305468e-01
-9.07034814e-01 -1.09009778e+00 -4.65851992e-01 2.30411983e+00
8.27945888e-01 -5.33945084e-01 5.64719796e-01 -4.30291779e-02
8.87403667e-01 -2.41967976e-01 -8.22461128e-01 -8.98424983e-01
3.07052970e-01 7.42879927e-01 4.23083693e-01 1.88698530e-01
-9.54929113e-01 -3.73047292e-02 9.04451180e+00 -8.33305299e-01
-1.30046868e+00 -7.86943510e-02 5.41996479e-01 -4.96852428e-01
2.90336460e-01 -4.91737574e-01 -3.57237399e-01 5.12391686e-01
1.90693951e+00 -3.58344823e-01 3.87297302e-01 7.66192436e-01
6.49570167e-01 -4.71500158e-01 -8.99548352e-01 1.30108154e+00
2.41397530e-01 -7.81422973e-01 -9.32082772e-01 -1.08878918e-01
-1.75546497e-01 -2.89772272e-01 -2.11964056e-01 -1.65158346e-01
-8.05028737e-01 -8.01605821e-01 -1.07831106e-01 9.09570634e-01
1.44484663e+00 -4.25067186e-01 3.64567369e-01 -4.37955946e-01
-9.49057937e-01 1.92842916e-01 -8.13342258e-02 -2.18732789e-01
1.13975190e-01 9.18108582e-01 -1.71613574e-01 -3.30708504e-01
6.82767212e-01 5.38796425e-01 -1.51881024e-01 1.00515211e+00
-7.53957704e-02 4.47111458e-01 -2.82416523e-01 1.15157127e-01
-7.41364896e-01 -3.99900317e-01 3.90720069e-01 1.16846073e+00
1.23656578e-01 6.79380953e-01 -2.40975693e-01 1.11626685e+00
3.35547924e-01 1.15733758e-01 -6.77566886e-01 -1.94565848e-01
3.97014260e-01 1.66941702e+00 -8.03472102e-01 -2.00346664e-01
-5.06130338e-01 9.03858364e-01 -5.42801797e-01 2.88334310e-01
-9.30267930e-01 -7.80147433e-01 8.20818067e-01 2.60564927e-02
-4.31107432e-01 -2.93067917e-02 -5.46746016e-01 -9.16466653e-01
-1.36724383e-01 -4.26800966e-01 2.30076000e-01 -7.10274756e-01
-7.84232914e-01 1.83736905e-01 -1.57399237e-01 -1.04581153e+00
6.43375963e-02 -2.43126690e-01 -7.08755016e-01 1.14770126e+00
-1.40858555e+00 1.98844925e-01 -9.36457753e-01 8.14588845e-01
-2.41766229e-01 4.28377181e-01 1.52936423e+00 2.74095774e-01
-1.00464606e+00 2.93497264e-01 -6.31389797e-01 1.59205813e-02
1.05166769e+00 -1.03810644e+00 -1.05795279e-01 3.53459328e-01
-6.03724480e-01 5.59566557e-01 3.99876356e-01 -3.32398713e-01
-1.65666461e+00 -6.71471477e-01 5.34832895e-01 -4.29960221e-01
2.80808479e-01 3.74130020e-03 -6.23752832e-01 2.53143877e-01
2.50342995e-01 4.97960180e-01 1.68489790e+00 -2.66220689e-01
-6.40975684e-02 -4.44082946e-01 -1.65884912e+00 1.39743714e-02
2.57259369e-01 -5.70380867e-01 -3.16144347e-01 -7.03967363e-02
1.42360151e-01 -4.37537342e-01 -1.73746419e+00 1.97971314e-01
1.34446609e+00 -8.51371586e-01 7.94753671e-01 -2.21595809e-01
1.58512276e-02 -3.54124163e-03 4.16503131e-01 -1.01652670e+00
-5.01997888e-01 -1.03255177e+00 -3.39077055e-01 1.07783842e+00
1.12015866e-02 -9.62975681e-01 3.90365124e-01 1.69912219e+00
3.25259596e-01 -2.01809570e-01 -7.44689345e-01 -5.85185528e-01
-4.36974585e-01 -1.21742971e-01 -7.03329146e-02 6.68506324e-01
1.08099139e+00 1.93781897e-01 -3.46654028e-01 -1.64393470e-01
6.60341740e-01 4.79600132e-02 1.88876599e-01 -1.35282969e+00
2.68862933e-01 3.25542726e-02 -4.96148705e-01 1.61490485e-01
-4.90605474e-01 -3.52784365e-01 -1.61505759e-01 -1.30119503e+00
-7.99596310e-02 2.97578245e-01 -8.09497714e-01 4.30135667e-01
-2.48044342e-01 5.90404153e-01 -1.20821372e-01 -7.57827163e-02
-1.41481876e-01 -1.13230655e-02 1.02173221e+00 5.20877182e-01
-8.74675989e-01 -2.37253010e-01 -7.22703755e-01 2.40456551e-01
1.11454558e+00 -3.75493586e-01 -2.61080384e-01 5.63836575e-01
-2.08673209e-01 5.72099507e-01 3.27095747e-01 -1.36657047e+00
-1.69703558e-01 -1.04124971e-01 1.29085052e+00 1.27748057e-01
2.67467082e-01 -8.58564556e-01 4.52036202e-01 8.51836741e-01
-1.67037055e-01 3.60137969e-02 4.59723651e-01 2.55841345e-01
-3.90000530e-02 5.17506540e-01 1.22730005e+00 -4.99940246e-01
-2.63495803e-01 1.75339848e-01 -9.01566923e-01 2.69795395e-02
1.23428333e+00 -4.52934116e-01 -4.20295507e-01 -3.77517551e-01
-9.71416533e-01 -1.44037426e-01 5.64424135e-02 2.33715698e-01
7.46115327e-01 -1.29707706e+00 -3.64096850e-01 6.21144593e-01
-5.87360002e-04 -7.79947042e-01 4.83129799e-01 1.45987153e+00
-5.13278902e-01 2.45143756e-01 -7.20789731e-01 -6.07789457e-01
-1.20645440e+00 4.35514867e-01 7.34070480e-01 8.19046497e-01
-6.81075692e-01 4.62913841e-01 -8.19779158e-01 5.79184890e-01
1.20866016e-01 9.07902643e-02 -4.54818279e-01 2.81100988e-01
1.09889364e+00 8.90429854e-01 -1.33367434e-01 -2.63142943e-01
-6.18049681e-01 7.14916110e-01 7.98091948e-01 -8.05224329e-02
6.84511721e-01 -7.56689250e-01 -1.75280541e-01 8.84817481e-01
1.24740219e+00 -2.34830096e-01 -8.84935558e-01 4.24337417e-01
-5.94669044e-01 -2.78716445e-01 3.15520465e-02 -8.76610518e-01
-1.25732148e+00 8.15810561e-01 1.29843616e+00 1.01510666e-01
1.37501442e+00 -6.36511624e-01 4.66192931e-01 -2.11685464e-01
2.70358734e-02 -1.16340947e+00 -1.41180873e-01 -4.82081831e-01
5.01824677e-01 -9.49595749e-01 1.57500505e-01 -3.42435211e-01
-7.26889133e-01 1.57858217e+00 3.86699975e-01 1.67832673e-02
9.79427397e-01 3.48404795e-01 7.62632251e-01 -1.59970209e-01
-5.03731191e-01 1.79763213e-02 1.65991738e-01 8.81889880e-01
8.45137894e-01 6.85369298e-02 -5.03891528e-01 1.50573269e-01
1.76075831e-01 6.23232543e-01 5.26427746e-01 8.30566823e-01
-4.57965493e-01 -3.51525575e-01 -3.16274673e-01 4.31086928e-01
-8.73039424e-01 2.42001384e-01 -3.53959680e-01 4.62021708e-01
-3.40268493e-01 1.45185912e+00 1.40692681e-01 -2.95991451e-01
5.47588050e-01 8.13416839e-01 7.50604272e-01 -2.86507666e-01
-5.50157726e-01 3.71411681e-01 -1.61831707e-01 -1.38003910e+00
-8.64072382e-01 -4.85610396e-01 -1.11640120e+00 -7.35911876e-02
3.53115380e-01 -3.02677453e-01 1.00526488e+00 3.70597154e-01
6.43689454e-01 3.53069007e-01 8.06930363e-01 -7.67265856e-01
-4.28622402e-03 -7.65683055e-01 -1.05191219e+00 3.29568833e-01
4.25754577e-01 -1.76678896e-01 -6.34726405e-01 2.49694794e-01] | [13.907142639160156, 2.8982560634613037] |
7e05c80e-5940-46b9-a794-217fd223233c | asm-adaptive-skinning-model-for-high-quality | 2304.09423 | null | https://arxiv.org/abs/2304.09423v1 | https://arxiv.org/pdf/2304.09423v1.pdf | ASM: Adaptive Skinning Model for High-Quality 3D Face Modeling | The research fields of parametric face models and 3D face reconstruction have been extensively studied. However, a critical question remains unanswered: how to tailor the face model for specific reconstruction settings. We argue that reconstruction with multi-view uncalibrated images demands a new model with stronger capacity. Our study shifts attention from data-dependent 3D Morphable Models (3DMM) to an understudied human-designed skinning model. We propose Adaptive Skinning Model (ASM), which redefines the skinning model with more compact and fully tunable parameters. With extensive experiments, we demonstrate that ASM achieves significantly improved capacity than 3DMM, with the additional advantage of model size and easy implementation for new topology. We achieve state-of-the-art performance with ASM for multi-view reconstruction on the Florence MICC Coop benchmark. Our quantitative analysis demonstrates the importance of a high-capacity model for fully exploiting abundant information from multi-view input in reconstruction. Furthermore, our model with physical-semantic parameters can be directly utilized for real-world applications, such as in-game avatar creation. As a result, our work opens up new research directions for the parametric face models and facilitates future research on multi-view reconstruction. | ['Wei Yang', 'Zhongqian Sun', 'Jingkai Zhou', 'Xinghan Chen', 'Tianyang Shi', 'Hong Shang', 'Kai Yang'] | 2023-04-19 | null | null | null | null | ['3d-face-reconstruction', 'face-model', 'face-reconstruction'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [-2.97558084e-02 -2.59992871e-02 8.41400027e-02 -2.22748250e-01
-5.69429874e-01 -4.82521325e-01 3.52044255e-01 -8.41426313e-01
1.01178311e-01 2.83919156e-01 1.14846043e-01 -3.35174077e-03
-9.81326476e-02 -7.34716713e-01 -7.51936972e-01 -6.14397943e-01
2.11365104e-01 6.94172561e-01 1.33162960e-01 -4.30559516e-01
-1.05307966e-01 9.35512304e-01 -1.68667567e+00 1.91581100e-01
2.49221772e-01 9.68468010e-01 1.87911898e-01 4.83333528e-01
1.10219099e-01 1.70377150e-01 -8.47112313e-02 -7.30941057e-01
5.59413671e-01 -2.01343521e-01 -6.69938862e-01 3.84816021e-01
7.33828306e-01 -4.54796821e-01 -1.98759809e-01 6.71111822e-01
7.32255638e-01 1.03075439e-02 3.27458858e-01 -1.18481255e+00
-5.01270473e-01 1.24841549e-01 -9.42952693e-01 -2.01721132e-01
3.08137774e-01 1.38507888e-01 7.32627809e-01 -1.31930208e+00
8.51139665e-01 1.40586579e+00 6.68532848e-01 9.23701108e-01
-1.07667744e+00 -7.62492597e-01 1.17131725e-01 1.45416886e-01
-1.49471772e+00 -8.03994417e-01 9.59150493e-01 -1.43857107e-01
6.97213113e-01 4.22017068e-01 8.00612748e-01 1.07973957e+00
-1.43370237e-02 4.48868334e-01 1.17212319e+00 -3.02265793e-01
2.71893535e-02 3.38461585e-02 -5.80627978e-01 1.10268438e+00
5.64081147e-02 1.54206529e-01 -9.52011168e-01 -1.61485270e-01
1.28855717e+00 -2.81275243e-01 -1.84504345e-01 -7.24887013e-01
-8.55619431e-01 5.80566764e-01 5.93850724e-02 -2.48027071e-01
-2.12225735e-01 3.11869204e-01 2.17913494e-01 2.24917784e-01
7.40362048e-01 9.23611894e-02 -5.27340949e-01 -1.76675558e-01
-1.00833690e+00 1.75429597e-01 5.28743505e-01 1.03025961e+00
7.68655777e-01 7.38317147e-02 5.47588170e-01 1.15789628e+00
5.05889177e-01 7.51465797e-01 -6.98927641e-02 -1.32360482e+00
2.41951674e-01 3.24199080e-01 -1.18000008e-01 -8.74052167e-01
-3.40136737e-01 -2.96340585e-01 -7.67201364e-01 2.21444756e-01
9.58333313e-02 1.96827605e-01 -7.00940073e-01 1.64732885e+00
8.56435537e-01 3.42152685e-01 -2.23867118e-01 1.08224142e+00
9.21614945e-01 3.05088311e-01 -2.89834529e-01 2.20586546e-03
1.41937375e+00 -9.85115647e-01 -3.25234890e-01 -1.53310254e-01
1.74175218e-01 -8.46894383e-01 1.26578534e+00 5.34713149e-01
-1.36297822e+00 -3.33340079e-01 -7.97450721e-01 -1.68216094e-01
2.40690336e-01 -1.17995761e-01 9.07942593e-01 8.68064940e-01
-1.35873497e+00 3.84984881e-01 -7.19881892e-01 -5.48295557e-01
6.02221668e-01 4.99871701e-01 -9.38606024e-01 -2.81238645e-01
-8.40779603e-01 7.34102905e-01 -2.86057115e-01 1.98017031e-01
-8.75092149e-01 -8.73457372e-01 -6.99228048e-01 -2.69450754e-01
4.82970953e-01 -1.22087991e+00 9.62355554e-01 -7.41696537e-01
-1.99206531e+00 1.29179573e+00 -4.85771187e-02 8.73085782e-02
5.66256523e-01 2.78988145e-02 -3.62608731e-01 4.13202316e-01
-3.14440638e-01 7.05579400e-01 1.07414496e+00 -1.74386489e+00
4.31585163e-02 -7.64678478e-01 1.96662396e-01 3.63127291e-01
-3.58323246e-01 1.02975331e-01 -9.11181986e-01 -5.62488794e-01
3.25625330e-01 -1.20260203e+00 -5.17844781e-02 7.48065889e-01
-1.71914250e-01 3.92949045e-01 9.30832088e-01 -6.07967257e-01
8.33438039e-01 -2.07451749e+00 5.45983672e-01 1.32325351e-01
3.87579858e-01 -5.72474226e-02 -2.10760340e-01 3.46667767e-01
1.64684623e-01 2.04914466e-01 -1.43471390e-01 -7.53077984e-01
-1.58804625e-01 3.14134002e-01 -6.39733598e-02 6.11203134e-01
-2.02280566e-01 8.20243776e-01 -4.72994387e-01 -6.99892879e-01
3.18012923e-01 7.87109733e-01 -9.69804764e-01 1.92758754e-01
-8.57133325e-03 7.17928708e-01 -3.45362455e-01 8.14504325e-01
1.17767453e+00 -3.91048968e-01 2.65896618e-01 -3.70379865e-01
1.75231636e-01 -3.63714516e-01 -1.19649434e+00 2.06209207e+00
-6.96192265e-01 1.06622145e-01 7.96111584e-01 -5.41919291e-01
8.77836883e-01 3.72076929e-02 7.14020848e-01 -4.81700271e-01
-8.96029621e-02 5.77072576e-02 -2.87609845e-01 -4.04798865e-01
6.41003430e-01 -5.79605639e-01 1.55273244e-01 3.23501170e-01
1.44207133e-02 -4.10891056e-01 -5.28111160e-01 1.98762536e-01
6.06176198e-01 4.35736746e-01 1.80153828e-02 -3.53532612e-01
2.93730170e-01 -5.85371673e-01 4.34801787e-01 9.48549435e-02
-3.95296365e-02 8.63634229e-01 2.09328204e-01 -4.24482793e-01
-9.34892595e-01 -1.18750072e+00 -3.05497348e-01 9.95415568e-01
3.05954456e-01 -5.95117092e-01 -7.63951302e-01 -6.14928603e-01
1.14226021e-01 1.50711626e-01 -7.02343822e-01 1.05444141e-01
-7.75415778e-01 -6.85596049e-01 4.90178883e-01 4.49682415e-01
4.77211833e-01 -7.36052632e-01 -4.76785809e-01 -2.61980981e-01
-1.08625852e-01 -1.30766416e+00 -6.21041834e-01 -4.99119014e-01
-1.01489043e+00 -1.00910139e+00 -5.80405772e-01 -5.07196546e-01
7.34545410e-01 4.86270130e-01 1.10373306e+00 4.16148782e-01
-1.43060848e-01 8.47398102e-01 -1.32256791e-01 -1.38093485e-02
-3.72695684e-01 -8.56106728e-02 2.17920333e-01 1.89726636e-01
-3.47617716e-01 -9.06032443e-01 -7.12951064e-01 7.12422073e-01
-8.28800321e-01 4.72205758e-01 2.70466417e-01 6.58253789e-01
8.38696182e-01 -4.83081937e-01 4.50796306e-01 -1.04204082e+00
6.22104108e-02 -3.71291250e-01 -4.75102633e-01 2.73170501e-01
-5.31130135e-01 -7.25591881e-03 5.01271546e-01 -2.50352055e-01
-1.22659433e+00 -3.71429920e-02 -4.56152588e-01 -7.18962133e-01
2.16771051e-01 1.57206208e-02 -6.29216850e-01 -5.71626484e-01
2.94961423e-01 1.03509106e-01 2.81633973e-01 -6.39805257e-01
6.47749841e-01 4.52331841e-01 3.33055198e-01 -9.86899078e-01
8.36367369e-01 9.67602074e-01 3.41043055e-01 -1.06476200e+00
-3.64874512e-01 -1.78680688e-01 -6.56914473e-01 -6.03525281e-01
5.29144168e-01 -1.00946343e+00 -7.34148920e-01 5.35594463e-01
-9.97014344e-01 -3.70072395e-01 -1.95124581e-01 3.14136589e-04
-8.55969906e-01 7.02202559e-01 -5.22223830e-01 -6.01206481e-01
-3.69947255e-01 -1.20363748e+00 1.53637624e+00 2.40627155e-02
1.62614897e-01 -9.62613404e-01 -4.29801159e-02 9.81465220e-01
4.71021324e-01 1.37075588e-01 7.74961412e-01 2.60998845e-01
-7.88259804e-01 2.46894598e-01 -6.65550008e-02 1.63071036e-01
-9.53715388e-03 8.81873071e-02 -1.17736650e+00 -4.54833388e-01
-4.69388627e-02 -2.89228976e-01 6.60810888e-01 1.14368290e-01
1.47020793e+00 -9.67383683e-02 -2.88080573e-01 1.22247839e+00
1.35046756e+00 -3.48886758e-01 8.07741225e-01 -1.85865872e-02
9.40896332e-01 5.39455831e-01 3.07789683e-01 5.94565809e-01
5.75010777e-01 1.13535094e+00 7.86728323e-01 -1.28430709e-01
-4.11449909e-01 -4.48929489e-01 1.67076081e-01 1.14909351e+00
-5.09448588e-01 -1.46851763e-01 -6.58615649e-01 1.85607821e-01
-1.51665640e+00 -8.23651135e-01 8.55742991e-02 2.11966205e+00
5.34666359e-01 -2.73649931e-01 6.80694804e-02 -2.77936816e-01
5.08764744e-01 3.20780575e-01 -6.34101748e-01 -1.87279135e-01
-2.46978372e-01 4.63692516e-01 2.28279591e-01 4.84689802e-01
-7.62366056e-01 1.04945815e+00 6.44604731e+00 1.03923535e+00
-1.08642352e+00 4.40980434e-01 5.62168241e-01 -2.30869085e-01
-6.21791005e-01 3.84568125e-02 -7.69757867e-01 2.33234406e-01
7.33145714e-01 1.16362602e-01 8.17548275e-01 7.61906922e-01
2.04169035e-01 -1.57649219e-02 -8.00044358e-01 1.33112478e+00
3.49495143e-01 -1.52223325e+00 2.59862274e-01 4.35966820e-01
6.75444484e-01 -9.83528569e-02 2.02195257e-01 -2.41590180e-02
-1.86536655e-01 -1.07667089e+00 8.15991282e-01 6.03833437e-01
1.32133746e+00 -6.33803606e-01 1.17410891e-01 1.78661540e-01
-1.40763998e+00 2.59837180e-01 -3.80051911e-01 2.86446005e-01
2.91215390e-01 3.05143476e-01 -4.93843704e-01 7.42250085e-01
5.59346378e-01 5.81288993e-01 -5.49321771e-01 5.66260219e-01
2.39647299e-01 3.72561812e-01 -3.84332836e-01 4.79694128e-01
-3.81072462e-01 -2.48959631e-01 4.50521946e-01 7.23509908e-01
3.49572390e-01 1.89805865e-01 -5.97195588e-02 6.80883527e-01
-2.73680449e-01 1.53348699e-01 -5.88090777e-01 4.44521934e-01
5.41044593e-01 1.40102077e+00 -7.55552411e-01 8.16600993e-02
-4.53857034e-01 1.06613934e+00 4.75206554e-01 4.31282781e-02
-9.73115504e-01 5.68334222e-01 8.22437346e-01 7.48809993e-01
2.06962124e-01 -4.12974507e-01 -2.00764269e-01 -1.34005308e+00
1.02822874e-02 -7.30212390e-01 1.09396122e-01 -8.85243952e-01
-1.23634565e+00 6.30603254e-01 1.23449184e-01 -1.18292522e+00
1.04949504e-01 -6.42059803e-01 -1.98994517e-01 5.94670534e-01
-1.50835133e+00 -1.65291893e+00 -4.24229354e-01 6.27304554e-01
6.06974125e-01 -9.35420915e-02 1.01460564e+00 4.76065427e-01
-5.54394543e-01 7.04007566e-01 -1.55271843e-01 -3.23404938e-01
6.74541950e-01 -7.12838590e-01 4.74860102e-01 4.66210157e-01
1.50648698e-01 4.77983892e-01 3.56053978e-01 -5.23253441e-01
-2.10988927e+00 -9.58274186e-01 1.10389091e-01 -7.14749098e-01
2.01553240e-01 -4.39637423e-01 -7.07490683e-01 5.24602592e-01
-3.88494991e-02 5.00226617e-01 8.23041856e-01 -5.48234321e-02
-4.04733658e-01 -1.49386287e-01 -1.37350929e+00 5.05233765e-01
1.67317605e+00 -5.24358153e-01 1.04571760e-01 1.33488551e-01
5.71994543e-01 -6.83094501e-01 -1.05586016e+00 4.08908039e-01
9.28568065e-01 -1.22427249e+00 1.28427255e+00 -4.53444272e-01
3.38170648e-01 -1.01048820e-01 -5.16111732e-01 -1.12692869e+00
-8.99985731e-02 -7.94618607e-01 -2.90856689e-01 9.44056988e-01
5.35320565e-02 -4.64120507e-01 1.00927651e+00 6.23346090e-01
-1.00385450e-01 -1.17758024e+00 -1.17103732e+00 -7.55215168e-01
1.02059245e-01 -4.12403554e-01 8.93277287e-01 8.99941087e-01
-4.90221143e-01 2.02743888e-01 -7.10944712e-01 1.73755184e-01
7.06048667e-01 2.19094291e-01 1.12637758e+00 -1.26104951e+00
-3.92398775e-01 -5.25394082e-02 -2.44379357e-01 -1.17005777e+00
3.19039732e-01 -9.25772130e-01 -4.95459527e-01 -1.30588543e+00
4.07498956e-01 -7.15421081e-01 2.38451853e-01 2.46928170e-01
2.09465280e-01 5.57490706e-01 4.24845129e-01 1.63568288e-01
-4.86526012e-01 8.31153393e-01 1.73225999e+00 2.74728745e-01
1.92357823e-01 -1.67306662e-01 -7.76413798e-01 7.38040745e-01
6.45264983e-01 -2.34906048e-01 -4.57272232e-01 -6.48642361e-01
3.08369040e-01 3.29130739e-01 4.94343013e-01 -6.17208302e-01
-7.28477016e-02 -4.96202976e-01 1.00177459e-01 -2.30441928e-01
1.08891821e+00 -8.78559411e-01 6.41997337e-01 -6.33793175e-02
3.84860724e-01 2.61014223e-01 2.95064718e-01 5.60195267e-01
3.02024871e-01 9.36602131e-02 9.14473295e-01 -3.80819261e-01
-6.31406009e-01 6.68461204e-01 2.21950993e-01 1.01209976e-01
9.29762900e-01 -4.00009304e-01 -1.76436111e-01 -4.30422217e-01
-8.06004167e-01 -1.19801298e-01 1.02819347e+00 4.01093960e-01
9.15891588e-01 -1.40658331e+00 -7.33588040e-01 5.72518468e-01
2.43737716e-02 -4.70882617e-02 7.26831496e-01 6.56659484e-01
-6.98121667e-01 -5.96447177e-02 -2.06003115e-01 -6.30872905e-01
-1.42118311e+00 3.49412799e-01 3.97463381e-01 -6.37547374e-02
-8.37569654e-01 8.12999308e-01 3.56361777e-01 -7.20187366e-01
-1.80823982e-01 7.24751204e-02 2.41586357e-01 -2.62218446e-01
2.15705022e-01 4.75202203e-01 2.26727411e-01 -1.04449594e+00
-4.26309347e-01 1.10764349e+00 2.42731318e-01 -1.08638383e-01
1.47462416e+00 -1.90943226e-01 -1.38635546e-01 7.37233786e-03
9.69150364e-01 3.79544795e-01 -1.47224987e+00 3.88374105e-02
-6.97591245e-01 -7.63942778e-01 4.51120362e-02 -6.47988498e-01
-1.52241039e+00 9.08418775e-01 4.14534450e-01 -4.09004062e-01
1.21092701e+00 2.27942079e-01 8.89756739e-01 -3.27154733e-02
9.92942691e-01 -8.40439200e-01 3.20586205e-01 2.07459986e-01
1.26921070e+00 -9.83448207e-01 1.58647418e-01 -8.75942409e-01
-6.49898052e-01 1.04487228e+00 8.34079504e-01 1.18271969e-01
9.85519648e-01 5.30600011e-01 -8.83664489e-02 -5.26990950e-01
-6.54089808e-01 2.34531000e-01 6.49294481e-02 5.45357525e-01
7.01940199e-03 2.57952213e-01 1.97208419e-01 6.36609375e-01
-4.72521037e-01 -2.43853495e-01 4.32174176e-01 5.60681760e-01
-1.85900182e-02 -1.37964106e+00 -4.24281180e-01 2.92570233e-01
-1.51836589e-01 1.92211464e-03 -2.07052827e-01 8.54372144e-01
1.86015517e-02 6.73822999e-01 -7.69009739e-02 -4.80089724e-01
4.39573050e-01 1.16026765e-02 1.08720315e+00 -6.68461442e-01
-3.93760949e-01 6.29001409e-02 1.42248971e-02 -6.63528085e-01
-5.00470519e-01 -6.44543111e-01 -9.82464135e-01 -7.40477562e-01
-4.47352350e-01 -2.33642757e-01 7.09512353e-01 5.70711493e-01
8.98562431e-01 2.21400503e-02 6.85726523e-01 -1.12229121e+00
-3.48870993e-01 -5.42974234e-01 -6.03907108e-01 2.29828805e-01
1.45816440e-02 -1.01419055e+00 -6.42192587e-02 -7.34194890e-02] | [13.122142791748047, -0.04140983521938324] |
3c76810a-117e-4f2a-82f7-97ac4ad42cf4 | id-free-person-similarity-learning | null | null | http://openaccess.thecvf.com//content/CVPR2022/html/Shuai_Id-Free_Person_Similarity_Learning_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Shuai_Id-Free_Person_Similarity_Learning_CVPR_2022_paper.pdf | Id-Free Person Similarity Learning | Learning a unified person detection and re-identification model is a key component of modern trackers. However, training such models usually relies on the availability of training images / videos that are manually labeled with both person boxes and their identities. In this work, we explore training such a model by only using person box annotations, thus removing the necessity of manually labeling a training dataset with additional person identity annotation as these are expensive to collect. To this end, we present a contrastive learning framework to learn person similarity without using manually labeled identity annotations. First, we apply image-level augmentation to images on public person detection datasets, based on which we learn a strong model for general person detection as well as for short-term person re-identification. To learn a model capable of longer-term re-identification, we leverage the natural appearance evolution of each person in videos to serve as instance-level appearance augmentation in our contrastive loss formulation. Without access to the target dataset or person identity annotation, our model achieves competitive results compared to existing fully-supervised state-of-the-art methods on both person search and person tracking tasks. Our model also shows promising results for saving the annotation cost that is needed to achieve a certain level of performance on the person search task. | ['Joseph Tighe', 'Kaustav Kundu', 'Xinyu Li', 'Bing Shuai'] | 2022-01-01 | null | null | null | cvpr-2022-1 | ['person-search'] | ['computer-vision'] | [ 1.76788852e-01 -2.58958280e-01 -7.03569353e-02 -4.68479604e-01
-5.71564674e-01 -7.68406630e-01 6.85041726e-01 8.26469585e-02
-7.56697297e-01 5.73476493e-01 -2.23361757e-02 2.39300743e-01
3.89977187e-01 -4.03347880e-01 -9.27770138e-01 -3.34186345e-01
1.68309987e-01 5.90914607e-01 5.14361747e-02 1.14378236e-01
-2.84393460e-01 2.78083503e-01 -1.72039223e+00 -9.80422869e-02
8.03300381e-01 8.91319096e-01 -7.16878697e-02 5.98995864e-01
2.82866240e-01 3.58553052e-01 -5.68118513e-01 -9.41439271e-01
6.97984576e-01 -1.67896599e-01 -5.99578857e-01 4.41725254e-01
1.40393043e+00 -6.28089011e-01 -5.78655958e-01 1.08425164e+00
4.51426357e-01 2.48251900e-01 4.96254146e-01 -1.32517123e+00
-5.99900126e-01 1.19130902e-01 -8.00664246e-01 5.17218187e-02
5.34076393e-01 4.15843397e-01 8.10300827e-01 -7.22084701e-01
4.71736997e-01 1.10375869e+00 9.57153201e-01 8.67146075e-01
-1.46935046e+00 -8.85967612e-01 4.35032606e-01 1.62103131e-01
-1.54680240e+00 -6.89184666e-01 5.49515963e-01 -6.19919538e-01
5.61767697e-01 1.58137187e-01 6.70721233e-01 1.17253125e+00
-5.77737808e-01 8.51257324e-01 8.14027131e-01 -3.32863688e-01
-2.39258409e-01 4.49669182e-01 4.62849081e-01 7.09244013e-01
5.65257430e-01 2.78186440e-01 -4.58072633e-01 -1.03638247e-01
6.89267755e-01 1.70876071e-01 -1.80410981e-01 -5.76796949e-01
-8.89638186e-01 4.16153014e-01 4.56922084e-01 -1.49844103e-02
-1.36503279e-01 2.25286707e-01 4.81821775e-01 1.11851096e-01
4.83265758e-01 1.41117096e-01 -2.43718207e-01 1.23209573e-01
-1.12575698e+00 3.76214027e-01 4.82358426e-01 1.05966353e+00
8.47270310e-01 -1.87601224e-01 -5.40049374e-01 8.70834351e-01
1.16813466e-01 7.13621497e-01 2.42025793e-01 -6.37295961e-01
2.93620199e-01 7.39088953e-01 4.68485504e-01 -6.91665769e-01
-2.33137950e-01 -7.84211159e-01 -5.93728900e-01 8.10924768e-02
8.44969869e-01 -4.72396947e-02 -1.01168740e+00 2.12986541e+00
4.85208571e-01 5.20440757e-01 -2.77370363e-01 8.45504463e-01
3.88629735e-01 1.92569911e-01 2.11924478e-01 7.91068822e-02
1.76928926e+00 -1.17752993e+00 -5.09162903e-01 -4.27343011e-01
4.63361591e-01 -4.90349144e-01 9.22271788e-01 3.18248570e-02
-9.71197009e-01 -8.84213567e-01 -8.53540897e-01 -1.13412067e-01
-3.56565595e-01 5.32405794e-01 5.99877417e-01 1.04940772e+00
-9.81367111e-01 3.39645982e-01 -7.56949425e-01 -6.07678652e-01
5.57168663e-01 4.89338428e-01 -5.08981526e-01 -2.00498298e-01
-8.42111409e-01 7.65958190e-01 1.53378755e-01 4.43649963e-02
-8.51122975e-01 -8.60419154e-01 -1.01533592e+00 4.55867127e-02
4.68401462e-01 -9.52699065e-01 1.10215962e+00 -1.05780160e+00
-1.03110147e+00 1.24431777e+00 -3.85762304e-01 -6.37879729e-01
8.15405190e-01 -4.77593899e-01 -4.04643714e-01 -3.55743691e-02
1.85405955e-01 8.02451134e-01 1.01396322e+00 -1.23819983e+00
-8.39400113e-01 -6.20897949e-01 2.64912993e-01 5.91699705e-02
-6.36886179e-01 2.16461107e-01 -1.06631100e+00 -6.30195260e-01
-5.11032224e-01 -1.24353659e+00 9.47707798e-03 3.28182787e-01
-2.07230732e-01 -3.78161550e-01 8.13323140e-01 -1.08626962e+00
8.90011072e-01 -1.97161245e+00 -6.54157698e-02 1.56448424e-01
1.38538033e-01 4.32765156e-01 -1.65583730e-01 -1.56337023e-01
1.11207739e-01 -2.73615092e-01 -5.96571527e-02 -1.03020322e+00
2.34725494e-02 -2.23706871e-01 1.85367814e-03 6.01014197e-01
1.40501171e-01 9.55531895e-01 -8.28031719e-01 -4.31200713e-01
1.85203910e-01 5.71776569e-01 -5.69662213e-01 2.79828578e-01
-4.89010289e-02 7.29817927e-01 1.52215110e-02 7.29410172e-01
6.40165448e-01 -3.65276009e-01 3.04316059e-02 -4.44145083e-01
4.85356301e-02 -1.16297685e-01 -1.20223200e+00 1.82632256e+00
-2.56042540e-01 5.37932038e-01 4.80950736e-02 -7.56597340e-01
5.04210293e-01 2.62182683e-01 5.84211588e-01 -3.73960853e-01
3.02900653e-02 8.30172561e-03 -1.76112533e-01 -4.30525988e-02
3.81930381e-01 2.09755778e-01 5.22556752e-02 4.53805983e-01
1.87913671e-01 6.90849960e-01 5.23921192e-01 2.05096394e-01
8.30442250e-01 2.92465210e-01 -6.47362694e-02 -9.00498554e-02
7.90881395e-01 -2.46839032e-01 5.18804729e-01 9.03167963e-01
-4.08020079e-01 3.93237978e-01 -2.70842433e-01 -5.25325060e-01
-1.24317181e+00 -8.48129392e-01 -1.29049614e-01 1.32873619e+00
2.41167292e-01 -3.75599474e-01 -8.59227598e-01 -8.58417571e-01
1.20022409e-01 9.79624614e-02 -5.72752655e-01 -7.54400492e-02
-5.96975267e-01 -7.75056839e-01 7.11561620e-01 6.61209881e-01
6.57391012e-01 -5.50954819e-01 -2.43840590e-01 -1.09137207e-01
-2.70806253e-01 -1.45525050e+00 -1.20102155e+00 -4.18376058e-01
-3.83933723e-01 -1.22217786e+00 -1.08893752e+00 -7.88655221e-01
1.04495823e+00 3.19103152e-01 9.24617231e-01 1.77491680e-01
-5.70816517e-01 9.08428550e-01 8.60831961e-02 -3.33172351e-01
9.26684439e-02 -2.83472668e-02 4.42314774e-01 4.19771373e-01
4.78284985e-01 -3.72515708e-01 -9.14511621e-01 3.67938787e-01
-4.39208716e-01 -5.82053959e-02 4.18290079e-01 7.81797647e-01
3.43560487e-01 -1.41926378e-01 2.83854336e-01 -6.09435558e-01
3.75888944e-02 2.16656346e-02 -8.47151875e-01 4.84015852e-01
-5.01036882e-01 -1.21001936e-01 2.54028231e-01 -7.11864829e-01
-1.12630618e+00 4.53371793e-01 7.88535699e-02 -4.79838878e-01
-1.23321973e-01 -1.80095375e-01 -2.40075529e-01 -3.87946546e-01
5.11215210e-01 2.58180559e-01 -1.24846548e-01 -6.94407284e-01
4.02682811e-01 5.16927421e-01 1.00977576e+00 -5.77828884e-01
1.28312016e+00 6.71548665e-01 -2.54186720e-01 -5.64358711e-01
-1.06618690e+00 -8.84593248e-01 -8.75901520e-01 -2.47328997e-01
8.29773545e-01 -1.36879170e+00 -9.69656408e-01 6.73921347e-01
-1.07663023e+00 -1.34728655e-01 -1.74105316e-01 3.60417932e-01
-2.31412992e-01 5.88630915e-01 -5.44170141e-01 -9.87767994e-01
-5.02047300e-01 -7.26616919e-01 1.26349521e+00 3.85590434e-01
-2.26387475e-02 -9.03532505e-01 1.43878078e-02 6.01751864e-01
2.54957050e-01 1.33414134e-01 5.01014329e-02 -5.76422155e-01
-6.88743293e-01 -7.00276673e-01 -3.97334546e-01 2.72433072e-01
2.09600225e-01 -5.32193780e-01 -1.15286922e+00 -8.68980110e-01
-4.03911233e-01 -3.94719511e-01 9.94186878e-01 1.54851615e-01
1.01305795e+00 -2.58843154e-01 -7.16190398e-01 8.28481495e-01
1.15752089e+00 -2.47277081e-01 1.44519523e-01 2.67896712e-01
1.05018413e+00 6.73871040e-01 3.61610979e-01 3.29094350e-01
4.93742794e-01 1.16713881e+00 -1.44422734e-02 -3.49040747e-01
-4.78446275e-01 -5.08476615e-01 3.64106208e-01 -8.43760446e-02
-3.17610860e-01 6.42723888e-02 -6.42637968e-01 5.71546316e-01
-2.04831624e+00 -1.11601436e+00 1.79633290e-01 2.63356066e+00
8.90895426e-01 -5.26659153e-02 7.35509574e-01 -1.88512206e-01
1.11109948e+00 -1.56936914e-01 -7.78574824e-01 6.56630337e-01
-1.26403859e-02 -1.37518227e-01 7.21001804e-01 4.41479176e-01
-1.59781408e+00 9.87918615e-01 5.52231789e+00 5.26247740e-01
-8.25000525e-01 2.31533155e-01 3.49975407e-01 -4.31238979e-01
3.07439417e-01 -1.07861914e-01 -1.28616953e+00 6.43437803e-01
6.52890801e-01 -5.84002063e-02 3.82634848e-01 8.19365025e-01
3.77258696e-02 1.70819536e-01 -1.45182526e+00 1.50003564e+00
2.83930540e-01 -1.04999912e+00 -1.24621101e-01 2.15198815e-01
7.43905723e-01 -3.26412290e-01 -2.18534656e-02 4.60607886e-01
2.38438889e-01 -7.94666350e-01 7.87573516e-01 4.65118766e-01
8.81812155e-01 -4.83357519e-01 6.29891515e-01 1.03235081e-01
-1.55535090e+00 -1.11879446e-01 -2.06053123e-01 9.39818248e-02
3.84390742e-01 2.33361349e-01 -3.85393113e-01 4.80693102e-01
8.43211710e-01 7.62011409e-01 -9.49208915e-01 1.32788563e+00
4.43530306e-02 3.39599788e-01 -5.50009489e-01 6.84251666e-01
-2.21353412e-01 -7.72922114e-02 4.08499658e-01 1.21233213e+00
1.27146110e-01 -1.84515104e-01 6.63806200e-01 8.43609154e-01
-3.81086975e-01 -1.68075144e-01 -2.89672911e-01 1.37488171e-01
4.25018519e-01 1.20942736e+00 -3.64976227e-01 -5.22743344e-01
-6.35669529e-01 1.36697030e+00 4.66948628e-01 3.65685999e-01
-9.29174125e-01 3.96209694e-02 8.53412211e-01 3.12887132e-01
1.36064574e-01 -9.10125077e-02 5.02532907e-02 -1.39302361e+00
3.12109202e-01 -7.03941822e-01 5.94633400e-01 -2.22447917e-01
-1.51818430e+00 3.37991983e-01 -2.27336027e-02 -1.12200749e+00
-2.58133948e-01 -4.17513281e-01 -3.91920477e-01 1.01811159e+00
-1.52491581e+00 -1.90012252e+00 -6.59446418e-01 8.17792177e-01
3.19768131e-01 -2.67446905e-01 4.67318386e-01 8.81660223e-01
-8.62818539e-01 1.24648309e+00 -2.60750115e-01 5.68315029e-01
1.16599882e+00 -1.13984454e+00 4.89848018e-01 1.13591278e+00
2.23809153e-01 9.62098777e-01 6.51953220e-01 -7.70965576e-01
-1.05548537e+00 -1.22560287e+00 7.34355152e-01 -8.02398801e-01
3.50579709e-01 -6.30466938e-01 -8.35580289e-01 9.54111159e-01
-2.08894610e-01 3.74662012e-01 7.77554929e-01 4.44269866e-01
-5.81829429e-01 -2.10872933e-01 -1.00065136e+00 4.79156792e-01
1.46660161e+00 -7.11878300e-01 -2.19259337e-01 4.53046173e-01
4.70981419e-01 -3.43535930e-01 -6.37923837e-01 3.16708714e-01
8.07377219e-01 -5.44139326e-01 1.38171601e+00 -5.43815911e-01
-3.36165488e-01 -5.89636862e-01 3.18861246e-01 -6.89650774e-01
-2.61577457e-01 -5.43813765e-01 -3.22334737e-01 1.51203263e+00
7.26666525e-02 -4.54135031e-01 1.07896316e+00 1.12970400e+00
2.95122832e-01 -1.80009499e-01 -6.65573895e-01 -1.13798165e+00
-3.66208673e-01 -1.44023612e-01 4.98343945e-01 7.91038692e-01
-4.11799699e-01 2.31990710e-01 -8.53970885e-01 4.62204635e-01
1.16711795e+00 -4.74715941e-02 1.19172668e+00 -1.37459362e+00
-3.72207105e-01 -3.68992001e-01 -4.59632188e-01 -1.16838932e+00
3.65233243e-01 -8.20718944e-01 -1.49660885e-01 -1.15080440e+00
7.80053437e-01 -5.13702393e-01 -3.26027662e-01 5.79940796e-01
-5.51022649e-01 6.46705091e-01 4.63017970e-01 5.48606396e-01
-8.19770336e-01 5.33537865e-01 6.45741403e-01 -4.05774385e-01
-1.53365463e-01 2.62031138e-01 -6.33317530e-01 6.16762042e-01
3.38004768e-01 -2.68204451e-01 -2.21009150e-01 -4.14074063e-01
-2.99593538e-01 -5.71999371e-01 9.84880328e-01 -1.20942402e+00
4.69086409e-01 2.93496162e-01 6.85509622e-01 -4.27789062e-01
6.07789874e-01 -7.99965084e-01 1.10359311e-01 4.26564604e-01
-2.91948527e-01 -1.64985433e-01 2.54053742e-01 7.73831427e-01
4.77743987e-03 -1.14371084e-01 8.79795313e-01 -6.40761778e-02
-7.66673148e-01 8.26292932e-01 3.94231111e-01 -1.09194674e-01
9.42600131e-01 -3.16128284e-01 -2.14655414e-01 -2.79438317e-01
-6.54950082e-01 4.33048248e-01 8.81795943e-01 5.36373854e-01
1.81661114e-01 -1.30858052e+00 -6.68160021e-01 9.87503752e-02
4.21959043e-01 -2.87522972e-01 2.78761417e-01 6.86994076e-01
6.03955686e-02 4.31331843e-01 -2.36115649e-01 -6.19954288e-01
-1.64621925e+00 8.06333542e-01 4.80998129e-01 -3.76207322e-01
-7.32375741e-01 7.55282462e-01 5.92081487e-01 -3.09765220e-01
5.56878805e-01 2.50185370e-01 -1.04487464e-01 -1.70823157e-01
8.14530492e-01 2.33906925e-01 -1.78068057e-01 -8.55970740e-01
-4.72532779e-01 6.13956630e-01 -3.49179238e-01 -9.26727653e-02
8.60031724e-01 -2.65176296e-01 2.43727982e-01 -5.17199673e-02
9.43678617e-01 1.76041834e-02 -1.64554191e+00 -5.45791149e-01
8.95332024e-02 -6.50041699e-01 -2.55205184e-01 -7.93358862e-01
-9.76818383e-01 5.71626067e-01 1.04734588e+00 -4.39069360e-01
7.58648276e-01 3.89193073e-02 1.03103316e+00 3.65195930e-01
4.58570689e-01 -1.09886682e+00 8.87619853e-02 1.67180166e-01
5.12274802e-01 -1.76662409e+00 6.35794699e-02 -4.39964980e-01
-2.61335552e-01 5.72570920e-01 7.86585927e-01 1.13591455e-01
3.10832143e-01 -9.84077752e-02 -1.61567777e-01 8.90358090e-02
-1.35763615e-01 -7.13231027e-01 7.39649832e-01 8.59658659e-01
3.27042103e-01 -6.33392930e-02 6.26002550e-02 6.01605535e-01
-8.43025446e-02 -5.40946014e-02 -1.11944318e-01 7.39801228e-01
-1.27689898e-01 -1.28905988e+00 -4.52471644e-01 2.29672641e-01
-4.26332891e-01 -1.23143662e-02 -3.78349781e-01 5.45529127e-01
2.42382079e-01 7.99882114e-01 1.31049708e-01 -1.65709123e-01
3.60723019e-01 1.07884429e-01 7.81429529e-01 -6.00324392e-01
-5.39265633e-01 -2.49705300e-01 2.95650382e-02 -3.26592386e-01
-5.38445055e-01 -8.08958650e-01 -6.15615726e-01 -3.40223879e-01
-3.38128388e-01 -7.29741976e-02 3.99819493e-01 9.52581227e-01
2.56901234e-01 2.39904568e-01 1.38017148e-01 -1.08169806e+00
-5.06642163e-01 -8.69965434e-01 -3.01419675e-01 9.91864145e-01
4.09864515e-01 -9.02721882e-01 2.96874605e-02 4.88410205e-01] | [14.749338150024414, 0.9558098912239075] |
9446d35b-5e84-41cc-94ce-7b503f0fb9a8 | meta-distribution-alignment-for-generalizable | null | null | http://openaccess.thecvf.com//content/CVPR2022/html/Ni_Meta_Distribution_Alignment_for_Generalizable_Person_Re-Identification_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Ni_Meta_Distribution_Alignment_for_Generalizable_Person_Re-Identification_CVPR_2022_paper.pdf | Meta Distribution Alignment for Generalizable Person Re-Identification | Domain Generalizable (DG) person ReID is a challenging task which trains a model on source domains yet generalizes well on target domains. Existing methods use source domains to learn domain-invariant features, and assume those features are also irrelevant with target domains. However, they do not consider the target domain information which is unavailable in the training phrase of DG. To address this issue, we propose a novel Meta Distribution Alignment (MDA) method to enable them to share similar distribution in a test-time-training fashion. Specifically, since high-dimensional features are difficult to constrain with a known simple distribution, we first introduce an intermediate latent space constrained to a known prior distribution. The source domain data is mapped to this latent space and then reconstructed back. A meta-learning strategy is introduced to facilitate generalization and support fast adaption. To reduce their discrepancy, we further propose a test-time adaptive updating strategy based on the latent space which efficiently adapts model to unseen domains with a few samples. Extensive experimental results show that our model outperforms the state-of-the-art methods by up to 5.1% R-1 on average on the large-scale and 4.7% R-1 on the single-source domain generalization ReID benchmark. | ['Heng Tao Shen', 'Wen Li', 'Feng Zheng', 'Xiaopeng Luo', 'Jingkuan Song', 'Hao Ni'] | 2022-01-01 | null | null | null | cvpr-2022-1 | ['generalizable-person-re-identification'] | ['computer-vision'] | [-5.33420704e-02 -3.71086448e-01 -4.14987803e-01 -8.35820436e-01
-7.27517128e-01 -6.96445882e-01 5.10590672e-01 -2.96027303e-01
-4.32790011e-01 1.16192222e+00 1.58131912e-01 2.43057951e-01
-2.60997236e-01 -7.59930670e-01 -7.67496347e-01 -6.44883811e-01
2.16080606e-01 8.56658936e-01 3.69946927e-01 -2.59229958e-01
-2.32845861e-02 2.71402627e-01 -1.40952575e+00 7.29487315e-02
1.46736896e+00 5.72669387e-01 2.43406028e-01 -5.14074452e-02
-1.25826538e-01 8.17900449e-02 -8.97609234e-01 -4.01240140e-01
4.41059023e-01 -2.63207644e-01 -4.73211646e-01 3.99328098e-02
3.69409472e-01 -4.84783322e-01 -4.95728016e-01 1.08782220e+00
6.70406342e-01 4.33807433e-01 9.25644457e-01 -1.29758155e+00
-1.09708166e+00 9.91485566e-02 -4.82969850e-01 2.21851811e-01
3.84531558e-01 6.20988756e-02 2.87601531e-01 -8.76678348e-01
6.50827467e-01 1.44279814e+00 4.07163948e-01 9.28598583e-01
-1.15889466e+00 -1.12392235e+00 6.63570762e-01 2.43263885e-01
-1.51689494e+00 -3.62447470e-01 9.22663271e-01 -2.29148373e-01
5.09980977e-01 -1.45742089e-01 1.57796130e-01 1.61947823e+00
3.39011103e-02 7.33126879e-01 1.13061917e+00 -1.91246737e-02
4.51962560e-01 4.29347128e-01 1.04372919e-01 1.80011883e-01
4.63386476e-01 6.53829575e-02 -6.03822351e-01 -3.33661407e-01
7.10160732e-01 1.38038024e-01 -1.36616364e-01 -6.10048950e-01
-1.08380353e+00 7.21827090e-01 3.91484797e-01 -3.87926474e-02
-1.89604700e-01 -5.35349250e-01 3.80488306e-01 5.59210658e-01
7.10144937e-01 2.04991605e-02 -8.10218632e-01 8.47478062e-02
-5.67139745e-01 4.64279294e-01 6.77188933e-01 1.45042658e+00
9.16460156e-01 -1.74702808e-01 -3.80080752e-02 1.17854214e+00
1.81717172e-01 7.33252883e-01 7.79932082e-01 -5.62485695e-01
8.16768765e-01 6.84288919e-01 2.88039863e-01 -6.49900377e-01
-1.71842337e-01 -6.28196478e-01 -9.87279832e-01 -9.79653299e-02
4.20108825e-01 -1.38682634e-01 -1.07136047e+00 2.07490087e+00
6.61854744e-01 4.22011971e-01 2.47143328e-01 7.50380397e-01
4.91795927e-01 5.54758668e-01 -5.98719204e-03 -1.58870146e-01
9.78881180e-01 -7.63691247e-01 -4.31455523e-01 -4.75335270e-01
4.37447697e-01 -4.06185299e-01 1.13398862e+00 2.73603648e-01
-6.84213758e-01 -8.65488470e-01 -1.17154193e+00 9.66373906e-02
-4.64856476e-01 1.76407844e-02 2.91749656e-01 7.39713967e-01
-7.47539878e-01 4.56563264e-01 -6.47134900e-01 -4.19404566e-01
5.31408787e-01 5.07577121e-01 -6.29595697e-01 -5.81612945e-01
-1.27477002e+00 7.17297792e-01 5.32228112e-01 -3.43735337e-01
-8.69942427e-01 -8.56899679e-01 -8.70197833e-01 -2.73236424e-01
2.25861892e-01 -8.33440125e-01 9.61828351e-01 -7.82569647e-01
-1.61056328e+00 7.06196368e-01 -5.25569618e-01 -1.50244489e-01
8.05570722e-01 -3.17438334e-01 -8.94409776e-01 -1.73089191e-01
5.43897748e-01 4.60427850e-01 9.15838420e-01 -1.13627064e+00
-7.09651947e-01 -4.71091807e-01 -1.37192413e-01 3.10478091e-01
-6.50446355e-01 -2.67769396e-01 -6.12708569e-01 -7.90361702e-01
1.40988514e-01 -9.23803866e-01 2.57406849e-02 -1.00409061e-01
-2.88232695e-02 -4.49329078e-01 6.27062976e-01 -5.73635101e-01
8.75382185e-01 -2.19661450e+00 1.78862408e-01 2.84904420e-01
-8.55061784e-03 1.75757214e-01 -3.89551401e-01 6.90859556e-02
1.85180847e-02 -1.76942855e-01 -2.85180658e-01 -3.74191791e-01
9.86858979e-02 3.01664352e-01 -4.54710454e-01 6.24564648e-01
-1.94038171e-02 5.17036557e-01 -9.48820949e-01 -3.53458524e-01
-6.74823895e-02 3.36515248e-01 -5.94234288e-01 3.38034123e-01
-5.58277033e-02 7.30966330e-01 -7.21863747e-01 6.50674522e-01
1.27693927e+00 -1.39994085e-01 1.53062999e-01 1.29056290e-01
1.06795251e-01 2.48536840e-01 -1.20628750e+00 2.00174785e+00
-2.27827281e-01 2.28691511e-02 -3.37857872e-01 -1.05487013e+00
1.34007335e+00 -5.87132014e-02 2.80991077e-01 -8.14661562e-01
-2.39419147e-01 3.59166354e-01 -1.59491017e-01 -8.71229246e-02
2.82527268e-01 1.75293849e-03 -3.00320178e-01 1.98293656e-01
1.06675982e-01 4.18575943e-01 -4.80781868e-02 -7.38690272e-02
8.70446682e-01 1.95983037e-01 2.02722609e-01 -3.58848423e-01
7.03781247e-01 -1.16909519e-01 1.18323374e+00 7.40245879e-01
-2.53605634e-01 5.66776276e-01 2.44518921e-01 -4.16253239e-01
-1.03694975e+00 -1.40055549e+00 -2.58982688e-01 1.20490122e+00
6.07580423e-01 -4.66608629e-02 -5.81994891e-01 -1.11454189e+00
4.21412587e-01 7.37210274e-01 -6.67166591e-01 -4.83820409e-01
-6.88896179e-01 -7.68546402e-01 1.94131926e-01 6.84699059e-01
8.01277518e-01 -6.01160645e-01 2.36258119e-01 2.75462717e-01
-1.28303259e-03 -8.47854376e-01 -6.26000881e-01 -2.31869355e-01
-8.26608717e-01 -8.12086463e-01 -1.21139407e+00 -9.57644403e-01
1.00889277e+00 3.89066547e-01 7.97699571e-01 -4.31645960e-01
2.12109029e-01 2.32517764e-01 -3.61661524e-01 -2.37042174e-01
-1.91334829e-01 3.63256186e-01 6.77511930e-01 3.24251540e-02
7.69015014e-01 -8.36589634e-01 -6.27760291e-01 8.79153728e-01
-6.81554615e-01 -2.81729341e-01 7.97745109e-01 1.11103725e+00
6.40969574e-01 1.00393474e-01 1.12080836e+00 -8.67874682e-01
5.52645266e-01 -9.22360837e-01 -6.35151863e-01 3.87784690e-01
-5.33188403e-01 1.68546528e-01 8.01545262e-01 -1.06954646e+00
-1.37996888e+00 -1.04874223e-01 1.98455557e-01 -4.55644459e-01
-3.24030399e-01 2.24878281e-01 -8.32756102e-01 5.43031059e-02
7.83630073e-01 5.26825488e-01 -2.05260381e-01 -6.47984207e-01
1.82090610e-01 7.12771773e-01 7.19067872e-01 -1.11401844e+00
1.27100408e+00 3.95337343e-01 -3.62155676e-01 -3.26853544e-01
-8.12181234e-01 -5.00637352e-01 -8.41167390e-01 3.18377733e-01
4.67931539e-01 -1.27809536e+00 -1.94780141e-01 5.57738423e-01
-9.20478463e-01 -3.51228923e-01 -1.04490593e-01 5.14806867e-01
-3.91040742e-01 4.81654793e-01 -9.94038731e-02 -3.46109003e-01
-4.20429595e-02 -8.14623892e-01 8.39843035e-01 4.13945347e-01
1.89444069e-02 -8.28316689e-01 1.39139995e-01 8.43565241e-02
4.34445888e-01 1.30979180e-01 7.27292895e-01 -9.64419901e-01
-4.24297273e-01 -1.77106708e-01 -2.80927449e-01 3.62318248e-01
4.75902647e-01 -8.01694512e-01 -8.68292570e-01 -6.96758926e-01
-2.08898168e-02 -2.88551033e-01 7.56170630e-01 1.30974010e-01
1.24618459e+00 -1.49277747e-01 -6.89035594e-01 7.70205915e-01
1.00877607e+00 8.61587897e-02 3.36776614e-01 4.97996747e-01
4.15633678e-01 4.95759785e-01 9.72164333e-01 4.43519950e-01
6.72924817e-01 7.90451765e-01 -1.36451721e-01 2.22964466e-01
-1.70782711e-02 -5.33399165e-01 5.21797299e-01 5.45071959e-01
2.01762483e-01 -1.08909629e-01 -7.28634596e-01 7.25296736e-01
-1.97349751e+00 -7.64891267e-01 3.98069769e-01 2.51543999e+00
9.90823090e-01 3.50713044e-01 3.04351538e-01 -4.07281667e-01
9.22986805e-01 -1.94618672e-01 -1.21401620e+00 3.27089310e-01
-1.88859910e-01 -2.47280835e-03 4.56555009e-01 2.40172699e-01
-1.09470832e+00 8.95828903e-01 5.62047482e+00 8.65498006e-01
-8.55567455e-01 2.84217179e-01 3.02317917e-01 1.40783901e-03
-2.84480363e-01 -1.92672849e-01 -1.19857359e+00 9.55462337e-01
6.72408104e-01 -6.54520154e-01 3.54986668e-01 1.31820977e+00
-3.18267912e-01 3.36874813e-01 -1.26048303e+00 1.02492189e+00
5.14294393e-02 -7.48374820e-01 1.40161350e-01 1.53636366e-01
9.87096250e-01 -1.32589728e-01 2.94962674e-01 8.42834353e-01
2.74632007e-01 -5.22972107e-01 4.88109440e-01 4.96428937e-01
9.89618719e-01 -9.31896567e-01 6.49962187e-01 5.66703916e-01
-1.17820740e+00 -1.46236777e-01 -1.12272727e+00 2.44227350e-01
-2.57075392e-02 3.79280984e-01 -7.46604562e-01 7.60644436e-01
6.95020854e-01 8.43136251e-01 -5.86548626e-01 1.13208854e+00
-1.81091949e-01 3.19830567e-01 -3.36933404e-01 1.98793709e-01
-3.61202270e-01 4.37279195e-02 5.57336867e-01 8.53271604e-01
5.98036647e-01 1.00135349e-01 5.44372499e-01 7.69527435e-01
-1.27394721e-01 -1.48312211e-01 -5.63356161e-01 4.28841233e-01
9.48485255e-01 7.03801692e-01 -1.87686339e-01 -3.95478219e-01
-4.46867764e-01 1.47358775e+00 4.03236866e-01 6.10749066e-01
-8.31994295e-01 -3.94027352e-01 9.70362842e-01 1.22455910e-01
2.23209590e-01 -2.22721368e-01 1.27706686e-02 -1.57768083e+00
3.85162413e-01 -7.60420084e-01 5.83112001e-01 -2.15741992e-01
-2.16277146e+00 5.66748559e-01 4.22458678e-01 -1.67239451e+00
-4.13193822e-01 -4.60041702e-01 -4.40574199e-01 1.12059188e+00
-1.88683748e+00 -1.20877707e+00 -3.97250891e-01 1.06615198e+00
5.20615578e-01 -7.31962681e-01 8.06525171e-01 3.73625875e-01
-5.19451499e-01 1.27028883e+00 5.43564498e-01 1.32590439e-03
1.35840952e+00 -1.12596667e+00 6.64494693e-01 6.87278628e-01
-6.08864009e-01 1.01665103e+00 4.29543674e-01 -9.86805499e-01
-1.24937797e+00 -1.36750138e+00 6.43953919e-01 -5.50771534e-01
2.57316381e-01 -5.51868141e-01 -1.22897792e+00 8.35337579e-01
-2.25848332e-01 2.16090977e-01 7.51127601e-01 2.46600032e-01
-6.20337784e-01 -4.63962138e-01 -1.49044895e+00 3.52422386e-01
1.50921655e+00 -4.15702939e-01 -9.96244311e-01 3.56897920e-01
6.74539685e-01 -5.92766285e-01 -7.17287600e-01 3.95420998e-01
3.32740724e-01 -5.53179264e-01 1.10946369e+00 -6.32967055e-01
-1.90688536e-01 -5.44428408e-01 -9.78506356e-02 -1.52840948e+00
-5.09180844e-01 -5.29238582e-01 -6.56729266e-02 1.51918757e+00
9.25180167e-02 -1.15439701e+00 9.01381135e-01 7.44949877e-01
-7.80857801e-02 -1.31698430e-01 -1.06112218e+00 -1.39150250e+00
2.81565219e-01 3.39903915e-03 1.08733356e+00 1.08936036e+00
-2.86278933e-01 9.50962380e-02 -3.78558218e-01 5.39782643e-01
9.02493536e-01 1.47418156e-01 1.17458427e+00 -1.59583485e+00
-8.86501670e-02 1.73345068e-03 -4.05838549e-01 -1.51047182e+00
6.09863102e-01 -8.11476946e-01 -1.90800637e-01 -1.14640975e+00
2.34333709e-01 -7.29673624e-01 -6.98274076e-01 3.49129885e-01
-3.93056214e-01 -3.56879920e-01 -2.23580733e-01 3.58808249e-01
-6.92249119e-01 8.36667836e-01 1.04890454e+00 -2.33094990e-01
-4.29562420e-01 1.68264493e-01 -8.97992313e-01 4.25843090e-01
7.63953388e-01 -6.20559096e-01 -8.06957662e-01 -4.64826435e-01
-4.41709489e-01 -3.85528862e-01 1.88836068e-01 -1.15746653e+00
4.51017708e-01 -3.52085888e-01 9.57276344e-01 -5.56630850e-01
3.59845698e-01 -6.25118256e-01 -4.00810614e-02 6.92859516e-02
-9.46289301e-02 -1.21387318e-01 2.74359018e-01 8.71628046e-01
-4.36029211e-02 1.01345666e-01 7.77251482e-01 1.84558988e-01
-9.68834758e-01 6.00747824e-01 2.98221946e-01 2.00692825e-02
1.08069241e+00 -2.16243193e-01 -5.04366934e-01 -1.60729155e-01
-4.48337466e-01 5.42732000e-01 8.28652263e-01 7.19289303e-01
6.22560620e-01 -1.64364314e+00 -9.06526089e-01 5.68957984e-01
4.34281677e-01 2.23598853e-01 5.92880428e-01 1.40197024e-01
1.51660800e-01 1.58613563e-01 -5.04106343e-01 -6.58795238e-01
-7.94383585e-01 7.00702012e-01 1.53690860e-01 -5.88124655e-02
-5.59175611e-01 8.71143401e-01 6.51899159e-01 -7.94586718e-01
1.39739752e-01 3.82263027e-02 -5.64852506e-02 -1.84854731e-01
7.84555078e-01 2.88369805e-01 -2.38818541e-01 -4.86505121e-01
-5.95520496e-01 6.11854076e-01 -6.80252850e-01 9.62698087e-03
1.33527780e+00 -1.60638615e-01 2.09216669e-01 1.04600720e-01
1.16165042e+00 2.96379086e-02 -1.55076015e+00 -6.79446399e-01
-8.29073659e-04 -8.54870260e-01 -5.64113736e-01 -8.37963641e-01
-5.56384325e-01 4.82886791e-01 7.31411994e-01 -2.64860958e-01
1.20592463e+00 3.91905978e-02 7.54160464e-01 5.57448268e-01
5.72446406e-01 -1.21762860e+00 2.00444624e-01 3.94544929e-01
8.83696079e-01 -1.26119637e+00 1.96077079e-02 -3.50859314e-01
-6.53546274e-01 8.97959113e-01 1.10063970e+00 -1.58138394e-01
4.99112397e-01 -2.22551867e-01 -2.04137340e-01 4.21893954e-01
-4.67591673e-01 8.22548866e-02 3.57231915e-01 1.13274395e+00
-1.28634334e-01 5.31633869e-02 -5.47636002e-02 1.03470874e+00
-9.92180333e-02 8.73861909e-02 8.65023509e-02 7.89245009e-01
-3.21835339e-01 -1.61176491e+00 -3.08748066e-01 1.97916061e-01
7.42877275e-02 3.12088877e-01 -2.00383782e-01 8.38983178e-01
2.21478209e-01 7.38489926e-01 -1.53284565e-01 -3.75653833e-01
5.73760033e-01 5.01130335e-02 3.68660957e-01 -7.50321805e-01
1.25910640e-01 -1.95848614e-01 -3.00576866e-01 -3.78234029e-01
-1.56473741e-01 -8.85381043e-01 -9.75077152e-01 -3.57327670e-01
-1.26183229e-02 1.59599945e-01 2.81344980e-01 8.65773201e-01
6.24285877e-01 1.98470801e-01 7.91591525e-01 -6.41099334e-01
-9.55323100e-01 -9.34900820e-01 -6.75054729e-01 4.86668229e-01
3.68623078e-01 -1.03175187e+00 -2.16326118e-01 -8.42077136e-02] | [14.733835220336914, 1.1135196685791016] |
008f0bbc-8339-48bb-996c-a094315b5d10 | eco-driving-trajectory-planning-of-a | 2205.09618 | null | https://arxiv.org/abs/2205.09618v1 | https://arxiv.org/pdf/2205.09618v1.pdf | Eco-driving Trajectory Planning of a Heterogeneous Platoon in Urban Environments | Given the increasing popularity and demand for connected and autonomous vehicles (CAVs), Eco-driving and platooning in highways and urban areas to increase the efficiency of the traffic system is becoming a possibility. This paper presents Eco-driving trajectory planning for a platoon of heterogeneous electric vehicles (EVs) in urban environments. The proposed control strategy for the platoon considers energy consumption, mobility and passenger comfort, with which vehicles may pass signalized intersections with no stops. For a given urban route, first, the platoon's leader vehicle employs dynamic programming (DP) to plan a trajectory for the anticipated path with the aim of balancing energy consumption, mobility and passenger comfort. Then, every other following CAV in the platoon either follows its preceding vehicle using a PID-based cooperative adaptive cruise control or plans its own trajectory by checking whether it can pass the next intersection without stopping. Furthermore, a heavy-duty vehicle that cannot efficiently follow a light-weight vehicle would instead employ the DP-based trajectory planner. Simulation studies demonstrate the efficacy of the proposed control strategy with which the platoon's energy consumption is shown to reduce while the mobility is not compromised. | ['Javad Mohammadpour Velni', 'Jidong J. Yang', 'Sahand Mosharafian', 'Hao Zhen'] | 2022-05-19 | null | null | null | null | ['trajectory-planning'] | ['robots'] | [-4.05764282e-01 6.75392985e-01 -3.11022133e-01 -2.16179773e-01
-6.73636943e-02 -5.01988590e-01 3.15084994e-01 5.94753958e-02
-4.65975136e-01 1.04669166e+00 -6.98654830e-01 -7.66592920e-01
-5.09193122e-01 -1.19731307e+00 -3.78295839e-01 -1.16653240e+00
-2.17586473e-01 4.31269199e-01 5.43972969e-01 -1.96547791e-01
-4.02584262e-02 8.47116709e-01 -1.21774900e+00 -7.09156215e-01
1.17582679e+00 8.66898537e-01 4.87897068e-01 3.63554180e-01
3.34443241e-01 2.03302681e-01 2.82853171e-02 -9.34856907e-02
1.60868436e-01 -1.87094435e-02 -3.01238179e-01 -4.66762111e-02
-7.27445841e-01 -4.07146871e-01 -4.88254666e-01 6.46996379e-01
1.11604780e-01 4.80417430e-01 6.21231258e-01 -2.61192703e+00
4.93306965e-01 4.22192551e-02 -2.68373430e-01 -9.60511491e-02
-4.15355504e-01 4.21691120e-01 5.20495653e-01 -2.02539861e-01
5.39891362e-01 8.74529600e-01 5.40112890e-02 3.43628407e-01
-8.58690858e-01 -7.54270017e-01 2.42702171e-01 5.33768415e-01
-1.49522769e+00 -2.45714337e-01 6.63860619e-01 -8.97018518e-03
8.29143167e-01 3.54668647e-01 1.14947343e+00 1.25458064e-02
6.66095495e-01 6.61536515e-01 4.17091429e-01 -5.81407212e-02
4.71507281e-01 2.98529953e-01 -1.89485997e-01 2.22208336e-01
5.73559999e-01 1.01182140e-01 4.79385376e-01 1.18038304e-01
-5.48946083e-01 -2.10150927e-01 2.35304087e-01 -4.47155148e-01
-7.89107800e-01 6.72424436e-01 8.83940235e-02 1.69372365e-01
-9.75293517e-01 2.17875421e-01 1.43975392e-01 -2.49005836e-02
4.80528995e-02 -6.35937899e-02 -8.57494771e-02 -2.94022292e-01
-1.14133859e+00 4.92364824e-01 6.15650952e-01 1.51336515e+00
5.46672761e-01 2.40951255e-01 2.30342131e-02 2.12610915e-01
5.30715823e-01 9.85697985e-01 -5.37134886e-01 -1.24357748e+00
2.48590201e-01 6.27846360e-01 5.15328705e-01 -7.34698772e-01
-7.82530308e-01 -9.21665430e-02 -5.84180474e-01 5.46560168e-01
7.88520947e-02 -6.32513762e-01 -3.22330773e-01 1.41845262e+00
6.12311423e-01 -1.82194680e-01 3.62998873e-01 6.17643714e-01
-3.71825211e-02 1.14882922e+00 4.30418551e-01 -6.42939448e-01
1.28442252e+00 -6.35640085e-01 -7.80963480e-01 -2.23468505e-02
5.53642094e-01 -3.35122019e-01 -1.23947792e-01 9.63631868e-02
-1.41644859e+00 -7.79461563e-02 -1.24504840e+00 4.09802139e-01
-3.46824408e-01 1.12047475e-02 -3.91687900e-01 6.01229370e-01
-8.71614635e-01 2.05001295e-01 -8.70451272e-01 -3.37579668e-01
3.52423847e-01 5.24619460e-01 1.61815435e-01 -1.49780530e-02
-1.11914349e+00 9.99254823e-01 3.13078612e-01 -2.75449026e-02
-9.15737212e-01 -6.26846433e-01 -5.70446014e-01 6.89067468e-02
2.99038678e-01 -2.46892482e-01 8.72478366e-01 -3.89885575e-01
-1.19330347e+00 7.07998946e-02 -6.80923015e-02 -3.89086902e-01
7.32914805e-01 6.86327815e-01 -6.95957839e-01 1.60574257e-01
1.36005387e-01 7.31732607e-01 1.51281385e-02 -1.18896949e+00
-1.48921561e+00 4.13165949e-02 6.21072343e-03 3.61867398e-01
1.17928192e-01 -3.24863613e-01 -2.91279852e-01 4.15805280e-01
-4.09272343e-01 -1.74879575e+00 -4.39960629e-01 -1.57367527e-01
-7.23875046e-01 -8.74696076e-01 1.30185413e+00 -9.46945176e-02
1.23972654e+00 -2.19663239e+00 -3.27813834e-01 7.62181938e-01
-1.86622098e-01 1.62125006e-01 5.99119775e-02 7.38607049e-01
5.93261838e-01 -9.17410851e-02 2.76797414e-02 -1.76712289e-01
2.08578900e-01 4.74960744e-01 1.00842535e-01 5.49032569e-01
-5.06313220e-02 3.96497130e-01 -8.92695427e-01 -4.77635741e-01
4.10423309e-01 3.79041880e-01 -4.23071176e-01 2.03866839e-01
-8.13642591e-02 2.04733208e-01 -7.90167689e-01 2.31650442e-01
1.12548184e+00 5.80435157e-01 3.15912724e-01 -5.96343307e-03
-1.16579437e+00 -1.57668784e-01 -1.16198874e+00 4.93243396e-01
-5.75545669e-01 1.03439629e+00 5.35279989e-01 -7.89608836e-01
6.41687989e-01 2.30598420e-01 9.71438944e-01 -7.56270409e-01
2.60655105e-01 4.39294100e-01 9.60455742e-03 -4.23006713e-01
8.59451234e-01 2.05670252e-01 -2.45010890e-02 2.61955678e-01
-6.30328774e-01 1.25414163e-01 3.96431744e-01 3.07843238e-01
9.54553545e-01 -2.07922086e-01 -1.37265056e-01 -6.09527707e-01
7.92999208e-01 4.13151294e-01 7.27091193e-01 6.01152480e-02
-4.89935905e-01 -6.95390046e-01 4.36754882e-01 1.93143502e-01
-1.11551607e+00 -8.82625461e-01 -1.15161434e-01 5.48093796e-01
9.83257711e-01 2.55622923e-01 -6.81623995e-01 -9.03608277e-02
6.99706227e-02 1.61483908e+00 -2.43034810e-02 -1.62769884e-01
-5.99779189e-01 -2.26025611e-01 2.37140302e-02 2.56623793e-03
5.80618680e-01 -5.06606877e-01 -1.22247493e+00 5.43760598e-01
-6.61421269e-02 -1.04818892e+00 -5.10693967e-01 -6.83586597e-02
-1.05618231e-01 -1.16591179e+00 -2.24378213e-01 -6.72292292e-01
9.49122667e-01 6.71360672e-01 -4.20479551e-02 2.69283563e-01
5.34260631e-01 3.70851338e-01 9.41819251e-02 -6.90446019e-01
-5.78760803e-01 1.96459800e-01 3.18430248e-03 1.65127680e-01
3.87302846e-01 -2.44522154e-01 -9.69883144e-01 6.92930043e-01
-4.06638622e-01 2.98183531e-01 -1.91131178e-02 7.60494024e-02
6.94736719e-01 8.06815982e-01 9.88708496e-01 -7.65828341e-02
2.82050759e-01 -1.04560184e+00 -1.04534614e+00 -1.41727226e-02
-8.64757657e-01 -3.63179386e-01 6.65349782e-01 1.10005423e-01
-9.97202277e-01 1.18577272e-01 -8.40495974e-02 8.27340260e-02
1.27314718e-03 -1.74679369e-01 -6.73811495e-01 -1.51852369e-01
-4.15098578e-01 2.93603063e-01 1.84290379e-01 2.55412042e-01
1.64543018e-01 6.32665515e-01 2.80172408e-01 -5.07669011e-03
1.11545408e+00 4.11892444e-01 7.28823006e-01 -8.31581175e-01
5.47988355e-01 -3.01509976e-01 -2.87492275e-01 -1.04917657e+00
9.01750803e-01 -8.65948856e-01 -1.64988911e+00 2.46805653e-01
-1.09327388e+00 -1.55960798e-01 1.19854860e-01 5.82582951e-01
-5.44995546e-01 1.59661938e-02 2.02993616e-01 -1.40553272e+00
-5.54408245e-02 -1.26774013e+00 3.18077981e-01 6.56239748e-01
1.32098913e-01 -6.60557628e-01 -2.20770180e-01 -1.13585360e-01
4.77837712e-01 5.85008264e-01 8.13354731e-01 -3.05896312e-01
-1.05118966e+00 -1.41803354e-01 -4.94487621e-02 -1.10225581e-01
-2.02727541e-01 2.11156428e-01 -2.55156428e-01 -5.64209580e-01
-6.36195064e-01 5.12628615e-01 -2.32833877e-01 3.45106989e-01
5.06041586e-01 -5.56622088e-01 -1.12838531e+00 2.26275280e-01
1.64531207e+00 1.31260014e+00 6.62422061e-01 5.64921081e-01
1.54234037e-01 1.06009579e+00 1.15300333e+00 3.69910628e-01
1.09633768e+00 7.93986797e-01 1.06078029e+00 5.61853014e-02
4.59241211e-01 -1.28579527e-01 2.87294090e-01 2.35158384e-01
-3.98910046e-02 -7.04421699e-01 -6.20853961e-01 8.25673699e-01
-1.69004858e+00 -8.66082549e-01 -5.78400671e-01 2.09776783e+00
1.16657555e-01 2.67025549e-02 2.04620525e-01 3.40367377e-01
7.63180971e-01 -3.23909402e-01 -6.95374548e-01 -9.45643544e-01
1.94820121e-01 -7.26923287e-01 1.30521333e+00 4.50187832e-01
-2.32686818e-01 3.51599991e-01 3.98823500e+00 1.02149630e+00
-6.91540301e-01 -3.48578952e-02 4.97777700e-01 -1.40670180e-01
-6.14250243e-01 2.12151468e-01 -8.73987257e-01 9.27369237e-01
1.42382181e+00 -1.07940030e+00 4.57812607e-01 8.22910488e-01
1.29112041e+00 -6.52380168e-01 -6.52071536e-01 2.11572677e-01
-6.85657680e-01 -1.19823086e+00 -6.54935300e-01 2.97425896e-01
7.21013963e-01 -1.33850038e-01 -5.26736140e-01 1.82801589e-01
1.59454182e-01 -3.16971630e-01 9.82240379e-01 5.63766003e-01
3.15530002e-01 -1.66194546e+00 3.83543372e-01 6.82578683e-01
-1.54511786e+00 -3.06733280e-01 2.13030428e-01 5.13398826e-01
1.16133547e+00 1.69625610e-01 -7.19412088e-01 6.52082086e-01
2.25051135e-01 5.38057685e-02 1.72497079e-01 1.20528722e+00
5.11987582e-02 3.23733121e-01 -5.43740273e-01 -6.07652724e-01
7.62392640e-01 -6.21711969e-01 1.05901468e+00 8.32898021e-01
6.18002057e-01 3.74028593e-01 2.87445843e-01 5.04038513e-01
2.74972141e-01 -3.07310969e-02 -5.77003419e-01 4.14105803e-01
9.42947209e-01 1.29801118e+00 -7.07198322e-01 -3.12790036e-01
-2.10158691e-01 2.25994170e-01 -6.07257307e-01 4.74018782e-01
-1.29139578e+00 -9.70086336e-01 9.37880218e-01 5.18882513e-01
-9.99994110e-03 -2.81830877e-01 -1.68652818e-01 2.00461388e-01
-1.91820666e-01 2.82898396e-01 -1.37185708e-01 -8.60696554e-01
-2.56589949e-01 5.42779863e-01 3.56532335e-01 -1.41779244e+00
2.23521627e-02 1.17614202e-03 -1.10321426e+00 5.64894378e-01
-1.97432017e+00 -7.86092699e-01 -1.72805637e-01 4.03565437e-01
4.33758110e-01 -6.17375523e-02 -1.48798600e-01 6.32418275e-01
-6.22913420e-01 1.74439624e-01 4.91126657e-01 -6.08487427e-01
-7.35639259e-02 -6.16900504e-01 -2.26710662e-01 8.41380119e-01
-1.31477225e+00 -9.32542458e-02 1.04249072e+00 -6.32694244e-01
-1.87698972e+00 -1.33172190e+00 1.10422254e+00 5.43948531e-01
4.48115081e-01 1.00946434e-01 -4.69995677e-01 1.39896527e-01
7.86291063e-01 -5.06829023e-01 4.38384060e-03 -1.10660326e+00
1.01261413e+00 -4.46352184e-01 -1.49617851e+00 9.50558543e-01
7.07839608e-01 2.76966542e-01 3.04750293e-01 1.91175699e-01
4.25044358e-01 8.78623500e-03 -4.39153016e-01 3.32322389e-01
4.04878557e-01 -3.21933657e-01 2.61618733e-01 -1.23092696e-01
-5.15884340e-01 -6.25428617e-01 4.77468148e-02 -1.49192357e+00
-2.82874376e-01 -8.52083027e-01 3.78633261e-01 1.26018178e+00
3.74802172e-01 -6.43522561e-01 5.78157604e-01 1.10570788e+00
-4.58865285e-01 -7.41833508e-01 -1.72470951e+00 -9.70883787e-01
5.82401939e-02 -5.76299846e-01 7.60377467e-01 1.33326933e-01
3.08397412e-01 -2.32871160e-01 -1.89244658e-01 6.45655870e-01
8.96776021e-01 -2.89470226e-01 7.08593428e-01 -7.77743578e-01
8.22294474e-01 -3.70450199e-01 1.29185589e-02 -7.15754271e-01
3.35992277e-01 -1.03334832e+00 4.67497826e-01 -1.90162110e+00
-3.41734350e-01 -5.69489181e-01 1.76739141e-01 1.33095682e-01
5.63001931e-01 -3.26259762e-01 3.74196768e-01 -1.34065866e-01
-5.33098400e-01 7.17557430e-01 1.09503376e+00 -2.64625371e-01
-4.01991367e-01 5.12040317e-01 3.16260606e-02 3.44395906e-01
1.01869226e+00 -5.98580718e-01 -8.39131415e-01 4.61412460e-01
-1.38412025e-02 6.44152045e-01 1.60937697e-01 -8.48886430e-01
7.77687669e-01 -6.96778655e-01 -6.75346255e-01 -1.22190917e+00
1.70776978e-01 -1.72428024e+00 9.73909974e-01 1.07150185e+00
9.53575820e-02 2.18959168e-01 2.78844774e-01 8.27233911e-01
2.65578628e-01 5.07343337e-02 1.12711024e+00 8.08867335e-01
-6.32871866e-01 4.51936275e-01 -1.56752694e+00 -4.76936013e-01
2.24964428e+00 -4.91888821e-01 -1.67808473e-01 -1.68960333e-01
-2.87066311e-01 1.46653605e+00 1.95562467e-01 4.68052447e-01
2.62125880e-01 -1.30696380e+00 -3.78114760e-01 -1.40419453e-01
-1.72021046e-01 -5.15686333e-01 6.38099611e-01 1.12473762e+00
-3.45936894e-01 7.96183586e-01 -3.46447647e-01 -3.29509497e-01
-1.21038353e+00 5.16647637e-01 4.46225256e-01 1.35436073e-01
-3.99509758e-01 -2.60638088e-01 -2.31617764e-01 2.75696903e-01
-1.55260786e-01 4.23425213e-02 -2.78416812e-01 1.62431091e-01
1.42100140e-01 1.35984468e+00 -3.37187856e-01 -9.41018403e-01
-7.06526160e-01 2.62260497e-01 6.77314699e-01 -3.54274474e-02
9.58630025e-01 -8.62021089e-01 1.37167856e-01 -1.31870270e-01
9.86131608e-01 -1.70794368e-01 -1.41893601e+00 4.52992618e-01
-1.24350421e-01 -1.55694813e-01 3.29329729e-01 -5.67025065e-01
-1.35406673e+00 2.14967102e-01 4.18408096e-01 4.19996828e-01
1.03795767e+00 -3.94296020e-01 1.06400394e+00 4.06675860e-02
7.72218108e-01 -1.74913442e+00 -7.57615387e-01 2.19134241e-01
3.80740225e-01 -6.06030285e-01 -3.22188854e-01 -3.72965634e-01
-7.67597497e-01 1.04070771e+00 6.96587324e-01 8.86538345e-03
9.10310507e-01 3.07234228e-01 -7.13598430e-01 7.95341358e-02
-8.84081841e-01 -1.80018812e-01 -3.86449844e-01 4.39407021e-01
-6.22682452e-01 3.52296919e-01 -9.00230169e-01 3.53977799e-01
1.76144212e-01 4.75362949e-02 9.72485125e-01 7.44853497e-01
-8.92716169e-01 -8.13011825e-01 -9.02414843e-02 2.01677859e-01
1.00502275e-01 6.70304954e-01 7.40770698e-01 1.06401920e+00
5.02197921e-01 1.46964288e+00 2.60271102e-01 -1.49370670e-01
6.74828589e-01 -2.41654783e-01 -3.65119189e-01 2.79034525e-01
-6.75258934e-02 -2.40695298e-01 5.33435166e-01 -3.53313327e-01
-2.85988510e-01 -8.45183849e-01 -1.88604844e+00 -7.54455745e-01
-6.16172962e-02 8.45142782e-01 1.19806421e+00 9.98888016e-01
4.14697170e-01 3.91987324e-01 1.41630220e+00 -4.78603810e-01
2.04729661e-01 -2.27646351e-01 -7.18954086e-01 -4.81783301e-01
2.06332192e-01 -7.18561888e-01 -3.73182267e-01 -5.74933648e-01] | [5.542645454406738, 1.6505099534988403] |
77ee6d07-c217-4e5e-9d83-91f852f3a28c | counterfactual-debiasing-inference-for | null | null | https://dl.acm.org/doi/abs/10.1145/3474085.3475472 | https://dl.acm.org/doi/pdf/10.1145/3474085.3475472?casa_token=vpmtrdT6DSMAAAAA:E97KG5JVQqGGGmptKQpIIxOrOpAJD6wkStOHKsmh4sDJ6qVB7DVxxkOKXrG-WgCb3CtmEz_nl9dXlg | Counterfactual Debiasing Inference for Compositional Action Recognition | Compositional action recognition is a novel challenge in the computer vision community and focuses on revealing the different combinations of verbs and nouns instead of treating subject-object interactions in videos as individual instances only. Existing methods tackle this challenging task by simply ignoring appearance information or fusing object appearances with dynamic instance tracklets. However, those strategies usually do not perform well for unseen action instances. For that, in this work we propose a novel learning framework called Counterfactual Debiasing Network (CDN) to improve the model generalization ability by removing the interference introduced by visual appearances of objects/subjects. It explicitly learns the appearance information in action representations and later removes the effect of such information in a causal inference manner. Specifically, we use tracklets and video content to model the factual inference by considering both appearance information and structure information. In contrast, only video content with appearance information is leveraged in the counterfactual inference. With the two inferences, we conduct a causal graph which captures and removes the bias introduced by the appearance information by subtracting the result of the counterfactual inference from that of the factual inference. By doing that, our proposed CDN method can better recognize unseen action instances by debiasing the effect of appearances. Extensive experiments on the Something-Else dataset clearly show the effectiveness of our proposed CDN over existing state-of-the-art methods. | ['Chuang Gan', 'Lixin Duan', 'Wen Li', 'Xunsong Li', 'Bo Wu', 'Pengzhan Sun'] | 2021-10-17 | null | null | null | acm-international-conference-on-multimedia-2 | ['counterfactual-inference'] | ['miscellaneous'] | [ 5.34464717e-01 1.17398426e-01 -3.56810331e-01 -1.79887116e-01
-8.56299102e-02 -4.67597425e-01 9.38796759e-01 -4.75776881e-01
-1.65202603e-01 7.51194656e-01 5.81709921e-01 1.38770686e-02
-7.02018440e-02 -6.31856799e-01 -1.18705881e+00 -9.20975387e-01
1.91415370e-01 -3.53243016e-02 2.48046309e-01 2.36321837e-01
1.29748315e-01 9.87893268e-02 -1.71677923e+00 6.47987068e-01
5.41442513e-01 8.85850787e-01 8.70516151e-02 1.35007337e-01
-1.14027718e-02 1.25631595e+00 -5.44293165e-01 -5.77838898e-01
4.94555056e-01 -5.22625625e-01 -4.04713809e-01 4.21651959e-01
9.96567309e-01 -8.16277623e-01 -7.69150555e-01 1.24517095e+00
2.30450362e-01 2.79962748e-01 4.44623023e-01 -1.60404944e+00
-9.44091558e-01 6.89478874e-01 -9.94669139e-01 3.53581458e-01
3.74804437e-01 5.71408510e-01 9.32729900e-01 -5.65032899e-01
7.16978014e-01 1.70360565e+00 1.63659230e-01 6.64010942e-01
-1.27671623e+00 -1.02106214e+00 9.32185054e-01 6.54519916e-01
-9.49529350e-01 -4.93380725e-01 1.26099575e+00 -5.78808606e-01
4.41982120e-01 2.51849025e-01 7.85524309e-01 1.71870232e+00
7.56870508e-02 1.32271516e+00 1.06959951e+00 -7.17583895e-02
2.07872614e-01 -1.59820840e-01 -7.98694417e-02 5.76093733e-01
3.80265355e-01 2.72278637e-01 -5.25252819e-01 5.38995676e-03
6.95603132e-01 5.43202102e-01 -5.48144877e-01 -5.14844298e-01
-1.42083001e+00 4.73350734e-01 3.99632156e-01 8.59434009e-02
-6.43276572e-01 3.04980189e-01 3.23357999e-01 1.36545435e-01
4.64544773e-01 5.68011254e-02 -4.52476740e-01 2.01181307e-01
-6.66849077e-01 3.44357014e-01 3.15659523e-01 8.49089503e-01
4.29289281e-01 1.28301129e-01 -6.41242445e-01 5.82200110e-01
1.16439372e-01 5.11567891e-01 4.97002512e-01 -9.43438292e-01
4.86539453e-01 7.52169967e-01 1.53969660e-01 -1.25798833e+00
-6.67726696e-02 -3.17162216e-01 -7.72894859e-01 2.53009349e-01
5.52471042e-01 -1.34591028e-01 -9.32065666e-01 2.07045555e+00
6.19036138e-01 8.41310859e-01 -5.98379560e-02 1.05148041e+00
8.36911261e-01 5.06795168e-01 2.62709528e-01 -5.85706651e-01
1.29518700e+00 -9.82435822e-01 -9.65555251e-01 -2.55278289e-01
3.23848993e-01 -5.25041282e-01 8.93383443e-01 2.63484955e-01
-6.91854298e-01 -5.89304149e-01 -9.19541717e-01 1.07898876e-01
-4.47819121e-02 7.37607852e-02 6.56195104e-01 2.95600325e-01
-4.18403506e-01 4.87226605e-01 -7.08306909e-01 -3.65792513e-02
7.32187867e-01 6.92950785e-02 -3.51361454e-01 -4.44136471e-01
-1.05913043e+00 4.52594161e-01 4.78215456e-01 1.59359932e-01
-1.08154929e+00 -8.74423623e-01 -6.93969846e-01 2.93234624e-02
1.14043450e+00 -8.21374357e-01 1.00511265e+00 -1.53530884e+00
-1.20275426e+00 4.93256599e-01 -4.54451069e-02 -3.80500287e-01
8.38753939e-01 -2.81554729e-01 -3.23331624e-01 6.58616796e-02
3.10621429e-02 5.20838201e-01 1.28216159e+00 -1.36306787e+00
-6.37821734e-01 -6.01206958e-01 5.69552958e-01 2.10441455e-01
-5.66058233e-02 -3.85829985e-01 -4.62760925e-01 -8.75089943e-01
1.63173769e-02 -9.16166008e-01 1.52240962e-01 3.34821254e-01
-4.37125891e-01 -1.97090387e-01 1.09094381e+00 -6.80550575e-01
9.53416228e-01 -2.30326748e+00 1.00160487e-01 -4.05985594e-01
3.48986208e-01 1.70562744e-01 -3.39478731e-01 1.41221350e-02
-2.58277565e-01 -9.92332473e-02 3.43381949e-02 -3.63869257e-02
-4.25842637e-03 1.94768280e-01 -4.74998206e-01 5.54134429e-01
1.04770184e-01 8.79875183e-01 -1.19389272e+00 -5.06054878e-01
3.27301502e-01 4.29787338e-01 -5.59540212e-01 3.54555659e-02
-5.27491808e-01 5.29313326e-01 -4.96643722e-01 5.18287838e-01
8.97260964e-01 -1.07453443e-01 4.30868775e-01 -6.34385407e-01
1.80082709e-01 1.09360382e-01 -1.24297690e+00 1.49574184e+00
-1.04335472e-01 3.69042307e-01 -9.74160507e-02 -1.08243251e+00
2.94967502e-01 3.44114572e-01 5.32266736e-01 -7.68834114e-01
4.45703082e-02 -2.89095402e-01 2.44234145e-01 -9.14808750e-01
-1.23800658e-01 -2.27600858e-01 3.15059006e-01 3.26355189e-01
2.22855192e-02 4.49984968e-01 9.98880938e-02 3.04108471e-01
1.02308631e+00 5.72937667e-01 3.15988779e-01 1.94099888e-01
3.35093021e-01 -1.28198743e-01 1.01305628e+00 8.70580316e-01
-4.17595446e-01 3.89407814e-01 4.13813174e-01 -5.68795621e-01
-6.62620187e-01 -1.00618684e+00 2.78155565e-01 1.10371053e+00
3.66786748e-01 -1.47442132e-01 -6.63722456e-01 -1.24527669e+00
1.54659823e-01 9.37680483e-01 -1.00435948e+00 -4.25083548e-01
-4.99742031e-01 -7.12668538e-01 1.30386546e-01 5.41660726e-01
7.58408368e-01 -1.14655042e+00 -3.64079714e-01 1.54108815e-02
-4.36348498e-01 -1.16975594e+00 -6.98896766e-01 -5.10479689e-01
-6.63290083e-01 -1.35465181e+00 -4.43971843e-01 -1.91774562e-01
5.58297575e-01 5.99799871e-01 7.68218338e-01 8.17526430e-02
-1.95501387e-01 3.85647148e-01 -3.58902454e-01 -4.26712424e-01
-1.12977162e-01 -7.49264002e-01 5.97356856e-02 7.30291903e-01
3.37056398e-01 -6.83962047e-01 -8.89369965e-01 1.51798069e-01
-1.05088615e+00 4.03715163e-01 6.91957891e-01 7.46528685e-01
4.44147587e-01 3.21374744e-01 3.43100995e-01 -1.02602494e+00
6.88335523e-02 -5.87195516e-01 -3.78508449e-01 3.26050133e-01
-2.35686436e-01 4.63883951e-02 6.77159846e-01 -9.31017935e-01
-1.60186708e+00 1.23587050e-01 5.03979862e-01 -8.94511521e-01
-2.38044769e-01 8.46801773e-02 -5.72305560e-01 4.24237460e-01
2.45342985e-01 2.77173132e-01 -1.40082136e-01 -4.48585838e-01
4.15313423e-01 2.29181513e-01 4.93963391e-01 -4.88131493e-01
8.26804996e-01 1.00405622e+00 7.39610419e-02 -4.55479801e-01
-1.22635627e+00 -2.78636277e-01 -5.60656190e-01 -4.26630616e-01
9.58608031e-01 -8.88146996e-01 -7.90062785e-01 4.62472439e-01
-1.20352483e+00 9.49977860e-02 -1.81191713e-01 5.13497055e-01
-4.31353629e-01 3.87412876e-01 -1.46804363e-01 -9.32696879e-01
1.87212214e-01 -9.94544327e-01 1.01290977e+00 -7.59340357e-03
1.24841079e-01 -6.56261265e-01 -1.85040236e-01 6.17571473e-01
-7.71139562e-02 4.77354646e-01 8.41616631e-01 -5.14856935e-01
-8.57904971e-01 -1.10905312e-01 -3.09826344e-01 1.94748089e-01
4.24697757e-01 -4.21720147e-02 -1.14102423e+00 -8.54672194e-02
2.45823935e-01 7.06605241e-03 1.09002769e+00 4.69268203e-01
1.18096483e+00 -8.12909663e-01 -4.55466747e-01 2.54067183e-01
1.22694111e+00 3.31202209e-01 5.25381684e-01 5.25685661e-02
1.12174678e+00 7.43285179e-01 6.63183272e-01 3.25971365e-01
2.63951480e-01 7.74923563e-01 7.62245417e-01 4.68035825e-02
-4.05672222e-01 -4.87541974e-01 4.87220317e-01 7.93966427e-02
-3.09596837e-01 -2.08026826e-01 -3.58958691e-01 5.65383911e-01
-2.30379248e+00 -1.50801301e+00 -1.70874104e-01 2.05291581e+00
6.30924702e-01 -9.55258962e-04 1.42370425e-02 -1.23083048e-01
9.36458468e-01 4.30153549e-01 -7.63784587e-01 3.29864442e-01
-6.65639639e-02 -4.60076362e-01 2.61966020e-01 7.71750063e-02
-1.14682651e+00 7.42882967e-01 4.70040560e+00 8.04350197e-01
-8.28490913e-01 1.92641035e-01 4.01317775e-01 -5.23883581e-01
-1.99280426e-01 3.24978828e-02 -4.38090563e-01 6.39269650e-01
2.59758145e-01 1.58630311e-02 4.59748566e-01 8.09659660e-01
5.41367829e-01 -8.97287801e-02 -1.33720469e+00 8.09846282e-01
1.24933489e-01 -1.21658325e+00 5.84485412e-01 -1.14135183e-02
7.31513917e-01 -4.73051935e-01 3.17035727e-02 2.72243023e-01
3.76332611e-01 -5.44764161e-01 1.07026792e+00 7.05222785e-01
3.84321481e-01 -2.68865645e-01 5.81985235e-01 3.89925361e-01
-1.04014564e+00 -4.23835605e-01 -1.85342893e-01 -3.84656876e-01
-6.85975105e-02 4.86581951e-01 -4.66408163e-01 5.69228053e-01
5.79677939e-01 1.01493275e+00 -4.58698601e-01 8.39845181e-01
-3.92546147e-01 6.36281908e-01 1.56767759e-02 2.57742852e-01
1.18771516e-01 -1.02026425e-01 7.97610283e-01 8.34434092e-01
-1.51086211e-01 2.06179321e-01 2.01297075e-01 9.80496228e-01
-1.33404166e-01 -1.39649615e-01 -7.64175355e-01 9.32792947e-02
3.03640813e-01 8.12976062e-01 -5.40401161e-01 -5.34498513e-01
-7.62776673e-01 1.05924201e+00 2.66921014e-01 6.42706156e-01
-9.14903045e-01 3.70021850e-01 6.49120688e-01 7.34139457e-02
5.52424729e-01 3.16395521e-01 1.00585572e-01 -1.58902299e+00
3.05622309e-01 -9.81494904e-01 5.65481544e-01 -9.52090681e-01
-1.37307656e+00 -9.30027589e-02 3.32661122e-01 -1.35053575e+00
1.14747785e-01 -5.23622155e-01 -5.84452152e-01 2.82318652e-01
-1.37787592e+00 -1.32891309e+00 -2.16491938e-01 5.94145000e-01
9.39298570e-01 2.52317488e-01 1.56798020e-01 1.15959182e-01
-8.00754786e-01 2.67132282e-01 -1.46601453e-01 1.08600549e-01
7.25045383e-01 -1.02153254e+00 -1.38116121e-01 1.00112629e+00
2.57385582e-01 7.02666104e-01 7.97944307e-01 -8.88592541e-01
-1.44198763e+00 -1.17980719e+00 4.35027689e-01 -4.51580495e-01
6.81886077e-01 -2.77997226e-01 -9.04421985e-01 9.15582359e-01
1.37457982e-01 1.34345159e-01 2.50496894e-01 -3.44142155e-03
-7.29298532e-01 -1.93236068e-01 -1.01232433e+00 8.87279510e-01
1.49776340e+00 -3.37910950e-01 -8.79582167e-01 3.32313329e-01
8.50107789e-01 -1.11832894e-01 -2.58614779e-01 4.85718817e-01
8.28558028e-01 -9.19107914e-01 1.03222704e+00 -9.20781374e-01
6.98652208e-01 -5.18878460e-01 -1.45136639e-01 -1.08898163e+00
-4.46723849e-01 -3.49032760e-01 -3.31089348e-01 1.19913089e+00
-4.23291419e-03 -3.63389671e-01 5.14157712e-01 7.00352609e-01
7.32400343e-02 -4.20765042e-01 -8.57813716e-01 -7.58231401e-01
-3.29640359e-01 -3.34035635e-01 6.75404370e-01 1.18292129e+00
-3.17647070e-01 3.64843518e-01 -7.60256350e-01 3.18122238e-01
7.39566445e-01 4.14896220e-01 8.13211679e-01 -9.77350533e-01
-4.93571132e-01 -1.97849512e-01 -4.34502333e-01 -8.60257089e-01
3.46145838e-01 -5.18131256e-01 -1.19631931e-01 -1.28277397e+00
7.34771073e-01 3.01235706e-01 -4.38640654e-01 6.26695335e-01
-5.63220620e-01 7.01923892e-02 4.95884269e-01 2.50575572e-01
-7.35825360e-01 5.89482903e-01 1.67201555e+00 -5.02367079e-01
7.63029046e-03 -2.12913215e-01 -6.14234269e-01 1.09262478e+00
3.10964584e-01 -5.45460522e-01 -6.54394746e-01 -4.08919752e-01
-1.04176544e-01 -8.12835153e-03 9.48760867e-01 -5.75784326e-01
4.36495245e-02 -5.50770104e-01 5.00320375e-01 -3.99258077e-01
2.96724319e-01 -1.09571993e+00 2.60307580e-01 3.18761349e-01
-3.78305942e-01 -3.64450723e-01 1.14044711e-01 1.33112097e+00
1.88463069e-02 1.06220581e-01 5.01112998e-01 -3.52894336e-01
-7.85882711e-01 3.18073750e-01 -1.31047770e-01 -1.26088277e-01
1.15118480e+00 -2.71769375e-01 -4.58617359e-01 -2.06322566e-01
-7.12458909e-01 1.11818060e-01 3.40579927e-01 6.80240631e-01
4.97774005e-01 -1.35912979e+00 -5.88093638e-01 -4.32161391e-02
1.84793711e-01 -2.42192075e-01 7.70272851e-01 1.05996549e+00
3.89993966e-01 7.96320513e-02 -2.21937552e-01 -3.67626637e-01
-1.35086894e+00 1.08219457e+00 2.37361744e-01 -6.79816827e-02
-7.57469833e-01 6.36380255e-01 1.12822294e+00 4.32748981e-02
2.07102656e-01 -2.66458660e-01 -3.24873388e-01 2.13187277e-01
6.85328901e-01 3.95615846e-01 -3.40113848e-01 -6.32515073e-01
-2.77640492e-01 2.78315067e-01 -3.16739440e-01 1.78421423e-01
1.31538570e+00 -1.61273852e-01 1.13382623e-01 5.45406044e-01
1.02165735e+00 -7.98530653e-02 -1.74689579e+00 -3.27289373e-01
-4.03848261e-01 -7.54589021e-01 1.27837792e-01 -9.74797189e-01
-1.32819474e+00 5.84407568e-01 4.11190808e-01 -2.35750843e-02
1.31601059e+00 3.09157874e-02 4.61468518e-01 7.88528547e-02
1.89070493e-01 -8.25050354e-01 2.21576184e-01 -1.49431422e-01
9.68371153e-01 -1.32758892e+00 2.85581350e-02 -4.88912195e-01
-5.97793639e-01 7.82314718e-01 7.65498340e-01 -1.18067726e-01
3.81521553e-01 -1.24136351e-01 -2.70053864e-01 -3.48455071e-01
-9.08669829e-01 -1.91699624e-01 4.35517013e-01 5.56708038e-01
7.60224462e-02 4.35313210e-02 -4.38280761e-01 6.62208438e-01
3.24484497e-01 1.11582197e-01 3.90507072e-01 7.28275001e-01
8.94923508e-03 -6.14702284e-01 -4.23251122e-01 4.19031084e-01
-3.88623327e-01 -1.51618943e-01 -6.97162211e-01 8.97375762e-01
4.35552597e-01 9.00244415e-01 3.38145792e-02 -4.08505291e-01
4.43834603e-01 8.10298994e-02 5.80976009e-01 -4.80144054e-01
-1.15775205e-01 1.27494216e-01 3.21081728e-02 -7.98661113e-01
-9.27108347e-01 -9.23201859e-01 -9.94270980e-01 -8.82342309e-02
-3.44572842e-01 -3.71322989e-01 2.30305508e-01 1.19948685e+00
3.40601057e-01 6.81024730e-01 3.93176317e-01 -6.61998808e-01
-5.55381238e-01 -8.65902305e-01 -5.93202353e-01 1.01036155e+00
5.97821057e-01 -1.14281154e+00 -6.80734634e-01 4.49157566e-01] | [8.656194686889648, 0.7551302313804626] |
04a4fb03-d348-499b-8a1b-c0138221f8b2 | image-quality-assessment-for-omnidirectional | 1904.04960 | null | http://arxiv.org/abs/1904.04960v2 | http://arxiv.org/pdf/1904.04960v2.pdf | Image Quality Assessment for Omnidirectional Cross-reference Stitching | Along with the development of virtual reality (VR), omnidirectional images
play an important role in producing multimedia content with immersive
experience. However, despite various existing approaches for omnidirectional
image stitching, how to quantitatively assess the quality of stitched images is
still insufficiently explored. To address this problem, we establish a novel
omnidirectional image dataset containing stitched images as well as
dual-fisheye images captured from standard quarters of 0$^\circ$, 90$^\circ$,
180$^\circ$ and 270$^\circ$. In this manner, when evaluating the quality of an
image stitched from a pair of fisheye images (e.g., 0$^\circ$ and 180$^\circ$),
the other pair of fisheye images (e.g., 90$^\circ$ and 270$^\circ$) can be used
as the cross-reference to provide ground-truth observations of the stitching
regions. Based on this dataset, we further benchmark six widely used stitching
models with seven evaluation metrics for IQA. To the best of our knowledge, it
is the first dataset that focuses on assessing the stitching quality of
omnidirectional images. | ['Yu Zhang', 'Yifan Zhao', 'Long Xu', 'Kaiwen Yu', 'Jia Li'] | 2019-04-10 | null | null | null | null | ['image-stitching'] | ['computer-vision'] | [ 2.12006211e-01 -3.31867397e-01 2.94261694e-01 6.59741415e-03
-2.39109918e-01 -7.38845825e-01 5.26634574e-01 -2.58472979e-01
-3.47600758e-01 4.46208060e-01 -5.74768335e-02 -5.02964020e-01
-6.62182048e-02 -7.79751956e-01 -6.83491886e-01 -6.17632926e-01
-2.25817651e-01 -1.34422824e-01 2.07019731e-01 -5.54881930e-01
5.55812478e-01 1.16792731e-01 -1.68880534e+00 -7.87421390e-02
9.08567011e-01 1.06649113e+00 3.43027264e-01 7.19949007e-01
3.94435555e-01 2.13449419e-01 -5.85859001e-01 -5.32837272e-01
5.50995946e-01 -1.80512145e-01 -2.05944896e-01 -9.75564048e-02
7.32731283e-01 -7.47079492e-01 -4.93562639e-01 1.25711000e+00
3.43341351e-01 1.22488484e-01 3.45069110e-01 -1.20493138e+00
-6.18788183e-01 1.81451127e-01 -9.38364387e-01 3.37305367e-01
7.77796805e-01 7.53074586e-02 7.64286637e-01 -6.47030652e-01
8.50335836e-01 7.53544211e-01 2.88664281e-01 9.04171467e-02
-5.72195590e-01 -1.11185038e+00 -1.09849155e-01 -2.42310949e-02
-1.43522882e+00 -3.81281674e-01 9.31722164e-01 -3.08777571e-01
3.90335083e-01 2.68982977e-01 7.52023101e-01 8.69173288e-01
6.06832206e-01 4.85098988e-01 1.11450279e+00 -2.30372518e-01
6.95296004e-02 1.83810547e-01 -1.33446500e-01 7.21918166e-01
1.62444279e-01 6.13320112e-01 -1.91706046e-01 1.92686275e-01
7.87784636e-01 -1.48403160e-02 -6.41596854e-01 -4.98722911e-01
-1.00432563e+00 6.50009453e-01 4.56437975e-01 2.54361957e-01
1.32343337e-01 -4.47552763e-02 2.82905251e-01 3.74323606e-01
2.81935573e-01 3.39661479e-01 9.38330516e-02 -2.68674403e-01
-7.97020316e-01 8.82519335e-02 1.59265444e-01 1.24041641e+00
6.81263268e-01 3.92716974e-01 6.99314594e-01 9.33077157e-01
1.41833112e-01 8.74307632e-01 8.24145451e-02 -8.26633513e-01
6.01167440e-01 1.83458477e-01 3.57519269e-01 -1.45828593e+00
-5.07975854e-02 -3.15494299e-01 -7.76652336e-01 4.58454043e-01
2.11516708e-01 -3.89121920e-02 -6.65695488e-01 1.23070109e+00
5.58863729e-02 3.71150225e-01 2.27657273e-01 1.17574453e+00
1.09032631e+00 6.93075895e-01 -6.27117515e-01 -1.56443179e-01
1.37678909e+00 -7.64949977e-01 -4.58640337e-01 -2.17133716e-01
3.19308460e-01 -1.13126981e+00 9.67914164e-01 6.32223606e-01
-1.01128793e+00 -6.47225857e-01 -1.31099951e+00 3.27032149e-01
-2.43934751e-01 -2.25615710e-01 4.12324786e-01 9.72615480e-01
-1.01222563e+00 -6.17197230e-02 -1.70306444e-01 1.21981956e-01
-1.07978053e-01 -7.51657858e-02 -6.60189033e-01 -7.37556219e-01
-1.16877198e+00 6.77970350e-01 -4.08289023e-02 -6.47348166e-02
-7.97038972e-01 -7.70077765e-01 -1.13510799e+00 -2.54688591e-01
5.11416197e-01 -1.90454319e-01 9.30040479e-01 -6.70927405e-01
-1.25629938e+00 8.21454585e-01 1.92634851e-01 -8.48750100e-02
3.43032628e-01 -5.06064072e-02 -7.63050735e-01 2.66998470e-01
2.71293968e-01 5.26066542e-01 7.18551278e-01 -1.85370111e+00
-9.00793135e-01 -4.89810735e-01 2.45004892e-01 3.95087272e-01
1.31232515e-01 3.64610762e-03 -6.44273579e-01 -6.27251744e-01
3.25184971e-01 -8.71341169e-01 1.18120641e-01 -3.06224048e-01
-4.19334501e-01 5.46628475e-01 1.29771757e+00 -6.59721911e-01
1.00645626e+00 -2.36687136e+00 -2.23448604e-01 8.29374120e-02
2.96793431e-01 3.48571748e-01 -9.60004926e-02 3.81299406e-01
7.33114406e-02 1.17189988e-01 1.44907518e-03 -2.88882345e-01
-4.83090788e-01 1.05063021e-01 -3.29953820e-01 6.05334044e-01
-5.54649055e-01 5.41738033e-01 -1.19650328e+00 -1.81832150e-01
7.66614497e-01 3.19778740e-01 -5.08855402e-01 1.37003630e-01
3.18441182e-01 5.28339267e-01 -1.30234748e-01 7.75260985e-01
1.33689141e+00 4.32257116e-01 -8.22381005e-02 -2.40202904e-01
-6.77791655e-01 -2.92621911e-01 -1.03644919e+00 1.56172967e+00
-5.18708706e-01 8.36335540e-01 2.53760427e-01 -8.57315361e-01
1.11379802e+00 2.74475455e-01 6.36311889e-01 -1.24984574e+00
1.64156586e-01 3.32366347e-01 -2.70356119e-01 -4.45800692e-01
1.23584390e+00 -1.83245406e-01 -4.49844480e-01 3.43776643e-01
-3.88228834e-01 -8.85703504e-01 1.58819780e-01 2.33959690e-01
8.59077394e-01 -1.43924624e-01 8.29296038e-02 -7.95525089e-02
3.60167056e-01 -1.70860559e-01 4.80715185e-01 7.10403025e-01
-3.08022320e-01 1.11105549e+00 -8.79299175e-03 -5.14211476e-01
-1.39312828e+00 -1.25149608e+00 -1.97480291e-01 3.49317580e-01
1.10739362e+00 -3.94828349e-01 -5.32747269e-01 -3.48415524e-01
-3.34305912e-01 6.59731686e-01 -4.46704537e-01 -6.49715737e-02
-4.84869063e-01 -4.85762283e-02 6.06021464e-01 6.25803396e-02
1.13018477e+00 -8.70088577e-01 -9.72927630e-01 -3.22553649e-04
-4.28352267e-01 -1.27540553e+00 -8.03396702e-01 -3.73361319e-01
-4.69299018e-01 -1.13454866e+00 -8.48007441e-01 -6.02810204e-01
4.04185742e-01 1.35138357e+00 1.13824117e+00 1.53925970e-01
-2.91379541e-01 3.23671013e-01 -6.69814646e-01 -2.01971784e-01
-1.87304288e-01 -7.54424572e-01 -9.39509049e-02 -2.84488171e-01
-2.38480672e-01 -5.02195358e-01 -1.01428294e+00 1.00577009e+00
-1.34889185e+00 1.88411698e-01 1.92286253e-01 6.77031875e-01
3.03884298e-01 4.94932324e-01 -2.11628705e-01 -4.45740223e-01
3.51592958e-01 -3.80785316e-01 -6.92531526e-01 -1.40735339e-02
-3.62141192e-01 -5.08449376e-01 7.79048741e-01 -1.49064824e-01
-9.45890784e-01 -4.49988872e-01 -7.86490217e-02 -5.17713428e-01
-9.43209231e-02 4.59476113e-01 2.09514834e-02 -5.06931394e-02
4.26934361e-01 3.67535084e-01 -8.62438083e-02 -1.14057913e-01
3.51500690e-01 7.42778063e-01 9.15520370e-01 -1.40902355e-01
8.83426845e-01 9.61247265e-01 -4.23357636e-01 -1.26204014e+00
5.23354709e-02 -6.54932559e-01 -7.89237544e-02 -6.20536804e-01
6.09546900e-01 -7.39345253e-01 -4.63142723e-01 6.93041861e-01
-8.56843472e-01 7.93130398e-02 7.21930563e-02 6.55936420e-01
-5.46268761e-01 8.78184676e-01 -1.98164895e-01 -6.65496230e-01
-1.93879217e-01 -1.53371620e+00 1.25994074e+00 4.25247818e-01
1.38759464e-01 -6.83771968e-01 1.39958739e-01 5.62282503e-01
2.59593964e-01 1.41464218e-01 3.92090678e-01 4.59309161e-01
-4.35611188e-01 -3.02858204e-01 -4.85199809e-01 2.85534590e-01
2.84434199e-01 3.24990600e-01 -4.90232617e-01 -4.78349060e-01
-2.21382156e-02 -1.34575129e-01 4.18895245e-01 5.69198191e-01
4.63931739e-01 -3.80182825e-02 -2.08626390e-01 8.51889670e-01
1.32517970e+00 9.18217182e-01 1.40992010e+00 4.93845612e-01
2.60427237e-01 3.94842714e-01 1.09547079e+00 5.86537302e-01
4.11203384e-01 1.02673841e+00 7.26920843e-01 -7.73099735e-02
7.29527399e-02 -2.81866193e-01 1.77870005e-01 3.91687661e-01
-2.46172756e-01 -4.96560246e-01 -7.41525173e-01 4.18481737e-01
-1.04070950e+00 -9.72089112e-01 1.39182046e-01 2.31346130e+00
2.24077538e-01 -6.48374259e-02 -1.98502555e-01 2.92176425e-01
7.95287609e-01 6.41179383e-01 -8.88414383e-02 -4.54684675e-01
-1.47733361e-01 3.58711407e-02 4.49508995e-01 5.43642700e-01
-8.96314085e-01 8.02560568e-01 4.88606405e+00 7.31686175e-01
-1.32508790e+00 -2.98468977e-01 2.57873178e-01 3.19906116e-01
-6.90539420e-01 1.17020287e-01 -4.06465471e-01 8.13730299e-01
3.15427840e-01 4.37725848e-03 3.11537206e-01 6.88039184e-01
3.02179247e-01 -6.96719110e-01 -4.40784752e-01 1.32435822e+00
3.47406328e-01 -1.16323578e+00 -2.86238164e-01 2.83805966e-01
1.01711309e+00 -2.30739996e-01 5.04287064e-01 -1.94180384e-02
5.19628882e-01 -8.44408572e-01 9.20004606e-01 -1.50830043e-03
1.28410971e+00 -7.92561591e-01 6.90456152e-01 2.42734954e-01
-1.45450306e+00 2.39167176e-02 -3.21003437e-01 2.21892357e-01
5.68274319e-01 5.95250905e-01 -4.57474172e-01 1.01925302e+00
1.09162271e+00 6.83934152e-01 -2.08009005e-01 9.34378743e-01
6.34568511e-04 -2.12646797e-02 -1.17017977e-01 3.60686898e-01
3.69941086e-01 -5.48046947e-01 7.58929968e-01 7.95869589e-01
8.78495991e-01 2.40985096e-01 -1.45546779e-01 4.85691607e-01
2.60237128e-01 -1.95666954e-01 -1.41534781e+00 2.28169531e-01
4.61403459e-01 9.29051220e-01 -6.53341711e-01 -1.46209776e-01
-3.38320225e-01 9.51072693e-01 -2.97423482e-01 3.27470005e-01
-1.02755117e+00 -5.08895159e-01 8.01448584e-01 1.24333300e-01
1.58871472e-01 -6.10372424e-01 -7.97937885e-02 -1.40172613e+00
-1.01502925e-01 -9.95561421e-01 1.81825116e-01 -1.33693147e+00
-4.36252356e-01 7.10853279e-01 1.06561750e-01 -1.87942898e+00
-1.18207246e-01 -1.97178975e-01 -4.48024273e-01 5.13847351e-01
-1.30591035e+00 -7.30753899e-01 -8.10827374e-01 5.85722983e-01
3.97674412e-01 -1.27843410e-01 3.58466268e-01 2.37479657e-01
-1.39766410e-01 5.20994008e-01 4.13084328e-01 4.36803997e-02
4.94706720e-01 -6.41140044e-01 3.10010523e-01 9.54585731e-01
8.26430470e-02 5.72214067e-01 9.98088360e-01 -4.49111313e-01
-1.36630321e+00 -5.34207702e-01 1.35552809e-01 5.77726215e-02
1.38201773e-01 -6.59344494e-02 -5.74369550e-01 2.98308194e-01
2.43184701e-01 1.15772627e-01 3.09763134e-01 -5.52139699e-01
-4.93126363e-01 -2.26285085e-01 -1.37297320e+00 9.37386632e-01
1.16108847e+00 -3.42112780e-01 -8.03450793e-02 -4.92957532e-01
4.12379682e-01 -7.03006804e-01 -6.78836167e-01 4.50756580e-01
9.16377068e-01 -1.73836684e+00 1.17936742e+00 3.55524540e-01
8.37421775e-01 -2.83205956e-01 -3.82764369e-01 -1.58735359e+00
1.69448644e-01 -4.02148575e-01 6.70636058e-01 8.32179010e-01
-1.68016732e-01 -8.72347653e-01 8.22409093e-01 -1.47673693e-02
-5.38660645e-01 -2.28315756e-01 -1.12365210e+00 -7.41708875e-01
-9.75997597e-02 -3.93856913e-01 4.38787371e-01 8.25946689e-01
1.84145838e-01 -3.34378123e-01 -7.25856125e-01 2.40218535e-01
8.30047846e-01 1.42217666e-01 1.03833055e+00 -6.88473046e-01
-4.40679044e-02 -2.56486446e-01 -6.10194445e-01 -1.35831487e+00
-3.22946608e-01 -3.21228266e-01 5.50106056e-02 -1.12299907e+00
-5.68718510e-03 -8.32416296e-01 2.83406436e-01 -3.10970932e-01
7.82660097e-02 7.59980321e-01 1.95633799e-01 2.76310086e-01
-2.44863927e-01 4.61210489e-01 1.49069250e+00 -1.80460036e-01
1.46681201e-02 -2.64664114e-01 -6.36969388e-01 5.26799023e-01
8.50916743e-01 -1.00909673e-01 -6.58138692e-01 -4.72742260e-01
1.18873700e-01 8.40055585e-01 3.44849527e-01 -8.84798646e-01
-3.24641801e-02 -2.04051137e-01 -1.49852857e-01 -7.39253700e-01
4.86103356e-01 -7.39159942e-01 4.61943120e-01 5.51942110e-01
7.92705938e-02 3.66736472e-01 8.04195646e-03 4.56088692e-01
-5.50011575e-01 -3.12165707e-01 9.39842522e-01 -1.18273258e-01
-9.12551224e-01 1.84211373e-01 -3.21055591e-01 -1.14899538e-01
1.09530318e+00 -9.88195360e-01 -5.33501685e-01 -9.05765116e-01
-1.05879353e-02 -4.47393022e-02 9.15543854e-01 6.36338651e-01
1.28103554e+00 -1.22467721e+00 -4.45314735e-01 5.86738050e-01
3.03863764e-01 -2.60968395e-02 7.41350293e-01 5.49238741e-01
-7.69746184e-01 5.11373699e-01 -5.43193817e-01 -6.65234327e-01
-1.54892647e+00 6.04232550e-01 1.55151397e-01 -1.76700577e-01
-6.42972350e-01 6.34286284e-01 4.05508101e-01 -2.74503291e-01
-1.73711777e-01 -1.96919456e-01 -2.32566684e-01 -4.06752884e-01
5.21937311e-01 2.61726946e-01 2.38578953e-02 -8.60326588e-01
-2.16005459e-01 9.62657809e-01 -8.69107321e-02 -6.76451564e-01
8.37987900e-01 -5.18266916e-01 2.10573062e-01 -2.11103708e-01
1.40778089e+00 5.73465168e-01 -1.27986526e+00 3.61310840e-02
-6.85885489e-01 -1.34337628e+00 -6.47572652e-02 -3.16575378e-01
-1.37219298e+00 7.50963867e-01 5.46877325e-01 1.28129587e-01
1.02664697e+00 -2.24877641e-01 7.38768458e-01 -3.53673577e-01
1.06906283e+00 -8.49956751e-01 4.62970018e-01 3.44976455e-01
8.13697159e-01 -1.13333035e+00 -4.15507071e-02 -6.30868852e-01
-6.55763924e-01 9.40687835e-01 5.30416131e-01 -1.37679234e-01
3.34223390e-01 4.01920557e-01 3.17935854e-01 -2.46587187e-01
-1.99887291e-01 2.17623115e-01 3.82874757e-02 7.87558317e-01
2.17395902e-01 -1.37270182e-01 -3.56551081e-01 1.21375900e-02
-5.27643979e-01 -1.01286821e-01 9.47866142e-01 1.01446176e+00
-4.12976235e-01 -8.12308908e-01 -6.44513547e-01 1.04069479e-01
-1.26462281e-01 -1.18906349e-01 1.30222499e-01 8.77054751e-01
3.81522924e-02 1.39840281e+00 4.33250964e-02 -9.07656550e-01
3.86019975e-01 -8.00025165e-01 3.80533934e-01 -1.39152288e-01
-3.29120815e-01 2.02509575e-02 1.58879787e-01 -2.92876482e-01
-2.01198056e-01 -3.78843755e-01 -1.08335853e+00 -8.20198059e-01
-3.27035099e-01 3.45181525e-01 7.80074775e-01 4.97849047e-01
3.15770954e-02 4.58827130e-02 1.06243408e+00 -1.00392509e+00
-7.25316554e-02 -4.45907235e-01 -9.64168489e-01 5.53747118e-01
2.19945520e-01 -8.94773126e-01 -6.23133600e-01 -2.07766116e-01] | [9.592329978942871, -2.344963312149048] |
59b56b0c-9e94-4407-9c81-b3e4853859e5 | x-reid-cross-instance-transformer-for | 2302.02075 | null | https://arxiv.org/abs/2302.02075v1 | https://arxiv.org/pdf/2302.02075v1.pdf | X-ReID: Cross-Instance Transformer for Identity-Level Person Re-Identification | Currently, most existing person re-identification methods use Instance-Level features, which are extracted only from a single image. However, these Instance-Level features can easily ignore the discriminative information due to the appearance of each identity varies greatly in different images. Thus, it is necessary to exploit Identity-Level features, which can be shared across different images of each identity. In this paper, we propose to promote Instance-Level features to Identity-Level features by employing cross-attention to incorporate information from one image to another of the same identity, thus more unified and discriminative pedestrian information can be obtained. We propose a novel training framework named X-ReID. Specifically, a Cross Intra-Identity Instances module (IntraX) fuses different intra-identity instances to transfer Identity-Level knowledge and make Instance-Level features more compact. A Cross Inter-Identity Instances module (InterX) involves hard positive and hard negative instances to improve the attention response to the same identity instead of different identity, which minimizes intra-identity variation and maximizes inter-identity variation. Extensive experiments on benchmark datasets show the superiority of our method over existing works. Particularly, on the challenging MSMT17, our proposed method gains 1.1% mAP improvements when compared to the second place. | ['Guiguang Ding', 'Yuchen Guo', 'Tao He', 'Leqi Shen'] | 2023-02-04 | null | null | null | null | ['person-re-identification'] | ['computer-vision'] | [ 7.71200284e-02 -5.13370275e-01 -1.39293931e-02 -6.99014068e-01
-5.04809916e-01 -4.27143455e-01 5.32525539e-01 7.26426160e-03
-5.86146712e-01 6.85855448e-01 8.36639702e-02 3.56277436e-01
3.86539288e-02 -7.34343052e-01 -8.33929420e-01 -8.23589027e-01
3.24320912e-01 1.93733886e-01 6.47040680e-02 2.38305493e-03
1.76723912e-01 1.77929416e-01 -1.56942260e+00 3.37644070e-01
1.05062675e+00 1.04341316e+00 3.13745558e-01 -4.17254083e-02
-3.79285477e-02 4.11618710e-01 -5.84133744e-01 -6.13998353e-01
5.02962947e-01 -3.71381313e-01 -5.75278401e-01 1.01352058e-01
8.01564813e-01 -3.37880701e-01 -4.89738226e-01 1.38906217e+00
5.01022696e-01 1.59294829e-01 4.41903323e-01 -1.42786682e+00
-1.06147325e+00 2.35336587e-01 -1.09255731e+00 1.98609367e-01
2.04524741e-01 3.46305162e-01 6.50395691e-01 -7.63515413e-01
4.16943848e-01 1.50615227e+00 6.78484917e-01 5.44618249e-01
-1.12524211e+00 -1.18309164e+00 5.88876128e-01 7.13433087e-01
-1.79046941e+00 -4.91394073e-01 6.55591369e-01 -2.91788042e-01
5.59698164e-01 2.44594499e-01 4.06184375e-01 9.48687017e-01
-2.99833208e-01 1.00684237e+00 1.31149137e+00 -1.34907171e-01
-3.80642742e-01 4.36649740e-01 2.26600602e-01 4.03354496e-01
2.95550048e-01 1.65181905e-01 -3.01944345e-01 2.59409547e-01
5.98835051e-01 2.99477518e-01 -3.55849206e-01 -1.96082205e-01
-1.18444633e+00 5.12192905e-01 8.75714064e-01 8.01135749e-02
-2.74539709e-01 -2.62265772e-01 4.60960120e-01 1.07644789e-01
2.28785217e-01 2.31583659e-02 -2.88095236e-01 7.08118379e-02
-5.96547246e-01 1.12534970e-01 1.12040132e-01 1.10875452e+00
1.14882696e+00 -4.45910126e-01 -4.11108673e-01 1.18081105e+00
1.31392658e-01 4.64340508e-01 5.38383365e-01 -5.52798569e-01
6.32002175e-01 8.08567286e-01 7.35387355e-02 -1.15674865e+00
-1.60086319e-01 -6.42785370e-01 -9.83266592e-01 -3.73893864e-02
2.97781736e-01 -9.48061273e-02 -8.10497940e-01 1.94939804e+00
4.57848459e-01 4.68400449e-01 -7.99667761e-02 1.09620035e+00
9.32432771e-01 6.01466238e-01 2.97698498e-01 6.70860261e-02
1.45322895e+00 -1.10203159e+00 -4.22867119e-01 -3.90696913e-01
2.18428478e-01 -7.05752909e-01 7.12575674e-01 -1.85792401e-01
-7.65661538e-01 -1.02246797e+00 -1.03676319e+00 3.82839106e-02
-5.72711945e-01 4.33980495e-01 2.48976156e-01 5.90688348e-01
-7.81862020e-01 2.79430449e-01 -1.88128769e-01 -4.13772345e-01
6.55230522e-01 5.38913906e-01 -8.05637956e-01 -3.91339988e-01
-1.16818345e+00 7.65063763e-01 5.40603340e-01 2.57581919e-01
-3.46143842e-01 -7.84629583e-01 -9.49885070e-01 1.71592757e-02
3.09491009e-01 -6.30331457e-01 7.33606637e-01 -1.31797242e+00
-1.14526582e+00 1.00289023e+00 -5.12272775e-01 1.27073629e-02
4.84143376e-01 -6.49283007e-02 -5.84281445e-01 -9.10296515e-02
5.30145228e-01 7.80614495e-01 5.62992513e-01 -1.57653129e+00
-9.32914376e-01 -6.11441195e-01 1.43480137e-01 2.76452512e-01
-2.93196291e-01 8.01201761e-02 -8.10326278e-01 -6.16956770e-01
-4.73037548e-02 -8.27015281e-01 1.80093035e-01 -1.27257630e-01
-4.13836420e-01 -3.53134811e-01 6.93417907e-01 -8.34386647e-01
9.06167269e-01 -2.09195137e+00 -2.01325994e-02 1.13938354e-01
4.21643555e-02 4.04595226e-01 -3.62233192e-01 4.99069914e-02
-1.92846775e-01 2.68655065e-02 7.23604858e-02 -3.78561437e-01
-4.95286882e-02 3.65499780e-02 1.31799176e-01 4.07900512e-01
4.39965338e-01 9.08876896e-01 -7.81276941e-01 -5.84640205e-01
3.38480055e-01 5.15855968e-01 -2.72941530e-01 1.30454257e-01
4.51001674e-01 6.25776112e-01 -4.26468968e-01 6.55240297e-01
1.14380205e+00 -1.80295169e-01 -9.36007425e-02 -5.63419282e-01
-1.15211882e-01 -1.26880154e-01 -1.21025443e+00 1.42703521e+00
-5.79044342e-01 3.11861098e-01 -7.03126788e-02 -1.02290559e+00
8.03591669e-01 -1.18160926e-01 2.56291240e-01 -1.02848446e+00
1.02621213e-01 -5.41658960e-02 -2.49856021e-02 -3.55808318e-01
3.32920343e-01 7.72259384e-02 -6.30788952e-02 2.40463287e-01
2.00447794e-02 9.93268013e-01 1.22716546e-01 8.61611068e-02
4.08957928e-01 8.29487070e-02 1.79906607e-01 -1.67381406e-01
9.91787732e-01 -4.07072574e-01 9.93875623e-01 7.41908371e-01
-4.80326176e-01 7.24882245e-01 1.27424628e-01 -4.50374424e-01
-9.48594511e-01 -9.64293838e-01 -2.93707937e-01 9.49563503e-01
9.71431971e-01 -1.14935957e-01 -6.66194379e-01 -7.93450952e-01
2.47592553e-01 2.12503850e-01 -9.51253712e-01 -1.10617168e-01
-5.81206918e-01 -8.69999707e-01 2.95160741e-01 6.90863013e-01
1.01550364e+00 -8.10849428e-01 2.35134870e-01 5.58153763e-02
-3.21954429e-01 -1.14506745e+00 -9.81632352e-01 -5.79744041e-01
-2.53452212e-01 -9.78779137e-01 -9.13304806e-01 -8.57837558e-01
8.51557434e-01 6.79124475e-01 8.19154561e-01 2.39675239e-01
-3.96667063e-01 1.40498832e-01 -3.28540504e-01 -1.99344456e-01
1.76677480e-01 -9.06766057e-02 1.51456773e-01 8.11453938e-01
6.50182366e-01 -3.86062771e-01 -8.63600254e-01 5.59812009e-01
-4.40804988e-01 5.41822538e-02 7.50357985e-01 1.13335502e+00
6.60879493e-01 3.33284959e-02 7.94847071e-01 -6.43582165e-01
1.88346684e-01 -6.38974667e-01 -2.50469863e-01 6.20266795e-01
-3.77829283e-01 -2.66403139e-01 6.26852155e-01 -6.17320776e-01
-1.28133774e+00 4.06580456e-02 1.39157981e-01 -3.67647976e-01
-2.30892003e-01 3.79398875e-02 -8.80589902e-01 -2.73967147e-01
1.03815690e-01 5.03167033e-01 -5.18088043e-02 -4.19584185e-01
2.26605207e-01 7.89009690e-01 6.55936718e-01 -7.20913351e-01
8.04433644e-01 3.99472415e-01 -3.54661912e-01 -4.50204402e-01
-8.45101118e-01 -6.34690583e-01 -7.51186907e-01 -1.79810897e-01
8.22050214e-01 -1.20121801e+00 -9.61755812e-01 8.31482410e-01
-1.01644599e+00 1.04626998e-01 1.05751203e-02 3.28674585e-01
-6.30677715e-02 4.95705396e-01 -4.40466523e-01 -6.20917559e-01
-2.47053191e-01 -1.27585709e+00 1.19271636e+00 1.03449667e+00
2.21114546e-01 -7.55208552e-01 -3.39931399e-01 4.80549395e-01
2.80698836e-01 1.11750670e-01 4.99915510e-01 -5.89041471e-01
-6.16256416e-01 -3.92090023e-01 -9.84912157e-01 3.41840506e-01
3.81871104e-01 -3.03977519e-01 -9.05100644e-01 -4.32777673e-01
-4.24656838e-01 -2.24779814e-01 8.79176676e-01 5.28654009e-02
1.31326914e+00 -2.35599652e-01 -5.73568583e-01 7.61128306e-01
1.47823048e+00 6.08640313e-02 6.16457701e-01 5.05130708e-01
1.06133783e+00 5.85683048e-01 6.95798159e-01 2.15392798e-01
7.89947033e-01 1.00609672e+00 1.15284093e-01 -1.95924118e-01
-2.21622661e-01 -3.54392767e-01 1.56109899e-01 3.92720878e-01
-2.52763182e-01 1.39401376e-01 -4.05001134e-01 6.58691883e-01
-1.82870340e+00 -1.24395514e+00 -1.59689672e-02 2.32446647e+00
6.80347502e-01 -2.50309914e-01 3.84865791e-01 -2.86037952e-01
1.31608498e+00 -9.62881595e-02 -7.81783462e-01 2.18469277e-01
-2.95464933e-01 -6.28467798e-02 5.14984846e-01 2.95623749e-01
-1.39566958e+00 7.44185627e-01 4.95882654e+00 1.09229195e+00
-9.40646768e-01 2.10340381e-01 6.91294670e-01 -7.83589780e-02
4.85429866e-03 -1.94971353e-01 -1.02490067e+00 1.07223499e+00
2.03400776e-01 -3.40759605e-01 2.95155317e-01 7.65404522e-01
-8.95585269e-02 9.05917212e-02 -9.59516346e-01 1.33206427e+00
2.52481103e-01 -9.12086308e-01 1.61539704e-01 1.73388511e-01
7.73147702e-01 -4.01256263e-01 1.48758009e-01 5.74463785e-01
1.23902649e-01 -9.32475090e-01 5.33634484e-01 6.13062918e-01
7.43661880e-01 -1.01048386e+00 1.02178407e+00 1.11229494e-01
-1.67732382e+00 -5.68095632e-02 -5.85408807e-01 1.38908446e-01
-1.90653163e-03 1.61991835e-01 -2.88114041e-01 1.00528228e+00
8.67881417e-01 9.05365348e-01 -8.07386816e-01 1.25640917e+00
3.17306332e-02 3.18751507e-03 -1.65447116e-01 3.06860119e-01
1.01720549e-01 -2.50153244e-01 2.14662969e-01 1.18569100e+00
2.28970006e-01 2.35935152e-01 5.68583965e-01 8.04399073e-01
-6.37353882e-02 2.29918778e-01 -2.32523665e-01 6.52365685e-01
5.88143289e-01 1.26781273e+00 -2.25241795e-01 -5.56657851e-01
-6.91557288e-01 1.41556001e+00 4.78395015e-01 4.30105805e-01
-9.58097696e-01 -4.12770182e-01 1.18368328e+00 -8.83856788e-02
3.49413186e-01 2.77241290e-01 -1.28668383e-01 -1.36335945e+00
3.58169019e-01 -7.42287993e-01 5.74419439e-01 -4.85199928e-01
-1.88050115e+00 5.64662397e-01 -1.53177187e-01 -1.32110071e+00
2.09816188e-01 -4.61458027e-01 -6.97303593e-01 1.34675133e+00
-1.59123969e+00 -1.63684404e+00 -7.23584592e-01 6.50860548e-01
4.41042453e-01 -2.51596659e-01 5.31126499e-01 6.97585881e-01
-9.84911263e-01 1.37325835e+00 1.65002614e-01 5.80369353e-01
9.49501157e-01 -1.01619351e+00 3.44748229e-01 8.16040814e-01
-2.01585039e-01 8.75582933e-01 1.53005242e-01 -5.67659795e-01
-1.14057803e+00 -1.38359129e+00 7.29281902e-01 -3.66186947e-01
2.83042938e-01 -1.34919629e-01 -1.07834911e+00 6.19372964e-01
-5.07624671e-02 2.48567954e-01 7.91880548e-01 2.69144922e-01
-7.25500882e-01 -5.36628664e-01 -1.26077855e+00 3.91044766e-01
1.28948951e+00 -6.16828263e-01 -4.38824683e-01 1.59259185e-01
3.33417326e-01 -1.56857371e-01 -9.13163126e-01 4.67758477e-01
6.87630951e-01 -8.62480342e-01 1.27972472e+00 -4.88160789e-01
9.74597335e-02 -7.33033478e-01 -5.80680417e-03 -1.18921745e+00
-6.22863889e-01 -1.13040708e-01 2.39838541e-01 1.69843161e+00
-5.09650186e-02 -8.97558749e-01 4.86449480e-01 7.93411613e-01
1.04746237e-01 -5.53719640e-01 -9.10106540e-01 -1.05803728e+00
1.24276079e-01 2.33295128e-01 1.00621700e+00 1.12472236e+00
-3.20101410e-01 1.79221541e-01 -5.79347074e-01 4.48606938e-01
8.22934330e-01 3.95773113e-01 9.09679770e-01 -1.10682189e+00
-1.96593359e-01 -5.93474388e-01 -8.41381252e-01 -1.02314293e+00
3.15938056e-01 -9.58033025e-01 4.26822621e-03 -1.23010325e+00
8.63842487e-01 -7.83997893e-01 -6.49676263e-01 5.06519616e-01
-9.28339422e-01 5.33397973e-01 4.70024675e-01 2.62364686e-01
-6.56529725e-01 5.82285047e-01 1.24989581e+00 -5.77328920e-01
2.75651608e-02 -2.00085491e-01 -9.06169116e-01 3.68575722e-01
5.80948770e-01 -2.64873832e-01 -1.99965715e-01 -4.60514247e-01
-5.80922604e-01 -4.47482079e-01 6.26425028e-01 -1.15484214e+00
3.80026788e-01 -5.54754771e-02 9.68989670e-01 -4.73037273e-01
3.11729997e-01 -6.20185554e-01 2.00651288e-01 1.15068734e-01
-4.19469513e-02 -1.18224002e-01 2.24286124e-01 6.52831554e-01
-3.17941666e-01 -1.10860080e-01 9.02659833e-01 -1.41487122e-01
-1.11133516e+00 7.15757728e-01 5.12556732e-01 -1.44732103e-01
1.23130369e+00 -4.17816162e-01 -3.67099702e-01 -8.38974193e-02
-4.63357776e-01 5.49872220e-01 7.16619611e-01 7.73714602e-01
4.12870020e-01 -1.69926941e+00 -1.01085532e+00 2.11540923e-01
4.62083071e-01 -2.55776048e-01 9.20951784e-01 6.14972591e-01
8.54410008e-02 1.20198801e-01 -6.44475043e-01 -7.44801521e-01
-1.52439368e+00 7.99965262e-01 4.74573612e-01 8.27875733e-02
-4.38354582e-01 9.66178238e-01 9.51134324e-01 -4.63821173e-01
-8.10116082e-02 5.48470497e-01 -3.03764254e-01 1.16174690e-01
1.03297734e+00 2.27992982e-01 -2.51373708e-01 -1.34312534e+00
-5.79113603e-01 1.03820026e+00 -5.67537785e-01 3.49501163e-01
9.79596853e-01 -3.59792531e-01 -1.37896597e-01 1.02373101e-01
1.43081796e+00 -1.33441672e-01 -1.42305279e+00 -4.90595222e-01
-2.98628122e-01 -9.47040200e-01 -3.16477984e-01 -8.71596754e-01
-1.26719165e+00 8.45210671e-01 8.50400329e-01 -2.48745710e-01
1.22082996e+00 -1.29655913e-01 8.68919551e-01 -7.94155374e-02
5.14329910e-01 -1.00350296e+00 -2.26157859e-01 7.00251535e-02
6.58756793e-01 -1.66963601e+00 -7.61244521e-02 -4.85479891e-01
-6.68150067e-01 7.58353114e-01 1.08446300e+00 -3.81140634e-02
3.12582314e-01 -3.42054665e-01 2.51940116e-02 3.22790891e-01
-1.52233690e-01 -4.74240869e-01 5.10249913e-01 7.66672194e-01
1.29436359e-01 3.34281832e-01 -1.17923245e-01 8.84320796e-01
1.66272223e-01 -2.32414216e-01 -6.41413108e-02 4.81531978e-01
-1.82688415e-01 -1.35951662e+00 -4.58738357e-01 4.45676684e-01
-1.33293688e-01 2.62211058e-02 -2.81506151e-01 8.84019375e-01
7.50377774e-01 9.97277796e-01 1.75448105e-01 -6.49489462e-01
3.74726295e-01 -1.79156572e-01 6.22447610e-01 -3.09387326e-01
-5.92394173e-01 -4.00645405e-01 -1.42580420e-01 -4.37747866e-01
-4.44728494e-01 -9.71177459e-01 -9.11892235e-01 -4.97259200e-01
-2.21465871e-01 -1.08556442e-01 3.82748038e-01 8.70386660e-01
5.49403727e-01 3.62504095e-01 8.12059224e-01 -8.47396076e-01
-2.77764857e-01 -9.00622129e-01 -3.98787677e-01 8.46956730e-01
1.76882297e-01 -9.61495340e-01 -1.59549136e-02 2.09816042e-02] | [14.686485290527344, 0.9470586776733398] |
7138edd5-2026-4b09-b53f-8f4a95c0e419 | controlling-high-dimensional-data-with-sparse | 2303.09446 | null | https://arxiv.org/abs/2303.09446v1 | https://arxiv.org/pdf/2303.09446v1.pdf | Controlling High-Dimensional Data With Sparse Input | We address the problem of human-in-the-loop control for generating highly-structured data. This task is challenging because existing generative models lack an efficient interface through which users can modify the output. Users have the option to either manually explore a non-interpretable latent space, or to laboriously annotate the data with conditioning labels. To solve this, we introduce a novel framework whereby an encoder maps a sparse, human interpretable control space onto the latent space of a generative model. We apply this framework to the task of controlling prosody in text-to-speech synthesis. We propose a model, called Multiple-Instance CVAE (MICVAE), that is specifically designed to encode sparse prosodic features and output complete waveforms. We show empirically that MICVAE displays desirable qualities of a sparse human-in-the-loop control mechanism: efficiency, robustness, and faithfulness. With even a very small number of input values (~4), MICVAE enables users to improve the quality of the output significantly, in terms of listener preference (4:1). | ['Zack Hodari', 'Tian Huey Teh', 'Devang Savita Ram Mohan', 'Dan Andrei Iliescu'] | 2023-03-14 | null | null | null | null | ['text-to-speech-synthesis', 'speech-synthesis'] | ['speech', 'speech'] | [ 3.37018043e-01 4.84766066e-01 6.68499665e-03 -4.50174958e-01
-9.40494895e-01 -9.47090149e-01 4.41092610e-01 -3.62785339e-01
1.22780561e-01 5.28164446e-01 4.96674746e-01 -1.33116588e-01
2.02470079e-01 -6.22750938e-01 -5.71389675e-01 -4.42641854e-01
2.55584657e-01 5.74800909e-01 -2.47441649e-01 -2.34964743e-01
-1.14624120e-01 2.18743697e-01 -1.55807984e+00 4.38739657e-01
7.53542006e-01 6.04503691e-01 3.67631018e-01 8.39652896e-01
1.92543298e-01 6.04986370e-01 -6.50046408e-01 -2.01724425e-01
1.59636989e-01 -8.05176079e-01 -3.89847219e-01 2.26487234e-01
2.96851918e-02 -5.60214967e-02 1.73027501e-01 8.62205863e-01
5.54150164e-01 1.97349519e-01 6.44824862e-01 -1.23149121e+00
-7.03893602e-01 9.05453026e-01 8.05179551e-02 -2.48464391e-01
4.69535768e-01 3.54221135e-01 1.20501041e+00 -8.05524707e-01
5.85513771e-01 1.36213982e+00 4.07121032e-01 7.42622793e-01
-1.83478963e+00 -5.51016390e-01 -1.27464747e-02 -5.56748986e-01
-1.37995172e+00 -1.05574930e+00 7.33217418e-01 -6.07197285e-01
7.83605635e-01 5.78152001e-01 5.68499446e-01 1.09359956e+00
-8.97579119e-02 5.54248571e-01 6.35352969e-01 -4.69940513e-01
4.09603119e-01 3.01242024e-01 -4.07075584e-01 5.68282604e-01
-2.70157337e-01 1.70038015e-01 -8.07712793e-01 -2.38867491e-01
1.00932407e+00 -5.15871882e-01 -3.25978786e-01 -2.90166408e-01
-1.09903789e+00 9.10104454e-01 1.14643313e-01 1.24182731e-01
-1.94398910e-01 2.46337116e-01 4.25927751e-02 3.72129291e-01
1.45899609e-01 9.17989612e-01 -3.81005108e-01 -4.15134668e-01
-9.81187284e-01 4.08597380e-01 9.85020995e-01 1.01143003e+00
4.61356491e-01 5.39294124e-01 -3.45880061e-01 8.12089622e-01
3.32148492e-01 5.54063022e-01 4.55285549e-01 -1.39165080e+00
3.96920860e-01 1.16626836e-01 1.32781759e-01 -7.63393462e-01
4.36189473e-02 -3.14678311e-01 -7.25351930e-01 3.23010445e-01
9.37769562e-02 -4.00037020e-01 -8.41162145e-01 2.10821247e+00
9.65385810e-02 -3.00759703e-01 3.39530222e-02 8.15906227e-01
3.32475334e-01 9.56005037e-01 -2.58377530e-02 -5.15267313e-01
1.10383582e+00 -7.45349407e-01 -9.45021212e-01 -3.17819864e-01
1.88832834e-01 -7.22308993e-01 1.56378925e+00 4.81403321e-01
-1.48581350e+00 -6.48470581e-01 -9.09862280e-01 -1.48211792e-01
1.17687084e-01 2.33590156e-01 3.77823114e-01 6.09354258e-01
-1.07404542e+00 5.00065029e-01 -8.91079307e-01 8.09801072e-02
-6.51409924e-02 4.67427790e-01 -2.72571355e-01 4.01626021e-01
-9.79531527e-01 6.15177751e-01 2.46822283e-01 -8.71091187e-02
-9.09093857e-01 -7.58418679e-01 -1.20034003e+00 3.96984130e-01
1.86906099e-01 -8.20159793e-01 1.61700547e+00 -1.00051928e+00
-1.96561313e+00 4.48157698e-01 -1.85250193e-01 -6.11005053e-02
3.02657664e-01 -2.74980813e-01 -9.76158008e-02 -1.00200139e-01
8.64323899e-02 8.03709209e-01 1.15950704e+00 -1.34240890e+00
-1.61965847e-01 9.19202715e-02 -1.09114483e-01 1.99576452e-01
-1.36887103e-01 1.13684870e-03 -4.57303822e-01 -1.01333463e+00
-2.26544701e-02 -1.13896191e+00 -1.84276924e-01 -5.67963980e-02
-6.71726823e-01 9.08545032e-03 7.64054656e-01 -4.36693728e-01
1.40726781e+00 -2.33110905e+00 6.31393731e-01 2.94984043e-01
6.78440556e-02 -7.29662552e-03 -1.66438460e-01 2.52674103e-01
-7.39596337e-02 3.20334762e-01 -2.65404165e-01 -7.19362736e-01
3.10764283e-01 3.33229274e-01 -4.48155195e-01 -2.26036063e-03
3.01912814e-01 7.69089580e-01 -7.26321995e-01 -3.93906683e-01
-1.80502348e-02 5.59161782e-01 -1.16408408e+00 6.97733998e-01
-5.11220813e-01 7.09694684e-01 -3.32405448e-01 3.53446752e-01
2.55196244e-02 -1.66230321e-01 3.39928359e-01 -1.64805740e-01
-1.93657264e-01 4.25912172e-01 -1.28439462e+00 1.75727296e+00
-6.64855421e-01 6.55500293e-01 3.22179317e-01 -3.38155895e-01
8.18554223e-01 6.91292048e-01 6.67915866e-02 -2.57757843e-01
4.36757132e-02 8.76451880e-02 5.02352975e-02 -3.01805586e-01
4.83334064e-01 -3.54501486e-01 -4.36489701e-01 7.55951583e-01
-3.75710502e-02 -7.90404201e-01 1.72517985e-01 1.08710259e-01
8.57748747e-01 4.92893197e-02 3.18381697e-01 -1.70894057e-01
-3.16328462e-03 -5.52513480e-01 5.36452413e-01 6.04672313e-01
3.63463819e-01 9.74042833e-01 7.19631374e-01 4.30361368e-02
-1.22238088e+00 -1.27704394e+00 1.32840067e-01 1.30587184e+00
-5.89286745e-01 -6.11610413e-01 -9.25404191e-01 1.29986554e-02
-1.64921328e-01 1.08790517e+00 -4.14404422e-01 -5.81621155e-02
-6.53290629e-01 -1.81762964e-01 4.91232902e-01 5.95101535e-01
-4.16896194e-02 -1.36346996e+00 -7.60047495e-01 2.90486813e-01
-3.45059693e-01 -8.98885310e-01 -9.96852934e-01 3.80843937e-01
-6.37073219e-01 -3.46913636e-01 -3.72470886e-01 -6.61635637e-01
7.18841493e-01 -4.22067583e-01 1.18791306e+00 -3.03615987e-01
1.02036726e-02 2.36766905e-01 -2.16805890e-01 -1.61997765e-01
-9.82394695e-01 1.39967546e-01 1.00319877e-01 -9.14558545e-02
-2.95342326e-01 -8.93251479e-01 -2.32029587e-01 2.49519661e-01
-9.57451105e-01 5.11292398e-01 2.66857266e-01 7.32888579e-01
6.11038268e-01 -1.47771955e-01 6.34404719e-01 -8.10659587e-01
9.42745507e-01 -2.05379561e-01 -5.12794375e-01 4.54165861e-02
-3.10709894e-01 5.79037964e-01 7.59828031e-01 -7.09648192e-01
-8.76761675e-01 5.76362669e-01 -1.68442115e-01 -5.39677501e-01
4.01646271e-03 4.02705908e-01 -6.08496845e-01 3.59491467e-01
6.02422297e-01 7.23640397e-02 -9.42781288e-03 -4.29506153e-01
9.62314427e-01 7.18430936e-01 8.37774992e-01 -7.29158044e-01
7.66606688e-01 -1.49471983e-01 -3.71564448e-01 -7.02475846e-01
-6.40592933e-01 1.29883409e-01 -5.80936611e-01 5.11240438e-02
9.42669392e-01 -8.95812511e-01 -6.86150491e-01 3.66521371e-03
-1.12916505e+00 -6.94305539e-01 -7.19280958e-01 2.19134182e-01
-1.05107129e+00 -1.12339363e-01 -6.06947899e-01 -9.40694034e-01
-2.68466532e-01 -1.32644260e+00 1.21206570e+00 -1.96139750e-04
-9.87265766e-01 -7.61160493e-01 -1.34831043e-02 3.62912677e-02
5.25214791e-01 2.55723298e-01 1.21342528e+00 -2.27785349e-01
-5.15704513e-01 3.51683684e-02 4.26785499e-01 3.44862700e-01
3.01982224e-01 -2.06877990e-03 -1.08606946e+00 -2.38375917e-01
1.28478259e-01 -4.06041563e-01 4.13640797e-01 3.09589714e-01
1.22372520e+00 -6.97906494e-01 5.68486676e-02 5.88174641e-01
7.92128861e-01 2.37488329e-01 5.34405887e-01 -4.17377740e-01
6.20035291e-01 4.94922400e-01 7.91766122e-02 6.48312271e-01
2.73776412e-01 8.97784472e-01 3.71630080e-02 -1.44361574e-02
7.25080771e-03 -6.01441383e-01 4.47612017e-01 1.16486001e+00
1.31781265e-01 -5.32303989e-01 -5.34974694e-01 4.70617741e-01
-1.59401071e+00 -9.11535561e-01 4.34535384e-01 1.97602606e+00
1.45350873e+00 1.82224974e-01 -2.03022826e-03 2.94783890e-01
5.87858379e-01 7.74104372e-02 -4.03339922e-01 -6.98325694e-01
2.32031301e-01 3.99256676e-01 -2.18114361e-01 8.89050007e-01
-8.41517448e-01 8.69303584e-01 6.81814146e+00 5.94836891e-01
-1.10847497e+00 -1.28523260e-01 5.41764021e-01 -3.60871702e-01
-8.76512527e-01 -3.52202058e-02 -6.04312718e-01 4.94288176e-01
1.06996751e+00 -1.32535463e-02 8.79548490e-01 6.42717779e-01
6.01707339e-01 3.23292583e-01 -1.60594082e+00 1.00964653e+00
3.37082483e-02 -1.27965319e+00 2.28199035e-01 -4.80489545e-02
6.25501990e-01 -6.49016380e-01 3.90253037e-01 1.85013801e-01
3.14613193e-01 -1.33204412e+00 1.18351126e+00 4.98943686e-01
1.34655058e+00 -5.85318267e-01 -1.32047281e-01 3.84159923e-01
-1.04987991e+00 -4.90217358e-02 4.65121977e-02 -8.08650702e-02
5.67653179e-01 2.57308841e-01 -7.32802033e-01 -1.80676371e-01
2.01380759e-01 2.46352375e-01 -1.76731884e-01 4.40548301e-01
-3.85135114e-01 7.31450200e-01 -3.50874513e-01 1.62023485e-01
-1.69842646e-01 5.03001697e-02 6.12095535e-01 1.16673112e+00
6.66281581e-01 4.18490410e-01 2.12145731e-01 1.39187956e+00
-8.80360678e-02 -2.11701736e-01 -5.76652706e-01 -4.14051592e-01
5.44276714e-01 9.14646804e-01 -3.53857100e-01 -1.10169873e-01
1.51487112e-01 9.93624926e-01 5.48273288e-02 3.81858498e-01
-5.82021296e-01 -2.69114763e-01 6.76959157e-01 2.77015716e-01
4.24060583e-01 -3.94203424e-01 -4.81544703e-01 -1.12086046e+00
-2.06169233e-01 -1.31462252e+00 -4.69261445e-02 -1.08846319e+00
-1.03437173e+00 7.75493443e-01 2.58739721e-02 -8.82118344e-01
-9.47012663e-01 -1.68095514e-01 -4.95996237e-01 9.55711126e-01
-6.04413390e-01 -1.00799382e+00 -8.05576071e-02 3.98746461e-01
7.54385650e-01 -2.07268581e-01 1.19197798e+00 1.74906269e-01
-3.25225800e-01 6.22685611e-01 -3.57878178e-01 -1.81759104e-01
4.64040071e-01 -1.34114766e+00 6.29974782e-01 6.76878512e-01
3.73013496e-01 8.45893621e-01 1.05045950e+00 -4.73381609e-01
-1.31921279e+00 -9.32814717e-01 1.03162122e+00 -6.66784942e-01
3.61778826e-01 -7.35230029e-01 -6.89412951e-01 6.33552849e-01
3.22352618e-01 -3.03675562e-01 9.17609274e-01 -2.18869410e-02
-2.35421821e-01 6.72698691e-02 -9.18634832e-01 7.56985962e-01
8.22482824e-01 -8.44306707e-01 -6.04655921e-01 -5.08488603e-02
1.08908319e+00 -5.74123919e-01 -6.80938780e-01 2.58084118e-01
5.20609677e-01 -6.65882051e-01 7.85733700e-01 -4.67787623e-01
4.05096024e-01 -5.21798432e-01 -3.55966210e-01 -1.49856305e+00
-4.21478212e-01 -1.26306415e+00 -1.35077462e-01 1.25513649e+00
6.47173882e-01 -7.99351782e-02 4.60264623e-01 8.65274012e-01
-2.93827385e-01 -3.77020657e-01 -6.55515671e-01 -2.76570022e-01
-8.55102465e-02 -5.81466556e-01 6.87677860e-01 6.27972782e-01
1.80125073e-01 6.31376445e-01 -5.81471741e-01 1.17378063e-01
2.40902647e-01 1.15548141e-01 6.93633735e-01 -9.94950593e-01
-8.34017992e-01 -2.72988498e-01 8.31404179e-02 -1.09049618e+00
9.67767909e-02 -7.27570474e-01 5.49940467e-01 -1.07504845e+00
-6.44218102e-02 -4.75676358e-01 7.23083466e-02 6.84747756e-01
-1.44074503e-02 1.90782666e-01 6.51975036e-01 1.03220053e-01
-2.30658278e-01 6.17716908e-01 1.26285839e+00 1.26895845e-01
-6.27840519e-01 5.97721972e-02 -8.20554137e-01 5.31936765e-01
6.35449171e-01 -4.79857057e-01 -8.03751886e-01 -5.20262063e-01
3.09268624e-01 5.64113617e-01 3.06781322e-01 -8.78096581e-01
9.73564610e-02 -2.54486561e-01 2.37815231e-01 -2.50303686e-01
6.39298558e-01 -5.01374245e-01 6.34911776e-01 8.13140199e-02
-8.04547787e-01 1.31846040e-01 1.59207687e-01 2.62522370e-01
-2.83918917e-01 -1.25338316e-01 7.13808239e-01 -6.69424236e-02
-4.78425324e-02 -2.34017409e-02 -5.71898341e-01 1.55335546e-01
5.58556139e-01 9.49950144e-02 -2.12430512e-03 -8.63280535e-01
-1.02928245e+00 -9.46689397e-02 3.42593223e-01 3.61401170e-01
4.85013843e-01 -1.45720851e+00 -5.03916144e-01 7.14729965e-01
-7.02463761e-02 -1.18045174e-02 -1.70439884e-01 1.59004137e-01
-3.01087350e-01 2.79486090e-01 -3.14289294e-02 -4.68983352e-01
-1.01541388e+00 2.76777238e-01 2.48817161e-01 7.78443217e-02
-5.16050994e-01 7.98994005e-01 4.02186960e-01 -3.59669745e-01
2.62120098e-01 -6.18902385e-01 1.13094263e-01 7.59459138e-02
2.94395894e-01 -4.64437529e-02 -2.78315276e-01 -4.17126864e-01
-5.88301830e-02 2.56620914e-01 4.24577743e-01 -7.25906432e-01
1.21358681e+00 1.27738230e-02 -5.85201336e-03 6.71539843e-01
9.80949461e-01 2.71050185e-01 -1.55575871e+00 2.67700940e-01
-4.11720216e-01 -3.44257385e-01 -1.74865589e-01 -8.20070267e-01
-7.34377861e-01 8.90639782e-01 2.22915187e-01 3.42859387e-01
1.04900396e+00 -2.85501219e-02 7.22705245e-01 2.42327139e-01
6.89019337e-02 -9.07429874e-01 3.66516471e-01 5.51483810e-01
1.24488962e+00 -6.87063217e-01 -4.32001293e-01 -3.16594541e-01
-1.05762756e+00 8.60959530e-01 3.41682285e-01 1.21623099e-01
5.58630526e-01 7.17397213e-01 1.73428103e-01 8.83994326e-02
-1.13724589e+00 7.23637491e-02 3.47648382e-01 5.97948551e-01
6.44168139e-01 2.14977056e-01 2.49001592e-01 7.09741414e-01
-8.37109983e-01 -6.96770921e-02 4.02659744e-01 5.68466663e-01
-3.70236099e-01 -1.26912439e+00 -2.55635381e-01 1.17766179e-01
-2.69245416e-01 -1.70408517e-01 -2.32957050e-01 3.72714847e-01
9.28683504e-02 1.06059325e+00 2.87930310e-01 -2.78721184e-01
3.20406199e-01 2.48452947e-01 3.31400663e-01 -1.00326109e+00
-6.40322089e-01 5.24578810e-01 1.98418379e-01 -4.28660870e-01
-8.77177194e-02 -5.76561213e-01 -1.14199877e+00 5.75690866e-02
-1.58670619e-01 3.22835237e-01 4.30100322e-01 7.42260873e-01
3.76316696e-01 5.80592752e-01 6.51832044e-01 -9.34373796e-01
-5.53280890e-01 -9.38075364e-01 -4.39130455e-01 2.59011418e-01
4.76920933e-01 -3.57226670e-01 -3.78051907e-01 6.53353095e-01] | [15.417756080627441, 6.320740222930908] |
f0457019-b943-4f96-a788-e1aa4bb66a30 | exploring-graph-structured-passage | 1809.02040 | null | http://arxiv.org/abs/1809.02040v1 | http://arxiv.org/pdf/1809.02040v1.pdf | Exploring Graph-structured Passage Representation for Multi-hop Reading Comprehension with Graph Neural Networks | Multi-hop reading comprehension focuses on one type of factoid question,
where a system needs to properly integrate multiple pieces of evidence to
correctly answer a question. Previous work approximates global evidence with
local coreference information, encoding coreference chains with DAG-styled GRU
layers within a gated-attention reader. However, coreference is limited in
providing information for rich inference. We introduce a new method for better
connecting global evidence, which forms more complex graphs compared to DAGs.
To perform evidence integration on our graphs, we investigate two recent graph
neural networks, namely graph convolutional network (GCN) and graph recurrent
network (GRN). Experiments on two standard datasets show that richer global
information leads to better answers. Our method performs better than all
published results on these datasets. | ['Yue Zhang', 'Mo Yu', 'Zhiguo Wang', 'Radu Florian', 'Linfeng Song', 'Daniel Gildea'] | 2018-09-06 | null | null | null | null | ['multi-hop-reading-comprehension'] | ['natural-language-processing'] | [ 8.62025172e-02 8.42077255e-01 -5.18870711e-01 -2.95433521e-01
-6.59844637e-01 -5.80408096e-01 6.19638741e-01 8.68876874e-01
-1.51767045e-01 6.77547991e-01 9.22248185e-01 -8.08171928e-01
-3.45035732e-01 -1.28566742e+00 -8.73280883e-01 4.83633205e-02
-6.72211647e-02 8.28284383e-01 7.49891222e-01 -6.33891165e-01
1.77876249e-01 1.35455981e-01 -9.80962098e-01 4.90659475e-01
9.74748731e-01 6.56874657e-01 -6.40273765e-02 1.02130604e+00
-5.75889647e-01 1.64531553e+00 -7.66341448e-01 -1.12831509e+00
-4.72707361e-01 -7.78600335e-01 -1.57496011e+00 -8.17306042e-01
1.04572499e+00 -1.98301479e-01 -6.20726764e-01 1.18749619e+00
4.44611847e-01 4.00189072e-01 4.26654398e-01 -9.40684617e-01
-1.00416017e+00 1.45068538e+00 -1.28832221e-01 6.46646321e-01
1.00979733e+00 -2.55844712e-01 1.76843810e+00 -1.69292256e-01
8.31033230e-01 1.61570978e+00 6.60722554e-01 2.89305449e-01
-8.78700614e-01 -3.01611394e-01 3.93686086e-01 8.96487892e-01
-6.51150823e-01 8.26779902e-02 1.08828497e+00 2.88196146e-01
1.48379445e+00 4.04967248e-01 6.35079026e-01 1.43101382e+00
4.96697307e-01 5.70378244e-01 6.46284282e-01 -4.62965608e-01
-6.40829355e-02 -7.83244848e-01 9.69782352e-01 1.07985270e+00
5.81127942e-01 -1.32912114e-01 -5.36292553e-01 -8.19060728e-02
5.07783949e-01 -3.77030790e-01 -6.81405544e-01 8.12942013e-02
-1.05919707e+00 9.98668134e-01 1.19958770e+00 5.26602566e-01
-3.33132803e-01 3.51022989e-01 3.82220864e-01 6.63429201e-01
1.43116072e-01 6.16160035e-01 -2.04147071e-01 3.03071439e-01
-2.95132548e-01 1.37092322e-01 1.35915697e+00 7.62004137e-01
5.41325867e-01 -2.18190014e-01 -6.35632336e-01 6.76936507e-01
3.18995446e-01 1.81950748e-01 1.26450002e-01 -1.15683019e+00
8.23144436e-01 9.37011302e-01 -5.34547389e-01 -1.55806100e+00
-8.37932587e-01 -8.60107005e-01 -8.79018784e-01 -3.67599040e-01
4.85792756e-01 1.62824735e-01 -6.30743325e-01 1.91443801e+00
4.33582664e-02 1.48538187e-01 5.53876795e-02 1.03797042e+00
1.75028479e+00 4.02966678e-01 8.38167742e-02 1.19733967e-01
1.60512662e+00 -1.33796430e+00 -1.14661086e+00 -4.46247548e-01
6.88496530e-01 -1.79546848e-01 8.34255159e-01 2.31705025e-01
-1.18340826e+00 -3.96834403e-01 -1.14594269e+00 -6.57820642e-01
-5.37907958e-01 -8.28128278e-01 6.33082628e-01 5.77241145e-02
-1.44976294e+00 4.64835286e-01 -4.89207283e-02 -3.99325520e-01
1.02387749e-01 1.76106304e-01 -4.16787714e-01 -3.08448583e-01
-1.97510946e+00 1.29609656e+00 7.15552568e-01 1.04885079e-01
-5.69331467e-01 -3.28448802e-01 -1.26079607e+00 5.85795999e-01
8.83428097e-01 -1.24888051e+00 1.41856432e+00 -3.30544889e-01
-1.23365605e+00 7.77558327e-01 -4.94592078e-02 -4.91674095e-01
-5.38546816e-02 -1.11643612e-01 -5.31937659e-01 3.46337855e-01
-4.32329960e-02 2.55337864e-01 3.08461279e-01 -1.00255632e+00
-8.15753937e-02 -2.26594016e-01 4.73391384e-01 1.71638384e-01
2.56494313e-01 -4.39120829e-02 -4.07923460e-01 -3.72970223e-01
1.53583944e-01 -2.37574801e-01 5.84709011e-02 -7.61792839e-01
-6.10252857e-01 -9.43707585e-01 5.38793206e-01 -6.76006854e-01
1.52865815e+00 -1.28747106e+00 4.65123355e-01 2.36172080e-01
1.01919925e+00 2.16274366e-01 -5.23118794e-01 6.83052719e-01
-8.11370537e-02 1.25407219e-01 -1.11038208e-01 -5.22649735e-02
1.85150713e-01 6.41981125e-01 -2.05734402e-01 -5.63446507e-02
1.74024716e-01 1.60500443e+00 -1.45909417e+00 -6.58922315e-01
-3.31099629e-01 4.95821387e-02 -5.96853733e-01 3.51121604e-01
-7.28380799e-01 -1.67413950e-02 -3.56542766e-01 5.09145081e-01
5.21759748e-01 -7.83436894e-01 6.44759059e-01 -2.17175990e-01
5.24911284e-01 9.95763958e-01 -7.76535809e-01 1.83442855e+00
-8.06415603e-02 4.71550971e-01 -8.00363999e-03 -1.01835299e+00
8.21191490e-01 2.38382980e-01 -3.91184866e-01 -1.09931755e+00
3.82273048e-01 5.66247962e-02 3.07000369e-01 -4.97515917e-01
5.59506238e-01 2.89587468e-01 -6.54211119e-02 5.14373124e-01
6.15736365e-01 5.00088148e-02 3.98318976e-01 1.05488408e+00
1.48085785e+00 -1.56861335e-01 4.99084383e-01 -2.11282358e-01
4.22266781e-01 -2.87821561e-01 1.95293427e-01 1.14689255e+00
-7.14908391e-02 4.54830229e-01 1.10041428e+00 -1.91364080e-01
-3.00215185e-01 -1.07124317e+00 4.41996664e-01 1.27784836e+00
3.74826789e-02 -7.86315501e-01 -5.20263553e-01 -1.14983487e+00
-7.89691135e-02 9.53140914e-01 -7.23221302e-01 -3.14408809e-01
-1.01765680e+00 -1.80998862e-01 6.72067821e-01 7.64206886e-01
5.63395679e-01 -1.39799762e+00 -2.33995110e-01 3.22934568e-01
-6.06334209e-01 -1.05670297e+00 -5.50933421e-01 5.94440885e-02
-6.98019981e-01 -1.61648965e+00 -9.74639207e-02 -8.59217763e-01
8.81271735e-02 2.27819934e-01 2.16049337e+00 8.45082283e-01
2.87425250e-01 5.11791050e-01 -5.16360402e-01 -1.39073133e-01
-3.89858752e-01 3.18208724e-01 -6.79785252e-01 -5.48870623e-01
4.26490992e-01 -6.24893665e-01 -5.11178315e-01 -1.44858599e-01
-6.36854827e-01 -1.59616172e-01 3.63689244e-01 7.94367969e-01
9.75240991e-02 -6.07619226e-01 5.41207194e-01 -1.17413950e+00
1.43231392e+00 -7.82375574e-01 -2.57446527e-01 6.33786738e-01
-5.84221125e-01 3.68968070e-01 5.41898489e-01 7.22006932e-02
-1.12058294e+00 -1.23058748e+00 -2.87702829e-01 -1.23655632e-01
7.79781193e-02 1.06869781e+00 -2.23286431e-02 6.84986413e-02
7.96426475e-01 -3.98336679e-01 -3.03543776e-01 -2.48609453e-01
8.92753541e-01 1.35669196e-02 1.04272175e+00 -8.97021830e-01
1.20769121e-01 -2.58231193e-01 1.80201396e-01 -1.81900725e-01
-1.22455049e+00 -4.39688027e-01 -4.33778882e-01 -2.21824139e-01
9.51166630e-01 -2.39173263e-01 -1.17918909e+00 -4.39892709e-02
-1.64868438e+00 -3.56471896e-01 -7.94265494e-02 2.58052111e-01
-2.56236613e-01 6.38669074e-01 -1.02377415e+00 -4.24268007e-01
-5.11579931e-01 -6.87302768e-01 7.75110304e-01 2.16040105e-01
-4.35781389e-01 -1.55515504e+00 5.66272855e-01 4.91578400e-01
4.34805751e-01 3.67955536e-01 1.15321219e+00 -1.03059936e+00
-6.36143088e-01 1.42878920e-01 -5.44478953e-01 -2.83454180e-01
-2.46190354e-01 -2.37351269e-01 -7.10093200e-01 3.69215496e-02
-2.28904828e-01 -5.92630684e-01 1.28064084e+00 1.58327624e-01
8.85475278e-01 -5.74071884e-01 -3.67502987e-01 3.39969069e-01
1.16813803e+00 -2.23419189e-01 6.28629208e-01 1.42523587e-01
1.00753272e+00 6.06960237e-01 -1.04920007e-01 -3.46093923e-01
1.13739502e+00 3.21272105e-01 7.70675898e-01 8.70856419e-02
-7.94192553e-01 -5.40715396e-01 1.30887881e-01 1.30537677e+00
-3.48721296e-01 -9.11361814e-01 -1.10075974e+00 5.75421989e-01
-2.20329523e+00 -1.07709289e+00 -8.61668169e-01 1.39058983e+00
8.42910290e-01 9.42913070e-02 -1.58090696e-01 -1.64703429e-01
6.30621135e-01 5.21243393e-01 -2.62805194e-01 -7.32828200e-01
-5.24354339e-01 5.42240918e-01 8.30359757e-03 1.03050494e+00
-6.33152843e-01 8.75376523e-01 6.59993219e+00 3.40146899e-01
-4.01937664e-01 9.01620090e-02 1.13890469e-01 4.77951288e-01
-8.28015447e-01 1.05348937e-01 -3.16510886e-01 -9.93831754e-02
1.23617804e+00 1.26208171e-01 2.78210878e-01 2.25791544e-01
-5.72913408e-01 -1.93855108e-03 -1.06925762e+00 5.01625240e-01
3.31362367e-01 -1.75733256e+00 4.69126135e-01 -4.96767282e-01
3.40390056e-01 2.62798429e-01 -4.85618293e-01 6.68965697e-01
1.16019654e+00 -1.15990841e+00 3.29063416e-01 8.34764242e-01
3.19404840e-01 -4.80584025e-01 1.12704098e+00 2.21928000e-01
-1.30551231e+00 9.83088315e-02 -3.37035269e-01 -2.70802647e-01
2.02635169e-01 2.67512530e-01 -6.64670587e-01 1.33061612e+00
6.63934112e-01 6.22030258e-01 -8.06836724e-01 7.98944831e-01
-1.13663399e+00 8.06886017e-01 -1.60313547e-01 -3.40810925e-01
3.81926775e-01 -6.84046894e-02 6.91841125e-01 1.40071547e+00
-1.51812509e-01 3.45700115e-01 8.22603106e-02 9.40615296e-01
-7.27107346e-01 -6.07694760e-02 -7.27921188e-01 2.06208885e-01
3.89886260e-01 1.14050579e+00 -3.27333838e-01 -4.37624097e-01
-5.07382035e-01 6.58166647e-01 1.35314107e+00 4.71559644e-01
-5.74107945e-01 -4.07434881e-01 -1.31767333e-01 -2.63315588e-01
-9.59891751e-02 -1.46140903e-01 -6.35244846e-02 -1.26338983e+00
-2.24472195e-01 -9.15076792e-01 1.59188199e+00 -1.05419159e+00
-1.56540596e+00 5.75713515e-01 4.01567435e-03 -1.87785029e-01
-3.94598842e-01 -6.50854051e-01 -1.13558078e+00 7.33677506e-01
-1.76035225e+00 -1.17508078e+00 -4.49838430e-01 8.08633804e-01
-1.96994301e-02 1.88782513e-01 9.35385764e-01 -4.35586534e-02
-2.88692445e-01 6.84634387e-01 -7.86721826e-01 4.81781155e-01
6.04939580e-01 -1.69364083e+00 4.18730140e-01 8.34894836e-01
4.20830995e-01 8.66684854e-01 4.95369196e-01 -7.76863754e-01
-1.45101917e+00 -7.40823805e-01 1.29264545e+00 -7.46681750e-01
9.59575295e-01 -8.85096639e-02 -1.52276695e+00 1.13912654e+00
1.16840053e+00 -2.01174483e-01 4.54970539e-01 7.94276714e-01
-9.18237567e-01 2.27464765e-01 -7.82601714e-01 5.46424985e-01
1.48850620e+00 -6.26423836e-01 -1.48137784e+00 1.00167975e-01
1.29900634e+00 -5.52295148e-01 -1.04985499e+00 4.41354543e-01
-1.62896477e-02 -9.61393714e-01 7.60476589e-01 -1.04270065e+00
3.93193394e-01 -6.39523193e-02 -2.45233215e-02 -1.53501749e+00
-8.10660779e-01 -7.34027863e-01 -8.33059788e-01 9.23635304e-01
5.28019011e-01 -9.11800385e-01 2.78567582e-01 1.16386466e-01
-5.60064793e-01 -4.46349502e-01 -9.16993618e-01 -3.29231799e-01
1.68788224e-01 -2.84352481e-01 6.79228306e-01 1.22711873e+00
6.31343484e-01 1.13233447e+00 1.25806585e-01 7.00800866e-02
5.55206776e-01 3.74633193e-01 4.41238314e-01 -1.48672271e+00
-1.98346138e-01 -8.09742332e-01 7.84859434e-03 -9.63053524e-01
6.38917267e-01 -1.56290901e+00 -2.76029587e-01 -2.31181145e+00
3.47420305e-01 3.14405650e-01 -5.83594918e-01 5.79752803e-01
-6.98792458e-01 -2.23469630e-01 3.57411616e-02 -3.38449568e-01
-1.11236596e+00 3.98748934e-01 1.64461362e+00 -5.49399436e-01
3.54492188e-01 -5.56222558e-01 -8.18476498e-01 4.38777894e-01
6.82051122e-01 -2.69299060e-01 -5.30582666e-01 -6.67970836e-01
9.46753740e-01 3.64573568e-01 4.59154844e-01 -5.10034919e-01
8.24840665e-01 2.08126262e-01 5.28294891e-02 -5.72376609e-01
-8.70020166e-02 -3.06120366e-01 -3.43860656e-01 3.67606908e-01
-4.97602284e-01 3.74968648e-01 1.42053545e-01 7.51352668e-01
-5.20029664e-01 -5.27611151e-02 1.73770621e-01 -3.23778629e-01
-6.94175482e-01 1.77668512e-01 -4.64124568e-02 5.16348362e-01
1.21104538e-01 3.00679624e-01 -1.29741693e+00 -8.85623813e-01
-6.83430135e-01 6.28410995e-01 -2.07725763e-01 4.99424338e-01
8.55114698e-01 -1.20827842e+00 -1.02717960e+00 -2.66951919e-01
1.12583891e-01 3.24460566e-02 1.19403936e-01 8.74817967e-01
-4.64968830e-01 7.24814951e-01 3.97053510e-02 -3.72103602e-01
-1.06308198e+00 7.51067460e-01 4.28377688e-01 -8.23746920e-01
-7.09042907e-01 9.11846936e-01 -1.62988216e-01 -7.26693630e-01
1.21747874e-01 -7.28402495e-01 -8.41591239e-01 1.81553245e-01
3.19408774e-01 3.70176613e-01 2.23838612e-01 -9.56887379e-02
-2.74167389e-01 4.72662508e-01 -2.30337959e-02 3.01646322e-01
1.02184689e+00 1.16018988e-01 -7.58785486e-01 1.66948333e-01
8.95564795e-01 8.45166519e-02 -5.50747871e-01 -4.30316210e-01
3.73628765e-01 1.24346845e-01 -4.64258827e-02 -9.43656385e-01
-7.96195626e-01 7.71181464e-01 -4.16130930e-01 7.30066538e-01
7.30276406e-01 5.05379558e-01 7.49174297e-01 7.26668239e-01
2.51191497e-01 -5.97822666e-01 2.94723094e-01 1.12131381e+00
1.34619164e+00 -1.05482101e+00 -2.56414771e-01 -3.93115461e-01
-3.06551963e-01 1.34136093e+00 8.51196945e-01 -6.67327121e-02
3.29425603e-01 3.31297494e-03 -1.25827327e-01 -1.04270256e+00
-1.07506251e+00 -4.32995439e-01 8.05110931e-01 4.67758656e-01
4.89995331e-01 1.38085261e-01 -2.73401201e-01 7.71598339e-01
-3.86223972e-01 -4.64628190e-01 5.26650131e-01 5.05381107e-01
-5.51364958e-01 -8.35900784e-01 -1.60615772e-01 3.51784438e-01
-3.59610409e-01 -4.12130713e-01 -9.58906412e-01 8.78526449e-01
-4.49212492e-01 1.32260692e+00 -5.58581427e-02 -1.54727623e-01
5.25840521e-01 1.93289757e-01 7.75303423e-01 -4.48869020e-01
-9.14641380e-01 -5.43870211e-01 8.98184597e-01 -1.00484860e+00
-4.16654468e-01 -1.54506527e-02 -1.42232907e+00 -4.99266893e-01
-2.24482298e-01 4.64277029e-01 -8.96506459e-02 9.63262081e-01
3.03224504e-01 1.16010499e+00 -1.84124649e-01 -3.55323181e-02
-9.44632962e-02 -1.16652155e+00 -1.74637198e-01 5.74342549e-01
4.36137587e-01 -5.23982227e-01 -3.30590993e-01 -7.00191379e-01] | [10.778260231018066, 7.933235168457031] |
9549f963-f1b0-4377-ac3f-09a1a5ab8bfc | papooling-graph-based-position-adaptive | 2111.14067 | null | https://arxiv.org/abs/2111.14067v1 | https://arxiv.org/pdf/2111.14067v1.pdf | PAPooling: Graph-based Position Adaptive Aggregation of Local Geometry in Point Clouds | Fine-grained geometry, captured by aggregation of point features in local regions, is crucial for object recognition and scene understanding in point clouds. Nevertheless, existing preeminent point cloud backbones usually incorporate max/average pooling for local feature aggregation, which largely ignores points' positional distribution, leading to inadequate assembling of fine-grained structures. To mitigate this bottleneck, we present an efficient alternative to max pooling, Position Adaptive Pooling (PAPooling), that explicitly models spatial relations among local points using a novel graph representation, and aggregates features in a position adaptive manner, enabling position-sensitive representation of aggregated features. Specifically, PAPooling consists of two key steps, Graph Construction and Feature Aggregation, respectively in charge of constructing a graph with edges linking the center point with every neighboring point in a local region to map their relative positional information to channel-wise attentive weights, and adaptively aggregating local point features based on the generated weights through Graph Convolution Network (GCN). PAPooling is simple yet effective, and flexible enough to be ready to use for different popular backbones like PointNet++ and DGCNN, as a plug-andplay operator. Extensive experiments on various tasks ranging from 3D shape classification, part segmentation to scene segmentation well demonstrate that PAPooling can significantly improve predictive accuracy, while with minimal extra computational overhead. Code will be released. | ['Tingfa Xu', 'Ying Wang', 'Lihe Ding', 'Jianan Li', 'Jie Wang'] | 2021-11-28 | null | null | null | null | ['3d-shape-retrieval', 'scene-segmentation'] | ['computer-vision', 'computer-vision'] | [-6.73904270e-02 -1.40996128e-01 4.08787616e-02 -4.74789590e-01
-4.42192882e-01 -5.88831186e-01 4.42985386e-01 6.22773767e-01
-1.43079087e-01 3.23889524e-01 -3.67978871e-01 -2.30895609e-01
-3.13875139e-01 -1.39916265e+00 -8.09011102e-01 -6.15341783e-01
-4.47246373e-01 5.01952648e-01 7.72801638e-01 -4.06892300e-02
4.57347840e-01 1.57568288e+00 -1.59605277e+00 2.02624485e-01
8.64717782e-01 1.36624920e+00 3.52121264e-01 5.08178174e-01
-6.98868930e-01 1.25882611e-01 -4.43105459e-01 -1.07995473e-01
4.53864306e-01 4.90534455e-01 -7.12975204e-01 1.09299898e-01
5.88558495e-01 -8.27400312e-02 -2.02481732e-01 9.42693651e-01
4.44080055e-01 2.17667386e-01 4.35288131e-01 -1.30415344e+00
-5.96423984e-01 2.21625000e-01 -6.32526577e-01 1.52533963e-01
1.52107567e-01 2.69781411e-01 1.09509981e+00 -1.12236130e+00
4.09143448e-01 1.24327886e+00 8.43803108e-01 8.87584984e-02
-1.09670496e+00 -5.70473015e-01 4.10511076e-01 -1.72663078e-01
-1.69825113e+00 1.04863241e-01 9.87178445e-01 -2.74261296e-01
1.30196023e+00 4.84356612e-01 8.64056528e-01 4.44450639e-02
1.18270926e-01 5.11461616e-01 4.45083767e-01 1.03603639e-01
1.75300255e-01 -4.84143704e-01 1.40322030e-01 8.12268198e-01
1.62852660e-01 -3.81852627e-01 -3.99734735e-01 -4.54920918e-01
1.24065924e+00 4.14784193e-01 -2.42922176e-02 -6.59282506e-01
-1.33150947e+00 6.52917266e-01 1.23239887e+00 1.10040076e-01
-6.68183684e-01 2.56612301e-01 1.05405264e-02 1.42132059e-01
5.24591386e-01 4.08490568e-01 -5.10074079e-01 1.93670481e-01
-7.97043920e-01 5.41146576e-01 4.24504310e-01 1.24787986e+00
1.51802945e+00 -3.48455787e-01 -2.12352604e-01 8.49071026e-01
2.67769814e-01 3.76124084e-01 8.85048881e-02 -8.58474076e-01
6.12688959e-01 1.40009463e+00 -1.86815247e-01 -1.30520797e+00
-6.56303644e-01 -2.03181565e-01 -7.72897899e-01 4.54356283e-01
4.03881483e-02 8.36830139e-02 -1.14226699e+00 1.16657639e+00
5.68160951e-01 4.33113068e-01 -4.28825557e-01 8.64611983e-01
1.16070163e+00 6.99144483e-01 6.32945746e-02 4.09177721e-01
1.27205670e+00 -6.12435043e-01 8.56962204e-02 -1.24106489e-01
4.79070127e-01 -6.75590813e-01 8.87052000e-01 -8.84540379e-02
-1.20291233e+00 -5.47158062e-01 -9.48192298e-01 -2.95057535e-01
-7.21941948e-01 -1.89676404e-01 1.00083888e+00 2.19262421e-01
-1.27697730e+00 8.46680343e-01 -8.08519959e-01 -1.07245490e-01
8.29388142e-01 9.21425521e-01 -4.12290394e-01 6.65959567e-02
-5.60083210e-01 2.70399928e-01 4.13286120e-01 2.61740655e-01
-2.87683550e-02 -1.04040504e+00 -9.85030830e-01 2.47649595e-01
1.90331623e-01 -8.55300248e-01 9.32548821e-01 -2.68579781e-01
-1.11146677e+00 7.26970851e-01 -1.15321226e-01 -2.96142191e-01
9.18687358e-02 5.27237505e-02 1.86934173e-02 9.64805856e-02
2.44412661e-01 1.06095827e+00 7.70700634e-01 -1.23686850e+00
-7.66277492e-01 -6.55430317e-01 7.29670227e-02 3.79515707e-01
8.96099210e-02 -1.87794536e-01 -8.28568697e-01 -3.95682871e-01
7.56653905e-01 -5.83946764e-01 -6.69292271e-01 1.87305361e-01
-3.48970085e-01 -6.71223819e-01 1.23223364e+00 -5.16992249e-02
1.02299035e+00 -2.26964855e+00 -2.52659023e-01 7.62169957e-01
6.42569780e-01 2.17233852e-01 -6.25710189e-02 3.81216735e-01
6.90210983e-02 1.57966942e-01 -4.20183122e-01 -2.84814656e-01
-3.69372368e-02 3.00955266e-01 -7.44379684e-02 2.54399002e-01
6.10764027e-01 1.27703393e+00 -9.26889300e-01 -6.11048341e-01
7.69347608e-01 4.86396700e-01 -6.91830635e-01 -7.23554641e-02
-2.37591371e-01 -5.69733903e-02 -9.82365012e-01 9.96391475e-01
1.19652534e+00 -3.77662182e-01 -4.92970347e-01 -2.97143608e-01
-3.22236151e-01 1.05334319e-01 -1.34661651e+00 1.66083252e+00
-2.10266739e-01 1.19624220e-01 3.77989002e-02 -5.62290013e-01
1.25216091e+00 -1.96019351e-01 7.86951840e-01 -3.45560372e-01
4.86778766e-02 1.22046053e-01 -4.09985393e-01 1.12965098e-02
7.84584105e-01 2.52300799e-01 -1.10873289e-01 9.78890155e-03
5.62000787e-03 -5.93102217e-01 -9.08719972e-02 1.71330735e-01
1.26152444e+00 -9.53916013e-02 8.57456848e-02 -1.45674020e-01
3.91572535e-01 9.83001739e-02 2.76744634e-01 7.38077819e-01
3.92223708e-03 9.73205447e-01 2.56561100e-01 -5.95813751e-01
-6.82260394e-01 -1.12263548e+00 -1.67400539e-01 8.38463902e-01
4.82798636e-01 -5.59289753e-01 -4.61870760e-01 -5.88309765e-01
5.63612223e-01 1.19966500e-01 -3.21098119e-01 1.93243220e-01
-7.19385207e-01 -6.38320327e-01 1.24659136e-01 6.70769274e-01
5.95535994e-01 -1.28204608e+00 -5.37297964e-01 3.26318651e-01
3.95785660e-01 -9.57816839e-01 -4.18352664e-01 3.16495717e-01
-1.16880667e+00 -8.37780178e-01 -4.42094058e-01 -7.33051538e-01
9.27927673e-01 7.02331543e-01 1.15982199e+00 3.83519709e-01
-2.52520055e-01 1.67765021e-01 -2.46742517e-01 -2.92719752e-01
5.20431221e-01 2.68064201e-01 -4.06641930e-01 -3.42251435e-02
4.02009904e-01 -8.36882353e-01 -7.11030006e-01 2.53760099e-01
-8.15726340e-01 1.41835975e-04 6.94893122e-01 3.62597883e-01
1.04350817e+00 -4.54393849e-02 3.12542506e-02 -9.06283200e-01
4.76773709e-01 -3.56244773e-01 -6.90745652e-01 -8.85563344e-02
-6.93490449e-03 -1.64217249e-01 3.85529667e-01 -2.73366757e-02
-5.35379589e-01 2.96300113e-01 -3.62465531e-01 -7.42707968e-01
-3.02234650e-01 4.88013864e-01 -3.47617626e-01 -7.75915205e-01
3.33365381e-01 6.22638799e-02 -1.92993581e-01 -4.53068793e-01
4.42795724e-01 4.24839020e-01 4.43354905e-01 -7.40598917e-01
8.95459414e-01 5.45062125e-01 2.83813536e-01 -9.15973425e-01
-3.06193680e-01 -6.89125419e-01 -9.49734569e-01 -1.21777415e-01
7.90022731e-01 -7.33804345e-01 -9.21038210e-01 4.54047650e-01
-1.18063807e+00 -2.24139437e-01 -5.50344288e-01 1.27205014e-01
-4.33322579e-01 1.25648439e-01 -3.57079387e-01 -4.21579719e-01
-2.49838531e-01 -1.00023019e+00 1.51421762e+00 4.17619288e-01
3.16359736e-02 -7.44164348e-01 -2.80425280e-01 -7.27687180e-02
2.46174827e-01 5.00031054e-01 9.44047153e-01 -3.25130880e-01
-1.19680297e+00 -4.96580541e-01 -7.35870779e-01 5.25254682e-02
1.32078663e-01 2.65709609e-01 -8.20554614e-01 -6.24329150e-02
-4.92081404e-01 2.28632569e-01 5.78545034e-01 5.25304973e-01
1.78910983e+00 4.00323011e-02 -5.06425261e-01 8.79081607e-01
1.45203817e+00 -3.65995131e-02 5.98687828e-01 1.55965537e-01
1.18979216e+00 2.32850701e-01 2.87603617e-01 4.24465209e-01
5.03668547e-01 5.52920878e-01 8.30265999e-01 -3.17051262e-01
-1.14230841e-01 -1.32363945e-01 -2.70238787e-01 6.01405621e-01
-4.51867163e-01 -1.10711176e-02 -8.82980824e-01 4.30347651e-01
-1.90428686e+00 -6.59416258e-01 -3.87910247e-01 2.08283949e+00
2.03747869e-01 1.76160157e-01 -7.69981090e-03 -1.82109490e-01
6.73639119e-01 4.09503192e-01 -4.67179775e-01 -3.49627763e-01
-8.28167945e-02 5.97200871e-01 8.76528561e-01 3.05229634e-01
-1.18185854e+00 1.16567242e+00 5.72463989e+00 9.93583858e-01
-1.14730287e+00 -2.33624965e-01 3.67590636e-01 2.52306405e-02
-2.72505552e-01 -7.68539160e-02 -7.41628468e-01 2.79668719e-01
2.50715941e-01 1.86742470e-01 1.66727990e-01 1.02981329e+00
3.23184058e-02 1.79347739e-01 -7.63853490e-01 1.09196246e+00
-3.39812070e-01 -1.73396838e+00 3.67175400e-01 1.98895603e-01
5.87207794e-01 4.57072139e-01 -1.73886850e-01 5.34857363e-02
2.76479691e-01 -7.75848627e-01 5.84084392e-01 4.31613982e-01
5.00002503e-01 -9.37385380e-01 5.56148946e-01 1.97144434e-01
-1.71700203e+00 1.94257632e-01 -7.14988530e-01 -1.33687511e-01
7.83561617e-02 7.16819942e-01 -8.44245195e-01 7.92350709e-01
9.29151297e-01 6.93934083e-01 -7.22613037e-01 1.27011144e+00
3.83540243e-02 1.23849519e-01 -8.27314436e-01 -5.19066900e-02
5.58045864e-01 -1.46199986e-01 5.13465881e-01 1.09638369e+00
3.63337487e-01 2.34617606e-01 3.62202317e-01 8.86164010e-01
-1.82245940e-01 1.64717883e-01 -6.45722568e-01 2.57448316e-01
6.17434502e-01 1.51191008e+00 -1.31373811e+00 -2.29579508e-01
-4.81126845e-01 7.19136298e-01 5.98154247e-01 2.96208054e-01
-5.01506090e-01 -6.83482409e-01 9.98459697e-01 3.86123538e-01
5.72298110e-01 -5.94271183e-01 -6.86917305e-01 -7.01916873e-01
1.00057900e-01 -1.27727136e-01 2.44892627e-01 -8.08894634e-01
-1.26568496e+00 6.32057011e-01 -8.03101510e-02 -1.24318230e+00
2.62954891e-01 -6.58486545e-01 -9.43297982e-01 1.01984584e+00
-1.46103990e+00 -1.26136291e+00 -6.27497017e-01 8.37956727e-01
3.45078170e-01 2.91201800e-01 4.58205521e-01 2.62025088e-01
-2.64278471e-01 2.94750243e-01 -4.97338295e-01 6.45418763e-02
6.23102151e-02 -1.38092113e+00 8.06100726e-01 5.72134495e-01
9.96931493e-02 6.23611927e-01 1.63607180e-01 -8.54264796e-01
-1.35454357e+00 -1.34335220e+00 7.78092265e-01 -3.80507886e-01
5.28714418e-01 -5.20022035e-01 -1.12039340e+00 4.54079598e-01
-3.44035119e-01 5.81381500e-01 3.08569968e-01 1.56654249e-04
-6.43750504e-02 -8.63803178e-02 -1.32759762e+00 6.43941760e-01
1.35313809e+00 -3.37893665e-01 -3.95084769e-01 3.13240230e-01
9.41107392e-01 -7.08864391e-01 -1.02534604e+00 6.62619770e-01
1.77322641e-01 -8.46783400e-01 1.35800970e+00 -2.85773635e-01
1.00778252e-01 -7.31778502e-01 -2.24449858e-01 -1.02557456e+00
-7.93370247e-01 -4.54342246e-01 9.09943134e-02 1.04710495e+00
3.50733042e-01 -8.01148236e-01 1.15971851e+00 6.01778984e-01
-6.88501716e-01 -9.89224136e-01 -1.03837609e+00 -4.75651622e-01
-2.58193314e-01 -7.64940619e-01 1.39359057e+00 7.66937912e-01
-4.45586771e-01 -8.98481682e-02 4.33775902e-01 7.57272720e-01
4.14064556e-01 4.10166681e-01 9.82743919e-01 -1.42929721e+00
3.27511668e-01 -7.56222606e-01 -1.05858934e+00 -1.25481474e+00
-1.76008642e-01 -1.05608499e+00 -3.39173488e-02 -1.80244994e+00
-3.79275143e-01 -9.16985393e-01 -3.18573266e-01 5.98131716e-01
-1.82596579e-01 7.09567487e-01 2.00533405e-01 1.73522040e-01
-5.53668797e-01 3.92630637e-01 1.41021812e+00 4.30982700e-03
-4.89486307e-01 1.34551704e-01 -5.47724068e-01 7.57578313e-01
6.15690589e-01 -1.18557021e-01 -1.35748297e-01 -4.28821206e-01
5.89138828e-02 -3.90300930e-01 6.30063295e-01 -1.27713704e+00
4.14659619e-01 -2.29833186e-01 6.11354709e-01 -1.19521368e+00
5.32228887e-01 -1.00067878e+00 1.46562949e-01 -1.01893544e-01
3.47235084e-01 2.39593238e-01 3.59468460e-01 2.27775946e-01
-3.01985592e-01 -1.77713409e-02 4.23632324e-01 -2.45760813e-01
-1.01153851e+00 1.03774202e+00 1.69344097e-01 -4.71707702e-01
1.12430561e+00 -7.11377025e-01 -1.23470351e-01 1.63472846e-01
-7.01239884e-01 4.23503011e-01 5.33070445e-01 2.45798424e-01
9.19595063e-01 -1.47159147e+00 -4.37678814e-01 6.84646130e-01
2.24998608e-01 7.86285281e-01 4.69749540e-01 4.43140507e-01
-9.21141028e-01 2.52948105e-01 -1.65701360e-02 -1.01408517e+00
-1.11297190e+00 4.31455314e-01 9.70746279e-02 2.71921605e-02
-9.79248285e-01 1.13700628e+00 2.68727303e-01 -6.47535861e-01
-8.60337317e-02 -9.48843777e-01 -4.52237017e-02 -1.03066333e-01
1.69825703e-01 2.36417219e-01 5.43694854e-01 -7.24131346e-01
-4.99988288e-01 1.07716763e+00 -2.35205423e-02 4.65723336e-01
1.41997409e+00 9.45642591e-02 -2.83915281e-01 1.33320093e-01
1.05604231e+00 2.53089890e-02 -1.33699369e+00 -1.38901249e-01
-1.73832774e-01 -7.36175537e-01 1.38526946e-01 -2.22697139e-01
-1.26090574e+00 7.31666565e-01 2.35397667e-01 6.49948955e-01
1.02001822e+00 3.83362502e-01 7.66867399e-01 3.34527075e-01
6.31446004e-01 -6.10935509e-01 -3.99769634e-01 6.09221458e-01
8.46749663e-01 -9.01644170e-01 2.42683869e-02 -9.43697453e-01
-2.14888215e-01 1.04297400e+00 6.66166186e-01 -6.36846066e-01
9.67510283e-01 2.68910050e-01 -1.81152314e-01 -6.88540876e-01
-3.33910316e-01 -3.88834953e-01 4.56918985e-01 5.71604311e-01
4.52345014e-02 2.12873667e-01 2.33671233e-01 3.11139822e-01
-4.29846555e-01 -2.22280473e-01 -1.73548907e-01 1.07366502e+00
-6.39459550e-01 -9.13189173e-01 -4.53016996e-01 8.91544044e-01
-3.13453041e-02 -2.54137218e-02 -3.25526178e-01 7.99970627e-01
4.36223418e-01 3.91044706e-01 7.08782375e-01 -4.24718082e-01
7.12713659e-01 -1.67354822e-01 2.60200500e-01 -8.40066969e-01
-7.76869237e-01 -3.64349186e-02 -4.65956479e-01 -8.65156829e-01
-2.09247902e-01 -5.90989113e-01 -1.60781336e+00 -4.43830729e-01
-1.79819122e-01 -2.25937888e-02 6.26246631e-01 7.39974737e-01
6.96277201e-01 6.02538645e-01 6.01149619e-01 -1.61694312e+00
1.84911609e-01 -5.34178138e-01 -6.03603542e-01 1.37149677e-01
2.68865824e-01 -7.51819432e-01 -8.81073102e-02 -4.94810700e-01] | [7.9206862449646, -3.5113942623138428] |
def27bac-6ea4-42a8-9566-b6f65a94bbd9 | emfet-e-mail-features-extraction-tool | 1711.08521 | null | http://arxiv.org/abs/1711.08521v1 | http://arxiv.org/pdf/1711.08521v1.pdf | EMFET: E-mail Features Extraction Tool | EMFET is an open source and flexible tool that can be used to extract a large
number of features from any email corpus with emails saved in EML format. The
extracted features can be categorized into three main groups: header features,
payload (body) features, and attachment features. The purpose of the tool is to
help practitioners and researchers to build datasets that can be used for
training machine learning models for spam detection. So far, 140 features can
be extracted using EMFET. EMFET is extensible and easy to use. The source code
of EMFET is publicly available at GitHub
(https://github.com/WadeaHijjawi/EmailFeaturesExtraction) | ["Ala' M. Al-Zoubi", 'Ibrahim Aljarah', "Ja'far Alqatawna", 'Hossam Faris', "Wadi' Hijawi", 'Maria Habib'] | 2017-11-22 | null | null | null | null | ['spam-detection'] | ['natural-language-processing'] | [-3.95922154e-01 -4.45259362e-01 -1.35140330e-01 -4.27810341e-01
-5.57977498e-01 -6.84512615e-01 6.16186261e-01 2.69772977e-01
-3.34546357e-01 5.79029262e-01 2.08324656e-01 -5.38517714e-01
-5.25128804e-02 -7.11650550e-01 3.51035967e-02 -2.85845309e-01
3.21445495e-01 3.27643096e-01 3.41293752e-01 -3.66054207e-01
6.39643192e-01 5.09325027e-01 -1.44984913e+00 5.76269209e-01
6.28585637e-01 7.17697084e-01 3.42166394e-01 7.74600148e-01
-4.67052758e-01 4.72336024e-01 -7.20364273e-01 -7.67886460e-01
-2.79197600e-02 -2.23895654e-01 -1.18307602e+00 -3.84189308e-01
1.58683494e-01 -3.57161939e-01 -5.80212653e-01 7.01815963e-01
6.97234273e-01 -2.05862597e-01 4.33598369e-01 -1.20965052e+00
-3.33671004e-01 4.42388177e-01 -5.83741106e-02 3.98847044e-01
8.05725574e-01 5.93873039e-02 1.04014266e+00 -8.06359112e-01
6.84275568e-01 1.32546449e+00 7.65734792e-01 3.27811867e-01
-9.08551037e-01 -7.15646386e-01 -3.87590796e-01 3.70382845e-01
-9.46920872e-01 -3.87936413e-01 4.75210369e-01 -4.36941266e-01
1.03029358e+00 7.48533607e-01 3.40122312e-01 1.48571885e+00
3.30260694e-01 8.77647221e-01 1.28545654e+00 -3.47661734e-01
-1.81156084e-01 5.37151814e-01 7.20709801e-01 8.31722796e-01
1.96198463e-01 -1.41739592e-01 -5.25886118e-01 -9.53988552e-01
3.35012347e-01 2.29784101e-01 -1.00697435e-01 3.42102617e-01
-7.65777647e-01 1.20214128e+00 2.04635590e-01 5.66311777e-01
-7.25363195e-02 -2.26443857e-01 7.05313742e-01 8.24387670e-01
4.48861390e-01 3.94465595e-01 -7.15521634e-01 -5.98689973e-01
-6.51953518e-01 3.44314158e-01 1.19206977e+00 9.52607691e-01
6.86817646e-01 -5.90477109e-01 6.79030642e-02 1.09622526e+00
3.83993536e-01 4.10775006e-01 7.30277121e-01 -6.93783283e-01
4.09140915e-01 8.04507136e-01 2.20865682e-02 -9.02886033e-01
-3.67573321e-01 3.69674489e-02 -2.53571332e-01 -2.32439101e-01
4.26400453e-01 -2.55667746e-01 -3.80369127e-01 9.40953135e-01
2.82432050e-01 -2.17028201e-01 -5.44166207e-01 5.88954449e-01
1.44728100e+00 3.90351772e-01 -2.88814176e-02 2.01428384e-01
1.53123450e+00 -5.49095869e-01 -5.80849349e-01 3.22164327e-04
8.46893728e-01 -1.40670586e+00 9.31320906e-01 1.13468908e-01
-7.63570964e-01 -2.18526825e-01 -5.00996053e-01 -4.10924107e-02
-8.79800141e-01 -4.27572653e-02 7.72059798e-01 8.73922110e-01
-8.43721211e-01 7.32726395e-01 -6.32477701e-01 -7.97409356e-01
3.69868755e-01 3.97080183e-01 -4.15141791e-01 9.64768454e-02
-1.27189934e+00 9.30474222e-01 2.74319232e-01 -4.70777631e-01
-1.11807488e-01 -2.50987828e-01 -5.16076744e-01 -1.72513098e-01
1.74435839e-01 -6.18506372e-01 1.45850551e+00 -7.17516124e-01
-1.16800427e+00 7.77997196e-01 -3.75083953e-01 -1.67573363e-01
4.90554661e-01 -2.18807086e-01 -4.46226865e-01 2.86109865e-01
-2.36573350e-02 -8.31115097e-02 8.43669415e-01 -8.12273562e-01
-6.17270470e-01 -3.22110087e-01 -2.80474156e-01 -3.21609914e-01
-7.02251136e-01 8.85511518e-01 -3.89725089e-01 -6.76051974e-01
-3.85549784e-01 -8.25293899e-01 2.96517432e-01 -3.50758821e-01
-4.31792110e-01 -4.53010976e-01 1.27239335e+00 -1.06007874e+00
1.60225558e+00 -1.80952215e+00 -3.70477140e-01 7.08930641e-02
2.38354281e-01 7.38863766e-01 -1.06112137e-01 1.14286351e+00
1.63683891e-01 2.42988274e-01 2.14005992e-01 -5.23527525e-02
1.92265794e-01 -1.49539679e-01 -9.63657349e-02 3.12682688e-01
2.80652605e-02 9.24306095e-01 -9.15981650e-01 -6.09321892e-01
4.94158834e-01 4.90281373e-01 -2.09573150e-01 3.83770227e-01
1.17761768e-01 8.99342597e-02 -8.45626235e-01 5.87085247e-01
6.94532812e-01 -2.03435957e-01 1.07424460e-01 2.02360645e-01
4.65348326e-02 7.40093589e-01 -8.59967887e-01 8.92445028e-01
-4.99085635e-01 6.31771982e-01 5.43078855e-02 -3.83539706e-01
8.37160826e-01 2.51976311e-01 3.86382222e-01 -1.74996722e-02
7.19177544e-01 1.25763774e-01 -3.63762140e-01 -5.53610682e-01
3.79095435e-01 2.77120322e-01 -1.02926672e-01 8.18478644e-01
4.16565895e-01 1.48918465e-01 5.26962936e-01 5.26934683e-01
1.45331252e+00 -2.95271188e-01 2.96971589e-01 -1.50359601e-01
7.66202211e-01 -2.28003249e-01 3.49860266e-02 6.80595577e-01
-6.05104901e-02 1.58454210e-01 5.04608691e-01 -3.42920989e-01
-8.87540102e-01 -9.87479925e-01 -5.40928781e-01 1.26484394e+00
-3.55798036e-01 -1.15354025e+00 -7.49991536e-01 -1.13764191e+00
3.75378877e-01 5.24056494e-01 -3.59749615e-01 9.83927771e-02
-4.45115656e-01 -6.57834709e-01 4.00787205e-01 2.29905590e-01
3.34801883e-01 -1.23899841e+00 -1.98994532e-01 1.13884389e-01
-4.27330554e-01 -8.02337766e-01 -4.80462551e-01 1.70314729e-01
-8.42251778e-01 -1.23508990e+00 -2.01785386e-01 -7.24089801e-01
5.10482788e-01 4.03171659e-01 1.15765905e+00 6.38042927e-01
-6.48535907e-01 5.02449811e-01 -8.04968596e-01 -5.32569647e-01
-4.96447176e-01 4.72428590e-01 -1.74439564e-01 -4.04145360e-01
9.53613043e-01 -3.49600703e-01 -5.21351874e-01 4.98421460e-01
-7.51462877e-01 -4.07495320e-01 3.11657965e-01 9.89408076e-01
-2.06954122e-01 -3.15660089e-01 8.51371706e-01 -1.14706647e+00
1.03712618e+00 -7.64214694e-01 -3.36849809e-01 -1.05334371e-01
-4.84982669e-01 -2.65117943e-01 6.56542540e-01 -6.10923246e-02
-9.23840344e-01 -1.52335957e-01 -7.87987530e-01 3.90492260e-01
-3.51891577e-01 2.59274542e-01 6.67457208e-02 -2.05174387e-01
4.49903190e-01 1.73306778e-01 1.86735451e-01 -9.89898026e-01
1.60793021e-01 1.70175838e+00 -2.23344371e-01 -3.67140979e-01
6.69514477e-01 7.61284977e-02 -5.47958255e-01 -9.24349368e-01
-5.81512749e-01 -9.48103070e-01 -8.45864534e-01 -1.51902556e-01
1.85108915e-01 -2.99752802e-01 -7.14131594e-01 6.35334194e-01
-6.17373109e-01 1.16178222e-01 2.09997565e-01 -6.72577322e-03
-3.29932958e-01 8.55930626e-01 -1.05596209e+00 -5.84816933e-01
-8.56334090e-01 -5.82416534e-01 7.39157259e-01 2.83888519e-01
-6.91925824e-01 -1.17767227e+00 -1.38261780e-01 5.79521656e-01
4.41422671e-01 -1.93795696e-01 7.30525970e-01 -1.16835690e+00
-1.19632751e-01 -7.72543728e-01 -1.52481601e-01 3.38423908e-01
2.28797287e-01 2.32887030e-01 -8.63215148e-01 -5.35922706e-01
-9.69226211e-02 -2.91306853e-01 8.79827619e-01 -7.56317303e-02
1.24438822e+00 -5.79385698e-01 -5.56442916e-01 3.34359735e-01
1.10062933e+00 -8.54929760e-02 5.78833640e-01 6.85391426e-01
7.38766193e-02 4.10455018e-01 5.86305082e-01 6.43612385e-01
3.02881896e-01 4.79753047e-01 5.36485948e-02 2.97144860e-01
1.45075947e-01 -9.98299569e-02 3.93071622e-01 9.22202826e-01
3.18130016e-01 -2.63022900e-01 -7.69207239e-01 2.90987223e-01
-1.84423721e+00 -1.15585542e+00 -6.00047708e-01 1.94882786e+00
7.49643087e-01 6.77932352e-02 5.66167712e-01 1.34279383e-02
8.20553064e-01 -5.27357012e-02 -4.55175787e-02 -7.78685451e-01
2.34364957e-01 4.33397055e-01 2.52415478e-01 3.65970939e-01
-1.30344069e+00 8.81538332e-01 6.95804214e+00 8.26283097e-01
-1.00491178e+00 2.35725582e-01 1.88528270e-01 -5.22894561e-02
2.19163850e-01 2.55901311e-02 -1.18523824e+00 9.67299700e-01
1.53018415e+00 -4.96179193e-01 2.16249853e-01 7.70946503e-01
1.41175613e-01 -7.64474273e-02 -5.06704688e-01 6.59617960e-01
-2.06778601e-01 -1.31090093e+00 -1.21276662e-01 1.48964878e-02
1.17679901e-01 4.12023723e-01 -1.61070675e-02 2.79120445e-01
4.14308429e-01 -6.64197326e-01 8.86103958e-02 3.20757359e-01
5.32296002e-01 -6.87356293e-01 1.10788238e+00 2.66657799e-01
-8.78430247e-01 -3.52480292e-01 -4.53409821e-01 3.49478051e-02
4.41486835e-02 6.24610662e-01 -1.03684783e+00 6.76142931e-01
8.37296784e-01 6.71961188e-01 -1.09592581e+00 1.34043574e+00
-3.81705188e-03 9.73790884e-01 -1.61841884e-01 -3.74096096e-01
8.64390656e-02 -2.50692904e-01 5.73713005e-01 1.73432064e+00
2.45077416e-01 -3.63524616e-01 6.36920780e-02 4.72848088e-01
8.91373605e-02 4.47601438e-01 -5.10062575e-01 -3.07564288e-01
6.03847086e-01 1.73177028e+00 -6.50443733e-01 -2.71287113e-01
-6.30820513e-01 9.90818441e-01 2.64992535e-01 -6.33416185e-03
-6.34570062e-01 -9.75838363e-01 5.85287273e-01 4.96490628e-01
3.48018974e-01 -1.69838265e-01 -4.96692322e-02 -1.07903528e+00
-5.18153012e-02 -9.63512123e-01 7.93042123e-01 -5.18816829e-01
-1.83966446e+00 7.74784923e-01 -9.00149494e-02 -9.86943543e-01
-2.96961188e-01 -1.03741157e+00 -8.11794639e-01 9.98603344e-01
-1.00012541e+00 -1.20104349e+00 -3.39634448e-01 5.87232590e-01
2.90146917e-01 -4.21313584e-01 1.11047995e+00 2.94374973e-01
-6.33714378e-01 6.54456496e-01 5.80322921e-01 3.77315700e-01
1.00894260e+00 -1.17359257e+00 6.07234478e-01 1.96065411e-01
-1.36431158e-01 1.16437352e+00 5.22989571e-01 -7.19556272e-01
-1.21661472e+00 -1.10368478e+00 1.30255985e+00 -9.31415856e-01
9.77003098e-01 -5.14233291e-01 -7.81546533e-01 6.83970153e-01
3.32121342e-01 -3.45739424e-01 1.04363370e+00 2.59767294e-01
-3.60477835e-01 3.71978700e-01 -1.32037842e+00 2.64419079e-01
8.06239665e-01 -4.94055539e-01 -6.94901645e-01 5.05540371e-01
-1.53070673e-01 -7.96709880e-02 -1.10529804e+00 -9.39223170e-02
6.70583427e-01 -1.11197209e+00 7.40931690e-01 -5.97616792e-01
1.16673857e-01 2.43186682e-01 3.45269710e-01 -1.04579258e+00
-4.77394581e-01 -7.24883258e-01 -3.65750641e-01 1.40575969e+00
4.13178593e-01 -1.29525971e+00 4.65195924e-01 2.16345355e-01
2.85257727e-01 -1.01392233e+00 -8.27500641e-01 -8.05357993e-01
3.23089927e-01 -1.49225742e-01 7.03973949e-01 7.63951302e-01
5.38059711e-01 3.85989159e-01 -6.81434646e-02 -5.95298946e-01
2.40310699e-01 1.56725883e-01 8.79756749e-01 -1.51910806e+00
-7.41699189e-02 -4.40668464e-01 -3.70880455e-01 -9.25791800e-01
1.97877154e-01 -1.29530191e+00 -6.99244857e-01 -1.60385108e+00
5.98345757e-01 -3.56977552e-01 -2.35646591e-01 5.88061094e-01
-2.32497171e-01 1.76398367e-01 2.41675168e-01 2.46142283e-01
-4.64984506e-01 2.60142863e-01 8.84349406e-01 2.01971561e-01
1.45255551e-01 4.74471927e-01 -8.92878234e-01 5.09140074e-01
1.27455044e+00 -6.60492778e-01 2.47580245e-01 3.24496686e-01
-2.86274582e-01 -4.39841419e-01 3.82922828e-01 -5.35687387e-01
-2.72482008e-01 1.29980789e-02 5.11123538e-01 -7.35255480e-01
1.94797486e-01 -4.98513043e-01 -1.44999951e-01 3.77947360e-01
-2.61940472e-02 1.85707197e-01 6.99973106e-02 9.73508060e-02
-3.48200649e-02 -7.92796254e-01 5.79586029e-01 -2.20262721e-01
-3.75043124e-01 1.24852754e-01 -8.37600470e-01 -7.19355643e-02
7.91835487e-01 1.91871636e-02 -7.55558908e-01 -5.83008111e-01
-4.59181905e-01 1.27216801e-01 7.00953066e-01 8.69258583e-01
5.48152506e-01 -1.17413568e+00 -6.35984838e-01 3.72234672e-01
8.65829438e-02 -1.15121591e+00 -2.73369183e-03 8.42103720e-01
-5.25308967e-01 6.90024436e-01 -3.72421741e-01 -2.31335297e-01
-1.81153190e+00 3.76384228e-01 7.49791914e-04 -2.50749677e-01
-8.52587819e-01 5.92663944e-01 -5.58674753e-01 -7.62785614e-01
1.10072076e-01 2.56145656e-01 -2.11951882e-01 1.30341034e-02
7.84973919e-01 5.00543475e-01 3.30849349e-01 -6.37503028e-01
-4.07294631e-01 -1.08580582e-01 -4.65626836e-01 3.33269060e-01
1.49063599e+00 -1.16652258e-01 -4.32644755e-01 2.00032651e-01
1.49369740e+00 2.31346399e-01 -3.32794458e-01 -1.37607679e-01
2.60033876e-01 -9.62497592e-01 -1.48024112e-01 -8.46989155e-01
-5.84120214e-01 5.30138016e-01 5.94790518e-01 5.29728472e-01
8.27592552e-01 2.20621735e-01 1.26894414e+00 4.40741956e-01
3.02753359e-01 -1.19963932e+00 -1.87540706e-02 8.15989971e-01
8.52165282e-01 -1.14030683e+00 -7.60819390e-02 -5.55522740e-01
-4.76528943e-01 1.34393704e+00 3.50298226e-01 2.19371114e-02
1.00921071e+00 2.65131533e-01 1.62392154e-01 -2.30740547e-01
-8.39184165e-01 -2.40925342e-01 2.59203404e-01 4.64901268e-01
1.02127838e+00 -7.06783384e-02 -8.65235031e-01 7.82949626e-01
-5.04346669e-01 -2.25184366e-01 4.11018312e-01 1.27193367e+00
-6.16317868e-01 -1.81640077e+00 -3.83923411e-01 1.29706144e+00
-8.61308753e-01 -3.42480950e-02 -9.25834000e-01 5.24831116e-01
-2.22174227e-01 1.24604666e+00 -4.18347180e-01 -6.19882226e-01
1.06922835e-01 4.11281675e-01 2.95478612e-01 -5.77718258e-01
-1.13981390e+00 -2.24057212e-01 5.73353946e-01 -3.11281621e-01
8.57521743e-02 -7.51320660e-01 -9.60682392e-01 -9.14536476e-01
-6.18100584e-01 4.95580912e-01 4.71223086e-01 5.25860727e-01
5.74935257e-01 1.86021924e-02 8.58573556e-01 -5.41220784e-01
-6.16411507e-01 -1.28255486e+00 -3.89997482e-01 5.44077158e-01
-3.94750945e-02 -6.82834625e-01 -4.70996201e-01 -2.39705473e-01] | [7.940494537353516, 9.982787132263184] |
6d43d10a-0314-4b92-80ed-c77f1141aff6 | disaggregating-hops-can-we-guide-a-multi-hop | null | null | https://openreview.net/forum?id=RxOWqx2hwgz | https://openreview.net/pdf?id=RxOWqx2hwgz | Disaggregating Hops: Can We Guide a Multi-Hop Reasoning Language Model to Incrementally Learn at each Hop? | Despite the success of state-of-the-art pre-trained language models (PLMs) on a series of multi-hop reasoning tasks, they still suffer from their limited abilities to transfer learning from simple to complex tasks and vice-versa. We argue that one step forward to overcome this limitation is to better understand the behavioral trend of PLMs at each hop over the inference chain. Our critical underlying idea is to mimic human-style reasoning: we envision the multi-hop reasoning process as a sequence of explicit one-hop incremental reasoning steps. Using the SHINRA and ConceptNet resources jointly, we provide automatically generated datasets built upon a set of inference heuristics on relevant phrases and distractors, allowing us to teach the models incremental reasoning skills. We empirically show the effectiveness of the proposed models on multiple-choice question answering (MCQA) and reading comprehension (RC), with a relative improvement of $68.4\%$ and $16.0\%$ accuracy improvement w.r.t. classic PLMs, respectively. | ['Anonymous'] | 2022-01-16 | null | null | null | acl-arr-january-2022-1 | ['multiple-choice-qa'] | ['natural-language-processing'] | [ 2.39527360e-01 4.99392211e-01 1.71973761e-02 -3.89567733e-01
-8.49318027e-01 -6.09046638e-01 5.17022848e-01 4.45959747e-01
-5.47901630e-01 8.28133225e-01 1.59531713e-01 -9.28604722e-01
-2.71590918e-01 -9.64022040e-01 -8.98383498e-01 -6.49435371e-02
1.50460705e-01 7.76110113e-01 4.89109516e-01 -5.39368808e-01
5.19002259e-01 1.88076764e-01 -1.27741861e+00 5.13085246e-01
1.45187783e+00 7.52512991e-01 2.17967540e-01 8.19932342e-01
-5.79909027e-01 1.55901062e+00 -4.12772149e-01 -9.84663010e-01
-6.96761236e-02 -4.84225810e-01 -1.27646983e+00 -4.12467003e-01
4.74629909e-01 -5.34469545e-01 -1.31198183e-01 1.01586497e+00
2.52021819e-01 3.28663021e-01 6.49351001e-01 -1.01034212e+00
-9.70754445e-01 9.94734406e-01 -2.79332340e-01 3.57384592e-01
6.24495983e-01 2.51274884e-01 1.37618113e+00 -7.00171590e-01
3.52920771e-01 1.35207570e+00 5.24611056e-01 7.42824078e-01
-1.11233592e+00 -5.33880293e-01 3.59692037e-01 5.78928709e-01
-9.40315306e-01 -1.81225926e-01 5.91512501e-01 -2.65122861e-01
1.08703709e+00 4.60329056e-02 3.50859225e-01 9.58238065e-01
-8.03773105e-02 1.21252763e+00 1.30622518e+00 -6.76407218e-01
1.91852078e-01 1.61975771e-01 5.32724679e-01 9.23053205e-01
1.52037561e-01 -3.02923262e-01 -5.24856687e-01 1.03826366e-01
4.25441265e-01 -6.07192852e-02 -6.49349168e-02 -4.56777811e-02
-8.92306209e-01 8.28400433e-01 3.85160327e-01 1.77811533e-01
-3.26825678e-01 -4.41157036e-02 1.63658515e-01 4.47416455e-01
-7.25367144e-02 8.44418228e-01 -7.48066843e-01 -1.95631161e-01
-8.10297966e-01 4.65234190e-01 1.04776525e+00 1.01092935e+00
5.65666795e-01 -3.41723174e-01 -5.94610155e-01 8.16790342e-01
2.51004487e-01 5.40333807e-01 3.66713017e-01 -1.13202584e+00
7.78944969e-01 7.29066908e-01 2.13845670e-01 -7.21900225e-01
-3.04849774e-01 -4.53949124e-01 -4.37719166e-01 -3.72866601e-01
5.66899300e-01 -5.79086058e-02 -5.78345835e-01 1.84953701e+00
9.28211678e-03 -3.40536349e-02 1.70145154e-01 5.26824713e-01
5.29195786e-01 5.45922220e-01 4.66130108e-01 -1.19531266e-02
1.53931248e+00 -1.31850636e+00 -5.86578488e-01 -3.90315443e-01
5.25867462e-01 -4.71035123e-01 1.65157485e+00 4.55934286e-01
-1.58157396e+00 -5.99348903e-01 -8.44185829e-01 -3.27208400e-01
-2.31996760e-01 -2.98306316e-01 6.93447471e-01 2.88631827e-01
-9.09287274e-01 3.85783523e-01 -4.71612781e-01 -1.80362776e-01
4.65897948e-01 -9.64760780e-02 2.64479369e-01 -5.37197113e-01
-1.50020730e+00 1.08592474e+00 3.45331699e-01 -1.01831011e-01
-8.92110229e-01 -9.67055619e-01 -5.46337545e-01 5.08665860e-01
8.05252731e-01 -8.32325518e-01 1.77607238e+00 -6.37854278e-01
-1.57712734e+00 6.88052356e-01 -2.20790297e-01 -5.59643269e-01
6.05741084e-01 -6.95118248e-01 -2.95701563e-01 3.72365594e-01
2.04919547e-01 7.00791717e-01 4.84118134e-01 -1.11027324e+00
-7.82907486e-01 7.07119331e-02 7.31262207e-01 3.13336402e-01
-1.61770463e-01 -2.30448067e-01 -1.75960630e-01 -4.39879239e-01
-7.19352812e-02 -5.81279218e-01 5.34303486e-03 -1.08273365e-01
-2.55774200e-01 -6.23091996e-01 8.89435187e-02 -7.50810981e-01
1.31169319e+00 -1.56910455e+00 2.02697843e-01 -7.88615271e-02
2.90845037e-01 4.52997446e-01 -4.35635835e-01 5.56506515e-01
2.27314189e-01 1.67511567e-01 -1.85870558e-01 -1.07518084e-01
2.20767125e-01 1.64542153e-01 -5.28449774e-01 -5.24365485e-01
2.90782124e-01 1.17110121e+00 -1.28042376e+00 -5.63210487e-01
-1.09420016e-01 -2.18813598e-01 -9.17101264e-01 6.17485821e-01
-9.53916848e-01 1.97078288e-01 -4.37968314e-01 5.95980585e-01
5.68500042e-01 -5.60425103e-01 2.81392515e-01 1.35576218e-01
3.18094939e-01 7.86745906e-01 -7.78750539e-01 1.79893923e+00
-6.20940447e-01 3.29121947e-01 -2.82378495e-01 -1.00406206e+00
5.75131476e-01 4.92703319e-02 -2.01076016e-01 -1.14816630e+00
-1.43383592e-01 7.72479251e-02 5.26425183e-01 -7.48061359e-01
4.18010503e-01 -3.45792860e-01 4.82011773e-02 7.04891264e-01
1.53941602e-01 -1.77742750e-03 3.89320344e-01 3.98113191e-01
1.21240556e+00 1.66032508e-01 3.28109682e-01 -2.64611185e-01
9.09354687e-01 9.48481262e-02 2.19518349e-01 1.14173031e+00
-2.22668797e-01 -3.01860273e-02 5.37377357e-01 -2.55126387e-01
-7.36397982e-01 -1.32275653e+00 2.59001106e-01 1.54251993e+00
-5.73601983e-02 -3.28003645e-01 -8.25851679e-01 -7.16196597e-01
-9.08191130e-02 1.39194584e+00 -3.63988072e-01 -1.84164867e-01
-7.83898890e-01 -2.64551938e-01 7.47788787e-01 6.98605955e-01
1.00853074e+00 -1.22206414e+00 -7.20253170e-01 3.07180822e-01
-3.78110766e-01 -1.19627035e+00 4.62534018e-02 -1.75663263e-01
-8.46274018e-01 -1.09376383e+00 -3.17348450e-01 -6.02783084e-01
6.02082908e-01 2.77463496e-01 1.59592772e+00 6.25379801e-01
2.52466630e-02 4.65466589e-01 -3.93159926e-01 -3.23789150e-01
-4.62069660e-01 1.49323717e-01 -1.92354411e-01 -5.74755788e-01
6.51677549e-01 -5.35483360e-01 -7.20237136e-01 9.18941572e-02
-7.37659454e-01 3.76171678e-01 8.69510174e-01 8.15080822e-01
1.69592261e-01 -1.10407598e-01 7.53922164e-01 -1.13078880e+00
1.16726923e+00 -6.21918321e-01 -4.46237803e-01 7.85034597e-01
-7.67297268e-01 3.66660386e-01 9.93912995e-01 -3.27160507e-01
-1.40631509e+00 -8.06764960e-01 -2.65774488e-01 1.30968951e-02
-2.24775597e-01 6.76426053e-01 1.30221993e-01 3.35533112e-01
5.81880093e-01 5.00176251e-01 -2.86556453e-01 -3.30656379e-01
7.28500605e-01 2.28183255e-01 3.96780223e-01 -1.35068226e+00
8.71951163e-01 -1.23577761e-02 -2.98138201e-01 -3.79354119e-01
-1.43789220e+00 -1.52977094e-01 -2.85828859e-01 9.31397900e-02
8.63197923e-01 -7.31208384e-01 -1.16242731e+00 1.82695329e-01
-1.19920099e+00 -6.68616831e-01 -4.30540368e-02 5.92995575e-03
-4.73835438e-01 3.13029796e-01 -8.29261601e-01 -8.59765410e-01
-4.54015404e-01 -9.58673179e-01 5.03227949e-01 4.60916609e-01
-3.94414663e-01 -1.16344011e+00 -3.44519280e-02 9.14858699e-01
6.42024875e-01 -5.09187877e-01 1.81679475e+00 -9.19544518e-01
-7.92338610e-01 1.23813361e-01 -3.88001502e-01 4.65175390e-01
-2.04879403e-01 -4.09126610e-01 -9.11574841e-01 -4.58990894e-02
6.85604140e-02 -7.75530100e-01 5.92229784e-01 -8.27417225e-02
1.43030334e+00 -3.65359306e-01 -2.23432407e-02 1.02366500e-01
1.31028795e+00 1.76951319e-01 6.03100181e-01 2.04448476e-01
4.24700797e-01 4.78904068e-01 5.58772326e-01 -5.97390272e-02
9.03858066e-01 1.85369134e-01 6.15726113e-02 4.56536084e-01
-8.77462998e-02 -8.43409956e-01 2.55109370e-01 1.06018901e+00
4.58069658e-03 -7.59663731e-02 -1.33140910e+00 5.70739806e-01
-1.63593185e+00 -9.27540004e-01 3.23594600e-01 1.82360923e+00
1.19748926e+00 4.71259385e-01 -1.23349480e-01 -5.06173223e-02
2.56543785e-01 -8.27908441e-02 -5.84365785e-01 -5.44366896e-01
3.41823161e-01 6.36210501e-01 -9.57587361e-02 7.37693727e-01
-4.09036547e-01 1.18314230e+00 6.32066965e+00 7.70061910e-01
-7.33569384e-01 -1.19372204e-01 4.88436908e-01 2.14939430e-01
-4.16930437e-01 4.57665548e-02 -7.54076600e-01 1.53535634e-01
1.07732844e+00 -8.97663385e-02 7.29802072e-01 7.07272649e-01
-2.48624504e-01 -2.65512764e-01 -1.36348689e+00 5.04022598e-01
1.79519296e-01 -1.30974841e+00 5.33890843e-01 -6.59925699e-01
7.31406987e-01 -3.34856302e-01 -5.93461059e-02 1.02464473e+00
5.85760832e-01 -1.19311702e+00 5.66514313e-01 6.85625255e-01
2.57372499e-01 -5.51821172e-01 5.99946320e-01 8.75612378e-01
-7.53982961e-01 -3.53737622e-01 -3.22522700e-01 -5.41079879e-01
-9.21203867e-02 8.54554623e-02 -1.01674867e+00 4.53126818e-01
5.06764591e-01 1.92827508e-01 -7.64864683e-01 5.54192543e-01
-9.08380747e-01 8.25311899e-01 1.29266381e-01 -5.88330209e-01
2.12497443e-01 -1.25319958e-01 6.30828515e-02 1.05573595e+00
-3.00009213e-02 5.46530843e-01 -1.82517450e-02 1.17638922e+00
-2.93936700e-01 -7.81688318e-02 -1.18759632e-01 -1.64238825e-01
7.92490005e-01 9.46393728e-01 -3.80059183e-01 -5.71766555e-01
-7.10561454e-01 8.32462847e-01 9.30679083e-01 4.84087825e-01
-8.45987320e-01 -3.36329371e-01 4.12810773e-01 7.29755685e-02
2.37250492e-01 -1.56232387e-01 -3.88108313e-01 -1.29691839e+00
1.24746911e-01 -1.29305446e+00 5.37001431e-01 -1.06527591e+00
-1.61496902e+00 2.56805658e-01 1.00491382e-01 -5.77740431e-01
-1.81066900e-01 -7.67565787e-01 -6.76897407e-01 9.18449342e-01
-1.95867896e+00 -9.71015334e-01 -3.69659781e-01 3.75796497e-01
7.67167449e-01 -1.50058940e-01 9.53631997e-01 5.23314364e-02
-3.19811642e-01 8.01841974e-01 -3.31769615e-01 2.90791869e-01
4.38482642e-01 -1.32664394e+00 4.24410552e-01 7.98926413e-01
3.24445479e-02 1.13771927e+00 6.15160346e-01 -2.73211271e-01
-1.51703680e+00 -6.37091637e-01 9.82632756e-01 -7.46202052e-01
9.08297956e-01 -2.42468953e-01 -1.14617705e+00 8.75528753e-01
4.97928381e-01 -5.44908345e-01 7.29482412e-01 3.12702924e-01
-6.17734551e-01 -2.45733894e-02 -1.08915496e+00 1.14858937e+00
1.15031612e+00 -7.05811560e-01 -1.50769305e+00 2.76753724e-01
8.76841187e-01 -3.77688646e-01 -7.65642345e-01 1.58030972e-01
3.45606804e-01 -9.72457528e-01 1.09724665e+00 -1.17726672e+00
1.02275908e+00 -7.24459887e-02 -1.87765732e-01 -1.21727586e+00
-2.74406493e-01 -4.51024830e-01 -3.50036472e-01 9.89694774e-01
5.22275329e-01 -5.45731068e-01 4.30957377e-01 7.51738191e-01
-4.78224382e-02 -8.41931581e-01 -5.63539922e-01 -4.23424065e-01
6.28451228e-01 -5.18642902e-01 5.92872143e-01 8.12678635e-01
2.01838031e-01 7.67316401e-01 8.50227848e-02 1.54827803e-01
4.59677815e-01 2.27311701e-01 6.47216678e-01 -1.05818677e+00
-6.57853425e-01 -5.53028047e-01 4.12268609e-01 -1.66754937e+00
3.64428222e-01 -8.85789454e-01 2.23932303e-02 -1.59013319e+00
3.42634290e-01 -4.49976295e-01 -4.98493493e-01 3.64588648e-01
-7.17121005e-01 -4.56551552e-01 2.75237709e-01 -1.21289164e-01
-9.25296307e-01 2.84065902e-01 1.51283503e+00 -4.50925566e-02
2.91715711e-01 -1.12037539e-01 -9.91597831e-01 7.50232875e-01
8.10741544e-01 -2.31506780e-01 -8.74199510e-01 -7.76186526e-01
6.55302048e-01 3.44286591e-01 3.15911829e-01 -9.76997972e-01
3.84633064e-01 -3.61251324e-01 8.88252407e-02 -4.25616473e-01
4.93430533e-02 -5.67010283e-01 -7.26459384e-01 5.27086318e-01
-9.30249989e-01 3.79631728e-01 4.07219410e-01 3.86237860e-01
-9.75370035e-02 -4.02915061e-01 5.59271514e-01 -4.84452635e-01
-7.60519028e-01 -1.45365596e-01 -2.23729759e-01 6.99991405e-01
6.71461523e-01 3.28740448e-01 -7.20817685e-01 -4.55728889e-01
-4.19984400e-01 5.73369741e-01 1.37709798e-02 3.33840668e-01
6.04690492e-01 -8.98703218e-01 -6.41749978e-01 -4.70701680e-02
1.00525640e-01 1.07362561e-01 2.68127888e-01 7.08563864e-01
-7.28410304e-01 8.01201284e-01 -3.63682181e-01 -2.67352909e-01
-8.27385366e-01 6.29825234e-01 3.95083457e-01 -6.41300440e-01
-3.81023794e-01 1.07349086e+00 -1.38633698e-01 -5.77815354e-01
2.61721641e-01 -5.43645561e-01 -9.44395587e-02 -3.04923385e-01
6.29033685e-01 2.79633373e-01 -1.32547230e-01 3.73281032e-01
-4.29453421e-03 3.63629192e-01 -4.11202937e-01 -7.13928193e-02
1.01116180e+00 -2.35161018e-02 -2.51156002e-01 4.30794299e-01
6.25034928e-01 -5.42959794e-02 -1.00503516e+00 -5.24587572e-01
3.40234876e-01 -1.37832612e-01 -5.54170191e-01 -1.52884161e+00
-4.23905432e-01 1.22973049e+00 -9.31274239e-03 6.70504943e-02
9.27422285e-01 -3.84970717e-02 7.60785997e-01 1.06567109e+00
4.91834074e-01 -9.78812456e-01 3.96909297e-01 7.56923437e-01
7.09414244e-01 -1.21060336e+00 -6.72223270e-02 -2.81623036e-01
-5.90795159e-01 1.03757679e+00 9.51069355e-01 -7.62282982e-02
1.74317271e-01 -1.34878814e-01 5.14179394e-02 -1.07772090e-01
-1.21237946e+00 9.10238270e-03 1.19190887e-01 3.82285833e-01
7.07993805e-01 8.55444670e-02 -3.31018955e-01 7.53962100e-01
-2.68025994e-01 1.25763416e-01 2.96223819e-01 1.02469528e+00
-5.12733638e-01 -9.65689838e-01 -1.16557933e-01 3.59481484e-01
-2.35446990e-01 -5.02923906e-01 -2.59615391e-01 9.18025136e-01
-1.52960390e-01 9.79799747e-01 -1.73133165e-01 -1.60842035e-02
4.54115242e-01 6.75700426e-01 7.04051673e-01 -5.49031794e-01
-6.25122070e-01 -7.53360868e-01 1.07005268e-01 -4.69689041e-01
-1.46288157e-01 -3.86756063e-01 -1.38040841e+00 -4.71281648e-01
1.80520862e-01 2.07421109e-01 1.48330316e-01 1.32643104e+00
2.17165455e-01 7.84035027e-01 1.97090465e-03 4.98478338e-02
-1.24236906e+00 -1.09143758e+00 4.25069109e-02 5.88752508e-01
1.24791674e-01 -5.36671400e-01 -1.50387958e-01 -1.27527550e-01] | [9.855378150939941, 7.507110118865967] |
f3d51a1d-21ec-491f-991d-eef4ef692f4c | blind-image-super-resolution-with-semantic | 2202.13142 | null | https://arxiv.org/abs/2202.13142v2 | https://arxiv.org/pdf/2202.13142v2.pdf | Real-World Blind Super-Resolution via Feature Matching with Implicit High-Resolution Priors | A key challenge of real-world image super-resolution (SR) is to recover the missing details in low-resolution (LR) images with complex unknown degradations (e.g., downsampling, noise and compression). Most previous works restore such missing details in the image space. To cope with the high diversity of natural images, they either rely on the unstable GANs that are difficult to train and prone to artifacts, or resort to explicit references from high-resolution (HR) images that are usually unavailable. In this work, we propose Feature Matching SR (FeMaSR), which restores realistic HR images in a much more compact feature space. Unlike image-space methods, our FeMaSR restores HR images by matching distorted LR image {\it features} to their distortion-free HR counterparts in our pretrained HR priors, and decoding the matched features to obtain realistic HR images. Specifically, our HR priors contain a discrete feature codebook and its associated decoder, which are pretrained on HR images with a Vector Quantized Generative Adversarial Network (VQGAN). Notably, we incorporate a novel semantic regularization in VQGAN to improve the quality of reconstructed images. For the feature matching, we first extract LR features with an LR encoder consisting of several Swin Transformer blocks and then follow a simple nearest neighbour strategy to match them with the pretrained codebook. In particular, we equip the LR encoder with residual shortcut connections to the decoder, which is critical to the optimization of feature matching loss and also helps to complement the possible feature matching errors. Experimental results show that our approach produces more realistic HR images than previous methods. Codes are released at \url{https://github.com/chaofengc/FeMaSR}. | ['Shihui Guo', 'Tao Yang', 'Xiaoguang Han', 'Xiaoming Li', 'Yipeng Qin', 'Xinyu Shi', 'Chaofeng Chen'] | 2022-02-26 | null | null | null | null | ['texture-synthesis'] | ['computer-vision'] | [ 6.24353230e-01 7.60861486e-02 -7.93312211e-03 -9.58561823e-02
-1.23169708e+00 -3.89221191e-01 2.45042086e-01 -7.90126979e-01
8.74590501e-02 7.92101502e-01 4.91418481e-01 8.33500549e-02
-6.32289797e-02 -9.81796384e-01 -9.28997695e-01 -8.08090150e-01
4.32121933e-01 -8.65650624e-02 -1.74948588e-01 -4.36212957e-01
8.03833008e-02 5.33516049e-01 -1.63752329e+00 1.86304867e-01
9.87972856e-01 9.48286295e-01 4.44916695e-01 6.20451093e-01
2.98448205e-01 7.62491465e-01 -4.04795259e-01 -3.38828832e-01
6.26319885e-01 -6.81135237e-01 -7.24460423e-01 2.01196864e-01
4.15214360e-01 -5.98097563e-01 -8.51108730e-01 1.25974357e+00
4.44682300e-01 2.32286900e-01 2.07046896e-01 -8.66001308e-01
-1.02912295e+00 3.30802947e-01 -6.37827992e-01 8.28868598e-02
4.87529695e-01 2.45096147e-01 7.80820549e-01 -9.83526647e-01
7.13373482e-01 1.23517132e+00 5.05657554e-01 7.54234731e-01
-1.28104925e+00 -7.09451258e-01 -3.27402771e-01 1.66515291e-01
-1.62963700e+00 -8.27408731e-01 9.59127963e-01 -2.32407637e-02
4.18030709e-01 4.78196353e-01 2.91320801e-01 1.22540307e+00
1.57192256e-02 2.26348788e-01 1.20362318e+00 -2.68557698e-01
-4.87044007e-02 6.20251447e-02 -4.87735599e-01 4.45872784e-01
-1.26259163e-01 4.15587276e-01 -4.07825679e-01 1.16427556e-01
1.24955034e+00 2.96535999e-01 -8.25786948e-01 -2.76159607e-02
-1.10083961e+00 7.32574642e-01 7.21136212e-01 3.32295209e-01
-3.76472652e-01 -3.46895643e-02 -1.88356638e-01 3.06778818e-01
2.39865661e-01 2.90149033e-01 -7.16330633e-02 3.40590715e-01
-8.11861992e-01 -9.80874430e-03 1.80184856e-01 1.01535439e+00
1.02006471e+00 2.57894784e-01 -1.63731456e-01 1.10577166e+00
-8.21098983e-02 5.58787823e-01 4.97797877e-01 -1.30915916e+00
5.41235387e-01 2.24829651e-02 1.28030032e-01 -1.08242309e+00
1.50025249e-01 -5.51943779e-01 -1.42596626e+00 1.72758922e-01
-2.27869246e-02 2.81092852e-01 -9.50303435e-01 1.76181841e+00
1.60795778e-01 2.83796608e-01 3.80970985e-01 1.16892743e+00
7.99348116e-01 7.85609365e-01 -3.41195583e-01 -1.94163635e-01
1.03937232e+00 -8.22387755e-01 -6.76358879e-01 -1.54100254e-01
-3.35946195e-02 -8.34474385e-01 1.17686224e+00 2.23427176e-01
-1.20848143e+00 -9.43564653e-01 -1.09596813e+00 -4.51839387e-01
3.04669403e-02 5.01894206e-02 1.42190337e-01 2.65116870e-01
-1.14670885e+00 7.31182277e-01 -4.85540062e-01 -7.17452914e-03
5.27939618e-01 1.52008370e-01 -5.46776414e-01 -6.92494214e-01
-1.30110765e+00 6.44825220e-01 1.95478469e-01 2.76358306e-01
-9.82185543e-01 -7.08890021e-01 -1.20164657e+00 -2.29702573e-02
3.14616263e-01 -7.93178320e-01 7.09638000e-01 -9.96773124e-01
-1.61630821e+00 6.56343460e-01 -1.13900952e-01 -2.62325913e-01
5.04245281e-01 -2.72940323e-02 -5.76267660e-01 3.39956939e-01
9.32950601e-02 5.04924536e-01 1.32335246e+00 -1.52999890e+00
-3.69716287e-01 -2.18194991e-01 1.56523157e-02 3.17495078e-01
-1.26364706e-02 -1.75483346e-01 -4.20354784e-01 -8.60513508e-01
2.73283780e-01 -6.01367295e-01 -2.49407738e-01 -1.27498761e-01
-5.22705019e-01 5.41082382e-01 8.07868242e-01 -1.09408140e+00
8.99411678e-01 -2.35823083e+00 3.34161848e-01 6.98144138e-02
3.04194719e-01 9.27057490e-02 -3.38695616e-01 1.97955772e-01
-1.78780764e-01 1.76605090e-01 -5.72874546e-01 -3.95578355e-01
-2.93806434e-01 2.92630672e-01 -6.14872754e-01 5.90312243e-01
2.13420659e-01 9.23537910e-01 -6.98207855e-01 -3.57157618e-01
4.59156036e-01 1.13150239e+00 -3.70556772e-01 4.28330272e-01
1.14891842e-01 9.46788371e-01 -3.84365201e-01 5.70277095e-01
9.51719999e-01 -1.96689144e-01 -8.63350034e-02 -5.08071542e-01
5.15396940e-03 -6.77788304e-03 -1.09960663e+00 1.74811625e+00
-6.86301291e-01 3.24574739e-01 5.39878421e-02 -8.48374665e-01
1.04952252e+00 1.39572904e-01 3.75466019e-01 -9.47913408e-01
-1.64287880e-01 1.61713853e-01 -4.12793994e-01 -1.54475108e-01
4.47726190e-01 -2.33740121e-01 9.88924876e-03 3.53587233e-03
4.35906537e-02 -3.43651593e-01 -3.28162372e-01 1.54765293e-01
9.37315881e-01 2.05776542e-01 2.14026004e-01 2.56481498e-01
8.18168700e-01 -4.82970804e-01 8.40145528e-01 4.60321784e-01
2.62106389e-01 1.41229200e+00 1.02390954e-02 -1.87593207e-01
-1.48250914e+00 -1.11396289e+00 -2.66194344e-01 3.79400045e-01
2.96125650e-01 -1.36683628e-01 -6.20650172e-01 -3.79172921e-01
-4.05309558e-01 5.63683689e-01 -4.66763437e-01 -3.39021355e-01
-6.48615837e-01 -4.46994781e-01 3.06777358e-01 2.36152589e-01
8.66088927e-01 -1.08237374e+00 -2.62089938e-01 1.31485865e-01
-4.47985411e-01 -1.32284236e+00 -6.10876799e-01 -2.09266230e-01
-7.03751922e-01 -9.75906909e-01 -8.63018572e-01 -6.11577272e-01
7.76826084e-01 4.10830110e-01 9.42622721e-01 3.27027857e-01
-4.63083297e-01 1.22573100e-01 -4.68971431e-01 4.00455683e-01
-5.40122569e-01 -3.21887791e-01 -1.54823095e-01 3.57424498e-01
-3.57062012e-01 -7.56164491e-01 -8.44236255e-01 2.70542800e-01
-1.22239780e+00 2.52794623e-01 7.64608979e-01 1.18779516e+00
1.01266670e+00 4.95743990e-01 5.04502594e-01 -8.08385253e-01
8.94800108e-03 -4.15726990e-01 -5.94131529e-01 1.32681681e-02
-4.89119202e-01 -2.08985191e-02 1.05394840e+00 -1.74132407e-01
-1.20179796e+00 4.57004122e-02 -4.19200093e-01 -9.33136165e-01
-1.26322329e-01 -9.39386487e-02 -5.27957916e-01 -3.59243929e-01
4.72497582e-01 7.41691172e-01 -5.63765913e-02 -5.82023740e-01
5.17096817e-01 6.19499505e-01 9.65463102e-01 -2.96638340e-01
1.46424544e+00 5.71079373e-01 -7.28344470e-02 -7.27514744e-01
-9.65433419e-01 -1.71482727e-01 -3.67760628e-01 5.69063686e-02
7.61336565e-01 -1.12501311e+00 -3.01025480e-01 5.03796935e-01
-8.26583922e-01 -3.17097783e-01 -5.36414981e-01 3.47956747e-01
-8.56419206e-01 4.52235907e-01 -8.37353230e-01 -4.59567189e-01
-3.92367691e-01 -1.10984468e+00 1.14404464e+00 4.20582622e-01
3.83857399e-01 -6.75229609e-01 -2.84126043e-01 6.15036607e-01
6.69391572e-01 4.10886168e-01 6.83588147e-01 1.25641376e-01
-8.19254696e-01 5.50854839e-02 -3.69552016e-01 6.83639169e-01
3.56355786e-01 -4.48590577e-01 -9.86132801e-01 -5.71835518e-01
2.74801046e-01 -2.88541287e-01 8.66168439e-01 2.16086343e-01
1.54949427e+00 -5.51271617e-01 -6.11996204e-02 1.24853146e+00
1.81363785e+00 -4.83887084e-02 1.19637918e+00 2.41250694e-01
9.01534677e-01 4.04591858e-01 6.83372498e-01 3.28315109e-01
3.07476759e-01 7.69878507e-01 4.54232991e-01 -4.62281942e-01
-5.98144293e-01 -5.06602049e-01 4.04708743e-01 6.69967711e-01
-2.32700855e-01 -9.14875045e-02 -2.22201377e-01 3.67416859e-01
-1.35874510e+00 -1.06871808e+00 2.56159395e-01 2.33164692e+00
1.03861392e+00 -3.08225632e-01 -3.65595818e-01 1.95823908e-01
8.65599215e-01 3.97217572e-01 -7.08244741e-01 8.91895145e-02
-4.99824286e-01 4.09856081e-01 5.63334286e-01 6.19090319e-01
-8.40671122e-01 9.25719976e-01 4.80411339e+00 8.82453680e-01
-1.05604136e+00 2.32063457e-01 7.63761640e-01 5.22655882e-02
-5.68241239e-01 2.76920833e-02 -5.83413064e-01 5.31278968e-01
6.83675349e-01 -5.51682040e-02 1.03351903e+00 4.99166429e-01
1.00769512e-01 2.80518562e-01 -7.52816141e-01 1.21121562e+00
1.54164910e-01 -1.31682265e+00 1.60435751e-01 6.39580935e-02
8.37795615e-01 -2.33361632e-01 3.24459046e-01 1.15261450e-01
2.15228155e-01 -1.32167447e+00 5.35550654e-01 7.46495843e-01
1.52657175e+00 -9.23222184e-01 6.09941602e-01 1.16097115e-01
-1.13195050e+00 -1.54465437e-01 -7.44087100e-01 5.30239165e-01
1.71144783e-01 6.30430937e-01 -1.33569866e-01 9.59415257e-01
9.92490947e-01 9.62699890e-01 -3.41404557e-01 5.29966831e-01
-2.86938310e-01 2.49945208e-01 3.95949632e-02 1.14188671e+00
-2.64884681e-01 -4.15824056e-01 6.71395540e-01 6.90473914e-01
5.76073527e-01 4.49039817e-01 -3.07041556e-02 1.11317992e+00
-3.24095935e-01 -2.12881014e-01 -6.66640162e-01 3.00058633e-01
4.95506346e-01 1.30867445e+00 -2.11437792e-01 -1.17729783e-01
-4.16273117e-01 1.45527947e+00 1.53042823e-01 6.12236559e-01
-7.62411416e-01 -3.92655224e-01 6.94454968e-01 1.53017953e-01
4.18212324e-01 9.17216614e-02 3.62056755e-02 -1.39980364e+00
1.73234180e-01 -1.16874397e+00 2.36872554e-01 -9.43367362e-01
-1.29768455e+00 8.98575068e-01 -3.73362899e-01 -1.54368508e+00
-1.87728435e-01 1.34259433e-01 -3.43692571e-01 1.12872136e+00
-2.01271129e+00 -1.24886549e+00 -6.31714404e-01 1.03956771e+00
5.90374172e-01 1.33262640e-02 6.22534215e-01 3.61037225e-01
-5.06598115e-01 6.73408091e-01 7.22760260e-02 7.61556700e-02
6.40189528e-01 -8.86398077e-01 3.01379859e-01 1.05726278e+00
-1.02307394e-01 4.09159869e-01 6.16747022e-01 -5.51191211e-01
-1.52505744e+00 -1.41819322e+00 3.64261985e-01 -1.00738525e-01
1.94124922e-01 -2.54566610e-01 -1.15270805e+00 7.61039972e-01
-5.33369072e-02 5.26400328e-01 2.15837955e-01 -7.97867060e-01
-4.48286802e-01 -1.91226825e-01 -1.49054348e+00 3.93084675e-01
1.24641240e+00 -7.88343847e-01 -3.90352130e-01 1.27071306e-01
1.05561352e+00 -5.57662368e-01 -1.09121084e+00 4.59528893e-01
2.55706757e-01 -1.05590963e+00 1.38565373e+00 -8.39396864e-02
6.65536463e-01 -5.89863241e-01 -4.77150202e-01 -1.31270349e+00
-3.27381909e-01 -6.65262282e-01 -1.92634985e-02 1.34279633e+00
5.85765429e-02 -7.68329740e-01 4.56368208e-01 2.28324905e-01
-1.86882526e-01 -4.99507517e-01 -7.63857007e-01 -7.41152108e-01
-9.44574475e-02 6.01250753e-02 8.47403765e-01 1.00411427e+00
-7.21335411e-01 -9.21588466e-02 -7.99844980e-01 5.26297629e-01
1.09627569e+00 2.70354390e-01 5.73459446e-01 -8.10084939e-01
-4.06521261e-01 -1.41524496e-02 -3.01221460e-01 -8.33774924e-01
1.70326829e-01 -7.43133008e-01 1.57510817e-01 -1.32134008e+00
2.10724428e-01 -5.33994973e-01 -2.91581124e-01 2.91634977e-01
-1.25863999e-01 7.86652744e-01 1.77749723e-01 4.06306356e-01
-2.57236928e-01 8.59070480e-01 1.51815307e+00 1.81175452e-02
-8.26748088e-02 -1.94398463e-01 -7.93409705e-01 4.21327710e-01
7.36260116e-01 -4.58531380e-01 -2.71987140e-01 -3.17891449e-01
-8.72242674e-02 5.39185345e-01 7.26935685e-01 -1.00530970e+00
3.08967661e-02 -7.77951702e-02 6.89715803e-01 -2.75308937e-01
5.63572943e-01 -7.10859895e-01 5.70022166e-01 1.27053067e-01
-2.32547298e-01 -3.40228200e-01 -1.47718593e-01 5.16210794e-01
-5.08433998e-01 9.21265781e-02 1.18960285e+00 -2.46450141e-01
-4.98048246e-01 5.99757612e-01 2.38809958e-01 -9.03718695e-02
7.68778443e-01 -2.39538968e-01 -4.02199507e-01 -5.10730565e-01
-6.29819393e-01 -8.30052644e-02 9.66312647e-01 4.02249485e-01
1.11296523e+00 -1.37435281e+00 -8.56948733e-01 5.00288963e-01
-5.65375201e-02 3.22448462e-01 7.65929759e-01 5.12457907e-01
-3.29956651e-01 7.39389157e-04 -4.08368766e-01 -2.52411962e-01
-9.40175176e-01 7.41307735e-01 4.98307556e-01 -3.60976756e-02
-1.20532560e+00 4.53776509e-01 3.94841045e-01 -2.97306627e-01
-6.51508868e-02 1.59923226e-01 -2.81199217e-01 -4.77119476e-01
7.10361183e-01 1.91183850e-01 -1.01413406e-01 -9.47059929e-01
-1.96066126e-02 9.88772094e-01 8.22343603e-02 -4.04766994e-04
1.51139295e+00 -6.46902740e-01 -6.52234405e-02 -1.11592993e-01
1.34684372e+00 2.11896658e-01 -1.53214562e+00 -5.57862699e-01
-5.36492229e-01 -9.58965480e-01 2.77808666e-01 -4.22205418e-01
-1.57599950e+00 5.51984727e-01 6.46626055e-01 -1.47182480e-01
1.65722370e+00 8.03870559e-02 1.11174619e+00 -2.05606282e-01
4.59004104e-01 -6.35101318e-01 4.31779176e-02 -4.80689050e-06
1.22361922e+00 -1.20047510e+00 -7.68092796e-02 -5.17454207e-01
-5.25525212e-01 1.03551757e+00 3.80293399e-01 -2.69304574e-01
2.31985614e-01 4.90603335e-02 1.53516332e-04 8.11577141e-02
-4.41415250e-01 -1.99195206e-01 1.56461656e-01 7.12242782e-01
-8.49889666e-02 -1.55298606e-01 1.66777503e-02 4.92390841e-01
-4.63988781e-01 -2.04993654e-02 7.55378842e-01 4.12004560e-01
-6.17903136e-02 -9.99246657e-01 -4.98351365e-01 2.28688762e-01
-5.74357629e-01 -2.87312716e-01 1.99293897e-01 5.11387646e-01
9.05933008e-02 9.86018896e-01 -1.21209212e-01 -4.87060279e-01
3.33076000e-01 -4.44518954e-01 3.43413800e-01 -4.50204909e-01
-1.23730876e-01 9.17869993e-03 -3.65812182e-01 -9.55099523e-01
-2.25128159e-01 -5.18796563e-01 -1.13913989e+00 -3.12539697e-01
-1.36918277e-02 -3.55282165e-02 4.87295419e-01 5.64299881e-01
3.74707580e-01 5.58677197e-01 1.18760669e+00 -8.76815259e-01
-4.98899251e-01 -6.76362872e-01 -6.86413229e-01 6.39440835e-01
7.57217467e-01 -4.03915524e-01 -7.30261087e-01 1.43396407e-01] | [11.07148265838623, -2.006082534790039] |
5d7e702d-07db-4391-9cb8-248daba71d64 | percol0-un-systeme-multimodal-de-detection-de | null | null | https://aclanthology.org/F12-1070 | https://aclanthology.org/F12-1070.pdf | Percol0 - un syst\`eme multimodal de d\'etection de personnes dans des documents vid\'eo (Percol0 - A multimodal person detection system in video documents) [in French] | null | ['Stephane Ayache', 'Remi Auguste', 'Frederic Bechet', 'Delphine Charlet', 'Corinne Fredouille', 'Christophe Levy', 'Georges Linares', 'Benoit Favre', 'Jean Martinet', 'Geraldine Damnati'] | 2012-06-01 | percol0-un-systeme-multimodal-de-detection-de-1 | https://aclanthology.org/F12-1070 | https://aclanthology.org/F12-1070.pdf | jeptalnrecital-2012-6 | ['person-recognition'] | ['computer-vision'] | [-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01
-8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01
-2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00
-3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01
-9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01
7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01
7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00
-1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01
3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01
9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01
5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00
-5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01
1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00
7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01
-1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02
-1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01
1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00
1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01
3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01
8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01
9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01
-9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01
-1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01
-2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01
-8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00
1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01
8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00
-6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01
-6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01
3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01
4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02
4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01
1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01
9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01
-1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01
-7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00
-1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02
1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01
-1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01
-1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01
-4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01
-1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01
6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01
-1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00
-7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01
-1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01
3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01
-1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01
1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01
3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00
-2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01
2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01
-1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01
4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01
2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00
1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01
-3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00
7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01
3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01
-2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01
7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01
-2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01
6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00
7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00
-3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01
-6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00
3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01
1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01
-5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01
-5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01
-2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01
-1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01
5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01
-2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01
-4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01
5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01
-6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00
1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01
8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02
-6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01
-1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01
1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01
8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03
3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01
8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01
-1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01
6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01
-9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01
-1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00
-5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02
-5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01
1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01
-4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01
1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01
6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01
5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01
1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01
1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01
1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01
1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00
-8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01
-1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01
6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01
4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00
-1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00
-7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00
2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01
-4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02
1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01
3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01
9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01
7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01
6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01
1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01
-1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00
3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01
9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01
-4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01
4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00
2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01
5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01
2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01
-2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01
-6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01
-1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01
-5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01
-1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00
-6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01
-1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01
1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02
1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01
5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01
-4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01
-1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01
6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01
7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01
1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01
1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01
8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01
8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00
-4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01
-5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02
5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01
-2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00
-3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01
6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01
5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00
3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01
-8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00
-8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01
3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01
-3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05
-1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01
8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02
6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01
5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00
1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01
2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02
-1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01
-4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02
9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01
-8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02
7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01
-3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01
-1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01
-1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01
-2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01
1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03
6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01
-4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01
9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01
8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01
1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01
-1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01
8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01
5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01
9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01
-4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01
1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01
-9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00
5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01
-7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00
-6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01
-5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01
-5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01
4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01
-9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00
-8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00
-5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01
-2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00
-3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02
2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01] | [-7.214419364929199, 3.7613120079040527] |
329d6766-412b-4f75-8433-ce2894e8086c | few-shot-text-classification-with | 1908.06039 | null | https://arxiv.org/abs/1908.06039v3 | https://arxiv.org/pdf/1908.06039v3.pdf | Few-shot Text Classification with Distributional Signatures | In this paper, we explore meta-learning for few-shot text classification. Meta-learning has shown strong performance in computer vision, where low-level patterns are transferable across learning tasks. However, directly applying this approach to text is challenging--lexical features highly informative for one task may be insignificant for another. Thus, rather than learning solely from words, our model also leverages their distributional signatures, which encode pertinent word occurrence patterns. Our model is trained within a meta-learning framework to map these signatures into attention scores, which are then used to weight the lexical representations of words. We demonstrate that our model consistently outperforms prototypical networks learned on lexical knowledge (Snell et al., 2017) in both few-shot text classification and relation classification by a significant margin across six benchmark datasets (20.0% on average in 1-shot classification). | ['Menghua Wu', 'Shiyu Chang', 'Yujia Bao', 'Regina Barzilay'] | 2019-08-16 | null | https://openreview.net/forum?id=H1emfT4twB | https://openreview.net/pdf?id=H1emfT4twB | iclr-2020-1 | ['few-shot-text-classification'] | ['natural-language-processing'] | [ 5.23990512e-01 3.49304937e-02 -6.87763989e-01 -3.64668429e-01
-7.32367992e-01 -4.75064665e-02 1.06308305e+00 6.77875161e-01
-6.61796689e-01 6.63488269e-01 5.72369397e-01 -2.14273527e-01
3.64709347e-02 -7.98403263e-01 -4.19673711e-01 -3.82665485e-01
3.04581374e-01 2.55200535e-01 2.69129481e-02 -4.69238460e-01
4.06403780e-01 -2.30427518e-01 -1.75989234e+00 6.35548890e-01
5.29209614e-01 6.53068066e-01 -1.39862094e-02 6.50551915e-01
-5.91475606e-01 1.16887116e+00 -4.26786274e-01 -4.17202443e-01
-2.20419422e-01 -4.56292659e-01 -9.69866574e-01 -2.50974119e-01
7.55501866e-01 -7.34046772e-02 -5.21454215e-01 1.02948165e+00
4.25728321e-01 6.36851609e-01 1.12357461e+00 -8.95984173e-01
-1.00186431e+00 1.14700854e+00 -6.42579973e-01 4.31382567e-01
1.66248783e-01 5.11412211e-02 1.83689022e+00 -1.25517905e+00
5.62093198e-01 1.16541696e+00 7.40478516e-01 5.97141445e-01
-1.21614861e+00 -6.35844529e-01 2.88113147e-01 4.39608246e-01
-9.90477800e-01 -6.16392255e-01 5.09630382e-01 -5.52878141e-01
1.39521396e+00 -4.44638878e-02 4.74424213e-01 1.50613976e+00
4.65263665e-01 8.53926301e-01 8.59400332e-01 -7.59013891e-01
4.17740464e-01 -6.64987490e-02 9.17958736e-01 6.00838721e-01
3.99108201e-01 -2.45454043e-01 -8.69781911e-01 -1.61697373e-01
-8.97247419e-02 2.89639801e-01 3.27217276e-03 -8.80695283e-02
-1.03730202e+00 1.06035054e+00 2.26262689e-01 3.55655491e-01
-3.83115739e-01 1.61876231e-01 7.07084894e-01 2.83948511e-01
9.42770422e-01 5.02090037e-01 -5.41253030e-01 -3.00607920e-01
-7.65445411e-01 1.00796208e-01 8.47252965e-01 8.69048297e-01
7.80357420e-01 -1.57020763e-01 -6.46234393e-01 1.13739157e+00
3.18089202e-02 2.08599731e-01 9.33624446e-01 -3.98686737e-01
3.84082586e-01 5.87566435e-01 -3.25571805e-01 -8.59740257e-01
-4.30267513e-01 -2.26010382e-01 -7.07112610e-01 -1.30544096e-01
-6.05845870e-03 -3.06011289e-01 -1.08750474e+00 1.60671043e+00
4.47003804e-02 2.84267545e-01 9.80074629e-02 3.29605609e-01
1.14379525e+00 7.53885210e-01 4.90599036e-01 -3.13253284e-01
1.50778735e+00 -1.08002055e+00 -8.10761809e-01 -5.44299185e-01
9.50157583e-01 -4.68635440e-01 1.26849985e+00 2.60653086e-02
-8.30509424e-01 -3.57222319e-01 -1.07147527e+00 -2.91259557e-01
-5.92721701e-01 -5.01125276e-01 5.14202893e-01 4.07057077e-01
-6.65538609e-01 7.33311176e-01 -4.86731052e-01 -7.61325717e-01
7.39046693e-01 -6.73238114e-02 -4.06590365e-02 -1.48624778e-02
-1.41573048e+00 1.11926305e+00 5.75508058e-01 -7.97201872e-01
-7.32228935e-01 -1.00316906e+00 -8.98621619e-01 3.97226453e-01
3.88813764e-01 -7.03924775e-01 1.36986256e+00 -7.33676910e-01
-1.10354459e+00 1.02022195e+00 -3.42907220e-01 -7.27057338e-01
1.72621220e-01 -1.44643128e-01 -4.01879102e-01 -1.03562273e-01
1.56981975e-01 5.04887581e-01 1.05253088e+00 -8.39891434e-01
-9.00445819e-01 -2.30563492e-01 -6.99737892e-02 1.31203383e-01
-9.50631082e-01 -3.97165008e-02 -5.83195388e-02 -6.75468445e-01
-4.42060918e-01 -5.39959669e-01 6.87809065e-02 -1.96364015e-01
-2.78626949e-01 -5.64849198e-01 8.87720585e-01 -2.47030392e-01
1.32324994e+00 -1.81618416e+00 1.63223580e-01 -2.12326452e-01
3.99410009e-01 2.36531109e-01 -1.74441993e-01 5.21984577e-01
-1.12936027e-01 5.84658831e-02 -1.06434450e-01 -4.06494617e-01
1.34620760e-02 1.16108879e-02 -5.73074579e-01 2.44690746e-01
3.49910818e-02 1.38025856e+00 -1.11715901e+00 -5.39649963e-01
2.20862612e-01 2.72129118e-01 -2.80253202e-01 -1.10193186e-01
-3.82389903e-01 -4.06920969e-01 -2.25258976e-01 4.39481616e-01
-2.13355795e-02 -4.15701151e-01 1.79797187e-01 -2.25175247e-01
2.00490788e-01 3.89759779e-01 -4.49127555e-01 1.62201822e+00
-7.30467200e-01 9.65282619e-01 -5.00106812e-01 -1.18454635e+00
6.58222079e-01 2.54076451e-01 3.88341695e-01 -5.09641290e-01
4.21190977e-01 -2.48173863e-01 1.04379235e-02 -4.95097458e-01
7.22967386e-01 -3.27648520e-01 -1.16894573e-01 8.80855262e-01
5.37769318e-01 2.24780127e-01 3.11835051e-01 4.14694935e-01
1.26121962e+00 -2.21571237e-01 8.63627434e-01 -3.43583405e-01
2.13327631e-01 2.30431799e-02 7.96214193e-02 9.83509302e-01
-2.37062708e-01 2.28114679e-01 3.61426115e-01 -3.87809455e-01
-1.00723720e+00 -7.70720422e-01 -2.38959506e-01 1.95920658e+00
-1.64492145e-01 -6.94084406e-01 -3.79258811e-01 -9.21923161e-01
2.74540544e-01 1.16333151e+00 -1.18443227e+00 -6.56234801e-01
-1.91312239e-01 -8.12408268e-01 3.58856052e-01 6.25691950e-01
-6.03561429e-03 -1.14293516e+00 -5.85365415e-01 2.58724838e-01
1.71691492e-01 -8.73863757e-01 -6.22243524e-01 6.38298571e-01
-7.58727193e-01 -9.37781572e-01 -5.72649360e-01 -8.21620107e-01
3.76513302e-01 6.73893154e-01 1.26714730e+00 1.27314646e-02
-6.39229715e-01 6.44078851e-01 -6.37843013e-01 -6.56589210e-01
-3.29653651e-01 3.85561764e-01 -2.82547041e-03 -7.54437596e-02
1.00896311e+00 -5.38169444e-01 -2.30281234e-01 -2.07580045e-01
-5.71364522e-01 -9.91363302e-02 3.30399603e-01 1.38003218e+00
2.44592652e-01 -2.98936069e-01 7.86702931e-01 -1.41391158e+00
8.18536639e-01 -8.95295799e-01 7.10372031e-02 3.29970211e-01
-9.12912309e-01 -4.68343049e-02 5.97298563e-01 -5.04542947e-01
-9.64404881e-01 -3.64908844e-01 2.05387652e-01 -3.25234741e-01
-1.00982107e-01 6.05805397e-01 1.63948506e-01 3.65530074e-01
1.04606962e+00 2.26647139e-01 -1.63799629e-01 -1.48020074e-01
6.79750800e-01 9.58587825e-01 1.28895953e-01 -5.14813244e-01
5.78932106e-01 5.64342856e-01 -1.46032512e-01 -1.10153615e+00
-1.79431355e+00 -7.38773942e-01 -7.15338647e-01 1.32614821e-01
8.43571901e-01 -9.06842351e-01 -2.84855098e-01 6.54301196e-02
-9.92591321e-01 -2.93898433e-01 -5.17201066e-01 2.74145603e-01
-4.67579812e-01 6.97369277e-02 -6.17315769e-01 -6.00773871e-01
-5.82419574e-01 -5.42784035e-01 9.09340560e-01 5.14520071e-02
-7.15576231e-01 -1.52920139e+00 2.28393167e-01 5.28757833e-02
2.76025772e-01 -1.50447950e-01 1.20220625e+00 -1.13569152e+00
2.50138670e-01 -1.27321035e-01 -1.69065133e-01 -5.53601682e-02
5.11943579e-01 -1.77047059e-01 -1.41751802e+00 -3.37020218e-01
-2.44149372e-01 -7.12025106e-01 1.51226079e+00 4.51006174e-01
1.09854615e+00 -2.61878461e-01 -4.79926109e-01 3.73848349e-01
1.14874852e+00 -6.91931769e-02 1.78075671e-01 3.31169933e-01
8.35211277e-01 6.49249256e-01 5.21628320e-01 5.14323235e-01
2.49176234e-01 4.46822852e-01 4.37815674e-02 2.98621923e-01
-2.49636799e-01 -3.06169748e-01 3.09620529e-01 6.87500417e-01
2.26555884e-01 -2.67489374e-01 -1.00024366e+00 5.72563112e-01
-2.17193437e+00 -1.39948583e+00 8.81645754e-02 1.74227202e+00
1.07645226e+00 3.39948654e-01 2.44185459e-02 -1.82866499e-01
7.82679677e-01 5.31124473e-01 -6.82918072e-01 -3.72581989e-01
-1.79547027e-01 4.51946110e-01 1.17713496e-01 4.05589461e-01
-1.34297347e+00 1.09226263e+00 6.55536890e+00 8.23060095e-01
-9.35487747e-01 4.03735757e-01 4.89864558e-01 -5.46817183e-01
-3.00424814e-01 -1.06026277e-01 -9.79109228e-01 3.00977230e-01
1.13995290e+00 -8.64517391e-01 1.72983319e-03 9.34353769e-01
-3.58970016e-01 2.20801815e-01 -1.40400028e+00 8.30024660e-01
5.95447838e-01 -1.61747193e+00 4.33514953e-01 -9.95254293e-02
1.09441781e+00 1.36613786e-01 1.67864129e-01 8.94788682e-01
5.50775886e-01 -1.15208495e+00 3.86050135e-01 2.99182653e-01
5.98023653e-01 -8.94502699e-01 4.86088306e-01 3.20779026e-01
-9.63978589e-01 -9.96487811e-02 -7.44473755e-01 -1.60450891e-01
-8.47927034e-02 4.93340999e-01 -8.70217204e-01 1.24181740e-01
3.93467873e-01 1.23436737e+00 -6.62369967e-01 4.20941889e-01
-1.75728068e-01 7.22591102e-01 3.61953944e-01 -4.55496579e-01
2.58490413e-01 4.18025643e-01 3.50690722e-01 1.59532416e+00
1.95615571e-02 8.64009336e-02 2.49811634e-01 6.15955889e-01
-5.67214489e-01 4.81417090e-01 -7.95265317e-01 -2.32149422e-01
4.66175348e-01 1.47385216e+00 -5.21135807e-01 -7.04885781e-01
-7.62146533e-01 8.10035169e-01 6.09444559e-01 2.14279264e-01
-4.47082520e-01 -6.82369947e-01 7.72170603e-01 -1.95775822e-01
3.82194430e-01 2.10217208e-01 -5.40327191e-01 -1.34647059e+00
-5.17393827e-01 -5.97636342e-01 6.98590279e-01 -4.74379599e-01
-1.82653844e+00 3.49152237e-01 -1.53130814e-02 -9.92494822e-01
-4.72330034e-01 -8.48255455e-01 -8.52226913e-01 7.12729037e-01
-1.44187033e+00 -1.29035735e+00 -2.33628005e-01 3.39786440e-01
1.09250581e+00 -5.53090930e-01 9.95354772e-01 -2.30372459e-01
-4.87502187e-01 6.85698450e-01 2.00781703e-01 2.40922943e-01
9.47309852e-01 -1.32284081e+00 7.39724219e-01 4.83170599e-01
5.99450946e-01 7.52498209e-01 7.02269793e-01 -7.39207625e-01
-1.37650812e+00 -1.24428093e+00 8.26641202e-01 -7.93931067e-01
1.10535741e+00 -2.92673022e-01 -1.26039982e+00 8.65363836e-01
4.83929604e-01 8.36273357e-02 1.10062039e+00 7.65052915e-01
-8.97538006e-01 2.84020454e-01 -7.20830739e-01 7.53088117e-01
1.08476901e+00 -7.89894044e-01 -1.16451085e+00 3.73432785e-01
8.60349178e-01 6.99458271e-02 -6.63337827e-01 5.65478904e-03
5.87798357e-01 -5.70880115e-01 7.99065351e-01 -1.37823081e+00
9.07763302e-01 4.02812541e-01 -3.38243753e-01 -1.67896748e+00
-5.29444873e-01 -2.48999327e-01 -7.48415649e-01 1.17380524e+00
6.03807688e-01 -3.55292708e-01 6.68838084e-01 1.97490305e-01
-2.27836724e-02 -6.98951304e-01 -7.28894949e-01 -6.88945293e-01
5.14153183e-01 -4.51809734e-01 2.22372353e-01 1.49260545e+00
6.66828454e-01 1.10316396e+00 -3.49018127e-01 -4.35680240e-01
8.40108037e-01 1.65131167e-01 4.53845382e-01 -1.57043767e+00
-2.42521062e-01 -8.06981146e-01 -2.64335513e-01 -5.85962951e-01
9.23681915e-01 -1.47975361e+00 7.17759058e-02 -1.49679410e+00
7.93202460e-01 1.83671936e-01 -6.97316945e-01 5.43882966e-01
-6.67098284e-01 1.47647947e-01 3.63003947e-02 1.59693852e-01
-6.98315620e-01 6.29064381e-01 7.37058461e-01 -6.17066383e-01
4.20874208e-02 -3.66627276e-01 -8.73924792e-01 7.73612142e-01
8.04187655e-01 -4.60136414e-01 -4.05902058e-01 -2.50396907e-01
2.14975327e-01 -4.73732948e-01 1.54802904e-01 -6.62619531e-01
4.78842080e-01 -2.77021915e-01 4.37356651e-01 -3.85713667e-01
2.56134748e-01 -3.52855802e-01 -7.53296256e-01 3.69245231e-01
-9.93573129e-01 -3.05701554e-01 7.37813562e-02 8.25386345e-01
2.11189128e-02 -4.94913697e-01 8.22699785e-01 -1.82463139e-01
-9.68572080e-01 3.12788963e-01 -3.16106856e-01 5.68726122e-01
9.81878400e-01 2.02090796e-02 -6.01155937e-01 -3.42123836e-01
-5.68550348e-01 1.52818084e-01 2.88511515e-01 7.02851772e-01
6.13684833e-01 -1.27084053e+00 -7.07844138e-01 -1.13525607e-01
7.59724915e-01 -5.23267269e-01 2.84637958e-01 5.44673145e-01
2.71909177e-01 4.56915259e-01 -6.84202760e-02 -4.34955478e-01
-1.16749024e+00 7.15831757e-01 1.20366953e-01 3.08288028e-03
-8.59089851e-01 7.74304032e-01 1.81453943e-01 -1.56186759e-01
2.34262496e-01 -8.14339668e-02 -2.97922939e-01 6.96551800e-01
9.28001821e-01 3.41875613e-01 7.25745261e-02 -4.19770092e-01
-1.88004464e-01 4.79482949e-01 -7.54937649e-01 -5.19794859e-02
1.34441876e+00 -5.77523187e-02 8.34447443e-02 1.16396034e+00
1.31564438e+00 -2.91918814e-01 -9.42732632e-01 -7.05801010e-01
4.87391233e-01 -1.96895212e-01 3.38849545e-01 -5.70717871e-01
-4.59788382e-01 1.07691467e+00 1.21009246e-01 8.88046995e-02
4.94389951e-01 1.60299689e-01 7.74924338e-01 8.77392173e-01
1.85339689e-01 -1.26346910e+00 3.85440767e-01 1.05062389e+00
3.67288888e-01 -1.43493783e+00 1.66807905e-01 1.23935893e-01
-6.40600145e-01 1.22632289e+00 6.58374488e-01 -8.73599015e-03
9.12355065e-01 2.18593314e-01 -1.76983535e-01 -2.39949092e-01
-1.57906640e+00 -4.05296803e-01 4.45637792e-01 4.87018168e-01
8.30852628e-01 -1.13878399e-01 -9.50239375e-02 5.14117122e-01
7.00666010e-02 -2.16750622e-01 5.41423202e-01 1.13116562e+00
-1.08765996e+00 -8.01330924e-01 6.27092794e-02 9.72665370e-01
-4.47942048e-01 -4.92410868e-01 -4.58819628e-01 5.22098243e-01
-3.24241906e-01 8.13141108e-01 3.93380731e-01 -4.22590822e-01
2.46668667e-01 7.39360034e-01 3.68415862e-01 -1.43485355e+00
-5.35947442e-01 -4.06808317e-01 1.14368096e-01 -1.55328751e-01
-3.12745273e-01 -6.43043756e-01 -1.08479118e+00 -2.49206215e-01
-1.34509683e-01 1.09207273e-01 3.05042952e-01 1.15654540e+00
2.57906199e-01 8.28504145e-01 4.19683069e-01 -5.16212702e-01
-8.70292544e-01 -1.27591443e+00 -4.75973189e-01 4.21871066e-01
3.02583307e-01 -7.18685448e-01 -4.00909901e-01 8.01204666e-02] | [10.685150146484375, 7.747781753540039] |
249b5d4f-0dd7-4b83-9fec-6787bcc5a641 | deep-embedding-using-bayesian-risk | 1812.02466 | null | http://arxiv.org/abs/1812.02466v1 | http://arxiv.org/pdf/1812.02466v1.pdf | Deep Embedding using Bayesian Risk Minimization with Application to Sketch Recognition | In this paper, we address the problem of hand-drawn sketch recognition.
Inspired by the Bayesian decision theory, we present a deep metric learning
loss with the objective to minimize the Bayesian risk of misclassification. We
estimate this risk for every mini-batch during training, and learn robust deep
embeddings by backpropagating it to a deep neural network in an end-to-end
trainable paradigm. Our learnt embeddings are discriminative and robust despite
of intra-class variations and inter-class similarities naturally present in
hand-drawn sketch images. Outperforming the state of the art on sketch
recognition, our method achieves 82.2% and 88.7% on TU-Berlin-250 and
TU-Berlin-160 benchmarks respectively. | ['Ajeet Kumar Singh', 'Anand Mishra'] | 2018-12-06 | null | null | null | null | ['sketch-recognition'] | ['computer-vision'] | [-1.24879695e-01 -7.06894398e-02 -2.78177857e-01 -8.25017869e-01
-9.75355506e-01 -3.53649735e-01 6.68337047e-01 -2.53409743e-01
-5.48010409e-01 5.93447983e-01 -1.27848357e-01 -1.65358171e-01
-3.17183822e-01 -7.90579379e-01 -8.27101469e-01 -1.66869104e-01
-5.03100492e-02 5.78425467e-01 -1.08455427e-01 3.78741533e-01
5.40234864e-01 9.56154764e-01 -1.14253914e+00 3.19836169e-01
5.24860799e-01 1.46584713e+00 -1.90903172e-01 8.02614510e-01
-1.88538253e-01 6.60831034e-01 -6.91729367e-01 -8.86241078e-01
3.77573818e-01 -1.92890931e-02 -5.81222236e-01 4.25546728e-02
1.18752158e+00 -7.66283512e-01 -1.02191079e+00 8.23688090e-01
5.12931883e-01 2.88725555e-01 1.19463110e+00 -1.23700655e+00
-1.01993132e+00 2.11401328e-01 -4.00888145e-01 4.02026586e-02
3.49025205e-02 -1.48380902e-02 1.09805417e+00 -1.38586795e+00
5.13981521e-01 1.31846046e+00 6.23560548e-01 6.36462271e-01
-1.23995364e+00 -8.04761410e-01 2.27180734e-01 1.02695771e-01
-1.58246613e+00 -3.47236931e-01 9.80209768e-01 -5.10190308e-01
7.67380476e-01 -1.09604634e-01 3.05639543e-02 1.43633306e+00
7.22852796e-02 1.20258796e+00 6.46328032e-01 -2.92143583e-01
2.58276492e-01 -1.20379791e-01 -2.21929953e-01 1.06369710e+00
1.09057471e-01 2.12741315e-01 -4.71206963e-01 -3.09352964e-01
1.08432126e+00 4.19430584e-01 2.09076911e-01 -4.53559935e-01
-8.91287684e-01 9.19018984e-01 3.89760226e-01 6.38808385e-02
-2.13765740e-01 7.38966942e-01 4.13375795e-01 3.16780716e-01
3.81057113e-01 1.98528931e-01 -2.28546664e-01 -1.76053628e-01
-1.24034357e+00 2.20139429e-01 8.57060194e-01 1.01585734e+00
4.90659744e-01 1.01393566e-01 -5.21980703e-01 1.22648215e+00
4.58423525e-01 5.03556907e-01 1.66027024e-01 -8.85102153e-01
6.05328262e-01 2.10273907e-01 5.67595698e-02 -1.09400308e+00
2.19860628e-01 -3.05759937e-01 -1.07135522e+00 6.20816410e-01
4.46301818e-01 1.84257954e-01 -1.11671460e+00 1.51795328e+00
-2.28487492e-01 1.07466757e-01 -6.01193190e-01 7.73543715e-01
5.02342820e-01 3.05653840e-01 -1.38919041e-01 5.11843324e-01
7.73912311e-01 -9.51388001e-01 -5.59831977e-01 1.09142531e-02
-1.64138451e-01 -8.77487481e-01 1.01250076e+00 6.28916800e-01
-1.02171373e+00 -7.99589097e-01 -1.33148777e+00 -3.68407458e-01
-4.46118414e-01 6.16938770e-01 2.87745446e-01 6.87937200e-01
-4.94169384e-01 1.15359986e+00 -7.01830626e-01 1.37379050e-01
1.19025743e+00 1.97914630e-01 -3.83478224e-01 -3.23037177e-01
-6.88940227e-01 7.32109010e-01 -1.43761456e-01 1.11183792e-01
-1.24099112e+00 -9.10651982e-01 -5.42155504e-01 1.08910218e-01
1.19624026e-01 -4.11067307e-01 1.16230690e+00 -5.28497100e-01
-1.76089501e+00 8.93384457e-01 -5.83819970e-02 -2.19494253e-01
1.19325900e+00 -3.29501808e-01 -2.38994464e-01 -2.13607736e-02
-1.50354892e-01 5.57686448e-01 1.31473112e+00 -9.86014307e-01
-4.62506562e-01 -3.42446148e-01 -1.10672228e-01 -4.97304648e-01
-2.22888634e-01 -3.22806120e-01 -3.97490948e-01 -1.13344991e+00
-6.67194054e-02 -6.61104202e-01 1.47284165e-01 9.98784065e-01
-3.44834566e-01 -6.87900364e-01 8.65264773e-01 -5.35347104e-01
9.18310761e-01 -2.09257030e+00 -4.54504825e-02 2.70231634e-01
3.32851022e-01 3.41861010e-01 -4.09484953e-01 3.57960284e-01
2.65473932e-01 -3.65656233e-05 -3.85757953e-01 -6.13951385e-01
7.28097916e-01 3.21827084e-01 -8.19396019e-01 6.18443191e-01
4.89327997e-01 7.88934708e-01 -9.41100657e-01 -1.55804247e-01
4.20152456e-01 4.23477441e-01 -3.61161470e-01 5.49255013e-01
-2.68252164e-01 -4.39345479e-01 -2.68642396e-01 7.12348998e-01
1.03811610e+00 -3.17451209e-02 1.50122773e-03 -5.46017215e-02
4.80468124e-01 3.24983686e-01 -1.18300045e+00 1.86331367e+00
-7.90712357e-01 1.00442064e+00 -2.65421927e-01 -9.40064490e-01
1.39805079e+00 -9.39213485e-02 -1.44922584e-01 -5.55451870e-01
-2.05774546e-01 2.47518629e-01 -4.78096277e-01 -1.86394915e-01
3.74719471e-01 -2.18258500e-01 4.33240049e-02 5.06472945e-01
4.86872882e-01 -1.35244071e-01 -2.13988632e-01 2.00541526e-01
1.15447855e+00 3.51875573e-02 -1.49555936e-01 -7.85561576e-02
4.21288848e-01 -9.30584252e-01 1.61716610e-01 1.04583752e+00
-1.21624872e-01 1.03797317e+00 7.00948477e-01 -8.01014066e-01
-1.16713572e+00 -1.64942193e+00 -2.26653636e-01 1.03036165e+00
-2.41156474e-01 -2.86658198e-01 7.14659551e-03 -1.29426348e+00
8.49298239e-01 5.66546202e-01 -1.00520527e+00 -1.40016720e-01
-3.43265533e-01 6.78650383e-03 8.44041228e-01 9.37494576e-01
3.78453583e-01 -8.01816523e-01 -1.67405829e-01 3.00893158e-01
5.33241630e-01 -8.51644397e-01 -6.23191833e-01 2.48035118e-02
-6.88534915e-01 -1.01677024e+00 -9.85529482e-01 -7.00940192e-01
6.90527320e-01 -8.31136927e-02 1.19507813e+00 -2.36406717e-02
-8.41378450e-01 4.61432487e-01 1.21528469e-01 -3.05796266e-01
-1.19338848e-01 4.47887369e-02 -1.75701901e-01 1.00873984e-01
2.58066654e-01 -4.85167831e-01 -7.19736934e-01 4.35662568e-01
-6.74164236e-01 -4.66255754e-01 5.72358429e-01 1.22952247e+00
5.42945564e-01 -4.56356019e-01 4.91511554e-01 -5.99689782e-01
6.97830737e-01 -1.42405912e-01 -6.55525565e-01 5.12643576e-01
-5.50522864e-01 4.07658577e-01 6.84018493e-01 -4.59250003e-01
-8.37354779e-01 -5.76993711e-02 -1.09404452e-01 -9.37777996e-01
-1.63574964e-01 9.32073966e-03 1.50038237e-02 -2.01309696e-01
4.17812586e-01 1.03155181e-01 -1.34271115e-01 -7.04110622e-01
5.44148326e-01 7.68152237e-01 5.75771511e-01 -1.10606587e+00
7.80140042e-01 5.01236260e-01 8.17626566e-02 -7.07376778e-01
-8.64218175e-01 1.42676393e-02 -6.09720588e-01 -5.43538295e-02
4.32397902e-01 -6.32215619e-01 -5.92319846e-01 3.63730192e-01
-1.34395683e+00 -3.81227732e-01 -3.68176758e-01 1.92131296e-01
-6.02500081e-01 5.10942876e-01 -4.29109216e-01 -1.01825237e+00
-3.57758760e-01 -9.37049508e-01 1.21887922e+00 -8.94136168e-03
-1.33270666e-01 -8.24211478e-01 8.18231776e-02 -1.07098214e-01
4.75868076e-01 9.68984216e-02 8.45109761e-01 -6.08948052e-01
-6.26682162e-01 -8.05952072e-01 -8.06082904e-01 9.23199177e-01
2.29482487e-01 2.27676496e-01 -1.22092497e+00 -2.04622880e-01
-6.25076950e-01 -6.55791640e-01 1.16789734e+00 -1.21272421e-02
1.90393817e+00 -2.71973163e-01 3.04946061e-02 6.79311335e-01
1.35091078e+00 -1.58315465e-01 7.10024238e-01 -1.57249719e-01
5.69987297e-01 1.70888230e-01 2.09093332e-01 7.13568151e-01
9.42930430e-02 5.49049318e-01 2.47896045e-01 4.02967364e-01
-2.29338273e-01 -7.28149593e-01 2.19787750e-02 2.88769662e-01
2.37766758e-01 -1.61289141e-01 -6.50906861e-01 4.52376485e-01
-1.82967615e+00 -9.67918634e-01 4.47507650e-01 2.29608870e+00
6.29922867e-01 2.93989062e-01 2.80778911e-02 4.32714038e-02
4.98493016e-01 4.82043415e-01 -5.37642419e-01 -4.63326901e-01
6.27886057e-02 7.07858562e-01 3.63857299e-01 4.78234947e-01
-1.21772480e+00 7.15764880e-01 6.19397736e+00 1.17272937e+00
-9.37205911e-01 -2.49943227e-01 8.23988199e-01 -1.68220997e-01
-1.83513045e-01 -4.29594100e-01 -4.23848838e-01 6.62031531e-01
4.89023209e-01 7.95245916e-02 6.92572236e-01 1.00254190e+00
-4.77399230e-01 1.89469844e-01 -1.64334309e+00 1.30678821e+00
2.35680774e-01 -1.50968170e+00 1.99613348e-01 -7.31801912e-02
7.58967936e-01 -1.18386090e-01 4.18237120e-01 5.06896317e-01
1.97029576e-01 -1.48096883e+00 6.99172378e-01 8.52769971e-01
1.08382034e+00 -9.29418027e-01 6.09705448e-01 -1.19723915e-03
-9.07320917e-01 1.44284144e-01 -8.30393374e-01 7.61280581e-02
-2.92841703e-01 7.76672661e-01 -6.07839525e-01 2.41165474e-01
2.42383003e-01 7.43194759e-01 -3.58717054e-01 1.04837406e+00
-5.76313436e-01 3.88429582e-01 -3.47423166e-01 -1.49042770e-01
4.07773852e-01 -6.81191757e-02 1.83161229e-01 1.49733353e+00
1.44857317e-01 -2.88719416e-01 -1.62474290e-01 1.38657880e+00
-8.23841751e-01 -5.40562212e-01 -6.39320374e-01 -2.08225504e-01
4.86030430e-01 9.87505019e-01 -2.88007349e-01 -5.90227485e-01
-2.03760147e-01 1.48210478e+00 7.55312085e-01 2.86134005e-01
-7.39555478e-01 -1.07480979e+00 8.78610611e-01 -2.87798494e-01
8.03351343e-01 -2.49751389e-01 -4.66700613e-01 -9.58769739e-01
4.61497724e-01 -5.58560610e-01 2.59147376e-01 -2.90688038e-01
-2.14514971e+00 3.21465939e-01 -4.09301758e-01 -8.66142750e-01
6.15634657e-02 -1.18889856e+00 -8.92277181e-01 9.83029962e-01
-1.38842416e+00 -9.09289539e-01 -2.64021158e-01 1.61033556e-01
6.03255212e-01 -4.69636351e-01 8.70141506e-01 4.76794392e-01
-5.45753062e-01 1.37947905e+00 4.04431462e-01 6.04538918e-01
8.83859038e-01 -1.54191685e+00 8.43009591e-01 3.24103206e-01
4.28266406e-01 5.47318816e-01 1.25431478e-01 -2.78861910e-01
-1.51705706e+00 -1.07058215e+00 8.74425352e-01 -5.81382573e-01
7.02222884e-01 -6.49668515e-01 -8.72393548e-01 3.13959867e-01
-2.33533740e-01 6.88674986e-01 3.95791531e-01 1.53177351e-01
-1.25446010e+00 -2.52829820e-01 -1.48678863e+00 2.62774348e-01
1.17164230e+00 -1.19209325e+00 -7.60018587e-01 2.66249299e-01
-1.03254514e-02 -2.81620145e-01 -8.87836456e-01 2.83660498e-02
1.36844218e+00 -6.84766889e-01 1.23956060e+00 -1.13946772e+00
7.02478409e-01 1.63568169e-01 -4.68171507e-01 -1.08583188e+00
-3.60651910e-01 -6.72743499e-01 -3.46656024e-01 1.02979410e+00
2.17347413e-01 -2.72054642e-01 1.14592361e+00 8.72142375e-01
1.97370395e-01 -9.53282475e-01 -1.22259247e+00 -1.17760634e+00
5.34960091e-01 -5.13748765e-01 4.24832106e-01 6.21647298e-01
-3.28039467e-01 -2.82686859e-01 -5.32480359e-01 -4.61544655e-03
1.07528913e+00 1.49653450e-01 8.67346168e-01 -1.17700279e+00
-1.59128264e-01 -8.20836425e-01 -6.81441665e-01 -1.29012680e+00
5.43156505e-01 -9.65773106e-01 7.75017887e-02 -1.31601477e+00
8.73651132e-02 -3.87766451e-01 -6.49760783e-01 1.85552061e-01
-1.25020638e-01 1.20960377e-01 3.59002799e-01 -2.56735802e-01
-6.98324323e-01 9.02753472e-01 7.53291309e-01 -6.30543828e-01
5.67323565e-01 1.63819566e-02 -2.42665615e-02 4.70252573e-01
6.25674427e-01 -7.31476843e-01 1.37008596e-02 -7.25956619e-01
-1.53102830e-01 -1.95053980e-01 5.97152412e-01 -7.72688806e-01
2.43948132e-01 1.70751400e-02 7.84247041e-01 -7.97515273e-01
4.25158799e-01 -8.26865971e-01 -5.16555369e-01 3.31793249e-01
-7.35425651e-01 -3.02344829e-01 1.54124439e-01 9.10275221e-01
-3.23781446e-02 -3.48987609e-01 9.18009639e-01 1.75163835e-01
-2.28160545e-01 5.42360365e-01 -1.28378332e-01 2.82665014e-01
5.56319773e-01 9.17027518e-02 -2.79695839e-01 -2.26235524e-01
-3.85760456e-01 -8.43493920e-03 1.29260927e-01 5.30767322e-01
1.04719937e+00 -1.78964758e+00 -6.71900570e-01 2.86624044e-01
1.67170838e-01 -2.71126330e-01 2.22836971e-01 1.05316006e-01
-3.99410576e-01 2.59538293e-01 -3.10599566e-01 -2.11282775e-01
-7.21922696e-01 3.52821022e-01 4.07768339e-01 -2.55668350e-02
-5.58516800e-01 1.27846229e+00 -1.00120313e-01 -4.05639946e-01
1.03777552e+00 -4.36660141e-01 8.80512059e-01 -1.12773687e-01
6.67507648e-01 7.49189198e-01 1.45103252e-02 1.61797091e-01
-4.74825174e-01 2.99021661e-01 -2.46574536e-01 -1.16315261e-01
1.35779417e+00 5.51166117e-01 3.01359385e-01 3.91957909e-01
1.78360713e+00 -2.24161893e-01 -1.55498040e+00 -3.21524680e-01
1.43871903e-01 -1.02474678e+00 1.13407910e-01 -9.73387241e-01
-1.03537202e+00 1.27193069e+00 6.84723198e-01 -2.76303440e-01
2.53448725e-01 -2.46403754e-01 7.44027495e-01 1.02628267e+00
2.11134598e-01 -1.19227040e+00 3.16032708e-01 3.91768247e-01
1.36841464e+00 -1.42781603e+00 2.13892832e-01 3.76049012e-01
-3.15605164e-01 1.60712290e+00 3.31919760e-01 -8.58046770e-01
8.21753442e-01 2.30962530e-01 -3.20698947e-01 5.52298166e-02
-7.70137250e-01 3.19313109e-01 6.49413466e-01 5.04430890e-01
3.32511067e-01 2.43801042e-01 2.63573468e-01 6.11322999e-01
3.52566779e-01 3.54990900e-01 1.83455013e-02 7.88348258e-01
-1.80974483e-01 -1.09206140e+00 5.66116418e-04 7.78182507e-01
-2.03828007e-01 2.03057602e-01 -6.50569499e-01 5.61024129e-01
-1.52460679e-01 5.17852008e-01 2.33532578e-01 -2.91785240e-01
5.21990955e-01 2.01636285e-01 9.87823069e-01 -2.97855973e-01
-1.65408105e-01 -5.92844009e-01 -4.03995030e-02 -5.22868991e-01
3.03636551e-01 -3.32693338e-01 -5.58377624e-01 -4.28307295e-01
-3.33240777e-02 -2.14406520e-01 7.08644509e-01 6.43758774e-01
3.26005071e-01 3.41422200e-01 8.44038367e-01 -1.11809337e+00
-1.46737814e+00 -9.61616457e-01 -5.60559094e-01 3.88983756e-01
4.77517486e-01 -8.34542513e-01 -4.45329666e-01 -4.99650389e-01] | [11.655019760131836, 0.5384122133255005] |
c9f1ad43-2434-41e1-b0a6-216817eb84ec | counting-with-adaptive-auxiliary-learning | 2203.04061 | null | https://arxiv.org/abs/2203.04061v1 | https://arxiv.org/pdf/2203.04061v1.pdf | Counting with Adaptive Auxiliary Learning | This paper proposes an adaptive auxiliary task learning based approach for object counting problems. Unlike existing auxiliary task learning based methods, we develop an attention-enhanced adaptively shared backbone network to enable both task-shared and task-tailored features learning in an end-to-end manner. The network seamlessly combines standard Convolution Neural Network (CNN) and Graph Convolution Network (GCN) for feature extraction and feature reasoning among different domains of tasks. Our approach gains enriched contextual information by iteratively and hierarchically fusing the features across different task branches of the adaptive CNN backbone. The whole framework pays special attention to the objects' spatial locations and varied density levels, informed by object (or crowd) segmentation and density level segmentation auxiliary tasks. In particular, thanks to the proposed dilated contrastive density loss function, our network benefits from individual and regional context supervision in terms of pixel-independent and pixel-dependent feature learning mechanisms, along with strengthened robustness. Experiments on seven challenging multi-domain datasets demonstrate that our method achieves superior performance to the state-of-the-art auxiliary task learning based counting methods. Our code is made publicly available at: https://github.com/smallmax00/Counting_With_Adaptive_Auxiliary | ['Yalin Zheng', 'Xiaowei Huang', 'Xiaoyun Yang', 'Yihong Qiao', 'Yitian Zhao', 'Meng Wei', 'Joshua Bridge', 'Yanda Meng'] | 2022-03-08 | null | null | null | null | ['object-counting', 'auxiliary-learning'] | ['computer-vision', 'methodology'] | [ 1.65158689e-01 -2.33094066e-01 -1.46906272e-01 -5.92168808e-01
-5.91504812e-01 -2.84158200e-01 7.30514824e-01 1.43138126e-01
-9.39873219e-01 9.33688164e-01 1.39226332e-01 -3.69422808e-02
-5.65606840e-02 -7.84366488e-01 -7.77620912e-01 -7.78832555e-01
1.50360405e-01 7.09816873e-01 3.54649544e-01 1.79552138e-02
3.21233034e-01 2.83085495e-01 -1.34287930e+00 5.12205660e-02
1.06721365e+00 1.01147819e+00 4.34279233e-01 7.50746012e-01
-3.62167865e-01 1.06751227e+00 -4.48565692e-01 -4.51123893e-01
2.04791918e-01 4.43954878e-02 -8.73673201e-01 -1.25508562e-01
7.29266346e-01 -4.65813339e-01 -3.55043679e-01 1.00581801e+00
5.00226498e-01 2.69294828e-01 6.06262445e-01 -1.26455986e+00
-1.16510534e+00 4.19133633e-01 -8.53682518e-01 8.61942828e-01
-3.44654649e-01 4.29901510e-01 9.03654754e-01 -1.01794946e+00
1.52820200e-01 1.21111917e+00 8.12547624e-01 4.43951696e-01
-1.24279380e+00 -8.03651094e-01 4.75597024e-01 9.24745947e-02
-1.21980393e+00 -7.49111846e-02 7.15834379e-01 -4.00958478e-01
9.55048501e-01 -2.23483562e-01 6.27524555e-01 1.10747898e+00
-3.49896818e-01 1.01551402e+00 9.85665262e-01 -1.14713863e-01
-9.80895013e-02 -1.55237615e-02 3.94829780e-01 1.03063846e+00
4.76894051e-01 -2.05110490e-01 -3.56075674e-01 -1.24511346e-01
1.07050478e+00 1.72761649e-01 1.08226284e-01 -1.59052774e-01
-1.08759439e+00 9.21035230e-01 8.28171730e-01 2.76588768e-01
-3.08012307e-01 5.68137348e-01 5.83129525e-01 -1.60500869e-01
9.55890000e-01 -8.28624666e-02 -5.92675328e-01 1.33111896e-02
-6.90212727e-01 3.41312557e-01 3.73932719e-01 1.07627964e+00
9.08784688e-01 1.41864538e-01 -9.00553405e-01 9.59662259e-01
1.02455661e-01 5.68508863e-01 -6.06202744e-02 -1.04042101e+00
7.94048369e-01 7.53190160e-01 -2.23177369e-03 -7.43188739e-01
-5.47799766e-01 -5.67936301e-01 -1.02181041e+00 -2.24498603e-02
6.99193001e-01 -7.44060725e-02 -1.06276608e+00 1.85876799e+00
5.93531609e-01 4.52949166e-01 -6.14491940e-01 7.30740547e-01
8.64444852e-01 2.06345811e-01 5.64404726e-01 3.39214921e-01
1.61004567e+00 -1.24918950e+00 -5.29636681e-01 -3.51084918e-01
5.36046445e-01 -3.74362767e-01 1.33492827e+00 -5.79945557e-02
-1.09420061e+00 -7.25633264e-01 -4.85405505e-01 -7.01053262e-01
-5.34912586e-01 3.19127202e-01 8.67231607e-01 4.61630613e-01
-9.61380184e-01 3.62745047e-01 -7.45570779e-01 -8.01542774e-02
1.34182286e+00 4.50199425e-01 -7.71965012e-02 -4.25230451e-02
-8.57079744e-01 5.67849815e-01 2.98152745e-01 8.24608281e-02
-5.94832599e-01 -9.21329081e-01 -9.10174906e-01 1.89558014e-01
3.66172761e-01 -1.20128822e+00 1.23686874e+00 -7.36774385e-01
-9.73329663e-01 1.02959049e+00 -2.41951212e-01 -2.18198642e-01
6.47431195e-01 -1.92004681e-01 3.35233122e-01 1.91657797e-01
6.50742590e-01 7.59240389e-01 8.37736905e-01 -1.21694148e+00
-9.19780493e-01 -6.42536461e-01 1.25079110e-01 4.93938588e-02
-6.33600593e-01 -1.67972688e-02 -6.48231268e-01 -7.88850546e-01
-5.97831607e-01 -4.81719166e-01 -2.06474885e-01 4.18315172e-01
-3.07720304e-01 -5.28603017e-01 8.72319520e-01 -6.15546644e-01
9.57410693e-01 -1.82714581e+00 1.47715071e-02 -2.02788413e-01
6.71074152e-01 3.70329440e-01 -2.70053685e-01 -2.24611849e-01
4.41498578e-01 -1.70571640e-01 -5.37482142e-01 -1.00143540e+00
-3.51196490e-02 2.80415326e-01 2.07472056e-01 4.87160265e-01
5.86561203e-01 1.43925786e+00 -1.26952052e+00 -9.35714602e-01
5.42884231e-01 8.27360988e-01 -6.17307484e-01 1.87886819e-01
-3.12448859e-01 9.02670979e-01 -7.14919627e-01 9.80736732e-01
9.15004253e-01 -5.81535935e-01 -3.27279925e-01 1.11125056e-02
1.91744342e-02 -5.38927913e-02 -6.98298037e-01 1.66890919e+00
-7.31168270e-01 2.67149150e-01 4.55588579e-01 -1.11383760e+00
8.77972782e-01 -2.46754363e-01 2.37591073e-01 -8.96513700e-01
3.19160074e-01 3.08573898e-02 -3.19496751e-01 -1.79485247e-01
6.55578852e-01 2.79101074e-01 -1.47937015e-01 2.88494140e-01
4.73330200e-01 -2.19044089e-02 2.72517145e-01 3.52206200e-01
9.35407043e-01 9.79011580e-02 2.87593603e-01 -4.91014451e-01
4.91283923e-01 -2.19934300e-01 4.88129020e-01 8.56642902e-01
-5.39154351e-01 4.06160533e-01 4.54581201e-01 -5.33607721e-01
-1.08697283e+00 -9.04782653e-01 -3.54153216e-01 1.72679162e+00
-2.68034004e-02 2.16930196e-01 -6.84161961e-01 -8.03808630e-01
3.20448607e-01 4.21319753e-01 -1.21103382e+00 2.75872231e-01
-7.48095572e-01 -8.39942813e-01 5.51619947e-01 1.19362819e+00
9.67148483e-01 -1.11098027e+00 -2.09404454e-01 1.88612491e-01
-2.68431842e-01 -1.28327358e+00 -9.14395094e-01 3.72714281e-01
-8.00586939e-01 -1.20677233e+00 -9.00031745e-01 -6.99286103e-01
6.35960281e-01 5.27456284e-01 1.51151168e+00 5.39033353e-01
-5.77337623e-01 4.73737270e-01 -2.30147764e-01 -4.12834495e-01
1.69949919e-01 3.63370091e-01 -4.56117511e-01 -1.15293682e-01
5.49585283e-01 -7.67747641e-01 -8.59336913e-01 1.20615430e-01
-8.09807897e-01 -2.46703550e-01 2.63735920e-01 9.46027935e-01
6.75068378e-01 -6.21924579e-01 5.97065032e-01 -1.04284060e+00
6.56818688e-01 -8.42556357e-01 -8.10175896e-01 2.10107699e-01
-2.83268958e-01 -3.09187621e-02 6.86909199e-01 -2.65155554e-01
-1.23276556e+00 6.13122620e-02 8.33459257e-04 -4.61922079e-01
-4.71130423e-02 -2.03321055e-01 -4.70952876e-02 -1.02935039e-01
5.56446552e-01 3.61013055e-01 -2.76495755e-01 -3.22446138e-01
5.01199484e-01 2.96709985e-01 7.00935662e-01 -8.65414202e-01
8.31939220e-01 7.75029123e-01 -6.33822381e-02 -5.29564321e-01
-1.30207169e+00 -6.26057386e-01 -1.03872406e+00 -1.49306595e-01
1.15312040e+00 -1.01020074e+00 -8.37971807e-01 7.84994543e-01
-1.22077751e+00 -8.39629889e-01 -5.08247674e-01 8.39825347e-02
-4.71781433e-01 3.06064367e-01 -7.10336149e-01 -7.95627117e-01
-5.19705713e-01 -8.17963779e-01 1.57772374e+00 3.72583002e-01
3.82752925e-01 -1.43933892e+00 9.42374393e-03 6.46428883e-01
6.67085946e-01 3.37397635e-01 7.80700326e-01 -6.14994228e-01
-7.11627305e-01 1.43401742e-01 -9.90937769e-01 3.66807371e-01
1.45365819e-01 -1.98794842e-01 -1.06285834e+00 -1.05312273e-01
-4.30892736e-01 -6.96090519e-01 1.55217350e+00 7.56464541e-01
1.78884852e+00 -1.37002766e-01 -2.12200329e-01 9.36022162e-01
1.35007608e+00 -5.60960054e-01 3.76230687e-01 2.85842955e-01
1.11450255e+00 3.35801214e-01 4.40732092e-01 6.47857845e-01
9.99067068e-01 4.55330163e-01 6.86454713e-01 -3.93391907e-01
-2.60248065e-01 -2.36934215e-01 -2.84934193e-01 3.03898662e-01
-4.44526523e-01 -1.79456085e-01 -8.30669224e-01 7.39724219e-01
-2.07535911e+00 -1.03910458e+00 -3.41012925e-01 1.86375070e+00
5.05842328e-01 -4.92100269e-02 4.79975164e-01 -2.13781536e-01
8.97004664e-01 2.05484942e-01 -8.23994637e-01 -2.56640434e-01
-1.82459757e-01 3.99537057e-01 6.37770534e-01 3.71235043e-01
-1.38302815e+00 1.10200095e+00 4.94895840e+00 1.14675117e+00
-4.90389109e-01 6.35882437e-01 9.47537303e-01 -4.38714512e-02
-1.35516256e-01 -4.84393775e-01 -9.45945144e-01 4.74103659e-01
1.98604092e-01 2.34563783e-01 4.05474305e-01 8.22162390e-01
-5.91177754e-02 -1.17636658e-01 -7.14241445e-01 7.45180368e-01
-1.45786330e-01 -1.47017407e+00 -7.29331374e-02 7.66473040e-02
8.41288269e-01 3.46534461e-01 1.79208040e-01 6.93014145e-01
7.03096271e-01 -1.04532325e+00 7.56713450e-01 4.09609765e-01
8.22256982e-01 -7.04692602e-01 8.39344800e-01 2.21471995e-01
-1.77370441e+00 -2.93483794e-01 -4.30939943e-01 -2.16122389e-01
1.40861079e-01 5.91132998e-01 -2.29691267e-01 2.00983703e-01
8.72100115e-01 6.79234922e-01 -7.08744287e-01 8.45377207e-01
-2.11293459e-01 4.51569140e-01 -2.54913628e-01 1.00282179e-02
4.90647227e-01 -1.66596115e-01 1.83704138e-01 1.60646021e+00
-1.03369042e-01 1.33361042e-01 3.44593823e-01 1.19665897e+00
-5.22969067e-01 -7.03697354e-02 -5.67676604e-01 4.44541842e-01
6.95215166e-01 1.52688193e+00 -9.71773982e-01 -4.11103874e-01
-6.55913413e-01 8.58190894e-01 1.03494895e+00 1.81147709e-01
-1.10880256e+00 -2.10918009e-01 5.75863838e-01 1.78490520e-01
6.47679210e-01 -5.82489856e-02 -4.96388912e-01 -9.56520677e-01
2.38781981e-02 -1.25608698e-01 6.61130190e-01 -3.42093885e-01
-1.87707531e+00 2.92302966e-01 1.22007795e-01 -6.48417652e-01
3.04886103e-01 -5.97979009e-01 -9.87448275e-01 9.38226283e-01
-2.05843282e+00 -1.79024315e+00 -6.42562747e-01 1.13898861e+00
6.54821873e-01 -1.45690337e-01 4.50744480e-01 4.12512034e-01
-5.06277978e-01 7.95204401e-01 -6.77713007e-02 5.23231387e-01
3.42377275e-01 -1.54942691e+00 5.46666801e-01 5.35366178e-01
-2.10437268e-01 2.91989654e-01 -9.33655724e-02 -4.62568521e-01
-8.07214200e-01 -1.78902292e+00 5.98230541e-01 -7.35379636e-01
6.29804313e-01 -5.97144902e-01 -9.39630866e-01 7.49459267e-01
2.17628870e-02 7.78741956e-01 3.80317837e-01 2.02829733e-01
-6.13849580e-01 1.58053547e-01 -1.43593872e+00 1.42535165e-01
1.48859048e+00 -4.45503443e-01 -1.86940253e-01 6.67712688e-01
8.80075157e-01 -2.82389611e-01 -6.02878809e-01 3.43565494e-01
1.21871315e-01 -9.22580242e-01 1.18529463e+00 -4.78820741e-01
5.87022662e-01 1.94097757e-02 -9.64061841e-02 -8.42955828e-01
-7.59750605e-01 5.86686702e-03 -3.95076275e-01 1.25377011e+00
2.24540457e-01 -7.21982658e-01 8.29190314e-01 3.25788796e-01
-1.48644269e-01 -8.15031946e-01 -1.14863372e+00 -7.83542633e-01
5.09450734e-01 -3.11855912e-01 5.22849023e-01 8.91045332e-01
-6.26954734e-01 4.60381091e-01 -2.98163652e-01 3.24393697e-02
9.70029414e-01 -2.19240710e-02 8.88175309e-01 -1.25526345e+00
-1.43795073e-01 -7.25727975e-01 -1.37615308e-01 -1.10913360e+00
3.85767162e-01 -9.98050928e-01 -1.39481559e-01 -1.54549360e+00
7.43301451e-01 -6.54706120e-01 -3.39981079e-01 6.89244688e-01
-7.63998568e-01 5.01522839e-01 3.13690513e-01 -8.78750384e-02
-1.03144121e+00 7.63766646e-01 1.49182963e+00 -2.98422962e-01
-8.38573575e-02 8.07513818e-02 -7.60012627e-01 6.96803212e-01
1.01015544e+00 -6.34955585e-01 -2.40115002e-01 -8.39542747e-01
-1.10619128e-01 -3.45681489e-01 9.15588260e-01 -7.37619400e-01
1.22962497e-01 -2.56431177e-02 5.38830817e-01 -4.82343197e-01
3.74091566e-01 -6.10207617e-01 -7.48292267e-01 4.10424352e-01
-8.61002132e-02 -1.25927329e-01 3.39971274e-01 9.00198817e-01
1.19966283e-01 9.01588984e-03 9.53762352e-01 -3.92389417e-01
-5.47499418e-01 8.61879408e-01 2.09861591e-01 5.83403528e-01
8.07410121e-01 -3.95017773e-01 -6.44927859e-01 -3.16050947e-02
-4.91223842e-01 6.02579117e-01 2.03682303e-01 1.49670124e-01
4.22374874e-01 -1.21193290e+00 -1.06445253e+00 -1.04008324e-01
2.21148685e-01 5.31856954e-01 4.80005085e-01 8.92722785e-01
-2.24549696e-01 2.86923021e-01 -1.70060918e-01 -8.79650772e-01
-1.03738666e+00 4.46538776e-01 1.47365451e-01 -7.73187459e-01
-4.70856875e-01 1.30489779e+00 3.89371991e-01 -9.30936277e-01
2.44255841e-01 -4.02523339e-01 -7.53370151e-02 -1.16613738e-01
4.81615245e-01 5.29830813e-01 -1.61243320e-01 -5.17979324e-01
-2.78309733e-01 4.94753867e-01 -1.06244989e-01 4.62081850e-01
1.67016256e+00 -1.17119085e-02 5.76442704e-02 1.03044488e-01
1.10384595e+00 -3.29310954e-01 -1.73109293e+00 -7.85144746e-01
-2.10953251e-01 -7.12375522e-01 3.77793945e-02 -6.17783248e-01
-1.45973551e+00 9.84389842e-01 3.67842942e-01 -9.69514698e-02
1.09633482e+00 2.39482895e-01 7.14872897e-01 2.42823765e-01
3.57190549e-01 -1.18911982e+00 4.95582521e-01 7.03147769e-01
5.75714707e-01 -1.69162083e+00 -4.20560986e-02 -2.48521596e-01
-6.86097860e-01 7.69513488e-01 1.06731427e+00 -3.45179200e-01
6.36216640e-01 4.23877537e-01 -3.64176124e-01 -4.12289143e-01
-5.27192950e-01 -6.02925479e-01 8.08486491e-02 9.13505137e-01
3.29120904e-01 7.50333667e-02 3.70309502e-02 6.03074551e-01
2.24227905e-01 7.40687922e-02 -2.02920642e-02 8.51945817e-01
-4.87779319e-01 -6.77361608e-01 -3.14994961e-01 8.48238111e-01
-3.30360681e-01 -4.27241921e-01 -8.80696923e-02 1.02696240e+00
5.22330701e-01 5.54049253e-01 4.47164983e-01 2.17723012e-01
2.14075252e-01 -2.40875602e-01 6.63652837e-01 -7.16159940e-01
-8.58407795e-01 -4.42652911e-01 -2.65364379e-01 -5.00238121e-01
-7.04784930e-01 -4.99922395e-01 -1.11197674e+00 -6.35324180e-01
-4.30080563e-01 -4.68160033e-01 2.62556344e-01 7.91375697e-01
1.94613874e-01 6.65825725e-01 4.00068343e-01 -1.24810016e+00
-2.89371401e-01 -1.11806154e+00 -5.29091537e-01 3.51284325e-01
3.88557076e-01 -8.63775432e-01 2.37546186e-03 -1.96231052e-01] | [8.951532363891602, 0.2501373887062073] |
a2f4e19d-6c19-41cb-b078-919ca3754103 | vs-transgru-a-novel-transformer-gru-based | 2307.03918 | null | https://arxiv.org/abs/2307.03918v1 | https://arxiv.org/pdf/2307.03918v1.pdf | VS-TransGRU: A Novel Transformer-GRU-based Framework Enhanced by Visual-Semantic Fusion for Egocentric Action Anticipation | Egocentric action anticipation is a challenging task that aims to make advanced predictions of future actions from current and historical observations in the first-person view. Most existing methods focus on improving the model architecture and loss function based on the visual input and recurrent neural network to boost the anticipation performance. However, these methods, which merely consider visual information and rely on a single network architecture, gradually reach a performance plateau. In order to fully understand what has been observed and capture the dependencies between current observations and future actions well enough, we propose a novel visual-semantic fusion enhanced and Transformer GRU-based action anticipation framework in this paper. Firstly, high-level semantic information is introduced to improve the performance of action anticipation for the first time. We propose to use the semantic features generated based on the class labels or directly from the visual observations to augment the original visual features. Secondly, an effective visual-semantic fusion module is proposed to make up for the semantic gap and fully utilize the complementarity of different modalities. Thirdly, to take advantage of both the parallel and autoregressive models, we design a Transformer based encoder for long-term sequential modeling and a GRU-based decoder for flexible iteration decoding. Extensive experiments on two large-scale first-person view datasets, i.e., EPIC-Kitchens and EGTEA Gaze+, validate the effectiveness of our proposed method, which achieves new state-of-the-art performance, outperforming previous approaches by a large margin. | ['Yanning Zhang', 'Lingtong Min', 'Qinyi Lv', 'Ze Sun', 'Congqi Cao'] | 2023-07-08 | null | null | null | null | ['action-anticipation'] | ['computer-vision'] | [ 1.70299158e-01 -1.15067005e-01 -2.32628003e-01 -6.20421410e-01
-4.26342189e-01 -6.32354524e-03 7.26801634e-01 -3.42473179e-01
-3.39430571e-01 4.55300838e-01 7.83742249e-01 2.76597857e-01
8.24825466e-02 -4.82627183e-01 -5.93200147e-01 -4.40746784e-01
2.76473939e-01 5.52354865e-02 5.70728146e-02 -2.33541191e-01
8.24597180e-02 -1.88241482e-01 -1.77128673e+00 5.52128315e-01
8.64857376e-01 1.05429828e+00 4.07910794e-01 4.36369449e-01
1.95517559e-02 1.11488235e+00 1.14194185e-01 -4.46901888e-01
1.32422477e-01 -5.06216049e-01 -4.53340590e-01 2.61843145e-01
1.89065468e-02 -5.17630100e-01 -6.61470890e-01 7.75980294e-01
5.46477854e-01 2.06459746e-01 2.45073110e-01 -1.16358459e+00
-5.06053209e-01 4.97455746e-01 -5.11692464e-01 9.27159842e-03
5.88835120e-01 4.64585930e-01 8.93154442e-01 -7.92234480e-01
6.39482796e-01 1.24994385e+00 5.16837895e-01 6.74431801e-01
-8.20012033e-01 -4.82541263e-01 7.85170376e-01 8.43472123e-01
-1.10248363e+00 -5.36932707e-01 1.04848516e+00 -4.12248731e-01
1.00333297e+00 4.50908057e-02 1.02220798e+00 1.45074677e+00
1.97944716e-01 1.16955972e+00 1.08142209e+00 -2.92014599e-01
-4.80572656e-02 2.86530908e-02 -8.01976621e-02 7.57139266e-01
-5.23858666e-01 1.77049652e-01 -8.08283567e-01 3.88633847e-01
4.79421705e-01 5.40321887e-01 -2.26303339e-01 -3.63338023e-01
-1.26545131e+00 5.64279139e-01 6.33067966e-01 1.50565594e-01
-7.48913407e-01 3.56089532e-01 5.77652633e-01 -8.87917951e-02
5.69929719e-01 -9.07427743e-02 -2.17399418e-01 -5.91428399e-01
-7.66486228e-01 1.51016749e-02 2.24471331e-01 7.64724970e-01
5.73063314e-01 4.76805568e-02 -4.28608954e-01 7.03234136e-01
5.25343895e-01 5.38054883e-01 6.67525947e-01 -8.44444990e-01
6.46331906e-01 8.60759735e-01 3.18024755e-02 -9.64309633e-01
-4.33814615e-01 -4.69540566e-01 -6.78212404e-01 1.18145183e-01
1.18690543e-01 -2.07610112e-02 -1.13512862e+00 1.97808981e+00
2.40808815e-01 4.10263211e-01 2.37857521e-01 9.96436834e-01
6.92856729e-01 5.50660610e-01 3.02452207e-01 -2.11375490e-01
1.34159076e+00 -1.27748024e+00 -7.77917802e-01 -5.08625090e-01
7.76254654e-01 -3.39856893e-01 1.06005466e+00 9.64310765e-02
-9.60164368e-01 -8.66501093e-01 -9.54927564e-01 -1.69823706e-01
-2.36885533e-01 4.05853093e-01 7.25618780e-01 1.42917901e-01
-7.90292919e-01 4.44130123e-01 -1.04042721e+00 -2.75913805e-01
4.95122999e-01 1.03855565e-01 -4.05771255e-01 -1.77789241e-01
-1.28788960e+00 7.67930448e-01 4.99500632e-01 4.22792912e-01
-8.95064235e-01 -3.39080513e-01 -9.77906227e-01 2.59179384e-01
5.08851111e-01 -9.47715580e-01 1.01464307e+00 -1.26175284e+00
-1.52287591e+00 4.37451094e-01 -3.75718325e-01 -6.54907048e-01
6.57182336e-01 -4.64886963e-01 -4.04359967e-01 9.90123525e-02
-1.32350966e-01 7.94134915e-01 7.38079667e-01 -8.86514902e-01
-7.22105801e-01 -5.73644757e-01 2.87041754e-01 7.57099092e-01
-5.23848951e-01 -2.43339121e-01 -5.66436946e-01 -4.53300744e-01
-5.95126413e-02 -1.09692276e+00 -2.65449286e-01 -2.52196789e-01
-1.81950167e-01 -2.15399355e-01 5.74309886e-01 -9.00103748e-01
1.19867885e+00 -2.16489482e+00 3.03091675e-01 -1.09537452e-01
1.38109431e-01 4.22764122e-01 -1.13477968e-01 3.61970335e-01
-1.56881511e-01 -3.89224142e-01 9.70434323e-02 -7.28516579e-01
1.43833030e-02 1.52731627e-01 -2.48089254e-01 2.58888215e-01
-1.25693768e-01 1.27303874e+00 -8.14159870e-01 -4.23505366e-01
5.62623024e-01 6.03878140e-01 -5.06481051e-01 3.22905391e-01
-2.09609598e-01 6.62396193e-01 -5.60543478e-01 4.18615282e-01
2.41546586e-01 -3.66493404e-01 5.60725443e-02 -2.59236693e-01
-8.05837587e-02 1.58450946e-01 -8.85357082e-01 2.18873596e+00
-6.51589274e-01 4.09499288e-01 -6.11938417e-01 -9.71989572e-01
7.57106185e-01 3.61367762e-01 5.41597009e-01 -1.03312874e+00
1.34553993e-02 -9.35859159e-02 -2.70411789e-01 -6.65104628e-01
3.61428022e-01 -4.78371978e-02 -1.65524613e-02 1.53573185e-01
-1.84515566e-02 6.78203225e-01 3.97618860e-02 1.80369854e-01
8.00165415e-01 8.27135324e-01 2.60364026e-01 4.61815119e-01
7.38111496e-01 -2.09764838e-01 8.64083767e-01 4.05547142e-01
-2.28835747e-01 5.76141357e-01 3.90061617e-01 -5.15840888e-01
-8.23133647e-01 -6.55171514e-01 5.16258895e-01 1.07586670e+00
3.22459459e-01 -7.02049434e-01 -6.00793600e-01 -7.82705605e-01
-3.81064475e-01 8.42139959e-01 -7.60733724e-01 -3.91036123e-01
-5.31955719e-01 -5.33350587e-01 -1.26058906e-02 1.03575134e+00
8.91491532e-01 -1.19822454e+00 -9.10353839e-01 1.40189543e-01
-6.08525455e-01 -1.30765033e+00 -3.25629383e-01 -4.73940134e-01
-7.24050820e-01 -8.28993380e-01 -7.57468045e-01 -1.86845392e-01
4.45757151e-01 2.79796362e-01 7.35949755e-01 -2.00599790e-01
-1.92243280e-03 6.65243447e-01 -5.12546539e-01 -2.51107305e-01
9.27666426e-02 -1.64918154e-01 -9.66140162e-03 4.13028955e-01
3.72075766e-01 -7.29940355e-01 -8.30882370e-01 2.87080407e-02
-4.58349109e-01 6.82146609e-01 7.95795143e-01 8.88749361e-01
4.13802087e-01 -2.39062890e-01 4.02277201e-01 -5.51902592e-01
1.67278558e-01 -3.42698902e-01 -1.31260425e-01 4.68132854e-01
-6.40682220e-01 3.15727741e-02 4.62370843e-01 -2.75540531e-01
-1.61963820e+00 2.68097669e-01 -1.81872129e-01 -8.66784513e-01
-5.98840155e-02 5.92203617e-01 -2.13758573e-01 3.34055007e-01
2.75828212e-01 5.43389499e-01 -3.22347321e-02 -4.52844292e-01
5.97759604e-01 4.84977126e-01 5.15657544e-01 -3.17142461e-04
3.05537283e-01 6.81021273e-01 -1.50912151e-01 -2.87699729e-01
-8.96902144e-01 -5.30493975e-01 -6.10884130e-01 -5.88273466e-01
1.06329501e+00 -1.41453171e+00 -9.08533394e-01 6.00854814e-01
-1.05157161e+00 -1.86835572e-01 -1.98968813e-01 7.61976361e-01
-7.96466470e-01 5.29035866e-01 -2.78438002e-01 -1.02091312e+00
-2.99972385e-01 -1.13400149e+00 1.02075934e+00 3.34639281e-01
7.26898238e-02 -9.20435429e-01 -1.85026526e-02 6.01014316e-01
1.37526110e-01 2.83436298e-01 3.86938959e-01 -3.76112878e-01
-6.74539030e-01 -2.65423089e-01 -2.57197976e-01 2.83097148e-01
-1.44725546e-01 -4.67452526e-01 -8.96224797e-01 -1.07889168e-01
5.32376133e-02 -3.00213039e-01 1.20564306e+00 3.05549294e-01
1.15059114e+00 -2.16475204e-01 -4.28301603e-01 6.57504916e-01
1.37334800e+00 1.84094176e-01 8.80349517e-01 3.03352118e-01
1.07547069e+00 6.05835795e-01 8.52929413e-01 7.24715233e-01
9.59584534e-01 8.34087849e-01 5.34615219e-01 1.89265367e-02
-1.22115836e-01 -6.58196211e-01 6.89247191e-01 6.61264956e-01
-4.79166448e-01 -3.34639668e-01 -7.23697484e-01 5.56142867e-01
-2.47540927e+00 -1.16874838e+00 1.24679588e-01 2.01002860e+00
3.02799493e-01 1.12965681e-01 8.40635747e-02 -1.67591423e-01
5.53558767e-01 5.09117842e-01 -6.71869874e-01 -1.32164344e-01
8.37287530e-02 -4.10880893e-01 2.93002605e-01 1.72708064e-01
-1.06835830e+00 1.03394735e+00 5.37732315e+00 7.72936583e-01
-1.05676603e+00 1.28923625e-01 4.99825001e-01 -2.63793975e-01
-2.29185373e-01 1.64676443e-01 -7.29876637e-01 7.09048510e-01
8.27186286e-01 -1.66821331e-02 3.59084249e-01 8.89034986e-01
4.12230015e-01 -8.65910798e-02 -1.02059293e+00 1.18552649e+00
3.36766869e-01 -1.02804637e+00 -6.78045973e-02 -1.03543010e-02
4.81013238e-01 -3.59196588e-02 1.24934882e-01 4.94951308e-01
1.27447575e-01 -7.87106335e-01 6.89911366e-01 1.27019560e+00
5.31895816e-01 -7.21255958e-01 7.34987855e-01 5.04015386e-01
-1.27375388e+00 -4.70164090e-01 -5.15175238e-02 -3.48577648e-01
6.27184272e-01 2.93371856e-01 -4.81425583e-01 8.69677186e-01
6.49110138e-01 1.43081903e+00 -6.55686796e-01 7.44584441e-01
-4.41624790e-01 3.34963202e-01 -6.48908615e-02 1.54275134e-01
3.75849962e-01 -7.84808397e-02 5.51427186e-01 7.95728683e-01
3.30807477e-01 1.92575172e-01 1.87966853e-01 5.55116773e-01
3.41608852e-01 -2.07835361e-02 -5.78928769e-01 -6.74961358e-02
-3.35334614e-02 1.11479712e+00 -2.68118799e-01 -5.36827624e-01
-5.77370107e-01 1.14204609e+00 4.54723626e-01 4.25939053e-01
-1.11218822e+00 4.25332710e-02 5.99177301e-01 -9.37156528e-02
5.25177836e-01 -1.96121991e-01 -1.37396812e-01 -1.48688543e+00
2.96222866e-01 -6.23746216e-01 4.30000067e-01 -9.66035903e-01
-7.96729565e-01 4.81519908e-01 5.51800653e-02 -1.51604402e+00
-8.69207025e-01 -1.84813574e-01 -5.29808879e-01 6.31597221e-01
-1.46716428e+00 -1.61766958e+00 -5.43296933e-01 5.69409072e-01
8.19752812e-01 -1.90248773e-01 5.19934535e-01 1.21048734e-01
-5.86943448e-01 4.23718929e-01 -7.70256221e-02 -1.50480449e-01
5.14303386e-01 -8.42752874e-01 2.81094104e-01 9.43774521e-01
1.64216369e-01 3.19631666e-01 5.30121148e-01 -6.49792671e-01
-1.10692501e+00 -9.37450349e-01 9.08694804e-01 -3.10453534e-01
2.82293051e-01 -3.17427516e-01 -6.74672842e-01 9.17534530e-01
1.84956819e-01 -2.13937268e-01 5.21505117e-01 1.91836059e-01
-1.65977508e-01 -2.18067333e-01 -6.81715608e-01 7.41555750e-01
1.49623966e+00 -5.15762389e-01 -6.68815315e-01 2.78095394e-01
6.25591457e-01 -3.57497275e-01 -5.69014430e-01 5.21804631e-01
8.48961234e-01 -1.29275548e+00 1.05646849e+00 -6.29440129e-01
7.60576367e-01 -1.99433535e-01 -5.12232147e-02 -1.19551158e+00
-3.61559302e-01 -4.19451207e-01 -2.89084017e-01 1.17614257e+00
-7.85569400e-02 -5.58706939e-01 6.57944202e-01 7.44789243e-01
-1.41298383e-01 -8.45439076e-01 -7.61212826e-01 -4.60652471e-01
-7.07613051e-01 -5.62452018e-01 5.46716094e-01 5.28225839e-01
6.57119974e-02 3.99940223e-01 -1.06069767e+00 -1.84581637e-01
4.87692744e-01 2.13814244e-01 8.19757044e-01 -8.63459229e-01
-2.18505114e-01 -7.98919797e-02 -6.98760390e-01 -1.25998116e+00
3.11888516e-01 -6.70931756e-01 3.51303704e-02 -1.62567914e+00
3.86536062e-01 4.96562906e-02 -5.75145602e-01 6.59187675e-01
-3.79022479e-01 1.74676239e-01 5.26058316e-01 2.26317436e-01
-9.45863247e-01 1.11607468e+00 1.18927062e+00 1.17458537e-01
-2.51285076e-01 -4.03259732e-02 -6.09309316e-01 9.83049154e-01
5.19176781e-01 -1.22577310e-01 -6.82098687e-01 -5.09068668e-01
1.91690981e-01 2.19513938e-01 7.60851920e-01 -1.24290502e+00
3.01089317e-01 -1.03625432e-01 4.81632233e-01 -7.16414630e-01
7.73189247e-01 -6.96771264e-01 6.36898503e-02 2.61761695e-01
-4.46545690e-01 -9.37877223e-03 -1.28707111e-01 1.05472958e+00
-2.44599208e-01 1.84577242e-01 3.98528963e-01 -2.45863229e-01
-1.36533809e+00 3.24383885e-01 -1.39263747e-02 -2.77436078e-01
1.13361275e+00 -3.59522998e-01 -1.64026156e-01 -6.74197912e-01
-9.55260634e-01 5.75221062e-01 3.33506584e-01 6.79021657e-01
8.39289248e-01 -1.60573018e+00 -5.77262402e-01 1.24495178e-01
2.98489302e-01 -4.78971243e-01 9.87503409e-01 1.16974783e+00
7.32123479e-02 4.82400596e-01 -4.93213087e-01 -5.67021728e-01
-1.33967817e+00 6.73095703e-01 8.10195580e-02 -4.67544824e-01
-9.29859102e-01 6.93582416e-01 5.87611437e-01 -7.14404276e-03
5.67585081e-02 2.40599345e-02 -6.64486051e-01 6.59373030e-02
5.87931633e-01 3.16125602e-01 -4.67105657e-01 -9.49489892e-01
-4.40610677e-01 5.57841480e-01 -1.21832397e-02 -1.82523906e-01
1.38708532e+00 -6.12542152e-01 3.08346540e-01 5.63370526e-01
1.12563515e+00 -4.82791483e-01 -1.69544542e+00 -4.30083871e-01
-3.83883119e-01 -7.27370679e-01 5.08171506e-02 -7.50548422e-01
-1.11925030e+00 1.11989570e+00 8.07237804e-01 -3.43940645e-01
1.42548990e+00 -1.39816776e-01 9.38397229e-01 1.79234505e-01
1.77711219e-01 -1.18516099e+00 2.77742505e-01 2.12195501e-01
8.34527552e-01 -1.30006051e+00 6.97534978e-02 -2.57279575e-01
-1.26152468e+00 1.02108097e+00 6.71206474e-01 1.85059123e-02
2.46609151e-01 -4.45737898e-01 -2.36921832e-02 -6.67738616e-02
-9.27390277e-01 -5.34394622e-01 4.46567565e-01 4.03250247e-01
8.17758739e-02 1.97593961e-02 -2.61458933e-01 5.93183100e-01
9.27895308e-02 4.57332462e-01 1.68175355e-01 6.51475370e-01
-2.34039620e-01 -9.04538155e-01 1.48813576e-01 2.92577446e-01
-1.35689616e-01 -9.13437456e-03 -1.67923998e-02 4.68778104e-01
2.09707633e-01 7.50902116e-01 -4.13437225e-02 -7.11723447e-01
2.28696138e-01 2.72666156e-01 3.83457541e-01 -1.77384436e-01
-2.72166193e-01 4.77665067e-02 2.00001135e-01 -1.06473136e+00
-8.25045347e-01 -7.95538545e-01 -1.08982551e+00 -1.18201599e-02
-3.19286101e-02 -3.05957526e-01 4.91684318e-01 1.35723758e+00
6.54508829e-01 7.58378863e-01 6.00675046e-01 -8.03475738e-01
-4.12563115e-01 -9.14625168e-01 -3.35901052e-01 4.94615436e-01
1.23685405e-01 -9.82122481e-01 -1.61248788e-01 1.07331499e-01] | [8.24383544921875, 0.4918098449707031] |
e74437a9-1dc9-40b4-843b-534524ab0f92 | ahead-a-triple-attention-based-heterogeneous | 2208.08200 | null | https://arxiv.org/abs/2208.08200v1 | https://arxiv.org/pdf/2208.08200v1.pdf | AHEAD: A Triple Attention Based Heterogeneous Graph Anomaly Detection Approach | Graph anomaly detection on attributed networks has become a prevalent research topic due to its broad applications in many influential domains. In real-world scenarios, nodes and edges in attributed networks usually display distinct heterogeneity, i.e. attributes of different types of nodes show great variety, different types of relations represent diverse meanings. Anomalies usually perform differently from the majority in various perspectives of heterogeneity in these networks. However, existing graph anomaly detection approaches do not leverage heterogeneity in attributed networks, which is highly related to anomaly detection. In light of this problem, we propose AHEAD: a heterogeneity-aware unsupervised graph anomaly detection approach based on the encoder-decoder framework. Specifically, for the encoder, we design three levels of attention, i.e. attribute level, node type level, and edge level attentions to capture the heterogeneity of network structure, node properties and information of a single node, respectively. In the decoder, we exploit structure, attribute, and node type reconstruction terms to obtain an anomaly score for each node. Extensive experiments show the superiority of AHEAD on several real-world heterogeneous information networks compared with the state-of-arts in the unsupervised setting. Further experiments verify the effectiveness and robustness of our triple attention, model backbone, and decoder in general. | ['Minnan Luo', 'Jun Zhou', 'Qinghua Zheng', 'Zhaoxuan Tan', 'Shangbin Feng', 'Binchi Zhang', 'Shujie Yang'] | 2022-08-17 | null | null | null | null | ['graph-anomaly-detection'] | ['graphs'] | [ 2.34154817e-02 2.52346426e-01 -1.50029525e-01 -3.17832559e-01
1.58781826e-01 -2.89167970e-01 4.79531139e-01 5.96212447e-01
6.67982325e-02 3.52909446e-01 2.43512258e-01 -4.64057289e-02
-1.70658335e-01 -9.43326056e-01 -6.12329006e-01 -5.70274770e-01
-1.93158224e-01 4.39970821e-01 2.86550254e-01 -1.82935178e-01
-2.81077549e-02 3.09009224e-01 -1.02890623e+00 -1.84586033e-01
1.01152968e+00 9.98145044e-01 -3.74369174e-01 3.09629321e-01
-1.93673342e-01 9.81324852e-01 -4.25738901e-01 -7.28000224e-01
2.83953338e-03 -4.14650112e-01 -3.81468445e-01 3.58074278e-01
3.26335371e-01 -2.09163189e-01 -9.74474669e-01 1.63067400e+00
3.79335403e-01 -1.83824092e-01 7.43219614e-01 -1.77924371e+00
-8.79543900e-01 9.09103930e-01 -8.47088516e-01 5.31251609e-01
9.73973125e-02 1.85297191e-01 1.47433925e+00 -6.69212520e-01
4.68700618e-01 1.26978421e+00 5.29276371e-01 2.30394602e-01
-1.18145418e+00 -5.93586504e-01 8.41560304e-01 4.08737987e-01
-1.56084263e+00 -3.40648800e-01 1.10090756e+00 -3.57329249e-01
5.83028197e-01 8.90617073e-02 4.59558278e-01 1.31654978e+00
1.24088109e-01 8.10088754e-01 3.90363067e-01 1.25853464e-01
6.02714298e-03 -8.73283818e-02 1.82795212e-01 9.06123281e-01
7.50036299e-01 -2.67954111e-01 -4.00842190e-01 -1.33275345e-01
4.07263398e-01 3.88156205e-01 -2.13888690e-01 -3.52152258e-01
-1.11910582e+00 6.83677971e-01 7.31383324e-01 1.79442719e-01
-5.38905144e-01 1.81969047e-01 6.78986430e-01 5.75453818e-01
7.57399201e-01 -6.22542296e-03 -2.43142515e-01 1.57483056e-01
-3.04811656e-01 -4.57888506e-02 8.54479671e-01 1.24037087e+00
6.41264677e-01 2.38931119e-01 -3.62620652e-01 7.00846136e-01
6.65819585e-01 2.77106345e-01 3.15001875e-01 -3.50593656e-01
4.59105045e-01 1.06210804e+00 -6.21068895e-01 -1.61785984e+00
-3.73596221e-01 -9.42135274e-01 -1.47141445e+00 -3.96290749e-01
1.41397357e-01 -1.31876785e-02 -8.73765945e-01 2.01541781e+00
3.83302957e-01 6.68446779e-01 -1.58710063e-01 6.21621966e-01
9.05500114e-01 3.17330450e-01 8.69484805e-03 1.12201665e-02
1.16067231e+00 -9.85831380e-01 -8.31625462e-01 -2.01992780e-01
7.89676964e-01 -1.59787357e-01 7.86396623e-01 -8.61981437e-02
-8.44634414e-01 -1.35229051e-01 -9.38894749e-01 1.53470337e-01
-2.80235380e-01 -3.85050207e-01 4.46640909e-01 4.18900996e-01
-8.32762837e-01 4.59744632e-01 -6.49928093e-01 -3.60038668e-01
6.71632588e-01 1.56123668e-01 -3.19323123e-01 -8.19685757e-02
-1.25025022e+00 2.47233137e-01 4.14069891e-01 1.43482417e-01
-8.54174495e-01 -6.40391469e-01 -1.01127493e+00 3.81304681e-01
8.92009795e-01 -6.03509068e-01 8.82265389e-01 -9.38848794e-01
-8.53929341e-01 6.10537469e-01 -1.09868661e-01 -3.34509373e-01
3.60298276e-01 1.24015577e-01 -9.93669152e-01 -7.73165897e-02
3.41136992e-01 1.00188656e-02 7.48355627e-01 -1.09822726e+00
-6.47561431e-01 -5.91192901e-01 -6.26435131e-02 -3.39985848e-03
-6.72223806e-01 -4.47895706e-01 -6.80584669e-01 -7.72486567e-01
3.61651570e-01 -7.86359370e-01 -8.50082189e-02 -7.96119720e-02
-1.14226103e+00 -2.70063102e-01 8.65059853e-01 -3.45695227e-01
1.59563804e+00 -2.41530085e+00 2.08983213e-01 8.08435023e-01
1.08134580e+00 -3.10795046e-02 -1.14253886e-01 2.97561407e-01
-1.43864527e-01 3.68295431e-01 -2.83654958e-01 -1.58242166e-01
8.54809061e-02 2.77558595e-01 8.10428187e-02 4.94836003e-01
2.98464745e-01 9.21778202e-01 -1.09293330e+00 -4.27494079e-01
-2.35795721e-01 2.14277789e-01 -5.02734721e-01 1.97640181e-01
-1.21613204e-01 4.82619584e-01 -7.44527578e-01 1.03745961e+00
5.24772465e-01 -6.53275132e-01 1.51243493e-01 -2.06346378e-01
4.35304463e-01 1.21734239e-01 -1.01461506e+00 1.42299497e+00
-4.50292369e-03 4.78328049e-01 1.43870667e-01 -1.03400958e+00
7.57864118e-01 2.09602103e-01 3.51198256e-01 -6.00649416e-01
-4.03357372e-02 2.89365798e-01 6.33902192e-01 -3.87060553e-01
3.06931645e-01 2.89323419e-01 9.19978097e-02 4.23163980e-01
1.00194849e-01 8.19915891e-01 2.56021589e-01 7.14175522e-01
1.65470135e+00 -6.35950267e-01 1.96270570e-01 -1.30026460e-01
6.61202490e-01 -5.68641007e-01 8.02714884e-01 7.05129743e-01
-4.08369303e-01 4.32478130e-01 1.23103762e+00 -3.42533827e-01
-9.18152809e-01 -1.00332844e+00 3.62785980e-02 1.09238362e+00
3.50814551e-01 -6.45114779e-01 -3.76043737e-01 -1.04258156e+00
1.99734703e-01 3.52894306e-01 -7.34569848e-01 -6.79999113e-01
-3.57723773e-01 -9.39806521e-01 5.60203493e-01 3.46935719e-01
6.62333131e-01 -1.05467474e+00 4.42027330e-01 1.76653549e-01
-5.19664176e-02 -1.40261269e+00 -5.83039284e-01 -2.25480467e-01
-7.16249883e-01 -1.18322742e+00 -2.12790564e-01 -5.90605080e-01
8.69299889e-01 2.31315166e-01 1.39252222e+00 5.14513791e-01
-5.34004048e-02 3.77545834e-01 -4.32715029e-01 -1.41056046e-01
-3.32737476e-01 5.71247160e-01 1.33494753e-02 6.01283073e-01
3.04514825e-01 -1.10041046e+00 -5.49684227e-01 2.82403350e-01
-1.03183484e+00 -3.35921764e-01 1.04447830e+00 6.96745574e-01
3.88616204e-01 2.78863013e-01 5.56047559e-01 -1.59136546e+00
7.27567315e-01 -1.24168146e+00 -1.74566433e-01 1.52632117e-01
-7.86949933e-01 1.69661224e-01 6.99158192e-01 -1.82917804e-01
-6.12845063e-01 -4.47520167e-01 -4.42634635e-02 -4.92434323e-01
-1.66452765e-01 7.96195626e-01 -7.15331793e-01 1.17380962e-01
3.23874414e-01 2.01022342e-01 -3.76242064e-02 -3.33305568e-01
6.40844554e-02 3.43087345e-01 4.33209449e-01 -4.47328687e-01
1.06191576e+00 3.09389710e-01 1.54633090e-01 -7.63602674e-01
-7.41711438e-01 -3.20791870e-01 -3.66734892e-01 -1.01058148e-01
4.29248691e-01 -8.55200469e-01 -5.50364971e-01 6.21153235e-01
-8.90174210e-01 1.59673184e-01 -1.79272160e-01 3.95533405e-02
1.73089758e-01 5.59521317e-01 -5.74830115e-01 -4.13972855e-01
-2.02092156e-01 -1.16021824e+00 8.34685624e-01 3.99079509e-02
5.86615652e-02 -1.27708018e+00 -1.37361392e-01 -1.51655048e-01
3.77301991e-01 3.77356350e-01 1.25341618e+00 -1.30829155e+00
-6.78199589e-01 -2.84829795e-01 -5.69962442e-01 1.35871666e-02
2.97593445e-01 -5.12624439e-03 -7.10483432e-01 -2.68169761e-01
-5.45222044e-01 4.25301529e-02 1.01908064e+00 -7.32258409e-02
1.50745857e+00 -5.56384206e-01 -5.45464337e-01 7.38871038e-01
1.18183184e+00 -2.12231800e-01 4.49392617e-01 6.74749836e-02
1.36543620e+00 4.58962768e-01 -1.55280388e-04 5.41450083e-01
6.82321966e-01 5.13955593e-01 9.66555417e-01 3.53139564e-02
5.85863087e-03 -4.01598483e-01 3.46244216e-01 9.39908922e-01
1.77393258e-02 -6.85203731e-01 -8.44892144e-01 4.94413584e-01
-1.91414022e+00 -8.59162211e-01 -3.60752344e-01 2.06827402e+00
3.48030180e-01 3.85209978e-01 1.94514483e-01 -5.31102940e-02
1.07454276e+00 5.34510136e-01 -9.78489697e-01 8.61210376e-02
-2.34922096e-01 -5.18043876e-01 4.04125631e-01 1.99226141e-01
-9.54125881e-01 6.63702965e-01 4.89089584e+00 6.94116533e-01
-6.81828022e-01 -3.32585871e-02 8.05126250e-01 2.21519440e-01
-6.48584187e-01 -7.29072094e-02 -4.37556684e-01 9.50090230e-01
8.50438893e-01 -3.77358049e-01 2.51481712e-01 7.66928077e-01
-2.75555700e-01 5.95026076e-01 -1.12537432e+00 7.81830311e-01
-2.09017447e-03 -8.38033915e-01 3.25052679e-01 2.96332538e-01
5.43243408e-01 1.93751931e-01 1.15235694e-01 3.82265508e-01
3.30268443e-01 -9.31065977e-01 3.71708751e-01 4.63438720e-01
4.14452672e-01 -7.50499785e-01 8.53673995e-01 1.56592622e-01
-1.48481226e+00 -1.42218033e-02 -2.89134115e-01 2.92223036e-01
-1.09821901e-01 8.49846244e-01 -4.81933653e-01 7.84219623e-01
6.05029941e-01 1.32783473e+00 -8.05288136e-01 8.24178338e-01
-1.65440008e-01 6.82432175e-01 -2.38750651e-01 2.00649589e-01
3.65500152e-01 -2.26439029e-01 9.64303911e-01 9.61640596e-01
1.64542004e-01 -2.21955046e-01 3.02940577e-01 9.76966083e-01
-6.98279321e-01 5.90157695e-02 -7.84549057e-01 -2.04535887e-01
7.82344282e-01 1.37841308e+00 -6.43658400e-01 -1.53265417e-01
-7.07323492e-01 1.05755866e+00 6.08558118e-01 4.60995942e-01
-7.42668450e-01 -3.70257825e-01 7.84572124e-01 1.50263369e-01
1.86662734e-01 1.41671270e-01 1.04690716e-01 -1.36516285e+00
3.52234185e-01 -8.70628536e-01 6.72299087e-01 -1.65450111e-01
-1.85889828e+00 7.20011652e-01 -3.61397564e-01 -1.19685054e+00
-6.35312754e-04 -3.38475764e-01 -9.98587847e-01 4.24159050e-01
-1.47239614e+00 -1.09096956e+00 -7.35729277e-01 7.83193469e-01
2.03118026e-01 -5.19758999e-01 5.39152563e-01 5.20745814e-01
-1.18447351e+00 9.86321270e-01 1.07211962e-01 6.13075733e-01
5.17759144e-01 -1.45269215e+00 6.53801382e-01 1.01013851e+00
2.55932838e-01 3.71140450e-01 4.50258911e-01 -7.55370140e-01
-1.42775488e+00 -1.46589112e+00 5.41452169e-01 -1.21189982e-01
9.12681162e-01 -3.65594536e-01 -1.31290340e+00 9.73157883e-01
-7.86756426e-02 5.59856474e-01 7.07068503e-01 4.54752535e-01
-4.99928206e-01 -2.75106490e-01 -9.74464536e-01 7.49946892e-01
1.41429961e+00 -4.97064233e-01 7.30887651e-02 3.23299676e-01
8.69413316e-01 -2.88800895e-01 -8.87642205e-01 5.48842609e-01
2.40888327e-01 -9.78643596e-01 7.63615370e-01 -7.96144545e-01
4.45730180e-01 -2.71892071e-01 1.79429501e-02 -1.58211315e+00
-6.26629651e-01 -6.13822043e-01 -7.21135974e-01 1.50324285e+00
6.12377465e-01 -1.02042115e+00 7.33903348e-01 3.95839930e-01
-3.28260422e-01 -8.04586768e-01 -7.68499315e-01 -4.98338193e-01
-3.76366854e-01 -2.56458789e-01 8.88904214e-01 1.13376772e+00
-2.30345488e-01 7.16084301e-01 -4.27382261e-01 4.86513227e-01
7.07880557e-01 -7.97659010e-02 6.85366213e-01 -1.72047758e+00
-8.77557024e-02 -8.03943336e-01 -1.07434428e+00 -8.42073739e-01
4.78981763e-01 -1.14185047e+00 -3.49899411e-01 -1.30074751e+00
2.13567868e-01 -5.25754631e-01 -6.76068187e-01 3.28119814e-01
-5.72670221e-01 4.69294451e-02 -1.85792327e-01 3.89274746e-01
-9.65915799e-01 8.26710343e-01 9.64333534e-01 -3.89427692e-01
6.27168864e-02 -2.90257912e-02 -1.02246976e+00 8.77612412e-01
7.42456436e-01 -4.87135261e-01 -5.21644473e-01 -4.12770897e-01
4.89616901e-01 -2.67524421e-01 3.47157031e-01 -9.43987131e-01
4.06612962e-01 3.69570494e-01 1.89091608e-01 -2.50533104e-01
-1.03434704e-01 -1.01824975e+00 -1.62818879e-01 3.13503057e-01
-4.11251217e-01 3.60619903e-01 -2.66676188e-01 1.34452116e+00
-3.53308260e-01 1.97607458e-01 3.79640073e-01 6.13124184e-02
-8.45948517e-01 1.14473009e+00 3.98566164e-02 6.05309069e-01
9.03687119e-01 3.07566449e-02 -4.77755696e-01 -5.56189477e-01
-6.85198307e-01 5.63084245e-01 3.65648687e-01 5.39770424e-01
5.88676870e-01 -1.57609487e+00 -9.89871085e-01 3.66772830e-01
6.24742270e-01 2.77830303e-01 3.72531027e-01 1.20058429e+00
-1.66606963e-01 -2.14742780e-01 -4.87329662e-02 -6.34453118e-01
-8.48434627e-01 5.53072989e-01 3.28702062e-01 -4.19286430e-01
-8.31152618e-01 6.35499418e-01 4.27975178e-01 -3.00041467e-01
3.04656267e-01 3.01150884e-02 -2.32265666e-01 -1.77199885e-01
2.39330739e-01 3.60913694e-01 -1.52162733e-02 -8.36866975e-01
-4.36570078e-01 1.59512639e-01 -5.37167251e-01 4.73213255e-01
1.00900102e+00 -2.33303696e-01 -3.39796126e-01 5.72670698e-01
1.00013256e+00 7.66755566e-02 -8.92710805e-01 -8.86040747e-01
2.55625427e-01 -4.50814605e-01 -9.56231281e-02 -3.63181323e-01
-1.72848463e+00 5.93709707e-01 6.49278089e-02 6.99600339e-01
1.08002460e+00 1.51348040e-01 7.23912001e-01 3.76425982e-01
-1.36276782e-01 -6.44772708e-01 6.15228526e-02 4.32462871e-01
4.88190472e-01 -1.46208477e+00 -1.77092060e-01 -7.43156672e-01
-6.65510654e-01 8.66134882e-01 1.05192494e+00 -2.83341911e-02
8.87964427e-01 -1.70720592e-02 -3.26574236e-01 -5.98004103e-01
-8.49096537e-01 -1.40898764e-01 3.98632705e-01 4.44301397e-01
4.36641246e-01 -7.64627708e-03 2.03396827e-01 5.07914186e-01
1.89658090e-01 -7.79560387e-01 4.90312576e-01 5.30479193e-01
-3.20205778e-01 -8.93513203e-01 2.22279429e-01 9.47826147e-01
-6.17630303e-01 -2.60015905e-01 -5.84444225e-01 7.41781116e-01
-1.52446896e-01 7.83919990e-01 1.69832945e-01 -6.39687657e-01
3.97387713e-01 -1.17932454e-01 -1.25859052e-01 -5.35257995e-01
-3.95533502e-01 -3.31371129e-01 5.60847633e-02 -5.42752326e-01
-9.55282450e-02 -3.92193228e-01 -1.04795456e+00 -7.02949464e-01
-2.23684758e-01 2.67722040e-01 2.01343726e-02 9.55572784e-01
7.54765928e-01 9.38902676e-01 8.52905750e-01 -2.29086906e-01
-3.38123173e-01 -1.10058630e+00 -9.66374755e-01 7.08913743e-01
5.05213022e-01 -7.49496520e-01 -6.89265132e-01 -5.83540916e-01] | [6.7357635498046875, 5.872530937194824] |
955731dd-fac8-4d66-bcea-ec9c89b8cf99 | multi-grained-spatio-temporal-modeling-for | 1908.11618 | null | https://arxiv.org/abs/1908.11618v2 | https://arxiv.org/pdf/1908.11618v2.pdf | Multi-Grained Spatio-temporal Modeling for Lip-reading | Lip-reading aims to recognize speech content from videos via visual analysis of speakers' lip movements. This is a challenging task due to the existence of homophemes-words which involve identical or highly similar lip movements, as well as diverse lip appearances and motion patterns among the speakers. To address these challenges, we propose a novel lip-reading model which captures not only the nuance between words but also styles of different speakers, by a multi-grained spatio-temporal modeling of the speaking process. Specifically, we first extract both frame-level fine-grained features and short-term medium-grained features by the visual front-end, which are then combined to obtain discriminative representations for words with similar phonemes. Next, a bidirectional ConvLSTM augmented with temporal attention aggregates spatio-temporal information in the entire input sequence, which is expected to be able to capture the coarse-gained patterns of each word and robust to various conditions in speaker identity, lighting conditions, and so on. By making full use of the information from different levels in a unified framework, the model is not only able to distinguish words with similar pronunciations, but also becomes robust to appearance changes. We evaluate our method on two challenging word-level lip-reading benchmarks and show the effectiveness of the proposed method, which also demonstrate the above claims. | ['Chenhao Wang'] | 2019-08-30 | null | null | null | null | ['lipreading'] | ['computer-vision'] | [-6.47934899e-02 -7.59486914e-01 -3.93906653e-01 -1.49034306e-01
-8.89647782e-01 -4.31668520e-01 5.95412254e-01 -2.66006202e-01
-7.17992783e-02 3.95233333e-01 7.35515773e-01 3.47288921e-02
2.82623798e-01 -1.48645937e-01 -6.38892949e-01 -8.80921423e-01
2.63664126e-01 -2.65326142e-01 2.83807129e-01 1.39682472e-01
2.90953904e-01 5.30795515e-01 -2.03764033e+00 4.59527880e-01
7.80332804e-01 1.16675079e+00 4.74733591e-01 6.63106859e-01
-4.56659019e-01 5.92419922e-01 -2.77726054e-01 3.87217454e-03
-2.82757908e-01 -4.52763140e-01 -4.33004767e-01 3.12360615e-01
7.78544724e-01 -1.56972677e-01 -2.81388730e-01 1.05623317e+00
8.04299116e-01 1.98177874e-01 7.10359931e-01 -1.08473825e+00
-6.98989034e-01 9.85445008e-02 -6.02932394e-01 1.54926404e-01
4.05853152e-01 5.40626585e-01 8.50294411e-01 -1.10397875e+00
3.06271702e-01 1.57072914e+00 3.59383345e-01 6.11874163e-01
-9.39455032e-01 -6.55699492e-01 4.37142313e-01 5.95229268e-01
-1.33693337e+00 -1.17584002e+00 9.37050343e-01 -5.01755238e-01
4.99158472e-01 1.42739490e-01 5.54870665e-01 1.35218811e+00
8.57422948e-02 1.06274927e+00 1.14583099e+00 -3.25178385e-01
6.24113251e-03 -4.60180081e-02 7.43225887e-02 5.28572798e-01
-2.58378536e-01 2.30015535e-02 -8.89240980e-01 3.21656793e-01
3.00059259e-01 1.30422805e-02 -7.38477767e-01 -3.75939220e-01
-1.36895216e+00 3.89664829e-01 1.55937448e-01 3.12905163e-01
-4.51150686e-01 -1.21243047e-02 4.39071923e-01 -2.31779203e-01
4.33964670e-01 -4.84553933e-01 -3.15456152e-01 -1.68394536e-01
-8.75912130e-01 -2.41617799e-01 2.45160371e-01 7.81136274e-01
6.14257276e-01 1.39030084e-01 -4.21702534e-01 1.15301156e+00
5.83550572e-01 7.04729617e-01 9.33954656e-01 -4.67919946e-01
5.39056182e-01 1.13790564e-01 -4.47056778e-02 -9.06266332e-01
-1.60557196e-01 -1.25225887e-01 -8.79069865e-01 2.98483551e-01
4.34927732e-01 1.25818193e-01 -9.04631853e-01 2.16827869e+00
3.48599881e-01 4.32802081e-01 9.01373774e-02 7.70417869e-01
1.05150557e+00 6.03329241e-01 1.75880238e-01 -4.92309451e-01
1.55769348e+00 -9.64822590e-01 -1.02470398e+00 -1.62258610e-01
8.50491226e-03 -7.51797080e-01 1.27805340e+00 -1.06203131e-01
-1.21932268e+00 -9.99429405e-01 -6.96548283e-01 -1.56420380e-01
-3.68992776e-01 1.77278787e-01 9.12291482e-02 4.56831694e-01
-1.18410194e+00 5.65229878e-02 -5.12807131e-01 -2.62655079e-01
4.45152074e-01 5.06687202e-02 -3.28064650e-01 -5.23709431e-02
-1.02184713e+00 5.16698301e-01 -1.39420614e-01 3.05343866e-01
-6.91623092e-01 -5.23259938e-01 -1.21887386e+00 3.01881246e-02
2.00745091e-01 -4.82813209e-01 9.86074746e-01 -1.00691879e+00
-1.84040809e+00 6.73078120e-01 -1.16060054e+00 9.37888399e-02
5.42539001e-01 1.76328152e-01 -6.14937842e-01 1.49777710e-01
-3.75656821e-02 7.00542390e-01 1.26419377e+00 -1.23959959e+00
-6.06633782e-01 -4.38772291e-01 -4.48058993e-01 4.53710437e-01
-2.34753594e-01 2.49011815e-01 -7.75906384e-01 -7.41152167e-01
-4.25199308e-02 -7.36657739e-01 6.26718998e-01 2.41760924e-01
-3.12468201e-01 -4.05558735e-01 9.79036808e-01 -8.33157361e-01
9.51081097e-01 -2.40949750e+00 6.44897074e-02 -3.57201159e-01
9.64303091e-02 3.68045866e-01 -4.27981198e-01 -2.51520681e-03
5.28872497e-02 2.88779605e-02 1.38770878e-01 -5.82521200e-01
-1.51105644e-02 -1.13316089e-01 -1.75878957e-01 5.83634853e-01
7.24514052e-02 1.12180567e+00 -7.20004261e-01 -6.90693438e-01
4.15194958e-01 7.08416045e-01 -3.90124060e-02 1.78387672e-01
-6.25001937e-02 3.56002569e-01 -2.41046727e-01 8.54040444e-01
7.77683616e-01 -1.61285768e-03 -1.95832819e-01 -4.06654894e-01
-1.19889237e-01 2.61019439e-01 -1.01849794e+00 1.57870495e+00
-7.35805094e-01 1.06281435e+00 3.58302891e-01 -6.82561696e-01
6.76905751e-01 5.29276907e-01 2.72084773e-01 -8.81502569e-01
-2.21008640e-02 1.51379526e-01 -1.72100142e-01 -7.85529554e-01
5.74719422e-02 -9.48521309e-03 3.01100850e-01 3.16827029e-01
-7.57914037e-02 1.64565027e-01 -2.84671873e-01 -2.91149944e-01
4.36303288e-01 -1.34391636e-01 1.19510837e-01 -1.82014287e-01
9.97200012e-01 -8.92856061e-01 4.25717175e-01 2.79386759e-01
-6.21469557e-01 6.42586648e-01 4.67005670e-02 -1.12791739e-01
-6.25249863e-01 -1.22407913e+00 -1.74070358e-01 1.17814589e+00
4.94535893e-01 8.16819444e-02 -6.44560099e-01 -5.72602808e-01
-7.32610375e-02 3.56043875e-01 -6.66321635e-01 -1.20942853e-01
-2.95773566e-01 -2.35276029e-01 3.94419163e-01 4.30538684e-01
5.59833765e-01 -1.44801927e+00 -3.28252375e-01 -7.58402571e-02
-5.78627050e-01 -1.34218609e+00 -1.11480951e+00 -3.03124785e-01
-1.50669336e-01 -7.89897919e-01 -1.12506258e+00 -1.13775432e+00
2.39734948e-01 5.87126255e-01 8.56319308e-01 -1.56675845e-01
-2.02813894e-01 3.60134572e-01 -1.34341463e-01 -1.93034887e-01
-4.34455067e-01 -2.97481030e-01 2.67577589e-01 7.28565633e-01
2.25184903e-01 -4.44912136e-01 -6.97247624e-01 4.21989352e-01
-6.22293055e-01 1.04899414e-01 4.41666633e-01 8.22601438e-01
5.33151209e-01 -1.15666583e-01 7.64266431e-01 2.48984501e-01
6.49215639e-01 -2.52948523e-01 -2.06246808e-01 5.00968039e-01
-3.38969380e-02 -8.34270716e-02 5.42836905e-01 -1.00007212e+00
-1.06785035e+00 -2.08875701e-01 -2.22431168e-01 -5.78952551e-01
-5.01749456e-01 -6.77632689e-02 -8.18922877e-01 -2.45357566e-02
-2.08629272e-03 7.41880834e-01 2.89277643e-01 -5.37899733e-01
3.75304103e-01 1.09532893e+00 7.79582202e-01 -2.53085822e-01
4.24170107e-01 5.74824989e-01 -2.10825428e-01 -1.32491839e+00
-4.15764958e-01 -5.19284070e-01 -5.74006617e-01 -2.85441875e-01
1.14481235e+00 -9.31355119e-01 -1.02383220e+00 1.16946971e+00
-1.25472832e+00 -4.30568635e-01 -5.29874079e-02 5.07940948e-01
-6.29579544e-01 5.53127408e-01 -5.62016547e-01 -8.02781761e-01
-9.74847525e-02 -1.52311838e+00 1.14662981e+00 3.34916174e-01
1.57761410e-01 -1.01056695e+00 -8.80314484e-02 2.80795783e-01
3.58094007e-01 -1.75772756e-01 8.82028818e-01 -3.64742458e-01
-4.33563143e-01 2.59477466e-01 -3.92200828e-01 4.29898858e-01
6.73062682e-01 2.18325970e-03 -1.23649669e+00 -1.72894165e-01
-1.14973262e-01 -3.00274163e-01 9.63088155e-01 7.40436494e-01
1.13168764e+00 -3.61458182e-01 -3.39918137e-01 6.33397520e-01
1.03085899e+00 9.67045650e-02 5.00129879e-01 -2.20486581e-01
8.52872849e-01 9.14398313e-01 3.95433426e-01 5.67649379e-02
4.21510935e-01 1.16739655e+00 4.17748183e-01 -2.93716103e-01
-6.52668297e-01 -3.19613248e-01 6.78899109e-01 8.49879742e-01
3.04731876e-01 -3.39397788e-01 -6.85998440e-01 8.07118952e-01
-1.59413612e+00 -1.25880730e+00 1.81397781e-01 2.10752201e+00
7.73789167e-01 -2.59465456e-01 1.95659533e-01 5.19611686e-02
1.14856291e+00 5.34101486e-01 -7.26757944e-01 -3.36047739e-01
-3.60526770e-01 -8.95369425e-02 -2.30936781e-02 5.99405408e-01
-9.21435595e-01 1.05687535e+00 5.70127249e+00 1.21253383e+00
-1.55865955e+00 1.08144255e-02 5.77820599e-01 -9.56413045e-04
-3.19434643e-01 -5.73951721e-01 -7.09886193e-01 8.35827708e-01
5.97242713e-01 5.22706062e-02 3.61125588e-01 2.93692559e-01
6.61870539e-01 9.72558334e-02 -9.14209604e-01 1.22080898e+00
4.32831138e-01 -1.22706079e+00 -2.59986818e-02 1.30003184e-01
5.15777707e-01 2.51087565e-02 3.40557665e-01 -2.87222164e-03
-1.09915674e-01 -1.13602436e+00 9.62497175e-01 6.07364178e-01
1.05484176e+00 -5.86620927e-01 3.17995489e-01 3.53179157e-01
-1.78502834e+00 -4.34104539e-02 1.98652688e-02 4.07804936e-01
1.58052519e-01 4.10168990e-02 -5.19488156e-01 2.75624901e-01
7.13033497e-01 8.81080925e-01 -3.21553797e-01 8.78812611e-01
-3.92303765e-02 2.87074506e-01 -7.10130250e-03 -8.11604504e-03
-9.72554684e-02 8.70704204e-02 5.09616852e-01 1.46457839e+00
2.24203438e-01 -2.53864557e-01 -3.73434015e-02 7.18642473e-01
5.49620437e-03 1.92961469e-01 -4.20157105e-01 1.45254284e-01
5.32635450e-01 1.03673911e+00 -2.31171474e-01 -2.57575065e-01
-6.15681112e-01 1.01242983e+00 1.76111937e-01 7.69795239e-01
-7.06055224e-01 -8.23693648e-02 1.33660793e+00 9.79966819e-02
4.28627878e-01 -2.67496407e-01 6.79232031e-02 -1.13850272e+00
2.61413962e-01 -1.03776968e+00 -1.98534727e-02 -9.48359430e-01
-1.15192163e+00 6.24752820e-01 -4.17208612e-01 -1.22732031e+00
-3.30235779e-01 -6.02309942e-01 -7.82805860e-01 1.14887285e+00
-2.01156569e+00 -1.27838397e+00 -4.99740541e-01 1.10707819e+00
1.11123443e+00 -1.40606701e-01 4.60839123e-01 4.64659743e-02
-6.78700924e-01 8.94524813e-01 1.77358001e-01 1.49476945e-01
8.87725711e-01 -7.92865217e-01 2.45408684e-01 7.71791339e-01
2.75229931e-01 3.43034357e-01 2.73875564e-01 -3.06849957e-01
-1.23163009e+00 -1.09938395e+00 9.48670805e-01 1.79937799e-02
4.77669179e-01 -5.28774679e-01 -1.10614324e+00 1.58601061e-01
2.23699585e-01 3.54433030e-01 3.99279892e-01 -2.36732006e-01
-5.05858719e-01 -2.45000407e-01 -7.61251450e-01 6.55305743e-01
9.16551232e-01 -1.06452370e+00 -5.20628273e-01 3.81446518e-02
6.21494353e-01 -1.20271593e-01 -3.53848964e-01 3.70636582e-01
8.89554501e-01 -1.07983267e+00 1.13320148e+00 -4.26402628e-01
7.53333867e-02 -2.38661990e-01 -2.13959217e-01 -1.15132606e+00
2.28740182e-02 -6.63903415e-01 -1.80501863e-02 1.56494629e+00
1.32770419e-01 -6.48796260e-01 3.31738561e-01 8.78310055e-02
4.33236510e-02 -7.71109581e-01 -1.03661275e+00 -7.23613620e-01
5.87016791e-02 -5.78727663e-01 5.91942847e-01 6.95783377e-01
-2.47972943e-02 1.83216363e-01 -4.64009255e-01 2.25714132e-01
3.93884927e-01 3.25540870e-01 6.33399189e-01 -1.03021371e+00
-4.74888161e-02 -6.85620189e-01 -3.63573134e-01 -1.45070732e+00
8.23073268e-01 -5.54288447e-01 3.74235809e-01 -1.38370442e+00
2.00503692e-01 -1.37022465e-01 -3.65413040e-01 2.53242135e-01
-3.77166927e-01 2.66186297e-01 1.99473098e-01 3.87935430e-01
-3.94181639e-01 7.85026610e-01 1.35900974e+00 -2.89763987e-01
-2.53620774e-01 2.57524222e-01 -2.81163335e-01 7.99835443e-01
4.55850691e-01 1.38064176e-01 -3.29112411e-01 -4.36152518e-01
-5.25131047e-01 2.45361924e-01 6.30339801e-01 -8.65322471e-01
3.29991966e-01 -3.35376590e-01 3.08971226e-01 -7.24527240e-01
6.88346446e-01 -6.29387081e-01 -3.41180712e-01 1.78176001e-01
-4.17454422e-01 -1.57968581e-01 3.19615215e-01 7.04483151e-01
-4.58595127e-01 2.17686132e-01 1.15094268e+00 2.12240666e-01
-7.53861487e-01 5.77666342e-01 -3.15868974e-01 1.74663588e-01
8.50437820e-01 -4.77217793e-01 -1.37512922e-01 -6.73137248e-01
-5.97479463e-01 -6.82008490e-02 5.83110631e-01 7.83977866e-01
6.23179615e-01 -1.40467012e+00 -7.30554461e-01 4.78864551e-01
1.37507260e-01 -3.72390181e-01 7.55852401e-01 8.66070449e-01
1.71422049e-01 3.81338447e-01 -2.74405330e-01 -9.55110312e-01
-1.51863205e+00 8.44049990e-01 5.32636702e-01 3.28078449e-01
-5.88735878e-01 7.76295125e-01 7.35127687e-01 1.24001756e-01
4.96350795e-01 -4.07789826e-01 -3.78710985e-01 2.75178730e-01
6.42576218e-01 1.09259054e-01 -1.50916159e-01 -1.27428508e+00
-4.93336231e-01 1.25262880e+00 2.00208008e-01 -7.74299800e-02
7.61884332e-01 -7.50876844e-01 4.17023808e-01 6.00612760e-01
1.48379529e+00 6.41223371e-01 -1.59344149e+00 -4.74763662e-01
-4.40354586e-01 -4.43186611e-01 -3.20760347e-02 -5.78501403e-01
-1.12093639e+00 1.44662988e+00 6.44676387e-01 -3.43212187e-02
1.26136732e+00 1.84293464e-01 7.58290410e-01 -4.10080165e-01
-8.67739767e-02 -8.30930054e-01 2.99809635e-01 2.43180335e-01
9.16076899e-01 -1.34802806e+00 -5.83107054e-01 -2.35838696e-01
-7.37826884e-01 1.04669631e+00 2.33262613e-01 5.78785837e-01
4.75648552e-01 4.68323082e-02 3.34467858e-01 3.62609148e-01
-6.38667703e-01 -6.44077897e-01 6.94795847e-01 8.49782467e-01
1.59529537e-01 -4.42386754e-02 2.83870727e-01 3.84112716e-01
7.86680430e-02 -2.36658752e-01 -5.59957176e-02 2.09967002e-01
-3.99182349e-01 -7.95192480e-01 -2.87661791e-01 -2.37339929e-01
-2.76335269e-01 -1.85796633e-01 -2.48544678e-01 5.52806735e-01
2.49218509e-01 1.21908760e+00 1.92848071e-01 -3.14393818e-01
1.22025661e-01 2.29472801e-01 1.47399649e-01 -1.67076200e-01
-3.00525930e-02 2.31918246e-01 -3.43513250e-01 -5.83227932e-01
-3.10259968e-01 -8.80460680e-01 -9.51151609e-01 -1.23777270e-01
-1.48995608e-01 -2.19545737e-01 7.81008899e-01 9.68123436e-01
5.41850746e-01 4.14127588e-01 7.91530609e-01 -1.13861358e+00
-4.73412335e-01 -9.64015603e-01 -5.33868730e-01 6.37873352e-01
1.01162875e+00 -8.07445943e-01 -6.95704818e-01 1.16691276e-01] | [14.30845832824707, 4.972666263580322] |
d8f7c195-04bb-40dd-893c-4b9096442c17 | simple-and-effective-augmentation-methods-for | 2211.10790 | null | https://arxiv.org/abs/2211.10790v2 | https://arxiv.org/pdf/2211.10790v2.pdf | Simple and Effective Augmentation Methods for CSI Based Indoor Localization | Indoor localization is a challenging task. Compared to outdoor environments where GPS is dominant, there is no robust and almost-universal approach. Recently, machine learning (ML) has emerged as the most promising approach for achieving accurate indoor localization. Nevertheless, its main challenge is requiring large datasets to train the neural networks. The data collection procedure is costly and laborious, requiring extensive measurements and labeling processes for different indoor environments. The situation can be improved by Data Augmentation (DA), a general framework to enlarge the datasets for ML, making ML systems more robust and increasing their generalization capabilities. This paper proposes two simple yet surprisingly effective DA algorithms for channel state information (CSI) based indoor localization motivated by physical considerations. We show that the number of measurements for a given accuracy requirement may be decreased by an order of magnitude. Specifically, we demonstrate the algorithm's effectiveness by experiments conducted with a measured indoor WiFi measurement dataset. As little as 10% of the original dataset size is enough to get the same performance as the original dataset. We also showed that if we further augment the dataset with the proposed techniques, test accuracy is improved more than three-fold. | ['Andreas F. Molisch', 'Daoud Burghal', 'Ju-Hyung Lee', 'Omer Gokalp Serbetci'] | 2022-11-19 | null | null | null | null | ['indoor-localization'] | ['computer-vision'] | [ 1.54580712e-01 -2.96995938e-01 -1.45065576e-01 -5.26661277e-01
-9.51232493e-01 -5.38861215e-01 2.97996879e-01 2.42130756e-01
-5.79734504e-01 1.31849480e+00 -1.85247958e-01 -6.62746429e-01
-2.60997564e-01 -8.46602380e-01 -8.18710804e-01 -9.38090205e-01
-1.96300477e-01 1.54160196e-02 2.28092838e-02 1.02055304e-01
3.21730897e-02 7.74121463e-01 -1.36325037e+00 -3.40932369e-01
1.13462663e+00 1.15635741e+00 2.70530909e-01 7.32399404e-01
-7.46347830e-02 5.12720585e-01 -8.94616485e-01 1.64331779e-01
3.09571892e-01 -2.59285659e-01 -4.11185622e-01 -7.64534995e-02
3.33840251e-01 -5.16317636e-02 -4.36154455e-01 8.40769053e-01
7.18517065e-01 2.37037957e-01 1.60226539e-01 -1.15981531e+00
-1.40052348e-01 5.17906845e-01 -1.49150863e-01 4.65539470e-02
5.58335423e-01 -5.57668269e-01 4.58570421e-01 -5.73062122e-01
-1.18311616e-02 5.87344110e-01 1.01087654e+00 1.34644955e-01
-1.06213355e+00 -8.01788330e-01 -5.90417068e-03 4.52587530e-02
-1.87627435e+00 -3.94265026e-01 7.99070477e-01 7.50847012e-02
3.64424527e-01 5.49294651e-01 5.11674404e-01 1.00461090e+00
9.84627679e-02 3.50791693e-01 1.17095697e+00 -6.56614482e-01
5.19993782e-01 2.86189049e-01 4.08881456e-02 4.44088548e-01
6.44038677e-01 -1.08333997e-01 -2.11381257e-01 -2.31798202e-01
6.78437352e-01 2.61329591e-01 -3.05732697e-01 -5.02349794e-01
-1.15535665e+00 4.64934260e-01 7.53642976e-01 6.14511192e-01
-2.98737228e-01 2.58557200e-01 4.77443375e-02 2.15716094e-01
1.54183999e-01 7.58751810e-01 -6.03025079e-01 -3.16688150e-01
-1.10373068e+00 -4.53698337e-02 7.66700983e-01 1.05822146e+00
9.21843410e-01 6.21819757e-02 3.45171273e-01 7.07174182e-01
-2.21696775e-03 7.46465325e-01 2.24443778e-01 -8.54744673e-01
5.44751942e-01 2.02008724e-01 4.81257826e-01 -1.22288656e+00
-5.98263979e-01 -1.02709436e+00 -1.04889119e+00 -3.84691298e-01
5.74580729e-01 -5.82424939e-01 -6.67154312e-01 1.79057360e+00
1.34450093e-01 2.65277058e-01 -8.55928846e-03 3.82338017e-01
2.72139490e-01 6.23988628e-01 -2.60482460e-01 -2.19309911e-01
7.36503839e-01 -6.38476312e-01 -7.88709879e-01 -2.90704280e-01
6.98518932e-01 -7.79403329e-01 8.20862591e-01 4.01033342e-01
-6.37476683e-01 -7.40654111e-01 -1.38516808e+00 5.23777962e-01
-4.71350729e-01 2.55815089e-01 9.91306722e-01 1.27574730e+00
-9.25040781e-01 4.19842869e-01 -9.43473876e-01 -5.44924319e-01
8.38519484e-02 9.04742062e-01 -4.20865983e-01 -3.39266568e-01
-1.07653308e+00 5.57255447e-01 2.58182943e-01 3.24629813e-01
-2.01268151e-01 -2.89150476e-01 -8.37383866e-01 3.66695300e-02
2.35785589e-01 -3.33122373e-01 1.01098967e+00 -4.22715098e-01
-1.40576053e+00 -2.06338972e-01 -4.45171684e-01 -4.92018342e-01
2.12720975e-01 -2.13101462e-01 -7.60705411e-01 -3.10282260e-01
2.51437068e-01 1.10993341e-01 2.82922596e-01 -1.40972614e+00
-6.47659540e-01 -2.77994663e-01 1.89477071e-01 -1.74326017e-01
-4.57945913e-01 -5.35353839e-01 -3.65807742e-01 -4.04956639e-01
6.88522518e-01 -1.18004787e+00 -7.17888832e-01 -4.43955332e-01
-3.03435475e-01 4.14163172e-01 5.53096652e-01 -3.81948143e-01
1.30182207e+00 -1.97942781e+00 -6.12617016e-01 7.70468652e-01
-4.96108793e-02 3.17322090e-02 8.24717209e-02 4.22078043e-01
3.03971529e-01 3.75535414e-02 -2.86192633e-02 -4.60397542e-01
-1.61163077e-01 5.41384816e-01 1.08325526e-01 6.85545743e-01
-4.20911193e-01 4.27678198e-01 -9.67033029e-01 -6.36901706e-02
5.18539965e-01 4.13708270e-01 -6.41346455e-01 8.48328397e-02
5.05609095e-01 9.09731567e-01 -4.64056164e-01 5.81188917e-01
8.71526659e-01 -2.85941392e-01 3.15589398e-01 -1.89856827e-01
-1.35604262e-01 3.27498853e-01 -1.65772688e+00 1.56969500e+00
-1.09527016e+00 7.83293009e-01 -1.82368770e-01 -9.89838302e-01
1.14367270e+00 2.59936303e-01 6.17139041e-01 -7.27880418e-01
1.75479978e-01 5.46746194e-01 -8.34917463e-03 -3.57552707e-01
4.89110500e-01 2.67170101e-01 -4.46877033e-01 1.49630681e-01
-1.41156048e-01 1.35408849e-01 -7.38527849e-02 -2.16963470e-01
1.21777415e+00 -1.97015241e-01 4.78485286e-01 -2.52183020e-01
6.87041402e-01 -3.82773191e-01 5.61392665e-01 1.24352646e+00
-8.07345137e-02 1.30186483e-01 -3.12239766e-01 -3.80753756e-01
-7.90017426e-01 -9.14368868e-01 -3.57799739e-01 7.61315048e-01
4.83374983e-01 -3.49847823e-01 -6.20348752e-01 -4.64095682e-01
-3.00069582e-02 5.34717202e-01 -3.64896476e-01 1.20603003e-01
-5.26970208e-01 -8.45673144e-01 7.73968995e-01 5.73235869e-01
9.60618556e-01 -3.52022916e-01 -2.21456781e-01 3.74642402e-01
-4.14027035e-01 -1.37994552e+00 1.61015898e-01 5.86371720e-01
-7.38088846e-01 -7.21012235e-01 -6.02045774e-01 -6.40673757e-01
9.57304180e-01 6.22048140e-01 4.94741887e-01 3.60588878e-02
9.75646973e-02 2.69491911e-01 -3.58947277e-01 -6.36572838e-02
-1.24148957e-01 3.67111683e-01 5.67436337e-01 -3.89579870e-02
2.86736712e-02 -8.62812519e-01 -4.38145310e-01 2.97295302e-01
-4.14015323e-01 -4.10713613e-01 7.04094172e-01 7.43079126e-01
2.86981791e-01 4.09923196e-01 6.74168944e-01 -7.88071811e-01
2.83877134e-01 -4.43035781e-01 -5.72394431e-01 4.09383141e-02
-5.64070165e-01 -1.88271620e-03 9.74961102e-01 -3.53972852e-01
-9.63384390e-01 3.52563530e-01 -4.42627668e-01 2.16468051e-01
-4.72580314e-01 4.34648633e-01 -3.50463867e-01 -9.73594189e-01
8.11181784e-01 3.17975849e-01 -5.75320661e-01 -4.76001799e-01
3.83472800e-01 9.22586143e-01 5.25044560e-01 -5.75447977e-01
9.14160192e-01 3.96822751e-01 2.39348546e-01 -1.04623938e+00
-8.24475050e-01 -6.01330698e-01 -8.41233969e-01 1.86094400e-02
2.21930996e-01 -7.21566617e-01 -7.34803557e-01 5.34271896e-02
-7.82373309e-01 -1.15072310e-01 1.02299713e-01 7.39607275e-01
-1.55332163e-01 3.74037385e-01 -2.10639298e-01 -9.65361357e-01
1.55937195e-01 -1.07136369e+00 6.34493053e-01 3.18382978e-01
7.43042901e-02 -9.96119499e-01 2.45721955e-02 2.23352641e-01
8.45240712e-01 2.79305637e-01 4.25189763e-01 -5.40712118e-01
-5.57684720e-01 -7.98993886e-01 -1.49071906e-02 2.30721310e-01
6.37671769e-01 -4.41426337e-01 -1.07597995e+00 -3.88408184e-01
4.51682284e-02 2.10754961e-01 2.74390250e-01 3.21273297e-01
1.41068244e+00 -3.70940596e-01 -6.10420942e-01 6.55850887e-01
1.52925766e+00 5.15557587e-01 5.50140619e-01 4.41250741e-01
5.82650602e-01 -9.34936032e-02 6.62820339e-01 3.89321387e-01
3.35238904e-01 8.39653850e-01 2.13677809e-01 -4.57754195e-01
7.62251094e-02 -2.26194501e-01 -9.80802402e-02 7.97896802e-01
-1.14637703e-01 -3.27859372e-01 -8.75618815e-01 4.03976709e-01
-1.69992840e+00 -7.84857690e-01 -1.57498509e-01 2.47572470e+00
3.47703606e-01 1.91478580e-01 -2.79475629e-01 7.22607911e-01
4.26501185e-01 -4.80897613e-02 -4.96021435e-02 -3.89654897e-02
1.58925608e-01 2.67815977e-01 1.00805438e+00 7.31749535e-01
-1.23221588e+00 5.80304027e-01 6.37223673e+00 5.93028963e-01
-1.28823578e+00 2.81201769e-02 3.04431856e-01 3.45883459e-01
1.30317703e-01 -7.87758753e-02 -7.62320757e-01 5.40540040e-01
1.01608801e+00 2.36600563e-01 2.72291303e-01 1.13980567e+00
2.75926769e-01 -4.17078346e-01 -8.24974537e-01 1.26472855e+00
-8.05686861e-02 -1.31237793e+00 -4.76014823e-01 2.70761997e-01
9.00267243e-01 -3.05605624e-02 -1.30164325e-01 4.36311096e-01
-1.36866242e-01 -6.01792395e-01 2.63272673e-01 2.00668275e-01
4.88181561e-01 -8.56297493e-01 1.26296210e+00 4.62053001e-01
-1.44816005e+00 -3.77335936e-01 -4.78501171e-01 -4.88718182e-01
1.49408579e-01 7.99905837e-01 -9.42954600e-01 7.74979234e-01
5.49089789e-01 3.17650348e-01 -8.07929695e-01 1.57785940e+00
-1.25008255e-01 6.63812518e-01 -6.74690723e-01 -3.36851776e-01
2.69242436e-01 3.96740250e-02 1.38203591e-01 1.11952257e+00
7.12748230e-01 -5.72214536e-02 4.42976385e-01 1.48356602e-01
-7.21498812e-03 -5.61891012e-02 -8.53063524e-01 5.36553502e-01
1.00492334e+00 1.14894164e+00 -6.12084448e-01 -1.51262611e-01
-2.88340330e-01 9.89752293e-01 -5.26852645e-02 5.20001888e-01
-9.53561664e-01 -6.43713832e-01 2.87227154e-01 2.61524897e-02
8.47340301e-02 -7.26388335e-01 -3.01623791e-01 -9.11289752e-01
-8.19002539e-02 -4.51384515e-01 -1.57618627e-01 -2.09307298e-01
-7.85323441e-01 6.41954839e-01 -2.89402574e-01 -1.29346657e+00
-3.02778453e-01 -3.24715674e-01 -1.39950141e-01 6.35958254e-01
-1.40638947e+00 -9.46175337e-01 -7.95931458e-01 5.90783298e-01
1.92479849e-01 1.49380639e-01 1.08070767e+00 1.05315423e+00
-4.84970301e-01 8.87044549e-01 7.04071283e-01 5.58556579e-02
4.83040959e-01 -1.11208403e+00 1.30261749e-01 1.07834756e+00
4.06949848e-01 8.72716725e-01 8.38230669e-01 -3.40920538e-01
-1.38732433e+00 -1.04199302e+00 9.40329790e-01 -2.56483406e-01
4.17643219e-01 -6.54100180e-01 -4.26751494e-01 5.94855845e-01
-3.09206337e-01 1.71996474e-01 8.81624162e-01 4.08638299e-01
1.65151328e-01 -5.23813248e-01 -1.23744357e+00 4.64791298e-01
9.49554324e-01 -3.57729882e-01 -9.01505426e-02 3.80846500e-01
3.65823656e-01 -4.86280292e-01 -9.17994559e-01 5.30811608e-01
5.76193511e-01 -7.78038561e-01 9.28510308e-01 8.70882347e-02
-5.91745973e-01 -6.44998431e-01 -7.22296119e-01 -1.02835977e+00
-2.87351608e-01 -4.04066890e-01 -9.31755602e-02 1.16482186e+00
4.15577203e-01 -7.88445890e-01 1.04324913e+00 3.32135439e-01
4.97555919e-02 -6.15328848e-01 -1.05032372e+00 -1.11369026e+00
-5.72103024e-01 -8.57564569e-01 7.77945399e-01 8.46267045e-01
-1.66770443e-01 8.65444019e-02 -5.64598918e-01 7.36173570e-01
6.47347510e-01 -1.77998111e-01 1.24153101e+00 -1.19317234e+00
-3.14341217e-01 2.12657750e-01 -5.56379616e-01 -1.53080463e+00
-2.49957144e-01 -4.96017188e-01 3.07244211e-01 -1.58399546e+00
-3.13126028e-01 -1.29826510e+00 -4.73732084e-01 3.53482902e-01
1.09877184e-01 4.28880960e-01 -6.89871088e-02 -1.08261760e-02
-6.18661702e-01 3.27114940e-01 8.15225065e-01 6.34967908e-02
-4.15908813e-01 7.50599384e-01 -4.64095354e-01 6.92568481e-01
9.83349621e-01 -3.66068065e-01 -3.69382858e-01 -2.67419070e-01
3.64393331e-02 -2.64795143e-02 1.53072864e-01 -1.86012781e+00
4.47699040e-01 3.64163145e-02 7.94880152e-01 -4.98714238e-01
6.26257360e-01 -1.26121354e+00 3.50096613e-01 6.03614986e-01
-7.41105387e-03 -1.55262932e-01 3.47451776e-01 5.50790846e-01
-1.21471733e-01 9.97530576e-03 5.51258206e-01 3.15211505e-01
-7.18588531e-01 9.84088108e-02 -3.42970669e-01 -7.25291848e-01
8.26095760e-01 -3.53984654e-01 -4.64350097e-02 -6.88221872e-01
-6.57955706e-01 -2.27780148e-01 2.90225863e-01 9.32367444e-02
3.17920029e-01 -1.39169586e+00 9.79923010e-02 3.38155508e-01
-3.81887294e-02 -2.94083953e-01 1.46424800e-01 8.18817914e-01
-4.94079530e-01 7.60569811e-01 3.44794504e-02 -6.99892342e-01
-1.13085151e+00 3.45955700e-01 3.50874722e-01 -2.46303216e-01
-2.85229236e-01 7.46089995e-01 -4.09067512e-01 -7.10789442e-01
5.28243363e-01 -5.41174352e-01 5.85950129e-02 -5.35508573e-01
3.94523114e-01 4.92104203e-01 2.46977076e-01 -3.97171676e-01
-5.33323884e-01 6.13469779e-01 3.37660253e-01 1.92464050e-02
1.09111667e+00 -4.35171932e-01 1.44232571e-01 2.68801361e-01
1.11109293e+00 6.44914985e-01 -8.88451695e-01 -1.19166091e-01
8.91573206e-02 -6.22972071e-01 -2.14698073e-02 -6.80769384e-01
-6.37966692e-01 6.76600516e-01 9.45923865e-01 4.12706912e-01
1.11524117e+00 -2.25955412e-01 7.14011014e-01 8.47602069e-01
1.29768801e+00 -7.16300607e-01 -3.88800174e-01 3.97790939e-01
2.31050968e-01 -1.26161706e+00 7.02188769e-03 -5.44390082e-01
2.06131145e-01 9.83633459e-01 2.28543714e-01 -5.00997230e-02
6.33465052e-01 2.02666610e-01 -1.63833387e-02 3.27472180e-01
1.83668271e-01 -6.46632388e-02 1.13572948e-01 6.61580563e-01
3.10926884e-01 2.46588901e-01 -6.22420274e-02 6.02838516e-01
-4.08380210e-01 -1.05519488e-01 5.05404711e-01 1.06735182e+00
-6.74408436e-01 -1.46950233e+00 -6.92410171e-01 4.80714917e-01
-3.76762629e-01 1.55626774e-01 2.81325370e-01 9.52978730e-01
3.82419407e-01 1.37018847e+00 -3.86155635e-01 -6.46221936e-01
1.89840242e-01 -3.79658133e-01 5.72332263e-01 -3.65623236e-01
-4.39613685e-02 -1.25344470e-01 9.82299075e-02 -6.19702995e-01
-4.47371036e-01 -3.44289422e-01 -1.17781997e+00 -4.10012007e-01
-6.19745135e-01 4.89992410e-01 1.00468349e+00 1.13525653e+00
2.70765662e-01 6.84669256e-01 9.16389287e-01 -9.98781025e-01
-1.03188172e-01 -6.74585581e-01 -5.27836323e-01 -1.79551259e-01
4.89543974e-01 -7.76989043e-01 -3.49524170e-01 -3.78545880e-01] | [6.415460109710693, 0.9722769260406494] |
2d42b481-2ddd-4d53-a0f7-9d86ed10e32d | wearable-based-human-activity-recognition | 2212.02233 | null | https://arxiv.org/abs/2212.02233v1 | https://arxiv.org/pdf/2212.02233v1.pdf | Wearable-based Human Activity Recognition with Spatio-Temporal Spiking Neural Networks | We study the Human Activity Recognition (HAR) task, which predicts user daily activity based on time series data from wearable sensors. Recently, researchers use end-to-end Artificial Neural Networks (ANNs) to extract the features and perform classification in HAR. However, ANNs pose a huge computation burden on wearable devices and lack temporal feature extraction. In this work, we leverage Spiking Neural Networks (SNNs)--an architecture inspired by biological neurons--to HAR tasks. SNNs allow spatio-temporal extraction of features and enjoy low-power computation with binary spikes. We conduct extensive experiments on three HAR datasets with SNNs, demonstrating that SNNs are on par with ANNs in terms of accuracy while reducing up to 94% energy consumption. The code is publicly available in https://github.com/Intelligent-Computing-Lab-Yale/SNN_HAR | ['Priyadarshini Panda', 'Youngeun Kim', 'Hyoungseob Park', 'Ruokai Yin', 'Yuhang Li'] | 2022-11-14 | null | null | null | null | ['human-activity-recognition', 'human-activity-recognition'] | ['computer-vision', 'time-series'] | [ 2.63528854e-01 -4.99765426e-01 -4.53446694e-02 -2.51665056e-01
-2.37876981e-01 -4.45740581e-01 1.11497208e-01 2.54231114e-02
-5.22339821e-01 9.50531781e-01 3.64142470e-02 -2.01731529e-02
5.25740311e-02 -8.31412435e-01 -7.19704509e-01 -6.92342520e-01
-2.69571364e-01 -3.72983634e-01 -1.86601076e-02 2.25458711e-01
-1.82513222e-01 3.10477525e-01 -1.48232746e+00 2.84846991e-01
5.21616638e-01 1.47679746e+00 -2.04581618e-01 6.72297895e-01
4.44186658e-01 7.15207160e-01 -3.65461767e-01 6.57653287e-02
1.62859671e-02 -5.70572674e-01 -1.06479600e-01 -5.70012271e-01
-3.21146488e-01 9.23931897e-02 -7.32249439e-01 7.05008745e-01
7.12019861e-01 8.93136337e-02 2.27332279e-01 -1.46274877e+00
-5.20670176e-01 3.41835409e-01 -1.20421238e-01 6.83807731e-01
4.39748138e-01 4.56920743e-01 5.57044566e-01 -9.51450944e-01
1.72311977e-01 3.31130385e-01 1.27787912e+00 6.26792729e-01
-1.36267102e+00 -7.92230725e-01 -4.11321342e-01 9.25422683e-02
-1.60071445e+00 -7.64335275e-01 6.98190033e-01 -1.72217429e-01
1.34881151e+00 2.79663503e-01 1.20259535e+00 1.84900188e+00
5.12172461e-01 8.66919100e-01 9.81345773e-01 2.02547118e-01
8.11170816e-01 -5.18116593e-01 3.73864323e-01 4.41843420e-01
6.93882942e-01 -1.16696276e-01 -1.08849263e+00 -3.03031296e-01
9.47868586e-01 6.27107382e-01 -1.28580645e-01 2.67479420e-01
-1.34265339e+00 2.47893229e-01 6.08592749e-01 5.08362055e-01
-9.76197958e-01 6.75976574e-01 2.13152528e-01 -1.86061949e-01
1.50122151e-01 2.01716155e-01 -4.32213306e-01 -6.22671902e-01
-8.39475274e-01 1.30630329e-01 8.69150877e-01 7.14897394e-01
1.57869384e-01 2.87865639e-01 -2.52713412e-01 5.43982804e-01
-1.38230473e-01 7.32805908e-01 6.51500046e-01 -8.84583890e-01
7.88245872e-02 6.78081512e-01 1.27661303e-01 -8.51160526e-01
-9.71967578e-01 -3.74232471e-01 -1.43578696e+00 -3.18047047e-01
3.97132337e-01 -3.07431251e-01 -7.08243787e-01 1.70292628e+00
-1.46154791e-01 6.19740129e-01 -1.38491660e-01 6.53352797e-01
6.09672189e-01 4.71911460e-01 3.31875652e-01 -2.59099275e-01
1.43174827e+00 -7.31079161e-01 -8.63143325e-01 -4.20958102e-01
3.41256976e-01 3.69469970e-01 9.56884742e-01 3.50486487e-01
-1.19363797e+00 -4.04770821e-01 -1.25325644e+00 1.11495689e-01
-5.32977521e-01 -2.15420835e-02 7.40198076e-01 6.40250921e-01
-9.29365039e-01 7.38894045e-01 -1.54754162e+00 -6.07612610e-01
9.73335087e-01 6.75314069e-01 2.53381412e-02 5.31379044e-01
-9.31767583e-01 2.31853902e-01 7.33212084e-02 2.45431408e-01
-5.81223905e-01 -4.73735780e-01 -6.16531134e-01 2.40948364e-01
-2.89231360e-01 -8.59598696e-01 1.29479003e+00 -6.56683326e-01
-1.37063885e+00 4.40050632e-01 -5.10288656e-01 -9.37540412e-01
-6.66729510e-02 -6.40106276e-02 -8.41620862e-01 -6.89887479e-02
-2.48343021e-01 2.40856305e-01 3.23969424e-01 -3.62192929e-01
-7.69393146e-02 -5.68853438e-01 -7.06377745e-01 -4.13936198e-01
-6.41214132e-01 -2.02550203e-01 1.12473033e-01 -8.29476953e-01
2.59329695e-02 -1.06539607e+00 -9.93283391e-02 -6.72713444e-02
-1.47692189e-01 -1.79163367e-01 6.15148425e-01 -5.91092825e-01
1.47426045e+00 -2.16545439e+00 -4.33669776e-01 2.26842970e-01
3.91226113e-01 2.48432264e-01 7.86780119e-02 1.88557118e-01
1.35805696e-01 -1.28139168e-01 -4.44942206e-01 -1.66440591e-01
-4.61159870e-02 2.65276939e-01 -2.10180193e-01 4.23007131e-01
2.02870861e-01 1.68282032e+00 -7.85638332e-01 7.45862708e-05
5.54391695e-03 7.21232116e-01 -9.49001312e-02 2.93205958e-02
3.73756550e-02 4.77061689e-01 -4.42702651e-01 1.05893803e+00
1.73531443e-01 -7.72790015e-01 2.53666252e-01 -6.34672567e-02
3.12251188e-02 3.72112483e-01 -7.67164290e-01 1.87716961e+00
-2.73739874e-01 6.60866141e-01 -3.93293768e-01 -9.64662850e-01
9.33555603e-01 3.69324535e-01 7.88502991e-01 -1.31125593e+00
5.47569335e-01 3.20408136e-01 -6.92666918e-02 -3.81152332e-01
-1.95844188e-01 5.53473175e-01 -3.78360122e-01 5.60085118e-01
8.90991315e-02 8.89904976e-01 1.23099476e-01 -3.68429273e-01
1.89488697e+00 1.31941006e-01 4.67024833e-01 -1.97274253e-01
-2.04797953e-01 -4.03733432e-01 8.07144701e-01 8.35803449e-01
-6.05952442e-01 3.48355860e-01 6.40188996e-03 -7.91383386e-01
-7.75621057e-01 -1.40872407e+00 -7.29555413e-02 9.65778410e-01
9.66852084e-02 -4.37297016e-01 -8.69221985e-01 -7.46359751e-02
-7.02657178e-02 3.93589109e-01 -6.04495764e-01 -1.02386869e-01
-5.47518015e-01 -1.08470750e+00 1.07073057e+00 1.14357913e+00
8.13751221e-01 -1.49708402e+00 -1.47950101e+00 5.47109962e-01
-2.43429437e-01 -1.32876420e+00 -1.25609428e-01 6.53834224e-01
-9.87876058e-01 -4.79688644e-01 -5.02597749e-01 -5.39235950e-01
3.65979761e-01 -1.70221701e-01 9.11562920e-01 -1.39370576e-01
-6.49164557e-01 3.24211061e-01 2.36704610e-02 -8.03640664e-01
6.12943053e-01 2.55837917e-01 4.37912256e-01 9.62066129e-02
8.65296006e-01 -1.51376677e+00 -1.02030134e+00 1.28815666e-01
-5.84313393e-01 1.24246515e-02 4.81400609e-01 3.83846819e-01
7.90588617e-01 -3.22119206e-01 8.52195799e-01 -3.05341870e-01
5.53536236e-01 -7.13981807e-01 -3.44832361e-01 -3.61089259e-02
-5.29249430e-01 -6.45584315e-02 7.80840993e-01 -6.88830078e-01
-4.48916286e-01 4.35493648e-01 -5.28350519e-03 -1.82279631e-01
-2.13287950e-01 2.30435833e-01 2.58141041e-01 3.61138210e-02
9.14239228e-01 5.42922616e-01 -4.40775424e-01 -2.69307971e-01
-4.77324933e-01 6.55055165e-01 8.08749735e-01 -3.00937980e-01
2.25053921e-01 5.43627203e-01 3.92904654e-02 -6.55201614e-01
-6.73341632e-01 -2.39168465e-01 -3.12735200e-01 -5.44414632e-02
8.71407568e-01 -1.11917508e+00 -1.20869410e+00 6.39080763e-01
-1.04353619e+00 -6.78883672e-01 -3.56984794e-01 3.52607191e-01
-6.23497069e-01 -3.91423702e-01 -9.09114242e-01 -1.18239439e+00
-9.64959979e-01 -2.53159404e-01 7.55444229e-01 8.08227539e-01
-7.55299687e-01 -6.52858973e-01 4.50995052e-03 7.21525103e-02
7.88477004e-01 5.46414852e-01 1.66869104e-01 -6.51254296e-01
-3.31014782e-01 -1.98240876e-01 2.55703032e-02 -1.95587158e-01
-8.78848583e-02 -4.92026597e-01 -1.51566768e+00 1.18899690e-02
-6.98493561e-04 -2.23138675e-01 5.56648910e-01 6.67984426e-01
1.82282448e+00 -3.82390887e-01 -6.06248140e-01 8.24370801e-01
1.26048470e+00 2.81539053e-01 9.05117393e-01 7.15089887e-02
3.50661963e-01 -1.93469062e-01 -1.84100419e-01 7.81136215e-01
3.36601496e-01 4.47287679e-01 1.62676290e-01 -3.10743283e-02
4.33013104e-02 -2.13927776e-01 5.26567578e-01 9.05980527e-01
-4.21648324e-01 -3.48363161e-01 -1.01576030e+00 6.02769196e-01
-2.23408008e+00 -1.10236704e+00 -1.11025713e-01 2.01505351e+00
6.84260488e-01 1.38393164e-01 5.65648973e-01 3.33965033e-01
3.95420849e-01 -2.43519470e-02 -9.97673213e-01 -2.83250123e-01
-2.35300601e-01 5.96872449e-01 4.25037473e-01 -3.90850961e-01
-1.00573206e+00 2.00849444e-01 5.65407848e+00 2.36396372e-01
-1.08142138e+00 3.74103546e-01 6.24943614e-01 -6.81089103e-01
2.32339472e-01 -6.28079832e-01 -5.71025312e-01 1.07826185e+00
1.79569006e+00 -3.27271298e-02 9.07958806e-01 6.30198002e-01
2.17179701e-01 -5.79023883e-02 -9.42593217e-01 1.56392264e+00
-2.49001890e-01 -1.39071858e+00 -5.12909889e-01 9.13425684e-02
5.50740540e-01 4.85799819e-01 -1.37908354e-01 2.50959903e-01
-2.24017575e-02 -1.35470259e+00 3.71559739e-01 9.51946199e-01
6.40791416e-01 -6.38687134e-01 6.14627779e-01 3.98742855e-01
-1.50431061e+00 -3.17430019e-01 -1.11546041e-02 -4.50987756e-01
6.55356124e-02 8.93337488e-01 -2.28297338e-01 -1.46086946e-01
1.32181191e+00 7.94948816e-01 -5.33280969e-01 1.01476121e+00
1.77643299e-01 7.98643470e-01 -6.57381535e-01 -5.03945470e-01
-2.34875783e-01 2.21980568e-02 8.98256823e-02 1.41072571e+00
6.68645203e-01 4.26219672e-01 -1.96586058e-01 1.13994586e+00
-4.53325957e-01 -5.22117972e-01 -7.55266726e-01 -4.89849389e-01
6.76194310e-01 1.37374079e+00 -1.01253200e+00 -8.46235305e-02
-2.45895341e-01 1.27589405e+00 9.79746953e-02 4.09787148e-01
-1.02867413e+00 -7.95603633e-01 6.26865029e-01 2.39334911e-01
5.04442751e-02 -4.89317715e-01 -9.06531096e-01 -1.09317493e+00
4.02169853e-01 -2.70604312e-01 1.69921428e-01 -9.23890233e-01
-1.20249856e+00 4.64273334e-01 -7.05769658e-01 -1.13661003e+00
-2.53327042e-01 -4.75701809e-01 -6.98942363e-01 4.12098765e-01
-9.50240493e-01 -8.05506408e-01 -7.76963472e-01 7.78653979e-01
3.09767038e-01 3.36787403e-01 1.09838724e+00 3.18348199e-01
-8.98924828e-01 6.10454917e-01 -3.25251557e-02 4.55947667e-01
-6.64264429e-03 -7.62362063e-01 7.20115721e-01 7.67649829e-01
2.27678642e-01 7.27658331e-01 3.01857233e-01 -4.98618662e-01
-1.59449053e+00 -1.15184736e+00 1.04496372e+00 -7.18259573e-01
5.83528459e-01 -8.83148193e-01 -8.39964151e-01 7.08768368e-01
1.43574327e-01 8.40704262e-01 1.01538217e+00 -2.48437494e-01
-9.04517472e-02 -1.44114092e-01 -1.19821858e+00 6.58626854e-01
1.83662987e+00 -6.24807060e-01 -1.84747681e-01 4.93056290e-02
2.96792358e-01 4.93684299e-02 -1.11302507e+00 4.19267982e-01
1.12674606e+00 -7.59034157e-01 9.73354757e-01 -3.22100073e-01
-1.78149212e-02 -2.37211689e-01 -7.57355243e-02 -9.26739991e-01
-3.62313002e-01 -7.44722664e-01 -1.04820704e+00 7.67065585e-01
2.56156594e-01 -7.56122947e-01 7.91919291e-01 6.35455430e-01
2.28611752e-01 -8.94677401e-01 -1.03822577e+00 -1.06818378e+00
-7.44844079e-01 -5.39281368e-01 4.77684319e-01 7.43813396e-01
3.55665833e-01 2.75323510e-01 -1.95061579e-01 -5.89417033e-02
5.98189116e-01 -4.75198552e-02 1.57939449e-01 -1.40111327e+00
-2.01998651e-01 -1.99570641e-01 -4.81316626e-01 -4.33090419e-01
-1.98205143e-01 -6.72285616e-01 3.02584469e-02 -1.37629592e+00
1.62670225e-01 5.66917956e-02 -1.02451491e+00 1.09847796e+00
3.70623708e-01 8.58589768e-01 -1.36461943e-01 6.48142397e-02
-9.88467395e-01 4.72912103e-01 3.30745727e-01 5.88031486e-02
-4.84343469e-01 -4.33841087e-02 -5.52142262e-01 7.42828369e-01
1.35973608e+00 -6.40948594e-01 -9.76460576e-02 -2.43654907e-01
2.92101175e-01 2.53437143e-02 7.48981535e-01 -1.85998666e+00
6.01938725e-01 9.97757986e-02 1.29785395e+00 -2.14811966e-01
6.01990581e-01 -7.33554244e-01 3.43905330e-01 8.51746202e-01
-1.34527534e-01 2.43900463e-01 3.67151678e-01 5.90818882e-01
4.02301401e-01 5.88151813e-01 4.45547700e-01 1.13445729e-01
-5.04967153e-01 4.03245121e-01 -7.03335166e-01 -1.29822090e-01
9.77685392e-01 -4.41635400e-01 -5.09549439e-01 -8.55477303e-02
-8.44074428e-01 -1.53187588e-01 1.26135483e-01 3.02693486e-01
5.60010374e-01 -1.52717221e+00 -1.86182782e-01 5.04352152e-01
3.01852703e-01 -4.37125236e-01 2.38084942e-01 1.28017116e+00
-3.67752351e-02 4.49745268e-01 -6.47720933e-01 -4.70100135e-01
-8.10208142e-01 3.37612629e-01 3.30728710e-01 5.36996126e-03
-6.72622144e-01 4.56898063e-01 -5.47547996e-01 -9.34365168e-02
3.15900415e-01 -4.53056067e-01 1.51099727e-01 -2.53438920e-01
6.97003603e-01 5.41677952e-01 1.62253141e-01 1.26223136e-02
-7.72173882e-01 9.17128548e-02 6.31311476e-01 4.87165200e-03
1.68751562e+00 1.21672161e-01 1.35062128e-01 8.06894958e-01
9.47552562e-01 -6.14009082e-01 -1.17942238e+00 -2.00170353e-01
2.50751823e-01 1.12590104e-01 -7.99944848e-02 -8.66652608e-01
-8.83406401e-01 8.73081446e-01 1.16993427e+00 2.36258402e-01
1.46148312e+00 -3.05190533e-01 1.49585843e+00 7.24291384e-01
5.36417544e-01 -1.13995349e+00 9.91810560e-02 2.97059834e-01
5.28758287e-01 -7.76433468e-01 -4.18793619e-01 2.33464032e-01
-2.64463514e-01 9.55354154e-01 5.89320421e-01 -5.21209419e-01
7.96552002e-01 8.16273272e-01 -5.21422327e-01 -1.73566610e-01
-8.20820391e-01 -1.15684718e-01 -7.81683028e-02 6.56806290e-01
2.70114094e-01 2.92831361e-01 -8.18527490e-02 1.44565129e+00
-1.52465394e-02 9.08496797e-01 2.27775082e-01 1.14024186e+00
-2.16510028e-01 -4.51288939e-01 4.83758114e-02 9.48402405e-01
-4.25753176e-01 -7.99662024e-02 -2.12292477e-01 7.74897076e-03
1.29397452e-01 9.56886828e-01 3.12118858e-01 -7.86159754e-01
3.06520313e-01 4.23940957e-01 4.95012820e-01 -1.59523562e-01
-8.32638741e-01 -2.52432585e-01 -1.22140013e-01 -1.09027243e+00
-3.07090133e-01 -6.04802191e-01 -1.56772220e+00 -4.89726841e-01
5.22458136e-01 -3.64360392e-01 7.94104993e-01 8.26183081e-01
1.08622706e+00 7.01638162e-01 3.06566030e-01 -9.19974685e-01
-1.06595866e-01 -8.31238151e-01 -6.49800360e-01 2.75801957e-01
2.89500058e-01 -3.20764571e-01 -9.03554484e-02 2.18651533e-01] | [8.254141807556152, 2.3837924003601074] |
46d01c7b-d614-42a8-824f-f7f7180f3738 | multi-view-subspace-clustering-networks-with | 2010.09323 | null | https://arxiv.org/abs/2010.09323v3 | https://arxiv.org/pdf/2010.09323v3.pdf | Multi-view Subspace Clustering Networks with Local and Global Graph Information | This study investigates the problem of multi-view subspace clustering, the goal of which is to explore the underlying grouping structure of data collected from different fields or measurements. Since data do not always comply with the linear subspace models in many real-world applications, most existing multi-view subspace clustering methods that based on the shallow linear subspace models may fail in practice. Furthermore, underlying graph information of multi-view data is always ignored in most existing multi-view subspace clustering methods. To address aforementioned limitations, we proposed the novel multi-view subspace clustering networks with local and global graph information, termed MSCNLG, in this paper. Specifically, autoencoder networks are employed on multiple views to achieve latent smooth representations that are suitable for the linear assumption. Simultaneously, by integrating fused multi-view graph information into self-expressive layers, the proposed MSCNLG obtains the common shared multi-view subspace representation, which can be used to get clustering results by employing the standard spectral clustering algorithm. As an end-to-end trainable framework, the proposed method fully investigates the valuable information of multiple views. Comprehensive experiments on six benchmark datasets validate the effectiveness and superiority of the proposed MSCNLG. | ['Zhiqiang Tian', 'Zhongyu Li', 'Yuanyuan Ma', 'Jihua Zhu', 'Qinghai Zheng'] | 2020-10-19 | null | null | null | null | ['multi-view-subspace-clustering'] | ['computer-vision'] | [-2.87349761e-01 -3.94026995e-01 -9.92064923e-02 -2.08826452e-01
-2.93303370e-01 -4.94873226e-01 3.79538953e-01 -5.07764697e-01
1.31947502e-01 1.02864012e-01 5.03719926e-01 1.07539259e-01
-5.16464353e-01 -4.35904711e-01 -3.27931434e-01 -1.15606725e+00
3.94574970e-01 2.39514008e-01 -1.91802531e-01 1.76759064e-01
-7.68724605e-02 2.65311271e-01 -1.24790335e+00 2.52015769e-01
8.43758047e-01 5.30445457e-01 2.09879979e-01 2.33445317e-01
1.53793450e-02 5.01269400e-01 -2.34604860e-03 -1.42864222e-02
9.81962979e-02 -2.83667535e-01 -2.71253884e-01 6.45834565e-01
2.92681634e-01 -9.90461260e-02 -4.84653413e-01 1.28102219e+00
5.93670845e-01 2.37646475e-01 4.78919119e-01 -1.29823256e+00
-6.79660738e-01 6.42313719e-01 -7.82046139e-01 -6.16181716e-02
1.44436210e-01 -1.68918297e-01 9.06522334e-01 -1.05682206e+00
5.32356858e-01 1.38863742e+00 2.70110875e-01 3.34733099e-01
-1.13818336e+00 -5.47824085e-01 5.14262497e-01 3.11450273e-01
-1.63715208e+00 -3.77361715e-01 1.39312541e+00 -4.48086411e-01
3.42731506e-01 1.37359947e-01 6.44829929e-01 1.05923665e+00
-4.45522070e-02 9.18508589e-01 1.09140122e+00 8.83313641e-02
4.37500179e-02 1.02796480e-01 1.29881531e-01 4.66702223e-01
2.69034147e-01 -2.00334847e-01 -2.16291666e-01 -2.16740087e-01
5.80506504e-01 5.97339272e-01 -5.38829625e-01 -1.16156435e+00
-1.43790293e+00 8.22074413e-01 4.05993104e-01 4.58403200e-01
-4.13008481e-01 -3.70147616e-01 4.62779373e-01 -3.56449652e-03
3.99319053e-01 -2.81087071e-01 -4.40119393e-03 4.18896973e-01
-8.18106115e-01 -2.00188488e-01 5.28167367e-01 1.10144770e+00
7.23747611e-01 3.67611289e-01 3.27163815e-01 7.63031662e-01
7.02784240e-01 5.12299001e-01 5.28046668e-01 -8.83609831e-01
5.23388803e-01 1.07736933e+00 -2.55566180e-01 -1.38504267e+00
-4.44965273e-01 -8.19101393e-01 -1.46154547e+00 -2.44046301e-01
-1.79952998e-02 9.92819294e-03 -5.47150910e-01 1.58412635e+00
5.74517608e-01 4.35472578e-01 4.63719189e-01 1.03708804e+00
1.02413213e+00 6.41885161e-01 -3.45510125e-01 -5.60921788e-01
9.77770150e-01 -9.14305151e-01 -7.34106719e-01 3.08016371e-02
4.53713179e-01 -6.13396168e-01 7.48689651e-01 4.49153870e-01
-6.31930292e-01 -6.35383308e-01 -9.96699631e-01 3.88216853e-01
-1.67236656e-01 1.65881813e-01 6.42556369e-01 3.96055073e-01
-9.31986272e-01 2.19139457e-01 -8.97029817e-01 -3.84857893e-01
2.26516694e-01 2.57910460e-01 -6.77087247e-01 -5.14596105e-01
-7.69330978e-01 8.75426009e-02 8.39554369e-01 4.69471306e-01
-6.92489326e-01 -2.63122171e-01 -8.26236486e-01 6.49425909e-02
6.75534904e-01 -7.11101592e-01 4.10857618e-01 -7.09259212e-01
-1.05760705e+00 4.13534671e-01 -2.59951293e-01 2.48522073e-01
1.71796560e-01 -2.01520726e-01 -7.96304822e-01 2.39087701e-01
1.22197606e-01 4.36597578e-02 7.46915936e-01 -1.74556017e+00
-1.95969969e-01 -8.74905765e-01 -3.24230373e-01 6.58381999e-01
-4.99202996e-01 -1.74137414e-01 -8.30497324e-01 -4.19479936e-01
8.11451793e-01 -9.87722039e-01 -2.70483702e-01 -7.15510011e-01
-6.16576433e-01 -1.11197412e-01 1.36754942e+00 -6.98321521e-01
1.28084278e+00 -2.19277048e+00 7.96776652e-01 2.97106564e-01
4.22822267e-01 -6.76938444e-02 4.81350683e-02 5.76199472e-01
-3.19408983e-01 -2.23645300e-01 -1.19207434e-01 -1.58256322e-01
-1.02874428e-01 1.16042972e-01 -1.09625906e-01 7.70624399e-01
-4.51037496e-01 6.74860239e-01 -7.34131873e-01 -6.57144070e-01
3.75512034e-01 5.07220030e-01 -4.49866444e-01 3.76136065e-01
2.40832344e-01 8.12864423e-01 -6.08598530e-01 6.40998304e-01
8.32883894e-01 -6.63853645e-01 3.07117790e-01 -6.97950602e-01
1.54585361e-01 -6.13875985e-01 -1.63675511e+00 2.15736914e+00
-4.74262759e-02 3.11320517e-02 3.77488017e-01 -1.32484388e+00
6.33146763e-01 4.13078904e-01 8.87115180e-01 -2.01280694e-02
1.04935236e-01 6.59796670e-02 3.18461843e-02 -3.95195723e-01
1.69043541e-01 -1.07156426e-01 8.48477706e-02 3.38126719e-01
2.41806209e-01 4.47118461e-01 -5.92886806e-02 4.71094340e-01
3.72459114e-01 1.65432408e-01 3.77138644e-01 -2.61722058e-01
9.87032950e-01 -3.53455245e-01 7.40614951e-01 1.11756980e-01
-8.75575244e-02 6.05996370e-01 9.66673344e-02 -3.14815789e-01
-8.05843651e-01 -9.26381528e-01 1.00369938e-01 6.87935412e-01
4.23352152e-01 -4.24717486e-01 -7.27937043e-01 -6.39779449e-01
-3.17911774e-01 4.10589367e-01 -3.11306417e-01 -2.17944443e-01
-3.74226987e-01 -8.51507962e-01 2.64968425e-02 4.77448106e-01
5.39593518e-01 -6.05525613e-01 -5.26447631e-02 1.63138315e-01
-3.70395094e-01 -1.27091718e+00 -4.93781894e-01 -1.94598824e-01
-1.05239606e+00 -1.14739966e+00 -7.12008417e-01 -7.04536200e-01
7.40113795e-01 9.06909645e-01 6.65702164e-01 1.17333326e-02
1.24277391e-01 7.95531690e-01 -3.52094322e-01 1.38735875e-01
-8.41595158e-02 -8.84176046e-02 4.56953675e-01 6.94364429e-01
5.94711244e-01 -9.98486519e-01 -5.68284810e-01 3.79971534e-01
-9.98825788e-01 1.74988598e-01 5.54735005e-01 1.00571620e+00
7.95635939e-01 3.77305329e-01 4.42473054e-01 -7.24691153e-01
2.46409222e-01 -6.05284095e-01 -3.68656784e-01 2.86098540e-01
-6.22006357e-01 -2.17441037e-01 8.65345001e-01 -2.69338697e-01
-9.90631700e-01 1.75606251e-01 2.56170452e-01 -1.29402828e+00
-4.22011882e-01 7.91126847e-01 -9.48615968e-01 6.01625964e-02
4.44587246e-02 8.27753425e-01 7.60507807e-02 -4.81801450e-01
6.59566522e-01 6.00125670e-01 3.51851523e-01 -3.33999336e-01
9.69881237e-01 7.24946201e-01 -8.89820978e-02 -9.07885730e-01
-7.18972325e-01 -8.49315584e-01 -1.05957460e+00 -2.88063079e-01
1.11461747e+00 -1.44207811e+00 -6.05545759e-01 4.63657141e-01
-8.08470905e-01 4.04981345e-01 2.91909546e-01 8.31189930e-01
-4.29904193e-01 9.82942641e-01 -3.67158443e-01 -7.26014018e-01
-1.17041849e-01 -1.10522044e+00 1.05920279e+00 2.09897310e-01
3.86879534e-01 -1.28232765e+00 -5.65980673e-02 6.00760579e-01
-9.73613337e-02 3.42389911e-01 8.22498024e-01 -7.82331169e-01
-4.79437232e-01 -3.04898061e-02 -5.63352108e-02 2.87425071e-01
2.86987901e-01 -2.42886722e-01 -8.60187709e-01 -7.74632394e-01
3.89646739e-01 -5.20562902e-02 6.81327999e-01 5.41771054e-01
1.09779310e+00 -2.25436255e-01 -4.93537933e-01 9.05377030e-01
1.70433712e+00 1.98332682e-01 1.59684375e-01 7.30872899e-02
1.57367766e+00 6.89817667e-01 3.26468617e-01 4.05659854e-01
6.27356350e-01 4.52485919e-01 5.06361067e-01 -3.51692550e-02
4.37485963e-01 -3.06511849e-01 3.28507870e-01 1.86424935e+00
-1.68027177e-01 -2.04851061e-01 -7.28730023e-01 5.25268674e-01
-2.12224364e+00 -1.19636703e+00 -2.32878134e-01 1.99621081e+00
-2.02476978e-01 -2.49273062e-01 1.47545978e-01 1.58483144e-02
1.02924013e+00 4.54583466e-01 -7.51448989e-01 2.64776021e-01
-3.23624998e-01 -8.92275274e-01 1.63232386e-01 -2.90218294e-02
-1.30946863e+00 7.25081980e-01 4.79061985e+00 7.03698039e-01
-9.66648877e-01 2.32293829e-02 2.35292837e-01 -5.28147444e-03
-4.98715580e-01 1.43320948e-01 -4.85623211e-01 3.16272438e-01
6.16989553e-01 -2.08331153e-01 6.83555722e-01 9.09296155e-01
3.29235435e-01 4.97843862e-01 -9.66575325e-01 1.37093902e+00
3.73431772e-01 -9.74333286e-01 4.11411494e-01 2.70379603e-01
8.35789204e-01 -3.37236635e-02 -7.73669872e-03 1.42410472e-01
3.03817749e-01 -7.04361320e-01 3.00240070e-01 5.72972417e-01
5.22700429e-01 -9.61851895e-01 5.87446511e-01 7.17888355e-01
-1.51379526e+00 -2.69698381e-01 -5.31691551e-01 2.98491240e-01
2.11564511e-01 5.32302082e-01 -3.19607198e-01 1.53161967e+00
7.15502799e-01 1.53072643e+00 -6.19736314e-01 6.60079360e-01
1.72272757e-01 5.91675639e-01 -1.86292261e-01 4.47763234e-01
3.31963509e-01 -9.70152795e-01 8.08460593e-01 7.02854931e-01
4.55600321e-01 1.50395885e-01 6.77292049e-01 8.59270692e-01
2.64503866e-01 2.80520737e-01 -9.84824836e-01 -2.06940562e-01
3.57437402e-01 1.48870003e+00 -7.85191834e-01 -2.88755208e-01
-8.75686944e-01 9.39387262e-01 2.32468382e-01 7.78255463e-01
-5.89702964e-01 1.20358743e-01 2.29123831e-01 -3.19551349e-01
3.14918935e-01 -3.63416076e-01 -3.12605612e-02 -1.90868974e+00
1.47945449e-01 -1.12009680e+00 6.77722096e-01 -6.22937858e-01
-1.44785166e+00 4.92881924e-01 6.93861321e-02 -1.82060897e+00
-1.07915327e-01 -3.67090225e-01 -7.27407396e-01 7.08910167e-01
-9.92804766e-01 -1.73448634e+00 -4.74246025e-01 1.08905470e+00
3.97562683e-01 -5.58051825e-01 6.68920696e-01 3.53180379e-01
-9.01050746e-01 2.66583413e-01 5.77010036e-01 1.59736797e-01
5.49402952e-01 -1.19949257e+00 -2.17340320e-01 1.14813578e+00
3.69882166e-01 9.68763053e-01 3.15881997e-01 -5.97538173e-01
-2.03522038e+00 -1.28996086e+00 -6.16596229e-02 -1.28057063e-01
6.25047982e-01 -2.82937139e-01 -1.10626686e+00 8.26670110e-01
2.56684273e-01 3.57025838e-03 1.04439664e+00 2.52995312e-01
-3.82104278e-01 -3.29146013e-02 -5.48215210e-01 5.08979142e-01
1.01793253e+00 -7.46505380e-01 -5.59094310e-01 2.44342059e-01
7.42404640e-01 -1.38723433e-01 -1.16206467e+00 4.28888738e-01
4.15153444e-01 -1.13140225e+00 1.01496160e+00 -6.80819929e-01
2.71396279e-01 -6.99949622e-01 -5.43299019e-01 -1.52158678e+00
-6.52161896e-01 -1.34424895e-01 -3.08002949e-01 1.40228295e+00
-1.89551026e-01 -5.62854350e-01 7.07464874e-01 1.23045728e-01
-2.03371406e-01 -5.18119693e-01 -8.46725523e-01 -7.19616652e-01
-7.36392066e-02 -1.61216229e-01 6.02406144e-01 1.42844641e+00
-1.98655054e-01 6.23670161e-01 -6.36665404e-01 6.46410763e-01
1.17669630e+00 7.20070541e-01 8.90669703e-01 -1.33892000e+00
-4.12644267e-01 -2.76780635e-01 -4.71357197e-01 -8.21866453e-01
4.18907762e-01 -1.15073156e+00 -3.48842770e-01 -1.68056989e+00
7.27971852e-01 1.55374244e-01 -5.53095520e-01 -9.37717259e-02
-4.95115250e-01 -2.53189951e-01 2.14573100e-01 5.44467449e-01
-8.01920354e-01 9.40177798e-01 1.20925868e+00 -1.61749259e-01
-2.53403664e-01 -1.09007291e-01 -8.34363341e-01 8.75101149e-01
3.41813177e-01 2.05604006e-02 -8.61159980e-01 -3.59388411e-01
-1.43667400e-01 3.52171630e-01 3.56764823e-01 -1.06660044e+00
4.39424068e-01 -1.68581575e-01 4.20651078e-01 -9.32533920e-01
2.10609809e-01 -1.21674049e+00 4.82812017e-01 2.74598431e-02
1.61371738e-01 -7.62752071e-02 -1.73661977e-01 1.19047761e+00
-6.31844997e-01 3.91350091e-01 5.85689247e-01 -1.95946574e-01
-8.04348052e-01 6.30642414e-01 1.57649830e-01 -1.39273867e-01
1.03112197e+00 -3.83289248e-01 3.69384810e-02 -4.80148584e-01
-9.50229108e-01 5.08162379e-01 6.50073886e-01 5.01097023e-01
6.73368037e-01 -1.89279878e+00 -6.07858062e-01 2.13481903e-01
4.06960249e-01 2.52330810e-01 8.64190698e-01 1.07861149e+00
8.35859403e-03 3.73559415e-01 -1.16796874e-01 -1.01088381e+00
-1.22093368e+00 1.12933218e+00 8.23255926e-02 -1.40965566e-01
-7.87051916e-01 3.35925102e-01 6.66809440e-01 -7.11380601e-01
-1.00387715e-01 3.00884604e-01 -6.86020076e-01 1.39809847e-01
2.79110521e-01 3.47545207e-01 -4.06785488e-01 -1.17215371e+00
-2.28323221e-01 8.41682494e-01 8.32861587e-02 1.26715705e-01
1.44798803e+00 -5.78717887e-01 -3.04106146e-01 8.35259616e-01
1.45816231e+00 -1.10872298e-01 -1.01025653e+00 -4.53203022e-01
-1.82146668e-01 -2.87665814e-01 1.05818391e-01 -5.40772378e-02
-1.32818389e+00 1.09193432e+00 5.33544302e-01 5.48508577e-03
1.31401575e+00 -1.70628801e-01 5.09525120e-01 4.76814866e-01
4.29319829e-01 -1.00211608e+00 -1.65358573e-01 1.21693403e-01
6.98930025e-01 -1.32497025e+00 1.63909942e-01 -5.02611995e-01
-7.90965319e-01 1.01961172e+00 7.59145439e-01 -2.65220664e-02
7.26948321e-01 -4.91513282e-01 4.30165529e-02 -4.93681669e-01
-5.06604791e-01 1.08984478e-01 5.35315633e-01 3.92954677e-01
1.19424097e-01 5.33588640e-02 -3.09253391e-02 7.18497694e-01
1.49217620e-01 -4.24262524e-01 4.25791204e-01 5.07953465e-01
-1.36684403e-01 -8.01647544e-01 -5.12540221e-01 3.54638785e-01
-2.26922289e-01 1.82944089e-01 -3.58830482e-01 6.49457216e-01
-9.85338688e-02 1.01115012e+00 -3.29326153e-01 -7.52482891e-01
9.81600657e-02 1.06887214e-01 -4.70185876e-02 -5.27780473e-01
-1.13765104e-02 8.50575864e-01 -3.74382168e-01 -6.82488024e-01
-9.76790190e-01 -7.89543390e-01 -1.07347357e+00 2.71502770e-02
-3.41776282e-01 3.06446195e-01 2.79690593e-01 9.28939998e-01
4.33272362e-01 5.41760862e-01 9.51437831e-01 -8.23270798e-01
-4.95755881e-01 -8.16455543e-01 -1.06368315e+00 7.17998326e-01
3.59716378e-02 -6.64008856e-01 -4.20303494e-01 1.31898865e-01] | [8.278345108032227, 4.598697662353516] |
658a5af0-79b6-499a-afe5-08e151a917b6 | lrw-1000-a-naturally-distributed-large-scale | 1810.06990 | null | http://arxiv.org/abs/1810.06990v6 | http://arxiv.org/pdf/1810.06990v6.pdf | LRW-1000: A Naturally-Distributed Large-Scale Benchmark for Lip Reading in the Wild | Large-scale datasets have successively proven their fundamental importance in
several research fields, especially for early progress in some emerging topics.
In this paper, we focus on the problem of visual speech recognition, also known
as lipreading, which has received increasing interest in recent years. We
present a naturally-distributed large-scale benchmark for lip reading in the
wild, named LRW-1000, which contains 1,000 classes with 718,018 samples from
more than 2,000 individual speakers. Each class corresponds to the syllables of
a Mandarin word composed of one or several Chinese characters. To the best of
our knowledge, it is currently the largest word-level lipreading dataset and
also the only public large-scale Mandarin lipreading dataset. This dataset aims
at covering a "natural" variability over different speech modes and imaging
conditions to incorporate challenges encountered in practical applications. It
has shown a large variation in this benchmark in several aspects, including the
number of samples in each class, video resolution, lighting conditions, and
speakers' attributes such as pose, age, gender, and make-up. Besides providing
a detailed description of the dataset and its collection pipeline, we evaluate
several typical popular lipreading methods and perform a thorough analysis of
the results from several aspects. The results demonstrate the consistency and
challenges of our dataset, which may open up some new promising directions for
future work. | ['Yuan-Hang Zhang', 'Jing-Yun Xiao', 'Keyu Long', 'Xilin Chen', 'Mingmin Yang', 'Dalu Feng', 'Shuang Yang', 'Shiguang Shan', 'Chenhao Wang'] | 2018-10-16 | null | null | null | null | ['lipreading'] | ['computer-vision'] | [ 1.97462782e-01 -2.89361119e-01 -5.60873091e-01 -2.31132314e-01
-1.10948241e+00 -3.20196390e-01 7.20121503e-01 -2.23532543e-01
-3.60321760e-01 6.30240977e-01 4.04093534e-01 -2.38050386e-01
4.28785950e-01 1.70001201e-02 -5.21267831e-01 -7.91556895e-01
2.50483572e-01 2.16282889e-01 1.40153080e-01 5.91905303e-02
4.10505742e-01 2.97950327e-01 -2.08788586e+00 9.41421017e-02
7.61096954e-01 1.08775175e+00 3.34856659e-01 6.35858536e-01
1.25710145e-01 2.18055949e-01 -6.91221654e-01 -5.89622736e-01
-1.53129473e-01 -3.29428971e-01 -5.09740829e-01 3.18264365e-01
9.47825074e-01 -2.06778705e-01 -2.66577005e-01 1.03356636e+00
1.27305388e+00 3.33179720e-02 5.90145290e-01 -1.39284050e+00
-6.29145861e-01 3.63165349e-01 -6.47226632e-01 1.29950419e-01
5.12713373e-01 4.04663682e-01 8.03172171e-01 -1.14016998e+00
4.28219020e-01 1.34310913e+00 5.49234688e-01 8.91825020e-01
-7.86352396e-01 -5.68777084e-01 -5.38933836e-02 6.04553759e-01
-1.47877383e+00 -1.29375637e+00 8.13609958e-01 -1.97497398e-01
5.23278534e-01 2.05501959e-01 3.52558315e-01 1.46868658e+00
-3.52475554e-01 1.11601603e+00 1.31556547e+00 -5.72465897e-01
1.88889429e-01 2.10406199e-01 1.59860086e-02 2.93111682e-01
-5.10310463e-04 -5.46886586e-02 -8.91379416e-01 9.33494121e-02
3.07699859e-01 -6.26774192e-01 -6.32972717e-01 -5.26153684e-01
-1.43124783e+00 6.15478933e-01 -2.82244503e-01 -5.10415174e-02
1.32152468e-01 -3.12940478e-01 5.07935345e-01 7.21402187e-03
3.06836098e-01 -2.98671246e-01 -4.38296556e-01 -3.97007704e-01
-8.31426084e-01 7.93299079e-02 7.09464729e-01 1.04908478e+00
3.13330919e-01 -1.38145909e-02 -2.28962585e-01 1.54407597e+00
2.81076759e-01 9.08649147e-01 7.55814970e-01 -7.34915435e-01
6.38614357e-01 -6.58483133e-02 -1.47014350e-01 -5.82992136e-01
-2.01390088e-01 1.47687495e-01 -7.03353047e-01 1.70172900e-01
7.06314027e-01 3.28905471e-02 -7.28095233e-01 1.84078181e+00
1.96558699e-01 7.20224231e-02 -6.93150759e-02 8.27173650e-01
1.25757825e+00 2.82008082e-01 -6.83254674e-02 -3.72229218e-01
1.66842687e+00 -1.06518900e+00 -6.91612184e-01 -2.30108738e-01
-3.46553773e-01 -1.18695438e+00 1.40579724e+00 4.76076424e-01
-1.29453647e+00 -6.34664178e-01 -7.94636667e-01 -1.75629571e-01
-3.88771385e-01 4.92280662e-01 3.43429357e-01 1.09785306e+00
-1.33274055e+00 -8.03413540e-02 -2.01968387e-01 -7.68405735e-01
4.88154888e-01 1.25402823e-01 -3.33759487e-01 -2.12978274e-01
-8.30606937e-01 6.39234602e-01 -2.44095072e-01 -1.94025159e-01
-7.52176285e-01 -5.28698087e-01 -9.42839503e-01 -2.10036173e-01
4.46584642e-01 -2.69999713e-01 1.34261060e+00 -4.24983829e-01
-1.80074823e+00 1.26496708e+00 -7.32129991e-01 -1.45877048e-01
6.89962983e-01 1.76076338e-01 -7.16417551e-01 1.26243889e-01
-9.16166753e-02 9.23804879e-01 1.24200523e+00 -1.12202334e+00
-3.94210756e-01 -3.84232014e-01 -4.07052875e-01 2.32013524e-01
-2.99929500e-01 5.42045414e-01 -1.03556919e+00 -7.75694788e-01
-2.77576417e-01 -7.63368726e-01 5.69452286e-01 7.34731928e-02
-5.11174977e-01 -4.54905152e-01 5.89881420e-01 -7.06667840e-01
9.57926750e-01 -2.36374569e+00 -8.83817598e-02 -4.36773479e-01
1.82964265e-01 4.07824904e-01 -2.12715238e-01 1.68096334e-01
-1.41864968e-02 1.24945186e-01 -9.07986388e-02 -8.34531486e-01
2.04255864e-01 -2.16992170e-01 -1.32333383e-01 6.11780226e-01
-2.00302854e-01 9.25143480e-01 -5.03156185e-01 -7.45347321e-01
2.29064092e-01 5.94792545e-01 -1.44137993e-01 1.18088305e-01
6.91568926e-02 1.39708817e-01 9.83645469e-02 1.07453251e+00
1.09879708e+00 6.85724616e-02 -2.99761444e-01 -9.32898670e-02
-5.47158606e-02 2.03929976e-01 -9.26390409e-01 1.49748409e+00
-3.45154941e-01 1.25936961e+00 4.45980400e-01 -7.17107117e-01
7.63789117e-01 4.22336042e-01 2.11049065e-01 -6.61094010e-01
7.27054998e-02 2.98766375e-01 -1.05612330e-01 -6.64674938e-01
5.08678496e-01 1.13110438e-01 2.10530296e-01 1.95902854e-01
-1.25464544e-01 -2.11646855e-01 2.60988951e-01 -1.25080258e-01
4.40957248e-01 -5.30203164e-01 3.00698608e-01 -1.76150769e-01
8.19068074e-01 -7.49866068e-01 2.20847175e-01 4.63519156e-01
-8.98825169e-01 9.78518367e-01 4.83403832e-01 1.46511346e-01
-8.98501158e-01 -1.30910182e+00 -5.86527228e-01 1.04028594e+00
3.72920521e-02 -2.49178618e-01 -9.63579834e-01 -2.86783069e-01
-6.14096150e-02 1.96042299e-01 -5.54996431e-01 2.15226069e-01
-2.53655344e-01 -5.50236642e-01 7.89969683e-01 3.65139365e-01
7.28542566e-01 -1.35771072e+00 -2.44989559e-01 -4.55102533e-01
-5.64038396e-01 -1.51013112e+00 -9.91605818e-01 -4.89945948e-01
-4.03119415e-01 -1.07241380e+00 -1.40098631e+00 -1.17418778e+00
4.21917319e-01 3.01772982e-01 1.12813556e+00 -1.73469990e-01
-3.25412035e-01 4.67731029e-01 -2.50035971e-01 -5.80843270e-01
-5.53811729e-01 1.42191455e-01 3.22583914e-01 1.96891397e-01
4.28634763e-01 -3.59498039e-02 -4.76294130e-01 7.47690380e-01
-4.28847283e-01 -1.97259724e-01 4.93730336e-01 6.94867849e-01
4.33108658e-01 -2.22519860e-01 6.46344244e-01 3.06378826e-02
8.27802539e-01 -1.07972376e-01 -4.64181721e-01 3.64802003e-01
-3.05581778e-01 -4.08764213e-01 3.02878708e-01 -6.90438390e-01
-8.83451223e-01 -4.11204398e-01 -3.12824190e-01 -2.14813948e-01
-5.50414979e-01 -1.74372911e-01 -6.30501509e-01 -9.00210291e-02
2.33729675e-01 5.87396502e-01 3.43700260e-01 -8.40547860e-01
1.03404500e-01 1.38194001e+00 7.70010769e-01 -5.06766498e-01
3.73461485e-01 1.53806448e-01 -3.31194013e-01 -1.56897676e+00
-1.45798922e-01 -5.68205833e-01 -2.89627731e-01 -1.83115721e-01
6.21735990e-01 -8.37399721e-01 -1.20603931e+00 1.42826712e+00
-9.65928495e-01 -4.04717177e-01 -1.45767793e-01 4.00898963e-01
-7.36043751e-01 6.20471835e-01 -4.58467692e-01 -9.10167933e-01
-1.27204865e-01 -1.60711777e+00 1.26690161e+00 3.16727966e-01
1.46544978e-01 -7.31740773e-01 -1.09086968e-01 7.99923778e-01
4.85270023e-01 -4.81984615e-01 6.66417956e-01 -3.86432171e-01
-3.00970703e-01 1.84524894e-01 -3.65010232e-01 5.55778861e-01
2.37422466e-01 -2.14835312e-02 -1.35776389e+00 -4.16578114e-01
-1.10008128e-01 -5.47356725e-01 9.54682171e-01 6.72867835e-01
1.26887488e+00 6.53634837e-04 -1.79212525e-01 4.68189001e-01
8.96205664e-01 7.15967938e-02 7.71967351e-01 -1.13363631e-01
3.26717079e-01 6.70137942e-01 3.63836259e-01 4.28782403e-01
4.90822792e-01 1.01147258e+00 2.56022096e-01 -1.43652081e-01
-7.02823579e-01 -3.03252846e-01 5.28275132e-01 5.99085093e-01
-5.04310392e-02 -3.77427757e-01 -8.74893546e-01 5.69828391e-01
-1.14014816e+00 -1.02808177e+00 2.74574220e-01 2.52138090e+00
9.45175350e-01 -2.49293059e-01 6.36581779e-01 4.32466596e-01
1.20129418e+00 4.22288239e-01 -7.14438736e-01 -2.17346445e-01
-6.00714624e-01 6.11025281e-02 1.74277574e-01 5.13927162e-01
-1.03363478e+00 9.68638301e-01 7.09489822e+00 1.11236048e+00
-1.28399754e+00 -1.63760036e-02 9.87423658e-01 -7.96219036e-02
2.01155841e-01 -7.37047911e-01 -1.11006510e+00 6.93388999e-01
7.25349963e-01 -2.37288009e-02 6.70535505e-01 6.78451836e-01
2.56186962e-01 -2.71015525e-01 -9.51809347e-01 1.40764427e+00
8.55997145e-01 -1.06771839e+00 -1.27096206e-01 2.91141778e-01
4.40992236e-01 3.21035087e-01 5.64853907e-01 7.27293640e-02
-4.22188848e-01 -1.08730412e+00 7.31699586e-01 1.36778623e-01
1.55481374e+00 -4.29888010e-01 3.56898159e-01 2.15805843e-01
-1.24336576e+00 5.08201979e-02 -3.33594084e-01 4.10667598e-01
1.21853605e-01 6.16587214e-02 -5.70316136e-01 -4.86870222e-02
8.04962158e-01 6.07640266e-01 -6.67107344e-01 1.25967336e+00
2.78116390e-03 4.61184919e-01 -1.74410269e-01 -2.63020754e-01
-3.54653090e-01 2.79213071e-01 5.47721088e-01 1.07161427e+00
2.18138993e-01 -3.66855413e-01 -2.51831174e-01 3.85303646e-01
-3.29078108e-01 4.31228399e-01 -3.37416947e-01 1.27797097e-01
5.16649127e-01 1.01596129e+00 -3.62230390e-01 -1.68482870e-01
-5.95418334e-01 6.51975930e-01 -8.91698822e-02 4.55142438e-01
-7.80604482e-01 -2.65613705e-01 1.12293875e+00 2.26469234e-01
4.07154381e-01 -5.15855849e-02 -1.39088184e-01 -1.23579991e+00
3.03589970e-01 -1.34607852e+00 2.88511328e-02 -8.71892512e-01
-1.24306023e+00 5.37416458e-01 -9.13601965e-02 -1.04692388e+00
-1.18363693e-01 -8.88945222e-01 -5.31727612e-01 8.87230515e-01
-1.83688986e+00 -9.29128170e-01 -3.80938709e-01 9.29844081e-01
1.14282513e+00 -7.16374576e-01 6.26656532e-01 3.02811116e-01
-7.59622514e-01 1.18881905e+00 1.33530274e-01 1.73559725e-01
1.21200764e+00 -8.42539430e-01 4.89476711e-01 5.54974616e-01
1.17062621e-01 4.38784927e-01 5.48685491e-01 -2.17704147e-01
-1.14843154e+00 -5.05035222e-01 1.18194652e+00 -3.27061057e-01
4.80399340e-01 -6.94705546e-01 -5.66970885e-01 2.01699689e-01
4.52790648e-01 1.17141187e-01 5.45282423e-01 -6.89988658e-02
-3.44792515e-01 -2.74892867e-01 -1.27644539e+00 6.48182392e-01
1.07366168e+00 -5.65581918e-01 -2.56409466e-01 1.78915337e-01
3.64575773e-01 -4.00209606e-01 -4.73435938e-01 2.73285180e-01
6.53690398e-01 -1.27992249e+00 1.11224937e+00 -2.39308745e-01
1.29288271e-01 1.41464859e-01 -2.24727184e-01 -1.14512193e+00
2.91836858e-01 -9.35312986e-01 -7.07923993e-02 1.56104362e+00
2.45583430e-01 -7.47943163e-01 8.68416548e-01 2.99833745e-01
5.90835474e-02 -9.05610740e-01 -1.04688931e+00 -8.45167339e-01
3.27327877e-01 -5.63163459e-01 7.33669698e-01 4.26942855e-01
5.87328821e-02 1.06580272e-01 -5.54486632e-01 -2.05816627e-01
7.34206080e-01 -5.98773621e-02 8.35330367e-01 -1.11300647e+00
-1.46683618e-01 -7.34930515e-01 -3.81084770e-01 -1.55616200e+00
4.37188208e-01 -4.49241668e-01 1.80646494e-01 -1.14350903e+00
4.08256352e-01 -2.22579151e-01 7.65071362e-02 1.72578022e-01
-1.46188185e-01 4.51559693e-01 2.28905246e-01 1.51531696e-01
-3.24288994e-01 4.27760720e-01 1.35602665e+00 -4.41915184e-01
1.01734027e-01 4.82256234e-01 -6.17442310e-01 6.53481841e-01
8.73788953e-01 2.19048798e-01 -4.07889605e-01 -3.35563302e-01
-4.37196106e-01 3.67537811e-02 2.66267896e-01 -8.41025233e-01
3.22965682e-01 -1.32871300e-01 2.42172815e-02 -6.79486632e-01
7.34736919e-01 -3.60237122e-01 -5.29440343e-01 2.15715170e-02
-2.96583384e-01 2.59286282e-03 1.46943316e-01 2.72743762e-01
-3.08520496e-01 -2.71168109e-02 1.11675978e+00 2.64536202e-01
-7.28533804e-01 4.50554699e-01 -2.28428259e-01 5.99763036e-01
9.12741125e-01 -3.44099522e-01 -4.80856270e-01 -6.29729867e-01
-5.01498938e-01 -1.01948366e-01 5.82521617e-01 6.66264355e-01
6.79988623e-01 -1.05809891e+00 -8.55270088e-01 5.73511064e-01
3.66764098e-01 -4.84677523e-01 2.62535840e-01 8.88196170e-01
-5.69011346e-02 5.73957562e-01 -2.44730651e-01 -7.18523622e-01
-1.74123597e+00 5.53995132e-01 2.83326507e-01 2.83516496e-01
-3.15697372e-01 9.06628430e-01 -6.12759846e-04 -5.18448539e-02
7.57596850e-01 -1.64701864e-01 -2.82745838e-01 2.29026794e-01
7.16541767e-01 6.65978014e-01 8.69776011e-02 -1.11994088e+00
-5.68755746e-01 8.34667087e-01 1.30499914e-01 -1.75719529e-01
7.72353411e-01 -4.59117353e-01 1.45583764e-01 4.34047043e-01
1.26706874e+00 5.22304475e-01 -1.25915194e+00 -1.56116232e-01
-4.67820048e-01 -5.93632460e-01 -2.48776987e-01 -8.26848567e-01
-1.26621175e+00 1.29624164e+00 6.06136441e-01 -2.01483388e-02
1.08150351e+00 4.00675386e-01 7.67901480e-01 2.91814767e-02
3.63697797e-01 -1.10146689e+00 2.04476565e-01 3.32212716e-01
1.04683554e+00 -1.67636204e+00 -3.87569726e-01 -4.76960778e-01
-8.01473022e-01 7.51817882e-01 3.89499605e-01 7.68452704e-01
4.91782904e-01 2.40423873e-01 3.93522173e-01 5.50079525e-01
-4.29896683e-01 -2.89679557e-01 3.17567021e-01 1.13003218e+00
3.98115307e-01 1.54472768e-01 -1.04227930e-01 3.75824839e-01
-3.87953132e-01 -1.51256159e-01 2.47007951e-01 1.23798259e-01
-2.47207865e-01 -1.20948148e+00 -4.98310089e-01 1.61901131e-01
-3.58221740e-01 -2.53687739e-01 -3.67975086e-01 6.42896533e-01
4.06777672e-02 1.27742684e+00 3.56108546e-02 -1.01756305e-01
1.48107201e-01 1.47691309e-01 5.09420455e-01 -2.86485791e-01
7.80378133e-02 4.90092672e-02 -3.46833169e-02 -3.18411201e-01
-1.98455006e-01 -1.18881941e+00 -6.32871866e-01 -5.59307098e-01
-1.15492478e-01 -2.09842771e-01 9.38395917e-01 4.81840163e-01
2.32311293e-01 -2.23155990e-01 5.27049065e-01 -9.94559288e-01
-7.83416331e-01 -1.14188802e+00 -6.87522173e-01 3.52481365e-01
6.56408310e-01 -7.89473951e-01 -6.66096151e-01 7.28131905e-02] | [14.315237045288086, 4.988456726074219] |
6afd27eb-7d9d-40d4-adfd-6370d0794788 | bandit-algorithms-for-tree-search | 1408.2028 | null | http://arxiv.org/abs/1408.2028v1 | http://arxiv.org/pdf/1408.2028v1.pdf | Bandit Algorithms for Tree Search | Bandit based methods for tree search have recently gained popularity when
applied to huge trees, e.g. in the game of go [6]. Their efficient exploration
of the tree enables to re- turn rapidly a good value, and improve preci- sion
if more time is provided. The UCT algo- rithm [8], a tree search method based
on Up- per Confidence Bounds (UCB) [2], is believed to adapt locally to the
effective smoothness of the tree. However, we show that UCT is
"over-optimistic" in some sense, leading to a worst-case regret that may be
very poor. We propose alternative bandit algorithms for tree search. First, a
modification of UCT us- ing a confidence sequence that scales expo- nentially
in the horizon depth is analyzed. We then consider Flat-UCB performed on the
leaves and provide a finite regret bound with high probability. Then, we
introduce and analyze a Bandit Algorithm for Smooth Trees (BAST) which takes
into account ac- tual smoothness of the rewards for perform- ing efficient
"cuts" of sub-optimal branches with high confidence. Finally, we present an
incremental tree expansion which applies when the full tree is too big
(possibly in- finite) to be entirely represented and show that with high
probability, only the optimal branches are indefinitely developed. We illus-
trate these methods on a global optimization problem of a continuous function,
given noisy values. | ['Pierre-Arnuad Coquelin', 'Remi Munos'] | 2014-08-09 | null | null | null | null | ['game-of-go'] | ['playing-games'] | [ 1.26311198e-01 6.18210971e-01 -6.59349501e-01 -9.47934017e-02
-1.34972501e+00 -7.05547869e-01 5.54180220e-02 2.11048990e-01
-4.12142903e-01 1.24269950e+00 -6.59773424e-02 -6.51865065e-01
-6.25867784e-01 -9.58214521e-01 -8.94598067e-01 -7.90934384e-01
-2.90611118e-01 7.80845642e-01 1.64739430e-01 -3.50251421e-02
2.21703202e-01 3.98864657e-01 -1.06049669e+00 1.22829184e-01
9.96234655e-01 1.25388873e+00 -3.51403542e-02 9.58303154e-01
5.43748355e-03 5.89069128e-01 -2.97865182e-01 -7.07610250e-01
4.53249484e-01 -3.97402704e-01 -1.12135029e+00 1.30080208e-01
-1.08450569e-01 -3.44342291e-01 9.47026461e-02 1.20191157e+00
-1.22940894e-02 8.79856274e-02 5.42070508e-01 -8.09797943e-01
1.59841910e-01 1.30936015e+00 -7.66829014e-01 2.05394074e-01
-3.62958536e-02 -2.70449460e-01 1.34875512e+00 -8.58517643e-03
5.01703084e-01 1.26419318e+00 6.71989441e-01 3.48953635e-01
-1.50490832e+00 -3.73267263e-01 5.32719076e-01 1.81707695e-01
-8.29644799e-01 -1.12749547e-01 5.52943587e-01 -1.09024227e-01
5.02118945e-01 8.34470034e-01 9.53907549e-01 5.18624544e-01
5.26979715e-02 1.25642657e+00 1.21012592e+00 -6.90874219e-01
7.15062380e-01 -1.82264879e-01 5.05405255e-02 6.06518269e-01
3.66591096e-01 6.01780295e-01 -4.51969028e-01 -2.81980574e-01
5.74365556e-01 -3.11689258e-01 -1.92108124e-01 -4.90201950e-01
-5.81148565e-01 1.17025983e+00 6.39858961e-01 2.60707319e-01
-5.04943371e-01 5.38401306e-01 3.03444594e-01 3.33905816e-01
7.08350182e-01 2.59513140e-01 -4.76361990e-01 -4.56677556e-01
-1.23610258e+00 3.79080653e-01 9.32005227e-01 7.61623323e-01
4.90634918e-01 -2.88824797e-01 -3.76419157e-01 6.09708726e-01
-5.25060929e-02 1.65156230e-01 2.92072296e-02 -1.18335867e+00
6.17799163e-01 -2.37786956e-03 6.44308209e-01 -3.78971577e-01
-2.52802610e-01 -6.64373279e-01 -5.98124802e-01 4.50126678e-01
6.30310893e-01 -1.37530386e-01 -9.18566883e-01 1.68754780e+00
5.22581816e-01 -1.66294649e-01 -1.49143681e-01 5.13392627e-01
-4.15427387e-01 6.20102644e-01 -3.72011632e-01 -7.99241543e-01
8.86838436e-01 -9.20727968e-01 -4.81489480e-01 -1.40898556e-01
8.68053675e-01 -2.55546749e-01 6.72295570e-01 9.20168519e-01
-1.35738921e+00 1.37739718e-01 -8.21144879e-01 4.90734309e-01
1.67435348e-01 -4.83475685e-01 7.83683062e-01 1.12729108e+00
-8.53017211e-01 1.04982746e+00 -9.53275502e-01 -5.45648560e-02
6.31819427e-01 2.63330400e-01 1.62615657e-01 -6.52002394e-02
-6.16492093e-01 7.79983103e-01 6.29850805e-01 2.45622322e-01
-9.29667711e-01 -4.83170152e-01 -3.76218587e-01 2.27383569e-01
9.20083940e-01 -3.59897196e-01 1.81560922e+00 -1.07254076e+00
-1.58075714e+00 5.41449428e-01 -1.34519219e-01 -1.10862935e+00
1.09402716e+00 -2.71732748e-01 3.13396513e-01 6.32991791e-02
1.41405180e-01 3.06547523e-01 6.35156631e-01 -9.96266127e-01
-1.09426582e+00 -4.48833525e-01 3.31978858e-01 2.07516789e-01
5.28487526e-02 -1.99277312e-01 6.22745566e-02 -5.52977443e-01
2.68821150e-01 -7.93597937e-01 -7.64096737e-01 -1.56725496e-01
-4.45400804e-01 -5.06799994e-03 6.84880093e-02 -4.25391912e-01
1.39692497e+00 -1.65073442e+00 1.19447462e-01 4.14124012e-01
-1.90741703e-01 -2.27669641e-01 1.81819409e-01 3.84590983e-01
7.44763464e-02 3.63785297e-01 -3.37089449e-01 -1.55108748e-02
-8.16015303e-02 6.95102751e-01 -3.10550779e-01 4.01886851e-01
-4.84284759e-01 7.40250409e-01 -9.60062385e-01 -5.13222694e-01
-7.25770136e-03 -6.25682831e-01 -8.26504886e-01 -2.40011364e-02
-5.92176318e-01 3.00168078e-02 -7.34798133e-01 5.56451857e-01
5.23498714e-01 -9.40167718e-03 1.35403350e-01 5.42583704e-01
-3.48425984e-01 1.28064781e-01 -1.10094821e+00 1.24056923e+00
-5.83104968e-01 1.09482877e-01 3.66344273e-01 -1.39236319e+00
5.49974263e-01 3.96687128e-02 5.26067138e-01 -1.91118777e-01
2.30363593e-01 4.75783437e-01 -1.95088208e-01 -3.77612896e-02
2.00392932e-01 -4.62724239e-01 -1.00392122e-02 3.44447568e-02
-5.37088037e-01 -4.07181948e-01 2.67266780e-01 1.53315654e-02
1.22257411e+00 2.08114255e-02 3.90480965e-01 -3.59939307e-01
1.14977449e-01 1.19495906e-01 5.26575327e-01 1.27476895e+00
-4.87696230e-02 2.06698924e-01 9.71424818e-01 -2.48147815e-01
-7.98611045e-01 -7.79641151e-01 -1.72737226e-01 1.14333236e+00
-6.12503178e-02 -2.75075406e-01 -8.41941953e-01 -7.82298684e-01
1.38211653e-01 1.13424313e+00 -9.82637525e-01 1.58921182e-01
-3.29073608e-01 -6.86933458e-01 5.56420069e-03 2.84648508e-01
4.54501897e-01 -7.47139931e-01 -1.03924704e+00 7.00136721e-01
-2.29354203e-01 -5.73227763e-01 -3.40380847e-01 9.41782057e-01
-1.43601346e+00 -6.88961864e-01 -7.71906435e-01 -1.39119610e-01
3.42338592e-01 1.58919487e-02 9.87386346e-01 -2.02096865e-01
1.06095411e-01 1.27754480e-01 -5.39826095e-01 -3.77705812e-01
-3.59595776e-01 1.04319535e-01 -3.20183039e-01 -2.46846735e-01
-2.14021802e-01 -7.02074170e-01 -3.82993042e-01 1.36974335e-01
-3.92448753e-01 9.45740715e-02 4.24244970e-01 1.10383224e+00
7.69965291e-01 3.25907201e-01 3.33139151e-01 -1.00269878e+00
3.82128716e-01 -2.71453708e-01 -1.10491705e+00 3.73804420e-01
-7.71727860e-01 4.55346227e-01 3.80794436e-01 -3.23063016e-01
-9.47489381e-01 3.42872180e-02 -9.54602063e-02 -3.81893694e-01
3.96861255e-01 6.54835701e-01 2.60470331e-01 -6.87889606e-02
7.37788439e-01 -1.00233950e-01 -4.16522712e-01 -5.69707870e-01
5.37711561e-01 3.90274107e-01 3.89968246e-01 -1.12937951e+00
4.44704086e-01 3.95224124e-01 3.19805652e-01 -2.81149954e-01
-1.43777442e+00 -3.33222747e-01 -2.77092993e-01 -2.39728376e-01
3.71467203e-01 -2.95757055e-01 -8.95610690e-01 2.05523577e-02
-8.56265008e-01 -6.23236060e-01 -8.48833919e-01 2.79942900e-01
-1.23253572e+00 4.30815786e-01 -4.01151776e-01 -1.71485531e+00
-2.29749784e-01 -7.10587502e-01 7.73763895e-01 2.48823017e-01
6.63648546e-02 -7.20666230e-01 -3.82378548e-02 2.67840415e-01
-1.01848152e-02 3.88785988e-01 6.60513759e-01 -4.20448989e-01
-6.83723390e-01 -2.62428552e-01 -2.35037301e-02 1.52983442e-01
-2.24634036e-01 -2.10799977e-01 -5.72751582e-01 -3.26295495e-01
1.03009209e-01 -3.18042338e-01 1.08060384e+00 1.01870632e+00
1.32181585e+00 -7.89779305e-01 -4.18689489e-01 5.28324366e-01
1.31928694e+00 3.74969065e-01 2.96213806e-01 6.86923623e-01
-8.42156932e-02 5.77999890e-01 1.10323441e+00 7.59528518e-01
-1.69396058e-01 5.80060542e-01 8.33792448e-01 4.03110445e-01
5.75332582e-01 -3.85220051e-01 2.60575682e-01 2.09779590e-01
-3.21514338e-01 -1.68704644e-01 -4.85626668e-01 7.75141656e-01
-2.15455294e+00 -1.02739859e+00 -9.23268795e-02 2.65180540e+00
1.08471715e+00 7.37241030e-01 4.97774720e-01 3.17973316e-01
6.79373324e-01 1.66563783e-02 -7.60500371e-01 -9.77923095e-01
2.44742841e-01 3.89587641e-01 1.14587295e+00 8.03772330e-01
-8.18336368e-01 8.89506638e-01 6.79072332e+00 1.22122645e+00
-6.12152576e-01 1.92716956e-01 9.40105915e-01 -2.15907067e-01
-3.12296480e-01 4.18713450e-01 -7.42229521e-01 4.92319167e-01
8.65728438e-01 -4.55638289e-01 7.13692069e-01 1.35269153e+00
1.18608721e-01 -5.80324769e-01 -1.00167561e+00 5.06050110e-01
-7.78192818e-01 -1.31170309e+00 -4.81284231e-01 2.79387861e-01
6.82729661e-01 -5.61687909e-02 -1.36369959e-01 2.35120326e-01
9.93619621e-01 -8.82887423e-01 8.92385006e-01 1.33723646e-01
6.71160817e-01 -1.09722126e+00 4.89249706e-01 9.33829963e-01
-9.90329623e-01 -4.28253561e-01 -4.79424745e-01 -2.00208455e-01
3.83093685e-01 8.16596985e-01 -7.46959269e-01 7.05467880e-01
7.68287122e-01 1.56799987e-01 2.22694322e-01 1.38574696e+00
-1.99870959e-01 8.72803152e-01 -1.01194155e+00 -3.06008101e-01
7.14292169e-01 -4.48815107e-01 6.21198535e-01 1.00873435e+00
5.25905430e-01 2.16296911e-01 1.54435351e-01 6.14754677e-01
2.35309139e-01 8.19798112e-02 -1.90199330e-01 1.05908364e-01
3.62039119e-01 8.54233265e-01 -7.73572981e-01 -3.14790159e-01
8.39587599e-02 6.83465362e-01 3.88169497e-01 9.59267691e-02
-7.26850986e-01 -4.57062721e-02 2.40434721e-01 -6.90029338e-02
7.72584140e-01 2.21930102e-01 -5.94314694e-01 -8.64550591e-01
6.09263182e-02 -4.64882523e-01 8.21970165e-01 -5.35995245e-01
-8.56342256e-01 2.18885854e-01 3.13682079e-01 -8.28639269e-01
-5.81203043e-01 -3.62424105e-01 -4.81194466e-01 7.14991271e-01
-1.30757499e+00 -7.57955968e-01 4.20545489e-01 2.51789719e-01
7.07794905e-01 3.80795687e-01 4.79366362e-01 -3.66832256e-01
-3.10631394e-01 5.54849088e-01 5.62368691e-01 -4.92071033e-01
-2.88255841e-01 -1.44757044e+00 1.09866455e-01 8.76616359e-01
2.25815818e-01 1.06099963e-01 1.13703036e+00 -7.48849750e-01
-8.76866758e-01 -6.96708679e-01 5.25432110e-01 -1.97620075e-02
8.99083674e-01 -6.70991316e-02 -4.81976807e-01 8.07619095e-01
-2.63451576e-01 -1.24374926e-01 -5.41589828e-03 5.50938666e-01
-1.64128188e-02 -3.23087722e-01 -1.27790034e+00 4.44721907e-01
1.06091464e+00 1.62408605e-01 -3.58173013e-01 4.87004429e-01
6.74585521e-01 -6.25729620e-01 -7.56100178e-01 4.46873695e-01
6.51522875e-01 -1.32673430e+00 6.60009086e-01 -6.31164968e-01
1.34995908e-01 2.63727576e-01 -1.08451165e-01 -1.21833801e+00
-2.81519592e-01 -1.29006028e+00 -2.26588860e-01 7.83920944e-01
3.64810884e-01 -6.13641441e-01 1.32847214e+00 2.84880400e-01
-1.13911564e-02 -1.15009856e+00 -1.60285652e+00 -1.16353261e+00
4.39119846e-01 -5.97622573e-01 2.82196581e-01 3.60915244e-01
2.14569822e-01 -6.47026896e-02 -5.03042519e-01 -1.36487737e-01
8.16158533e-01 4.03385311e-01 4.71913010e-01 -1.23044538e+00
-7.46788025e-01 -6.12963378e-01 1.30552605e-01 -1.42092216e+00
-2.74382442e-01 -5.38694799e-01 2.30247974e-01 -1.25463378e+00
1.55138314e-01 -7.00018704e-01 -2.97106683e-01 4.08402026e-01
-2.52660527e-03 -5.56261778e-01 2.74073809e-01 7.29685947e-02
-4.17453140e-01 3.89066339e-01 1.29478204e+00 2.02623401e-02
-3.21189016e-01 8.74470294e-01 -7.88679183e-01 7.91776478e-01
7.27711141e-01 -8.08963656e-01 -2.04355732e-01 1.77429378e-01
3.74129593e-01 8.45213234e-01 -4.28189263e-02 -6.89365983e-01
-1.25031665e-01 -5.96188784e-01 -1.12464532e-01 -8.87904406e-01
3.25510859e-01 -8.59827638e-01 1.59474880e-01 9.22330678e-01
-5.96042395e-01 -5.16804755e-01 -9.37640294e-02 8.88442338e-01
1.19797677e-01 -1.05508208e+00 9.49075639e-01 -2.54038841e-01
9.70427766e-02 2.52814084e-01 -2.97229469e-01 -5.49721718e-02
9.79852080e-01 -3.44889879e-01 2.02646837e-01 -7.22361565e-01
-8.35750580e-01 5.26144087e-01 1.85249522e-01 -4.13256615e-01
-4.21038829e-02 -1.05812132e+00 -6.31263733e-01 -4.19862539e-01
-2.91697532e-01 2.07257062e-01 1.00318596e-01 7.71744549e-01
-4.08864081e-01 4.14811462e-01 2.00237989e-01 -4.22406852e-01
-1.17982674e+00 8.75902116e-01 3.78440917e-01 -9.42903697e-01
-5.22961855e-01 1.08444786e+00 7.48388618e-02 1.06396198e-01
4.26551580e-01 -7.99391627e-01 2.51510859e-01 -1.65955033e-02
2.18994245e-01 4.46369797e-01 -1.45826831e-01 1.72043771e-01
-6.48307800e-03 3.89361024e-01 -5.20135053e-02 -5.52266777e-01
1.29077435e+00 -2.07838684e-01 6.85128793e-02 3.95487607e-01
6.53360665e-01 3.99502926e-02 -1.44281864e+00 -1.77482650e-01
3.10164303e-01 -6.60724163e-01 1.49091974e-01 -9.20619249e-01
-8.69015574e-01 8.54274690e-01 2.56162763e-01 8.49134326e-01
1.21746504e+00 -3.99866179e-02 5.30609369e-01 5.91352224e-01
7.93977857e-01 -1.25582600e+00 -3.62501860e-01 3.25324744e-01
8.25538337e-01 -8.38004708e-01 1.69390813e-01 -3.40572268e-01
-4.62350249e-01 1.21013975e+00 4.21097539e-02 8.01295321e-03
3.51186842e-01 2.62264401e-01 -6.39848530e-01 2.79297680e-01
-9.16370034e-01 -5.28971195e-01 -2.11309358e-01 4.36156720e-01
-6.46957681e-02 4.21357840e-01 -6.05112195e-01 4.51633304e-01
-4.71495777e-01 1.78313583e-01 3.87977421e-01 7.92942643e-01
-9.87354100e-01 -1.08589768e+00 -6.22512162e-01 6.08828127e-01
-7.45063186e-01 -1.43817076e-02 -1.55384958e-01 7.53495753e-01
-2.45904341e-01 9.75039005e-01 -2.32693180e-01 -1.28928572e-02
6.99873865e-02 -1.68571204e-01 8.03454161e-01 -2.65560359e-01
-3.11897337e-01 4.81386721e-01 6.07936859e-01 -5.89963317e-01
-2.40912870e-01 -8.00677598e-01 -7.91751146e-01 -5.65023899e-01
-8.14467609e-01 6.18379414e-01 5.23871779e-01 9.33765233e-01
-3.53384882e-01 2.14512408e-01 7.95932710e-01 -6.26831949e-01
-1.25371993e+00 -7.06263185e-01 -8.33075702e-01 -3.89821202e-01
2.59004265e-01 -5.01097739e-01 -5.16778052e-01 -4.37566429e-01] | [4.517509460449219, 3.2748262882232666] |
8a0e7f2e-3e98-42db-9ea8-fb27a2c0dd2e | upar-unified-pedestrian-attribute-recognition | 2209.02522 | null | https://arxiv.org/abs/2209.02522v1 | https://arxiv.org/pdf/2209.02522v1.pdf | UPAR: Unified Pedestrian Attribute Recognition and Person Retrieval | Recognizing soft-biometric pedestrian attributes is essential in video surveillance and fashion retrieval. Recent works show promising results on single datasets. Nevertheless, the generalization ability of these methods under different attribute distributions, viewpoints, varying illumination, and low resolutions remains rarely understood due to strong biases and varying attributes in current datasets. To close this gap and support a systematic investigation, we present UPAR, the Unified Person Attribute Recognition Dataset. It is based on four well-known person attribute recognition datasets: PA100K, PETA, RAPv2, and Market1501. We unify those datasets by providing 3,3M additional annotations to harmonize 40 important binary attributes over 12 attribute categories across the datasets. We thus enable research on generalizable pedestrian attribute recognition as well as attribute-based person retrieval for the first time. Due to the vast variance of the image distribution, pedestrian pose, scale, and occlusion, existing approaches are greatly challenged both in terms of accuracy and efficiency. Furthermore, we develop a strong baseline for PAR and attribute-based person retrieval based on a thorough analysis of regularization methods. Our models achieve state-of-the-art performance in cross-domain and specialization settings on PA100k, PETA, RAPv2, Market1501-Attributes, and UPAR. We believe UPAR and our strong baseline will contribute to the artificial intelligence community and promote research on large-scale, generalizable attribute recognition systems. | ['Jürgen Beyerer', 'Mickael Cormier', 'Andreas Specker'] | 2022-09-06 | null | null | null | null | ['pedestrian-attribute-recognition', 'person-retrieval'] | ['computer-vision', 'computer-vision'] | [-3.34866196e-02 -6.49347723e-01 -2.85438627e-01 -7.76529431e-01
-8.70677114e-01 -6.60611331e-01 7.40127981e-01 1.05779216e-01
-2.56534636e-01 6.62281454e-01 3.17757905e-01 3.53662223e-01
-1.10973820e-01 -6.70626760e-01 -4.82763618e-01 -8.13655615e-01
1.06065586e-01 8.10639679e-01 -1.16263188e-01 -1.09685864e-02
-3.54889125e-01 4.80983883e-01 -1.80807889e+00 3.82254690e-01
5.61861515e-01 1.24772382e+00 -5.16299009e-01 4.43192005e-01
3.99561614e-01 1.16448909e-01 -4.48293507e-01 -1.19254529e+00
3.32618147e-01 3.45506556e-02 -5.58475435e-01 5.42316437e-02
1.27045584e+00 -3.63026470e-01 -5.99512398e-01 7.38006592e-01
9.03981686e-01 -4.42792028e-02 1.18873215e+00 -1.68018281e+00
-1.07054102e+00 9.70041230e-02 -5.21647453e-01 -1.35273173e-01
6.96204424e-01 4.05634969e-01 8.25836420e-01 -9.10752475e-01
4.14141446e-01 1.56472135e+00 1.09203458e+00 6.74473107e-01
-1.38577366e+00 -7.14301109e-01 2.70298570e-01 3.79537880e-01
-1.80756652e+00 -6.22931778e-01 5.23197055e-01 -3.88351679e-01
5.59796274e-01 4.20047611e-01 7.33666420e-01 1.84036624e+00
-5.71197607e-02 9.25624073e-01 1.30990505e+00 -1.05354615e-01
-1.38229564e-01 3.43400598e-01 3.64394277e-01 4.84195024e-01
4.66078490e-01 -6.32112324e-02 -6.87538505e-01 -3.70489895e-01
5.27764320e-01 -1.21926323e-01 -8.35284125e-03 -6.33047044e-01
-1.14797628e+00 5.83751202e-01 -1.55525105e-02 -3.95258099e-01
-2.95170605e-01 -5.89327514e-02 6.58830643e-01 3.14941764e-01
3.91845018e-01 8.03627670e-02 -5.28353930e-01 -1.18391231e-01
-5.33679783e-01 4.73581195e-01 7.80024946e-01 1.34800315e+00
4.47211117e-01 -9.70911309e-02 -4.40844178e-01 1.16691327e+00
2.34262824e-01 1.28297663e+00 -1.08587943e-01 -9.02164042e-01
2.12686434e-01 4.03213561e-01 1.25978142e-01 -1.06129968e+00
-2.52707571e-01 -2.08779365e-01 -8.87522280e-01 -5.72820269e-02
6.57452941e-01 1.71201676e-01 -7.80790210e-01 1.70780170e+00
2.65214384e-01 -1.53123543e-01 5.60031098e-04 1.00434339e+00
1.15642214e+00 2.06865028e-01 6.40019655e-01 2.22196087e-01
1.87018716e+00 -6.82082236e-01 -4.43324178e-01 -2.21794069e-01
2.41406754e-01 -7.53880084e-01 1.14930916e+00 2.24806249e-01
-7.76235819e-01 -7.08919287e-01 -8.88284802e-01 1.19094715e-01
-6.24919295e-01 4.55567002e-01 8.15250218e-01 1.08658588e+00
-8.10494661e-01 -4.46195155e-02 -2.28136197e-01 -9.88734424e-01
7.66945302e-01 4.07341361e-01 -7.09906518e-01 -2.88909584e-01
-1.15940094e+00 8.11036885e-01 -1.64900854e-01 -2.32665241e-01
-8.21845829e-01 -8.20602179e-01 -7.94148326e-01 -2.22788408e-01
2.52583236e-01 -1.15305793e+00 5.84175646e-01 -5.56654632e-01
-1.08712161e+00 1.44921196e+00 -1.47096932e-01 -3.68423074e-01
4.73618031e-01 -3.62461925e-01 -6.61085784e-01 -3.00395377e-02
1.77384913e-01 6.77934170e-01 7.50178039e-01 -1.12779570e+00
-4.60562050e-01 -8.66263032e-01 -1.51374608e-01 1.25713155e-01
-4.37075198e-01 2.38179386e-01 -5.44065773e-01 -7.07196712e-01
-2.63266325e-01 -1.10512376e+00 1.35163814e-01 8.46698657e-02
-3.50565463e-01 -2.88770497e-01 5.95750153e-01 -4.78020847e-01
6.36365235e-01 -2.28270030e+00 1.16513642e-02 4.21322696e-02
7.12561756e-02 1.80048212e-01 -2.81270295e-01 5.09900227e-02
1.45973638e-01 -1.22147232e-01 2.03304306e-01 -4.09274578e-01
2.96332568e-01 1.71593875e-01 -1.89746395e-01 6.79343164e-01
1.05336584e-01 1.01944959e+00 -5.75000226e-01 -7.90869594e-01
3.08976203e-01 6.83202684e-01 -3.05902869e-01 1.78379491e-01
2.27559209e-01 3.52822602e-01 -6.50893807e-01 1.32435226e+00
7.74742603e-01 1.66540802e-01 -2.67174691e-01 -6.61258161e-01
2.20508933e-01 -3.95920128e-01 -1.27908444e+00 1.33752167e+00
1.28659252e-02 5.40443122e-01 1.00619830e-01 -8.38749230e-01
1.04524684e+00 1.52977362e-01 6.71156168e-01 -5.94387174e-01
5.64811267e-02 6.71211109e-02 -6.03336036e-01 -3.59679908e-01
5.62598705e-01 2.83019900e-01 -4.90202248e-01 -6.32092655e-02
1.42739862e-01 3.46372128e-01 6.62890300e-02 7.05634207e-02
4.90555763e-01 1.35298073e-01 2.67437667e-01 -4.27676409e-01
7.88114250e-01 -2.39245161e-01 5.06628692e-01 1.23364401e+00
-7.73028791e-01 8.02447081e-01 2.57823020e-01 -9.92851436e-01
-1.25513482e+00 -1.27070403e+00 -6.63517356e-01 1.42766035e+00
1.90172076e-01 -3.41067046e-01 -6.03776991e-01 -6.00585997e-01
4.16526824e-01 1.14179283e-01 -7.64217913e-01 -1.13999076e-01
-3.78313422e-01 -1.08231366e+00 9.60395217e-01 8.41165841e-01
8.56286645e-01 -7.49313593e-01 -1.16856573e-02 -1.24502949e-01
-3.55368406e-01 -1.71438634e+00 -5.53625762e-01 -4.59034413e-01
-1.77533850e-01 -1.11735737e+00 -9.20439839e-01 -5.21129131e-01
2.79699624e-01 1.32005811e-01 1.55049229e+00 -2.41285905e-01
-3.68910402e-01 1.17096162e+00 -2.17956170e-01 -5.70317149e-01
-5.13936132e-02 3.52515951e-02 8.25079083e-01 3.03382993e-01
8.91658127e-01 -2.68917471e-01 -6.45563662e-01 8.16752255e-01
-1.36928603e-01 -3.81436884e-01 5.32629967e-01 7.38213539e-01
7.55681694e-01 -5.09488702e-01 5.22900045e-01 -5.56087434e-01
2.88242608e-01 -2.48976409e-01 -4.23661739e-01 5.27753532e-01
-5.21924078e-01 -3.09204519e-01 2.03016654e-01 -6.26642525e-01
-1.04181075e+00 1.20928586e-01 -2.83907680e-03 -1.83826253e-01
-7.05794394e-01 -3.17495644e-01 -8.02565932e-01 -2.13708416e-01
6.58160508e-01 2.37102449e-01 -5.65663502e-02 -3.05403382e-01
3.88704896e-01 7.90004969e-01 9.27564561e-01 -1.12835968e+00
6.20999396e-01 6.24326885e-01 1.37334898e-01 -9.83676791e-01
-8.57257247e-01 -5.15306234e-01 -7.82469749e-01 -2.73885667e-01
1.12211728e+00 -1.26551580e+00 -9.80899215e-01 9.93005514e-01
-8.03427339e-01 1.24528065e-01 -1.83215976e-01 3.36739928e-01
-5.86603343e-01 4.51238453e-01 -4.86022949e-01 -7.92559445e-01
-4.50877279e-01 -1.09874558e+00 1.33447492e+00 2.08862051e-01
-3.15946072e-01 -6.14366651e-01 -3.45752299e-01 7.80030370e-01
4.36710864e-01 2.69432813e-01 6.44712508e-01 -8.95697057e-01
-4.77898479e-01 -4.63065982e-01 -5.00916064e-01 1.39735535e-01
-3.82789969e-02 -8.09145719e-02 -1.19011664e+00 -4.10838008e-01
-7.14107513e-01 -6.90560400e-01 9.16343570e-01 5.06813586e-01
1.16188693e+00 3.87286209e-02 -4.75462705e-01 8.93238664e-01
8.50353062e-01 -1.85669422e-01 6.32467985e-01 4.82806325e-01
7.49761283e-01 6.41301453e-01 6.55584574e-01 4.93300706e-01
6.82023287e-01 1.16765213e+00 1.11411534e-01 -1.42501026e-01
-3.99999648e-01 8.10621828e-02 1.17210232e-01 1.11588001e-01
-4.24008995e-01 -1.22790135e-01 -7.35930324e-01 4.51812476e-01
-1.68474793e+00 -1.09947371e+00 -1.69388875e-01 2.40658140e+00
4.59860355e-01 -2.91878104e-01 5.86731672e-01 -1.15244076e-01
5.43215871e-01 1.17081031e-01 -3.59295726e-01 -3.98337878e-02
-8.32361639e-01 -2.53058761e-01 4.64389443e-01 2.59206221e-02
-1.80576634e+00 8.55336189e-01 6.55041933e+00 8.45899880e-01
-4.85567421e-01 -1.70165002e-01 8.72763038e-01 2.03927327e-02
3.56063038e-01 -5.97866774e-01 -1.32398355e+00 5.18592954e-01
6.95115685e-01 1.27545059e-01 2.38850296e-01 9.41278100e-01
-3.93189818e-01 2.44347498e-01 -1.42951608e+00 1.41913259e+00
3.91414821e-01 -8.95222187e-01 3.28640252e-01 5.94493337e-02
4.28387910e-01 -3.15112591e-01 4.59488928e-01 5.31306982e-01
1.08221337e-01 -1.11140513e+00 3.10214758e-01 6.20548666e-01
9.92754340e-01 -6.42161608e-01 1.02108872e+00 -3.64634961e-01
-1.34937322e+00 -2.01146281e-03 -5.66533983e-01 2.81388283e-01
-1.73752621e-01 5.55946454e-02 -1.70904100e-01 5.17594814e-01
1.22747076e+00 7.74982810e-01 -8.51018965e-01 8.29887033e-01
2.52207667e-01 4.40434933e-01 -4.28734124e-01 1.86295152e-01
-1.78640410e-01 -2.10752070e-01 5.23314893e-01 1.47341573e+00
1.95903722e-02 3.54623735e-01 5.25667846e-01 4.53052908e-01
-1.53379655e-03 3.61014605e-02 -5.21352232e-01 2.80922025e-01
7.26193488e-01 1.24429286e+00 -3.16439897e-01 -3.61082166e-01
-6.35255635e-01 9.10335958e-01 -4.19816598e-02 5.44133961e-01
-9.59864616e-01 2.49171034e-01 1.33334601e+00 1.77099720e-01
2.00407162e-01 -7.04258075e-03 -2.69162714e-01 -1.20435333e+00
-9.74569097e-03 -1.18757653e+00 8.99977624e-01 -5.15525460e-01
-1.93519831e+00 5.35605848e-01 3.33679080e-01 -1.08230257e+00
-8.28983635e-02 -8.49982798e-01 -6.16404153e-02 8.06045830e-01
-1.23775291e+00 -1.97849154e+00 -7.67598033e-01 9.43964005e-01
4.47806448e-01 -8.62250865e-01 1.12671244e+00 6.29719615e-01
-6.07891142e-01 1.30549264e+00 -5.50817400e-02 3.81059080e-01
1.32989597e+00 -9.51255441e-01 3.57325703e-01 3.11447263e-01
1.16639007e-02 5.29853106e-01 7.20108271e-01 -4.39387143e-01
-1.56026280e+00 -1.01755357e+00 4.54733700e-01 -1.25289309e+00
3.42738748e-01 -5.19603431e-01 -8.06459844e-01 6.29551589e-01
-5.37207797e-02 3.61799330e-01 9.66463923e-01 5.36629617e-01
-7.36133397e-01 -5.69221377e-01 -1.25902057e+00 2.69451618e-01
1.13348603e+00 -4.55136657e-01 -1.85393408e-01 1.52765885e-01
2.78381199e-01 -2.56661534e-01 -1.25343013e+00 5.94452560e-01
1.09219480e+00 -7.25988269e-01 1.81085694e+00 -7.16603279e-01
5.26687875e-02 -3.24795023e-02 -6.91475034e-01 -9.27180827e-01
-6.82106018e-01 -7.42608532e-02 -5.14417663e-02 1.76527548e+00
1.07784048e-01 -6.90318465e-01 8.58822405e-01 9.50890064e-01
1.44450679e-01 -6.94826543e-01 -8.78174067e-01 -9.84478593e-01
6.96006864e-02 -2.32102558e-01 9.46891129e-01 8.05792749e-01
-7.88051605e-01 1.68111473e-01 -7.93322146e-01 3.08065146e-01
1.29221833e+00 -9.39020421e-03 1.23985302e+00 -1.55229676e+00
6.94314539e-02 -2.68357605e-01 -9.40478623e-01 -7.60447860e-01
2.10235626e-01 -4.43525761e-01 -4.84566718e-01 -1.13549280e+00
7.90575445e-01 -6.91702366e-01 -2.21634716e-01 5.00893950e-01
-2.79744267e-01 7.66846955e-01 2.04601794e-01 3.70161265e-01
-8.44415367e-01 5.70905805e-01 7.89326310e-01 -5.49683392e-01
2.21474782e-01 -5.27283037e-03 -6.69871688e-01 6.61023915e-01
5.57178378e-01 -1.93210337e-02 -7.45110959e-02 -3.23970318e-01
-2.73358703e-01 -4.35078114e-01 8.25639069e-01 -1.12630308e+00
1.20112054e-01 -5.69925196e-02 1.11259556e+00 -4.92321372e-01
9.16157782e-01 -8.46879184e-01 1.29128814e-01 -8.48899856e-02
-1.02262527e-01 1.14423051e-01 1.50653034e-01 5.94807565e-01
-2.00798027e-02 4.33925718e-01 6.66746676e-01 6.88441172e-02
-1.06411898e+00 7.69228935e-01 -3.76526974e-02 2.55011916e-01
1.00226367e+00 -5.90623677e-01 -4.95275527e-01 -4.54060405e-01
-5.87147355e-01 2.82571286e-01 5.59780300e-01 5.11932075e-01
5.65148652e-01 -1.69808292e+00 -1.23396456e+00 2.93794364e-01
6.07906938e-01 -6.63796484e-01 3.01397473e-01 6.12306476e-01
6.91277906e-02 4.12449062e-01 -5.97164035e-01 -8.31324697e-01
-1.64447069e+00 6.74284697e-01 2.05966935e-01 4.79736999e-02
-4.89755571e-01 6.60692930e-01 3.25668693e-01 -4.00965393e-01
3.20387691e-01 6.05350018e-01 -3.14564019e-01 1.85931936e-01
5.40623963e-01 5.56360781e-01 -2.76778579e-01 -1.12133324e+00
-7.27332413e-01 8.41932118e-01 -1.15003914e-01 4.89520997e-01
9.79452431e-01 -1.94879532e-01 1.62701458e-01 1.20644718e-02
8.71709943e-01 -5.91892824e-02 -1.12890887e+00 -2.52285421e-01
-3.07423025e-01 -6.75004780e-01 -5.63115537e-01 -7.38768518e-01
-9.10544574e-01 4.79794711e-01 1.01385283e+00 -1.64944246e-01
1.07983112e+00 3.36220652e-01 6.38166845e-01 4.14069235e-01
5.19270122e-01 -1.04091477e+00 -3.10537685e-02 2.53087759e-01
7.49118865e-01 -1.65023088e+00 3.74346912e-01 -6.71161532e-01
-8.33130598e-01 6.90160453e-01 9.34300959e-01 2.75491893e-01
2.18617782e-01 2.20098004e-01 1.86897248e-01 -2.31527928e-02
-2.87777126e-01 -5.84816821e-02 6.29526913e-01 1.23118031e+00
4.38851744e-01 3.21442753e-01 9.14440900e-02 8.72854471e-01
-2.46470436e-01 -4.03772771e-01 3.97321656e-02 3.25308025e-01
6.54137805e-02 -1.12035787e+00 -6.07549071e-01 4.43574101e-01
-4.00691867e-01 1.13375753e-01 -4.97223765e-01 9.03722048e-01
8.34555402e-02 8.03616703e-01 -1.46172736e-02 -2.70377666e-01
6.55438304e-01 1.56203270e-01 6.05238020e-01 1.80829868e-01
-3.85193020e-01 -4.29019719e-01 4.75282192e-01 -5.73144794e-01
-4.87152636e-01 -9.33226466e-01 -3.44733626e-01 -6.52137160e-01
1.60072938e-01 -2.46340767e-01 2.54505903e-01 6.68623745e-01
4.45706904e-01 5.93473464e-02 2.19371006e-01 -6.95509553e-01
-6.37390852e-01 -7.49237299e-01 -3.83073658e-01 1.04851151e+00
5.29104657e-02 -9.65271413e-01 7.42809623e-02 1.22985907e-01] | [14.530478477478027, 0.9476910829544067] |
cfd7f9ec-ee80-47f7-9dc9-bc2a750a985f | robust-and-efficient-post-processing-for | null | null | https://arxiv.org/abs/2009.11050 | https://arxiv.org/pdf/2009.11050.pdf | Robust and Efficient Post-Processing for Video Object Detection (REPP) | Object recognition in video is an important task for plenty of applications, including autonomous driving perception, surveillance tasks, wearable devices or IoT networks. Object recognition using video data is more challenging than using still images due to blur, occlusions or rare object poses. Specific video detectors with high computational cost or standard image detectors together with a fast post-processing algorithm achieve the current state-of-the-art. This work introduces a novel post-processing pipeline that overcomes some of the limitations of previous post-processing methods by introducing a learning-based similarity evaluation between detections across frames. Our method improves the results of state-of-the-art specific video detectors, specially regarding fast moving objects, and presents low resource requirements. And applied to efficient still image detectors, such as YOLO, provides comparable results to much more computationally intensive detectors. | ['Luis Montesano', 'Alberto Sabater', 'Ana C. Murillo'] | 2020-10-01 | null | null | null | null | ['dense-object-detection'] | ['computer-vision'] | [ 2.85645455e-01 -6.24909341e-01 -4.70927842e-02 -2.10259154e-01
-4.57785636e-01 -3.62203687e-01 5.04622102e-01 2.11959258e-01
-8.89208078e-01 3.17320585e-01 -3.23320836e-01 3.25444750e-02
1.38851255e-01 -4.12556976e-01 -6.22907221e-01 -5.28714418e-01
-1.53503552e-01 6.98185414e-02 1.25519824e+00 -7.62437209e-02
2.88539380e-01 6.10887110e-01 -2.22633195e+00 5.54023445e-01
1.08477496e-01 1.29080534e+00 5.42355180e-01 1.00755155e+00
6.18919693e-02 8.00152898e-01 -6.05298698e-01 -1.02457479e-01
2.19856218e-01 -3.92896906e-02 -1.61779895e-01 2.05761820e-01
9.05803382e-01 -6.11273944e-01 -4.71764326e-01 1.05230439e+00
4.61401969e-01 1.04848035e-01 2.83583760e-01 -1.27033722e+00
-8.01085830e-02 -3.49565372e-02 -4.75971341e-01 9.49214399e-01
4.84813899e-01 2.76814818e-01 3.68873209e-01 -1.03299582e+00
5.12775183e-01 1.21668601e+00 7.38014162e-01 5.27734697e-01
-9.12378132e-01 -6.91990435e-01 9.36139002e-02 7.03342378e-01
-1.33763075e+00 -6.32402003e-01 3.56202602e-01 -5.61651647e-01
1.21010232e+00 1.52107477e-01 4.25601274e-01 8.89651895e-01
1.63977355e-01 8.10302317e-01 6.61714256e-01 -2.49138057e-01
1.96373135e-01 1.62307158e-01 1.18057296e-01 6.85894787e-01
7.09381163e-01 1.17367424e-01 -6.34374142e-01 7.07189515e-02
5.75927436e-01 3.04003656e-01 1.24973292e-02 -4.63938296e-01
-1.24307835e+00 6.01511657e-01 1.60867900e-01 3.66262525e-01
-4.87726122e-01 2.33532146e-01 6.52472854e-01 1.90864339e-01
2.72947967e-01 -5.59650846e-02 -4.16932851e-01 -2.36450016e-01
-1.30317485e+00 1.43430069e-01 4.85460430e-01 9.81015384e-01
7.21282005e-01 1.08243197e-01 -1.72060847e-01 4.04962450e-01
3.07245225e-01 5.65529764e-01 4.62556928e-01 -6.90685511e-01
3.76581281e-01 4.82390970e-01 1.24116845e-01 -8.78176153e-01
-7.49043226e-01 -3.78249139e-01 -5.54597557e-01 5.17558932e-01
4.84615654e-01 1.17202111e-01 -8.91069114e-01 9.35568213e-01
3.45538080e-01 3.65792334e-01 -2.20419113e-02 1.04197001e+00
1.17545283e+00 5.40942669e-01 1.81658655e-01 -3.27878147e-01
1.77074230e+00 -9.26176906e-01 -8.18021357e-01 -4.88829106e-01
4.53198254e-01 -9.84550416e-01 2.68618435e-01 4.51415002e-01
-8.14906061e-01 -1.05924344e+00 -1.24278450e+00 7.43297189e-02
-4.26909596e-01 4.57249016e-01 5.46049595e-01 1.06583154e+00
-8.58895302e-01 2.91738808e-01 -1.09843397e+00 -7.94209898e-01
5.38116038e-01 5.28651655e-01 -3.84136885e-01 -9.67937484e-02
-5.48171222e-01 1.03359079e+00 7.13984728e-01 -7.27253407e-02
-7.64054120e-01 -6.10946834e-01 -9.53589618e-01 -1.51263043e-01
6.36100471e-01 -5.98237693e-01 1.15187585e+00 -8.45047057e-01
-1.33295977e+00 8.94624650e-01 -2.24741995e-01 -7.94567466e-01
6.82487547e-01 -4.90572214e-01 -7.03278184e-01 3.14177990e-01
2.09474638e-02 7.83307076e-01 1.24141085e+00 -5.70727587e-01
-1.09324825e+00 -1.45127475e-01 -2.21956357e-01 -6.03036396e-02
-2.79584438e-01 5.34112990e-01 -7.88492501e-01 -3.91096681e-01
-1.14726081e-01 -8.56783867e-01 -1.84682339e-01 1.99367329e-01
2.25733817e-01 -3.16015631e-01 1.44757974e+00 -3.84603858e-01
1.03413618e+00 -2.22058463e+00 -2.10196689e-01 -2.02289447e-01
8.13886896e-02 8.86370897e-01 -1.19219393e-01 7.81495869e-02
1.48774326e-01 -6.12063646e-01 1.17537737e-01 -2.37505063e-01
-2.87146807e-01 3.23868729e-02 1.35402277e-01 7.06078351e-01
4.54258263e-01 6.63785994e-01 -8.26997519e-01 -4.06546086e-01
9.80772674e-01 5.40466130e-01 -3.14828962e-01 -6.79432377e-02
1.53928781e-02 2.08306447e-01 -1.80059403e-01 7.85160780e-01
7.68796265e-01 4.67714407e-02 -1.94577083e-01 -4.80791420e-01
-3.63245845e-01 -4.44748402e-02 -1.63344395e+00 1.48749566e+00
2.96492446e-02 1.14347267e+00 1.98993325e-01 -1.09423745e+00
8.09379876e-01 2.66203076e-01 4.74960029e-01 -6.90089405e-01
2.80916721e-01 2.68585116e-01 6.65859655e-02 -8.97689283e-01
8.51520002e-01 2.39378452e-01 3.13039601e-01 -1.04241960e-01
1.35791287e-01 1.09087214e-01 7.09123552e-01 -6.58887476e-02
1.27278185e+00 2.79780775e-01 4.80246753e-01 -1.60660878e-01
8.31078887e-01 6.48444593e-02 4.00394648e-01 1.04940355e+00
-5.17086446e-01 6.99297607e-01 -1.31048530e-01 -6.47793233e-01
-9.84694660e-01 -7.65354514e-01 -1.98568091e-01 1.05473137e+00
2.61450261e-01 -5.58242023e-01 -5.23276150e-01 -4.07328457e-01
8.76460969e-02 1.41238749e-01 -3.04864705e-01 1.12651154e-01
-7.37783492e-01 -7.14195430e-01 4.52778071e-01 6.53837144e-01
5.72558641e-01 -9.72644389e-01 -1.36827815e+00 4.01029170e-01
2.80474067e-01 -1.81359255e+00 -5.91986850e-02 -1.31434919e-02
-9.11224127e-01 -1.15413380e+00 -7.82159090e-01 -7.17190623e-01
3.62303227e-01 7.87001669e-01 9.86527622e-01 -3.40224802e-02
-8.05972993e-01 4.55124736e-01 -1.98528051e-01 -6.34990692e-01
-2.22764194e-01 -2.13572145e-01 3.46878052e-01 6.99628964e-02
6.61380529e-01 1.86802953e-01 -7.15968609e-01 4.96771604e-01
-9.17613804e-01 -2.97092706e-01 6.61349833e-01 4.92342144e-01
3.45206827e-01 1.34539470e-01 1.93943202e-01 -4.82170910e-01
4.17181812e-02 -4.68249694e-02 -1.11955106e+00 -1.22762481e-02
-1.10751726e-01 -1.57315746e-01 2.33349800e-01 -5.72784245e-01
-7.90067613e-01 5.87591946e-01 -8.35940167e-02 -5.66099942e-01
-3.77722055e-01 -1.17095232e-01 1.02813929e-01 -3.24176580e-01
7.08208263e-01 3.90814058e-02 6.02019615e-02 -4.84817147e-01
5.88431433e-02 7.04043567e-01 7.27579296e-01 1.22052990e-01
7.56070971e-01 7.72876561e-01 1.52755052e-01 -1.42713988e+00
-5.71114063e-01 -1.23180366e+00 -6.60882473e-01 -4.50785756e-01
8.86205912e-01 -1.22391975e+00 -7.87201822e-01 5.04875124e-01
-1.26677537e+00 2.50969410e-01 -6.24725483e-02 9.73367870e-01
-2.75594652e-01 5.30620694e-01 -3.90915006e-01 -9.46437895e-01
-1.55173123e-01 -1.19637346e+00 1.18452907e+00 4.60136622e-01
8.25532004e-02 -5.45766711e-01 -2.34176993e-01 3.33558857e-01
5.27418613e-01 5.93451113e-02 -2.39426389e-01 -2.81339735e-01
-7.89554060e-01 -4.78896290e-01 -4.83565658e-01 3.73347908e-01
-5.87166585e-02 8.48721638e-02 -1.07548475e+00 -3.48991662e-01
-2.10746765e-01 1.01853535e-01 1.32338321e+00 5.85992277e-01
7.69032598e-01 2.99560189e-01 -5.99736512e-01 3.65213573e-01
1.29737914e+00 3.16597581e-01 6.48024559e-01 6.61042988e-01
5.15131831e-01 4.46741104e-01 8.14633667e-01 4.15371001e-01
5.56935035e-02 1.04376614e+00 2.80818343e-01 -2.77765114e-02
-3.38375330e-01 4.47511107e-01 7.22526848e-01 1.94315732e-01
-1.81628779e-01 1.01709753e-01 -7.18618810e-01 4.73030448e-01
-2.12696624e+00 -1.24417078e+00 -5.47344267e-01 2.18652821e+00
7.23480619e-03 5.10092020e-01 4.64223146e-01 3.33421618e-01
7.76468396e-01 -3.59874405e-02 -3.51604342e-01 -1.70765758e-01
-9.33918506e-02 1.36636660e-01 9.86820340e-01 8.61338302e-02
-1.67313409e+00 7.82471955e-01 6.53873110e+00 6.99668288e-01
-1.19433391e+00 2.64398992e-01 -1.89813890e-03 -2.55767554e-01
9.39084589e-01 -3.00238103e-01 -1.28977418e+00 4.27644342e-01
9.17764068e-01 3.00022215e-01 -8.78090635e-02 1.03757977e+00
3.22096467e-01 -4.56690818e-01 -1.11686122e+00 1.56687188e+00
5.44536293e-01 -1.48903620e+00 -4.50759351e-01 -2.52925932e-01
4.52275157e-01 3.21971655e-01 -1.10732071e-01 1.21645562e-01
-4.33982104e-01 -5.78360140e-01 8.45640242e-01 2.62538522e-01
3.96797210e-01 -4.76263791e-01 8.77544224e-01 1.47395343e-01
-1.49100697e+00 -2.55525589e-01 -5.97103536e-01 -3.83874625e-01
2.98999727e-01 6.53286576e-01 -7.63935566e-01 3.28087717e-01
1.01540518e+00 8.30414176e-01 -8.15169156e-01 1.57078421e+00
1.04272172e-01 2.85813540e-01 -5.11746585e-01 -7.86727518e-02
2.50034899e-01 2.33451173e-01 6.45685256e-01 1.74717772e+00
2.95292497e-01 -4.11666073e-02 4.34082836e-01 1.32157907e-01
2.73083836e-01 -6.59009367e-02 -6.10111535e-01 2.79747218e-01
7.48808980e-02 1.42051113e+00 -1.06387722e+00 -6.38280511e-01
-7.60682106e-01 8.88566017e-01 -2.48261660e-01 -2.98626069e-03
-7.46350408e-01 -3.19283634e-01 7.87261426e-01 1.76895157e-01
8.62107456e-01 -5.40737629e-01 2.19156906e-01 -1.07307172e+00
1.68593332e-01 -6.53409719e-01 4.46565062e-01 -5.36482155e-01
-6.94834471e-01 4.26563293e-01 5.71206063e-02 -1.48483288e+00
-1.79532886e-01 -1.14761126e+00 -3.87293845e-01 1.95304111e-01
-1.65637672e+00 -1.01559246e+00 -5.74322045e-01 4.27936971e-01
9.37638104e-01 -3.38269025e-01 3.13027173e-01 7.66994834e-01
-7.24237204e-01 2.23457232e-01 1.11513361e-01 5.95293641e-02
6.76233411e-01 -6.92594290e-01 3.94334495e-01 1.20546877e+00
2.08960325e-01 8.31307992e-02 9.18972969e-01 -3.45918775e-01
-1.59418154e+00 -1.05317700e+00 5.58728397e-01 -5.11245489e-01
4.60639983e-01 -2.96358228e-01 -7.91377902e-01 3.14697921e-01
3.62460166e-02 5.02490342e-01 2.92351663e-01 -3.26968938e-01
-8.85809064e-02 -3.52961749e-01 -9.21280622e-01 1.25266701e-01
9.78859723e-01 -2.84596503e-01 -5.51225185e-01 4.90583777e-01
1.44435823e-01 -5.19703448e-01 -2.90734261e-01 4.36166495e-01
7.63493598e-01 -1.04537940e+00 1.26982784e+00 -2.90375769e-01
-3.39896321e-01 -8.53152096e-01 -2.49503516e-02 -6.00964487e-01
-3.91198665e-01 -5.54329693e-01 -3.04791093e-01 9.60061431e-01
-1.15342103e-01 -4.16781783e-01 9.45620716e-01 3.37650150e-01
-3.61130312e-02 1.30307496e-01 -1.08350217e+00 -1.12821722e+00
-1.02970123e+00 -7.05006659e-01 -6.22611679e-02 3.85739475e-01
-3.48134458e-01 1.10137247e-01 -4.01294023e-01 2.74386108e-01
8.56842220e-01 -1.37895688e-01 8.81546795e-01 -1.20721078e+00
-1.35495096e-01 -3.71043563e-01 -1.39338827e+00 -9.16164994e-01
-1.58574849e-01 -3.27797115e-01 9.16652679e-02 -1.46179998e+00
7.63344243e-02 7.31057376e-02 -3.32550973e-01 2.09889397e-01
-9.95632187e-02 1.02814972e+00 3.85243118e-01 8.92805681e-02
-1.16714430e+00 1.85788758e-02 5.58425188e-01 -1.44102380e-01
-1.55741170e-01 -8.76908973e-02 3.82535830e-02 9.32326853e-01
4.91246223e-01 -5.73935568e-01 1.07611537e-01 -1.80131093e-01
-2.18704771e-02 -3.64988655e-01 5.37732959e-01 -1.66933382e+00
6.47308588e-01 2.77870834e-01 6.12723827e-01 -9.61799860e-01
5.16055346e-01 -9.58966196e-01 1.08721137e-01 1.07172287e+00
2.06514150e-01 2.45413154e-01 5.24013042e-01 5.08836508e-01
-2.93013006e-01 -2.62414098e-01 9.20746565e-01 -1.26144692e-01
-1.61154509e+00 -1.71998665e-01 -8.21538389e-01 -3.19766074e-01
1.37596500e+00 -7.66297817e-01 -3.11958104e-01 3.74237564e-03
-5.02971709e-01 1.13269361e-02 1.02007166e-01 7.89626360e-01
7.10228324e-01 -1.07884109e+00 -8.42353344e-01 3.18800092e-01
2.88116843e-01 -3.10576767e-01 1.24194831e-01 1.05267549e+00
-8.48533571e-01 7.01176047e-01 -4.98389900e-01 -1.17918479e+00
-1.90553403e+00 7.08086848e-01 1.92827433e-02 1.69176832e-01
-7.75246978e-01 6.48370504e-01 1.00582659e-01 2.97052264e-01
2.79522687e-01 -4.87266630e-01 -3.42797935e-01 3.33207607e-01
1.00324345e+00 8.00710738e-01 4.99695450e-01 -8.47137690e-01
-8.89540315e-01 7.45736063e-01 -2.12062791e-01 2.27356091e-01
1.07088220e+00 -7.21081272e-02 3.43962610e-01 1.97020546e-01
9.07943964e-01 -4.17755425e-01 -1.37635314e+00 -7.23963156e-02
1.80319682e-01 -6.37996018e-01 3.12912643e-01 -1.06360331e-01
-9.39449131e-01 6.99778616e-01 1.18171334e+00 5.29657342e-02
9.85068500e-01 6.02565110e-02 3.92423302e-01 6.66441143e-01
2.82712460e-01 -1.25483608e+00 1.56978831e-01 4.04361963e-01
3.90132844e-01 -1.57904410e+00 3.91183853e-01 -4.52765316e-01
-2.51785755e-01 1.37162232e+00 4.67316389e-01 -9.64640602e-02
4.57152396e-01 4.86473173e-01 1.70396656e-01 3.47420722e-02
-5.65245509e-01 -8.13433111e-01 4.31623042e-01 7.63123155e-01
3.12124640e-01 -1.99755400e-01 -3.84644896e-01 -1.53143182e-01
4.12899882e-01 1.49007186e-01 4.77828860e-01 1.09950733e+00
-7.86402941e-01 -8.52962017e-01 -8.17090571e-01 3.99192363e-01
-6.61864340e-01 1.20194554e-01 1.59462243e-01 6.97236001e-01
4.06273335e-01 1.24519265e+00 4.37569141e-01 -1.17243886e-01
4.22818542e-01 -1.80061668e-01 5.47962308e-01 -4.87358838e-01
-5.87293684e-01 9.34235454e-02 5.30650169e-02 -8.75782549e-01
-9.24838603e-01 -9.18586314e-01 -9.10747170e-01 -3.50162424e-02
-4.47015673e-01 -3.77450287e-01 1.08595860e+00 9.43469822e-01
3.78602684e-01 6.00491822e-01 3.61817181e-02 -1.37336099e+00
-1.30763680e-01 -8.40960026e-01 -2.33560085e-01 4.82355744e-01
5.55031121e-01 -8.71047199e-01 7.93990213e-03 3.27137381e-01] | [8.5133056640625, -0.7111911177635193] |
4f2c547d-432d-46ed-8209-fcd337ad018a | interpretable-convolutional-filters-with | 1811.09725 | null | https://arxiv.org/abs/1811.09725v2 | https://arxiv.org/pdf/1811.09725v2.pdf | Interpretable Convolutional Filters with SincNet | Deep learning is currently playing a crucial role toward higher levels of artificial intelligence. This paradigm allows neural networks to learn complex and abstract representations, that are progressively obtained by combining simpler ones. Nevertheless, the internal "black-box" representations automatically discovered by current neural architectures often suffer from a lack of interpretability, making of primary interest the study of explainable machine learning techniques. This paper summarizes our recent efforts to develop a more interpretable neural model for directly processing speech from the raw waveform. In particular, we propose SincNet, a novel Convolutional Neural Network (CNN) that encourages the first layer to discover more meaningful filters by exploiting parametrized sinc functions. In contrast to standard CNNs, which learn all the elements of each filter, only low and high cutoff frequencies of band-pass filters are directly learned from data. This inductive bias offers a very compact way to derive a customized filter-bank front-end, that only depends on some parameters with a clear physical meaning. Our experiments, conducted on both speaker and speech recognition, show that the proposed architecture converges faster, performs better, and is more interpretable than standard CNNs. | ['Mirco Ravanelli', 'Yoshua Bengio'] | 2018-11-23 | null | null | null | null | ['distant-speech-recognition'] | ['speech'] | [ 1.37122259e-01 5.94086230e-01 -7.03663304e-02 -6.76716685e-01
-1.26855001e-01 -4.81064111e-01 5.62873125e-01 -7.08358884e-02
1.98457055e-02 5.70222914e-01 3.04738790e-01 -6.15267932e-01
-2.49832585e-01 -5.45435429e-01 -8.27049732e-01 -6.63539350e-01
-1.27253056e-01 4.50845063e-02 -1.53040364e-01 -3.88400614e-01
-2.85306592e-02 7.08111942e-01 -1.63155055e+00 4.95725065e-01
7.10642755e-01 1.12695312e+00 1.10260643e-01 5.38698792e-01
-7.78849870e-02 7.55517602e-01 -7.32838094e-01 -1.79560527e-01
-1.92706939e-02 -4.92113531e-01 -8.66416872e-01 -1.88580617e-01
1.92040578e-01 2.11336603e-03 -3.26834232e-01 1.04708719e+00
2.60163784e-01 8.43025446e-02 5.89121521e-01 -8.76908004e-01
-9.48918760e-01 1.21098089e+00 2.22220451e-01 3.71375978e-01
-5.30436449e-03 9.18700993e-02 1.17813587e+00 -7.34283566e-01
1.47366792e-01 1.35854948e+00 8.60493124e-01 7.53572345e-01
-1.32355654e+00 -5.56169987e-01 2.67370462e-01 3.53338420e-01
-1.17321789e+00 -4.28110808e-01 1.05789328e+00 -1.37218907e-01
1.12058246e+00 5.37524223e-01 7.03803957e-01 1.29712486e+00
-5.31110652e-02 5.96186042e-01 7.19829798e-01 -5.19208252e-01
2.88922846e-01 2.77001858e-01 2.78785646e-01 5.04059196e-01
1.05348028e-01 2.72308379e-01 -6.19334042e-01 1.58372410e-02
7.35068321e-01 -1.41145989e-01 -7.57770717e-01 -1.23420293e-02
-8.87806416e-01 8.91397119e-01 6.60212159e-01 7.27694392e-01
-3.94151419e-01 4.02879983e-01 1.74544290e-01 3.79467368e-01
3.09050918e-01 7.20990360e-01 -6.47968113e-01 -2.17734911e-02
-8.30042958e-01 -1.09339356e-01 9.05922055e-01 6.07467890e-01
6.49287462e-01 8.65164042e-01 2.24440351e-01 6.97948813e-01
2.58623272e-01 2.48035163e-01 6.45748615e-01 -6.70635760e-01
1.58444941e-01 4.12449092e-01 -3.21544558e-01 -9.82091725e-01
-6.36695683e-01 -1.17402196e+00 -1.17856872e+00 2.47499302e-01
2.92312145e-01 -1.64734185e-01 -7.61911154e-01 1.73319697e+00
-1.87438503e-01 1.27479598e-01 2.32374310e-01 1.05077553e+00
8.31032157e-01 9.53452408e-01 -9.23629776e-02 -4.24788222e-02
1.26000631e+00 -8.00524354e-01 -9.09024835e-01 -2.24846557e-01
2.72032887e-01 -4.19828624e-01 1.10855615e+00 7.75729001e-01
-1.09152210e+00 -8.90431345e-01 -1.39362729e+00 -1.88654304e-01
-5.71470380e-01 3.50676715e-01 7.72075772e-01 7.26859450e-01
-1.20516932e+00 9.44052875e-01 -5.45157433e-01 6.20878637e-02
3.00258279e-01 5.06739974e-01 -1.10869899e-01 7.34627008e-01
-1.32318008e+00 9.00509536e-01 5.55781662e-01 5.01447678e-01
-6.44027710e-01 -8.03444445e-01 -6.17109954e-01 6.80424392e-01
1.38420425e-02 -5.22954941e-01 1.36105502e+00 -1.45563281e+00
-1.85643315e+00 3.22116196e-01 -2.38510460e-01 -9.84873056e-01
-8.30831751e-02 -2.34618887e-01 -6.71619952e-01 1.03316031e-01
-7.68834174e-01 4.80082899e-01 1.27750564e+00 -1.17935920e+00
-2.29320064e-01 7.73801133e-02 1.41681135e-01 -5.54151475e-01
-6.73767567e-01 -2.34150723e-01 1.79696783e-01 -9.28501010e-01
2.69989491e-01 -5.85022926e-01 -1.22393779e-01 -3.32417758e-03
-4.28563058e-01 -3.35807472e-01 7.30735302e-01 -5.15332520e-01
1.24861133e+00 -2.13924670e+00 2.08899274e-01 2.90598601e-01
3.30198914e-01 6.52282834e-01 2.45042169e-03 2.49377385e-01
-6.82700694e-01 3.36048096e-01 -1.01555616e-01 -2.38449544e-01
4.39899936e-02 2.73349136e-01 -7.34629929e-01 1.23836182e-01
7.00486958e-01 6.59487665e-01 -6.91500485e-01 1.97471380e-01
1.59438595e-01 7.96900988e-01 -6.37441576e-01 1.40998781e-01
-2.94611633e-01 4.44230705e-01 -3.12323987e-01 1.61689758e-01
5.80531895e-01 -3.67003888e-01 2.46153072e-01 -4.56752598e-01
-2.08001673e-01 8.16384733e-01 -1.01094329e+00 1.36492050e+00
-6.80717111e-01 1.23153949e+00 -1.37758544e-02 -1.47773862e+00
1.29623401e+00 6.75428152e-01 4.75806817e-02 -5.20587265e-01
3.24809402e-01 3.00384849e-01 2.27824733e-01 -4.77029681e-01
6.69236854e-02 -1.58994928e-01 3.77544940e-01 3.92902315e-01
3.45114619e-01 2.90873554e-02 -2.53319860e-01 -2.45145857e-01
7.67475545e-01 -2.34567374e-01 3.35206538e-01 -5.08581936e-01
8.46279085e-01 -4.47689235e-01 4.91675347e-01 5.74521720e-01
2.97242552e-01 7.04850435e-01 4.26331460e-01 -1.07975185e+00
-1.02970839e+00 -1.11820948e+00 -2.56218910e-01 8.82335842e-01
-3.06391627e-01 -2.98657358e-01 -8.23146641e-01 -1.79053530e-01
-3.33956301e-01 8.54974926e-01 -5.35164595e-01 -3.67295533e-01
-7.97379076e-01 -5.32497227e-01 7.54881203e-01 6.40410423e-01
3.21551025e-01 -1.07267344e+00 -5.64398170e-01 4.33074355e-01
-6.65567145e-02 -9.71558452e-01 -8.22302774e-02 7.67152071e-01
-1.06319749e+00 -6.55144155e-01 -3.79559398e-01 -7.17529535e-01
5.11978567e-01 -5.29948622e-02 1.13582289e+00 3.01458001e-01
-8.13767985e-02 -1.60878196e-01 -3.30788046e-01 -7.54103899e-01
-5.65869033e-01 3.63250107e-01 1.97027892e-01 2.78097659e-01
2.14768723e-01 -1.03200233e+00 -3.12510312e-01 -9.78985503e-02
-9.30717409e-01 1.37863576e-01 6.63211524e-01 7.98678339e-01
1.60722211e-01 2.43722852e-02 8.64012957e-01 -5.67448497e-01
7.00284421e-01 -2.43699580e-01 -4.34691101e-01 2.10811850e-02
-3.71825576e-01 6.53642595e-01 1.26362169e+00 -5.90594232e-01
-1.09275019e+00 5.04845791e-02 -4.74661857e-01 -2.28557259e-01
-2.99930066e-01 4.25417483e-01 -1.33289620e-01 8.32024217e-02
8.90021145e-01 2.98560619e-01 -7.34988824e-02 -8.83900523e-01
4.45814520e-01 6.31718934e-01 7.37047493e-01 -3.79181117e-01
9.84722614e-01 3.91481370e-01 -2.40563333e-01 -1.15785456e+00
-7.19976246e-01 3.11121922e-02 -5.28864086e-01 -1.55048564e-01
6.61933184e-01 -5.93922734e-01 -9.77943659e-01 1.01810575e-01
-1.68328035e+00 8.53787549e-03 -4.76570100e-01 6.32662773e-01
-2.90014118e-01 -1.89283323e-02 -5.10206878e-01 -8.57813835e-01
-4.65818226e-01 -9.67865109e-01 6.38312817e-01 3.15702975e-01
-4.73042488e-01 -1.09808302e+00 -2.20160648e-01 -5.71322665e-02
6.55526996e-01 -4.20241058e-02 1.22432554e+00 -7.10880756e-01
-5.72544992e-01 1.29931226e-01 -2.94938311e-02 8.07409823e-01
4.37993109e-02 5.92050552e-02 -1.65000749e+00 -1.17965264e-03
3.61304343e-01 3.00493687e-02 9.19804037e-01 2.85401523e-01
1.64595807e+00 -6.58770084e-01 -5.44040115e-04 8.99780869e-01
1.08657289e+00 4.14306432e-01 6.71758473e-01 1.18332557e-01
3.25048685e-01 4.86478120e-01 -3.91037107e-01 3.04552317e-01
-2.15177715e-01 4.62909937e-01 5.69696724e-01 -3.93079072e-01
-1.55410632e-01 -8.85373130e-02 3.82436872e-01 1.15614438e+00
-4.00565445e-01 -2.88935840e-01 -6.58828735e-01 3.07676405e-01
-1.49528873e+00 -9.49238896e-01 -2.08061919e-01 1.81202328e+00
7.49895453e-01 3.85250002e-01 -2.65115649e-01 6.60520732e-01
4.41359937e-01 1.67292267e-01 -3.99406940e-01 -8.88545334e-01
-2.70941675e-01 8.76968384e-01 7.71169662e-02 4.93277043e-01
-8.02390218e-01 5.91607869e-01 6.57545710e+00 6.66083515e-01
-1.66078651e+00 -1.76833138e-01 4.70838964e-01 6.57759681e-02
-6.20171309e-01 -2.43093923e-01 -5.88415861e-01 1.25899076e-01
1.29164982e+00 -2.68932711e-02 5.75121820e-01 9.39418077e-01
4.89103764e-01 6.62669659e-01 -1.36919343e+00 8.78029585e-01
-2.96725452e-01 -1.81935382e+00 3.05517286e-01 -2.16203645e-01
3.17116976e-01 -2.72757500e-01 3.20415854e-01 1.70421213e-01
-1.63481846e-01 -1.33603942e+00 1.10101533e+00 6.01638854e-01
3.16616178e-01 -7.27631927e-01 5.40443599e-01 2.74064898e-01
-1.09228182e+00 -2.65913159e-01 -5.27268589e-01 -4.24918562e-01
-1.82039380e-01 7.08721459e-01 -1.00798190e+00 3.92906994e-01
7.65477955e-01 5.10785997e-01 -2.72054523e-01 7.50925481e-01
-4.71421868e-01 1.02512753e+00 -2.27535486e-01 -4.08447266e-01
3.52506310e-01 1.21353202e-01 4.96608377e-01 1.36961579e+00
2.79061347e-01 2.38882527e-02 -6.02137506e-01 1.38749230e+00
-2.05487818e-01 -2.57147700e-01 -4.42232490e-01 -1.70295909e-01
3.31384629e-01 1.22349644e+00 -4.65433687e-01 -2.32371822e-01
-2.56780595e-01 4.66891944e-01 2.01686099e-01 5.48579216e-01
-9.25688267e-01 -4.88744259e-01 8.62446666e-01 1.01067242e-03
4.19957310e-01 -2.91846037e-01 -4.67706621e-01 -1.01501346e+00
8.01325887e-02 -9.21308637e-01 -1.81179613e-01 -7.88518250e-01
-1.12610602e+00 1.10172129e+00 -2.26243034e-01 -1.19021344e+00
-6.15803182e-01 -9.04632926e-01 -8.82649958e-01 9.79639888e-01
-1.56161225e+00 -8.15518737e-01 -1.50977895e-01 4.41261947e-01
6.93707466e-01 -1.91790909e-01 9.95005846e-01 1.54908657e-01
-3.26597750e-01 5.41859567e-01 -3.70917842e-02 1.37217432e-01
6.24714978e-02 -1.10622001e+00 6.41426265e-01 7.28456676e-01
5.05311966e-01 9.05851841e-01 9.86526668e-01 1.74801007e-01
-1.30348897e+00 -8.68171215e-01 9.79984045e-01 -7.73416040e-03
6.19197071e-01 -5.47345638e-01 -1.27177906e+00 5.36518455e-01
5.34066141e-01 -2.16761574e-01 5.71698546e-01 2.32423767e-01
-5.03012300e-01 -4.66614544e-01 -6.62191510e-01 4.71126735e-01
8.63828063e-01 -6.60866261e-01 -8.91160846e-01 -1.76004618e-02
9.41424310e-01 -1.11249991e-01 -5.04026949e-01 1.86223388e-01
4.93067831e-01 -1.13584542e+00 1.12909997e+00 -8.06226373e-01
3.37466776e-01 -3.35018277e-01 2.13783875e-01 -1.53429902e+00
-3.59661341e-01 -9.51332927e-01 -2.28320509e-01 9.24468517e-01
8.19695354e-01 -7.52543449e-01 6.23567998e-01 1.08328097e-01
-6.76290512e-01 -8.00909936e-01 -9.39821243e-01 -8.81315947e-01
4.39906865e-02 -8.30285490e-01 8.99396658e-01 8.14733267e-01
1.60478596e-02 4.67104226e-01 -3.03854674e-01 5.00033081e-01
3.24232459e-01 6.93950355e-02 4.03961092e-01 -1.48189402e+00
-4.35762942e-01 -6.39323473e-01 -3.37627083e-01 -1.33701885e+00
2.43737623e-01 -8.30497980e-01 1.06681228e-01 -1.09342647e+00
-4.86258507e-01 -1.47563815e-01 -2.56379932e-01 4.39213365e-01
3.80292475e-01 6.10871846e-03 -4.20685718e-03 -1.54955626e-01
1.33330837e-01 5.61149240e-01 1.11768985e+00 -3.55660170e-01
-1.22707352e-01 1.80892840e-01 -7.28022039e-01 1.21470225e+00
1.10226524e+00 -3.26400757e-01 -4.84290242e-01 -6.76647305e-01
2.55214006e-01 -2.13502496e-01 5.93522310e-01 -1.29787290e+00
3.08665603e-01 1.31816566e-01 4.47982848e-01 -4.86269414e-01
5.09807825e-01 -8.26829135e-01 2.23441333e-01 5.77856719e-01
-6.02736115e-01 -1.90282807e-01 2.52756983e-01 2.48818144e-01
-4.74614263e-01 -4.27478224e-01 7.78328121e-01 5.07439673e-02
-5.36651850e-01 -5.02819344e-02 -7.06526875e-01 -2.60024697e-01
1.98035106e-01 -2.52659887e-01 -1.46693826e-01 -5.62679470e-01
-9.84083533e-01 -4.87632424e-01 -5.17360747e-01 5.44170439e-01
7.29670942e-01 -1.21735525e+00 -5.41615427e-01 4.91822064e-01
-3.49881858e-01 -1.27395198e-01 6.11387677e-02 4.56469268e-01
-3.77979338e-01 8.14322293e-01 -1.21145733e-01 -5.38906157e-01
-8.19233358e-01 4.44268405e-01 6.46452546e-01 1.31015942e-01
-7.11607456e-01 9.00907040e-01 2.76602149e-01 -1.67837322e-01
4.62827116e-01 -1.14495826e+00 -2.93028951e-01 -1.84243754e-01
6.98463917e-01 1.55754730e-01 2.33035743e-01 -2.68153071e-01
-1.67769462e-01 3.64527583e-01 2.33447000e-01 1.66945785e-01
1.66997492e+00 1.91042185e-01 -7.77465627e-02 4.23629582e-01
1.17294312e+00 -3.17053348e-01 -1.02769017e+00 -1.00759432e-01
1.18214414e-01 2.67653763e-02 1.26095429e-01 -6.07290804e-01
-1.29508638e+00 1.42801380e+00 4.26136702e-01 8.15807045e-01
1.36932456e+00 -6.70018718e-02 7.29003906e-01 7.37584770e-01
1.06773496e-01 -9.96837199e-01 1.52943581e-01 6.49964213e-01
1.30793428e+00 -7.34802902e-01 -3.94451112e-01 -4.11558479e-01
-2.43494986e-03 1.95322907e+00 3.33321661e-01 -8.88294503e-02
7.85055816e-01 3.92008692e-01 -3.29704024e-03 -6.55972734e-02
-8.26250136e-01 -2.19909288e-02 4.57134515e-01 6.42250538e-01
5.59442937e-01 -9.86079648e-02 -5.79616949e-02 1.06249380e+00
-6.52880788e-01 -2.37195417e-01 4.89070892e-01 3.59908700e-01
-7.32546508e-01 -1.00024045e+00 -5.03638446e-01 -1.99705106e-03
-3.67205739e-01 -2.37585321e-01 -3.63505930e-01 7.71886408e-01
3.14335912e-01 8.86693299e-01 1.80131078e-01 -5.22999167e-01
3.66455555e-01 1.08983606e-01 1.64223179e-01 -3.81712079e-01
-7.73289084e-01 -1.41106695e-01 1.98836923e-02 -3.35259438e-01
-3.28252256e-01 -2.19995916e-01 -1.49534810e+00 -2.76540071e-01
-2.67883718e-01 2.96150267e-01 8.30126941e-01 1.02785420e+00
2.08150372e-01 1.00903690e+00 6.47031844e-01 -9.36521590e-01
-6.56224310e-01 -9.16982472e-01 -3.83494198e-01 8.93734992e-02
8.24957252e-01 -4.71423596e-01 -5.91655076e-01 4.53017414e-01] | [15.295611381530762, 5.463650703430176] |
40b46a3e-698b-4949-9880-495e3ace347c | iwa-integrated-gradient-based-white-box | 2102.02128 | null | https://arxiv.org/abs/2102.02128v1 | https://arxiv.org/pdf/2102.02128v1.pdf | IWA: Integrated Gradient based White-box Attacks for Fooling Deep Neural Networks | The widespread application of deep neural network (DNN) techniques is being challenged by adversarial examples, the legitimate input added with imperceptible and well-designed perturbations that can fool DNNs easily in the DNN testing/deploying stage. Previous adversarial example generation algorithms for adversarial white-box attacks used Jacobian gradient information to add perturbations. This information is too imprecise and inexplicit, which will cause unnecessary perturbations when generating adversarial examples. This paper aims to address this issue. We first propose to apply a more informative and distilled gradient information, namely integrated gradient, to generate adversarial examples. To further make the perturbations more imperceptible, we propose to employ the restriction combination of $L_0$ and $L_1/L_2$ secondly, which can restrict the total perturbations and perturbation points simultaneously. Meanwhile, to address the non-differentiable problem of $L_1$, we explore a proximal operation of $L_1$ thirdly. Based on these three works, we propose two Integrated gradient based White-box Adversarial example generation algorithms (IWA): IFPA and IUA. IFPA is suitable for situations where there are a determined number of points to be perturbed. IUA is suitable for situations where no perturbation point number is preset in order to obtain more adversarial examples. We verify the effectiveness of the proposed algorithms on both structured and unstructured datasets, and we compare them with five baseline generation algorithms. The results show that our proposed algorithms do craft adversarial examples with more imperceptible perturbations and satisfactory crafting rate. $L_2$ restriction is more suitable for unstructured dataset and $L_1$ restriction performs better in structured dataset. | ['Vojislav B. Mišić', 'Jelena Mišić', 'Xiaolin Chang', 'Jiqiang Liu', 'Yixiang Wang'] | 2021-02-03 | null | null | null | null | ['dnn-testing'] | ['adversarial'] | [ 2.87527919e-01 9.29288790e-02 1.56506598e-01 2.86666155e-02
-4.47872519e-01 -9.86002445e-01 5.89696646e-01 -5.18018723e-01
-5.31758010e-01 1.02778625e+00 -1.15061566e-01 -5.53309023e-01
-1.71131194e-01 -1.07691455e+00 -8.94716322e-01 -7.75295973e-01
-1.02123119e-01 5.58951013e-02 1.76793233e-01 -5.44034660e-01
2.12151967e-02 6.62044287e-01 -1.13872945e+00 -6.85109571e-02
1.07958961e+00 7.18220174e-01 -1.14385620e-01 5.55099964e-01
-1.20880589e-01 7.26874828e-01 -1.11719143e+00 -5.49647927e-01
8.79487872e-01 -5.08922219e-01 -4.51610625e-01 -3.94877613e-01
2.65681356e-01 -5.04659176e-01 -4.60335374e-01 1.34766603e+00
8.43175650e-01 9.96346325e-02 5.00361085e-01 -1.58734274e+00
-5.01556635e-01 6.96083903e-01 -3.85702103e-01 1.50570095e-01
3.19513939e-02 6.17839456e-01 3.99409294e-01 -4.11223650e-01
4.35213119e-01 1.33996272e+00 6.16040587e-01 1.07064497e+00
-9.10448849e-01 -1.20266509e+00 3.91617894e-01 8.74127597e-02
-1.24209654e+00 -2.04170212e-01 1.29039359e+00 -2.71064162e-01
4.00157452e-01 5.68379343e-01 3.34430695e-01 1.31638825e+00
-5.29971793e-02 5.47244489e-01 1.08629572e+00 -1.76291332e-01
2.88507819e-01 2.28572458e-01 -3.56883138e-01 3.43334168e-01
4.12294388e-01 4.86283094e-01 2.49892399e-01 -1.57350197e-01
9.18354094e-01 1.03318565e-01 -5.11420190e-01 8.45678002e-02
-1.02095473e+00 8.66469562e-01 5.77147901e-01 7.22064599e-02
-1.83644697e-01 2.19167709e-01 5.17982900e-01 5.47121167e-01
8.57017264e-02 6.82883024e-01 -3.07512224e-01 -7.14957416e-02
-5.78042448e-01 4.50740904e-01 5.37591100e-01 1.13243103e+00
7.12898135e-01 6.18867755e-01 -1.57882109e-01 6.69853210e-01
1.99569941e-01 6.82427764e-01 4.66023892e-01 -6.04965448e-01
8.60051692e-01 5.30080616e-01 6.52802810e-02 -1.05815792e+00
-1.63384542e-01 -2.75874048e-01 -1.02779841e+00 7.47614503e-01
4.64041322e-01 -7.57071853e-01 -1.12200272e+00 1.97299945e+00
2.98544288e-01 2.73254573e-01 3.62430245e-01 8.86360586e-01
7.71362245e-01 7.50302613e-01 -1.42226547e-01 -1.60111442e-01
1.01664650e+00 -6.83255315e-01 -5.53587377e-01 -2.34966725e-01
3.66550148e-01 -7.86231279e-01 1.21448338e+00 1.89152524e-01
-7.36350179e-01 -5.76828480e-01 -1.17206824e+00 5.18330872e-01
-4.97896940e-01 -9.47535783e-02 3.66034776e-01 1.01031315e+00
-6.60955369e-01 4.82993692e-01 -4.74918455e-01 2.00776935e-01
5.90096831e-01 4.68233764e-01 -4.20570701e-01 -2.66166348e-02
-1.70823741e+00 6.85507894e-01 5.16665101e-01 3.54965240e-01
-1.05198216e+00 -8.47493470e-01 -7.31282830e-01 -3.38068426e-01
3.36346447e-01 -3.00660193e-01 6.59336627e-01 -9.55209196e-01
-1.44427717e+00 3.67442131e-01 3.68165702e-01 -4.68632698e-01
1.08139932e+00 -3.61318663e-02 -5.80175757e-01 -1.12984404e-02
-2.93846816e-01 5.49465537e-01 9.01942074e-01 -1.45033240e+00
-1.62269920e-01 -2.16450170e-01 3.84772778e-01 1.29711479e-01
-3.82643551e-01 -3.91916148e-02 8.64843875e-02 -1.07967639e+00
-2.04924524e-01 -8.92717183e-01 -4.82048064e-01 9.16550159e-02
-7.61283159e-01 1.65921912e-01 1.28884482e+00 -4.78229344e-01
1.05100608e+00 -2.01688075e+00 -2.35122234e-01 4.42214429e-01
1.78490043e-01 8.79407704e-01 -1.58150300e-01 3.30317080e-01
-2.31059104e-01 4.98976916e-01 -3.96443874e-01 1.99703187e-01
6.70435354e-02 2.72701681e-01 -5.61481059e-01 1.66143373e-01
2.72094697e-01 6.77718341e-01 -8.27594876e-01 -2.31520414e-01
1.89830482e-01 4.33793962e-01 -6.23274922e-01 2.81879842e-01
-2.03232586e-01 3.96833390e-01 -5.64264774e-01 5.06993175e-01
1.08933675e+00 5.27379036e-01 -4.61447150e-01 -3.00174147e-01
1.89669535e-01 -2.35739529e-01 -1.52049530e+00 1.03389144e+00
-2.72292346e-01 4.92459625e-01 -2.87904982e-02 -8.20648789e-01
1.28367865e+00 3.53025228e-01 -1.29548693e-02 -1.76170185e-01
2.52173811e-01 1.01243041e-01 2.31773153e-01 -4.45733279e-01
-4.99140322e-02 -2.67765969e-01 -9.79934540e-03 2.07126424e-01
-3.45249176e-01 -1.70791522e-01 -4.29850444e-02 1.00794345e-01
1.25006092e+00 -2.55310368e-02 -5.26190400e-02 -2.18805060e-01
8.30741167e-01 -2.61133552e-01 8.41450810e-01 7.58689821e-01
-4.00424004e-01 6.23066306e-01 7.47512162e-01 -4.90247637e-01
-1.06379557e+00 -1.08449888e+00 9.01545882e-02 4.95410472e-01
2.54529178e-01 1.71200439e-01 -7.59512722e-01 -1.04194260e+00
-1.75525740e-01 7.12536454e-01 -5.72220922e-01 -4.90720809e-01
-7.88932562e-01 -7.16462314e-01 1.04683411e+00 4.98257935e-01
1.10438061e+00 -1.37060499e+00 -1.61863640e-01 1.45622762e-02
1.23721898e-01 -8.00175428e-01 -5.44803441e-01 -2.93281395e-03
-6.29523516e-01 -9.69761491e-01 -7.05135822e-01 -6.09907866e-01
1.08520758e+00 -3.52902621e-01 5.53835988e-01 1.00208754e-02
-2.76186019e-01 -2.24346548e-01 -4.70261544e-01 -6.32817686e-01
-6.55790567e-01 -2.21385077e-01 1.77528739e-01 -3.48056778e-02
8.72800797e-02 -7.51224577e-01 -7.48587549e-01 4.41306680e-01
-1.27206123e+00 -3.54667962e-01 5.58887601e-01 8.60234559e-01
2.51201361e-01 1.95067346e-01 7.89531589e-01 -1.06583047e+00
7.72270560e-01 -2.93596566e-01 -6.50361001e-01 7.61770904e-02
-4.11828786e-01 1.13363959e-01 1.56039953e+00 -9.95180905e-01
-7.26114213e-01 -1.87779501e-01 -5.01628160e-01 -8.45533550e-01
-2.69828320e-01 1.58628263e-02 -7.75920331e-01 -4.67793643e-01
1.01823139e+00 2.73224175e-01 -6.90199509e-02 -2.39026457e-01
3.52765501e-01 3.71248543e-01 4.97666240e-01 -6.11541986e-01
1.53936398e+00 1.57280669e-01 -5.12148589e-02 -3.43554765e-01
-1.63412206e-02 3.55630785e-01 -1.43638685e-01 -1.67353526e-01
4.58549708e-01 -4.42893595e-01 -7.05435276e-01 6.77618444e-01
-9.05496538e-01 -2.42616653e-01 -2.73888022e-01 3.46475273e-01
-2.39774093e-01 4.51547980e-01 -3.24890435e-01 -8.24887335e-01
-5.86209953e-01 -1.53860867e+00 4.13087279e-01 2.33870819e-01
1.60699323e-01 -8.93637836e-01 -1.29314035e-01 2.83041839e-02
5.60787857e-01 9.06192482e-01 7.62359023e-01 -9.72730398e-01
-4.88923013e-01 -4.26165670e-01 -7.05970265e-03 6.68945074e-01
1.92022368e-01 4.42455783e-02 -8.42562914e-01 -4.35613841e-01
1.37536481e-01 -1.77522793e-01 4.82288033e-01 -1.75384447e-01
1.19634163e+00 -1.02800500e+00 -6.43163472e-02 8.88568938e-01
1.38280785e+00 6.85515821e-01 1.01893604e+00 5.90535641e-01
9.16903555e-01 2.38366082e-01 4.17272776e-01 3.17086428e-01
-2.30102539e-01 2.49548778e-01 7.78627157e-01 -1.71370015e-01
1.23883188e-01 -3.27455610e-01 3.13021392e-01 3.64839435e-01
-3.29929753e-03 -3.23658347e-01 -6.62839472e-01 3.47847015e-01
-1.32302225e+00 -1.06434524e+00 2.60471731e-01 2.18756580e+00
1.06099117e+00 5.47835350e-01 8.28063674e-03 5.00608563e-01
9.49654400e-01 3.53064686e-01 -7.55061030e-01 -5.16396880e-01
1.62932626e-03 3.39991689e-01 7.24065900e-01 4.37405527e-01
-9.38974380e-01 8.32480609e-01 5.51675892e+00 1.12794638e+00
-1.19197059e+00 -2.85243511e-01 6.58926070e-01 -7.71275535e-02
-6.93040311e-01 5.45056872e-02 -7.18809843e-01 9.72258925e-01
3.64827067e-01 -2.57926375e-01 4.83444601e-01 9.75453794e-01
1.36635557e-01 5.49762130e-01 -8.85761321e-01 7.17841744e-01
-2.25486860e-01 -1.18583262e+00 4.49571073e-01 -1.66901827e-01
8.11219394e-01 -3.76931459e-01 2.53776044e-01 4.09121037e-01
4.94056106e-01 -1.07429028e+00 4.25964266e-01 3.19915295e-01
7.84010708e-01 -1.10490727e+00 8.91198635e-01 4.33319211e-01
-9.65778410e-01 -1.29343033e-01 -5.62020540e-01 9.87752974e-02
-1.17366323e-02 3.85088533e-01 -7.05552876e-01 5.83232343e-01
4.25521582e-01 9.14483368e-02 -3.17598611e-01 8.22043240e-01
-4.49227512e-01 5.68641126e-01 -4.66406971e-01 -2.08925501e-01
3.51657003e-01 -1.95289984e-01 8.32414985e-01 9.09095764e-01
3.12766075e-01 1.41851321e-01 8.49361718e-03 9.69765306e-01
-2.56748289e-01 3.63181196e-02 -8.93020332e-01 1.51279300e-01
8.96685541e-01 1.08783102e+00 -3.23795855e-01 -4.66676690e-02
5.27308024e-02 8.12993348e-01 -1.23693138e-01 5.78212142e-01
-1.12684751e+00 -1.12366009e+00 7.75585771e-01 -1.13806967e-02
2.90670749e-02 5.53970337e-02 -1.20449178e-01 -9.35729444e-01
2.33919322e-01 -1.27393007e+00 2.64346361e-01 -5.74944437e-01
-1.31286013e+00 9.31664586e-01 1.63321812e-02 -1.49808443e+00
-1.95904076e-01 -6.39883578e-01 -1.24755907e+00 1.00981927e+00
-1.27907586e+00 -1.09274364e+00 -3.57786149e-01 9.61145043e-01
2.22399086e-01 -5.00340104e-01 6.52396560e-01 2.65308976e-01
-7.19671428e-01 1.36935687e+00 3.73948924e-02 5.32161951e-01
5.19955039e-01 -1.01095808e+00 4.34490114e-01 1.27544737e+00
-2.75857985e-01 7.99099147e-01 9.25239980e-01 -5.30730963e-01
-1.16870582e+00 -1.31181777e+00 1.66027665e-01 -2.07022578e-01
5.10534883e-01 -4.12442982e-01 -8.48488629e-01 6.48806691e-01
-1.76089197e-01 1.82375088e-01 3.61590475e-01 -6.39231324e-01
-2.89117992e-01 -5.72645783e-01 -1.85273063e+00 1.04380774e+00
8.41352999e-01 -3.31796736e-01 -4.23268765e-01 2.29471892e-01
1.07475388e+00 -4.46404308e-01 -8.63355935e-01 7.70792842e-01
1.57789022e-01 -7.86806524e-01 1.10480690e+00 -3.85367572e-01
4.62014496e-01 -5.88422894e-01 2.70122215e-02 -1.27347291e+00
7.90076852e-02 -1.09156942e+00 2.02095091e-01 1.42051077e+00
5.04150093e-01 -1.15575290e+00 8.64642620e-01 6.85444891e-01
-1.32871792e-01 -8.87091398e-01 -1.06722140e+00 -8.70380104e-01
3.79209608e-01 -3.34022224e-01 9.72038448e-01 1.01931453e+00
-2.06532612e-01 -2.79852241e-01 -5.25245845e-01 2.21580461e-01
5.47756493e-01 -4.62429404e-01 1.02685392e+00 -6.16694510e-01
-3.27720940e-01 -3.93801928e-01 -7.33197927e-01 -7.64746070e-01
-4.91003282e-02 -4.88624632e-01 3.34796589e-03 -1.12565637e+00
-4.15024638e-01 -7.44130433e-01 -2.52833366e-01 5.63524127e-01
-5.94592512e-01 2.93603152e-01 2.32777715e-01 -1.16642909e-02
1.35486349e-01 5.24967253e-01 1.36456728e+00 -1.73103228e-01
-2.82927096e-01 1.56501055e-01 -6.48959935e-01 6.38380051e-01
9.59947765e-01 -2.91321486e-01 -7.85376966e-01 -2.70402461e-01
1.98753141e-02 -2.08534509e-01 3.36668193e-01 -1.09368515e+00
3.29425856e-02 -2.09993154e-01 4.74872708e-01 -4.23716217e-01
1.42016247e-01 -9.64414120e-01 2.71004856e-01 5.73381305e-01
-2.12966785e-01 6.72373027e-02 4.22445536e-01 3.89131665e-01
-8.15989599e-02 -4.31474596e-01 8.93334687e-01 -1.35895967e-01
-5.01997471e-01 5.90656221e-01 -8.11370686e-02 2.27254406e-01
1.31071889e+00 -3.57792169e-01 -6.12582147e-01 -3.39506239e-01
-4.79585022e-01 2.57550389e-01 3.10431182e-01 3.65913659e-01
7.03611076e-01 -1.62502301e+00 -7.03954101e-01 5.20304382e-01
-2.35387549e-01 1.67424649e-01 2.65673310e-01 2.59265840e-01
-9.21776116e-01 -1.91516429e-01 -2.89773196e-01 -8.41618702e-02
-9.84814644e-01 9.19884205e-01 4.96630937e-01 -1.45902693e-01
-4.15638626e-01 1.06566274e+00 1.74477503e-01 -5.90404093e-01
2.99453765e-01 -1.00231059e-01 -1.56312078e-01 -4.63223487e-01
5.19743621e-01 4.56955612e-01 -3.61838639e-01 -3.59216094e-01
-2.65342623e-01 5.76537669e-01 -2.54620850e-01 -7.64170960e-02
9.14875090e-01 2.74209738e-01 -5.88982017e-04 -6.66695312e-02
1.20949960e+00 1.46407500e-01 -1.38889563e+00 8.33915398e-02
-5.64272821e-01 -5.38375378e-01 -4.61355090e-01 -6.34711623e-01
-1.24214470e+00 8.31025481e-01 5.95843136e-01 1.62878409e-01
1.35544360e+00 -5.48082590e-01 8.95912409e-01 3.11803043e-01
2.12859482e-01 -6.57996237e-01 1.16587793e-02 2.58538276e-01
9.22385752e-01 -1.01419294e+00 -1.11890286e-01 -2.99905807e-01
-6.02124393e-01 1.00366569e+00 1.15748906e+00 -3.29786003e-01
4.63495493e-01 4.11531299e-01 2.81411648e-01 1.97168544e-01
-3.71038556e-01 3.10396671e-01 1.76920015e-02 7.93876052e-01
-2.35324740e-01 -2.48170227e-01 -3.46086711e-01 5.59444129e-01
-5.52657068e-01 -2.96267360e-01 4.00145113e-01 8.29036832e-01
-2.94552058e-01 -1.13458622e+00 -6.78440630e-01 3.42430234e-01
-5.37395954e-01 -1.88641816e-01 -2.57996410e-01 1.02644980e+00
3.96938145e-01 8.41012955e-01 -2.66625136e-01 -7.18973160e-01
4.65355039e-01 -5.96227124e-02 1.38391182e-01 -4.86993305e-02
-7.21328497e-01 -2.94122934e-01 -2.30124518e-01 -2.27439016e-01
2.77633835e-02 -9.09163281e-02 -1.20870101e+00 -5.78435719e-01
-4.44056243e-01 1.66611761e-01 2.55506963e-01 7.65420556e-01
6.46322668e-02 5.61118543e-01 1.09239089e+00 -5.46103001e-01
-8.86832178e-01 -9.64652658e-01 -3.67764443e-01 5.45519412e-01
3.07850897e-01 -3.84029835e-01 -8.71519029e-01 -1.62934482e-01] | [5.534627914428711, 7.936845779418945] |
08d1a508-ba53-495a-bdf3-ebab3f0685f1 | a-comparison-of-deep-learning-architectures | 2111.04353 | null | https://arxiv.org/abs/2111.04353v1 | https://arxiv.org/pdf/2111.04353v1.pdf | A Comparison of Deep Learning Architectures for Optical Galaxy Morphology Classification | The classification of galaxy morphology plays a crucial role in understanding galaxy formation and evolution. Traditionally, this process is done manually. The emergence of deep learning techniques has given room for the automation of this process. As such, this paper offers a comparison of deep learning architectures to determine which is best suited for optical galaxy morphology classification. Adapting the model training method proposed by Walmsley et al in 2021, the Zoobot Python library is used to train models to predict Galaxy Zoo DECaLS decision tree responses, made by volunteers, using EfficientNet B0, DenseNet121 and ResNet50 as core model architectures. The predicted results are then used to generate accuracy metrics per decision tree question to determine architecture performance. DenseNet121 was found to produce the best results, in terms of accuracy, with a reasonable training time. In future, further testing with more deep learning architectures could prove beneficial. | ['Mattia Vaccari', 'Clement N. Nyirenda', 'Ezra Fielding'] | 2021-11-08 | null | null | null | null | ['morphology-classification'] | ['computer-vision'] | [-4.32505727e-01 -1.92331038e-02 4.98835891e-01 -2.58959800e-01
3.19868326e-02 -7.14795172e-01 9.03105736e-01 3.66948582e-02
-3.78520519e-01 4.38292354e-01 8.04063156e-02 -7.12535143e-01
-1.93349361e-01 -1.06129146e+00 -9.47794318e-02 -6.16232753e-01
8.34363848e-02 7.78052568e-01 5.27427614e-01 -1.60566103e-02
5.35758793e-01 6.36849403e-01 -1.45752418e+00 4.17368829e-01
4.37304020e-01 1.09768355e+00 4.06816453e-01 1.24232614e+00
1.03483148e-01 1.31940722e+00 -6.35850430e-01 -5.56942046e-01
3.14186424e-01 -2.39812881e-01 -1.31893754e+00 1.53835699e-01
3.26849908e-01 -6.66003209e-03 -5.07662147e-02 6.93763018e-01
6.19855762e-01 3.77907783e-01 4.82267678e-01 -6.27741873e-01
-2.60834754e-01 2.65751518e-02 1.36024341e-01 3.82577896e-01
-9.98694152e-02 5.11892319e-01 1.15661573e+00 -7.39814699e-01
4.43031698e-01 1.26969182e+00 8.91051590e-01 2.13215187e-01
-1.29575086e+00 -2.36463368e-01 -5.80244958e-01 3.79281163e-01
-1.12711537e+00 -2.79015243e-01 4.68682855e-01 -7.11054564e-01
1.45916855e+00 3.36032629e-01 1.03261268e+00 7.46571362e-01
2.40048110e-01 1.18916743e-01 1.28572309e+00 -3.70685339e-01
4.17171717e-01 3.24202657e-01 -1.35196075e-01 8.79931033e-01
2.35737294e-01 1.87691167e-01 1.38113908e-02 -1.14126049e-01
9.85402524e-01 -6.10626757e-01 2.38493338e-01 -6.06453754e-02
-6.49038553e-01 1.12948620e+00 8.05765510e-01 4.46278661e-01
-2.28907585e-01 -9.62932035e-03 3.50416124e-01 3.80167842e-01
6.54967308e-01 1.46503162e+00 -5.55229664e-01 -2.93479651e-01
-7.24803925e-01 5.07138610e-01 1.01797616e+00 3.01040739e-01
6.41008317e-01 1.80043519e-01 2.76943177e-01 1.21313643e+00
4.02588487e-01 -1.73798248e-01 7.55962014e-01 -1.32786632e+00
-3.49622160e-01 1.06590462e+00 -8.60478580e-02 -9.96996343e-01
-5.82096398e-01 -9.65326488e-01 -5.31449854e-01 6.13866091e-01
8.11441422e-01 -2.16383208e-02 -6.74287319e-01 9.60272253e-01
1.15508050e-01 -2.30173171e-01 -8.92347842e-02 7.74017155e-01
9.37647104e-01 4.18012142e-01 -1.40593007e-01 5.05857408e-01
1.13067734e+00 -1.03804326e+00 4.28309530e-01 -3.43701899e-01
7.11613655e-01 -1.04988754e+00 1.11453474e+00 6.41142726e-01
-9.19411778e-01 -8.09757173e-01 -9.34561729e-01 -1.79220676e-01
-4.52874929e-01 3.67464945e-02 9.23635304e-01 8.59544992e-01
-1.34455919e+00 7.54609406e-01 -8.07268620e-01 -8.44549000e-01
1.82848498e-01 6.60266817e-01 -1.56721428e-01 2.34469682e-01
-5.99236786e-01 9.41852272e-01 4.57308561e-01 -2.09972441e-01
-1.14198256e+00 -4.01767313e-01 -2.70973325e-01 3.45719792e-02
-1.03356205e-01 -8.84657204e-01 1.70071971e+00 -9.78458166e-01
-1.43685758e+00 1.22073221e+00 3.22836906e-01 -8.75036657e-01
3.17393690e-01 3.07833165e-01 6.40220642e-02 -5.50415739e-02
-1.63619205e-01 1.06319141e+00 4.49858338e-01 -7.75000453e-01
-8.33443224e-01 -3.49322148e-02 4.91137415e-01 -1.26606271e-01
-1.85653001e-01 2.97867835e-01 -1.22807980e-01 -2.67893255e-01
-1.91646814e-01 -1.05756140e+00 -2.28892788e-01 -4.93156612e-01
2.33214945e-02 -5.80083489e-01 4.96171802e-01 -5.97698569e-01
7.73351431e-01 -1.66926992e+00 -2.08998293e-01 4.55158670e-03
5.72651803e-01 3.81603897e-01 -5.29899485e-02 6.24222420e-02
1.02250382e-01 1.08336955e-01 1.82395101e-01 -1.99227020e-01
-2.65865892e-01 1.04578011e-01 2.22096387e-02 7.52814710e-02
1.07615851e-01 7.48915851e-01 -7.22664416e-01 -2.02458233e-01
4.13619071e-01 1.82390675e-01 -9.89553273e-01 6.21420920e-01
-3.11892748e-01 4.79615957e-01 9.54481482e-04 6.54433072e-01
2.63318032e-01 -5.11493504e-01 2.39204913e-02 3.28061394e-02
-3.12077940e-01 6.47808254e-01 -3.97671074e-01 1.16412485e+00
-6.77476227e-01 8.78629148e-01 4.19902578e-02 -8.10516179e-01
1.36367655e+00 1.13472797e-01 7.44557977e-02 -2.90629059e-01
4.51041311e-01 3.95831734e-01 9.07539666e-01 -4.29960161e-01
4.88319278e-01 -2.17187896e-01 2.17070729e-01 1.36759236e-01
3.17359060e-01 -4.91261482e-01 2.33924702e-01 -2.34489039e-01
1.38779449e+00 -6.88644648e-02 3.07401955e-01 -5.92850566e-01
5.26831150e-01 3.88700187e-01 2.01507062e-01 8.36948693e-01
-2.16933563e-01 7.79492259e-01 5.09809971e-01 -9.34038639e-01
-1.43130064e+00 -6.65697038e-01 5.51549233e-02 1.26657617e+00
-5.73794663e-01 -2.74287790e-01 -8.77352893e-01 -5.06515324e-01
-2.74500430e-01 8.32090735e-01 -6.27344251e-01 -4.35775891e-02
-2.32454911e-01 -9.44961786e-01 5.60753763e-01 2.75312513e-01
6.92838252e-01 -1.49960876e+00 -6.72803938e-01 2.13456348e-01
2.60154843e-01 -6.57028019e-01 2.51206189e-01 2.62663811e-01
-8.36422265e-01 -1.16368687e+00 -3.84110272e-01 -8.69472682e-01
3.05816382e-01 1.85276568e-01 1.28617370e+00 4.66888249e-01
-4.17908221e-01 5.60371764e-02 -3.49960536e-01 -3.94488066e-01
-7.06034124e-01 6.35088384e-01 -2.38845438e-01 -2.57561445e-01
4.49107170e-01 -7.14370549e-01 -7.37727523e-01 3.46619129e-01
-5.05402625e-01 1.60937145e-01 6.43996954e-01 7.96634257e-01
3.73142515e-03 1.07180439e-01 3.60570043e-01 -7.11288214e-01
7.17120647e-01 -5.40956974e-01 -7.66779542e-01 -7.32731074e-02
-7.93965518e-01 -1.58398852e-01 8.00693631e-01 -1.75100535e-01
-9.65930641e-01 -1.83883920e-01 -7.30720222e-01 -6.75145909e-02
-5.38833380e-01 3.94205958e-01 1.86079398e-01 -5.85461140e-01
9.91508782e-01 -2.14739874e-01 4.68147844e-02 -8.56136739e-01
-1.90660015e-01 6.84463739e-01 4.52186555e-01 -4.32998985e-01
4.13148582e-01 1.38287783e-01 3.26006510e-03 -8.55521083e-01
-8.92790377e-01 -5.47223926e-01 -6.23598337e-01 -2.96898186e-01
7.86620736e-01 -5.00408113e-01 -7.58852124e-01 5.71571290e-01
-9.88567531e-01 -7.19067454e-01 -1.22814037e-01 3.20196986e-01
-4.45759296e-01 -1.08300447e-01 -5.23449838e-01 -4.93181139e-01
-3.95155936e-01 -1.12398708e+00 6.11386716e-01 4.01996166e-01
-5.11904299e-01 -1.48560274e+00 3.07279974e-01 5.68126500e-01
7.24022806e-01 -1.11552045e-01 1.08331704e+00 -1.06746376e+00
-4.45850939e-01 -3.62539440e-01 -2.76377708e-01 4.56744075e-01
-2.02805623e-02 -1.06193461e-02 -1.18323231e+00 -3.00860345e-01
2.35962048e-01 -5.85802317e-01 6.82952642e-01 2.15503737e-01
1.35843849e+00 -2.17253104e-01 2.28215590e-01 7.58308709e-01
1.28711665e+00 2.29577556e-01 7.29605913e-01 8.95891428e-01
4.98267502e-01 7.33450711e-01 2.25607947e-01 2.75919795e-01
2.62417793e-01 4.22202647e-01 7.24555671e-01 5.52933710e-03
-3.89313459e-01 -9.58294608e-03 3.78738530e-02 7.29426742e-01
-3.92441303e-01 -3.10656905e-01 -1.25874567e+00 4.24082547e-01
-1.37507498e+00 -7.35868633e-01 -2.17643395e-01 1.95612514e+00
2.40145758e-01 3.47201079e-01 4.69431102e-01 3.21771577e-02
1.96460262e-01 -7.96320364e-02 -2.71699667e-01 -6.85757875e-01
-8.95755887e-02 4.46028948e-01 3.55291605e-01 4.96509314e-01
-9.20228601e-01 1.05592418e+00 6.59217596e+00 6.21200204e-01
-1.31890166e+00 1.70347646e-01 9.84240294e-01 7.97519553e-03
1.33872628e-01 2.79787660e-01 -6.27641261e-01 3.11299652e-01
1.25335646e+00 1.54718116e-01 6.80822790e-01 8.13393116e-01
3.21696281e-01 -3.60471845e-01 -7.33100235e-01 5.98402321e-01
-1.60724789e-01 -1.47653997e+00 -2.41471857e-01 3.58640194e-01
5.89614332e-01 3.73239189e-01 -1.81907400e-01 4.94345546e-01
7.13765562e-01 -1.22913229e+00 3.58066618e-01 4.37819660e-01
2.38910019e-01 -6.69165134e-01 9.40295696e-01 3.60765070e-01
-8.31289649e-01 -2.55562931e-01 -9.56056535e-01 -6.41185522e-01
-4.82906759e-01 2.39458740e-01 -1.87750769e+00 1.70688137e-01
7.17781186e-01 3.34742844e-01 -1.34958804e+00 1.63334250e+00
-7.73246810e-02 1.23467553e+00 -2.12819815e-01 -3.95549327e-01
4.81537789e-01 -1.72517881e-01 4.01409030e-01 1.24751031e+00
3.55888188e-01 -4.31439012e-01 9.67259109e-02 8.41525257e-01
1.08105704e-01 8.55628401e-02 -4.86452788e-01 -9.75166634e-02
2.67004818e-01 1.76635981e+00 -9.65868235e-01 -1.76720127e-01
-4.68161464e-01 3.79880250e-01 5.24680555e-01 -1.03896797e-01
-3.39612514e-01 -2.84141660e-01 5.35844326e-01 5.57583332e-01
1.17583513e-01 -1.98420346e-01 -4.98923153e-01 -4.73755479e-01
-8.89633954e-01 -8.91343176e-01 2.88785547e-01 -1.02845514e+00
-1.24158061e+00 6.45058215e-01 -4.05727863e-01 -8.49082410e-01
-4.33181167e-01 -1.11625218e+00 -1.04544008e+00 8.30760419e-01
-5.08408785e-01 -1.29545963e+00 -6.59121633e-01 -1.10553347e-01
6.85317636e-01 -7.96444416e-01 8.15231740e-01 9.72638130e-02
-3.71379584e-01 1.57222942e-01 3.38326991e-02 1.27214685e-01
4.29074764e-01 -1.68738258e+00 7.08782315e-01 4.74910289e-01
4.82044332e-02 2.36045927e-01 8.74190629e-01 -6.03806257e-01
-8.41753662e-01 -1.27994311e+00 7.45095849e-01 -3.74578893e-01
8.75232637e-01 -2.72944689e-01 -8.25468600e-01 4.46376175e-01
4.72300708e-01 -2.89913565e-01 5.80705464e-01 1.61586210e-01
8.21569562e-02 -1.67654268e-02 -1.08135557e+00 4.53848124e-01
7.73568928e-01 -4.27253604e-01 -4.99695003e-01 2.75770217e-01
3.59713554e-01 -7.39027783e-02 -9.91520226e-01 3.13411266e-01
4.71013635e-01 -1.40851319e+00 7.86463916e-01 -7.12066233e-01
4.30893213e-01 -1.55912861e-01 -4.39214297e-02 -1.40989196e+00
-8.03801358e-01 -4.12817419e-01 2.19044685e-01 1.06940186e+00
4.64099020e-01 -7.18792975e-01 1.24395347e+00 2.21130416e-01
-4.14402366e-01 -7.91293681e-01 -7.42403686e-01 -6.45764470e-01
3.48356545e-01 -2.65829384e-01 5.86881101e-01 5.85226834e-01
-4.76096511e-01 6.35146618e-01 -2.02970635e-02 -3.03446859e-01
1.68295741e-01 -4.08353582e-02 1.04317331e+00 -1.58110988e+00
-5.63360274e-01 -8.02498162e-01 -6.90518200e-01 -7.62200534e-01
-9.36479941e-02 -1.08824086e+00 -7.42081925e-02 -1.42491698e+00
1.70650631e-01 -4.69580501e-01 -3.32341045e-02 1.33191794e-01
1.01403601e-01 5.66990256e-01 -1.69607550e-02 1.71915069e-01
-4.80068177e-01 7.67672360e-02 1.12796032e+00 1.95313826e-01
5.81542291e-02 3.20685089e-01 -7.05550611e-01 1.03402603e+00
1.21156263e+00 -1.44763663e-01 -3.27222914e-01 -3.06249857e-01
9.38714817e-02 -5.08320391e-01 5.57873785e-01 -1.50652730e+00
1.53187186e-01 3.19531113e-02 6.38353467e-01 -3.23736817e-01
4.83684808e-01 -1.89813793e-01 1.41851977e-01 5.17725527e-01
-2.51102746e-01 2.19743967e-01 1.32744983e-01 -8.21440071e-02
-7.26411045e-02 -8.11033964e-01 1.09594595e+00 -6.82542264e-01
-4.77403015e-01 8.33515972e-02 -7.30051279e-01 -6.95794541e-03
5.09223938e-01 -1.89316690e-01 -3.57585430e-01 -3.90783459e-01
-7.82601476e-01 -9.70947146e-02 8.12873542e-01 1.07416697e-01
2.28914872e-01 -9.01774824e-01 -5.50801396e-01 4.30363640e-02
-1.50027156e-01 -9.03555825e-02 -2.18150005e-01 8.29380572e-01
-1.15491307e+00 6.12085283e-01 -3.71940732e-01 -5.49450576e-01
-1.33587861e+00 1.73065349e-01 9.60871577e-01 -5.08579135e-01
-5.71012869e-02 1.26729834e+00 2.67249763e-01 -6.48781300e-01
-1.34240434e-01 -2.46319696e-02 -2.80941039e-01 -1.40569016e-01
2.45033696e-01 7.26008058e-01 5.56385040e-01 -4.26880151e-01
9.62955356e-02 -4.67023998e-02 -4.81332950e-02 -9.73313376e-02
1.27507508e+00 3.14678043e-01 -3.71868461e-01 3.30215573e-01
9.57617164e-01 -5.27189150e-02 -1.01785123e+00 2.01942995e-01
4.23984110e-01 -5.59260726e-01 6.58497512e-02 -9.82963800e-01
-6.55314326e-01 1.08584547e+00 7.52497137e-01 8.58943939e-01
7.99738705e-01 1.16077192e-01 4.01852727e-01 5.46737611e-01
2.31769919e-01 -8.41892540e-01 2.16485307e-01 9.06005025e-01
9.83873069e-01 -1.11192906e+00 -2.22056270e-01 1.34335801e-01
-1.47739008e-01 1.27499628e+00 1.12180293e+00 -4.09055829e-01
4.15113091e-01 -4.16035317e-02 -1.00555792e-01 -3.72082382e-01
-1.20285273e+00 -2.79175371e-01 2.56325722e-01 4.92824435e-01
9.29696858e-01 1.58751950e-01 4.36748043e-02 2.30879188e-01
-8.94126356e-01 -3.29339534e-01 4.89263684e-01 5.17401695e-01
-9.42308068e-01 -1.22554231e+00 -4.84007895e-01 9.96617317e-01
-3.85488033e-01 3.28241922e-02 -1.20170176e+00 6.93656087e-01
4.38777268e-01 1.00061488e+00 3.74961078e-01 -5.93528092e-01
9.51361656e-02 5.56370839e-02 6.54891014e-01 -8.74359727e-01
-1.23625171e+00 -1.94260292e-02 5.63570559e-01 8.01169500e-02
1.35207340e-01 -6.28768146e-01 -7.38507092e-01 -6.33107722e-01
-2.21746668e-01 2.91516721e-01 4.65354025e-01 8.17334592e-01
2.10284337e-01 4.49585289e-01 5.75245321e-01 -1.02664030e+00
-5.24360359e-01 -1.60866952e+00 -2.39724576e-01 1.91757649e-01
-3.10947895e-01 -4.89340663e-01 -4.71939564e-01 -2.02406213e-01] | [7.9313883781433105, 2.9625141620635986] |
d8a2924d-6b9a-4643-90fb-f1f809f7b197 | styletalk-one-shot-talking-head-generation | 2301.01081 | null | https://arxiv.org/abs/2301.01081v2 | https://arxiv.org/pdf/2301.01081v2.pdf | StyleTalk: One-shot Talking Head Generation with Controllable Speaking Styles | Different people speak with diverse personalized speaking styles. Although existing one-shot talking head methods have made significant progress in lip sync, natural facial expressions, and stable head motions, they still cannot generate diverse speaking styles in the final talking head videos. To tackle this problem, we propose a one-shot style-controllable talking face generation framework. In a nutshell, we aim to attain a speaking style from an arbitrary reference speaking video and then drive the one-shot portrait to speak with the reference speaking style and another piece of audio. Specifically, we first develop a style encoder to extract dynamic facial motion patterns of a style reference video and then encode them into a style code. Afterward, we introduce a style-controllable decoder to synthesize stylized facial animations from the speech content and style code. In order to integrate the reference speaking style into generated videos, we design a style-aware adaptive transformer, which enables the encoded style code to adjust the weights of the feed-forward layers accordingly. Thanks to the style-aware adaptation mechanism, the reference speaking style can be better embedded into synthesized videos during decoding. Extensive experiments demonstrate that our method is capable of generating talking head videos with diverse speaking styles from only one portrait image and an audio clip while achieving authentic visual effects. Project Page: https://github.com/FuxiVirtualHuman/styletalk. | ['Xin Yu', 'Zhidong Deng', 'Yu Ding', 'Tangjie Lv', 'Changjie Fan', 'Zhipeng Hu', 'Suzhen Wang', 'Yifeng Ma'] | 2023-01-03 | null | null | null | null | ['talking-head-generation', 'talking-face-generation', 'face-generation'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [ 1.65284351e-01 -2.48061530e-02 -1.12605795e-01 -7.03834951e-01
-5.62115610e-01 -6.12432897e-01 5.30716598e-01 -1.13038814e+00
1.90431505e-01 4.52537864e-01 5.33365667e-01 2.94347495e-01
4.98564810e-01 -5.51486611e-01 -7.15631247e-01 -7.55135953e-01
6.52438998e-01 2.93815825e-02 -2.35779479e-01 -1.77255154e-01
-3.52239273e-02 4.67483848e-01 -1.69918346e+00 2.86561400e-01
6.93397641e-01 8.76396656e-01 3.21012735e-01 1.00606716e+00
-2.71843016e-01 7.71028042e-01 -5.85507572e-01 -4.57883567e-01
-2.73359241e-03 -8.70800674e-01 -4.67083752e-01 6.08885229e-01
4.81048703e-01 -7.21964896e-01 -4.82407808e-01 1.10535240e+00
7.81190395e-01 4.28577252e-02 4.27156806e-01 -1.30323088e+00
-6.15144432e-01 6.92649484e-01 -4.21145111e-01 -2.68727452e-01
8.54431152e-01 6.76500261e-01 6.47549331e-01 -9.11442339e-01
8.82854640e-01 1.69577396e+00 2.40627795e-01 1.23935282e+00
-1.07233572e+00 -9.59842026e-01 2.55597651e-01 1.09985419e-01
-1.39747632e+00 -1.14033282e+00 1.28056657e+00 -9.44399089e-02
2.13651597e-01 3.06208819e-01 9.92190659e-01 1.47809970e+00
-9.51624736e-02 1.00146484e+00 7.06454098e-01 -1.34612232e-01
4.21880931e-02 3.32766920e-02 -4.64166731e-01 6.18985713e-01
-5.69355607e-01 -6.46719187e-02 -6.96079850e-01 1.49289280e-01
6.83599353e-01 -1.58117458e-01 -7.22062349e-01 -2.91846693e-02
-1.28040826e+00 4.85616386e-01 1.74324527e-01 8.67046714e-02
-3.59534949e-01 2.49608546e-01 2.83237129e-01 2.05181777e-01
3.88535261e-01 -1.22363932e-01 1.23156987e-01 -2.46138424e-01
-1.03357255e+00 3.54970843e-01 8.13147366e-01 1.24815190e+00
5.57745516e-01 5.52582443e-01 -4.07714516e-01 1.02478826e+00
3.14875215e-01 1.00630569e+00 4.17557389e-01 -1.23965514e+00
2.58009464e-01 1.96709409e-01 2.51227580e-02 -9.52949464e-01
-6.87403753e-02 1.31058991e-01 -6.36215270e-01 9.33972746e-02
1.89483717e-01 -5.04534483e-01 -6.05581999e-01 2.04129982e+00
5.01499176e-01 1.79453656e-01 -4.22052667e-02 1.05922806e+00
9.74702895e-01 9.79409933e-01 -1.74172804e-01 -4.50922728e-01
1.38565421e+00 -9.08670843e-01 -1.11871195e+00 -8.75336677e-02
1.19753465e-01 -8.94031525e-01 1.44239628e+00 2.10283756e-01
-1.36918497e+00 -6.69369400e-01 -8.16547394e-01 -1.47305742e-01
2.79223293e-01 3.24512035e-01 -4.52427240e-03 4.66840804e-01
-1.07320297e+00 1.76838949e-01 -5.13198078e-01 -1.86882481e-01
2.94630647e-01 1.22756481e-01 -2.90005952e-01 1.08493485e-01
-9.76788878e-01 3.20203245e-01 -5.50138317e-02 -8.04838352e-03
-9.48344111e-01 -6.67697132e-01 -1.00916409e+00 -5.26540689e-02
2.21043959e-01 -9.76358056e-01 1.54121983e+00 -1.62041366e+00
-2.37488604e+00 7.79931843e-01 -6.25582516e-01 7.56494105e-02
5.71410596e-01 -7.50591457e-02 -5.41849554e-01 3.65676671e-01
-6.85890093e-02 1.02196741e+00 1.33602142e+00 -1.33590877e+00
-5.41066349e-01 -1.11024547e-02 -1.09207563e-01 4.48656082e-01
-5.14141560e-01 2.27889121e-01 -7.66190469e-01 -8.86363983e-01
-2.93076903e-01 -9.49103951e-01 3.49017322e-01 3.65767688e-01
-5.54291070e-01 9.47343260e-02 1.20278752e+00 -5.13064027e-01
1.14498401e+00 -2.36240506e+00 5.24069071e-01 -2.68878251e-01
1.45238265e-01 1.90788731e-01 -2.38162369e-01 -4.01939871e-03
-6.75193742e-02 -1.82509869e-01 -3.06113195e-02 -5.73049605e-01
-1.39707357e-01 2.00341538e-01 -3.40310127e-01 3.05089951e-01
1.73671961e-01 8.09508383e-01 -9.26213920e-01 -5.61448097e-01
1.97883099e-01 8.53603482e-01 -9.17229414e-01 5.11664689e-01
-3.24002862e-01 7.19112337e-01 -4.05956268e-01 5.96243024e-01
6.65523946e-01 2.06228167e-01 1.06330268e-01 -4.38533902e-01
-1.50977433e-01 1.03076689e-01 -9.67474639e-01 1.77625000e+00
-6.48638129e-01 7.16346145e-01 4.91677135e-01 -4.48183358e-01
1.00872719e+00 5.64269483e-01 1.91423863e-01 -3.25985134e-01
3.88536096e-01 -6.64289529e-03 -2.71924675e-01 -7.77222693e-01
2.58133858e-01 -4.17837709e-01 -3.72186676e-02 3.81856531e-01
1.71409100e-01 -5.16096532e-01 -1.50462892e-02 -5.66602536e-02
4.71646577e-01 2.73903668e-01 -9.16574895e-02 2.14639325e-02
8.65697742e-01 -6.94107056e-01 5.12011170e-01 -2.54758764e-02
-1.52736828e-01 8.76004159e-01 5.07840872e-01 -3.39762479e-01
-1.05808938e+00 -1.10802245e+00 3.06833565e-01 1.17395282e+00
-1.46463874e-03 -3.51469874e-01 -1.33059990e+00 -3.11937958e-01
-4.51823682e-01 7.03405857e-01 -4.78507280e-01 -2.15495706e-01
-7.68222988e-01 -4.27565910e-02 6.37253046e-01 2.90597141e-01
6.78925514e-01 -1.27337968e+00 -2.31286421e-01 1.86610609e-01
-5.07796466e-01 -1.07754707e+00 -1.37736821e+00 -6.33686125e-01
-3.81811529e-01 -5.71452677e-01 -1.19705498e+00 -9.44948018e-01
6.26066566e-01 1.27250642e-01 8.00194621e-01 -2.95993358e-01
7.69638866e-02 3.96918982e-01 -1.24822736e-01 -2.19005376e-01
-7.08472610e-01 -1.40191436e-01 3.23634177e-01 6.74621344e-01
-8.32605213e-02 -7.87266970e-01 -7.57032335e-01 2.67372280e-01
-8.92780602e-01 4.91420001e-01 5.21874465e-02 7.98292875e-01
3.31484586e-01 -6.07704103e-01 5.23659945e-01 -4.35928673e-01
6.62634313e-01 -1.44519433e-01 -3.96903455e-01 1.04183353e-01
2.95399707e-02 -9.03891996e-02 9.60440695e-01 -8.26440930e-01
-1.35638130e+00 2.16902822e-01 -5.10860980e-01 -9.74121511e-01
-1.14490308e-01 -1.98274851e-01 -7.38823950e-01 1.58847004e-01
4.54003692e-01 5.07709503e-01 3.99765342e-01 -1.74597561e-01
7.59845734e-01 9.66611385e-01 9.24006462e-01 -5.64145327e-01
7.30333626e-01 5.52745342e-01 -5.24798691e-01 -9.71643090e-01
-5.41347682e-01 1.57254681e-01 -3.36735666e-01 -7.07615972e-01
1.02844906e+00 -9.36714768e-01 -9.31922019e-01 8.12623799e-01
-1.14622748e+00 -4.90138263e-01 -2.13952407e-01 1.46882907e-01
-9.78396058e-01 2.46841758e-01 -6.60825491e-01 -5.79978049e-01
-4.83696580e-01 -1.53200829e+00 1.29189968e+00 4.97878224e-01
-4.52622026e-01 -7.44741797e-01 -9.30870846e-02 1.93407059e-01
3.90911013e-01 1.51595697e-01 4.74774092e-01 1.39933974e-01
-2.66135573e-01 4.44551520e-02 1.14845142e-01 2.62942493e-01
6.32366776e-01 4.14112866e-01 -1.14861774e+00 -1.11796334e-01
5.54012805e-02 -1.10355906e-01 3.96830231e-01 3.95983040e-01
1.13354588e+00 -8.55583131e-01 -4.37061526e-02 1.14361763e+00
7.37045527e-01 4.20750767e-01 5.39204478e-01 -2.97513604e-01
8.65124643e-01 6.42582476e-01 2.97252655e-01 4.64533925e-01
3.50309938e-01 9.25872147e-01 1.74619462e-02 7.91039839e-02
-5.43716609e-01 -6.19728029e-01 7.76746273e-01 9.15113628e-01
-1.42962066e-02 -3.25043976e-01 -4.77661133e-01 3.53954792e-01
-1.34053326e+00 -1.28849220e+00 3.92581940e-01 1.78284287e+00
1.19590080e+00 -1.50860935e-01 3.23214620e-01 2.70930044e-02
8.51227880e-01 3.16228598e-01 -7.66615272e-01 -5.09831548e-01
-8.59960318e-02 -1.66730374e-01 -1.73985556e-01 6.44558072e-01
-6.49019539e-01 1.15085018e+00 5.75568104e+00 8.05013716e-01
-1.83494473e+00 -1.33728519e-01 7.13534653e-01 -5.84367573e-01
-7.21483290e-01 -3.07556957e-01 -7.77041435e-01 8.02591860e-01
8.49327743e-01 -6.25250578e-01 6.92186475e-01 8.66806567e-01
7.15856671e-01 5.45732617e-01 -1.02473271e+00 1.31537211e+00
2.53521711e-01 -1.42911327e+00 4.25458908e-01 -2.99947828e-01
5.55934608e-01 -5.60781896e-01 2.05558464e-01 6.81157559e-02
-7.24147782e-02 -8.08698952e-01 1.12015760e+00 6.98990941e-01
1.47500396e+00 -7.42623687e-01 4.38649021e-02 5.86684197e-02
-1.17128050e+00 6.34873807e-02 1.02822650e-02 1.48285657e-01
4.84031409e-01 1.13722645e-01 -5.73932409e-01 1.56181842e-01
5.18998086e-01 8.01244676e-01 -6.07194304e-02 4.56030995e-01
-2.29681790e-01 4.92865503e-01 -1.40380561e-01 1.97149105e-02
-7.00775608e-02 -1.17973730e-01 7.98693717e-01 1.20300770e+00
4.64824826e-01 2.51657754e-01 -1.70570552e-01 1.03273141e+00
-3.66124898e-01 1.92802511e-02 -7.74770319e-01 1.39837321e-02
5.41821361e-01 1.22428191e+00 -2.31545419e-01 -4.91282046e-01
-3.01764578e-01 1.23857903e+00 -1.58253253e-01 4.70886886e-01
-9.77857113e-01 -2.77260393e-01 1.11391997e+00 1.57347739e-01
9.18226093e-02 8.62166286e-02 -4.50629629e-02 -1.37547469e+00
-8.59826654e-02 -1.17990696e+00 -2.41506532e-01 -1.25126445e+00
-8.46199095e-01 9.10892546e-01 -1.71344161e-01 -1.41907060e+00
-5.60924649e-01 -2.06067666e-01 -9.51267660e-01 7.53961861e-01
-9.86708939e-01 -1.10731554e+00 -5.50537109e-01 8.58329415e-01
1.08509421e+00 -1.73383474e-01 5.95527053e-01 1.60225838e-01
-6.13644242e-01 8.92440915e-01 -2.99282104e-01 1.28255039e-01
9.95920837e-01 -7.72399485e-01 3.64291131e-01 5.71774483e-01
-2.95804381e-01 4.71667796e-01 9.23782229e-01 -3.60733688e-01
-1.56184626e+00 -9.74706769e-01 5.89490354e-01 -1.25647625e-02
4.31396931e-01 -4.78584737e-01 -7.58221865e-01 5.33983886e-01
3.96589339e-01 -6.78687096e-02 4.35745448e-01 -7.66976118e-01
-1.90195590e-01 -3.81834298e-01 -1.14400566e+00 1.06318390e+00
1.15458155e+00 -6.39073789e-01 -3.40895683e-01 1.41301695e-02
8.80950749e-01 -5.26357293e-01 -6.90529585e-01 1.20906435e-01
8.85146201e-01 -9.32641208e-01 8.30558479e-01 -1.75859958e-01
5.74794173e-01 -3.65433484e-01 3.61070223e-02 -1.44610775e+00
-3.36065218e-02 -1.30095136e+00 9.59862247e-02 1.69332564e+00
1.56220168e-01 -3.90733421e-01 5.74494064e-01 4.57864761e-01
-1.55685037e-01 -5.36025882e-01 -5.81679225e-01 -3.86285126e-01
6.47163764e-02 -2.77225256e-01 8.46372366e-01 7.77267039e-01
3.60164233e-02 3.37985545e-01 -7.84699857e-01 -1.54788733e-01
4.81197506e-01 1.16574064e-01 1.10774660e+00 -5.26514947e-01
-1.26054153e-01 -4.87146735e-01 -3.65771316e-02 -1.18393469e+00
3.57555628e-01 -5.51943898e-01 2.37441644e-01 -1.12297928e+00
4.38008420e-02 -2.99948119e-02 4.95338708e-01 2.48399511e-01
-2.75087118e-01 1.50322556e-01 4.87084687e-01 1.21165469e-01
-1.72076315e-01 8.30271661e-01 1.82047272e+00 -1.63180783e-01
-2.38761202e-01 5.08236773e-02 -6.51371360e-01 8.74316812e-01
7.50827849e-01 -9.13060233e-02 -5.82796693e-01 -4.38374311e-01
-1.60896957e-01 5.73654890e-01 2.82394260e-01 -8.67058039e-01
8.66463408e-02 -2.64255166e-01 2.86524087e-01 -2.68258214e-01
5.39790690e-01 -4.32930797e-01 3.55594456e-01 2.69568592e-01
-5.74836552e-01 -3.33944969e-02 8.95137563e-02 1.55191526e-01
-2.24517509e-01 1.71033487e-01 1.05835044e+00 6.76276488e-03
-3.86548102e-01 4.74320531e-01 -4.63398486e-01 -2.95699835e-02
8.31859827e-01 -2.68070251e-01 -2.52180528e-02 -9.92323399e-01
-9.18769002e-01 1.01532310e-01 7.54237294e-01 6.00239396e-01
7.87383080e-01 -1.60748875e+00 -7.34892845e-01 5.95244706e-01
-1.44134030e-01 -4.33551110e-02 4.96813983e-01 4.94182348e-01
-4.33186591e-01 -7.62030557e-02 -3.96533400e-01 -6.50700867e-01
-1.39778793e+00 5.92360795e-01 4.83591110e-01 5.45243621e-01
-7.00014591e-01 8.25385332e-01 6.45368457e-01 -1.69847831e-01
2.13289246e-01 -2.96843350e-01 -8.01906511e-02 8.31270665e-02
8.87250423e-01 5.94337359e-02 -4.95797455e-01 -8.94826174e-01
-9.48844925e-02 9.07256186e-01 2.00058937e-01 -2.98786908e-01
1.00518227e+00 -5.06293118e-01 2.54912257e-01 4.66803491e-01
1.40989304e+00 3.82652998e-01 -1.72831607e+00 7.63503686e-02
-6.74201906e-01 -4.44659740e-01 -2.90939331e-01 -3.39986980e-01
-1.42688334e+00 8.30364883e-01 3.80270362e-01 -1.78190559e-01
1.45390785e+00 -4.39535454e-02 1.03917074e+00 1.98551845e-02
1.33918107e-01 -6.99401438e-01 2.79953808e-01 2.56324440e-01
1.24497235e+00 -7.95501351e-01 -5.77311635e-01 -4.08870518e-01
-9.60349739e-01 1.22401762e+00 4.64296699e-01 -8.16355739e-03
5.00806868e-01 5.85827470e-01 4.05367583e-01 1.72950238e-01
-8.23519886e-01 5.06584086e-02 1.59469530e-01 7.11941421e-01
4.11915511e-01 -3.51532735e-02 8.72495305e-03 4.79495645e-01
-7.53545880e-01 3.56299818e-01 4.97927606e-01 2.53447175e-01
-2.98700869e-01 -8.34262550e-01 -5.10772705e-01 -1.40643120e-01
-3.76680374e-01 9.11345929e-02 -3.96161258e-01 3.27399284e-01
-6.47347718e-02 9.06799376e-01 2.11547893e-02 -6.23257518e-01
3.84284586e-01 2.91793823e-01 3.94711494e-01 -2.81684220e-01
-3.85088414e-01 3.91414553e-01 2.36309301e-02 -6.86827660e-01
-2.68397838e-01 -6.19076431e-01 -1.07030082e+00 -6.64152086e-01
1.77794918e-01 -1.33338213e-01 3.22034419e-01 7.10942924e-01
3.26444775e-01 6.16789699e-01 1.00376296e+00 -1.31479645e+00
-1.06039710e-01 -9.06136453e-01 -2.67498791e-01 5.18490434e-01
5.59292495e-01 -4.39451396e-01 -5.17459512e-01 6.04411244e-01] | [13.177865028381348, -0.4083639681339264] |
97fada87-7702-40b7-a218-0af3bc9aa8c1 | translate-reverberated-speech-to-anechoic | 2007.08052 | null | https://arxiv.org/abs/2007.08052v1 | https://arxiv.org/pdf/2007.08052v1.pdf | Translate Reverberated Speech to Anechoic Ones: Speech Dereverberation with BERT | Single channel speech dereverberation is considered in this work. Inspired by the recent success of Bidirectional Encoder Representations from Transformers (BERT) model in the domain of Natural Language Processing (NLP), we investigate its applicability as backbone sequence model to enhance reverberated speech signal. We present a variation of the basic BERT model: a pre-sequence network, which extracts local spectral-temporal information and/or provides order information, before the backbone sequence model. In addition, we use pre-trained neural vocoder for implicit phase reconstruction. To evaluate our method, we used the data from the 3rd CHiME challenge, and compare our results with other methods. Experiments show that the proposed method outperforms traditional method WPE, and achieve comparable performance with state-of-the-art BLSTM-based sequence models. | ['Yang Jiao'] | 2020-07-16 | null | null | null | null | ['speech-dereverberation'] | ['speech'] | [ 4.87442106e-01 -5.56310313e-03 3.33262533e-01 5.15547208e-02
-9.10641789e-01 -4.03116584e-01 6.13244832e-01 -1.55465603e-01
-6.77996337e-01 7.90679455e-01 7.65822470e-01 -5.89301229e-01
2.81021982e-01 -3.25417757e-01 -7.46096790e-01 -7.38215566e-01
-9.60503146e-03 -1.25474393e-01 3.75785917e-01 -4.62056130e-01
-1.06977047e-02 1.84679911e-01 -1.34039438e+00 7.18924582e-01
6.11587167e-01 8.40318441e-01 7.89670765e-01 1.14391649e+00
4.49419729e-02 1.05412936e+00 -8.33489954e-01 -7.83586409e-03
-1.59286931e-01 -7.84963965e-01 -6.97006047e-01 -3.49646926e-01
-1.21822357e-02 -1.32514060e-01 -5.62162995e-01 7.87356734e-01
1.05197501e+00 3.43079537e-01 5.20154774e-01 -5.35702109e-01
-5.05995870e-01 8.27154040e-01 -7.93585405e-02 7.24810123e-01
5.18121898e-01 -6.06788844e-02 9.38636065e-01 -1.07390094e+00
5.64087987e-01 1.07920778e+00 8.94643188e-01 4.94477779e-01
-1.24125946e+00 -3.51403177e-01 9.46734175e-02 8.43300402e-01
-1.17258358e+00 -8.03810418e-01 9.40476954e-01 -1.05137825e-01
1.67992425e+00 3.32758248e-01 5.08813560e-01 1.66987479e+00
1.05118945e-01 9.49471354e-01 1.13044941e+00 -8.38949919e-01
6.18547983e-02 -2.00190619e-01 1.84387490e-01 2.74904698e-01
-6.24044299e-01 6.81289196e-01 -8.92461836e-01 8.36779252e-02
2.78017789e-01 -6.74102604e-01 -9.68722045e-01 1.91597968e-01
-1.14731991e+00 3.10721487e-01 3.82174581e-01 6.25743926e-01
-6.67129993e-01 2.26760581e-01 4.35627162e-01 4.61507589e-01
6.63604021e-01 1.35334745e-01 -4.60968852e-01 -4.22219783e-01
-1.15803945e+00 1.93862751e-01 8.45068157e-01 6.71969116e-01
3.75024408e-01 6.16405189e-01 -4.11554575e-01 9.83469129e-01
2.60693640e-01 3.24499369e-01 6.73452139e-01 -5.34851909e-01
5.59563518e-01 -6.52153373e-01 1.02324806e-01 -4.70796227e-01
-3.21013451e-01 -6.79853678e-01 -8.86836112e-01 -1.67355448e-01
2.59815920e-02 -1.36190280e-01 -1.01340806e+00 1.57028711e+00
-3.65725011e-02 6.46519780e-01 2.50648558e-01 8.91027987e-01
7.71525919e-01 1.16332710e+00 -2.36820042e-01 -5.83321929e-01
1.06087637e+00 -1.20454609e+00 -1.18983078e+00 -2.91259168e-03
1.94870830e-01 -9.87280428e-01 9.64267313e-01 6.72457159e-01
-9.92892921e-01 -9.14026260e-01 -1.23916161e+00 2.04344373e-02
-1.60073727e-01 -9.53174457e-02 -1.72597870e-01 6.40261531e-01
-1.36590314e+00 8.07274520e-01 -7.94118404e-01 -1.37231648e-01
-2.81040847e-01 -7.16219097e-02 4.04157080e-02 1.60069421e-01
-1.65075791e+00 9.67924535e-01 4.23000664e-01 3.63314718e-01
-1.20772374e+00 -8.35645914e-01 -7.11350203e-01 1.77675128e-01
5.20260772e-03 -5.47747135e-01 1.77263737e+00 -7.90959716e-01
-2.08168817e+00 2.24261761e-01 -5.00007093e-01 -1.04442823e+00
1.95334002e-01 -4.82729703e-01 -7.33911037e-01 3.13758701e-01
-6.90539002e-01 2.59453177e-01 1.11136699e+00 -1.11971807e+00
-3.61294866e-01 1.87501475e-01 -3.45834643e-01 2.28116915e-01
-1.27684578e-01 2.30660260e-01 1.09581478e-01 -9.28887308e-01
-2.59580910e-01 -4.16871995e-01 -1.43708035e-01 -7.65115857e-01
-1.24815114e-01 -3.62909064e-02 7.63788223e-01 -1.41999114e+00
1.51527429e+00 -2.18064880e+00 1.59225896e-01 -1.19463861e-01
-1.90362811e-01 6.87681615e-01 -3.09994966e-01 8.80944669e-01
-3.47206831e-01 -5.58564849e-02 -4.25572783e-01 -6.30036771e-01
2.58489400e-02 1.39604792e-01 -7.20445573e-01 2.80028790e-01
2.35304639e-01 6.36586547e-01 -8.05790901e-01 -6.31425902e-02
4.15336341e-01 9.46822226e-01 -4.59502399e-01 3.57072800e-01
-1.12412743e-01 5.49343824e-01 2.31676385e-01 1.05150202e-02
5.26170909e-01 2.95554966e-01 8.41957554e-02 -2.02312380e-01
-4.46333379e-01 1.20299590e+00 -8.05949688e-01 1.74791360e+00
-7.66031086e-01 9.86972332e-01 1.38672605e-01 -1.11071610e+00
7.68484950e-01 1.06805933e+00 7.49405697e-02 -8.92479837e-01
-2.92929504e-02 3.67571622e-01 2.48747259e-01 -4.61147904e-01
4.44451541e-01 -3.35497320e-01 5.66895485e-01 5.09652272e-02
4.64360654e-01 -1.21046424e-01 -1.72318313e-02 -1.03226259e-01
1.12848163e+00 4.14839327e-01 3.33542228e-01 -1.79862455e-01
7.80677497e-01 -4.49780762e-01 3.32536936e-01 5.54440022e-01
-2.13382214e-01 1.03342378e+00 3.35623547e-02 -2.32659094e-02
-1.16402113e+00 -1.19968784e+00 7.83720613e-02 8.11406672e-01
-2.38960937e-01 -4.15678084e-01 -8.31401110e-01 -1.18223883e-01
-6.54723644e-01 1.06773067e+00 -1.58672810e-01 8.58523175e-02
-1.03982508e+00 -5.45206308e-01 7.31746495e-01 4.74695861e-01
3.56317163e-01 -1.29350686e+00 -3.29096049e-01 8.33651066e-01
-7.63316393e-01 -1.27179825e+00 -5.93261898e-01 3.17005605e-01
-5.01328528e-01 -5.53354859e-01 -9.37109709e-01 -8.41458082e-01
-1.71136335e-01 2.12912127e-01 1.05639470e+00 -3.18827987e-01
2.98285317e-02 2.10854292e-01 -9.20637190e-01 -3.94601762e-01
-7.29038656e-01 7.43029267e-02 1.26341870e-02 1.25237837e-01
4.70393039e-02 -9.14369881e-01 -5.48819602e-01 -7.24885566e-03
-7.80891538e-01 5.98101802e-02 3.64472896e-01 8.19818318e-01
3.70483816e-01 -8.97964910e-02 7.41244197e-01 -3.50934923e-01
8.33312333e-01 -7.83404708e-02 -3.72914076e-01 2.00917684e-02
-3.84296983e-01 1.30041972e-01 8.42641056e-01 -4.77881879e-01
-1.34455776e+00 -3.29344869e-01 -9.48354721e-01 -3.64758998e-01
2.30164845e-02 5.70884347e-01 -1.94775239e-02 2.32416451e-01
6.97159529e-01 6.99082911e-01 -3.43614519e-01 -8.39647412e-01
2.69579947e-01 1.04308414e+00 7.54954159e-01 -2.54393965e-01
5.60767412e-01 5.73105812e-02 -4.37548339e-01 -1.29712725e+00
-3.91533077e-01 -6.88509703e-01 -3.68836820e-01 -1.56315625e-01
7.08653390e-01 -9.19062316e-01 -3.91340792e-01 5.23188770e-01
-1.79291654e+00 -5.77744424e-01 -1.67345434e-01 7.47037709e-01
-7.05007017e-01 7.19319940e-01 -1.07371283e+00 -1.08078671e+00
-5.91946423e-01 -9.61196363e-01 7.69841135e-01 -2.02015251e-01
-9.86068696e-02 -8.35204184e-01 4.30240899e-01 1.43457755e-01
6.25637531e-01 5.30763855e-03 5.20630896e-01 -4.58462179e-01
-4.43958223e-01 3.63639772e-01 2.33032912e-01 8.93095195e-01
-1.36279836e-01 -3.83245498e-01 -1.48235846e+00 -2.20693722e-01
4.62797552e-01 -1.18693914e-02 1.18802154e+00 3.48853707e-01
8.76011491e-01 -3.33133936e-01 -1.12753026e-02 5.11095881e-01
1.22872114e+00 4.05464947e-01 1.06147969e+00 9.73474458e-02
3.34901512e-01 4.04117912e-01 2.63464689e-01 2.88375288e-01
1.73247665e-01 6.44167721e-01 7.76611641e-02 -3.49569460e-03
-7.52336919e-01 -4.68269706e-01 8.88210714e-01 1.76050615e+00
-1.27873659e-01 -5.77187717e-01 -7.60221183e-01 6.84851825e-01
-1.61021006e+00 -1.19880176e+00 -1.19379573e-01 1.98377955e+00
9.75948274e-01 8.68295208e-02 -6.99738860e-02 6.40393198e-01
4.80614990e-01 5.58179975e-01 1.34108424e-01 -8.05301964e-01
-2.19505489e-01 7.82114208e-01 3.49911481e-01 1.00044155e+00
-8.33238721e-01 8.39742064e-01 6.54957676e+00 1.05477381e+00
-1.06956148e+00 5.68739712e-01 9.58622098e-02 7.10842833e-02
-3.35651636e-01 -2.36287549e-01 -5.57607412e-01 2.64631361e-01
1.66300154e+00 1.12356074e-01 7.57676244e-01 1.39182240e-01
5.58190525e-01 3.14757437e-01 -1.01019311e+00 8.00009310e-01
1.51943788e-01 -1.06735647e+00 -3.22438568e-01 -2.62070477e-01
4.92307872e-01 2.79500008e-01 -4.12319005e-02 4.66543138e-01
-5.18795326e-02 -8.94418716e-01 1.09913862e+00 4.83580589e-01
6.90708458e-01 -7.70344853e-01 5.64496577e-01 5.28095722e-01
-1.47437322e+00 6.64396062e-02 -2.47390792e-01 -1.97692901e-01
5.58086097e-01 6.64197087e-01 -1.19363463e+00 8.83635581e-01
6.00364506e-01 4.67225790e-01 1.76592812e-01 1.14684689e+00
-5.89949787e-01 1.22159803e+00 -2.26320729e-01 -7.94022307e-02
2.42096379e-01 1.58883885e-01 8.12682092e-01 1.67441308e+00
4.04985815e-01 2.26184372e-02 -3.93750757e-01 6.64833367e-01
1.00181513e-01 -1.85520902e-01 -4.23552364e-01 -1.32262796e-01
3.47464919e-01 8.19344103e-01 -1.81382690e-02 -2.05351517e-01
-4.85287547e-01 1.25106573e+00 1.51468709e-01 8.11583996e-01
-8.57425272e-01 -5.66703677e-01 4.05906022e-01 -1.77356794e-01
7.81340718e-01 -5.23998380e-01 2.39274904e-01 -9.35295403e-01
2.05975259e-03 -8.84923935e-01 -2.44632676e-01 -1.00654030e+00
-9.60148931e-01 1.02572620e+00 -2.88951308e-01 -1.19414151e+00
-6.27709270e-01 -6.62547886e-01 -4.27313566e-01 1.22797728e+00
-1.99894381e+00 -9.41309571e-01 2.37663224e-01 4.15624648e-01
8.42628837e-01 1.54255927e-01 9.30164635e-01 5.18614948e-01
-1.62513703e-01 3.12409669e-01 2.27041200e-01 -3.48919667e-02
5.31411588e-01 -1.14345539e+00 9.16881800e-01 1.18382537e+00
4.89504308e-01 4.38806981e-01 1.05937397e+00 -3.16213101e-01
-1.07907224e+00 -9.38648283e-01 1.34931338e+00 -1.97884709e-01
6.89560413e-01 -6.83232605e-01 -9.27810848e-01 5.69093466e-01
8.16619158e-01 -3.20814431e-01 4.14442003e-01 -1.87960148e-01
-3.35572839e-01 -1.01734564e-01 -5.77716231e-01 5.28841972e-01
9.72642481e-01 -8.95116031e-01 -1.03389692e+00 -5.39290719e-02
1.41487467e+00 -3.70168895e-01 -4.78054136e-01 4.05603111e-01
3.74087214e-01 -1.00288963e+00 1.11069465e+00 -5.45937121e-02
1.40332565e-01 -3.27059239e-01 -4.81870204e-01 -1.90000939e+00
-2.53848970e-01 -1.06127286e+00 -4.51805949e-01 1.13719726e+00
5.11507332e-01 -5.37097216e-01 1.58779562e-01 -4.90603715e-01
-8.00953150e-01 -3.99576217e-01 -1.16173971e+00 -9.61266696e-01
-7.45650083e-02 -8.11715364e-01 4.34838295e-01 3.93277645e-01
3.72497588e-02 4.92394060e-01 -6.41334057e-01 3.13498408e-01
4.06204730e-01 -3.63790661e-01 1.99279234e-01 -6.22849047e-01
-7.62663066e-01 -3.44990104e-01 5.91465607e-02 -1.33873367e+00
1.99502274e-01 -6.92995727e-01 5.67667365e-01 -1.59266758e+00
-4.42660451e-01 2.82607794e-01 -5.57762504e-01 5.39835393e-02
-1.39427036e-01 8.65714401e-02 3.62616748e-01 -2.09215924e-01
2.95926053e-02 9.80347216e-01 1.05349457e+00 -9.42573920e-02
-1.76647857e-01 -3.46335843e-02 -1.85067713e-01 4.16386604e-01
6.98859811e-01 -3.26540887e-01 -4.27428156e-01 -4.26535726e-01
-5.84929809e-02 3.52768213e-01 3.97484779e-01 -1.22600853e+00
2.02371880e-01 4.54287291e-01 5.26937507e-02 -7.68146217e-01
5.79534352e-01 -6.94267750e-01 -1.11877488e-03 6.07670188e-01
-3.11405361e-01 -8.84057954e-02 2.89147586e-01 4.38521922e-01
-5.93404651e-01 -1.51812211e-01 6.54976547e-01 1.36108641e-02
-6.18114054e-01 -8.22475702e-02 -7.75517464e-01 -1.59634247e-01
1.53569967e-01 3.02734144e-04 -2.28191033e-01 -3.89680386e-01
-7.75594771e-01 -3.22300553e-01 -3.11473757e-01 4.04824555e-01
7.93040454e-01 -1.22191954e+00 -9.37340617e-01 1.23048358e-01
-1.93345129e-01 -4.97361481e-01 2.72081465e-01 8.92235994e-01
-4.40962285e-01 6.73338413e-01 2.14213908e-01 -3.16602916e-01
-1.09168446e+00 7.01624334e-01 5.40929317e-01 -3.39858562e-01
-7.58347452e-01 8.92722070e-01 9.05621797e-02 -3.20180953e-01
2.89941639e-01 -6.26360178e-01 -2.84244984e-01 -9.52842236e-02
8.69021893e-01 4.16411579e-01 4.08696830e-01 -5.11795342e-01
-3.37373376e-01 7.81395137e-02 2.39546999e-01 -9.56422627e-01
1.29748034e+00 -3.06716472e-01 8.19934532e-02 8.04625332e-01
1.31739211e+00 -5.25790751e-02 -1.00155473e+00 -4.11617219e-01
-4.19318955e-03 -1.81625336e-02 2.18560785e-01 -8.91337037e-01
-4.47642684e-01 1.41461599e+00 6.78153694e-01 2.42554486e-01
1.39911246e+00 -5.70020854e-01 1.04198480e+00 9.72114950e-02
4.07123208e-01 -1.03829241e+00 8.59413743e-02 9.62111592e-01
1.17333949e+00 -5.75937748e-01 -6.22324109e-01 -2.66835302e-01
-3.30408394e-01 1.08637202e+00 1.10896558e-01 -9.08775926e-02
6.35587275e-01 5.15463829e-01 1.00254796e-01 4.38681871e-01
-9.44600463e-01 -3.75800639e-01 2.12461948e-01 8.59761178e-01
6.92262053e-01 -3.53618748e-02 -4.56879765e-01 5.98265111e-01
-5.07114351e-01 7.85900801e-02 4.06461775e-01 6.54864550e-01
-5.66848874e-01 -1.19441640e+00 -5.17619610e-01 -2.87531942e-01
-5.35055339e-01 -6.74713671e-01 -1.41345397e-01 3.70261997e-01
-1.38012707e-01 1.27994490e+00 -1.18616492e-01 -4.72246289e-01
5.14212251e-01 4.21949744e-01 6.80171013e-01 -3.16001624e-01
-9.82600451e-01 5.78783512e-01 4.81827617e-01 -3.64229321e-01
-5.27763546e-01 -5.82476258e-01 -1.14160621e+00 9.21517164e-02
-3.01315159e-01 1.98635265e-01 7.16830969e-01 9.53331113e-01
8.98693278e-02 1.33153343e+00 6.77306592e-01 -1.05211532e+00
-6.36482835e-01 -1.53658473e+00 -4.42349255e-01 -2.42383685e-02
1.01530147e+00 -1.16028018e-01 -4.61653471e-01 2.67939270e-01] | [14.958824157714844, 6.028085708618164] |
2f5a891e-c185-4dcd-8320-8ff85efa7263 | integrated-community-occupancy-models-a | 2109.01894 | null | https://arxiv.org/abs/2109.01894v1 | https://arxiv.org/pdf/2109.01894v1.pdf | Integrated community occupancy models: A framework to assess occurrence and biodiversity dynamics using multiple data sources | The occurrence and distributions of wildlife populations and communities are shifting as a result of global changes. To evaluate whether these shifts are negatively impacting biodiversity processes, it is critical to monitor the status, trends, and effects of environmental variables on entire communities. However, modeling the dynamics of multiple species simultaneously can require large amounts of diverse data, and few modeling approaches exist to simultaneously provide species and community level inferences. We present an "integrated community occupancy model" (ICOM) that unites principles of data integration and hierarchical community modeling in a single framework to provide inferences on species-specific and community occurrence dynamics using multiple data sources. We use simulations to compare the ICOM to previously developed hierarchical community occupancy models and single species integrated distribution models. We then apply our model to assess the occurrence and biodiversity dynamics of foliage-gleaning birds in the White Mountain National Forest in the northeastern USA from 2010-2018 using three independent data sources. Simulations reveal that integrating multiple data sources in the ICOM increased precision and accuracy of species and community level inferences compared to single data source models, although benefits of integration were dependent on data source quality (e.g., amount of replication). Compared to single species models, the ICOM yielded more precise species-level estimates. Within our case study, the ICOM had the highest out-of-sample predictive performance compared to single species models and models that used only a subset of the three data sources. The ICOM offers an attractive approach to estimate species and biodiversity dynamics, which is additionally valuable to inform management objectives of both individual species and their broader communities. | ['Elise F. Zipkin', 'Michael T. Hallworth', 'T. Scott Sillett', 'Wendy Leuenberger', 'Jeffrey W. Doser'] | 2021-09-04 | null | null | null | null | ['data-integration'] | ['knowledge-base'] | [ 5.64984158e-02 -7.53907561e-01 -2.02226818e-01 2.79605865e-01
2.05913991e-01 -6.94574773e-01 6.48116946e-01 6.71895146e-01
-8.09727967e-01 9.14680064e-01 6.20027542e-01 -5.93412161e-01
-2.80192375e-01 -9.77294564e-01 -5.17227113e-01 -4.89377886e-01
-8.08321774e-01 1.31836727e-01 3.61531198e-01 -1.71256796e-01
1.99904114e-01 4.83865142e-01 -1.81804490e+00 -2.78445512e-01
8.13152194e-01 2.37720646e-02 7.08757401e-01 4.68400389e-01
-4.17217351e-02 1.58428118e-01 -4.11754996e-01 1.67708322e-01
2.38722429e-01 -6.52457997e-02 -1.81228846e-01 -2.19046667e-01
4.38545913e-01 -7.05078304e-01 5.95930703e-02 8.03351760e-01
3.53459477e-01 -1.06116429e-01 6.20484948e-01 -1.15751982e+00
-3.88821214e-01 9.11174476e-01 -8.25157881e-01 3.58156115e-01
3.37064043e-02 4.89527255e-01 9.81483042e-01 -4.41691786e-01
6.69211149e-01 1.36591446e+00 1.22962594e+00 -4.17408496e-01
-2.32247066e+00 -1.06645322e+00 5.97562432e-01 3.78073528e-02
-1.74645615e+00 -2.65847653e-01 2.41062418e-02 -1.08829391e+00
1.23463714e+00 1.50852218e-01 1.51132333e+00 4.83302027e-01
3.98645550e-01 -6.97752014e-02 9.76052344e-01 1.78982750e-01
1.94853798e-01 -2.30192095e-01 -4.27856259e-02 1.27344877e-02
9.23333883e-01 5.08116245e-01 -5.08558869e-01 -7.10181713e-01
9.30556536e-01 4.75030631e-01 -6.46466538e-02 -8.59761797e-03
-1.00209486e+00 7.59045780e-01 7.89108157e-01 1.30889893e-01
-5.31290054e-01 2.23783687e-01 3.11807841e-01 -1.94341689e-01
5.52465737e-01 3.16325188e-01 -2.80270904e-01 -6.04342408e-02
-1.32859707e+00 3.86216372e-01 7.57911086e-01 3.88611078e-01
7.18376279e-01 4.18404713e-02 3.59744489e-01 9.99962568e-01
5.56927264e-01 1.03528941e+00 -1.72631875e-01 -8.15055490e-01
-9.90385711e-02 5.95625043e-01 3.04265738e-01 -1.31061518e+00
-4.64696348e-01 -7.35948205e-01 -6.95205331e-01 2.16387495e-01
2.63776690e-01 2.08417047e-02 -7.23261476e-01 2.03886485e+00
3.06068242e-01 4.33684140e-02 -5.81076741e-01 6.46822572e-01
2.79198885e-01 7.91271865e-01 5.67259610e-01 -4.80362892e-01
9.20805633e-01 -1.58164531e-01 -3.74868274e-01 -2.54871458e-01
-2.29628664e-02 -2.89589882e-01 5.96458375e-01 -1.79605260e-01
-5.95078409e-01 -2.79413372e-01 -8.46876204e-01 5.95228255e-01
-4.84682322e-01 -4.46599603e-01 5.88953018e-01 2.90207058e-01
-1.18123877e+00 4.43555892e-01 -1.08890736e+00 -7.87967324e-01
2.59274215e-01 -6.89329728e-02 -2.47194231e-01 2.93998778e-01
-8.94058168e-01 1.05675030e+00 -2.09573489e-02 2.48889387e-01
-9.13464785e-01 -9.96544600e-01 -5.96758187e-01 2.51257777e-01
3.71176773e-03 -4.61955607e-01 8.14365566e-01 -7.19457209e-01
-3.29182744e-01 4.34603721e-01 6.18732832e-02 -4.47227895e-01
2.04271957e-01 3.46915722e-01 -1.13151975e-01 -1.99269548e-01
8.54188144e-01 1.02505910e+00 -1.32041723e-01 -1.25046241e+00
-9.97328877e-01 -6.15410507e-01 -2.60709345e-01 2.81810611e-01
-6.53610975e-02 -7.07051903e-03 6.06395900e-01 -5.32241702e-01
2.50590853e-02 -8.39861631e-01 -3.96036983e-01 4.35248733e-01
5.60400367e-01 4.59489346e-01 2.94876397e-01 -5.92160881e-01
1.52931070e+00 -1.93062735e+00 2.03905627e-01 -8.55211020e-02
1.66903153e-01 2.14920938e-02 -1.40871331e-01 1.22688580e+00
5.22946656e-01 6.37241900e-01 -6.46780550e-01 3.29956949e-01
-2.82955289e-01 4.37343329e-01 2.95730168e-03 5.78492343e-01
6.37036115e-02 9.64329913e-02 -1.00781059e+00 -3.31949770e-01
5.46986878e-01 5.26016772e-01 -9.82869208e-01 -2.80022830e-01
-1.35859875e-02 2.10292682e-01 3.09920043e-01 7.08740711e-01
7.84371495e-01 3.09485430e-03 5.68851054e-01 4.20722276e-01
-9.80563521e-01 1.62694201e-01 -1.02988327e+00 8.92641068e-01
-3.21126908e-01 6.44344568e-01 6.81318641e-01 -2.64902383e-01
7.22836494e-01 -2.78434187e-01 5.24565578e-01 -1.48508593e-01
-4.48541790e-01 4.00424182e-01 8.35486650e-01 7.57844895e-02
5.46633780e-01 -5.51276386e-01 6.22018948e-02 4.00234133e-01
-8.17179605e-02 -1.54392481e-01 1.98839307e-01 -3.95262428e-02
8.39021027e-01 -9.29878354e-02 7.79879391e-01 -9.55589235e-01
-1.36817425e-01 5.29587626e-01 7.43524194e-01 7.08870769e-01
-5.39032817e-01 -2.35212684e-01 3.30532014e-01 -2.39727229e-01
-1.12669396e+00 -1.29811811e+00 -9.54723299e-01 1.04441500e+00
1.76650241e-01 -4.83676791e-01 1.37339935e-01 5.53449094e-01
9.66835082e-01 6.95173860e-01 -5.17845929e-01 -9.45908651e-02
2.99840003e-01 -1.17509615e+00 8.73367429e-01 3.58473510e-01
3.56268674e-01 -8.07374775e-01 -6.88502371e-01 5.17838418e-01
-2.33270615e-01 -3.66704971e-01 -2.89470911e-01 3.26066256e-01
-8.58930826e-01 -1.04391611e+00 -2.66486824e-01 -4.18593735e-01
3.24393928e-01 6.42417669e-01 9.03601170e-01 1.44506440e-01
-5.85266352e-01 1.36540160e-01 -1.79704383e-01 -4.07085605e-02
-2.79801190e-01 8.14533904e-02 5.44059098e-01 -5.68171501e-01
3.85271698e-01 -1.40325832e+00 -5.85430026e-01 4.21360284e-01
-8.46038520e-01 1.08215868e-01 6.42257333e-01 6.73844039e-01
2.47977808e-01 -1.37163967e-01 1.18339813e+00 -2.54449993e-01
3.92654955e-01 -9.06776488e-01 -5.38704395e-01 -9.42637846e-02
-5.93266845e-01 -3.25102419e-01 2.51737833e-01 -5.23240685e-01
-6.91648126e-01 -1.02445833e-01 8.73848796e-02 2.34292313e-01
-1.66125625e-01 1.28246963e+00 2.36041635e-01 1.75989285e-01
7.27935016e-01 3.92612331e-02 4.51704651e-01 -3.53702068e-01
1.27827730e-02 8.75971794e-01 3.76934171e-01 -7.23254442e-01
4.53808933e-01 3.56898278e-01 -9.14178640e-02 -1.33799827e+00
1.31899770e-03 -5.26333570e-01 -8.77847552e-01 -5.85328102e-01
4.05744791e-01 -1.42714953e+00 -7.41749525e-01 4.56232339e-01
-5.65382659e-01 -6.36901438e-01 -1.31481402e-02 7.12630630e-01
-1.38449937e-01 1.79235399e-01 -3.08423787e-01 -1.11255276e+00
-6.60156980e-02 -9.22055244e-01 5.77633083e-01 -2.16213688e-02
-1.91710114e-01 -1.00500882e+00 5.85073590e-01 -7.43222460e-02
6.80925012e-01 5.71181655e-01 6.95181072e-01 -1.42629566e-02
-4.07597542e-01 1.72253907e-01 -2.33822972e-01 -6.53837174e-02
6.96341276e-01 5.74836731e-01 -2.64361650e-01 -4.08296973e-01
-3.67113590e-01 2.74106920e-01 9.02031362e-01 7.42261350e-01
4.06552136e-01 -4.73464072e-01 -4.60641444e-01 5.26705563e-01
1.74793506e+00 1.02979869e-01 5.55367507e-02 2.43911460e-01
1.73471257e-01 7.25928724e-01 1.96070492e-01 5.18932998e-01
5.92773080e-01 5.12606263e-01 5.67739487e-01 5.48423342e-02
2.77541000e-02 -4.96069133e-01 6.26733541e-01 4.27115232e-01
9.88505110e-02 3.04035634e-01 -1.42649615e+00 1.01631784e+00
-1.77191865e+00 -1.69369304e+00 -6.88382208e-01 2.49710846e+00
8.34860086e-01 -1.99456424e-01 6.09519720e-01 -6.46262050e-01
9.56344664e-01 1.49718851e-01 -8.46026242e-01 -1.26242682e-01
-1.34305105e-01 -1.23219453e-01 1.00834131e+00 8.38207304e-01
-4.98510778e-01 8.00862253e-01 7.75491762e+00 2.59209752e-01
-9.16609645e-01 -2.02838764e-01 2.94915643e-02 -5.45809090e-01
-3.57005656e-01 3.36153656e-01 -9.01906788e-01 4.27421212e-01
1.00742638e+00 -7.69146383e-01 7.41284251e-01 4.07116294e-01
8.62770736e-01 -5.48224449e-01 -4.03865725e-01 1.97323471e-01
-2.73610532e-01 -1.02422833e+00 -1.10176831e-01 7.70732284e-01
7.15185523e-01 4.29692358e-01 -5.85695148e-01 2.68458456e-01
9.60493445e-01 -8.21548402e-01 6.37173772e-01 3.20600122e-01
7.05605984e-01 -5.98885059e-01 4.00768250e-01 5.40671468e-01
-1.68597412e+00 -2.51392394e-01 -3.85407478e-01 -1.11791670e+00
3.80925506e-01 2.24663511e-01 -4.75582361e-01 -4.53885309e-02
1.03079581e+00 1.02006781e+00 -6.73546910e-01 1.25160170e+00
2.34215945e-01 8.35843563e-01 -8.71815026e-01 1.54359475e-01
2.10277528e-01 -4.06867534e-01 6.16539359e-01 9.22873795e-01
1.16872594e-01 -1.29143119e-01 5.40928185e-01 1.17828178e+00
3.86483818e-01 -6.54740408e-02 -7.41632760e-01 -2.13532776e-01
1.19404328e+00 1.12419724e+00 -4.96643305e-01 4.31198329e-02
-3.98543447e-01 3.25341672e-02 1.94095403e-01 1.88905492e-01
-4.82829958e-01 9.13835913e-02 1.37125766e+00 6.22727454e-01
1.44879937e-01 -5.70260882e-01 2.48668194e-02 -1.05514038e+00
-5.39975107e-01 -6.69314563e-01 3.36080045e-01 -6.39713526e-01
-1.35666955e+00 -2.31473118e-01 5.35898447e-01 -9.68389869e-01
-1.88898447e-03 -1.79758474e-01 -3.57370645e-01 1.22868335e+00
-1.19164050e+00 -1.16000223e+00 -1.89829990e-01 -6.82585090e-02
8.99934396e-02 3.68481606e-01 5.82040608e-01 -3.96178890e-04
-4.56368089e-01 -2.18149915e-01 6.89933717e-01 -5.88509917e-01
3.61822695e-01 -9.90375757e-01 2.67204225e-01 7.69387484e-01
-4.54323709e-01 1.11046243e+00 5.03544450e-01 -1.45501637e+00
-8.95427763e-01 -9.50652957e-01 2.78552711e-01 1.74199894e-01
1.26503646e+00 -3.59570801e-01 -6.98386967e-01 8.82378817e-01
-1.82941884e-01 -6.21900439e-01 1.22980702e+00 5.16069710e-01
-1.08703330e-01 -3.14843923e-01 -1.32593226e+00 6.47343516e-01
1.11691940e+00 -5.03448129e-01 -2.93159753e-01 -1.65092126e-01
4.33093637e-01 5.40850103e-01 -1.43928730e+00 3.55252028e-01
1.46105218e+00 -4.86853868e-01 8.93425941e-01 -3.42880577e-01
3.25338989e-01 -7.91310847e-01 -7.78115034e-01 -1.44618976e+00
-8.47287595e-01 4.37662274e-01 7.03685105e-01 1.25555956e+00
2.73216754e-01 -8.26372683e-01 -5.00563048e-02 2.74662822e-01
1.76787600e-02 1.36551112e-01 -8.02646160e-01 -8.52139175e-01
5.44018090e-01 2.35273421e-01 4.69472200e-01 5.89973986e-01
6.09895214e-02 1.85997337e-01 -2.98250526e-01 7.03638971e-01
7.54169822e-01 -9.05696955e-03 9.71723616e-01 -1.66881680e+00
-1.12071089e-01 -7.46298552e-01 -6.20193899e-01 -2.73417771e-01
-4.71217856e-02 -8.10262799e-01 -1.06761014e-04 -1.62043643e+00
6.61031663e-01 -4.88330871e-01 3.55371594e-01 7.54776120e-01
-2.77389705e-01 -1.09755486e-01 3.59636247e-01 6.23281896e-01
4.53169316e-01 6.77076578e-01 5.20340562e-01 -6.27441332e-02
-4.62290347e-01 -2.58428037e-01 -7.76912153e-01 2.78237373e-01
5.17765045e-01 -4.76136059e-01 -7.81574398e-02 -2.45183125e-01
4.74140085e-02 1.57623485e-01 8.39916170e-01 -1.07770467e+00
-5.05287610e-02 -9.43699360e-01 3.40797752e-01 -8.25789690e-01
5.41127175e-02 -1.01143718e+00 1.02404261e+00 1.00577796e+00
-1.37661204e-01 -1.84413977e-02 8.02370846e-01 5.79007328e-01
2.71853060e-01 -1.31855886e-02 6.50128067e-01 -2.71152526e-01
-2.35206634e-01 2.81364977e-01 -1.10141838e+00 -4.58361179e-01
5.69526553e-01 -1.18192740e-01 -5.84746540e-01 -2.80011177e-01
-6.26523197e-01 5.07873356e-01 1.28806829e+00 1.98605642e-01
-7.57329389e-02 -9.72526550e-01 -8.88472795e-01 -7.01019540e-02
1.35905668e-01 -3.85734439e-01 3.27601224e-01 7.34590530e-01
-9.46278632e-01 1.81896910e-01 -7.06634402e-01 -8.25830579e-01
-1.05888736e+00 5.08328617e-01 6.77159488e-01 1.22061171e-01
-2.07627535e-01 9.35511515e-02 7.24559054e-02 -9.35869157e-01
-3.66861492e-01 -1.63086027e-01 1.73606768e-01 5.65681934e-01
4.54508573e-01 2.41753340e-01 -6.79533303e-01 -9.68102753e-01
-6.17786407e-01 1.28018886e-01 2.59838074e-01 -4.71896678e-01
1.72869563e+00 -4.35536742e-01 -3.88937980e-01 1.08588707e+00
5.04466712e-01 -1.55990154e-01 -1.44800413e+00 1.47143871e-01
-2.29939058e-01 -6.93625748e-01 2.07330748e-01 -8.62618804e-01
-5.43728292e-01 5.35367787e-01 6.09796286e-01 3.31615984e-01
6.89577162e-01 -3.45221817e-01 -2.16300085e-01 -6.26986474e-02
5.97841144e-01 -6.50092006e-01 -6.90159202e-01 -1.88381687e-01
1.07364273e+00 -7.75579929e-01 4.47599262e-01 1.58471510e-01
-1.37941450e-01 5.56720018e-01 7.78504074e-01 -1.18519505e-02
9.03075099e-01 1.87136710e-01 -5.34980059e-01 1.44747421e-01
-9.27940667e-01 -6.21097147e-01 -4.28831875e-01 6.69045985e-01
3.52370739e-01 7.96862900e-01 -8.09029818e-01 2.26945773e-01
-3.51867348e-01 -2.39932239e-01 9.02291298e-01 6.96206927e-01
-9.47763205e-01 -8.22809219e-01 -6.96615696e-01 9.68748868e-01
1.42659977e-01 -3.95652831e-01 -7.51066685e-01 8.54225039e-01
-5.41969612e-02 1.05983210e+00 3.42323989e-01 -3.27286124e-01
2.56081130e-02 -1.11301698e-01 6.57368824e-02 -8.43707979e-01
-9.37537909e-01 2.41346821e-01 5.59362723e-03 7.77199790e-02
-7.05065310e-01 -1.49436450e+00 -6.57203913e-01 -1.22124755e+00
-6.59764051e-01 2.61785672e-03 7.37693369e-01 3.42930138e-01
3.75849336e-01 9.36299264e-02 5.16923666e-01 -1.01383781e+00
-2.78805524e-01 -1.58025801e+00 -8.30513954e-01 -1.12731241e-01
1.83885962e-01 -7.70879090e-01 -7.98802733e-01 -4.73452270e-01] | [9.356217384338379, -1.4632439613342285] |
c25017e4-ec7c-4ced-99a5-260a2d3eeeae | iw-net-an-automatic-and-minimalistic | 1811.12789 | null | http://arxiv.org/abs/1811.12789v1 | http://arxiv.org/pdf/1811.12789v1.pdf | iW-Net: an automatic and minimalistic interactive lung nodule segmentation deep network | We propose iW-Net, a deep learning model that allows for both automatic and
interactive segmentation of lung nodules in computed tomography images. iW-Net
is composed of two blocks: the first one provides an automatic segmentation and
the second one allows to correct it by analyzing 2 points introduced by the
user in the nodule's boundary. For this purpose, a physics inspired weight map
that takes the user input into account is proposed, which is used both as a
feature map and in the system's loss function. Our approach is extensively
evaluated on the public LIDC-IDRI dataset, where we achieve a state-of-the-art
performance of 0.55 intersection over union vs the 0.59 inter-observer
agreement. Also, we show that iW-Net allows to correct the segmentation of
small nodules, essential for proper patient referral decision, as well as
improve the segmentation of the challenging non-solid nodules and thus may be
an important tool for increasing the early diagnosis of lung cancer. | ['Aurélio Campilho', 'António Cunha', 'Teresa Araújo', 'Isabel Ramos', 'Bram van Ginneken', 'Guilherme Aresta', 'Colin Jacobs'] | 2018-11-30 | null | null | null | null | ['lung-nodule-segmentation'] | ['medical'] | [-6.27833009e-02 5.80013037e-01 -4.83177677e-02 -2.58699119e-01
-9.11015451e-01 -2.76634932e-01 2.87345350e-01 3.30684274e-01
-6.06525064e-01 3.50293517e-01 -1.63986087e-01 -4.77958500e-01
-1.48460403e-01 -7.71036685e-01 -6.12415493e-01 -9.21220422e-01
1.76463619e-01 1.11085808e+00 9.35246229e-01 1.44881025e-01
-3.44976276e-01 7.51831353e-01 -9.84127641e-01 2.81546474e-01
7.98493505e-01 1.00349939e+00 2.95008153e-01 4.70869303e-01
-9.97886658e-02 6.37401104e-01 -7.10995216e-03 -2.56310105e-01
1.91443905e-01 -2.82125652e-01 -1.04259825e+00 -2.89293472e-02
2.15857387e-01 -2.08819270e-01 -6.25127852e-02 8.20182562e-01
3.94908905e-01 -1.37121454e-01 8.70581865e-01 -5.76581419e-01
-8.19568709e-02 6.21404469e-01 -3.02393734e-01 6.51899427e-02
-3.36166620e-01 3.80805939e-01 9.86627579e-01 -6.68444872e-01
5.19682646e-01 5.80636442e-01 8.55965257e-01 6.73091114e-01
-9.57489371e-01 -2.49883458e-01 -4.21693563e-01 2.09918022e-01
-1.17516220e+00 2.31437728e-01 2.97487617e-01 -6.27242684e-01
4.67217565e-01 3.54410857e-01 1.01658094e+00 5.43427289e-01
3.21253091e-01 6.41606152e-01 5.89466691e-01 -3.91264498e-01
1.14078119e-01 2.76611567e-01 5.65393679e-02 1.03858387e+00
3.04575205e-01 1.53466091e-02 3.52993459e-01 -1.42310739e-01
9.90471482e-01 -7.85322264e-02 -3.13704312e-01 -5.90480328e-01
-1.20156527e+00 7.37792075e-01 1.09828913e+00 5.85374892e-01
-5.43532610e-01 2.98219770e-01 2.48625025e-01 -3.87878269e-01
2.88459808e-01 4.37343776e-01 -2.26721779e-01 2.65460104e-01
-9.59227324e-01 2.58233544e-04 6.73840702e-01 1.81554526e-01
2.68515378e-01 -4.02609795e-01 -6.82834566e-01 5.61292470e-01
2.86652386e-01 2.73024321e-01 5.52810490e-01 -7.61950076e-01
-4.57749590e-02 7.82306373e-01 1.29750848e-01 -2.90324330e-01
-8.73150408e-01 -5.78239739e-01 -9.04148459e-01 2.66485989e-01
6.03533387e-01 2.15695053e-02 -1.33382511e+00 1.32290423e+00
5.03897071e-01 2.01403275e-01 -3.60726953e-01 9.73829985e-01
1.06805980e+00 1.46097854e-01 1.23426326e-01 4.68115099e-02
1.64253962e+00 -1.07504451e+00 -3.29115748e-01 7.02801719e-03
6.67269886e-01 -5.06016195e-01 1.04231513e+00 1.70096755e-01
-1.25067568e+00 -4.62350339e-01 -6.57598853e-01 1.06413767e-01
-3.19479816e-02 6.69721305e-01 2.29610592e-01 4.82309103e-01
-1.22249615e+00 6.78284228e-01 -1.19785523e+00 -3.83037925e-01
6.12164438e-01 7.61873305e-01 -6.02000132e-02 3.07150960e-01
-1.05884600e+00 1.00207555e+00 4.46767509e-01 -9.18846577e-02
-6.15976691e-01 -1.04198813e+00 -4.28133935e-01 3.06447953e-01
5.35960436e-01 -1.03950655e+00 1.46128368e+00 -6.48703337e-01
-1.36532819e+00 1.26722550e+00 1.30139023e-01 -6.86175108e-01
1.03330028e+00 1.47163048e-01 -1.64045568e-03 3.85101557e-01
3.63532752e-02 7.94820249e-01 6.40772700e-01 -1.04724205e+00
-5.68310142e-01 -2.12495148e-01 -2.01141685e-01 9.26528871e-02
5.17787077e-02 -4.64352489e-01 -9.37095046e-01 -3.34126443e-01
4.32065725e-02 -1.14087498e+00 -5.66024840e-01 3.72179180e-01
-6.71211541e-01 -2.91874647e-01 8.53275299e-01 -6.82518601e-01
8.09882879e-01 -2.06208587e+00 -6.46350067e-03 5.77297270e-01
5.74189007e-01 5.28363943e-01 3.31444472e-01 -2.69833535e-01
-1.16676547e-01 1.98439643e-01 -3.81560296e-01 -1.04625337e-01
-3.17230105e-01 7.79772028e-02 4.03813094e-01 3.05775791e-01
2.74085790e-01 1.21624875e+00 -5.74875891e-01 -5.97151160e-01
6.12343192e-01 5.94130635e-01 -4.54568624e-01 2.71735221e-01
-4.36715096e-01 7.53481269e-01 -5.31981587e-01 1.22093260e-01
5.53038120e-01 -8.10401738e-01 1.35512844e-01 -3.63651276e-01
-9.15333778e-02 1.32653177e-01 -8.92132342e-01 1.26050091e+00
-3.95033032e-01 2.18392849e-01 8.61873850e-02 -5.07090688e-01
4.92065310e-01 5.73869348e-01 9.62898254e-01 -3.00522655e-01
5.09980500e-01 4.05983001e-01 4.25047308e-01 -5.01545429e-01
-1.11044504e-01 -2.82589495e-01 4.41238016e-01 5.20706713e-01
-1.94936633e-01 -4.17450666e-01 8.07994902e-02 2.44684964e-02
1.07422566e+00 -3.09059918e-01 3.98263246e-01 -4.09499198e-01
9.20904994e-01 6.45232424e-02 8.28399882e-02 6.39241934e-01
-1.27114221e-01 8.55876029e-01 5.92247605e-01 -4.35810536e-01
-8.29488218e-01 -9.89536762e-01 -3.65105599e-01 3.54733437e-01
1.33300394e-01 1.81487024e-01 -9.91286993e-01 -1.06491995e+00
-3.89730856e-02 6.54272974e-01 -9.80870843e-01 -8.64589140e-02
-5.78308642e-01 -6.65654063e-01 9.85321775e-02 7.46583223e-01
4.24504995e-01 -1.14762616e+00 -7.64238179e-01 2.10852474e-01
-3.41108620e-01 -9.90414560e-01 -2.64129251e-01 4.03311998e-01
-9.31146562e-01 -1.26081038e+00 -8.96943033e-01 -5.96673429e-01
7.97954023e-01 4.80728522e-02 1.28348327e+00 5.22944689e-01
-6.24263167e-01 3.14073294e-01 -2.86689736e-02 -2.14157388e-01
-7.07702935e-01 4.02105361e-01 -4.57260579e-01 -2.94843316e-01
3.76831144e-02 -1.64701745e-01 -8.12343895e-01 5.04554689e-01
-7.80049205e-01 1.06679536e-01 7.45136023e-01 6.55718684e-01
9.42480385e-01 -1.25469342e-01 8.77285972e-02 -1.17153347e+00
1.66123912e-01 -3.48751307e-01 -6.81640625e-01 3.14753979e-01
-3.92300189e-01 1.37224659e-01 6.32818282e-01 -2.73473188e-02
-8.60614717e-01 3.77085507e-01 -5.52713096e-01 -2.44825423e-01
-1.10717796e-01 2.15777099e-01 3.80034804e-01 -4.82702792e-01
6.95004344e-01 -1.26229972e-01 9.02560651e-02 -1.85032353e-01
6.84440434e-02 2.10289046e-01 4.27915335e-01 -4.29969989e-02
8.52531135e-01 5.60345232e-01 3.46696496e-01 -5.93376935e-01
-9.79826748e-01 -8.36169183e-01 -9.22215581e-01 -3.32195818e-01
1.24981809e+00 -4.84952778e-01 -5.61466694e-01 6.86061606e-02
-9.60406661e-01 -4.30198789e-01 -7.59196162e-01 4.04084086e-01
-5.20322502e-01 9.22399685e-02 -6.07203424e-01 -1.46213099e-01
-6.33785307e-01 -1.57664037e+00 1.07207787e+00 3.06061417e-01
-4.91061322e-02 -1.19821012e+00 1.16405465e-01 2.95447528e-01
4.29618686e-01 1.06109038e-01 1.17179501e+00 -1.02925456e+00
-7.14447796e-01 -4.45728153e-01 -5.21628141e-01 2.42129132e-01
1.28497317e-01 1.91819251e-01 -9.28264439e-01 -1.46121934e-01
-9.51308236e-02 -2.03340262e-01 9.34522986e-01 8.82157743e-01
1.56730795e+00 3.81708443e-01 -6.27384484e-01 5.70412815e-01
1.37345600e+00 -7.98728913e-02 5.08268297e-01 6.70232326e-02
6.80607915e-01 1.19695939e-01 3.25984508e-01 1.48036942e-01
4.26690206e-02 6.69071376e-01 7.88948238e-01 -5.09818733e-01
-2.71956146e-01 2.39565313e-01 -4.33503509e-01 3.81506234e-01
-3.37895662e-01 -1.33504376e-01 -1.18396771e+00 2.74470150e-01
-1.67850530e+00 -4.59772348e-01 -6.98663354e-01 2.14832973e+00
6.38547659e-01 2.82842994e-01 1.49679273e-01 -1.24742381e-01
4.68598247e-01 -1.15522124e-01 -5.40629506e-01 4.79669571e-02
6.47423148e-01 4.91312712e-01 5.53987920e-01 4.89007175e-01
-1.24757612e+00 6.26802146e-01 5.93857765e+00 8.23831022e-01
-1.35509276e+00 2.93475270e-01 7.00201511e-01 2.38519669e-01
-8.01051930e-02 -5.20667970e-01 -5.39716542e-01 1.52630344e-01
5.03363371e-01 1.30252793e-01 -1.23815514e-01 7.53609478e-01
2.53651053e-01 -3.80222589e-01 -1.12263656e+00 5.93627214e-01
-2.26472735e-01 -1.52667344e+00 -6.00401834e-02 1.02525212e-01
5.71366012e-01 2.12144151e-01 -1.22210085e-01 1.20874159e-01
1.30101457e-01 -8.90486002e-01 2.70328611e-01 3.79383087e-01
9.66193378e-01 -5.67752063e-01 1.08400702e+00 3.83059919e-01
-1.14551222e+00 3.96686286e-01 -9.83356088e-02 8.19521844e-01
7.90515915e-02 8.06923509e-01 -1.50544155e+00 4.63170588e-01
4.81324464e-01 3.41069251e-01 -7.33988822e-01 1.44899631e+00
-2.16617405e-01 7.24413276e-01 -4.37930644e-01 1.21693641e-01
4.70003903e-01 -1.40247652e-02 5.00515163e-01 1.10715866e+00
2.56017953e-01 5.39659783e-02 2.19138488e-01 9.28072035e-01
-2.90374726e-01 2.42092803e-01 -3.38039920e-02 4.04657900e-01
-7.03201769e-03 1.62746310e+00 -1.20509410e+00 -3.38836074e-01
3.84367816e-02 8.66766989e-01 1.37797162e-01 -9.12678540e-02
-1.16684067e+00 8.29527900e-03 -1.30653670e-02 5.93491733e-01
3.75197530e-01 2.80694127e-01 -2.84502596e-01 -7.37174451e-01
-1.31887496e-01 -2.63042539e-01 2.72151560e-01 -6.86594486e-01
-8.89201820e-01 7.58509576e-01 -2.87594199e-01 -1.06981587e+00
-2.87081212e-01 -8.38068247e-01 -9.64979231e-01 7.34790325e-01
-1.38820767e+00 -1.11423051e+00 -7.29776323e-01 2.83767343e-01
2.89788067e-01 2.40448400e-01 6.88916385e-01 1.65775001e-01
-3.85286212e-01 3.20187688e-01 4.72533889e-02 9.63757336e-02
4.94355470e-01 -1.57675409e+00 2.81061530e-01 4.54087138e-01
-1.41929105e-01 -9.21421051e-02 3.48840624e-01 -5.06055117e-01
-6.43619597e-01 -1.34857357e+00 6.41666114e-01 -4.94759738e-01
6.47966385e-01 1.69547349e-01 -1.06986904e+00 4.96596426e-01
-5.62943257e-02 3.88028443e-01 5.81932127e-01 -4.56331015e-01
1.79298684e-01 1.18945502e-02 -1.35102093e+00 3.72860193e-01
4.39901888e-01 9.56454780e-03 -2.11293101e-01 7.21169591e-01
5.26706934e-01 -7.69172311e-01 -7.81719148e-01 6.21625662e-01
3.51206183e-01 -1.24193120e+00 1.10119951e+00 -2.10497096e-01
3.38024914e-01 -8.82078614e-03 6.07692838e-01 -1.05321944e+00
-6.17293358e-01 1.86329275e-01 2.08492890e-01 3.36978495e-01
6.24632955e-01 -3.56934518e-01 9.69968736e-01 5.26644707e-01
-1.80222392e-01 -1.16069460e+00 -9.20209825e-01 -4.18661028e-01
7.95389786e-02 -3.87906343e-01 2.74458844e-02 5.13739049e-01
-7.13912249e-01 -2.01841697e-01 3.50217938e-01 1.25468537e-01
5.39741278e-01 -1.49915084e-01 2.56402373e-01 -1.44822025e+00
-2.85919338e-01 -6.75902724e-01 -1.47384763e-01 -7.66931891e-01
-1.10083789e-01 -1.05640674e+00 1.54901385e-01 -1.92645299e+00
4.43532795e-01 -4.73412544e-01 -2.84982026e-01 4.25270081e-01
-2.41438910e-01 3.77818912e-01 6.16255812e-02 4.19335783e-01
-4.50607657e-01 1.40340075e-01 1.57639730e+00 4.79207709e-02
-1.88218638e-01 7.14326501e-01 -3.41602355e-01 9.56438303e-01
7.27669001e-01 -5.11896968e-01 -1.04396038e-01 -3.67375873e-02
-1.64385855e-01 1.32359356e-01 7.38528013e-01 -1.25174475e+00
6.93884641e-02 -1.42170186e-03 4.00900930e-01 -7.42245793e-01
1.50807232e-01 -1.11746156e+00 9.59724784e-02 1.14587712e+00
-2.81972706e-01 -6.27819240e-01 2.54900485e-01 4.39978570e-01
-7.58790076e-02 -6.14030063e-01 1.31408298e+00 -3.75703156e-01
-1.55540317e-01 5.17492294e-01 -3.56528789e-01 -5.43508353e-03
1.19176042e+00 4.26878743e-02 1.33736119e-01 -1.32988036e-01
-1.17186117e+00 2.82563299e-01 2.20582396e-01 -2.11858287e-01
2.55106211e-01 -1.11784673e+00 -6.19134426e-01 5.68281747e-02
-5.73287047e-02 4.25153643e-01 2.29109108e-01 1.37053347e+00
-9.77358103e-01 5.09308934e-01 2.45428279e-01 -1.01778984e+00
-1.36192477e+00 9.99014825e-02 1.05717480e+00 -9.14120018e-01
-7.13092089e-01 1.02179754e+00 2.62711287e-01 -3.63219619e-01
2.67786860e-01 -8.54158461e-01 -2.38644794e-01 -1.56366497e-01
1.73821598e-01 9.82985497e-02 5.32845795e-01 -3.02600354e-01
-4.72279191e-01 5.78852415e-01 -2.34664366e-01 1.43966630e-01
1.05983531e+00 3.26621562e-01 -1.27225742e-01 3.34462285e-01
7.93120444e-01 -1.92519039e-01 -1.03004062e+00 -2.43806854e-01
-5.51930442e-02 4.53124605e-02 2.93236822e-01 -1.10252810e+00
-1.27717090e+00 9.36452508e-01 9.25617516e-01 3.95821124e-01
8.87092590e-01 2.51764953e-01 8.73728514e-01 3.89881432e-01
-6.17762767e-02 -6.71670675e-01 1.34108499e-01 4.48231190e-01
6.41445160e-01 -1.49367583e+00 -2.78524067e-02 -7.22501576e-01
-5.89016438e-01 1.36235952e+00 6.45692766e-01 -1.26831457e-01
8.09113562e-01 2.45592773e-01 1.22889593e-01 -3.99488688e-01
-3.95066351e-01 -4.02487248e-01 7.88745940e-01 1.86073452e-01
6.60752177e-01 2.27299511e-01 -2.11706087e-01 5.13559699e-01
2.18190461e-01 2.10340887e-01 2.05172837e-01 6.00557506e-01
-7.68866301e-01 -9.70583320e-01 -2.85488099e-01 7.83138871e-01
-4.06543434e-01 5.89082800e-02 -4.29317296e-01 1.03067923e+00
2.44328082e-01 1.17797740e-01 3.86158340e-02 -2.30045561e-02
3.59229803e-01 3.60877812e-02 4.59164500e-01 -8.42349231e-01
-1.06591070e+00 1.81196481e-01 -3.00534785e-01 -4.75432813e-01
-2.62809515e-01 -3.84468496e-01 -1.68652010e+00 1.74611747e-01
-3.53613019e-01 1.74441874e-01 5.02012312e-01 1.01569808e+00
-6.63814768e-02 1.01739192e+00 5.08451283e-01 -7.93705285e-01
-5.89687169e-01 -6.79366171e-01 -3.34873468e-01 3.49710733e-01
1.23735711e-01 -5.39617836e-01 -3.53487253e-01 -1.36622548e-01] | [15.372638702392578, -2.1379261016845703] |
bcfab388-f6f1-4593-9933-5991daf6e337 | synthref-generation-of-synthetic-referring | 2106.04403 | null | https://arxiv.org/abs/2106.04403v2 | https://arxiv.org/pdf/2106.04403v2.pdf | SynthRef: Generation of Synthetic Referring Expressions for Object Segmentation | Recent advances in deep learning have brought significant progress in visual grounding tasks such as language-guided video object segmentation. However, collecting large datasets for these tasks is expensive in terms of annotation time, which represents a bottleneck. To this end, we propose a novel method, namely SynthRef, for generating synthetic referring expressions for target objects in an image (or video frame), and we also present and disseminate the first large-scale dataset with synthetic referring expressions for video object segmentation. Our experiments demonstrate that by training with our synthetic referring expressions one can improve the ability of a model to generalize across different datasets, without any additional annotation cost. Moreover, our formulation allows its application to any object detection or segmentation dataset. | ['Xavier Giro-i-Nieto', 'Carina Silberer', 'Miriam Bellver', 'Carles Ventura', 'Ioannis Kazakos'] | 2021-06-08 | null | null | null | null | ['referring-expression-segmentation'] | ['computer-vision'] | [ 4.49147731e-01 1.38530061e-01 -2.95978993e-01 -3.80816340e-01
-9.35972631e-01 -6.98951781e-01 3.91342074e-01 4.13762219e-03
-2.79999435e-01 7.17036784e-01 -1.19271770e-01 -2.95228988e-01
2.57311910e-01 -6.62583232e-01 -1.08469701e+00 -2.92474359e-01
2.93376625e-01 3.38427871e-01 4.76787806e-01 -3.30840237e-02
-1.56245939e-02 4.54266787e-01 -1.24076045e+00 3.72422576e-01
8.09114099e-01 9.42924142e-01 2.17242032e-01 5.24462163e-01
-2.21415550e-01 1.06742346e+00 -7.20025778e-01 -6.98189497e-01
1.54033944e-01 -5.10267615e-01 -1.27159250e+00 5.05785108e-01
6.49560630e-01 -3.76369804e-01 -1.71247289e-01 1.15581822e+00
1.61128134e-01 3.03616345e-01 4.51182544e-01 -1.69019604e+00
-7.44475067e-01 7.72320509e-01 -5.26555419e-01 2.71958429e-02
3.82972270e-01 1.74451873e-01 8.81633401e-01 -5.98293364e-01
8.66777718e-01 1.24728906e+00 4.66060817e-01 8.76957715e-01
-1.17195892e+00 -4.02304292e-01 5.30229449e-01 4.77526970e-02
-1.32293749e+00 -2.82544196e-01 7.66225100e-01 -5.80624223e-01
6.14066422e-01 7.31248707e-02 5.67617178e-01 1.08247626e+00
-6.05163395e-01 1.32688522e+00 6.72466815e-01 -2.55391419e-01
8.38846192e-02 -2.09459946e-01 8.28534588e-02 7.89353549e-01
1.42740279e-01 -4.80481476e-01 -2.14813441e-01 1.99013174e-01
8.45300078e-01 -2.85078496e-01 -3.78835738e-01 -4.00519013e-01
-1.22659922e+00 6.59601092e-01 4.57899898e-01 1.54198304e-01
-1.64872125e-01 5.59307575e-01 4.59831923e-01 -9.86831635e-03
5.37761390e-01 5.18736660e-01 -4.29661453e-01 -1.70420870e-01
-9.38950956e-01 3.87941390e-01 5.87329805e-01 1.37324095e+00
7.50982463e-01 1.23896845e-01 -2.95546055e-01 6.46999657e-01
1.30592668e-02 3.24374914e-01 5.89479171e-02 -1.28693664e+00
5.25545001e-01 6.01221144e-01 1.13891944e-01 -9.00627792e-01
-3.33341032e-01 -6.08180128e-02 -5.93245924e-01 -3.13640565e-01
5.25754511e-01 -2.29334384e-01 -9.73221958e-01 1.94003606e+00
3.22525293e-01 3.52679461e-01 1.05232708e-01 9.58229005e-01
1.06130052e+00 6.16955400e-01 3.98970395e-01 -1.13838524e-01
1.18807626e+00 -1.12976730e+00 -6.33502007e-01 -3.93063515e-01
9.30751920e-01 -5.34817815e-01 1.29910278e+00 2.62868583e-01
-1.05272794e+00 -5.34220397e-01 -7.11150289e-01 -3.69981617e-01
-2.55753398e-01 8.80427882e-02 7.32565343e-01 3.56749415e-01
-9.85941172e-01 2.27730006e-01 -7.09921062e-01 -4.26374853e-01
8.66640925e-01 2.89532900e-01 -2.59957045e-01 3.71927172e-02
-8.95640433e-01 5.06656528e-01 6.90328479e-01 5.84025159e-02
-9.11632657e-01 -5.55093229e-01 -1.07801473e+00 -1.54927373e-01
8.38430643e-01 -7.16166496e-01 1.44506097e+00 -1.48311496e+00
-1.14971316e+00 1.01582515e+00 -2.83911347e-01 -5.82275391e-01
5.86116970e-01 -3.79450977e-01 -4.94616404e-02 3.49114150e-01
3.06678832e-01 1.36149728e+00 7.69079566e-01 -1.44858456e+00
-5.02752066e-01 -1.63056269e-01 7.33296812e-01 -4.79055829e-02
-2.73631532e-02 3.20325643e-01 -1.00165415e+00 -7.62394130e-01
3.88443470e-02 -1.00092769e+00 -3.17022324e-01 1.62913024e-01
-6.93194449e-01 -4.94762450e-01 8.65432858e-01 -4.61753488e-01
8.91812742e-01 -2.16152501e+00 3.58580530e-01 -1.54358014e-01
1.97946712e-01 3.05913627e-01 -2.25184530e-01 -3.72149376e-03
2.37245336e-02 5.79180241e-01 -4.55500901e-01 -3.69557887e-01
-1.40015945e-01 5.14373124e-01 -4.28482324e-01 8.81389752e-02
6.10184371e-01 1.24673057e+00 -1.03339040e+00 -8.59930217e-01
-6.72277668e-03 2.47258410e-01 -5.88089108e-01 2.59188771e-01
-8.63662183e-01 6.90705419e-01 -6.81787491e-01 6.00446045e-01
4.50923830e-01 -4.40373629e-01 1.87086016e-02 -1.75173268e-01
2.16738671e-01 -2.18210742e-02 -8.03561866e-01 1.73700893e+00
-4.65379804e-01 8.19754660e-01 -1.68323442e-01 -1.21356320e+00
6.36385143e-01 1.51917458e-01 5.86520791e-01 -3.99103850e-01
1.35024890e-01 -6.66501448e-02 -2.71879435e-01 -7.34393299e-01
5.06296813e-01 1.60937104e-02 -1.84011638e-01 4.42494571e-01
3.23765762e-02 -3.66379887e-01 6.64879024e-01 4.37754720e-01
5.11246562e-01 5.70524871e-01 -3.74204740e-02 -7.85544589e-02
4.55533206e-01 2.53931582e-01 5.79482913e-01 6.16288364e-01
-2.04830915e-01 5.86183548e-01 8.42637241e-01 -4.35149670e-01
-9.16723192e-01 -8.22643101e-01 1.35368913e-01 1.19006622e+00
2.94203073e-01 -3.78544092e-01 -1.12215614e+00 -9.05389190e-01
-2.95766115e-01 6.05844021e-01 -3.43371779e-01 1.37923524e-01
-9.07820165e-01 -4.29247051e-01 6.31759584e-01 8.84808123e-01
5.62630057e-01 -1.14446664e+00 -6.10675037e-01 -1.46297961e-02
-4.58381712e-01 -1.88069022e+00 -4.89056557e-01 -1.72974467e-01
-8.04893851e-01 -1.24138868e+00 -7.46104896e-01 -1.03412521e+00
7.96425998e-01 1.65615588e-01 1.25918591e+00 4.36111316e-02
-8.26539323e-02 4.27767426e-01 -5.42773962e-01 -2.97702879e-01
-3.80175143e-01 1.01668097e-01 -2.34916583e-01 -4.83519994e-02
1.56595230e-01 -7.17206486e-03 -2.43194997e-01 2.54573286e-01
-1.12760735e+00 3.77912849e-01 1.81836501e-01 4.86717492e-01
8.21441591e-01 -2.83380747e-01 5.49062729e-01 -9.15683091e-01
3.97159129e-01 -2.26707086e-01 -8.30241680e-01 3.13375264e-01
1.54840529e-01 7.66263530e-02 5.07124782e-01 -3.97559166e-01
-9.45090473e-01 1.41108260e-01 -1.09234221e-01 -6.08214021e-01
-1.32322028e-01 5.07861376e-01 -3.04668665e-01 -1.08680886e-03
5.17076910e-01 7.11979419e-02 -4.05967444e-01 -2.24636942e-01
8.27261746e-01 3.61228615e-01 7.56882787e-01 -9.39183950e-01
5.76385915e-01 4.51325238e-01 1.00801811e-02 -7.79735923e-01
-1.29981565e+00 -2.75651664e-01 -8.78195167e-01 -1.31995574e-01
1.27269435e+00 -9.20912325e-01 -5.70359826e-01 2.78688192e-01
-1.49781132e+00 -6.73954010e-01 -2.46379241e-01 2.01018021e-01
-8.74243319e-01 3.89220536e-01 -4.78641272e-01 -5.52691460e-01
-5.80449067e-02 -1.41301310e+00 1.32409549e+00 2.41264850e-01
-1.73846856e-01 -9.24891829e-01 -5.49877405e-01 5.25768936e-01
-2.21023206e-02 4.41593558e-01 8.32384229e-01 -5.16386807e-01
-1.00241756e+00 9.49716493e-02 -5.57810724e-01 3.52240354e-01
2.60331091e-02 1.58579201e-01 -7.86458910e-01 4.01548781e-02
-3.08885813e-01 -7.55686700e-01 7.43236542e-01 3.55371773e-01
1.79559278e+00 -2.79720515e-01 -3.21161985e-01 7.38918960e-01
1.10487318e+00 1.55635998e-01 4.40540433e-01 1.67068765e-01
1.04835010e+00 7.13509202e-01 7.96614468e-01 1.49345905e-01
3.96192610e-01 8.27597380e-01 3.01262021e-01 -2.85888731e-01
-2.71721005e-01 -2.66120762e-01 1.43793955e-01 3.06384653e-01
-2.08361167e-02 -5.69801569e-01 -1.05870318e+00 7.49827206e-01
-1.91499484e+00 -8.42304826e-01 -1.02192320e-01 1.80079007e+00
8.60981166e-01 9.89944953e-03 2.23362043e-01 -1.41777873e-01
7.26994574e-01 1.73549265e-01 -6.65960550e-01 -1.49958283e-01
-8.09824318e-02 1.36729479e-01 2.50173748e-01 2.75837958e-01
-1.37693942e+00 1.54485404e+00 6.36421871e+00 6.71056271e-01
-1.12721050e+00 -1.90631337e-02 7.66535342e-01 1.59877256e-01
-2.31810182e-01 -2.90308204e-02 -6.71790838e-01 1.24115370e-01
5.36725581e-01 -2.14376405e-01 2.09430277e-01 9.55382049e-01
2.83699036e-01 -3.37780714e-02 -1.35817087e+00 1.14875531e+00
1.90141648e-01 -1.36107755e+00 4.29383218e-01 -4.06897902e-01
9.36446488e-01 -5.56531101e-02 -8.87831748e-02 1.55164227e-01
2.93571621e-01 -8.77598226e-01 8.86321783e-01 1.63105726e-01
7.14336514e-01 -6.00919187e-01 6.08223975e-01 1.70874938e-01
-1.07022548e+00 2.97376841e-01 -8.14835802e-02 1.59845203e-01
3.97212207e-01 2.26489633e-01 -7.68648565e-01 4.23446089e-01
4.30574566e-01 7.82064497e-01 -6.05879903e-01 8.60236406e-01
-4.35236424e-01 5.24993002e-01 -2.04333305e-01 5.99316880e-02
4.25526232e-01 -1.30194053e-01 2.41245240e-01 1.10186088e+00
7.03690872e-02 1.14446223e-01 5.79084933e-01 1.16747057e+00
-4.76146519e-01 6.53269663e-02 -6.54428124e-01 -3.34098071e-01
2.36735001e-01 1.05589092e+00 -9.70822573e-01 -5.90969086e-01
-4.13004845e-01 9.58326459e-01 2.21124843e-01 6.19223177e-01
-1.23485208e+00 -2.03393921e-01 5.30911863e-01 -5.43710869e-03
2.71592468e-01 -3.66350085e-01 -1.59049481e-01 -1.27778625e+00
1.82397991e-01 -8.71453643e-01 2.35417813e-01 -9.82926130e-01
-9.56815004e-01 5.90840638e-01 3.25377405e-01 -1.07781065e+00
-3.25508982e-01 -7.19445467e-01 -1.57965198e-01 4.43789452e-01
-1.32148874e+00 -1.38027406e+00 -4.19287384e-01 6.97615027e-01
8.32669258e-01 1.26807660e-01 3.34319144e-01 3.09933484e-01
-7.22935617e-01 4.81485099e-01 -5.73747933e-01 5.61422706e-01
4.15339619e-01 -1.01121879e+00 4.19877827e-01 9.99739349e-01
4.84056890e-01 5.32496393e-01 5.59140027e-01 -5.42255044e-01
-1.20611584e+00 -1.49767733e+00 4.64337319e-01 -4.38796312e-01
6.73818886e-01 -3.92167389e-01 -8.36451828e-01 1.05509222e+00
6.66769594e-02 3.41261715e-01 4.72470343e-01 -5.08652925e-02
-2.89170891e-01 2.64082283e-01 -9.28765237e-01 8.74350309e-01
1.35954416e+00 -6.00882232e-01 -4.13056970e-01 7.64022231e-01
1.15967917e+00 -7.28614748e-01 -6.73241735e-01 5.50112784e-01
1.80992097e-01 -6.56332731e-01 8.97339165e-01 -9.60927188e-01
5.61183989e-01 -3.96656752e-01 -2.54995316e-01 -8.25878501e-01
3.20300817e-01 -5.96519113e-01 2.77334936e-02 1.37263560e+00
3.50896150e-01 -1.42681465e-01 8.66244912e-01 8.32516372e-01
-2.60872900e-01 -6.85921192e-01 -5.92386723e-01 -8.22474539e-01
-2.53373990e-03 -7.91887403e-01 4.52377439e-01 8.22481930e-01
-3.54379922e-01 2.56361961e-01 -2.43130654e-01 8.75702575e-02
3.59541386e-01 2.70154685e-01 1.13408625e+00 -8.60216498e-01
-2.37666711e-01 -3.82653654e-01 -5.15515924e-01 -1.42251468e+00
5.88134468e-01 -8.14095438e-01 1.88480139e-01 -1.61169791e+00
1.83303103e-01 -5.12046039e-01 9.37899426e-02 6.89039111e-01
-2.79585540e-01 5.11510789e-01 4.93492395e-01 1.14169009e-01
-1.04115033e+00 4.45651352e-01 1.46359038e+00 -3.41971487e-01
-1.34426117e-01 -1.68066412e-01 -6.70561612e-01 1.08045995e+00
7.02213764e-01 -3.84711653e-01 -6.00554347e-01 -7.78640389e-01
5.38307801e-02 7.49879405e-02 6.13833606e-01 -8.21412146e-01
-1.58472821e-01 -3.81686985e-01 -1.30444085e-02 -4.93403196e-01
2.69972295e-01 -4.28749979e-01 -1.42639711e-01 9.65904668e-02
-4.86598372e-01 1.54735800e-02 3.19550604e-01 2.76432544e-01
-4.53611612e-01 -2.58691430e-01 7.46421337e-01 -1.00459330e-01
-1.09292245e+00 4.39047784e-01 -2.76680905e-02 5.88778257e-01
1.25906837e+00 -4.34422866e-02 -3.58113378e-01 -3.94464821e-01
-7.97492206e-01 4.09042031e-01 5.76310635e-01 5.60585022e-01
5.27209103e-01 -1.16953504e+00 -4.69863623e-01 -1.63090214e-01
2.14751303e-01 4.87966865e-01 4.12188396e-02 7.65542209e-01
-7.66442776e-01 3.76306742e-01 -7.66617209e-02 -8.79841864e-01
-1.15956867e+00 7.70208240e-01 2.91796893e-01 1.08408898e-01
-5.30085146e-01 1.11927032e+00 6.09255135e-01 -1.18811749e-01
1.09013185e-01 -6.14935637e-01 -9.21822116e-02 -5.61617501e-02
3.15460742e-01 -5.65152392e-02 -3.48267078e-01 -8.96141708e-01
-2.02569813e-01 8.13427210e-01 -1.02301100e-02 -2.32504178e-02
9.61112738e-01 -2.02173111e-03 -1.17346928e-01 4.48140740e-01
1.15688753e+00 -2.99824327e-01 -1.33482134e+00 -1.25735074e-01
1.98999420e-01 -4.61354434e-01 -1.58008188e-01 -2.71305680e-01
-1.24858010e+00 8.12028527e-01 9.03563499e-02 2.05729827e-01
1.20368981e+00 5.07485986e-01 8.74776065e-01 7.01662362e-01
4.92010176e-01 -8.31705570e-01 2.57237166e-01 4.72328037e-01
7.90347219e-01 -1.37857985e+00 -1.28951967e-01 -8.00025880e-01
-8.72223139e-01 9.12610292e-01 7.86159456e-01 8.06009620e-02
1.97354406e-01 1.47847638e-01 2.11906031e-01 -1.59752876e-01
-3.36713016e-01 -4.83417481e-01 1.91268414e-01 7.03942478e-01
5.81527174e-01 -8.65991563e-02 -2.62027055e-01 2.43588939e-01
2.72210985e-02 2.27579787e-01 5.27910769e-01 7.62877584e-01
-1.35129154e-01 -1.20518541e+00 -6.78485706e-02 1.37842789e-01
-6.30022049e-01 -6.06434047e-02 -5.25294125e-01 9.71606135e-01
1.95176855e-01 8.65691304e-01 1.14088230e-01 -3.62864602e-03
2.85664886e-01 -8.88946652e-02 5.55011511e-01 -7.48008251e-01
-1.00942478e-01 -1.16824564e-02 1.56948373e-01 -5.63751757e-01
-1.01661885e+00 -4.85363990e-01 -1.66893005e+00 1.44880891e-01
-1.37942582e-01 3.54341306e-02 2.70035416e-01 1.09950387e+00
1.97227761e-01 5.14634848e-01 1.20003879e-01 -8.27086747e-01
-2.19506398e-03 -5.90515792e-01 -1.01244569e-01 7.66020477e-01
1.82699427e-01 -6.85960531e-01 1.58072188e-01 6.81696415e-01] | [9.455619812011719, 0.5571905970573425] |
6fd79999-b694-4ce9-9593-7ead83406db3 | on-sir-type-epidemiological-models-and | 2210.11342 | null | https://arxiv.org/abs/2210.11342v1 | https://arxiv.org/pdf/2210.11342v1.pdf | On SIR-type epidemiological models and population heterogeneity effects | In this paper we elaborate on homogeneous and heterogeneous SIR-type epidemiological models. We find an unexpected correspondence between the epidemic trajectory of a transmissible disease in a homogeneous SIR-type model and radial null geodesics in the Schwarzschild spacetime. We also discuss modeling of population heterogeneity effects by considering both a one- and two-parameter gamma-distributed function for the initial susceptibility distribution, and deriving the associated herd immunity threshold. We furthermore describe how mitigation measures can be taken into account by model fitting. | ['Lucrezia Ravera', 'Silke Klemm'] | 2022-10-20 | null | null | null | null | ['type'] | ['speech'] | [-2.07238778e-01 2.03201979e-01 2.20607355e-01 -2.80441254e-01
2.78788656e-01 -2.83369511e-01 6.99894369e-01 -5.25936075e-02
-5.21579862e-01 8.97847295e-01 1.52380750e-01 -4.61071730e-01
-7.94690728e-01 -8.59165609e-01 -3.30771983e-01 -1.22914934e+00
-8.01555574e-01 6.58810139e-01 3.93902838e-01 -5.38542330e-01
-6.63683787e-02 3.95411313e-01 -8.68837059e-01 -4.86617923e-01
1.10253692e+00 1.39795393e-01 1.48770437e-01 9.96548593e-01
5.31449556e-01 4.19612348e-01 -9.31742609e-01 -4.98536438e-01
1.60073757e-01 -5.09381652e-01 -3.79412383e-01 -4.31536317e-01
-8.55852783e-01 -2.49268979e-01 -2.74571419e-01 6.49449408e-01
4.76216614e-01 -1.89768776e-01 1.20019829e+00 -1.07920396e+00
-7.39186645e-01 2.81223625e-01 -5.03402948e-01 5.63098013e-01
9.41484123e-02 8.14528763e-02 -6.99655414e-02 -2.52083167e-02
8.73146892e-01 1.25589180e+00 7.74989188e-01 4.12810534e-01
-9.63165760e-01 1.20068595e-01 -3.05830687e-01 -9.51451510e-02
-1.44106889e+00 1.13451101e-01 5.37729561e-01 -4.32393581e-01
6.99704289e-01 7.00310647e-01 8.41398358e-01 9.84019339e-01
1.14587176e+00 1.59015600e-02 1.14795244e+00 -1.71957985e-01
2.97468722e-01 -4.86872420e-02 2.31482029e-01 2.56919980e-01
8.50022852e-01 3.40313226e-01 5.19324899e-01 -8.01630914e-01
1.06785858e+00 -9.97657180e-02 2.84995884e-02 -2.50722408e-01
-9.32785690e-01 9.36030447e-01 9.84011143e-02 5.50169408e-01
-7.53796995e-01 -3.17681059e-02 9.51946806e-03 3.40210825e-01
9.24118876e-01 3.31351399e-01 -3.95891547e-01 4.29248840e-01
-5.01715779e-01 5.19285202e-01 7.98828900e-01 6.04584098e-01
1.57344028e-01 -1.57750994e-01 -1.49213776e-01 4.26988780e-01
3.07116300e-01 1.21249592e+00 -6.19059622e-01 -9.33797419e-01
-1.01388179e-01 7.22609237e-02 5.64456522e-01 -7.11537302e-01
-9.69816446e-01 -4.71743375e-01 -1.06489348e+00 -2.12407485e-01
4.35898513e-01 -7.33665407e-01 -4.82512057e-01 1.76678693e+00
6.07879817e-01 2.68229216e-01 6.47684708e-02 4.01378453e-01
3.95539284e-01 7.47898519e-01 2.74260342e-01 -7.75989354e-01
1.07635903e+00 -2.69910127e-01 -8.06708813e-01 6.08465493e-01
8.38519156e-01 -3.60853374e-01 -2.42039666e-01 -1.28400356e-01
-1.32378030e+00 2.86484182e-01 -4.38504010e-01 6.43115580e-01
-2.55692661e-01 -4.69389439e-01 3.56852055e-01 8.78944457e-01
-1.60779166e+00 4.88654047e-01 -7.97457397e-01 -9.13282394e-01
-3.06724787e-01 3.97716999e-01 1.62451372e-01 2.03929946e-01
-1.46034312e+00 1.07025671e+00 -4.98529881e-01 -1.00296445e-03
-5.96045494e-01 -8.96678269e-01 -3.14723492e-01 -3.43727529e-01
-2.54520267e-01 -1.33092189e+00 6.42646611e-01 -2.68433750e-01
-9.81537759e-01 6.78231657e-01 -1.10146413e-02 -2.95523554e-01
5.83239436e-01 3.89143676e-01 -7.38858521e-01 2.83205062e-01
-1.00015298e-01 -1.91878285e-02 2.58921444e-01 -1.13291764e+00
-4.58507761e-02 -5.08771360e-01 -3.66004393e-03 9.49669704e-02
2.13200554e-01 7.72198677e-01 3.46351177e-01 -6.24576747e-01
-5.14641762e-01 -9.33421969e-01 -6.46931469e-01 -4.90434259e-01
-3.21027398e-01 -1.72388881e-01 4.11522359e-01 -1.00863516e+00
1.16122031e+00 -1.70201385e+00 3.09563160e-01 3.29811394e-01
2.19060332e-01 1.53203994e-01 -2.18168586e-01 8.11561406e-01
4.08213586e-01 3.47769670e-02 -3.26779723e-01 1.53568834e-01
-3.80646288e-01 -3.62481736e-02 2.07726881e-01 1.09746230e+00
2.08132625e-01 7.46562541e-01 -9.81492817e-01 -3.19937855e-01
-1.03159085e-01 8.80322874e-01 -4.52617675e-01 2.22385734e-01
2.67066002e-01 8.90112519e-01 -6.90821707e-01 1.80673525e-01
1.29441154e+00 -1.42411171e-02 1.77101299e-01 6.72514558e-01
-6.33656502e-01 -2.44725525e-01 -4.03107226e-01 3.46826762e-01
-2.57585198e-01 1.64178073e-01 5.89504480e-01 -7.52527297e-01
6.17027283e-01 4.33063537e-01 7.27825940e-01 -3.09756786e-01
3.78719479e-01 1.72520950e-01 4.20674831e-01 -6.12169147e-01
2.56158680e-01 -2.35176876e-01 -2.09216122e-02 7.51080811e-01
-1.80338532e-01 2.97916502e-01 3.66510861e-02 1.65750667e-01
1.30338407e+00 -3.34380507e-01 -1.35658896e-02 -1.24134779e+00
2.43374437e-01 -1.29070729e-02 2.12253988e-01 8.09886694e-01
-1.12046838e-01 1.05722181e-01 7.00770676e-01 -3.84763956e-01
-1.09835982e+00 -1.24208665e+00 -7.55345404e-01 6.67876601e-01
5.53495288e-01 1.69902280e-01 -1.19485343e+00 -1.02308579e-02
-5.80210648e-02 7.58818269e-01 -1.03585124e+00 -5.07311165e-01
-8.27064037e-01 -2.00731492e+00 6.39108956e-01 -7.01171756e-02
2.90658534e-01 -6.47204041e-01 -5.99524796e-01 -2.98887249e-02
2.02539384e-01 -2.98658878e-01 1.36222932e-02 -7.21307769e-02
-7.93447793e-01 -9.73028779e-01 -1.37083280e+00 -3.94298673e-01
8.79497647e-01 8.28512460e-02 8.37439835e-01 2.89376557e-01
-3.04670423e-01 5.50856173e-01 -1.34275574e-02 -7.25698695e-02
-8.54179144e-01 -2.93650627e-01 3.59839916e-01 -2.44328111e-01
3.62093985e-01 -1.95067432e-02 -9.60343599e-01 7.46441662e-01
-6.85383260e-01 -4.67040867e-01 2.30248757e-02 1.50813222e-01
-3.40064406e-01 -6.86763003e-02 6.24501288e-01 -3.89250875e-01
8.26000094e-01 -1.22667909e+00 -5.16775310e-01 5.31465769e-01
-1.78885281e-01 -1.70309544e-01 1.24865413e-01 -4.61772591e-01
-1.55051732e+00 -8.09016764e-01 -1.56166852e-02 5.46999037e-01
-7.83192813e-02 1.14509605e-01 2.60382771e-01 -2.84853995e-01
3.85531574e-01 -9.62815210e-02 2.82373577e-02 -4.21473503e-01
2.68327236e-01 6.77202761e-01 -5.08698933e-02 -3.24402452e-01
6.94839656e-01 6.35789037e-01 4.95332271e-01 -1.53800738e+00
8.38893279e-02 -5.14670163e-02 -6.61120296e-01 -4.74949569e-01
1.11445975e+00 -3.51940721e-01 -7.41782010e-01 8.69143069e-01
-1.41792977e+00 -5.86583853e-01 -5.85297719e-02 8.43061566e-01
-6.76112771e-01 7.15318695e-02 -8.67716432e-01 -1.28812957e+00
1.90842614e-01 -6.50198936e-01 5.41393161e-01 3.14591601e-02
-1.11980841e-01 -1.90793204e+00 9.70258117e-01 -1.45571470e-01
9.84417677e-01 3.07922035e-01 9.41777349e-01 -4.96845514e-01
-3.54494989e-01 4.70483378e-02 -1.11605890e-01 -5.76572418e-01
-1.17220871e-01 1.79997325e-01 -2.25106671e-01 -3.04700226e-01
7.26783276e-01 6.17677212e-01 6.08046055e-01 1.18804348e+00
1.75900847e-01 -4.14459556e-01 -7.41127074e-01 6.77196145e-01
1.20014608e+00 5.17718554e-01 4.86353517e-01 1.38038918e-01
3.05632740e-01 1.22045600e+00 3.73269945e-01 3.43752980e-01
5.38455904e-01 7.31553078e-01 4.11876813e-02 -2.85030544e-01
3.56323689e-01 3.65332723e-01 7.05972165e-02 6.78535223e-01
-8.38373005e-01 -5.23246408e-01 -1.07495999e+00 5.48386097e-01
-1.63494515e+00 -1.24463105e+00 -9.71884787e-01 2.34604597e+00
6.41620517e-01 -4.86725003e-01 5.79602480e-01 -6.93847358e-01
1.11358321e+00 -2.39583462e-01 -1.24552667e-01 -5.08001447e-01
-5.81740022e-01 -3.29630345e-01 1.13642919e+00 1.05122495e+00
-8.38750660e-01 2.75112659e-01 8.62733364e+00 2.67779052e-01
-7.45012105e-01 4.71427321e-01 5.68094790e-01 -5.25566898e-02
-6.28160179e-01 -1.58736199e-01 -5.14196396e-01 6.35639846e-01
1.46544862e+00 -4.48469847e-01 1.35375142e-01 -9.31623504e-02
5.97008944e-01 -2.58869588e-01 -2.92198777e-01 1.49079813e-02
-1.85345978e-01 -8.34361017e-01 -4.92303729e-01 5.34336567e-01
9.82312024e-01 1.05153054e-01 7.79329687e-02 -2.34540790e-01
5.17091870e-01 -5.38612843e-01 4.03306894e-02 7.28276074e-01
5.01370013e-01 -7.09433377e-01 7.74428129e-01 3.89572471e-01
-9.20454264e-01 2.87880272e-01 -5.08636355e-01 -9.33406204e-02
8.12833250e-01 5.08627474e-01 -3.40747446e-01 5.61226785e-01
3.98403704e-01 2.23932922e-01 -4.77377802e-01 1.47470760e+00
1.86613098e-01 4.61697280e-01 -5.07835984e-01 -1.66001752e-01
6.92947879e-02 -8.81811917e-01 9.88646090e-01 9.82598484e-01
4.96866882e-01 4.44685906e-01 -7.23712146e-01 8.84986699e-01
7.97881603e-01 -1.60232663e-01 -7.67107189e-01 3.92219663e-01
-5.92341833e-02 7.35101104e-01 -1.02600074e+00 4.56403233e-02
-8.11636895e-02 6.39667034e-01 -3.68680596e-01 8.29292834e-01
-1.02519488e+00 -3.14470470e-01 6.11921847e-01 4.63954747e-01
8.13700259e-02 -9.90184173e-02 -5.54517843e-02 -1.03620660e+00
-3.77829760e-01 2.17732802e-01 8.83028209e-02 -2.90393591e-01
-1.11490381e+00 7.02341080e-01 9.91912365e-01 -4.53625381e-01
-3.89206231e-01 -3.92657846e-01 -9.70798850e-01 8.73139918e-01
-9.29232299e-01 -8.57933402e-01 5.87986052e-01 4.58725721e-01
-3.39078784e-01 1.79056972e-01 4.86663401e-01 2.28355139e-01
-6.37721419e-01 5.90866879e-02 7.89639235e-01 -6.14397228e-01
1.87270910e-01 -9.61029649e-01 7.32880712e-01 5.94755828e-01
-1.23425162e+00 1.08172965e+00 1.33257389e+00 -1.30283654e+00
-1.18035984e+00 -1.05337143e+00 1.11164963e+00 -5.21719038e-01
9.60542738e-01 -4.34045255e-01 -7.47365892e-01 4.56747115e-01
5.75983763e-01 -5.25499821e-01 4.67542291e-01 -1.74043804e-01
2.44313270e-01 5.04506469e-01 -1.65403509e+00 8.65670860e-01
1.22696090e+00 2.05001071e-01 -2.50205994e-01 6.84058309e-01
7.06166089e-01 4.42816466e-01 -9.16068971e-01 5.78616440e-01
3.20687383e-01 -9.16160047e-01 1.17431986e+00 -6.83026075e-01
-1.47920489e-01 2.71513641e-01 3.84239167e-01 -1.48413360e+00
-6.07681274e-01 -8.51806700e-01 4.32275146e-01 7.01122642e-01
2.98828423e-01 -1.17354429e+00 1.70641348e-01 4.51477319e-01
4.39796746e-01 -8.22457850e-01 -1.12894976e+00 -1.06476283e+00
6.78917408e-01 4.07706261e-01 4.68862832e-01 6.96771801e-01
5.56320958e-02 -2.82255262e-01 -6.09099865e-01 2.56222814e-01
1.20061517e+00 -6.29432380e-01 1.85823351e-01 -1.39845860e+00
-5.18161468e-02 -2.41748661e-01 -2.82041848e-01 -4.24440354e-01
-1.51706234e-01 -2.87663341e-01 -1.51934460e-01 -1.39124894e+00
3.10967952e-01 -7.03080654e-01 2.64225248e-02 -6.39747441e-01
-3.23634632e-02 1.96132576e-03 -4.53831434e-01 1.13765292e-01
-4.55448896e-01 3.25517207e-01 9.09083307e-01 4.03503984e-01
-4.03840780e-01 3.48089397e-01 -1.95128143e-01 6.18508458e-01
1.18410873e+00 -6.57343626e-01 -4.12589639e-01 -2.34582260e-01
5.67266941e-01 4.16667402e-01 6.89229965e-01 -4.02079076e-01
7.93279484e-02 -6.08842313e-01 -2.15386406e-01 -4.12760586e-01
1.71550408e-01 -4.44633484e-01 8.63403082e-01 1.01646674e+00
-5.30042239e-02 1.77771389e-01 -2.41905339e-02 5.07749617e-01
6.54141843e-01 -1.16353638e-01 9.23071861e-01 1.02286555e-01
6.02143764e-01 6.70300275e-02 -9.55930591e-01 -1.78944282e-02
1.49152029e+00 1.29417822e-01 -9.78985608e-01 -3.16961408e-01
-6.92243218e-01 2.58879244e-01 8.60852778e-01 1.47901550e-01
2.31863648e-01 -1.04261625e+00 -1.03387368e+00 1.16202377e-01
-4.39464390e-01 -9.83136594e-01 5.88914394e-01 1.54828370e+00
-9.88473892e-01 5.82276583e-01 -3.60997111e-01 -3.31396312e-01
-1.06342137e+00 1.04797924e+00 5.98177075e-01 -1.69001073e-01
-2.10726187e-01 6.99465632e-01 6.51161432e-01 -3.98457050e-01
-3.89394104e-01 2.99451053e-01 -2.14923635e-01 -2.02115670e-01
6.27715409e-01 1.08907139e+00 -4.96433824e-01 -1.11816287e+00
-7.23694444e-01 8.10006499e-01 3.85769546e-01 -3.12408686e-01
1.23702395e+00 -5.63712597e-01 -7.44093776e-01 3.89323920e-01
8.89787138e-01 -6.11341037e-02 -9.64521348e-01 4.03570712e-01
-6.64618686e-02 -9.24709160e-03 -3.23491871e-01 -6.10808671e-01
-7.27887690e-01 4.25040901e-01 2.05514118e-01 1.04437745e+00
7.81690180e-01 4.27058041e-01 2.95039982e-01 -2.22022101e-01
5.54287612e-01 -6.63777292e-01 -8.44562292e-01 1.59110606e-01
9.13553357e-01 -5.66497564e-01 -1.24510035e-01 -6.16241097e-01
-3.14144045e-01 3.72341573e-01 -7.39515647e-02 -4.74759459e-01
1.16968954e+00 5.40495276e-01 7.15667456e-02 -3.76383573e-01
-8.52759838e-01 5.64198904e-02 7.50506371e-02 1.14221239e+00
3.08121264e-01 2.77035356e-01 -1.15064120e+00 1.55191302e-01
1.89636037e-01 -2.35105753e-01 6.79187596e-01 6.09143198e-01
-4.29908872e-01 -1.12261081e+00 -7.19855428e-01 3.46050829e-01
-5.91382444e-01 4.80471328e-02 -5.70531726e-01 6.89797759e-01
4.61095050e-02 1.22367799e+00 3.63857985e-01 7.16174170e-02
7.44972611e-03 -3.45928460e-01 5.38979828e-01 -9.12592113e-02
-7.47958779e-01 9.60742757e-02 -1.52961267e-02 -1.52515262e-01
-4.83833730e-01 -8.34351361e-01 -7.35171497e-01 -8.22769940e-01
-2.58105278e-01 4.81243283e-01 4.98288989e-01 7.74865806e-01
1.68805286e-01 2.60295719e-01 6.80492282e-01 -6.06895745e-01
-4.23533589e-01 -9.17566597e-01 -1.00896657e+00 -2.54943132e-01
4.64806765e-01 -5.23260891e-01 -8.66524577e-01 -4.85241890e-01] | [5.94244384765625, 4.390457630157471] |
b6ab1aea-c088-4093-9432-34439ff6778f | combination-of-hidden-markov-random-field-and | 1705.04823 | null | http://arxiv.org/abs/1705.04823v4 | http://arxiv.org/pdf/1705.04823v4.pdf | Combination of Hidden Markov Random Field and Conjugate Gradient for Brain Image Segmentation | Image segmentation is the process of partitioning the image into significant
regions easier to analyze. Nowadays, segmentation has become a necessity in
many practical medical imaging methods as locating tumors and diseases. Hidden
Markov Random Field model is one of several techniques used in image
segmentation. It provides an elegant way to model the segmentation process.
This modeling leads to the minimization of an objective function. Conjugate
Gradient algorithm (CG) is one of the best known optimization techniques. This
paper proposes the use of the Conjugate Gradient algorithm (CG) for image
segmentation, based on the Hidden Markov Random Field. Since derivatives are
not available for this expression, finite differences are used in the CG
algorithm to approximate the first derivative. The approach is evaluated using
a number of publicly available images, where ground truth is known. The Dice
Coefficient is used as an objective criterion to measure the quality of
segmentation. The results show that the proposed CG approach compares favorably
with other variants of Hidden Markov Random Field segmentation algorithms. | ['Samy Ait-Aoudia', 'EL-Hachemi Guerrout', 'Ramdane Mahiou', 'Dominique Michelucci'] | 2017-05-13 | null | null | null | null | ['brain-image-segmentation'] | ['medical'] | [ 3.29303801e-01 9.22835469e-02 -1.89449042e-01 -4.08992767e-01
-7.02938735e-01 -3.03691894e-01 4.81112599e-01 3.80614221e-01
-7.93738008e-01 7.59067416e-01 -2.10114151e-01 -2.53011018e-01
2.21992228e-02 -6.05843961e-01 -1.49526700e-01 -9.43202317e-01
8.05569664e-02 5.62394142e-01 4.87196892e-01 2.43770301e-01
5.95801294e-01 7.36532927e-01 -9.35415983e-01 -1.79335132e-01
8.74853075e-01 5.94113946e-01 5.94927967e-01 7.11008310e-01
-1.90113947e-01 6.76968575e-01 -2.69769490e-01 -1.65569317e-02
2.05743402e-01 -6.65958524e-01 -1.29993951e+00 4.42678064e-01
-2.81068414e-01 -1.33533373e-01 3.94665152e-01 1.21740139e+00
4.28587407e-01 2.60315061e-01 9.87030685e-01 -6.95403576e-01
3.55720259e-02 3.42794210e-01 -4.70292002e-01 3.21409404e-01
2.75677502e-01 -4.32458855e-02 4.20323044e-01 -5.14597952e-01
9.25359488e-01 1.05146778e+00 5.14581680e-01 1.48469165e-01
-1.12878084e+00 4.21368740e-02 -4.92778331e-01 2.33242944e-01
-1.46703565e+00 1.31728619e-01 5.94036758e-01 -7.18018770e-01
6.51085794e-01 3.20229471e-01 7.33254373e-01 -1.99220944e-02
7.41710603e-01 8.71165454e-01 1.59838915e+00 -7.55653441e-01
3.98019135e-01 2.96505779e-01 3.70867550e-01 8.35771680e-01
1.89256564e-01 -1.88463315e-01 1.92081451e-01 -2.22299844e-01
6.66734755e-01 7.11474521e-03 -2.79880375e-01 -2.59350300e-01
-1.05983520e+00 8.42187285e-01 3.62027526e-01 6.55379295e-01
-6.21843219e-01 -8.60331208e-02 -3.64326197e-03 -2.85334766e-01
3.22612315e-01 1.61493316e-01 2.22708941e-01 1.40140429e-01
-1.50978887e+00 1.20082207e-01 5.90784669e-01 3.32813352e-01
7.28444040e-01 -2.75532097e-01 -1.34656236e-01 3.30208540e-01
9.04879093e-01 4.07012343e-01 6.13233685e-01 -1.05712163e+00
-1.90068513e-01 5.82968771e-01 7.55108520e-02 -1.12109768e+00
-3.77896428e-01 -2.71367937e-01 -7.52852142e-01 5.09679794e-01
6.00465536e-01 -6.52797073e-02 -1.13613856e+00 8.71078312e-01
6.45076573e-01 -5.97300269e-02 -1.11082293e-01 8.53705049e-01
5.10899842e-01 7.56642938e-01 8.32888782e-02 -5.17261267e-01
9.59585905e-01 -8.51767778e-01 -1.00694084e+00 3.92560303e-01
6.06038153e-01 -1.25284243e+00 3.70280951e-01 5.23197353e-01
-9.71866548e-01 -3.50282967e-01 -8.62021923e-01 9.91837233e-02
-3.38379532e-01 7.57643487e-03 5.11493146e-01 8.03292572e-01
-1.06279504e+00 8.85951102e-01 -1.10590088e+00 -4.48396772e-01
1.43949270e-01 5.20981371e-01 -2.35247880e-01 1.10909618e-01
-5.79501808e-01 9.52067554e-01 4.09440905e-01 2.88878500e-01
-6.52558446e-01 -3.67261358e-02 -5.38500965e-01 -3.13150316e-01
-1.04360759e-01 -3.77874702e-01 9.86966312e-01 -8.65868330e-01
-1.61085773e+00 1.22473729e+00 -2.55340904e-01 -3.78733695e-01
8.46620679e-01 2.38687962e-01 -1.61538683e-02 5.46781540e-01
1.11938559e-01 6.94245398e-01 7.98029184e-01 -1.09985220e+00
-4.02783960e-01 -1.90848708e-01 -4.33853298e-01 2.25228637e-01
4.06743854e-01 7.33157098e-02 -2.50374109e-01 -3.36726010e-01
5.23370266e-01 -1.06537449e+00 -6.83944881e-01 -2.18360037e-01
-7.10693896e-01 -2.38024872e-02 6.98395848e-01 -8.76639009e-01
1.09586465e+00 -1.92781401e+00 -3.27874703e-04 5.23611426e-01
2.76278079e-01 2.57556103e-02 7.10368156e-01 2.31284052e-01
1.48031646e-02 1.04379259e-01 -7.53302336e-01 -2.08298966e-01
-3.46090972e-01 1.92323878e-01 4.64313298e-01 9.98405993e-01
-3.24501753e-01 6.40792787e-01 -6.88842475e-01 -1.11929977e+00
4.64467436e-01 5.44664979e-01 -1.81264266e-01 -3.57767604e-02
9.70145836e-02 8.90090168e-01 -6.19860113e-01 3.06014776e-01
7.89168894e-01 -2.70670682e-01 -3.75854224e-02 5.02536446e-03
-3.78964692e-01 -3.21403295e-01 -1.24721527e+00 1.36504471e+00
1.63664654e-01 6.04496300e-01 2.17510797e-02 -1.01025391e+00
8.32304478e-01 5.29056013e-01 9.31895137e-01 -2.18838174e-02
5.47790945e-01 3.71388704e-01 -5.61429793e-03 -5.05106926e-01
1.33808613e-01 -3.21729630e-01 5.26408792e-01 3.64853770e-01
-1.51909441e-01 -3.88634622e-01 4.89371538e-01 6.71102181e-02
7.08861887e-01 3.20927873e-02 6.50703609e-01 -6.43490016e-01
8.04163516e-01 3.85377556e-01 4.27066922e-01 4.15038407e-01
-4.01552528e-01 7.62054622e-01 2.37764552e-01 -3.23088586e-01
-7.30667889e-01 -7.42527127e-01 -3.29994678e-01 6.98041171e-02
1.96064010e-01 2.13186800e-01 -1.21226382e+00 -4.23051745e-01
-4.76956546e-01 4.03558314e-01 -5.49424291e-01 4.32209283e-01
-3.52463543e-01 -1.02809250e+00 -4.03374620e-02 -1.33869857e-01
8.02544951e-01 -1.00133216e+00 -8.71974111e-01 3.27887774e-01
-2.06680536e-01 -7.11455047e-01 -1.74474612e-01 -2.58382466e-02
-1.35250318e+00 -1.01916194e+00 -1.42435431e+00 -7.09879935e-01
1.05191886e+00 6.91602156e-02 7.85300910e-01 2.54565567e-01
-4.35199469e-01 2.71213859e-01 -3.41509193e-01 -1.89141348e-01
-7.08082557e-01 7.09493160e-02 -4.80423212e-01 1.58948347e-01
1.55334234e-01 -1.20335758e-01 -6.99356139e-01 2.22614765e-01
-1.09366763e+00 -1.41523853e-01 8.11597705e-01 6.30955398e-01
1.06138623e+00 3.05969983e-01 -3.46852303e-01 -1.23282194e+00
4.45091784e-01 -2.46456280e-01 -9.89576519e-01 2.22036690e-01
-7.63765633e-01 4.67203036e-02 1.10965393e-01 -1.02979839e-01
-1.06790817e+00 6.01946175e-01 -2.53584027e-01 1.58117320e-02
-3.75553906e-01 4.80996430e-01 1.57167196e-01 -4.76823717e-01
3.60151440e-01 1.00486264e-01 1.33262113e-01 -5.38798928e-01
1.80849731e-02 5.75555384e-01 3.15152407e-01 1.41201198e-01
2.84839958e-01 7.16789186e-01 4.56472427e-01 -9.57994103e-01
-4.35310841e-01 -1.07445812e+00 -8.54014575e-01 -4.71488476e-01
1.46506119e+00 -1.51549429e-01 -4.48574811e-01 5.93809128e-01
-1.04199159e+00 -2.72246838e-01 -8.92918557e-02 9.33660746e-01
-7.64416277e-01 6.43163085e-01 -5.19801795e-01 -9.17347729e-01
-3.04924726e-01 -1.62692666e+00 6.56548202e-01 4.76319462e-01
-7.05202892e-02 -1.32457471e+00 2.12416023e-01 3.64831388e-01
1.52073473e-01 5.79412460e-01 6.65426970e-01 -3.21933836e-01
-5.55051327e-01 -4.71982926e-01 2.46849552e-01 3.05600911e-01
1.30625039e-01 1.98852777e-01 -7.23265409e-01 3.29426900e-02
4.90289360e-01 4.07948166e-01 6.44480765e-01 1.14416444e+00
6.97816372e-01 1.43087478e-02 -5.11729777e-01 4.52695698e-01
1.74962747e+00 4.49954540e-01 9.50913131e-01 3.91521335e-01
5.05530536e-01 5.34669042e-01 7.69451797e-01 1.65396348e-01
1.03164408e-02 3.34532320e-01 1.63494781e-01 -5.33730745e-01
-9.56683010e-02 1.23539798e-01 -1.06375627e-01 8.20560873e-01
-1.48525283e-01 -1.29237711e-01 -1.22122943e+00 4.35007930e-01
-1.44118464e+00 -8.44616830e-01 -7.47888803e-01 2.21292496e+00
5.94400764e-01 1.07007459e-01 1.03188172e-01 4.90558505e-01
8.45427394e-01 -3.28639388e-01 -7.65221613e-03 -3.95597398e-01
1.00417115e-01 4.75654304e-02 9.49888647e-01 9.67934072e-01
-1.15919685e+00 7.34221220e-01 6.62140465e+00 8.45546424e-01
-1.34320378e+00 2.91040003e-01 7.84990668e-01 8.95685732e-01
1.02016442e-01 2.46448562e-01 -8.11663270e-01 5.54935277e-01
5.65106213e-01 5.60029224e-02 -1.95134096e-02 5.20375848e-01
6.32521272e-01 -1.04952478e+00 -4.28987563e-01 8.39555860e-01
-2.32897371e-01 -1.06838143e+00 -2.19063923e-01 2.96719223e-01
9.65172827e-01 -2.54476696e-01 5.78533150e-02 -5.65677643e-01
-4.06327704e-03 -8.99336100e-01 3.70774686e-01 8.63581538e-01
3.25690597e-01 -5.61888218e-01 1.15565014e+00 3.79439533e-01
-8.36987317e-01 3.79977942e-01 -1.45728156e-01 2.63656765e-01
5.48980117e-01 8.77527714e-01 -1.19829524e+00 2.78961033e-01
3.86134118e-01 3.23713273e-01 -4.59707737e-01 1.57360303e+00
-8.88954028e-02 5.91980159e-01 -4.83784378e-01 -6.06348850e-02
4.20435756e-01 -6.40377700e-01 7.06132650e-01 1.09851933e+00
8.40942636e-02 -5.87791763e-02 1.38900638e-01 6.69356823e-01
5.37413776e-01 7.20837772e-01 -5.06446302e-01 -1.78846106e-01
-1.97036177e-01 1.29288983e+00 -1.73640931e+00 -4.63906199e-01
-7.84174204e-02 1.05256629e+00 -5.27811885e-01 2.15311304e-01
-3.45246613e-01 -3.94324601e-01 -4.58064437e-01 3.57059330e-01
-1.51061593e-02 -3.15927118e-01 -2.34429896e-01 -5.72820544e-01
-3.97083789e-01 -4.25013900e-01 3.61603320e-01 -5.74354887e-01
-7.06359088e-01 6.89380705e-01 3.13680440e-01 -1.13275802e+00
-4.36821222e-01 -5.04155338e-01 -4.52796936e-01 1.07256818e+00
-1.37303460e+00 -8.07331145e-01 -2.17187658e-01 4.49663550e-01
4.64023858e-01 1.96141690e-01 5.76252341e-01 1.79969683e-01
-3.69900733e-01 -8.98627378e-03 4.42369848e-01 4.67952155e-02
3.42947781e-01 -1.42993450e+00 -3.83138925e-01 1.07493174e+00
-2.34250918e-01 4.25484657e-01 9.58715618e-01 -9.76751804e-01
-6.16792202e-01 -5.82682669e-01 9.01822925e-01 1.11181542e-01
1.16912358e-01 3.72634351e-01 -7.31822014e-01 5.05017638e-01
2.75185943e-01 -9.71504720e-04 7.47183561e-01 -9.11296725e-01
9.58689570e-01 2.26465672e-01 -1.57118773e+00 2.18500003e-01
-8.62335935e-02 3.97621021e-02 -2.14616045e-01 4.56186503e-01
-1.00682467e-01 -2.20599905e-01 -1.04590058e+00 1.80681840e-01
2.44723886e-01 -1.13700163e+00 6.09378159e-01 2.22346768e-01
5.88624105e-02 -4.36893106e-01 1.33859947e-01 -1.01728523e+00
-1.25969024e-02 -8.86507630e-01 5.12497187e-01 9.33577597e-01
4.19371784e-01 -5.72621465e-01 1.00235248e+00 5.50080776e-01
1.77686766e-01 -7.00927258e-01 -6.91036046e-01 -4.77604449e-01
-6.65960312e-02 6.29037321e-02 3.20448205e-02 8.12688053e-01
-2.49672011e-01 -1.20517865e-01 1.42411783e-01 -9.81495678e-02
1.04849291e+00 -1.22737490e-01 2.80061156e-01 -1.13735569e+00
-3.07745427e-01 -3.64666820e-01 -6.54236794e-01 -8.37712765e-01
-2.00816363e-01 -7.55465627e-01 1.77136257e-01 -1.87094653e+00
3.00764646e-02 -5.20433068e-01 7.75468349e-02 -1.44056678e-01
-1.63928747e-01 1.59403399e-01 1.52899940e-02 5.47060847e-01
-2.39409000e-01 4.67382278e-03 1.56570375e+00 1.60048902e-02
-2.70045340e-01 4.70338166e-01 1.07478723e-01 9.44095790e-01
1.05086100e+00 -7.97821820e-01 -2.91099548e-01 4.06040177e-02
-2.49214724e-01 1.23286180e-01 1.15227476e-02 -1.09521222e+00
3.20208371e-01 -3.27394940e-02 3.74278307e-01 -7.64129817e-01
1.43420562e-01 -8.70356202e-01 4.05933589e-01 8.37079704e-01
-2.27132037e-01 1.02097362e-01 -3.65518093e-01 3.11180204e-01
-4.55646127e-01 -1.05744708e+00 1.19197440e+00 -5.89197218e-01
-7.32049704e-01 3.44314165e-02 -6.91503286e-01 -3.85872751e-01
1.25130665e+00 -4.80615228e-01 5.90729535e-01 -3.15511256e-01
-1.06156909e+00 -1.95165619e-01 4.88520354e-01 -6.58644915e-01
6.38053954e-01 -8.49957108e-01 -4.26699221e-01 -1.11368462e-01
-4.85079199e-01 -2.11229939e-02 1.02439456e-01 1.47697353e+00
-1.57268631e+00 4.37393904e-01 -2.23645478e-01 -8.45164835e-01
-1.48063517e+00 4.74784464e-01 6.04284406e-01 -5.97625971e-01
-4.75775331e-01 6.85230553e-01 -2.09854364e-01 -4.88721242e-04
-1.43646434e-01 -2.48454228e-01 -5.91945469e-01 -1.78057015e-01
3.08139354e-01 7.30766058e-01 -8.51649120e-02 -1.01530361e+00
-2.51578212e-01 8.02942038e-01 1.02165222e-01 -6.20827913e-01
8.94416749e-01 -3.96365613e-01 -4.15821284e-01 4.42671508e-01
1.31545806e+00 -1.06631145e-01 -8.26590776e-01 2.09649161e-01
2.96909451e-01 -3.73974919e-01 7.01646864e-01 -5.57174683e-01
-9.60464478e-01 1.13291943e+00 1.25749791e+00 2.75589287e-01
1.10300767e+00 -3.79566550e-01 6.64817631e-01 -5.38873821e-02
1.35977268e-01 -1.19999290e+00 -4.75030512e-01 -4.75945510e-02
3.17903996e-01 -1.28441072e+00 3.62969875e-01 -7.55253434e-01
-5.30533791e-01 1.18213153e+00 -2.58822978e-01 -1.50101304e-01
1.01266980e+00 1.47439435e-01 3.05039883e-01 -1.79295331e-01
1.79103896e-01 -3.44340295e-01 2.85945296e-01 3.78503203e-01
6.19985580e-01 1.10689491e-01 -1.10798657e+00 -2.89391041e-01
1.01325475e-01 2.45465964e-01 4.82145876e-01 1.27598536e+00
-6.47713602e-01 -1.14055371e+00 -7.08816469e-01 2.86489427e-01
-1.00798416e+00 2.12377965e-01 -1.62863374e-01 6.78918123e-01
1.08625934e-01 1.01551902e+00 -3.31813216e-01 2.24253893e-01
-2.89061248e-01 -1.78285107e-01 5.33367991e-01 -3.78577948e-01
-4.17626292e-01 4.87897933e-01 -5.49545348e-01 -3.85154098e-01
-8.36804330e-01 -7.67675400e-01 -1.61560154e+00 4.02430631e-02
-4.82728928e-01 5.24414480e-01 1.15793383e+00 1.00577188e+00
-3.12385589e-01 1.49623692e-01 5.84693134e-01 -8.40853810e-01
-1.55785382e-01 -7.04847991e-01 -9.27379072e-01 3.08888942e-01
1.86754048e-01 -6.32426858e-01 -2.52571374e-01 4.64688659e-01] | [14.394984245300293, -2.7938761711120605] |
3819b7eb-9409-4465-a709-bbea7cea2bff | a-survey-of-noma-state-of-the-art-key | 2306.06664 | null | https://arxiv.org/abs/2306.06664v1 | https://arxiv.org/pdf/2306.06664v1.pdf | A Survey of NOMA: State of the Art, Key Techniques, Open Challenges, Security Issues and Future Trends | Non-orthogonal multiple access (NOMA) systems can serve multiple users in contrast to orthogonal multiple-access (OMA), which makes use of the limited time or frequency domain resources. It can help to address the unprecedented technological advancements of the sixth generation (6G) network, which include high spectral efficiency, high flexibility, low transmission latency, massive connectivity, higher cell-edge throughput, and user fairness. NOMA has gained widespread recognition as a viable technology for future wireless networks. The main characteristic that sets NOMA apart from the conventional orthogonal multiple access (OMA) techniques is its ability to handle more users than orthogonal resource slots. NOMA techniques can serve multiple users in the same resource block by multiplexing users in power or code domain. The purpose of this paper is to provide a thorough overview of the promising NOMA systems. Initially, we discuss the state-of-the-art and existing literature on NOMA systems. This study also examines the practical deployment of NOMA implementation and key performance indicators. An overview of the most recent NOMA advancements and applications is also given in this survey. We also briefly discuss that multiple-input multiple-output (MIMO), visible light communications, cognitive and cooperative communications, intelligent reflecting surfaces (IRS), unmanned aerial vehicles (UAV), HetNets, backscatter communication, mobile edge computing (MEC), deep learning (DL), and other emerging and existing wireless technologies can all be flexibly combined with NOMA. This study surveys a thorough analysis of the interactions between NOMA and the aforementioned technologies. Lastly, we will highlight a number of difficult open problems and security issues that need to be resolved for NOMA, along with pertinent possibilities and potential future research directions. | ['Yanlong Li', 'Syed Agha Hassnain Mohsan'] | 2023-06-11 | null | null | null | null | ['edge-computing'] | ['time-series'] | [-4.02138988e-03 -5.40643409e-02 -2.20222965e-01 4.83314157e-01
1.73822239e-01 -4.55939800e-01 3.47842813e-01 -5.48169196e-01
-2.18378887e-01 1.33582413e+00 -4.36414331e-01 -1.03413260e+00
-6.31828725e-01 -8.47496510e-01 3.41955245e-01 -1.33908713e+00
-7.79704392e-01 -2.43349262e-02 -4.01088178e-01 -3.54137391e-01
-2.03856468e-01 7.73473918e-01 -1.32787061e+00 -4.35101032e-01
1.04672575e+00 1.29856467e+00 2.73857951e-01 4.64739174e-01
3.20458025e-01 -1.04559965e-01 -7.61078238e-01 1.69856578e-01
3.83566290e-01 -2.25404263e-01 -4.17429775e-01 1.02057852e-01
-5.28586619e-02 -1.51145145e-01 -4.21210021e-01 4.79939669e-01
9.82602358e-01 -3.61635797e-02 7.35231578e-01 -1.27161324e+00
-3.74656260e-01 4.50601429e-02 -6.03294671e-01 3.21957380e-01
-6.27842173e-02 -2.64871478e-01 5.81083179e-01 -8.47551823e-01
9.73263681e-02 8.53165984e-01 5.73135793e-01 5.58163643e-01
-2.84335554e-01 -6.67890310e-01 -2.82758862e-01 -2.45664408e-03
-1.62351167e+00 -5.98720193e-01 -8.85607377e-02 -3.46495807e-02
7.93709278e-01 6.07821643e-01 1.09833944e+00 3.68881732e-01
6.36494577e-01 5.27411222e-01 9.27803636e-01 -8.58981133e-01
2.02044681e-01 -6.60189241e-02 -1.08343370e-01 7.82421947e-01
8.84635270e-01 3.08434039e-01 -8.24258476e-02 -4.34722692e-01
6.51540637e-01 -3.62582713e-01 -5.55456638e-01 -3.29645872e-01
-1.01553321e+00 5.18740058e-01 4.60000932e-02 1.75728738e-01
-5.73605001e-01 1.42326891e-01 -2.84688711e-01 7.22721636e-01
5.23171425e-02 7.22266614e-01 -2.86274910e-01 2.23435551e-01
-9.17605460e-01 -9.03615914e-03 9.09277737e-01 1.36289537e+00
1.30297214e-01 8.88467073e-01 -2.64432821e-02 6.45080447e-01
5.77724993e-01 9.52620447e-01 4.20477986e-02 -7.68518388e-01
-3.42057459e-02 -4.01255697e-01 1.44535437e-01 -3.51480424e-01
-8.57886493e-01 -1.69196928e+00 -1.20274842e+00 4.64928269e-01
1.48697034e-01 -1.15345156e+00 -7.24196613e-01 1.21556306e+00
-5.64023964e-02 5.43735884e-02 6.33078754e-01 6.31595016e-01
4.88330454e-01 5.27588665e-01 -5.99705935e-01 -7.94327438e-01
1.32950234e+00 -8.66884589e-01 -6.71171129e-01 -2.34328657e-01
5.57551801e-01 -9.24679220e-01 -3.29554766e-01 4.38972145e-01
-1.16668248e+00 -1.16915911e-01 -1.48387289e+00 7.18020320e-01
-2.70752817e-01 3.47656518e-01 9.35788929e-01 1.52808797e+00
-1.19918358e+00 1.00933956e-02 -3.11364532e-01 -7.47683585e-01
5.94536185e-01 7.81081140e-01 3.23784083e-01 -2.03468859e-01
-1.29368746e+00 8.17761242e-01 2.72919148e-01 1.20320790e-01
-4.36399549e-01 -2.19223157e-01 -6.84149921e-01 1.52402535e-01
3.46554846e-01 -1.50554836e+00 8.81074905e-01 -1.99723154e-01
-1.71747708e+00 3.43762964e-01 -1.85062110e-01 -6.43813193e-01
-1.85272954e-02 -1.99972346e-01 -9.46230650e-01 -5.39554469e-02
-3.30122232e-01 3.41441721e-01 3.03645432e-01 -1.16329682e+00
-1.08456790e+00 -2.00555295e-01 2.32783541e-01 2.91666597e-01
-4.14711505e-01 -3.84978615e-02 5.99729344e-02 -3.39892417e-01
3.04802537e-01 -1.30900979e+00 -6.71473026e-01 -2.93843567e-01
-4.16729301e-01 2.63636857e-01 9.92097437e-01 1.20663844e-01
1.53169537e+00 -1.96860802e+00 9.41893905e-02 4.12463695e-01
2.75690407e-01 3.12725484e-01 1.36962131e-01 6.02638185e-01
3.66297722e-01 -7.10691810e-02 1.87818751e-01 8.89455061e-03
-5.30411959e-01 5.47366031e-02 -3.83609235e-02 5.72480142e-01
-7.94010162e-01 7.41815567e-01 -9.52653170e-01 3.00496757e-01
2.70011067e-01 1.18110985e-01 -8.29141065e-02 -6.06034636e-01
5.36435664e-01 5.97972989e-01 -4.23878044e-01 1.08474648e+00
9.85539317e-01 -2.91640639e-01 3.48602742e-01 1.09271862e-01
-6.03102922e-01 -1.57150492e-01 -9.42956507e-01 9.16711569e-01
-7.53748417e-01 8.96865785e-01 4.25682575e-01 -8.69070172e-01
5.45061469e-01 9.52268720e-01 4.84839976e-01 -4.77421403e-01
2.87032485e-01 8.99352849e-01 5.34362614e-01 -2.60411561e-01
6.60214424e-01 -1.59731969e-01 6.96789548e-02 5.42837903e-02
4.46070507e-02 3.84462684e-01 -1.20911077e-01 -8.97115991e-02
6.90240979e-01 -6.14886165e-01 1.12514198e+00 -7.94063866e-01
5.60827613e-01 -5.68174481e-01 5.51130712e-01 1.05733883e+00
-5.06438985e-02 2.12992616e-02 -1.86719656e-01 -4.40155491e-02
-6.28613889e-01 -1.03895652e+00 -5.37178338e-01 5.05577803e-01
7.96967864e-01 -8.37516133e-03 8.23903531e-02 -5.81494011e-02
-3.81285772e-02 5.80698609e-01 2.03770474e-01 1.68082237e-01
-2.05342770e-02 -1.41600204e+00 5.23774087e-01 1.05629407e-01
8.36005688e-01 -3.95934656e-03 -3.42329413e-01 1.06464595e-01
2.60837466e-01 -1.39711773e+00 3.03988308e-01 -2.01486629e-02
-7.69021869e-01 -6.17052794e-01 -1.22114158e+00 -7.88943708e-01
5.21302223e-01 1.26114094e+00 6.90632343e-01 7.93490410e-02
2.67925113e-01 4.53395933e-01 -3.17890316e-01 -6.32233977e-01
6.43696859e-02 3.07815552e-01 9.80353773e-01 -1.30795717e-01
1.12415394e-02 -9.04206038e-01 -5.12720525e-01 5.90538800e-01
-1.02203324e-01 1.30153254e-01 8.72690499e-01 7.64363647e-01
-3.11349966e-02 3.37020785e-01 1.19922078e+00 -5.06238699e-01
5.42369485e-01 -1.04505622e+00 -5.30683219e-01 1.64078504e-01
-4.94298041e-01 -7.35716045e-01 6.34217620e-01 1.77578017e-01
-8.29254210e-01 -4.54724699e-01 3.55802849e-02 3.40605229e-01
5.14848717e-02 7.86780536e-01 -3.94286633e-01 -7.94861376e-01
6.63943470e-01 7.54349604e-02 -2.92998254e-01 1.24206357e-01
1.25488639e-01 1.36413240e+00 -3.40526567e-05 -1.68569863e-01
1.00995100e+00 8.47045124e-01 6.62687063e-01 -1.76274025e+00
-8.77351642e-01 -7.52950430e-01 -3.74766201e-01 -5.63645065e-01
4.14991528e-01 -1.15100181e+00 -2.39354551e-01 6.07597888e-01
-9.97830749e-01 -2.41263281e-03 1.51838616e-01 1.06926274e+00
-2.46242359e-01 4.01963025e-01 -1.93078831e-01 -1.31175780e+00
-4.14192259e-01 -8.07973206e-01 3.28771263e-01 8.20105374e-01
1.70397349e-02 -1.17479825e+00 -8.66380215e-01 4.22691494e-01
9.17080700e-01 3.55688423e-01 5.94989419e-01 -1.21566102e-01
-8.03874493e-01 -1.42500505e-01 -6.14431826e-03 2.64338800e-03
4.13716316e-01 -8.82847071e-01 -1.03496134e+00 -9.33122337e-01
-1.93813443e-01 2.79637188e-01 7.24864900e-01 9.12352383e-01
6.22151077e-01 9.77665707e-02 -9.41818476e-01 7.60197043e-01
1.16538119e+00 6.73122048e-01 7.43208766e-01 1.77556008e-01
4.50861722e-01 -5.99886551e-02 6.99003458e-01 4.83370513e-01
1.18317269e-01 5.86340189e-01 7.24633515e-01 -1.04721047e-01
1.75624058e-01 7.69319952e-01 1.79473266e-01 8.88467252e-01
-6.89877093e-01 -1.26721954e+00 -3.66650969e-01 2.94507891e-01
-1.49042332e+00 -8.74481022e-01 -6.06395781e-01 2.41921020e+00
-2.66551346e-01 1.05337545e-01 -9.64984000e-02 1.04695223e-01
6.78806484e-01 9.49714184e-02 -4.26366270e-01 -2.16796696e-01
-6.03371620e-01 -1.38954129e-02 1.28858542e+00 6.17538154e-01
-1.10370958e+00 5.66458702e-01 6.15517569e+00 1.12553382e+00
-8.97551894e-01 1.80400163e-01 -2.07868847e-03 8.54261816e-02
-2.36946732e-01 -7.47982189e-02 -7.40702450e-01 2.09371686e-01
7.25266576e-01 -3.68414372e-01 2.34310955e-01 3.45915377e-01
4.93176073e-01 -3.73870105e-01 -5.91935396e-01 1.11992359e+00
-7.53533766e-02 -1.66210520e+00 8.34485590e-02 1.00262237e+00
1.30757380e+00 1.25143871e-01 1.54153228e-01 1.52711198e-01
-1.68083385e-01 -8.11240911e-01 1.79211661e-01 6.16359174e-01
1.02302825e+00 -7.07586110e-01 1.06528640e+00 4.13353503e-01
-1.20930636e+00 -5.95591605e-01 -3.52785379e-01 -7.59637654e-01
7.19133079e-01 1.17390025e+00 -3.85751188e-01 1.41900218e+00
7.10561723e-02 5.22663355e-01 3.92214715e-01 1.97165203e+00
2.63577588e-02 5.07513523e-01 -5.54800391e-01 -6.83126822e-02
3.52793306e-01 -1.52971327e-01 1.32092392e+00 8.37778568e-01
1.17639709e+00 2.67897904e-01 3.09251606e-01 1.47787079e-01
6.45771846e-02 -1.25532776e-01 -7.78833985e-01 2.39357278e-01
1.06645501e+00 1.24977601e+00 -6.59095943e-01 -1.65872902e-01
-8.10890436e-01 7.12942898e-01 -8.17118466e-01 6.52006090e-01
-3.24798882e-01 -7.48406172e-01 8.69413912e-01 6.54076561e-02
4.89242934e-02 -1.08864963e+00 -3.55427891e-01 -9.58690524e-01
-4.66809928e-01 -2.79911906e-01 2.98908561e-01 -3.77121627e-01
-9.56436634e-01 1.81569219e-01 -3.36056948e-01 -2.05229235e+00
2.03159451e-01 -8.64788532e-01 -6.19726598e-01 9.84774828e-01
-1.88842535e+00 -9.09426272e-01 -3.93084705e-01 -2.51731724e-01
3.93427253e-01 -1.00078499e+00 1.00322950e+00 6.14707649e-01
-3.39509994e-01 4.75142092e-01 8.45054209e-01 -4.84856963e-01
2.01522768e-01 -8.84171963e-01 2.49569863e-01 7.75736511e-01
-4.93112355e-01 4.47932959e-01 5.23501158e-01 -2.54382372e-01
-1.24895847e+00 -6.39520824e-01 3.84975821e-01 1.14712402e-01
1.83497220e-01 9.00072008e-02 4.49913107e-02 3.23891997e-01
4.62292433e-01 -2.61682540e-01 1.27292478e+00 1.12748802e-01
8.13549280e-01 3.50289971e-01 -8.56644869e-01 7.30139554e-01
1.10844624e+00 1.85806602e-01 2.07603976e-01 1.67998016e-01
-4.08824198e-02 -6.03318095e-01 -4.94467229e-01 4.40393925e-01
9.70946312e-01 -8.10424268e-01 8.37782383e-01 -2.68867224e-01
-5.56318402e-01 -5.62047541e-01 -2.90913135e-01 -1.55522811e+00
-6.41741633e-01 -1.22853625e+00 -5.47625422e-01 6.93184018e-01
3.25255990e-01 -9.97269511e-01 4.72756922e-01 -3.52216244e-01
-4.53226238e-01 -1.09135926e+00 -9.91763294e-01 -1.29733670e+00
8.17360580e-02 -1.88238740e-01 4.72084373e-01 1.16762722e+00
2.19644815e-01 6.49069130e-01 -9.95938122e-01 8.83644581e-01
7.29428113e-01 -1.01305261e-01 5.87773204e-01 -1.78759956e+00
-1.11916021e-01 -2.46104777e-01 -5.01989901e-01 -1.87199903e+00
-2.74382472e-01 -9.58908558e-01 -6.59148991e-01 -1.92008376e+00
-3.65519673e-01 -8.49737167e-01 -3.24420214e-01 -1.02077752e-01
4.96108353e-01 2.98883766e-01 1.60189942e-02 6.21556453e-02
-4.66724783e-01 5.06903529e-01 1.50729239e+00 2.44887099e-01
-3.20348978e-01 1.03059983e+00 -8.65945160e-01 9.25472617e-01
1.15356517e+00 1.92575231e-01 -4.54776794e-01 -4.33286071e-01
3.45120639e-01 4.55767274e-01 1.01858035e-01 -1.51362133e+00
3.47977906e-01 -3.04123312e-01 3.01645249e-01 -2.42200315e-01
6.10332549e-01 -8.91179144e-01 -2.21299857e-01 5.62528849e-01
6.72503650e-01 -6.00606143e-01 1.19349204e-01 6.64955974e-01
2.41001159e-01 -3.35625529e-01 9.67947960e-01 1.40754402e-01
-9.33025777e-01 3.08719546e-01 -1.33263016e+00 -3.28379542e-01
1.39528275e+00 -9.13689137e-01 -3.79347235e-01 -8.69138896e-01
-8.08043838e-01 8.64641309e-01 -9.45029035e-02 4.29218560e-02
4.63721424e-01 -1.13737929e+00 -5.78130543e-01 -1.29568890e-01
-1.65274456e-01 -5.79914451e-01 4.40802217e-01 1.22398436e+00
-3.53980213e-01 9.56936419e-01 -4.40941185e-01 -2.72322625e-01
-1.33126676e+00 -3.11427295e-01 8.51169169e-01 1.92284986e-01
-1.12299979e-01 6.67713702e-01 -5.90389334e-02 6.20687120e-02
-2.26776689e-01 4.37580436e-01 -5.40075898e-01 -9.69820544e-02
2.95323193e-01 8.34393442e-01 -1.91095546e-01 -5.92697203e-01
-2.04054400e-01 3.96728694e-01 2.17852607e-01 7.31550232e-02
8.02287579e-01 -8.24477077e-01 -3.87308300e-02 -9.95795242e-04
3.38736117e-01 6.83720946e-01 -2.92538822e-01 -1.85756572e-02
-4.82191294e-01 -4.80000883e-01 5.63387692e-01 -6.02989495e-01
-8.87051225e-01 6.73823953e-01 7.67354190e-01 6.83802187e-01
9.47320282e-01 -1.92157477e-01 1.00233865e+00 6.93001091e-01
1.25782430e+00 -8.87729287e-01 -4.30344224e-01 8.80913675e-01
2.12585926e-01 -7.34902859e-01 2.77344406e-01 -8.58770728e-01
1.62334606e-01 1.47224927e+00 5.65194547e-01 5.28582633e-01
1.07302427e+00 1.37783751e-01 9.01049450e-02 -1.54332176e-01
-2.61187911e-01 -7.14710534e-01 -2.27970947e-02 8.25060964e-01
4.61216748e-01 3.22570980e-01 -6.82909727e-01 3.41337621e-01
-4.20066714e-01 -2.65912741e-01 1.06170893e+00 8.91145110e-01
-9.18700814e-01 -1.19092274e+00 -3.39048833e-01 8.35415602e-01
-4.20346260e-01 -9.13428515e-02 3.17705363e-01 6.82816088e-01
3.00917923e-01 1.25412977e+00 -2.72109807e-01 -3.32961828e-01
-6.12023994e-02 -5.01992822e-01 3.53824556e-01 -7.15272963e-01
7.12131619e-01 -1.14370503e-01 7.36155391e-01 3.14094603e-01
-3.95874470e-01 -6.55690908e-01 -1.11994016e+00 -3.26913863e-01
-9.39390361e-01 3.74147147e-01 2.24277660e-01 1.49160123e+00
4.33450162e-01 8.68179560e-01 6.72242701e-01 -7.89925098e-01
7.22297132e-02 -4.02913392e-01 -1.14264631e+00 -1.11221611e+00
4.23308283e-01 -9.10782754e-01 -3.34404796e-01 -7.13717520e-01] | [6.19819974899292, 1.3782732486724854] |
4f162081-7c93-412d-ab5a-02d0e99cd871 | an-incremental-iterated-response-model-of | 1810.00367 | null | http://arxiv.org/abs/1810.00367v2 | http://arxiv.org/pdf/1810.00367v2.pdf | An Incremental Iterated Response Model of Pragmatics | Recent Iterated Response (IR) models of pragmatics conceptualize language use
as a recursive process in which agents reason about each other to increase
communicative efficiency. These models are generally defined over complete
utterances. However, there is substantial evidence that pragmatic reasoning
takes place incrementally during production and comprehension. We address this
with an incremental IR model. We compare the incremental and global versions
using computational simulations, and we assess the incremental model against
existing experimental data and in the TUNA corpus for referring expression
generation, showing that the model can capture phenomena out of reach of global
versions. | ['Christopher Potts', 'Reuben Cohn-Gordon', 'Noah D. Goodman'] | 2018-09-30 | an-incremental-iterated-response-model-of-1 | https://aclanthology.org/W19-0109 | https://aclanthology.org/W19-0109.pdf | ws-2019-1 | ['referring-expression-generation'] | ['computer-vision'] | [ 2.73970544e-01 1.05524707e+00 -9.83660817e-02 -4.74848062e-01
-8.40880930e-01 -7.69768834e-01 1.04681361e+00 -1.62447188e-02
-3.55152994e-01 6.03661001e-01 1.11100233e+00 -5.12449563e-01
-3.04063708e-01 -5.09189963e-01 -1.87607065e-01 -2.73655176e-01
-8.33273120e-03 6.83864117e-01 1.08664200e-01 -8.51058185e-01
3.76259983e-01 1.53576255e-01 -9.48737502e-01 4.84608769e-01
4.42563832e-01 -8.41621459e-02 4.27202940e-01 8.27560306e-01
-3.42480183e-01 1.52604330e+00 -6.49164438e-01 -1.22635089e-01
-1.24783874e-01 -7.90679395e-01 -1.73354471e+00 -1.16507210e-01
-7.85716355e-01 -4.52793509e-01 -1.80423170e-01 6.46996856e-01
3.02970380e-01 1.86519355e-01 3.18668455e-01 -8.33500624e-01
-8.07294786e-01 1.34342611e+00 1.04351118e-01 1.00209124e-01
1.19063652e+00 3.11637074e-01 1.38415432e+00 -2.13743612e-01
1.08241487e+00 1.74216878e+00 4.28550512e-01 1.01396823e+00
-1.05651653e+00 3.00764531e-01 1.96934417e-01 -2.76913464e-01
-9.24297392e-01 -7.52857149e-01 6.91403091e-01 -1.15960404e-01
1.33562136e+00 4.54116762e-01 5.62672317e-01 8.87795687e-01
-2.25219667e-01 9.96760845e-01 1.34844840e+00 -8.46726656e-01
7.84308687e-02 -2.33236118e-04 4.56141621e-01 1.17067292e-01
-3.36836129e-01 2.01957315e-01 -5.82266510e-01 -3.73589754e-01
5.97233415e-01 -6.74345315e-01 -4.01829690e-01 2.90523022e-01
-1.19389451e+00 8.60621631e-01 4.31225859e-02 7.07911849e-01
-7.50037611e-01 2.97234982e-01 3.15050691e-01 7.79671371e-01
7.91657865e-02 8.57370436e-01 -6.98377967e-01 -9.12021160e-01
-8.75841081e-02 5.24085104e-01 1.29207051e+00 8.46727192e-01
3.08848470e-01 -5.08816957e-01 -3.27251218e-02 1.05026448e+00
7.72315443e-01 4.42231297e-01 5.46653986e-01 -1.70500600e+00
1.46971568e-01 6.01550996e-01 6.14527524e-01 -5.73644996e-01
-6.82768106e-01 5.76960325e-01 3.10394913e-01 -2.63116539e-01
6.10964179e-01 -1.82525963e-01 9.49831232e-02 2.04023910e+00
3.35649222e-01 -6.71327174e-01 7.81139433e-01 6.33881807e-01
9.00533020e-01 6.66728199e-01 4.08633918e-01 -7.10228205e-01
1.34016311e+00 -7.04459548e-01 -9.61098790e-01 -1.82128727e-01
1.18600917e+00 -7.37176955e-01 1.27526605e+00 2.79722720e-01
-1.52968097e+00 1.22886591e-01 -6.27351165e-01 -3.27038080e-01
2.69795984e-01 -3.56125653e-01 7.95511544e-01 5.86650789e-01
-1.23295283e+00 4.84224558e-01 -7.28650033e-01 -6.48134470e-01
-1.23927683e-01 1.62755400e-01 -1.15825133e-02 4.56296623e-01
-1.23203468e+00 1.03132439e+00 4.16323096e-02 2.90441304e-01
-3.95262301e-01 3.88825685e-02 -6.77980125e-01 -3.03506553e-01
2.19710603e-01 -5.83592474e-01 2.37195063e+00 -1.14998114e+00
-2.45363474e+00 9.00116086e-01 -3.14794451e-01 -4.96097445e-01
1.67812675e-01 -2.78382450e-01 -5.89815602e-02 2.91355580e-01
-1.83101565e-01 5.61770856e-01 2.03441292e-01 -1.18938005e+00
-3.19301367e-01 -1.79448083e-01 7.40043521e-01 8.51850986e-01
5.65510392e-01 7.15088248e-01 2.58038968e-01 -8.67457986e-02
3.82522970e-01 -9.63688672e-01 -1.17834389e-01 -5.35397291e-01
1.53319659e-02 -8.32659185e-01 4.11458910e-01 -5.01715481e-01
1.13222194e+00 -1.93112338e+00 2.53702909e-01 -3.04863781e-01
1.57300711e-01 -2.04641059e-01 -2.39839822e-01 1.04550505e+00
4.39505093e-02 3.95931751e-01 3.67757827e-02 -1.06842741e-01
3.32820892e-01 5.97846746e-01 -4.29855406e-01 3.05283457e-01
7.01507404e-02 1.16865206e+00 -9.47595119e-01 -2.93964982e-01
5.41714206e-02 1.18458197e-01 -7.59171724e-01 4.58638161e-01
-6.09409511e-01 5.96554339e-01 -8.34027588e-01 3.12180072e-01
1.81078594e-02 -3.41986477e-01 9.32257295e-01 5.88786185e-01
-4.31369305e-01 1.24933958e+00 -6.84920311e-01 1.43554175e+00
-4.94342595e-01 4.91542071e-01 3.64147782e-01 -7.06137717e-01
4.69830960e-01 8.55081677e-01 1.75795034e-02 -7.73710072e-01
2.30719984e-01 2.90420920e-01 6.19832218e-01 -7.08722591e-01
6.40491128e-01 -4.93572235e-01 -4.46781635e-01 1.11903417e+00
-4.42056328e-01 -8.10797036e-01 -2.34009162e-01 4.15585309e-01
1.05423510e+00 3.34244430e-01 6.78120732e-01 -4.90241647e-01
6.50239468e-01 2.25569338e-01 1.94453478e-01 8.44054580e-01
-2.81803071e-01 1.16616813e-02 6.62882626e-01 -4.42828685e-01
-8.16168785e-01 -7.72044778e-01 6.56011328e-02 1.30216646e+00
2.67392278e-01 -4.68682081e-01 -1.08561838e+00 -1.52345359e-01
-7.49069214e-01 1.40303171e+00 -2.53259867e-01 2.54936874e-01
-1.09401560e+00 -4.92564917e-01 7.50036836e-01 1.44870207e-01
2.46198326e-01 -1.67982483e+00 -1.38029385e+00 5.57943761e-01
-6.00489020e-01 -1.06505215e+00 8.51356238e-02 -2.46902496e-01
-6.91020250e-01 -1.03763604e+00 5.23574390e-02 -4.04741108e-01
1.54471502e-01 1.64482266e-01 1.13130498e+00 7.49149144e-01
4.10971493e-01 8.61869693e-01 -7.99693644e-01 -3.08934718e-01
-1.35090721e+00 -2.54337162e-01 -1.16185777e-01 -6.18652165e-01
2.42482379e-01 -3.66666168e-01 -4.02463019e-01 3.20366174e-01
-5.88665426e-01 3.15842152e-01 1.12576541e-02 7.91121006e-01
-2.55828857e-01 -6.66748047e-01 7.07540751e-01 -8.78605902e-01
1.43315661e+00 -3.49288702e-01 -1.40580133e-01 1.95081249e-01
-3.08918118e-01 1.91688538e-01 1.51870817e-01 -2.14768559e-01
-1.82953107e+00 -4.62096781e-01 -3.63570720e-01 8.12746823e-01
-2.53053427e-01 4.62091953e-01 1.83245853e-01 2.81130314e-01
7.01299012e-01 7.78149590e-02 4.62834328e-01 -3.79901022e-01
7.67501831e-01 9.54730272e-01 -6.21839911e-02 -1.20245898e+00
1.11909330e-01 3.48247796e-01 -3.87014061e-01 -1.22006476e+00
-6.37924254e-01 -2.00046033e-01 -3.10833514e-01 -2.03964010e-01
6.03149891e-01 -6.67689800e-01 -1.03387594e+00 2.84321904e-01
-1.54965961e+00 -9.60415125e-01 -6.77306175e-01 4.62132037e-01
-1.32381880e+00 5.08888543e-01 -1.03063655e+00 -1.37391210e+00
-1.88397229e-01 -1.12487781e+00 1.05996001e+00 2.28761107e-01
-1.27365875e+00 -1.14786661e+00 1.78488895e-01 2.16223732e-01
4.98090684e-01 -3.27445328e-01 8.50295782e-01 -9.44934905e-01
-4.18920457e-01 4.43464398e-01 1.46861628e-01 -7.37191439e-02
1.86738998e-01 -1.00694440e-01 -7.13891208e-01 3.33908588e-01
5.63656390e-01 -5.83769500e-01 -1.22154906e-01 1.11024976e-01
-6.23902678e-02 -6.46352828e-01 -1.83179989e-01 -3.76683414e-01
9.27384496e-01 5.15025854e-01 8.29896688e-01 1.22113295e-01
-1.69417620e-01 1.08838308e+00 6.64818287e-01 1.55945554e-01
8.25709343e-01 6.42903864e-01 -2.58707613e-01 6.63815618e-01
-1.24478832e-01 -2.66986370e-01 4.14787114e-01 1.26468885e+00
-1.25266895e-01 -3.04679602e-01 -9.58849251e-01 6.76217556e-01
-1.97525048e+00 -1.14729261e+00 -2.53645033e-01 1.63764656e+00
1.01744914e+00 -6.09611720e-02 -1.16004072e-01 -9.62149799e-02
4.14309889e-01 2.50713080e-01 2.25843683e-01 -1.02922249e+00
-1.47180125e-01 9.50449333e-02 -2.58064777e-01 1.13395464e+00
-2.18696788e-01 1.38831306e+00 8.31102085e+00 1.21262059e-01
-6.30509675e-01 4.27144170e-01 3.67578626e-01 2.27957830e-01
-4.33681488e-01 3.89391005e-01 -1.87867790e-01 -1.68699414e-01
1.30670047e+00 -2.92408526e-01 9.37732339e-01 3.07067215e-01
5.69461226e-01 -4.50268894e-01 -1.17289710e+00 4.20568347e-01
-3.63731056e-01 -1.10376835e+00 -1.92499667e-01 -1.03797577e-01
2.88284957e-01 2.80131791e-02 -5.98664463e-01 3.04686189e-01
7.83599734e-01 -6.69922948e-01 7.76689470e-01 5.05741775e-01
1.96570888e-01 5.57024591e-02 4.93770957e-01 7.47542679e-01
-6.22798920e-01 -2.77309418e-01 1.49291798e-01 -7.67138481e-01
8.60944927e-01 -5.13042688e-01 -7.20277131e-01 1.71816647e-01
2.44302705e-01 5.23830168e-02 1.11628391e-01 1.41713977e-01
-5.44999838e-01 8.50142241e-01 -6.54193819e-01 -5.86442828e-01
3.84551615e-01 -4.16049093e-01 1.04995632e+00 6.84848964e-01
-1.88776270e-01 7.87221014e-01 -1.84116632e-01 7.83323884e-01
2.83053070e-01 2.91035444e-01 -5.90741754e-01 -2.78659254e-01
4.88031834e-01 6.17866397e-01 -5.31797171e-01 -3.47607851e-01
-3.38790774e-01 7.49554217e-01 2.94503659e-01 1.81621611e-01
-4.83716786e-01 3.95996302e-01 4.19018418e-01 -2.05508173e-01
-1.56855881e-01 -3.37635815e-01 -8.63079503e-02 -8.50692630e-01
-3.45789135e-01 -1.05438387e+00 7.65239969e-02 -9.95771527e-01
-9.38472688e-01 5.11294901e-01 4.27256674e-01 -5.14030337e-01
-8.36475968e-01 -4.73872662e-01 -4.44780856e-01 7.24301636e-01
-9.33414280e-01 -8.91307235e-01 3.27514976e-01 -3.65070510e-03
8.20639908e-01 4.38676208e-01 1.21996498e+00 -3.89550000e-01
2.01971289e-02 -1.15538366e-01 -4.65309262e-01 -2.25287959e-01
-6.10121619e-03 -1.12601602e+00 4.69249785e-01 2.13874519e-01
-3.35457213e-02 1.26457787e+00 1.07066667e+00 -3.70090008e-01
-1.42908382e+00 4.81687076e-02 1.15184069e+00 -7.88182974e-01
8.29758227e-01 1.53838947e-01 -8.93859208e-01 1.01925313e+00
8.59195590e-01 -6.19511664e-01 5.63805044e-01 3.15733045e-01
-5.44770211e-02 6.15753293e-01 -1.17239809e+00 8.88709247e-01
1.31649137e+00 -6.30584478e-01 -1.69199145e+00 4.73234117e-01
1.13725734e+00 -4.34208393e-01 -6.95117950e-01 1.09989889e-01
5.71769714e-01 -8.11833084e-01 4.07132655e-01 -6.32059991e-01
1.95985526e-01 -1.33511648e-01 -3.57495159e-01 -1.13643444e+00
2.89454758e-01 -1.48157728e+00 2.18688458e-01 7.58366704e-01
6.18494511e-01 -9.51345742e-01 1.10937305e-01 1.15493190e+00
-7.74174184e-02 -2.96106368e-01 -1.20492768e+00 -2.48718649e-01
3.06725621e-01 -6.09391332e-01 6.70230627e-01 6.47557437e-01
1.16475046e+00 8.47317934e-01 1.45994648e-01 -3.30351055e-01
-1.63049065e-02 -2.52237525e-02 8.10842812e-01 -1.14621592e+00
-4.61601168e-01 -3.75756651e-01 2.88894862e-01 -1.62258565e+00
2.17940778e-01 -4.96543825e-01 4.00309652e-01 -1.57918155e+00
-1.16205983e-01 -5.23035109e-01 4.87594336e-01 1.92231908e-01
2.35307053e-01 -5.60925484e-01 3.45561385e-01 5.32897651e-01
-5.51864684e-01 5.18251777e-01 1.26029873e+00 3.83186996e-01
-6.34638071e-01 -2.73738801e-01 -1.12854207e+00 1.07292962e+00
8.17102730e-01 -4.34845984e-01 -4.74833459e-01 -3.65813553e-01
6.74094200e-01 6.53570354e-01 1.21828638e-01 -2.07191184e-01
1.23084493e-01 -3.96491408e-01 -9.11035895e-01 2.08890904e-02
3.12958151e-01 -3.05911213e-01 2.39344984e-01 3.43889266e-01
-7.38814652e-01 1.74865127e-01 3.78972143e-02 4.08847362e-01
1.52528629e-01 -5.91876328e-01 4.19336110e-01 -3.98777008e-01
-4.73936111e-01 -8.19412231e-01 -9.58496869e-01 1.19015917e-01
6.77735686e-01 -1.10807987e-02 -3.97925049e-01 -8.21524501e-01
-9.24923956e-01 3.82049158e-02 3.90181392e-01 3.12489957e-01
4.43597287e-01 -6.83873415e-01 -5.68278253e-01 -3.54218662e-01
-1.00471310e-01 -6.10303916e-02 -7.46722445e-02 8.27131152e-01
-8.31088543e-01 5.73694468e-01 5.11328638e-01 -4.01874542e-01
-9.62141454e-01 3.76239240e-01 5.67927420e-01 -2.44831324e-01
-8.40431452e-01 5.82293034e-01 2.11860925e-01 -8.46635163e-01
-4.92097586e-01 -3.24418306e-01 -3.89025658e-01 -2.61168331e-01
6.46156430e-01 -5.82287349e-02 -6.84213459e-01 -9.80610013e-01
-1.47089452e-01 2.63290912e-01 -4.29121451e-03 -8.84889483e-01
9.70387101e-01 -7.37091720e-01 -5.80322385e-01 9.22964692e-01
6.11346662e-01 4.22437161e-01 -6.78946137e-01 -1.51549920e-01
3.48221093e-01 1.11611802e-02 -4.01769459e-01 -8.00326109e-01
1.54025331e-01 2.12495238e-01 -1.44113913e-01 7.07006097e-01
5.54572642e-01 6.06957912e-01 4.65916425e-01 7.71042526e-01
7.64457464e-01 -1.31704521e+00 -1.47577003e-01 6.48466289e-01
1.17289650e+00 -6.47255003e-01 1.18281310e-02 -5.63421905e-01
-8.72945487e-01 9.50351715e-01 1.85369760e-01 -5.21177500e-02
4.57781851e-01 1.63288891e-01 4.45666760e-01 -5.95186889e-01
-1.37457919e+00 2.52364911e-02 -6.66471899e-01 4.00729746e-01
7.14371264e-01 4.20805216e-01 -1.25361919e+00 4.84897822e-01
-4.48404402e-01 2.12043477e-03 9.80531394e-01 1.12998271e+00
-3.71362180e-01 -1.28915429e+00 -2.45472640e-01 -7.94372261e-02
-3.41451377e-01 -1.63116887e-01 -6.99032903e-01 1.00518131e+00
-6.96663022e-01 1.64786875e+00 1.08541280e-01 2.62507617e-01
3.34080219e-01 2.88200617e-01 6.18009984e-01 -7.50517488e-01
-9.42916453e-01 -1.08254030e-01 1.00262201e+00 -5.80661654e-01
-1.04874492e+00 -1.12317419e+00 -1.76742959e+00 1.58900060e-02
-5.72365403e-01 5.27174592e-01 4.05009568e-01 1.27676523e+00
1.28969684e-01 1.75810546e-01 5.78131080e-01 -4.57572758e-01
-9.40644383e-01 -1.33602834e+00 -9.49262604e-02 4.64111596e-01
1.62829444e-01 -2.69753605e-01 -6.79477870e-01 -1.34523407e-01] | [10.743995666503906, 8.443174362182617] |
649d0c09-e970-48db-b294-8b52dd4cee13 | modeling-dynamic-attributes-for-next-basket | 2109.11654 | null | https://arxiv.org/abs/2109.11654v1 | https://arxiv.org/pdf/2109.11654v1.pdf | Modeling Dynamic Attributes for Next Basket Recommendation | Traditional approaches to next item and next basket recommendation typically extract users' interests based on their past interactions and associated static contextual information (e.g. a user id or item category). However, extracted interests can be inaccurate and become obsolete. Dynamic attributes, such as user income changes, item price changes (etc.), change over time. Such dynamics can intrinsically reflect the evolution of users' interests. We argue that modeling such dynamic attributes can boost recommendation performance. However, properly integrating them into user interest models is challenging since attribute dynamics can be diverse such as time-interval aware, periodic patterns (etc.), and they represent users' behaviors from different perspectives, which can happen asynchronously with interactions. Besides dynamic attributes, items in each basket contain complex interdependencies which might be beneficial but nontrivial to effectively capture. To address these challenges, we propose a novel Attentive network to model Dynamic attributes (named AnDa). AnDa separately encodes dynamic attributes and basket item sequences. We design a periodic aware encoder to allow the model to capture various temporal patterns from dynamic attributes. To effectively learn useful item relationships, intra-basket attention module is proposed. Experimental results on three real-world datasets demonstrate that our method consistently outperforms the state-of-the-art. | ['Caiming Xiong', 'Julian McAuley', 'Markus Anderle', 'Chenxi Li', 'Chenghao Liu', 'Jia Li', 'Yongjun Chen'] | 2021-09-23 | null | null | null | null | ['next-basket-recommendation'] | ['miscellaneous'] | [-1.94395721e-01 -6.30644917e-01 -8.38367224e-01 -7.34243393e-01
-2.24753767e-01 -6.05238378e-01 3.05042237e-01 2.31134862e-01
-2.23274916e-01 9.19672251e-01 7.34233320e-01 4.38438915e-02
-3.62667948e-01 -1.03042829e+00 -8.71574223e-01 -2.67700136e-01
-4.82874751e-01 4.99429524e-01 7.23771229e-02 -4.39376652e-01
1.10482886e-01 8.66337642e-02 -1.79984164e+00 5.02560794e-01
8.29709888e-01 1.18210208e+00 2.89593667e-01 3.76806796e-01
-4.84796435e-01 8.13193977e-01 -5.11475086e-01 -3.68919194e-01
1.97798461e-01 -3.04460734e-01 -4.51919079e-01 -1.64754957e-01
-1.74016029e-01 -5.57871282e-01 -3.29103261e-01 6.07566178e-01
3.00025910e-01 2.83143133e-01 4.20586914e-01 -1.13315463e+00
-1.32186651e+00 1.31570470e+00 -5.55582345e-01 4.88169909e-01
4.68631983e-01 4.41909283e-02 1.32482648e+00 -7.33492136e-01
4.60783333e-01 9.66924727e-01 6.72787726e-01 2.49563187e-01
-9.94801164e-01 -8.55356455e-01 1.17351496e+00 4.99082118e-01
-9.92374003e-01 -2.63234705e-01 8.77563119e-01 -1.64050713e-01
8.07555258e-01 4.38296974e-01 1.02927220e+00 1.61078048e+00
-1.42223341e-02 1.29598689e+00 4.31975275e-01 1.71527460e-01
1.97603013e-02 5.90743087e-02 4.94245142e-01 -1.01880535e-01
3.00009251e-01 -1.56684577e-01 -6.42186224e-01 -1.44115165e-01
5.72586060e-01 8.91141236e-01 -2.20273018e-01 -1.44118696e-01
-1.31163275e+00 7.39864886e-01 3.23498935e-01 2.28474781e-01
-7.94266105e-01 -1.95426140e-02 3.44755828e-01 5.82428813e-01
5.32616377e-01 2.71089315e-01 -8.93884420e-01 -4.07515466e-01
-4.32605326e-01 2.31407717e-01 8.66023123e-01 1.37135839e+00
5.45780778e-01 1.63890436e-01 -3.25670838e-01 8.32378626e-01
3.00098956e-01 3.82513642e-01 7.96181083e-01 -3.81691873e-01
3.92862678e-01 5.21624088e-01 4.18226570e-01 -1.04929733e+00
-2.02942654e-01 -7.05939710e-01 -9.59106207e-01 -7.24554360e-01
3.01523298e-01 -1.45031780e-01 -8.53928745e-01 1.83714879e+00
1.98155582e-01 4.82836813e-01 -3.20078373e-01 7.98304498e-01
6.48611426e-01 6.86715722e-01 2.78934628e-01 -7.06117034e-01
1.13474429e+00 -6.20338023e-01 -9.96963084e-01 9.49421078e-02
3.02023798e-01 -5.26046813e-01 1.13089216e+00 2.22286433e-01
-1.19258356e+00 -6.18658602e-01 -7.36512303e-01 1.92295954e-01
-3.77249062e-01 -2.00575098e-01 9.87252176e-01 1.50301903e-01
-5.19373238e-01 5.57765484e-01 -8.03593338e-01 -1.50883242e-01
2.60479569e-01 5.03854275e-01 3.40532750e-01 2.91139275e-01
-1.70782769e+00 2.13333219e-01 2.17624798e-01 5.56478510e-03
-4.66400534e-01 -1.00608885e+00 -5.82804441e-01 3.79762501e-01
5.81365585e-01 -7.50053048e-01 1.38915312e+00 -1.07709467e+00
-1.29832721e+00 -2.77934615e-02 -2.62372464e-01 -4.56443846e-01
2.50911802e-01 -4.49458063e-01 -9.83515739e-01 -6.58493698e-01
-3.72409597e-02 -1.12531967e-01 5.65617621e-01 -1.00260031e+00
-1.17029393e+00 -3.61356825e-01 2.85342664e-01 3.47260803e-01
-7.31248140e-01 -9.56174657e-02 -6.59053326e-01 -1.02036834e+00
-2.85157233e-01 -7.20746577e-01 -2.42859945e-01 -5.57363510e-01
-1.30585253e-01 -2.74630636e-01 7.45977283e-01 -3.34019899e-01
1.99833763e+00 -1.93184793e+00 -2.23779157e-01 2.01599255e-01
4.05943170e-02 -8.51938222e-03 7.29791299e-02 4.20295388e-01
9.70667601e-02 7.81559944e-02 3.02125633e-01 -3.88187058e-02
6.54947609e-02 5.05989730e-01 -4.52066541e-01 3.15914676e-02
-3.40561569e-01 1.17999017e+00 -1.10174942e+00 -6.02453351e-02
2.43567415e-02 5.45244634e-01 -7.23855972e-01 2.79080659e-01
-4.29352671e-01 2.84630567e-01 -9.79007900e-01 7.71904767e-01
4.54201370e-01 -6.02704108e-01 2.36443862e-01 -2.11655855e-01
4.45610173e-02 6.42953157e-01 -1.29505312e+00 1.40038025e+00
-7.35584617e-01 1.41418695e-01 -3.14365804e-01 -1.01364112e+00
6.81583583e-01 2.98940778e-01 9.56338644e-01 -8.53197336e-01
-9.19790268e-02 -9.25290585e-02 9.08370763e-02 -4.64247793e-01
7.59708464e-01 2.71914840e-01 -1.70976594e-01 6.58747077e-01
-2.44489804e-01 1.14536595e+00 2.34994352e-01 9.96252075e-02
7.98997283e-01 -6.56686956e-03 4.34587210e-01 -7.74490237e-02
2.90344775e-01 -3.50071102e-01 9.99335229e-01 9.03668225e-01
1.82762537e-02 3.03948998e-01 1.37930050e-01 -8.70498598e-01
-9.14309323e-01 -8.50923538e-01 -2.25873739e-02 1.75693715e+00
3.44428897e-01 -5.21042943e-01 4.90122102e-02 -7.86713421e-01
2.79746562e-01 5.96293211e-01 -8.74534369e-01 -1.77733958e-01
-7.23960280e-01 -1.13300860e+00 -2.50733703e-01 8.83547604e-01
1.81256846e-01 -1.34261239e+00 -2.60642916e-01 9.58154678e-01
-4.00815189e-01 -6.93904698e-01 -9.79355395e-01 2.66853767e-03
-6.05907977e-01 -8.03099692e-01 -5.38570702e-01 -4.14515674e-01
1.66084170e-01 4.69909817e-01 1.76016450e+00 -4.29624580e-02
1.88730419e-01 2.45619774e-01 -7.71186411e-01 -6.28966808e-01
1.73643604e-01 2.62635946e-01 3.98074299e-01 4.57775444e-01
7.48071015e-01 -9.31679070e-01 -1.14258468e+00 4.90610927e-01
-8.30225646e-01 -1.16868436e-01 4.56595927e-01 8.17518234e-01
8.43335390e-01 1.13697080e-02 9.81713176e-01 -1.57167041e+00
8.09453189e-01 -1.17505395e+00 -1.40045613e-01 4.34866577e-01
-9.47917223e-01 -2.05515176e-01 1.08826911e+00 -1.03259039e+00
-1.11894989e+00 -4.16072994e-01 -1.29452705e-01 -3.28960747e-01
-1.39035523e-01 7.92398810e-01 -2.85064399e-01 8.13969076e-01
1.92224786e-01 2.60977060e-01 -6.49180532e-01 -7.03109801e-01
3.45058471e-01 7.26699293e-01 4.12630260e-01 -6.01510882e-01
2.99572676e-01 2.59431779e-01 -6.89682543e-01 -1.60245299e-01
-1.11002779e+00 -8.29103827e-01 -2.84741133e-01 3.58333290e-02
9.94437560e-02 -1.02297854e+00 -7.99961865e-01 3.52673978e-01
-7.30384290e-01 -1.21648125e-01 -5.99436879e-01 5.97441614e-01
-3.62845272e-01 -9.30267666e-03 -8.06448817e-01 -6.79084718e-01
-4.79244441e-01 -8.40261698e-01 7.26265490e-01 3.90004843e-01
-3.12716693e-01 -1.09847665e+00 -1.37805223e-01 -1.74087435e-01
6.50772870e-01 2.62569226e-02 7.67618120e-01 -8.22652698e-01
-5.94553590e-01 -1.26366109e-01 -3.13500501e-02 -3.34153861e-01
7.06628323e-01 -1.34392992e-01 -5.26524186e-01 -3.90017122e-01
-2.24012092e-01 1.28657490e-01 6.55478358e-01 4.13355768e-01
1.63532090e+00 -8.64537716e-01 -4.35514688e-01 7.46840179e-01
1.23642969e+00 6.82805955e-01 3.36292982e-01 2.37053156e-01
8.28420579e-01 1.61599398e-01 7.75576413e-01 1.04203165e+00
7.82592773e-01 9.57039297e-01 3.36834729e-01 1.10322669e-01
1.48388743e-01 -3.67137581e-01 3.54128391e-01 1.20294905e+00
-4.18176144e-01 -6.56477511e-01 -3.32076132e-01 6.01771057e-01
-2.27866340e+00 -1.41459751e+00 -1.66470036e-01 2.20920920e+00
9.43420112e-01 9.01854113e-02 5.01782715e-01 -1.36590809e-01
4.93239313e-01 2.73553450e-02 -1.14531231e+00 -1.89564124e-01
-2.46005245e-02 -1.98606834e-01 3.78607541e-01 4.61410992e-02
-9.54413593e-01 6.37566566e-01 5.98441839e+00 5.33131957e-01
-9.08799350e-01 -9.89167020e-03 6.15791261e-01 -4.21286255e-01
-7.29923308e-01 -4.55502778e-01 -1.06495166e+00 1.17167354e+00
9.54465926e-01 -5.28017282e-01 5.49911261e-01 7.85608709e-01
1.94469079e-01 5.48102021e-01 -1.12329924e+00 8.77066672e-01
-9.16806981e-02 -1.29314148e+00 2.87047625e-01 1.32906109e-01
7.51663029e-01 1.26164537e-02 2.13002220e-01 7.23868728e-01
7.59123206e-01 -5.19210756e-01 4.19023633e-01 8.40279877e-01
6.00709796e-01 -1.07908034e+00 6.09253585e-01 3.21686357e-01
-1.76183236e+00 -3.79069030e-01 -4.11823928e-01 1.02263680e-02
2.97599763e-01 7.37367392e-01 -2.45347455e-01 6.27643526e-01
1.01875401e+00 1.48206174e+00 -1.88885421e-01 9.32368577e-01
2.66661704e-01 8.87532711e-01 -2.79024929e-01 -1.18592389e-01
1.24464326e-01 -2.79571712e-01 3.49595368e-01 1.15861988e+00
6.29845560e-01 4.60922211e-01 2.75942177e-01 4.59270805e-01
-2.40061104e-01 2.35077441e-01 -4.60489869e-01 9.08946022e-02
6.06204331e-01 1.01275992e+00 -3.70277226e-01 -5.70563674e-01
-8.81913602e-01 7.84251809e-01 2.25980446e-01 5.34407377e-01
-9.61959779e-01 -1.68890268e-01 1.22086275e+00 4.65915114e-01
5.64877689e-01 1.36276871e-01 3.86247002e-02 -1.58996904e+00
1.20609678e-01 -9.53845203e-01 8.45732391e-01 -3.10149312e-01
-1.77735579e+00 4.86451894e-01 -9.83887985e-02 -1.54837036e+00
-4.57172930e-01 -4.91456268e-03 -6.15928710e-01 5.73512375e-01
-1.37285280e+00 -1.08233130e+00 -1.84363976e-01 9.26199377e-01
9.67752218e-01 -2.00481191e-01 5.34268677e-01 5.61502159e-01
-4.26502645e-01 8.70097756e-01 3.88672352e-01 -1.85666103e-02
6.35824680e-01 -1.29598117e+00 6.07313812e-01 4.93975908e-01
3.26361328e-01 8.81390214e-01 4.23555553e-01 -6.75316870e-01
-1.52372706e+00 -1.26816010e+00 8.87111247e-01 -4.93077934e-01
7.43637502e-01 -2.80938506e-01 -1.23845255e+00 9.58347738e-01
2.65907496e-02 -5.02001345e-02 1.11445570e+00 7.11294174e-01
-2.24567965e-01 -2.54256606e-01 -6.53080463e-01 5.57856977e-01
1.56862271e+00 -2.00713545e-01 -4.61915821e-01 2.47206971e-01
9.90734458e-01 -2.26865530e-01 -1.02660751e+00 4.09333557e-01
8.29370201e-01 -7.75607526e-01 1.23799610e+00 -1.02999520e+00
9.08180028e-02 -1.04215801e-01 -6.23706318e-02 -1.55795813e+00
-8.39303613e-01 -7.39401102e-01 -9.79574442e-01 1.29703593e+00
3.58154625e-01 -4.56070542e-01 7.12890327e-01 8.28168809e-01
-3.17947567e-02 -9.66170967e-01 -2.88190842e-01 -6.15534723e-01
-4.43142384e-01 -3.67688209e-01 1.53535426e+00 1.22285569e+00
1.72993466e-01 3.85303885e-01 -1.01108289e+00 -6.05993010e-02
2.58822829e-01 8.08600247e-01 5.27853370e-01 -1.40129387e+00
-5.30820310e-01 -5.37923217e-01 8.49987380e-03 -1.32305300e+00
-1.35655895e-01 -5.89140177e-01 -3.85294229e-01 -1.32133245e+00
2.18167678e-01 -7.85225153e-01 -1.05493939e+00 4.86538082e-01
-5.00645936e-01 -1.05077714e-01 -4.27687876e-02 3.92900795e-01
-8.90568733e-01 4.81435984e-01 1.25442886e+00 -1.43703744e-01
-7.05480754e-01 6.29466772e-01 -1.06100214e+00 5.57154596e-01
8.40589941e-01 -3.36487830e-01 -8.07908833e-01 -3.31067711e-01
6.54136956e-01 1.33303970e-01 -3.53324145e-01 -5.48059940e-01
1.85814902e-01 -4.96672571e-01 5.71713269e-01 -7.62094915e-01
1.61314294e-01 -9.17219698e-01 4.69366282e-01 -2.78248880e-02
-5.12677312e-01 3.38555306e-01 -2.06647620e-01 1.06844342e+00
-2.12259725e-01 3.50404948e-01 3.75454798e-02 -2.10685968e-01
-8.39010000e-01 1.17323792e+00 -1.32434353e-01 9.59346071e-02
7.87907958e-01 -1.72769397e-01 6.53717592e-02 -7.40788043e-01
-9.12197530e-01 5.55496156e-01 1.46475717e-01 8.96047473e-01
5.00078976e-01 -1.66751862e+00 -5.56389511e-01 2.09731430e-01
2.74266094e-01 -2.55537480e-01 3.85393381e-01 4.50332165e-01
4.36209261e-01 1.97558850e-01 -1.62840679e-01 -1.77423283e-01
-9.94293451e-01 1.00742912e+00 -2.63488721e-02 -4.86796439e-01
-6.04831278e-01 6.03783906e-01 2.26822257e-01 -1.08014971e-01
4.02542412e-01 -4.92873192e-01 -8.15186203e-01 4.95664269e-01
8.11696768e-01 1.68832615e-01 -1.38595909e-01 -4.13249016e-01
2.40753349e-02 3.40510219e-01 -6.42011940e-01 6.39786780e-01
1.59039545e+00 -7.02278733e-01 4.39846992e-01 7.83816278e-01
9.53870118e-01 -4.56304662e-02 -1.23285449e+00 -8.15020740e-01
-1.60468057e-01 -7.01973796e-01 -1.99318022e-01 -7.81417310e-01
-1.44174325e+00 3.81141990e-01 2.45630339e-01 8.26952934e-01
1.31543350e+00 -1.98087513e-01 1.41145432e+00 3.19742680e-01
7.04692423e-01 -1.11012292e+00 -1.49974316e-01 4.72544283e-01
6.00630760e-01 -1.27861261e+00 -1.02175072e-01 -1.58194184e-01
-8.25217724e-01 7.03132331e-01 7.69368470e-01 1.88806608e-01
1.14683759e+00 1.44415110e-01 -5.97328581e-02 1.68353170e-01
-1.18530965e+00 -3.11249256e-01 4.21467483e-01 4.00983244e-01
6.93463206e-01 3.00773352e-01 -3.20068777e-01 1.27219939e+00
-9.25422609e-02 9.87444744e-02 3.04060429e-01 6.43230438e-01
-1.66029289e-01 -1.31468785e+00 1.25949875e-01 1.10663152e+00
-7.95712292e-01 -2.88670808e-02 1.50212318e-01 5.33494413e-01
2.06773028e-01 6.14553332e-01 3.33799481e-01 -4.83997971e-01
5.23705900e-01 -9.39147696e-02 -6.02146946e-02 -4.46110904e-01
-7.50990033e-01 1.39985040e-01 2.53828522e-02 -6.50547266e-01
-3.21154386e-01 -8.11611831e-01 -8.72297823e-01 -5.74241042e-01
-1.56401813e-01 2.32303515e-01 9.61663797e-02 6.69928491e-01
5.92642009e-01 6.99232936e-01 9.81677949e-01 -4.94771481e-01
-3.56020838e-01 -8.12442482e-01 -6.69064939e-01 9.31564867e-01
4.44060266e-01 -7.85477817e-01 4.51752841e-02 2.13456854e-01] | [10.09586238861084, 5.588942527770996] |
10cff777-da23-497e-8cb8-b49bba33e01f | efficient-novelty-detection-methods-for-early | 2208.04732 | null | https://arxiv.org/abs/2208.04732v1 | https://arxiv.org/pdf/2208.04732v1.pdf | Efficient Novelty Detection Methods for Early Warning of Potential Fatal Diseases | Fatal diseases, as Critical Health Episodes (CHEs), represent real dangers for patients hospitalized in Intensive Care Units. These episodes can lead to irreversible organ damage and death. Nevertheless, diagnosing them in time would greatly reduce their inconvenience. This study therefore focused on building a highly effective early warning system for CHEs such as Acute Hypotensive Episodes and Tachycardia Episodes. To facilitate the precocity of the prediction, a gap of one hour was considered between the observation periods (Observation Windows) and the periods during which a critical event can occur (Target Windows). The MIMIC II dataset was used to evaluate the performance of the proposed system. This system first includes extracting additional features using three different modes. Then, the feature selection process allowing the selection of the most relevant features was performed using the Mutual Information Gain feature importance. Finally, the high-performance predictive model LightGBM was used to perform episode classification. This approach called MIG-LightGBM was evaluated using five different metrics: Event Recall (ER), Reduced Precision (RP), average Anticipation Time (aveAT), average False Alarms (aveFA), and Event F1-score (EF1-score). A method is therefore considered highly efficient for the early prediction of CHEs if it exhibits not only a large aveAT but also a large EF1-score and a low aveFA. Compared to systems using Extreme Gradient Boosting, Support Vector Classification or Naive Bayes as a predictive model, the proposed system was found to be highly dominant. It also confirmed its superiority over the Layered Learning approach. | ['Ernest Fokoué', 'Sèdjro Salomon Hotegni'] | 2022-08-06 | null | null | null | null | ['episode-classification'] | ['time-series'] | [ 1.65907994e-01 -1.90640818e-02 8.35183635e-02 -2.16291457e-01
-5.54154336e-01 -6.63381293e-02 4.72463518e-01 9.36517715e-01
-4.99851495e-01 9.37747359e-01 -5.01646101e-02 -4.57521081e-01
-6.79122508e-01 -7.08297431e-01 -3.62373628e-02 -7.73011923e-01
-5.24551451e-01 4.95535791e-01 1.22004963e-01 2.97962129e-01
3.99251461e-01 5.58323264e-01 -1.54655051e+00 5.84971309e-01
8.23172390e-01 1.11842251e+00 1.34908497e-01 5.22535384e-01
6.28527850e-02 1.11683476e+00 -4.88981783e-01 -2.41524521e-02
7.38459155e-02 -6.20726287e-01 -5.37592232e-01 -5.04861116e-01
-6.80869520e-01 -1.92678407e-01 4.21034604e-01 3.18020850e-01
6.88719571e-01 1.85794145e-01 7.41659760e-01 -1.12225449e+00
6.50685906e-01 4.22005147e-01 -1.19244657e-01 4.86997277e-01
4.56724286e-01 1.55960366e-01 5.65063834e-01 -8.39538991e-01
1.83660135e-01 6.96394622e-01 6.77692473e-01 -1.10890195e-02
-9.67457771e-01 -5.19998610e-01 -1.11924410e-01 4.37049925e-01
-1.17967296e+00 -8.17007050e-02 4.40689296e-01 -6.50071561e-01
1.09190667e+00 4.84691411e-01 9.28377569e-01 6.04756355e-01
7.67358303e-01 2.48515621e-01 1.07251287e+00 -5.79315305e-01
4.43654150e-01 4.07252252e-01 2.52351463e-01 3.94527644e-01
3.18062603e-01 3.84761572e-01 -3.47103566e-01 -4.83597100e-01
2.23335892e-01 2.77541131e-01 -1.85465425e-01 2.31988505e-01
-1.07878053e+00 7.30169237e-01 1.49548858e-01 3.88403863e-01
-9.90660429e-01 -5.28578520e-01 6.47045672e-01 1.71131864e-01
1.41437426e-01 4.15264249e-01 -3.95493925e-01 -2.83362746e-01
-7.88412392e-01 -9.94580165e-02 7.20770657e-01 1.14546128e-01
3.04595023e-01 -2.20491841e-01 -4.80527550e-01 4.72053081e-01
2.47828513e-02 5.53126782e-02 5.65627575e-01 -2.21543372e-01
2.35451281e-01 9.99124348e-01 1.81276664e-01 -9.59625542e-01
-8.35861623e-01 -6.24558389e-01 -9.31023061e-01 -4.99231368e-02
1.03587933e-01 -4.00920957e-01 -4.44112003e-01 1.38241553e+00
3.33519578e-01 5.56560792e-02 2.91548312e-01 5.51089227e-01
4.57416594e-01 7.32753396e-01 4.67722505e-01 -8.10963809e-01
1.28912675e+00 -2.40753904e-01 -6.72873557e-01 3.72980572e-02
6.82933033e-01 -7.41116166e-01 3.68851781e-01 5.76870263e-01
-8.45988095e-01 -3.35090697e-01 -7.92235315e-01 8.09083760e-01
-2.27331594e-01 2.11384937e-01 4.94025171e-01 3.46460968e-01
-3.62448782e-01 6.64933980e-01 -8.34567904e-01 -4.18765843e-01
1.76267661e-02 3.56458724e-01 -1.31730840e-01 1.87720597e-01
-1.32392526e+00 1.22438991e+00 6.79099381e-01 7.22048506e-02
-6.73033476e-01 -6.46470010e-01 -4.22988892e-01 1.43362939e-01
2.03350466e-02 -6.84950352e-01 6.02034271e-01 -6.13771260e-01
-8.72868836e-01 5.20632923e-01 -6.85057743e-03 -7.94510067e-01
6.03564322e-01 -3.59465629e-01 -3.80753279e-01 4.07719940e-01
-2.11088479e-01 -1.42150789e-01 5.72146058e-01 -8.31048846e-01
-8.88271749e-01 -3.71231169e-01 -1.45939335e-01 3.63744408e-01
-6.83625489e-02 1.31976649e-01 5.00093341e-01 -3.58173817e-01
4.47920039e-02 -7.32094407e-01 -3.08598697e-01 -6.78523362e-01
-1.07200488e-01 -1.81007564e-01 5.57256520e-01 -7.50823200e-01
1.64248013e+00 -2.09434414e+00 -3.06194097e-01 5.53869426e-01
4.57070209e-02 4.47386920e-01 6.71553433e-01 8.23102534e-01
-2.72553355e-01 -2.90817291e-01 -1.23385020e-01 4.45373058e-01
-5.86202025e-01 -1.22308675e-02 -9.23074484e-02 2.66443729e-01
1.06219381e-01 4.16213155e-01 -6.85128331e-01 -6.04528248e-01
6.89383566e-01 4.06505793e-01 -2.32723892e-01 6.88657165e-01
2.84085870e-01 4.22556967e-01 -4.35639739e-01 3.88070405e-01
7.42099062e-02 -2.17335254e-01 3.93375993e-01 -1.06883503e-01
-3.90710056e-01 2.86300927e-02 -1.10451663e+00 4.78757620e-01
-5.29020369e-01 1.51477024e-01 -6.18705332e-01 -1.10200274e+00
1.29245853e+00 8.56304884e-01 8.80380034e-01 -5.37389576e-01
2.01734036e-01 1.95518613e-01 5.74249476e-02 -8.20890665e-01
-1.38627112e-01 -2.84559339e-01 2.17778668e-01 1.95399299e-01
-4.45576221e-01 4.38209385e-01 1.26986891e-01 1.08823694e-01
1.04416609e+00 -2.16327429e-01 1.09993482e+00 -3.11002731e-01
6.03348672e-01 -2.89127287e-02 6.82169437e-01 5.43100834e-01
-8.43083113e-02 2.15459570e-01 6.05754197e-01 -6.91535532e-01
-4.47187781e-01 -8.12766492e-01 -4.06014353e-01 4.20752645e-01
-2.31766459e-02 -2.10063428e-01 -3.90217096e-01 -4.57495421e-01
-2.21664667e-01 1.06473708e+00 -3.71110648e-01 -5.20383894e-01
-4.26290959e-01 -1.34619653e+00 1.98345378e-01 3.62719268e-01
3.94685477e-01 -1.19285762e+00 -1.58617032e+00 5.22465169e-01
-3.87466282e-01 -4.69410121e-01 3.77455354e-01 6.31392002e-01
-1.02545607e+00 -1.40790629e+00 -5.17962515e-01 -3.04609418e-01
4.38084781e-01 -1.63598970e-01 9.46328342e-01 4.23646010e-02
-5.97556412e-01 5.10142818e-02 -5.34538388e-01 -6.47841692e-01
-4.65044349e-01 -3.07716876e-01 4.40949295e-03 1.16983131e-01
6.47032142e-01 -4.05398667e-01 -9.93176103e-01 2.31249332e-01
-6.39962077e-01 -1.22146374e-02 7.23498523e-01 8.46289158e-01
4.85832840e-01 -1.76544264e-02 1.03602827e+00 -7.28729367e-01
6.71139061e-01 -7.14562595e-01 -4.16579604e-01 3.31549615e-01
-1.04147494e+00 -3.28575730e-01 5.95350325e-01 -2.09955201e-01
-9.45301890e-01 -3.74400280e-02 -4.06063274e-02 3.29583511e-02
-2.93531001e-01 5.67173719e-01 3.07107329e-01 4.07666147e-01
6.59705460e-01 1.39914840e-01 -7.60132223e-02 -1.24950118e-01
-7.27197111e-01 7.45639682e-01 -8.65852013e-02 -7.20300153e-02
1.01174995e-01 1.84584446e-02 3.25857878e-01 -7.64804721e-01
-5.72583139e-01 -7.88362026e-01 -4.40937400e-01 -6.05727255e-01
7.43672371e-01 -7.96360731e-01 -7.43115008e-01 2.17569634e-01
-8.43466401e-01 1.91171587e-01 -1.38849854e-01 9.65175867e-01
-3.85552853e-01 2.54473597e-01 -1.71992376e-01 -1.26154578e+00
-6.99340641e-01 -7.17536271e-01 2.68358320e-01 8.15396234e-02
-7.21574247e-01 -8.78728211e-01 2.18652755e-01 -5.13611101e-02
2.10432962e-01 8.07242751e-01 1.18844700e+00 -1.11153519e+00
-1.23207711e-01 -5.29812396e-01 -3.17238048e-02 3.19562793e-01
1.73410743e-01 1.73543040e-02 -6.71882510e-01 -1.46656275e-01
1.12435140e-01 6.38021082e-02 5.91507196e-01 6.05218053e-01
7.24300623e-01 -2.26545006e-01 -6.08659267e-01 3.00590962e-01
1.60201228e+00 1.13677847e+00 6.17874265e-01 4.02437419e-01
4.96120192e-04 5.42695522e-01 9.94614065e-01 9.50721085e-01
3.91037427e-02 5.09894609e-01 4.25652474e-01 -1.81218952e-01
4.39401239e-01 1.78857192e-01 1.49278060e-01 5.51275253e-01
-3.55549961e-01 -1.96145475e-01 -1.18346298e+00 6.35846376e-01
-1.89727104e+00 -1.10563207e+00 -3.02147090e-01 2.56578898e+00
3.46742332e-01 2.50105292e-01 1.06762126e-01 7.13871062e-01
6.53437912e-01 -2.76709527e-01 -2.20978409e-01 -7.46700585e-01
1.33859679e-01 -1.29908040e-01 -4.73431731e-03 1.20716311e-01
-1.03614640e+00 9.26911086e-02 5.43571711e+00 2.84004539e-01
-9.67322290e-01 -1.74659714e-01 7.62026787e-01 -2.16387492e-02
4.63715851e-01 -3.42760608e-02 -6.71953678e-01 7.39902079e-01
1.29831207e+00 -1.60296053e-01 -5.44274002e-02 8.62358451e-01
4.63014066e-01 -5.12146235e-01 -9.57683563e-01 7.49938369e-01
-1.96147889e-01 -8.28934014e-01 -1.50960330e-02 -2.24872544e-01
3.11391532e-01 -3.93369287e-01 -4.46862191e-01 7.04268292e-02
-1.85766205e-01 -5.64377248e-01 2.06127986e-01 9.05408502e-01
5.96471965e-01 -9.30440903e-01 1.26149118e+00 4.92445648e-01
-1.00749385e+00 -4.15840745e-01 -7.27545768e-02 -2.30421901e-01
2.89538443e-01 1.03886652e+00 -1.23286283e+00 6.95621252e-01
5.25723815e-01 2.35181868e-01 -2.28868321e-01 1.38299739e+00
5.43813966e-02 6.49222195e-01 -4.31785822e-01 -1.90984622e-01
1.48244482e-02 2.30661929e-02 6.03587747e-01 1.20293975e+00
5.57922602e-01 3.45753431e-01 5.51977428e-03 1.30354136e-01
6.81152642e-01 5.79033375e-01 -7.22942591e-01 3.89793634e-01
3.75171721e-01 1.00725615e+00 -9.30406511e-01 -4.86916512e-01
-9.96749029e-02 4.83856410e-01 -1.72095731e-01 -8.60663131e-02
-8.77879441e-01 -4.48120028e-01 5.48647419e-02 3.15103620e-01
-2.53071450e-02 5.26549160e-01 -3.95930529e-01 -7.42162943e-01
-1.36152133e-01 -5.57260990e-01 8.36325109e-01 -4.75639194e-01
-8.34790945e-01 9.06213582e-01 2.38636434e-01 -1.50994635e+00
-5.69025815e-01 -1.75762251e-01 -6.68068826e-01 8.55931878e-01
-1.16079724e+00 -4.48428988e-01 -2.94372797e-01 5.56287169e-01
3.17650765e-01 -1.05827086e-01 1.15671957e+00 1.88373134e-01
-6.25349402e-01 1.90757021e-01 1.20169766e-01 -2.55845398e-01
4.08229560e-01 -9.60463226e-01 -6.83661282e-01 6.22450054e-01
-5.95584095e-01 3.51118237e-01 7.41219759e-01 -7.59088337e-01
-4.80653107e-01 -1.02411282e+00 1.30079353e+00 -1.99291334e-02
9.88318622e-02 4.09336150e-01 -7.57871270e-01 3.39292794e-01
-2.03209102e-01 -4.64438230e-01 8.22038889e-01 -8.53329673e-02
3.82600635e-01 -1.90537915e-01 -1.20607293e+00 2.25102857e-01
2.41999298e-01 1.25099853e-01 -7.98828661e-01 3.36523890e-01
1.06814645e-01 7.46928081e-02 -1.06375444e+00 8.46313179e-01
8.53130162e-01 -1.30972683e+00 8.59791756e-01 -5.86798310e-01
3.47858548e-01 4.81640324e-02 9.09578875e-02 -1.09395313e+00
-4.43350136e-01 -2.92537630e-01 2.61997562e-02 1.01810098e+00
4.44541603e-01 -8.63227308e-01 1.45763233e-01 6.12734199e-01
4.96243536e-02 -1.13462996e+00 -8.20311427e-01 -4.74870294e-01
-5.77749729e-01 -1.28744677e-01 4.71834034e-01 7.95203507e-01
3.00234884e-01 2.22755820e-01 -2.56409705e-01 7.81428367e-02
4.40785795e-01 3.14393848e-01 2.15252802e-01 -1.53574610e+00
-5.24595156e-02 -5.74008450e-02 -3.90019178e-01 1.83158994e-01
-6.65493548e-01 -5.86391449e-01 -2.05386296e-01 -1.64817822e+00
2.94944227e-01 -5.06786883e-01 -9.66269910e-01 3.46352160e-01
-4.57954258e-01 -1.60017788e-01 -4.80958931e-02 2.71721840e-01
-1.81808501e-01 3.83981168e-02 5.28693497e-01 4.63084459e-01
-5.58986664e-01 5.32476366e-01 -1.82381913e-01 8.25495362e-01
1.08819449e+00 -7.32517719e-01 -3.68468374e-01 3.66863042e-01
1.84722945e-01 6.97322190e-01 2.98118830e-01 -1.17866683e+00
4.36944775e-02 -2.46158674e-01 5.77144444e-01 -7.98531532e-01
2.47162431e-01 -1.03336835e+00 6.23320282e-01 1.18328118e+00
-2.59704053e-01 2.83123612e-01 2.09454969e-01 3.43708843e-01
-2.91811287e-01 -3.22317213e-01 7.33526349e-01 -1.45649806e-01
-4.41794217e-01 -1.74839824e-01 -6.95594490e-01 -3.33101243e-01
1.66417480e+00 -2.29317725e-01 2.56886352e-02 -1.04858711e-01
-9.91804600e-01 3.52086723e-02 -2.08580121e-01 6.45992532e-02
7.27399707e-01 -7.06328392e-01 -7.24100173e-01 5.80668375e-02
2.57199824e-01 -5.23645759e-01 6.04496717e-01 1.38245404e+00
-6.39625788e-01 5.09859324e-01 -2.60101795e-01 -4.41307127e-01
-1.52007258e+00 5.97837090e-01 1.92759499e-01 -7.97604322e-01
-7.20050216e-01 5.88535011e-01 8.69295001e-02 3.64568055e-01
2.60322243e-01 -2.04069307e-03 -6.54894710e-01 4.13330078e-01
6.64661288e-01 8.40031922e-01 3.12649846e-01 -3.60047698e-01
-6.36330128e-01 2.51885772e-01 4.17792611e-02 2.67721623e-01
1.20620918e+00 4.23035584e-02 -1.60809487e-01 4.83784825e-01
7.35628724e-01 -3.15155715e-01 -7.56973445e-01 2.01420486e-01
1.96111709e-01 -1.08956195e-01 5.71950302e-02 -1.13673103e+00
-4.28699046e-01 7.67213762e-01 8.60002935e-01 4.66718882e-01
1.52605486e+00 -4.43494469e-01 3.98464650e-01 2.78898418e-01
2.94511586e-01 -8.74837875e-01 -3.47622275e-01 3.22266435e-03
7.51754761e-01 -1.07082367e+00 9.11384970e-02 -2.15698749e-01
-8.56768966e-01 1.09862781e+00 2.29448408e-01 9.61835608e-02
8.65983009e-01 9.58781019e-02 6.84109330e-02 -1.55981973e-01
-1.12318766e+00 3.48819233e-02 8.45138729e-03 2.66064227e-01
3.95823449e-01 3.00074875e-01 -9.30216670e-01 6.43946946e-01
1.81228295e-01 3.87370855e-01 1.62034959e-01 8.27910304e-01
-5.10739684e-01 -6.15993321e-01 -3.31472278e-01 8.90337229e-01
-8.40055227e-01 -3.05363387e-02 -1.25426814e-01 7.82513201e-01
2.74863034e-01 1.10228407e+00 -8.58056098e-02 -5.28352559e-01
3.41056198e-01 4.49050963e-01 1.39686104e-03 -2.53158271e-01
-9.29982185e-01 -1.49221942e-01 2.81968743e-01 -4.03773993e-01
-3.37884486e-01 -5.91554999e-01 -1.18110001e+00 9.17996541e-02
-3.69310558e-01 6.00634992e-01 6.61832690e-01 8.68968129e-01
2.74309546e-01 8.13984990e-01 8.75709414e-01 -1.90281644e-01
-3.98476064e-01 -1.04201210e+00 -4.52933550e-01 1.18611060e-01
2.04708174e-01 -6.76013172e-01 -5.03812194e-01 -5.97489104e-02] | [8.446102142333984, 4.866647243499756] |
c999edf9-a3e3-4592-95e8-76f6b55bd90d | deep-clustering-for-unsupervised-learning-of | 1807.05520 | null | http://arxiv.org/abs/1807.05520v2 | http://arxiv.org/pdf/1807.05520v2.pdf | Deep Clustering for Unsupervised Learning of Visual Features | Clustering is a class of unsupervised learning methods that has been
extensively applied and studied in computer vision. Little work has been done
to adapt it to the end-to-end training of visual features on large scale
datasets. In this work, we present DeepCluster, a clustering method that
jointly learns the parameters of a neural network and the cluster assignments
of the resulting features. DeepCluster iteratively groups the features with a
standard clustering algorithm, k-means, and uses the subsequent assignments as
supervision to update the weights of the network. We apply DeepCluster to the
unsupervised training of convolutional neural networks on large datasets like
ImageNet and YFCC100M. The resulting model outperforms the current state of the
art by a significant margin on all the standard benchmarks. | ['Matthijs Douze', 'Armand Joulin', 'Piotr Bojanowski', 'Mathilde Caron'] | 2018-07-15 | deep-clustering-for-unsupervised-learning-of-1 | http://openaccess.thecvf.com/content_ECCV_2018/html/Mathilde_Caron_Deep_Clustering_for_ECCV_2018_paper.html | http://openaccess.thecvf.com/content_ECCV_2018/papers/Mathilde_Caron_Deep_Clustering_for_ECCV_2018_paper.pdf | eccv-2018-9 | ['unsupervised-semantic-segmentation'] | ['computer-vision'] | [-2.43007958e-01 -8.17628577e-02 -7.84204081e-02 -8.41149390e-01
-1.41341090e-01 -4.59052473e-01 8.38189840e-01 6.88027143e-02
-8.39900613e-01 -7.20743313e-02 1.27860293e-01 -2.39543654e-02
-8.48079175e-02 -2.72085339e-01 -5.39445758e-01 -8.68859112e-01
-2.42247909e-01 7.33496785e-01 3.99308801e-01 3.38765621e-01
2.98069477e-01 3.38287503e-01 -1.59583366e+00 6.12143755e-01
4.50334162e-01 9.71045494e-01 2.83429533e-01 6.31701767e-01
-9.59815755e-02 8.54911745e-01 -3.99461925e-01 6.71217665e-02
2.15083644e-01 -2.82985598e-01 -8.50890279e-01 4.48972732e-01
4.31290686e-01 9.78504196e-02 -2.70615935e-01 8.71046245e-01
2.15923011e-01 5.52948475e-01 9.55617726e-01 -1.26853943e+00
-3.84486675e-01 8.14272285e-01 -6.78718030e-01 2.07813635e-01
-5.27220190e-01 -9.03481096e-02 7.56055295e-01 -9.25693512e-01
6.03201687e-01 1.22478473e+00 6.52961016e-01 3.89385909e-01
-1.31473827e+00 -5.32496989e-01 1.41349778e-01 4.25162911e-01
-1.56635678e+00 -4.00660604e-01 6.67252898e-01 -8.21506202e-01
1.19843805e+00 -3.11267763e-01 4.09876078e-01 6.48722470e-01
-2.86108017e-01 7.00807095e-01 7.47498035e-01 -6.00579381e-01
4.82024640e-01 1.28492400e-01 3.07143182e-01 7.18045294e-01
-1.86003625e-01 4.21052948e-02 -5.79972416e-02 7.67997578e-02
4.02837753e-01 3.31005782e-01 1.91350013e-01 -8.01163793e-01
-1.05347681e+00 1.11119497e+00 8.61997485e-01 3.75516295e-01
-2.54672676e-01 5.51021218e-01 3.89748126e-01 2.05208227e-01
3.27551961e-01 3.86814713e-01 -6.28951550e-01 2.86662936e-01
-1.02237439e+00 -1.33302569e-01 5.90360940e-01 7.74738550e-01
1.08541954e+00 -1.15103438e-01 -7.37702753e-03 1.06808591e+00
6.95163310e-01 -4.87245470e-02 7.60481656e-01 -1.31698525e+00
3.30905803e-02 8.54787052e-01 -3.43777299e-01 -8.84281814e-01
-6.77508712e-01 -3.13272983e-01 -9.55845594e-01 5.30582428e-01
2.33170018e-01 -2.85236150e-01 -1.24527574e+00 1.38857484e+00
2.34915376e-01 3.56533289e-01 -8.84355232e-02 8.27289701e-01
7.95334756e-01 5.41308582e-01 5.27382568e-02 2.21914992e-01
7.14030087e-01 -1.31671524e+00 -2.38167271e-01 -3.32238615e-01
5.70351839e-01 -8.14676642e-01 6.43728673e-01 3.90937448e-01
-6.41536176e-01 -7.42905378e-01 -9.31655884e-01 2.25457296e-01
-5.95300436e-01 3.86887968e-01 7.83877730e-01 4.32303727e-01
-1.35138869e+00 7.42802799e-01 -1.21464729e+00 -6.25513256e-01
8.27341676e-01 7.26840496e-01 -3.59699517e-01 -1.15545057e-01
-4.65770453e-01 4.67255563e-01 8.40258956e-01 -2.43237205e-02
-9.34775829e-01 -5.94725728e-01 -6.10878468e-01 1.27547681e-01
6.48208633e-02 -2.50211209e-01 1.08365893e+00 -1.26991582e+00
-1.49525821e+00 1.01767182e+00 2.18326494e-01 -6.70826614e-01
1.08141527e-01 -1.32656798e-01 2.49620061e-02 2.80763477e-01
-1.59036353e-01 1.45113218e+00 8.53570819e-01 -1.34534276e+00
-7.66197622e-01 -3.12335879e-01 -5.93109131e-01 1.76087394e-02
-6.10999525e-01 1.79484054e-01 -9.98397648e-01 -4.70361382e-01
3.83579247e-02 -1.14121044e+00 -7.00850666e-01 -2.16356486e-01
-6.08682573e-01 -5.89471102e-01 8.77874792e-01 1.09180450e-01
8.14905226e-01 -2.47260284e+00 2.78397679e-01 6.13750339e-01
3.53319347e-01 2.92503864e-01 -1.04683891e-01 1.54778913e-01
-2.05315977e-01 -1.00098006e-01 -3.16559821e-01 -6.25376701e-01
6.73545972e-02 2.41172001e-01 1.54580981e-01 6.88796818e-01
1.78884506e-01 7.90426075e-01 -6.81673408e-01 -5.00616968e-01
4.74223197e-01 4.48806047e-01 -7.76544154e-01 2.60211170e-01
-1.13919578e-01 2.15271279e-01 -2.60984898e-02 1.15046889e-01
3.57904494e-01 -5.88627756e-01 3.33457738e-01 3.92814465e-02
-6.19039834e-02 -2.87226051e-01 -1.02994704e+00 1.77235138e+00
1.22125857e-01 9.62167323e-01 -1.79142326e-01 -1.39896905e+00
7.51672566e-01 -3.91758606e-02 6.11739814e-01 -1.76634535e-01
2.95044482e-01 -2.29286194e-01 2.55234510e-01 -4.03292626e-01
1.00632146e-01 2.36064434e-01 2.37587355e-02 4.43040907e-01
6.11787796e-01 3.00182432e-01 2.99057662e-01 3.96881014e-01
1.02647483e+00 -2.14705408e-01 -5.41248843e-02 -4.94700193e-01
2.33459368e-01 1.59492016e-01 1.64053097e-01 7.78531492e-01
-2.93369982e-02 8.07120025e-01 2.43108839e-01 -5.09410858e-01
-9.68316674e-01 -8.45690787e-01 -3.57614597e-03 1.50822294e+00
-3.41111809e-01 -4.88941312e-01 -9.99062717e-01 -7.15052426e-01
1.02270469e-01 1.92717895e-01 -9.03701901e-01 -2.38978758e-01
-2.34188408e-01 -7.65332043e-01 4.49331313e-01 7.29311347e-01
2.73430914e-01 -1.37915337e+00 -4.99287337e-01 1.26791537e-01
2.64391124e-01 -1.03640044e+00 -2.09479243e-01 8.21015358e-01
-8.64139557e-01 -1.27799511e+00 -2.50454038e-01 -1.32462323e+00
1.15924752e+00 2.03964472e-01 1.10675943e+00 4.64719146e-01
-6.13049746e-01 4.06483322e-01 -3.63812149e-01 -3.55110526e-01
-9.64288265e-02 3.57907116e-01 -2.80474797e-02 2.71917850e-01
7.12317884e-01 -5.40679097e-01 -7.07234681e-01 3.19959223e-01
-9.96237636e-01 -2.28446990e-01 5.03672063e-01 5.02604067e-01
5.79959393e-01 1.55050248e-01 3.33399236e-01 -1.21257579e+00
2.66339660e-01 -4.83093262e-01 -5.02228618e-01 -1.80520222e-01
-6.38275564e-01 2.67578214e-02 8.19367468e-01 -4.17976946e-01
-5.79220474e-01 7.67395437e-01 1.08010896e-01 -9.14436996e-01
-6.12093389e-01 5.00616729e-01 7.79751539e-02 7.77167920e-03
8.32897484e-01 -1.18382685e-01 -4.81025539e-02 -6.03968024e-01
8.04058552e-01 6.37843549e-01 8.80954385e-01 -3.04044068e-01
7.44434714e-01 6.94861233e-01 -3.39170247e-01 -6.63220167e-01
-7.49408662e-01 -1.12145567e+00 -1.18923807e+00 -1.40673384e-01
1.07028842e+00 -8.64607275e-01 -4.71723080e-01 5.64021766e-01
-7.18796611e-01 -9.91565645e-01 -1.27251297e-01 5.33512712e-01
-5.79563141e-01 2.15533584e-01 -5.12796283e-01 -1.18575446e-01
-1.11394793e-01 -1.04157853e+00 6.55701339e-01 1.69600412e-01
-1.78229555e-01 -1.10467935e+00 1.50886193e-01 -5.33442609e-02
2.72471189e-01 1.75648108e-01 8.49539340e-01 -8.31398666e-01
-2.21420035e-01 -1.33959591e-01 -1.97304741e-01 5.67224085e-01
9.78370830e-02 3.88736695e-01 -1.03750038e+00 -4.72096920e-01
-4.39093530e-01 -6.23710632e-01 1.59868956e+00 6.60536349e-01
1.56084418e+00 -3.42752435e-03 -5.99047422e-01 9.52712715e-01
1.53181088e+00 7.11721927e-02 6.31971359e-01 4.52711105e-01
9.44774210e-01 6.32991135e-01 1.46289304e-01 3.69126767e-01
3.11343461e-01 3.70988548e-01 5.14467537e-01 -3.11550975e-01
-5.26077002e-02 2.06948593e-01 1.59929376e-02 7.41077602e-01
9.07846242e-02 1.82432458e-01 -1.11508584e+00 8.21516752e-01
-2.16954184e+00 -8.52161765e-01 -9.88049656e-02 1.79606938e+00
7.29584575e-01 9.71631035e-02 1.18540592e-01 2.79136878e-02
8.27981532e-01 -9.79647636e-02 -5.07889450e-01 -1.21538050e-01
2.23335698e-01 1.94999978e-01 3.86020899e-01 1.82753474e-01
-1.67940176e+00 1.31871498e+00 6.66105223e+00 5.44851422e-01
-1.01801229e+00 -6.90212324e-02 8.23686123e-01 -8.60793665e-02
4.41553950e-01 -1.67642944e-02 -5.90844870e-01 3.40960771e-01
9.07777309e-01 4.98266965e-01 4.75855947e-01 1.10065794e+00
4.01977077e-02 2.07171291e-02 -1.29997027e+00 8.60109031e-01
1.71176389e-01 -1.36288452e+00 -1.59182876e-01 -2.78781354e-02
1.22979248e+00 7.79499710e-01 1.66915193e-01 3.83655101e-01
9.83334482e-01 -1.13983643e+00 3.61503661e-01 3.55993539e-01
4.33302909e-01 -1.05632401e+00 6.46941960e-01 2.30349511e-01
-8.58056664e-01 -2.43392691e-01 -8.46343160e-01 1.04423642e-01
-4.80843008e-01 5.16426325e-01 -9.53661084e-01 -2.45100945e-01
1.17981219e+00 1.03748536e+00 -1.00662136e+00 1.37180471e+00
-1.07171521e-01 9.49647427e-01 -4.52799201e-01 2.89978445e-01
8.35504115e-01 -1.57424048e-01 -2.53308713e-01 1.60958767e+00
-1.35931402e-01 -3.40097696e-01 5.96635401e-01 5.13336837e-01
-3.61412346e-01 -1.04934331e-02 -3.29219073e-01 1.27626136e-01
3.99525583e-01 1.70940924e+00 -1.38140988e+00 -5.16575813e-01
-3.38700503e-01 8.08546782e-01 8.68717611e-01 3.48685205e-01
-3.90178114e-01 -3.77550423e-01 4.57189888e-01 -1.75280973e-01
1.06328022e+00 -1.30436599e-01 -1.45397365e-01 -6.31314695e-01
-4.56692010e-01 -5.75493991e-01 5.90635300e-01 -5.20180583e-01
-1.43793833e+00 5.54141939e-01 -4.98371124e-02 -1.11946070e+00
-3.51282597e-01 -7.25311160e-01 -9.99769986e-01 2.14298069e-01
-1.43711782e+00 -7.64989972e-01 -4.28474844e-01 9.15578961e-01
4.51421469e-01 -5.76843083e-01 7.60189295e-01 1.17736742e-01
-6.80982649e-01 4.40592796e-01 7.90022492e-01 7.25600362e-01
9.72603917e-01 -1.62118101e+00 4.22175735e-01 5.81102073e-01
4.43674862e-01 5.98029315e-01 3.64268363e-01 -2.15786532e-01
-9.23167825e-01 -1.56469429e+00 4.75909501e-01 -3.50488573e-01
6.81215286e-01 -4.75408286e-01 -8.41677904e-01 6.08313084e-01
6.46633625e-01 5.01155019e-01 9.58546400e-01 9.67738628e-02
-5.61210096e-01 -1.56722397e-01 -7.87058771e-01 1.33428380e-01
6.14652812e-01 -1.35427311e-01 -4.00182545e-01 4.15868104e-01
4.87531900e-01 -1.80478301e-02 -8.87092233e-01 1.79841893e-03
2.90211082e-01 -9.51118112e-01 9.45584357e-01 -7.71999836e-01
4.29320872e-01 -4.66794461e-01 -1.14847369e-01 -1.72799361e+00
-7.33121634e-01 -3.72663587e-01 2.34988499e-02 1.15892625e+00
4.92474735e-01 -6.62118420e-02 9.68981028e-01 1.89036116e-01
-3.44961047e-01 -5.72831869e-01 -5.04741192e-01 -3.58492732e-01
2.89475739e-01 -3.24865222e-01 2.15968862e-01 1.22046173e+00
-8.69684219e-02 3.32340002e-01 6.75036162e-02 4.38231155e-02
7.75919139e-01 4.90889186e-03 8.17186773e-01 -1.61275065e+00
-2.20821306e-01 -6.88958943e-01 -6.27222538e-01 -5.46078324e-01
3.95682275e-01 -1.20466924e+00 3.90750557e-01 -1.53232229e+00
3.42707336e-01 -3.61992240e-01 -5.45885801e-01 7.40890443e-01
-1.18424952e-01 5.17410994e-01 2.61287063e-01 4.52156872e-01
-1.31123722e+00 1.86168969e-01 4.97064948e-01 -2.91786611e-01
-3.13337922e-01 -1.34592369e-01 -6.67380452e-01 8.90818894e-01
1.04380834e+00 -8.05137634e-01 -4.62192446e-01 -5.24607778e-01
-1.17976539e-01 -8.14482749e-01 1.89290479e-01 -1.09270990e+00
7.28744805e-01 1.23950846e-01 9.39562261e-01 -6.07448161e-01
-2.15279892e-01 -1.02639401e+00 -2.20575094e-01 1.57910094e-01
-7.16583073e-01 7.49714486e-03 2.57490613e-02 6.45065427e-01
-2.88119525e-01 -2.28507668e-01 1.03389108e+00 -2.69948542e-01
-9.54521537e-01 3.49532008e-01 -7.32028663e-01 9.14295390e-02
9.68573987e-01 -6.38803840e-02 -1.60662338e-01 -7.87943527e-02
-8.74314487e-01 4.89079148e-01 4.66351539e-01 5.54110169e-01
5.92551053e-01 -1.42544031e+00 -6.42834723e-01 1.48891658e-01
1.95697159e-01 3.21702003e-01 -3.12507711e-02 7.54192114e-01
-4.18134421e-01 1.50367305e-01 -2.13837028e-01 -1.04440975e+00
-1.20872915e+00 7.64878452e-01 3.05344880e-01 -7.54665658e-02
-6.07785881e-01 7.75333822e-01 2.02150092e-01 -5.84344804e-01
7.93324828e-01 -1.63575456e-01 -6.63370907e-01 4.68096770e-02
5.67566812e-01 2.32248455e-02 1.18675664e-01 -6.26286983e-01
-3.25142145e-01 4.69148457e-01 -4.20443356e-01 2.28851438e-01
1.76126337e+00 -1.29544465e-02 -2.89441913e-01 4.28560585e-01
1.64828658e+00 -7.34779894e-01 -1.54629672e+00 -3.96229029e-01
4.70673531e-01 2.95594577e-02 2.06925228e-01 -4.86565888e-01
-1.61071217e+00 8.65624964e-01 8.34211349e-01 1.25087621e-02
8.56986344e-01 2.86500931e-01 1.20155148e-01 1.01578355e+00
-1.20131992e-01 -1.46819806e+00 4.29993898e-01 6.32771969e-01
3.18853110e-01 -1.36585891e+00 -2.95547713e-02 4.89974730e-02
-7.74279475e-01 1.03028560e+00 7.09466100e-01 -7.81966746e-01
1.06315458e+00 3.80191922e-01 4.25678879e-01 -4.49894845e-01
-8.07550848e-01 -4.36675280e-01 3.90374422e-01 8.34139705e-01
3.67986917e-01 -8.01439360e-02 3.96149546e-01 3.40776384e-01
-5.31528816e-02 -2.34239787e-01 2.23097235e-01 7.79294431e-01
-7.18366504e-01 -7.99603879e-01 -3.37072253e-01 5.98589063e-01
-3.89970899e-01 -8.54370669e-02 -7.26500690e-01 6.31709635e-01
2.48214513e-01 9.54925537e-01 5.54207325e-01 -4.59904969e-01
3.16132233e-02 1.27390474e-01 2.04310462e-01 -8.11981380e-01
-8.07676077e-01 3.60170066e-01 -4.41516459e-01 -4.82472599e-01
-9.03549790e-01 -5.87217033e-01 -1.70759976e+00 1.31151319e-01
-1.49277017e-01 1.88757405e-01 6.27145231e-01 9.69374299e-01
3.61288071e-01 4.94881928e-01 8.44106615e-01 -1.20172381e+00
-9.34106410e-02 -1.15461433e+00 -4.18599427e-01 7.77013123e-01
1.98809877e-02 -6.37984693e-01 -3.55000824e-01 3.85042995e-01] | [9.20806884765625, 3.06827974319458] |
9747c8cc-3cc3-4888-a38c-16dc166c6a55 | spatial-temporal-prompt-learning-for | 2305.14244 | null | https://arxiv.org/abs/2305.14244v1 | https://arxiv.org/pdf/2305.14244v1.pdf | Spatial-temporal Prompt Learning for Federated Weather Forecasting | Federated weather forecasting is a promising collaborative learning framework for analyzing meteorological data across participants from different countries and regions, thus embodying a global-scale real-time weather data predictive analytics platform to tackle climate change. This paper is to model the meteorological data in a federated setting where many distributed low-resourced sensors are deployed in different locations. Specifically, we model the spatial-temporal weather data into a federated prompt learning framework that leverages lightweight prompts to share meaningful representation and structural knowledge among participants. Prompts-based communication allows the server to establish the structural topology relationships among participants and further explore the complex spatial-temporal correlations without transmitting private data while mitigating communication overhead. Moreover, in addition to a globally shared large model at the server, our proposed method enables each participant to acquire a personalized model that is highly customized to tackle climate changes in a specific geographic area. We have demonstrated the effectiveness of our method on classical weather forecasting tasks by utilizing three spatial-temporal multivariate time-series weather data. | ['Jing Jiang', 'Tianyi Zhou', 'Tao Shen', 'Guodong Long', 'Shengchao Chen'] | 2023-05-23 | null | null | null | null | ['weather-forecasting'] | ['miscellaneous'] | [-6.31753862e-01 -3.17920834e-01 1.09003283e-01 -5.97207248e-01
-4.06823725e-01 -7.52469540e-01 6.14431322e-01 5.22423565e-01
8.97570280e-04 7.25123823e-01 5.47561109e-01 -4.89657283e-01
-4.70099032e-01 -1.22919583e+00 -4.54518497e-01 -8.87784243e-01
-9.50749099e-01 2.70365834e-01 6.66640997e-02 -2.91011520e-02
-1.89867690e-01 4.66605365e-01 -1.57069993e+00 9.72383320e-02
1.07644081e+00 1.04818606e+00 3.37363482e-01 9.85079288e-01
-2.99990416e-01 8.37971985e-01 -6.20348930e-01 1.37098432e-01
1.95167422e-01 4.75145966e-01 -5.25623262e-01 -5.76548696e-01
3.49375933e-01 -3.60048026e-01 9.39640999e-02 3.05837691e-01
6.88209355e-01 4.62274998e-01 5.07402383e-02 -1.40217948e+00
-2.26403594e-01 7.29937136e-01 -2.73834437e-01 4.15459573e-01
2.94945508e-01 2.22065020e-02 7.52093077e-01 -3.75913233e-01
2.00804636e-01 8.13099682e-01 8.93922627e-01 -4.68386747e-02
-7.75237262e-01 -8.37383211e-01 8.87885451e-01 2.72401661e-01
-1.03560293e+00 -2.32456505e-01 6.74721837e-01 -3.92421961e-01
7.82591045e-01 8.76310825e-01 6.21155620e-01 6.91668630e-01
3.60892832e-01 1.61542058e-01 1.04379094e+00 1.55290619e-01
5.19472659e-01 -2.33841930e-02 4.99731563e-02 2.20980957e-01
8.31417553e-03 1.93654671e-01 -1.01360452e+00 -6.10668123e-01
1.86229572e-01 6.23907268e-01 -2.13334620e-01 2.39659905e-01
-1.28509915e+00 3.50128949e-01 6.09504163e-01 1.14616469e-01
-9.54135418e-01 1.35530084e-01 2.63942569e-01 1.62893653e-01
1.16816926e+00 1.26653602e-02 -1.03933406e+00 2.69746613e-02
-1.03599799e+00 3.09065759e-01 1.10229015e+00 5.27199149e-01
9.51867819e-01 6.40728921e-02 -8.59143287e-02 6.41226396e-02
1.92784548e-01 1.05135548e+00 -4.46367636e-03 -8.56083870e-01
5.39213121e-01 6.01166308e-01 4.93872285e-01 -1.44400859e+00
-9.59355533e-01 -4.60913867e-01 -1.39241421e+00 -1.55385852e-01
4.87793535e-02 -1.04637229e+00 -4.52884510e-02 1.67001641e+00
1.24985838e+00 1.05244935e+00 -8.97747502e-02 1.07655799e+00
6.47697449e-01 1.05730319e+00 1.92065120e-01 -2.75287718e-01
1.10403037e+00 -7.89931774e-01 -5.51377773e-01 4.86914903e-01
6.14539206e-01 -5.37691176e-01 6.57193363e-01 2.28371114e-01
-4.98112708e-01 -4.15442079e-01 -3.21553499e-01 4.61030990e-01
-7.31907070e-01 -1.49161205e-01 1.01095593e+00 4.71587986e-01
-1.26280785e+00 2.40844592e-01 -8.23065579e-01 -3.15436482e-01
-1.22995205e-01 8.81482214e-02 -3.03125620e-01 1.31379426e-01
-1.24816704e+00 3.84507388e-01 -5.35928598e-03 1.84182450e-01
-7.25291848e-01 -1.69000494e+00 -3.39867175e-01 2.59277761e-01
5.01661329e-03 -9.49905872e-01 9.77458417e-01 -3.00913692e-01
-1.28023541e+00 -1.59441486e-01 -2.11345717e-01 -2.37054154e-01
7.71914870e-02 -4.34908494e-02 -9.71751511e-01 -1.13739237e-01
-1.82140842e-01 -2.75371432e-01 4.89586562e-01 -9.83679056e-01
-1.05035770e+00 -5.95813513e-01 -5.92770241e-02 1.00563414e-01
-8.68510365e-01 7.05927610e-02 3.26416701e-01 -7.40657568e-01
-3.72609109e-01 -4.07995760e-01 -7.20959842e-01 -2.47514650e-01
1.92619972e-02 -3.62758219e-01 1.12067688e+00 -8.02553952e-01
1.38209879e+00 -1.68106115e+00 -3.63961309e-01 6.23168468e-01
5.22585869e-01 1.02450959e-01 -2.60555238e-01 1.00191605e+00
3.11196804e-01 5.65483421e-02 2.57352710e-01 -4.57377046e-01
-3.27798203e-02 2.88560599e-01 -1.01944518e+00 3.98182303e-01
-4.79200095e-01 5.77397883e-01 -8.39742482e-01 -2.23336175e-01
1.52482569e-01 5.13313115e-01 -2.82924652e-01 7.43804336e-01
-4.72472370e-01 9.70335126e-01 -8.79651964e-01 4.64077145e-01
1.09062421e+00 -3.76131088e-01 4.01479721e-01 9.32506323e-02
-5.58767736e-01 1.53400362e-01 -1.41720247e+00 1.50258768e+00
-8.58387768e-01 3.28281969e-02 7.28561580e-01 -8.40437591e-01
1.10811424e+00 6.47367597e-01 9.31519210e-01 -6.93492651e-01
-4.95854765e-01 -1.72120467e-01 -9.33573544e-01 -5.89246333e-01
7.05716312e-01 5.36593437e-01 -1.90526098e-01 1.19881392e+00
-4.87072706e-01 4.49543446e-01 -4.68547851e-01 2.77522475e-01
1.08190203e+00 -1.35636613e-01 -1.36674596e-02 -3.68642360e-01
3.72797132e-01 3.73893827e-02 7.07660377e-01 7.39377499e-01
6.72075083e-04 -2.57824421e-01 -2.91572154e-01 -1.45307505e+00
-5.55419981e-01 -8.22356522e-01 3.07112962e-01 1.70800030e+00
-9.13643278e-03 -7.18675673e-01 -2.84420177e-02 -3.26894045e-01
2.74892300e-01 3.69166315e-01 -4.26020116e-01 6.81157112e-01
-4.53707784e-01 -7.67953634e-01 3.99226576e-01 1.77275285e-01
3.96382362e-01 -4.94487911e-01 -8.01058531e-01 5.29472888e-01
-3.84188890e-01 -7.48394549e-01 -3.50483626e-01 -1.09230079e-01
-7.69673645e-01 -1.10535836e+00 -2.40040496e-01 -1.67607948e-01
3.85253906e-01 7.39150524e-01 1.09850681e+00 -9.87023860e-02
-1.15048215e-01 9.11791205e-01 -2.88435519e-01 -4.97744828e-01
2.16499895e-01 1.97901934e-01 2.22858489e-01 3.42386395e-01
-4.48773131e-02 -1.18320811e+00 -1.01229787e+00 2.38063395e-01
-8.48919988e-01 5.00574447e-02 -2.59285092e-01 8.11737105e-02
4.02808756e-01 1.90579206e-01 7.68982232e-01 -6.54114962e-01
5.48526525e-01 -1.32358086e+00 -8.04694831e-01 5.60643673e-01
-6.69846356e-01 -4.31300223e-01 1.11625302e+00 -1.53422847e-01
-1.34234834e+00 1.62163571e-01 5.34485936e-01 -1.27166109e-02
-4.50940698e-01 9.65397120e-01 3.50076944e-01 2.33820882e-02
5.37981391e-01 2.58936971e-01 -2.74040788e-01 -7.92103171e-01
6.86664224e-01 7.04120994e-01 6.46291673e-01 -8.12115252e-01
1.00053608e+00 6.06408834e-01 -5.67662492e-02 -3.84405106e-01
-5.27287602e-01 -6.48918748e-01 -2.11049840e-01 -4.20765519e-01
4.32339281e-01 -1.45783138e+00 -1.29886675e+00 4.84571576e-01
-1.14182925e+00 -4.88050044e-01 -4.13202308e-02 5.81266880e-01
1.55947089e-01 5.93838394e-02 -2.69097805e-01 -1.06528413e+00
-9.34709728e-01 -1.04814343e-01 7.49050617e-01 3.80786866e-01
6.18144423e-02 -1.56672788e+00 7.64903665e-01 9.23027396e-02
1.29598129e+00 6.65552676e-01 4.81040359e-01 -6.38017714e-01
-7.36812532e-01 7.89401308e-02 -7.81217664e-02 -5.66465199e-01
4.44677532e-01 -1.58012304e-02 -1.11849809e+00 -5.00630021e-01
-3.54228705e-01 -5.20900972e-02 3.92311335e-01 2.60652043e-02
1.34920132e+00 -1.04472280e+00 -4.42740172e-01 7.43708193e-01
1.11125791e+00 -5.29919803e-01 -2.96450704e-01 -8.41045752e-02
6.98521018e-01 8.93392026e-01 3.01513672e-01 1.20037186e+00
1.55627561e+00 4.22900110e-01 6.92435920e-01 -2.96269536e-01
4.12217796e-01 -1.74223170e-01 2.08533868e-01 9.59670961e-01
-1.37405604e-01 -2.84732878e-01 -1.03859162e+00 7.25966811e-01
-2.02923846e+00 -1.03469324e+00 -2.62978226e-01 2.12398577e+00
7.10629642e-01 -1.03100431e+00 1.53244764e-01 -3.71507019e-01
3.29623878e-01 7.50378728e-01 -5.73926866e-01 -2.41328120e-01
-1.86025098e-01 8.77574533e-02 5.61054707e-01 2.70943910e-01
-1.07183468e+00 4.67095613e-01 5.95078850e+00 2.79840887e-01
-1.49565601e+00 2.77546316e-01 4.10603553e-01 -2.63868004e-01
-6.36720836e-01 -4.38489914e-02 -5.27224898e-01 5.65732837e-01
1.58708143e+00 -4.44327831e-01 6.25784039e-01 7.16325045e-01
7.19833910e-01 9.36237574e-02 -4.92509842e-01 4.95355129e-01
-6.53312027e-01 -1.92826843e+00 -3.59259367e-01 3.94992344e-02
1.03975868e+00 5.75188577e-01 -1.62736297e-01 -1.99838445e-01
1.01441717e+00 -8.11709762e-01 9.59519669e-02 1.13558710e+00
4.85317260e-01 -5.14894068e-01 2.04662368e-01 7.58586228e-01
-1.86869001e+00 -3.44365925e-01 -1.66722298e-01 -3.86667818e-01
8.64130557e-02 1.18377507e+00 -6.31546617e-01 1.13138008e+00
1.21660137e+00 8.38274777e-01 -3.28130156e-01 9.99437988e-01
3.44908535e-01 1.05561650e+00 -8.08388710e-01 2.33420163e-01
-6.58149570e-02 -1.53651923e-01 4.49324280e-01 9.73642707e-01
6.51298046e-01 3.51333916e-01 7.24782228e-01 3.06255072e-01
1.10972613e-01 2.25176349e-01 -5.29350281e-01 5.47366619e-01
9.78492379e-01 1.69186854e+00 8.18180591e-02 -2.57672101e-01
-3.18218976e-01 2.94406712e-01 3.62277895e-01 5.47595024e-01
-4.08969760e-01 -3.07523552e-02 1.17194963e+00 -1.55454516e-01
2.98411041e-01 -4.54278857e-01 -1.91865966e-01 -1.18107426e+00
1.41283646e-02 -5.53323328e-01 7.75827587e-01 -5.81277013e-01
-1.64388967e+00 5.41980982e-01 -2.70761371e-01 -1.03075373e+00
-1.86054379e-01 2.35934034e-01 -1.49445176e+00 1.21477401e+00
-2.06034899e+00 -1.63778508e+00 -6.65652871e-01 1.36738825e+00
-2.66489506e-01 -1.07511103e-01 1.38860166e+00 2.06910416e-01
-5.92452824e-01 1.55776709e-01 4.87139672e-01 -3.82372975e-01
7.89397657e-01 -1.06499672e+00 4.79628146e-01 9.50915337e-01
-9.14550126e-02 5.65373600e-01 4.00164813e-01 -5.56296051e-01
-1.68723726e+00 -1.75923395e+00 1.38506663e+00 -1.57908022e-01
6.65032268e-01 -3.02145481e-01 -9.63294744e-01 3.82419586e-01
3.28718752e-01 5.10369956e-01 1.04968154e+00 4.23745424e-01
-6.27276003e-01 -9.56253231e-01 -1.10325563e+00 2.33984739e-02
6.47167683e-01 -3.96866739e-01 1.57082275e-01 7.16342092e-01
9.78919208e-01 -1.70665368e-01 -1.25163519e+00 4.96339053e-02
1.94216728e-01 -7.25308299e-01 5.32467127e-01 -6.86909020e-01
-3.07601660e-01 -5.94942868e-01 -1.96514815e-01 -1.69220543e+00
-4.29371357e-01 -1.16270232e+00 -3.39837700e-01 1.40210533e+00
1.13199055e-01 -1.18176496e+00 5.86020589e-01 1.00370693e+00
1.47239372e-01 -1.78473026e-01 -1.19832516e+00 -3.04067045e-01
-1.52439624e-01 -5.75164616e-01 1.65533578e+00 1.21011734e+00
1.21906035e-01 -3.48481864e-01 -7.12085664e-01 9.37816560e-01
5.93102098e-01 9.41252589e-01 1.08195889e+00 -1.56348002e+00
-1.54054195e-01 2.82383084e-01 2.72660822e-01 -7.61314869e-01
1.14141352e-01 -5.71210504e-01 -4.67363477e-01 -1.23639417e+00
-4.15623337e-01 -1.17612994e+00 -5.50142348e-01 8.55415881e-01
-1.03107862e-01 -2.49611393e-01 9.61923599e-02 4.57658201e-01
-7.26513684e-01 5.06727397e-01 7.15133488e-01 5.26545336e-03
-4.08276618e-01 3.35357517e-01 -5.06489217e-01 1.25710607e-01
9.81981993e-01 -3.52552533e-01 -4.45093334e-01 -7.73857176e-01
6.23760283e-01 4.06168967e-01 5.89770734e-01 -8.86267483e-01
9.91815209e-01 -7.92685866e-01 2.31719688e-01 -7.37449169e-01
-1.89571813e-01 -1.23677278e+00 7.93397546e-01 1.22056894e-01
-1.52572349e-01 4.13060904e-01 4.81352396e-02 8.44204187e-01
-1.24969343e-02 1.10882926e+00 -2.31043771e-01 2.45540187e-01
-6.35505080e-01 8.62841129e-01 -2.83917606e-01 -3.29473615e-01
1.14159989e+00 4.54319447e-01 -7.96635211e-01 -5.33578634e-01
-4.89246130e-01 1.05927563e+00 8.51460323e-02 2.83644497e-01
2.21611619e-01 -1.03911066e+00 -9.19403553e-01 2.46913418e-01
-2.76169111e-03 1.06630698e-01 9.91530120e-01 7.94885278e-01
-2.88030622e-03 3.28648895e-01 3.33269313e-02 -3.32316905e-01
-1.03310764e+00 3.76158834e-01 2.71546721e-01 -4.06167954e-01
-6.79050744e-01 4.43650723e-01 -4.89504457e-01 -1.12162673e+00
3.36723745e-01 -4.01339293e-01 -1.25841275e-01 3.10949773e-01
8.72348011e-01 6.96255386e-01 1.27049550e-01 -1.64052293e-01
-4.74901319e-01 1.28645152e-01 6.82879269e-01 2.80335814e-01
1.96057880e+00 -6.91980600e-01 -4.31675524e-01 2.44730860e-01
7.35624433e-01 1.70405537e-01 -1.41699517e+00 -8.20792496e-01
-2.03024581e-01 -4.11939949e-01 3.10964763e-01 -1.05318415e+00
-1.55457532e+00 3.31062704e-01 4.54164386e-01 4.36125338e-01
1.35028172e+00 -3.44879836e-01 1.15346301e+00 3.62416595e-01
4.65946347e-01 -8.23562264e-01 -6.33752108e-01 3.12136263e-01
5.22815287e-01 -1.00503361e+00 -7.10671619e-02 1.01572819e-01
-5.04168808e-01 1.13186467e+00 3.09326440e-01 1.40537977e-01
1.29723942e+00 3.89735401e-01 3.71996194e-01 -1.53002545e-01
-1.48287535e+00 1.14383690e-01 7.48463348e-02 9.10874546e-01
9.33878124e-02 7.59697258e-01 2.82514989e-01 8.11843872e-01
-2.50991493e-01 9.77470949e-02 2.77942061e-01 6.30604982e-01
-2.53622323e-01 -1.04663336e+00 -5.78088284e-01 2.78405935e-01
-1.29354954e-01 5.76157756e-02 2.20239520e-01 -6.92657232e-02
1.22733772e-01 1.37083375e+00 3.36159825e-01 -5.26545644e-01
-3.04991305e-02 -8.80527869e-02 -6.54565036e-01 -7.49466717e-02
-1.23462200e+00 -3.83120358e-01 -2.68202215e-01 -9.16726589e-01
-6.84082806e-01 -4.04244691e-01 -9.28530693e-01 -1.07948351e+00
3.99289221e-01 5.11773229e-01 1.02469206e+00 6.60978973e-01
1.34217715e+00 1.69309571e-01 1.62924099e+00 -8.65715921e-01
-3.74620527e-01 -6.50483370e-01 -6.13762438e-01 -1.03603244e-01
7.71693230e-01 -3.96533161e-02 -1.95699170e-01 -1.23607606e-01] | [6.652695655822754, 2.7524795532226562] |
b51db5a6-3ab2-4ef1-be4c-49f784acc444 | lstc-boosting-atomic-action-detection-with | 2110.09819 | null | https://arxiv.org/abs/2110.09819v1 | https://arxiv.org/pdf/2110.09819v1.pdf | LSTC: Boosting Atomic Action Detection with Long-Short-Term Context | In this paper, we place the atomic action detection problem into a Long-Short Term Context (LSTC) to analyze how the temporal reliance among video signals affect the action detection results. To do this, we decompose the action recognition pipeline into short-term and long-term reliance, in terms of the hypothesis that the two kinds of context are conditionally independent given the objective action instance. Within our design, a local aggregation branch is utilized to gather dense and informative short-term cues, while a high order long-term inference branch is designed to reason the objective action class from high-order interaction between actor and other person or person pairs. Both branches independently predict the context-specific actions and the results are merged in the end. We demonstrate that both temporal grains are beneficial to atomic action recognition. On the mainstream benchmarks of atomic action detection, our design can bring significant performance gain from the existing state-of-the-art pipeline. The code of this project can be found at [this url](https://github.com/TencentYoutuResearch/ActionDetection-LSTC) | ['Feiyue Huang', 'Jilin Li', 'Chengjie Wang', 'Weiyao Lin', 'Yabiao Wang', 'Jian Li', 'Boshen Zhang', 'Yuxi Li'] | 2021-10-19 | null | null | null | null | ['atomic-action-recognition'] | ['computer-vision'] | [ 4.32054490e-01 4.02202196e-02 -3.45761329e-01 -5.61378360e-01
-8.29715908e-01 -3.16089928e-01 8.37653399e-01 -8.63487273e-02
-2.70091355e-01 4.91955370e-01 6.55264914e-01 2.16247812e-02
6.37046024e-02 -4.56468821e-01 -6.60590649e-01 -5.58008790e-01
-2.17704743e-01 1.53835669e-01 6.23167217e-01 1.36626035e-01
1.03403427e-01 2.64101058e-01 -1.63060534e+00 9.38309133e-01
2.29865566e-01 9.23501730e-01 -1.38344288e-01 9.46825981e-01
2.45766059e-01 1.21111596e+00 -3.45883965e-01 -1.81888476e-01
3.17758799e-01 -7.00846732e-01 -8.39748442e-01 2.00271651e-01
5.70900083e-01 -6.22265220e-01 -4.21062380e-01 7.50308812e-01
4.26063240e-01 9.27106813e-02 3.38327765e-01 -1.28896487e+00
-2.77333595e-02 5.18495858e-01 -6.56341016e-01 4.89194989e-01
6.40178502e-01 7.06859767e-01 1.24079227e+00 -8.40446293e-01
5.91562212e-01 1.33257866e+00 4.65080678e-01 4.91159499e-01
-1.10810781e+00 -6.04714930e-01 7.45905101e-01 2.44594738e-01
-9.91102517e-01 -7.91685462e-01 5.39084256e-01 -6.60297215e-01
9.68208253e-01 1.91626921e-01 7.42946148e-01 1.45376897e+00
2.61728968e-02 1.35813880e+00 7.84225166e-01 -1.75071344e-01
1.94234699e-01 -3.87416035e-01 4.83353764e-01 6.76082850e-01
1.70619972e-02 2.21153155e-01 -7.11340249e-01 -9.88627970e-02
6.85934067e-01 1.21985964e-01 -6.17034212e-02 -6.71467558e-02
-1.28012788e+00 3.69116127e-01 1.44724935e-01 3.77284437e-01
-5.15781105e-01 5.22052646e-01 4.35313016e-01 -1.30217776e-01
2.93307841e-01 1.27955332e-01 -4.14649457e-01 -5.96155345e-01
-8.42041135e-01 2.28607476e-01 6.80824995e-01 5.02051413e-01
6.14866495e-01 -4.05180305e-01 -6.98431492e-01 4.70474035e-01
3.22525293e-01 1.50140509e-01 8.21318105e-02 -1.13159502e+00
3.95604372e-01 6.24276221e-01 1.27299011e-01 -7.60386229e-01
-2.61915296e-01 -2.36559212e-01 -1.88893691e-01 2.07934424e-01
6.45236552e-01 -1.51968613e-01 -7.17768192e-01 1.63544190e+00
5.87385595e-01 7.03672230e-01 -2.02902734e-01 9.30127561e-01
6.79899991e-01 4.16630030e-01 3.18736941e-01 -1.09489992e-01
1.46074486e+00 -1.10659850e+00 -6.51510060e-01 -5.37426889e-01
7.24860668e-01 -5.06960154e-01 8.87292862e-01 2.00613394e-01
-1.00890350e+00 -6.17466033e-01 -7.36029863e-01 -1.02991931e-01
-4.82971072e-02 3.05259377e-01 6.81747913e-01 1.30016282e-01
-7.16058910e-01 4.90285277e-01 -1.22390449e+00 -4.57716614e-01
5.98848045e-01 4.91673034e-03 -2.92299926e-01 -3.57546583e-02
-1.08224785e+00 5.42400181e-01 1.93045646e-01 1.17408954e-01
-1.05600262e+00 -4.13355112e-01 -6.92183077e-01 -1.17724076e-01
9.37199712e-01 -5.24266124e-01 1.39136779e+00 -1.23289371e+00
-1.19886923e+00 8.68207276e-01 -4.59376037e-01 -4.02614862e-01
6.61009550e-01 -5.18044889e-01 -2.16855347e-01 3.85329545e-01
3.10713112e-01 5.92242599e-01 8.10236216e-01 -8.71921062e-01
-1.10245585e+00 -4.53427225e-01 3.43483418e-01 2.29709849e-01
2.95603126e-02 3.04448754e-01 -9.40318704e-01 -6.94223583e-01
-1.87300399e-01 -9.65068936e-01 -4.16630693e-02 7.23330751e-02
-3.05303454e-01 -4.30928618e-01 6.49974644e-01 -6.02518797e-01
1.29870713e+00 -2.34359932e+00 9.55604166e-02 -1.19605005e-01
2.25978829e-02 9.42536443e-02 -1.27942428e-01 3.26175451e-01
-2.37668410e-01 -8.53096619e-02 -9.38919112e-02 -4.88674879e-01
-4.20320630e-02 1.30861521e-01 -2.49300167e-01 5.00793159e-01
2.73105741e-01 1.02468860e+00 -1.12977922e+00 -4.87656981e-01
1.80343509e-01 3.09795469e-01 -4.77631986e-01 2.56262153e-01
-3.86174440e-01 5.22327602e-01 -6.57156229e-01 8.49901319e-01
8.50226954e-02 -3.39875579e-01 2.34635279e-01 -1.60132766e-01
-6.67188540e-02 4.15897131e-01 -1.25032520e+00 1.63092458e+00
1.57709986e-01 6.02667868e-01 9.57511067e-02 -7.51944005e-01
1.81202173e-01 2.43911475e-01 8.06185722e-01 -4.89366591e-01
1.27407759e-02 -8.88670534e-02 -4.20375727e-02 -6.43963575e-01
2.15020299e-01 2.23703638e-01 1.47985397e-02 5.09219289e-01
-1.14422739e-01 6.05832338e-01 4.41913426e-01 3.59775901e-01
1.56053066e+00 6.39250219e-01 3.05534214e-01 1.83039472e-01
4.37326312e-01 -1.52308241e-01 9.00495946e-01 8.18266988e-01
-5.64013362e-01 4.94589061e-01 7.66482174e-01 -4.46843117e-01
-3.17660987e-01 -8.39609206e-01 3.26112390e-01 1.43418455e+00
3.65204513e-02 -6.83340967e-01 -5.98257542e-01 -9.62358773e-01
1.24812284e-02 5.09918928e-01 -7.87946522e-01 -3.68787535e-02
-7.14329958e-01 -3.53540063e-01 5.70059180e-01 8.18189204e-01
5.29119909e-01 -1.21247494e+00 -8.50380957e-01 4.20143008e-02
-3.48570287e-01 -1.20710206e+00 -5.65997183e-01 -1.07155249e-01
-6.50954783e-01 -1.25769818e+00 -4.07660842e-01 -1.30041033e-01
3.63167554e-01 1.95429683e-01 1.14177716e+00 1.92212984e-01
-3.36398244e-01 7.88067162e-01 -5.43841481e-01 -1.34290308e-01
-8.46620426e-02 -3.37356806e-01 -5.44382967e-02 4.95072156e-01
6.39860570e-01 -5.37391841e-01 -8.17019999e-01 4.34263825e-01
-6.98148727e-01 1.45878047e-01 6.67449296e-01 4.00602967e-01
5.55592000e-01 -1.46981835e-01 9.81213897e-02 -6.10915840e-01
6.90200329e-02 -5.66719353e-01 -3.83405715e-01 2.58039325e-01
-2.43262097e-01 -3.47052515e-02 2.35154890e-02 -4.14488167e-01
-1.03429055e+00 4.86178577e-01 1.74156744e-02 -4.52819228e-01
-5.10722280e-01 4.01322335e-01 -2.53106058e-01 5.97075343e-01
4.58701938e-01 2.24242657e-01 -2.14000404e-01 -4.89717633e-01
2.02060431e-01 3.40495765e-01 6.27346754e-01 -6.71910107e-01
3.05213153e-01 7.11141825e-01 -1.16366006e-01 -5.82331419e-01
-1.02158272e+00 -6.64231956e-01 -7.94342041e-01 -7.32448757e-01
1.33589470e+00 -9.81997728e-01 -5.78404963e-01 5.88570356e-01
-1.05646682e+00 -7.32042730e-01 -2.86307096e-01 5.28466702e-01
-5.36629438e-01 3.42915118e-01 -5.32203972e-01 -9.83008206e-01
1.84538677e-01 -1.16167593e+00 1.34771109e+00 1.43590644e-01
-4.07630384e-01 -6.45907283e-01 9.83284637e-02 5.27866185e-01
-2.05455914e-01 3.31996471e-01 2.42387682e-01 -6.44636571e-01
-6.93558931e-01 -2.44918585e-01 -2.42636837e-02 2.46558040e-01
1.09792046e-01 1.18534386e-01 -9.94028926e-01 -8.99430811e-02
-3.35685194e-01 -3.46582890e-01 1.11573982e+00 4.55674231e-01
8.56548548e-01 -1.16837412e-01 -5.34227669e-01 3.04039717e-01
8.92481685e-01 7.73490444e-02 7.21028030e-01 1.36683390e-01
7.78001368e-01 6.23231113e-01 9.14580107e-01 5.31468034e-01
2.78927833e-01 8.82875025e-01 3.56534153e-01 2.93229539e-02
-3.55539888e-01 -2.76957095e-01 9.04263854e-01 -7.59805441e-02
-2.17176661e-01 -1.15215942e-01 -9.04939413e-01 4.59185362e-01
-2.31680942e+00 -1.48907840e+00 -2.99107581e-01 2.25038147e+00
6.67450786e-01 3.15917671e-01 5.59036195e-01 -3.49268913e-02
5.97211182e-01 4.97970223e-01 -5.84238768e-01 9.28950757e-02
1.30693182e-01 -3.99115682e-01 1.46585956e-01 6.03130758e-01
-1.33712900e+00 8.21418047e-01 5.98346090e+00 6.77414834e-01
-8.74987185e-01 6.79422244e-02 6.04008615e-01 -5.96781373e-01
2.38576397e-01 2.42227048e-01 -1.00075305e+00 5.68813205e-01
7.96434641e-01 3.72002304e-01 6.38422966e-02 6.60909891e-01
5.66081703e-01 -5.24473846e-01 -1.60047984e+00 8.91921580e-01
1.13715157e-02 -1.01724005e+00 -2.39804029e-01 1.99007899e-01
3.17011535e-01 -3.30382399e-02 -1.78188547e-01 2.88432240e-01
3.77541661e-01 -6.35778427e-01 9.05728042e-01 8.03210318e-01
4.86462384e-01 -2.26966187e-01 2.12183282e-01 2.96633750e-01
-1.47723222e+00 -3.52055162e-01 3.72523069e-01 -3.12570900e-01
4.37629521e-01 6.23058736e-01 -4.04547691e-01 4.17133182e-01
7.52170384e-01 1.29069948e+00 -6.59361184e-01 7.63438225e-01
-5.53895593e-01 8.22509110e-01 -1.89759552e-01 3.58253002e-01
1.85940728e-01 -1.16722034e-02 9.34573829e-01 1.34816480e+00
-1.18043937e-01 5.65430582e-01 6.02878869e-01 5.33042192e-01
3.45194519e-01 -3.17434788e-01 -3.51761401e-01 -1.60883173e-01
1.16047740e-01 9.98275042e-01 -7.81100512e-01 -6.00031376e-01
-5.51077783e-01 1.09509218e+00 2.30038762e-01 3.84835094e-01
-1.11699057e+00 1.53316021e-01 8.26160848e-01 1.08334348e-01
5.14726281e-01 -1.21820599e-01 -1.43982187e-01 -1.33972359e+00
4.07976449e-01 -9.67990816e-01 8.84575486e-01 -7.60266840e-01
-1.04170918e+00 1.13198340e-01 6.79884851e-02 -1.27345097e+00
-2.62162805e-01 -4.84853655e-01 -6.51109517e-01 4.88333374e-01
-1.09535539e+00 -1.26495290e+00 -2.41618946e-01 6.41372323e-01
8.63235891e-01 1.12301275e-01 4.45588022e-01 2.68526524e-01
-9.79656160e-01 2.88412809e-01 -5.41988730e-01 4.36304778e-01
7.81118095e-01 -1.05594075e+00 1.45275518e-01 1.28845131e+00
2.11071357e-01 4.81144756e-01 6.95324242e-01 -9.41708684e-01
-1.23640168e+00 -1.01467443e+00 7.73107588e-01 -8.25678408e-01
6.59233928e-01 -4.04262841e-01 -6.51139498e-01 1.00868440e+00
-9.21087041e-02 2.60124296e-01 5.82744718e-01 3.06925237e-01
-4.83888566e-01 -2.14142520e-02 -6.07014596e-01 7.24498332e-01
1.34033680e+00 -5.26325583e-01 -6.28727317e-01 2.99266815e-01
4.34025049e-01 -1.65378734e-01 -6.15985990e-01 3.94793272e-01
9.99686360e-01 -1.26019776e+00 8.90857816e-01 -8.25912654e-01
6.87718749e-01 -4.57991600e-01 -1.54504418e-01 -5.70367455e-01
-4.81123596e-01 -5.30984759e-01 -5.70334256e-01 1.13151407e+00
3.15221190e-01 -4.15282398e-01 6.24734104e-01 8.52747798e-01
-5.51947542e-02 -8.62695634e-01 -7.38933444e-01 -7.62619317e-01
-4.70081240e-01 -6.68669045e-01 2.21397579e-01 6.79269195e-01
1.71174482e-01 4.57335413e-01 -5.05591929e-01 2.31954426e-01
3.74635547e-01 1.44337252e-01 9.86339509e-01 -8.92919898e-01
-8.42902124e-01 -5.75965703e-01 -3.92776489e-01 -1.32371330e+00
4.15482447e-02 -5.17838418e-01 1.76871061e-01 -1.60002983e+00
4.66323942e-01 -2.56513264e-02 -4.76674795e-01 9.87785637e-01
-4.84645605e-01 1.33989871e-01 2.74852693e-01 3.54090452e-01
-1.15227127e+00 2.88128406e-01 8.56639266e-01 -1.55322133e-02
-2.14030743e-01 8.12320039e-02 -5.19115269e-01 8.07041287e-01
4.69902396e-01 -5.33222854e-01 -3.77313942e-01 -3.74262422e-01
2.37691421e-02 1.16544269e-01 6.86242819e-01 -1.11162996e+00
2.24183708e-01 -3.78213257e-01 2.77758658e-01 -4.61769581e-01
5.72701633e-01 -5.94726503e-01 4.97630872e-02 4.90026236e-01
-6.02279842e-01 -2.91843265e-01 7.35323457e-03 8.23785186e-01
-1.21437714e-01 2.17006534e-01 5.71237624e-01 -2.31350616e-01
-8.67472768e-01 3.93736213e-01 -4.11415964e-01 -4.35261019e-02
1.15047789e+00 -3.31991017e-01 -2.42313713e-01 -5.26063323e-01
-9.54616606e-01 2.74669170e-01 3.32260370e-01 4.53528404e-01
3.98077041e-01 -1.22224712e+00 -7.93670118e-01 -1.47861853e-01
2.02910498e-01 -3.85441929e-01 2.49987066e-01 1.38658345e+00
1.63158283e-01 2.05484360e-01 5.89536577e-02 -6.54516160e-01
-1.55247521e+00 4.40600097e-01 3.92378420e-01 -3.73533040e-01
-7.01086402e-01 9.85562623e-01 4.31630433e-01 6.60101548e-02
4.89850760e-01 -2.80263305e-01 -1.23328030e-01 2.22813562e-01
8.39430273e-01 4.46259230e-01 -4.96118784e-01 -7.90503800e-01
-6.87923729e-01 3.05915415e-01 3.76654230e-02 -3.53762537e-01
1.21794903e+00 -4.55192886e-02 1.66450944e-02 6.21313870e-01
1.06510699e+00 2.64412211e-03 -1.76069379e+00 -1.48370087e-01
4.06613722e-02 -6.65878594e-01 -1.16046324e-01 -9.27172184e-01
-1.05290008e+00 4.89604533e-01 4.14104462e-01 3.91571037e-02
1.14256632e+00 2.91025698e-01 5.44576347e-01 1.13988318e-01
1.75312042e-01 -1.06028187e+00 2.20186234e-01 5.27895927e-01
9.23704684e-01 -1.30790341e+00 1.84832990e-01 -4.65955406e-01
-7.10209548e-01 8.27350914e-01 7.15014279e-01 -3.89019698e-02
4.64754701e-01 2.50274360e-01 -1.30387321e-01 -4.55833972e-01
-1.01394665e+00 -5.38360894e-01 2.93036819e-01 2.48988435e-01
4.48807448e-01 -8.07341859e-02 -2.14330316e-01 5.82863748e-01
5.66054285e-01 2.25037098e-01 1.92467913e-01 1.18026638e+00
-1.49380952e-01 -1.05237556e+00 -2.74178743e-01 3.71102244e-01
-5.03420353e-01 1.35860801e-01 -6.45276606e-01 5.37163615e-01
1.90502211e-01 9.42963362e-01 7.41418153e-02 -4.07095790e-01
3.04714739e-01 1.53998435e-01 2.69353807e-01 -6.59160972e-01
-4.61908787e-01 2.14680791e-01 5.62370420e-01 -1.38933015e+00
-6.86074257e-01 -1.16788661e+00 -1.32065403e+00 1.73169211e-01
5.68564199e-02 -2.74274886e-01 1.38632566e-01 1.18581557e+00
6.06346846e-01 4.54635590e-01 2.80717760e-01 -9.65440392e-01
-1.86646193e-01 -8.41068089e-01 -2.89788395e-01 6.10100806e-01
3.75539213e-01 -7.27155983e-01 -4.16668534e-01 4.45709497e-01] | [8.339824676513672, 0.5771890878677368] |
93be1b6f-b419-49c5-9455-b8f6a00ec4cb | towards-autoformalization-of-mathematics-and | 2301.02195 | null | https://arxiv.org/abs/2301.02195v1 | https://arxiv.org/pdf/2301.02195v1.pdf | Towards Autoformalization of Mathematics and Code Correctness: Experiments with Elementary Proofs | The ever-growing complexity of mathematical proofs makes their manual verification by mathematicians very cognitively demanding. Autoformalization seeks to address this by translating proofs written in natural language into a formal representation that is computer-verifiable via interactive theorem provers. In this paper, we introduce a semantic parsing approach, based on the Universal Transformer architecture, that translates elementary mathematical proofs into an equivalent formalization in the language of the Coq interactive theorem prover. The same architecture is also trained to translate simple imperative code decorated with Hoare triples into formally verifiable proofs of correctness in Coq. Experiments on a limited domain of artificial and human-written proofs show that the models generalize well to intermediate lengths not seen during training and variations in natural language. | ['David Juedes', 'Razvan C. Bunescu', 'Garett Cunningham'] | 2023-01-05 | null | null | null | null | ['mathematical-proofs', 'semantic-parsing'] | ['miscellaneous', 'natural-language-processing'] | [ 2.00668737e-01 7.03641593e-01 1.73948094e-01 -3.13895524e-01
-7.62896895e-01 -1.37813890e+00 6.62270606e-01 1.12278536e-01
2.28329629e-01 8.86716127e-01 -3.81494880e-01 -1.59943044e+00
-2.02828526e-01 -1.12648976e+00 -1.19919670e+00 1.54155686e-01
-3.39900017e-01 4.66314614e-01 3.74296635e-01 -1.70281827e-01
2.96873987e-01 1.10257916e-01 -1.34141219e+00 4.85633552e-01
1.07856596e+00 3.63251537e-01 -5.27251773e-02 1.34071231e+00
-3.24631125e-01 1.25696945e+00 -3.41514677e-01 -9.06665802e-01
5.02835624e-02 -4.81756210e-01 -1.35182226e+00 -5.13936937e-01
6.14110172e-01 -5.32608449e-01 -4.10297848e-02 1.16949272e+00
-5.01872897e-01 -7.81658530e-01 4.31463689e-01 -1.90115917e+00
-8.17440450e-01 1.17236888e+00 3.03461373e-01 -2.39919811e-01
6.90574944e-01 1.31261125e-01 1.32013702e+00 -3.22665781e-01
6.43572092e-01 1.47115052e+00 9.04692709e-01 7.75377691e-01
-1.58383846e+00 -5.15844226e-01 -3.68740320e-01 3.62502813e-01
-1.34846640e+00 -1.25527501e-01 4.46321785e-01 -4.95452672e-01
1.38519764e+00 3.87829274e-01 2.19450906e-01 5.66635311e-01
6.25078022e-01 2.86380678e-01 1.15541756e+00 -7.03388631e-01
-3.88903730e-02 7.03917384e-01 4.29472864e-01 1.25987589e+00
6.21864915e-01 2.42686823e-01 2.00495254e-02 -4.60837126e-01
6.26339138e-01 -5.59816480e-01 -7.15760048e-04 -6.29072070e-01
-1.40066350e+00 1.09748220e+00 1.51463868e-02 3.21329325e-01
2.57430792e-01 7.38181233e-01 8.91610205e-01 8.42578411e-01
-5.01990497e-01 7.18090236e-01 -8.37902546e-01 -4.52795438e-02
-6.50044858e-01 6.04329944e-01 1.44079125e+00 1.15745223e+00
4.43593889e-01 -1.96920022e-01 2.47216269e-01 -1.41739771e-01
4.39143896e-01 8.42468739e-01 -7.83230439e-02 -1.40139997e+00
3.61366183e-01 5.14652073e-01 2.20935673e-01 -9.51097012e-01
-4.11802046e-02 -6.39861450e-02 -3.07215333e-01 4.06974941e-01
5.63442647e-01 6.28846064e-02 -1.64909393e-01 1.73465347e+00
7.73628280e-02 -3.84193391e-01 6.39007747e-01 5.42184711e-01
5.89277089e-01 7.72558093e-01 1.63427624e-03 -1.17622212e-01
1.37388623e+00 -6.98198378e-01 -4.51133341e-01 1.99485660e-01
7.97350168e-01 -1.71515822e-01 1.01662862e+00 6.17655277e-01
-1.37611926e+00 -1.90457240e-01 -1.24025953e+00 -4.10091192e-01
-4.84975874e-01 -3.72000456e-01 1.22981846e+00 5.13221145e-01
-1.24610806e+00 4.92883831e-01 -5.15795231e-01 1.46093801e-01
1.37879089e-01 3.10820550e-01 -3.85327011e-01 1.19581960e-01
-1.45853150e+00 9.14681077e-01 7.92332828e-01 -1.37565017e-01
-1.19392169e+00 -6.63425148e-01 -1.25841486e+00 2.14164481e-01
3.65950406e-01 -9.53728199e-01 2.01022744e+00 -1.09443402e+00
-1.37177420e+00 1.02712917e+00 1.15168123e-02 -8.97988081e-01
6.46050215e-01 3.73450547e-01 -2.09281340e-01 2.27738783e-01
2.65124202e-01 5.14209449e-01 5.71750820e-01 -1.16498148e+00
-6.30293667e-01 3.50255403e-03 1.08932829e+00 -6.50535762e-01
3.90699625e-01 3.16733986e-01 3.77185434e-01 2.03583296e-02
-2.03066126e-01 -6.82030618e-01 1.64190948e-01 8.63707289e-02
-1.62702456e-01 -5.39742708e-01 5.16692340e-01 -6.59974873e-01
8.29312801e-01 -1.74136353e+00 1.44149855e-01 2.99535066e-01
4.79795903e-01 5.23859002e-02 4.50045168e-02 5.54685771e-01
-4.20822203e-02 4.72443819e-01 -1.89663723e-01 3.30999494e-01
9.56403613e-01 4.23298180e-01 -6.72979951e-01 1.38672248e-01
3.86524677e-01 1.29466856e+00 -1.16280520e+00 -8.49236310e-01
-7.87989274e-02 -8.71727765e-02 -8.83316219e-01 2.28400782e-01
-9.63202119e-01 -2.48124078e-01 -3.71591806e-01 3.62615198e-01
7.04606235e-01 -4.68464136e-01 6.03443563e-01 5.10223866e-01
-2.25197986e-01 8.50962460e-01 -9.09888208e-01 1.77216482e+00
-6.26400888e-01 7.05873013e-01 2.90450573e-01 -9.16910291e-01
3.86336923e-01 5.79567134e-01 -5.26045740e-01 -5.31957030e-01
1.71508521e-01 3.99002194e-01 7.75347129e-02 -3.58117491e-01
8.83355811e-02 -5.56508899e-01 -2.74694532e-01 8.50485981e-01
-3.44407074e-02 -6.52772665e-01 5.16998887e-01 5.86520255e-01
1.07165444e+00 4.19741541e-01 2.04564154e-01 -2.12380007e-01
9.48138714e-01 5.20439386e-01 2.35355198e-01 1.02737808e+00
1.11698955e-01 -2.10907266e-01 9.38640296e-01 -6.88136578e-01
-1.34062982e+00 -1.34107792e+00 4.64420393e-02 7.50707805e-01
-2.54893750e-01 -9.08460498e-01 -1.05149424e+00 -9.26934898e-01
9.47875604e-02 1.03973675e+00 -1.70936391e-01 -1.94868043e-01
-2.83012182e-01 5.12320995e-01 1.03665900e+00 3.75718981e-01
4.64599013e-01 -1.13722730e+00 -8.56390476e-01 1.51777789e-01
-2.82943636e-01 -1.27097642e+00 2.62493163e-01 1.84238870e-02
-6.85033798e-01 -1.51348758e+00 2.20435739e-01 -7.62229621e-01
6.55938387e-01 -4.57760021e-02 1.26607835e+00 7.66157806e-01
-1.49430484e-01 3.96237850e-01 -2.28455096e-01 -5.78224897e-01
-1.48199701e+00 -2.63104856e-01 -1.93883374e-01 -8.84134471e-01
1.55765831e-01 -5.03533244e-01 2.45275244e-01 -1.65468842e-01
-9.01202679e-01 5.48647940e-01 2.26523444e-01 6.56968594e-01
-1.37468278e-01 4.61318105e-01 6.02011383e-02 -8.61250341e-01
5.17158687e-01 1.10766724e-01 -1.36688030e+00 4.51549143e-01
-5.48433006e-01 5.44018090e-01 1.39504921e+00 -1.44825757e-01
-6.12462699e-01 -4.52090651e-01 2.28008166e-01 -1.54215559e-01
-3.27845216e-02 6.06510818e-01 7.94646740e-02 -1.15312122e-01
6.46913290e-01 3.48937929e-01 3.24543603e-02 3.79760504e-01
4.18792546e-01 3.34921658e-01 8.27036202e-01 -1.12440038e+00
1.67753696e+00 -1.31737858e-01 5.23334682e-01 1.67061817e-02
-5.43000758e-01 4.92165715e-01 -3.26191157e-01 3.22699100e-01
3.62096041e-01 -4.61179733e-01 -1.32501960e+00 7.15222582e-02
-1.67175448e+00 -7.23789513e-01 -5.86898625e-02 1.17953077e-01
-1.09100521e+00 6.68643177e-01 -5.19160092e-01 -8.63170207e-01
-5.33628762e-01 -1.02912748e+00 8.59291255e-01 -2.14774534e-01
-5.36661863e-01 -9.53766763e-01 1.05923630e-01 1.99266851e-01
1.19744554e-01 8.05812702e-03 1.53710938e+00 -3.89609426e-01
-8.87744725e-01 -2.00180888e-01 -7.22550452e-01 5.88303745e-01
-2.07509741e-01 2.25344330e-01 -6.52960002e-01 5.03236018e-02
-5.82988620e-01 -4.61100638e-01 -6.11162893e-02 -2.28867576e-01
7.26626158e-01 -7.58492112e-01 -3.98923531e-02 2.91665047e-01
1.36728716e+00 -1.65526986e-01 6.48948491e-01 4.53202844e-01
-7.88211599e-02 9.93594378e-02 3.38698626e-01 -2.18933836e-01
6.40600622e-01 5.03530130e-02 2.46181935e-01 2.92120278e-01
1.56120613e-01 -6.58907473e-01 3.88363719e-01 1.19726233e-01
9.57494974e-02 4.33752209e-01 -1.14287603e+00 3.92896563e-01
-1.64497888e+00 -1.39843774e+00 -1.86347961e-01 1.69223964e+00
1.32593358e+00 3.84110212e-01 -1.49678707e-01 5.13428211e-01
1.43540338e-01 -6.90460980e-01 1.94986194e-01 -1.35670316e+00
2.16712594e-01 4.48793501e-01 1.69287905e-01 1.04755557e+00
-7.78629661e-01 8.78707886e-01 6.83384323e+00 1.42668948e-01
-7.60640144e-01 8.53116736e-02 -2.27070913e-01 8.00127625e-01
-6.47210896e-01 6.08873248e-01 -9.40791592e-02 8.98410454e-02
1.30637038e+00 -5.84764361e-01 9.24303770e-01 8.39807808e-01
1.72867682e-02 -5.23121580e-02 -1.71365070e+00 4.94320631e-01
-3.41710478e-01 -1.72400463e+00 1.76455319e-01 -4.03373539e-01
1.47607744e-01 -4.30854499e-01 -2.88711935e-01 7.43490696e-01
6.63484693e-01 -1.07585049e+00 1.26517773e+00 2.65463322e-01
8.72082651e-01 -6.48024321e-01 7.60206521e-01 5.28389156e-01
-8.14785480e-01 -3.78468454e-01 -1.81297436e-01 -6.06855631e-01
-3.59118670e-01 -1.88836902e-01 -1.29430890e+00 6.64677560e-01
4.30148780e-01 3.48150462e-01 -4.52211171e-01 5.00699222e-01
-7.79657662e-01 5.14126003e-01 -1.56731412e-01 -3.48083586e-01
3.60936522e-01 -1.44448087e-01 3.57937425e-01 1.50415707e+00
-6.88818917e-02 3.55719507e-01 -2.45063081e-01 1.67237878e+00
2.74502993e-01 -4.33039010e-01 -8.68542314e-01 -4.50517952e-01
-7.34500140e-02 1.01895869e+00 -4.36097890e-01 -7.15112984e-01
-4.70656693e-01 7.43335962e-01 3.26782048e-01 1.99702308e-01
-1.20336425e+00 -5.98198056e-01 1.93745807e-01 5.93298487e-02
2.51111418e-01 -4.49321657e-01 -9.04527605e-02 -1.19484735e+00
1.27102092e-01 -1.21892583e+00 1.51976392e-01 -1.37472689e+00
-8.26334953e-01 1.86979979e-01 3.31582725e-01 -6.03581607e-01
-5.84889174e-01 -9.26132381e-01 -7.00038791e-01 8.93324494e-01
-1.73597777e+00 -1.37386668e+00 2.33148724e-01 3.38133365e-01
-3.79723459e-02 1.90966308e-01 1.29975545e+00 6.75428882e-02
7.29418695e-02 6.51768208e-01 -5.06113648e-01 3.58263791e-01
-6.12278543e-02 -1.49088871e+00 4.47476447e-01 9.90032375e-01
-1.24478668e-01 1.15201950e+00 1.17739296e+00 -9.93146375e-02
-2.18195629e+00 -9.12182629e-01 1.53563464e+00 -6.12132609e-01
1.31621265e+00 -4.51279938e-01 -7.03518987e-01 1.09044182e+00
2.17502236e-01 -2.39828229e-01 1.47584036e-01 5.43896444e-02
-1.21421885e+00 2.53359526e-02 -1.07561374e+00 5.40519774e-01
9.40177321e-01 -1.26234937e+00 -1.25402713e+00 4.21957076e-01
8.52303028e-01 -2.50402689e-01 -5.00206292e-01 1.25274152e-01
8.47484767e-01 -6.70968950e-01 7.22618580e-01 -9.67265368e-01
7.38851547e-01 -7.12614775e-01 -1.68320775e-01 -6.56550467e-01
-1.35489151e-01 -9.16178763e-01 -1.87136054e-01 9.75633800e-01
5.94445765e-01 -8.82832348e-01 2.46982753e-01 6.47219718e-01
1.11495771e-01 -8.26237425e-02 -7.76492596e-01 -7.64315367e-01
4.82636839e-01 -6.99693382e-01 4.98383760e-01 8.81000698e-01
9.16665256e-01 6.27982736e-01 4.81135756e-01 3.98472548e-01
6.40294015e-01 7.04613328e-01 1.02219546e+00 -1.19167292e+00
-4.47795570e-01 -5.47990739e-01 -4.48977768e-01 -5.74370742e-01
8.50872457e-01 -1.55213344e+00 1.42289013e-01 -1.58332086e+00
4.96634424e-01 -3.64916712e-01 3.43022108e-01 8.00970376e-01
4.22630370e-01 -1.03673458e-01 -2.69379001e-03 -5.15955091e-01
-7.59281635e-01 -8.92830566e-02 9.60858405e-01 -5.97187042e-01
4.78705168e-01 -3.02123487e-01 -5.86266577e-01 7.35347152e-01
5.00717282e-01 -3.63548726e-01 -5.76122582e-01 -4.46196109e-01
1.09089887e+00 4.13787395e-01 1.06855190e+00 -8.39497566e-01
7.51205757e-02 -4.05029655e-01 -4.78695422e-01 -2.23388523e-01
-4.96160597e-01 -9.89314735e-01 3.03211838e-01 7.08774209e-01
-3.32815975e-01 3.13303530e-01 5.52233219e-01 -8.99631977e-02
6.39102682e-02 -4.78521347e-01 3.17149639e-01 -2.05986232e-01
-4.38677490e-01 -2.11500451e-01 -5.98465860e-01 9.98567324e-03
8.77674818e-01 -9.54713970e-02 -1.51923984e-01 -2.29093373e-01
-4.76675600e-01 2.94861943e-01 6.94431424e-01 -9.29305446e-04
5.94886243e-01 -6.71569645e-01 -5.83624303e-01 1.30298957e-01
-2.82760127e-03 -2.90264517e-01 -3.14117998e-01 7.98107505e-01
-1.00216782e+00 9.03778076e-01 -1.08983390e-01 -2.57390648e-01
-1.49088407e+00 9.40262437e-01 6.58496737e-01 -3.13816160e-01
-5.72465003e-01 3.61567199e-01 1.15326129e-01 -9.14988279e-01
7.80886933e-02 -8.74176145e-01 5.58497787e-01 -9.82144177e-01
7.59618461e-01 -1.73557356e-01 -2.38372147e-01 -5.93700074e-03
-5.39589286e-01 3.61403465e-01 3.23564738e-01 -1.66112736e-01
1.09703279e+00 2.44415879e-01 -8.96827996e-01 1.19304933e-01
8.22862446e-01 2.09129117e-02 -3.41784835e-01 1.27694998e-02
9.02470946e-02 1.03020296e-01 -4.19687867e-01 -9.89731312e-01
-9.94087160e-02 8.42346013e-01 -2.48664498e-01 6.18780851e-01
5.93201399e-01 1.76860362e-01 5.05980670e-01 1.16932857e+00
7.19009757e-01 -4.85230803e-01 -5.76543987e-01 7.72109866e-01
7.95754135e-01 -9.34574306e-01 2.10625548e-02 -3.85603338e-01
-4.33294885e-02 1.40347230e+00 1.27156436e-01 -3.34154107e-02
2.48757135e-02 6.84020817e-01 -1.08707905e-01 -2.06129149e-01
-1.02899075e+00 4.48015809e-01 -3.73075992e-01 6.28820300e-01
7.00147152e-02 1.72059774e-01 -2.26612091e-01 6.76622152e-01
-5.01959145e-01 5.97902000e-01 7.66247571e-01 1.18285024e+00
-1.33860737e-01 -1.24296916e+00 -5.71862102e-01 -3.63013834e-01
-3.85427028e-01 -2.82727152e-01 -5.72015941e-01 1.20253801e+00
7.47649893e-02 9.32851315e-01 -2.69116163e-01 9.35027823e-02
-1.44118164e-02 3.19105059e-01 1.34587717e+00 -5.55463612e-01
-4.41342592e-01 -8.38121116e-01 5.02745867e-01 -4.28323656e-01
-2.19228894e-01 -4.58453387e-01 -2.01711416e+00 -1.00989258e+00
-1.61539674e-01 8.73622119e-01 4.69303936e-01 9.47248161e-01
-4.60988581e-02 3.34538490e-01 3.53788227e-01 -1.97234616e-01
-8.40921342e-01 -3.89870197e-01 -2.25275695e-01 1.52771652e-01
7.26168334e-01 -1.18002251e-01 -3.97027731e-01 4.92827922e-01] | [8.940459251403809, 7.046481609344482] |
ba1dd0f2-11ce-4847-9948-d7cd6a12b1cb | spotlights-probing-shapes-from-spherical | 2205.12564 | null | https://arxiv.org/abs/2205.12564v3 | https://arxiv.org/pdf/2205.12564v3.pdf | Spotlights: Probing Shapes from Spherical Viewpoints | Recent years have witnessed the surge of learned representations that directly build upon point clouds. Though becoming increasingly expressive, most existing representations still struggle to generate ordered point sets. Inspired by spherical multi-view scanners, we propose a novel sampling model called Spotlights to represent a 3D shape as a compact 1D array of depth values. It simulates the configuration of cameras evenly distributed on a sphere, where each virtual camera casts light rays from its principal point through sample points on a small concentric spherical cap to probe for the possible intersections with the object surrounded by the sphere. The structured point cloud is hence given implicitly as a function of depths. We provide a detailed geometric analysis of this new sampling scheme and prove its effectiveness in the context of the point cloud completion task. Experimental results on both synthetic and real data demonstrate that our method achieves competitive accuracy and consistency while having a significantly reduced computational cost. Furthermore, we show superior performance on the downstream point cloud registration task over state-of-the-art completion methods. | ['Laurent Kneip', 'Soren Schwertfeger', 'Tao Sun', 'Xinyu Jiang', 'Minghao Xu', 'Wenqing Jiang', 'Ran Cheng', 'Lige Liu', 'Jiaxin Wei'] | 2022-05-25 | null | null | null | null | ['point-cloud-completion', 'point-cloud-registration'] | ['computer-vision', 'computer-vision'] | [ 6.60244226e-02 2.11859688e-01 2.55273879e-01 -1.74603745e-01
-8.46403062e-01 -7.86294520e-01 9.38679993e-01 -9.18787941e-02
-5.56641594e-02 1.11474909e-01 7.34664723e-02 -1.76294714e-01
1.26877785e-01 -8.00004721e-01 -1.10486257e+00 -4.40895557e-01
1.27340958e-01 1.11132348e+00 1.18224330e-01 2.33212635e-02
3.42784137e-01 9.36698556e-01 -1.52997422e+00 -3.10749728e-02
5.35414398e-01 8.07446301e-01 3.88244092e-01 4.60150033e-01
-2.14330330e-02 3.75625379e-02 -2.80112684e-01 -3.41638446e-01
5.70384324e-01 3.46294492e-01 -4.61219251e-01 4.20347333e-01
4.93394583e-01 -4.13469940e-01 -3.54685277e-01 7.63509095e-01
1.32384494e-01 -6.00452423e-02 7.34575152e-01 -1.18074608e+00
-6.30059779e-01 -2.79356778e-01 -9.37107146e-01 -4.19809222e-01
7.59156168e-01 -1.27123550e-01 9.14377987e-01 -1.50377607e+00
8.03869486e-01 1.07052457e+00 6.99886084e-01 3.54972810e-01
-1.29623652e+00 -4.28382903e-01 1.06121838e-01 -3.60160053e-01
-1.41286039e+00 -2.53919393e-01 9.66228247e-01 -4.11874503e-01
7.95030951e-01 2.86002219e-01 7.29048729e-01 8.18895519e-01
-5.50775044e-02 4.51768458e-01 6.93045199e-01 -3.30065787e-01
4.75968450e-01 -2.02206105e-01 -1.94820672e-01 5.79921186e-01
4.31592464e-01 -1.11720525e-01 -5.06813407e-01 -6.46952331e-01
1.36991978e+00 4.74324763e-01 -3.17095876e-01 -1.32629037e+00
-1.43568134e+00 8.04234087e-01 6.22533619e-01 -1.68585986e-01
-3.61807436e-01 2.82334000e-01 -2.03978479e-01 -1.23588540e-01
6.15657449e-01 2.23933816e-01 -2.07737416e-01 1.99562714e-01
-7.51728058e-01 4.46533293e-01 6.84377789e-01 1.37004554e+00
9.40828562e-01 -2.02323705e-01 5.16153693e-01 5.81047118e-01
4.99899685e-01 7.68156171e-01 -2.27807105e-01 -1.22095621e+00
6.77372515e-01 8.35449278e-01 3.81320983e-01 -9.14556623e-01
-2.09124267e-01 -3.46259475e-01 -8.21139097e-01 4.85690981e-01
1.51393503e-01 3.38293493e-01 -7.05360115e-01 1.15202725e+00
5.92452228e-01 5.05393326e-01 -1.61872491e-01 1.07062626e+00
6.09790623e-01 5.11718154e-01 -6.70967460e-01 1.48099959e-01
1.13140237e+00 -5.57755947e-01 8.95235315e-02 -9.31246728e-02
2.64424354e-01 -6.65670216e-01 7.46896505e-01 4.88623381e-01
-1.30395448e+00 -2.73560673e-01 -9.24009800e-01 -2.56350905e-01
1.58376426e-01 -6.17665909e-02 6.72222912e-01 2.69152939e-01
-1.21891403e+00 4.55273062e-01 -1.08445334e+00 -1.38655931e-01
4.94180202e-01 2.68743247e-01 -5.44078171e-01 -3.95986438e-01
-2.22095221e-01 4.31934893e-01 -5.60862005e-01 7.48914853e-03
-9.01879549e-01 -9.84516501e-01 -9.16945338e-01 -2.10245121e-02
8.04481879e-02 -1.21658611e+00 1.20195997e+00 -3.32782000e-01
-1.30285645e+00 1.27248871e+00 -4.73017067e-01 -1.44907951e-01
5.48107386e-01 -3.20177704e-01 2.41872445e-01 3.18741471e-01
3.63687634e-01 5.70959449e-01 7.60202944e-01 -1.81022298e+00
-1.54934868e-01 -8.00800502e-01 1.51930019e-01 5.11777639e-01
3.40317667e-01 -1.50694326e-01 -5.97046196e-01 -1.49948850e-01
8.45011771e-01 -1.15798187e+00 -5.27944684e-01 5.03838181e-01
-4.28566247e-01 -1.24546520e-01 6.01658702e-01 -6.09634332e-02
2.22718596e-01 -2.11223054e+00 2.78954387e-01 3.63630414e-01
4.14871603e-01 -2.48420253e-01 3.68728973e-02 6.72521174e-01
-9.92300212e-02 -8.03650841e-02 -3.19025367e-01 -1.00719035e+00
-6.33359551e-02 2.47536898e-01 -6.62927687e-01 9.91665483e-01
-8.45771879e-02 6.63015008e-01 -9.99607027e-01 -3.02298535e-02
5.09787142e-01 8.49737585e-01 -7.30606258e-01 2.13068143e-01
-1.79531291e-01 6.63767636e-01 -6.21585310e-01 6.49075210e-01
1.11203349e+00 -5.54890573e-01 -2.24536970e-01 1.27048075e-01
-1.32844388e-01 2.71366954e-01 -1.17816687e+00 2.43351102e+00
-5.05076170e-01 2.32425630e-01 3.12718451e-01 -5.76559544e-01
1.05743408e+00 2.58392811e-01 6.33767903e-01 -1.42355606e-01
-1.80707186e-01 3.85378599e-02 -5.64419448e-01 8.26876983e-02
5.36579072e-01 -3.97173129e-02 1.45815462e-01 3.44209701e-01
-3.75485092e-01 -9.34510231e-01 -5.27593493e-01 4.03405428e-01
1.08463955e+00 3.51560026e-01 6.70822784e-02 -4.73535061e-02
2.75314629e-01 6.47655949e-02 2.41995811e-01 3.89865041e-01
3.79610807e-01 1.46524012e+00 2.26967677e-01 -6.59428298e-01
-1.20206523e+00 -1.46526802e+00 -2.55402207e-01 4.01213229e-01
4.55205023e-01 -3.59399229e-01 -5.14824808e-01 -1.72460854e-01
2.71620721e-01 5.35706699e-01 -5.24477839e-01 2.94225007e-01
-4.92965162e-01 -8.47126618e-02 -9.48088691e-02 3.90300989e-01
1.61079258e-01 -5.78285515e-01 -7.94731975e-01 -1.69132575e-01
-5.31865098e-03 -1.11431837e+00 -2.26408228e-01 -3.15166175e-01
-1.22823179e+00 -1.27014911e+00 -8.01455617e-01 -6.67967319e-01
1.01929581e+00 9.82028604e-01 1.34829772e+00 -8.50321949e-02
3.16863731e-02 9.98536825e-01 -2.26903394e-01 -3.98856997e-01
-1.59360960e-01 -2.41467521e-01 1.15835644e-01 6.03000410e-02
2.07404509e-01 -9.54858124e-01 -9.07694578e-01 2.36325651e-01
-8.48287106e-01 1.71452805e-01 2.06958994e-01 2.77636856e-01
9.47398305e-01 -7.37946928e-01 -2.68078268e-01 -4.79174197e-01
7.57338703e-02 -6.67576611e-01 -8.24680507e-01 -1.94009855e-01
-1.26013741e-01 -1.59161359e-01 2.43684843e-01 1.99444313e-02
-6.14298046e-01 3.60753596e-01 2.54553169e-01 -1.12335479e+00
-1.18555613e-01 1.31601617e-01 1.11783125e-01 -1.99317738e-01
5.67319632e-01 3.80495697e-01 2.03534663e-01 -6.98508441e-01
4.81222928e-01 2.81301975e-01 5.12364209e-01 -6.35208786e-01
1.02007222e+00 1.35603154e+00 3.05070728e-01 -9.37676430e-01
-4.62669283e-01 -7.20073044e-01 -8.91131103e-01 -1.74646437e-01
5.22958934e-01 -1.29608119e+00 -8.04927766e-01 1.32162154e-01
-1.57611775e+00 6.61483929e-02 -3.61646384e-01 2.86222935e-01
-8.49094927e-01 4.16007727e-01 -1.60023302e-01 -7.06696689e-01
-2.11302847e-01 -1.06970656e+00 1.93378603e+00 -1.19397067e-01
1.35282492e-02 -8.23382139e-01 4.05028701e-01 2.32696339e-01
2.89917234e-02 4.23632711e-01 5.58040321e-01 -1.22723073e-01
-1.23161364e+00 -3.04026783e-01 -8.62908736e-02 -7.17049465e-02
-1.05891921e-01 -3.51965427e-02 -9.79847908e-01 -4.18743223e-01
1.83931008e-01 -8.92215967e-03 4.85805899e-01 3.77247006e-01
1.27230895e+00 1.42546460e-01 -6.69549584e-01 1.11772788e+00
1.54731750e+00 -3.53683382e-01 6.12067640e-01 3.06624830e-01
6.91875696e-01 3.47138226e-01 2.45386779e-01 6.75128996e-01
5.20415902e-01 6.91221952e-01 1.19127429e+00 1.27049964e-02
6.11541867e-02 -3.95842314e-01 -1.29017487e-01 7.16241896e-01
-3.38702887e-01 1.58848375e-01 -1.11044443e+00 5.64173222e-01
-1.53323245e+00 -6.33751810e-01 -3.65460187e-01 2.47088027e+00
1.45143703e-01 -2.53979295e-01 -1.78813279e-01 1.12679601e-03
5.17796159e-01 2.44808793e-01 -5.91951609e-01 -3.74669172e-02
5.86967357e-02 2.85041481e-01 4.10989195e-01 5.46798050e-01
-7.36271501e-01 4.99386162e-01 5.95710182e+00 7.28022773e-03
-7.85453439e-01 -9.91637353e-03 2.11814761e-01 -1.80270642e-01
-7.87050009e-01 1.44139484e-01 -6.98505342e-01 2.45068923e-01
3.65107447e-01 -1.51177198e-01 4.15695578e-01 1.00155461e+00
-1.11434460e-02 1.43196229e-02 -1.44459164e+00 1.39439285e+00
1.92401022e-01 -1.63509893e+00 2.53960520e-01 3.63585860e-01
8.52448404e-01 6.77683830e-01 2.16233745e-01 -1.94883212e-01
4.00307000e-01 -8.98049295e-01 7.95604765e-01 5.54425657e-01
1.09466386e+00 -4.45480525e-01 1.69386677e-02 6.72622085e-01
-1.01033735e+00 2.70087779e-01 -6.95304096e-01 -2.46228307e-01
2.11881682e-01 5.10544658e-01 -1.00560224e+00 7.26251245e-01
5.61291993e-01 7.76230752e-01 -2.25330845e-01 1.16197777e+00
-1.14115216e-02 1.52936265e-01 -5.91951787e-01 3.54388356e-01
2.30399892e-01 -7.77054727e-01 7.98811495e-01 5.95005631e-01
5.98458827e-01 2.65521407e-01 -4.44122329e-02 1.10688567e+00
-9.64782462e-02 -3.18496108e-01 -1.26687443e+00 7.50782371e-01
6.76311016e-01 1.18411851e+00 -5.99368393e-01 -6.22481965e-02
-7.58418441e-01 9.06528652e-01 4.26072538e-01 4.16966856e-01
-5.16481042e-01 1.56435922e-01 9.65082049e-01 5.27193546e-01
3.82255167e-01 -6.20040953e-01 -5.03582001e-01 -1.20003986e+00
3.44862312e-01 -3.84354383e-01 -2.14475632e-01 -1.34647489e+00
-1.18984151e+00 5.33711553e-01 4.68819365e-02 -1.77073836e+00
-1.22848518e-01 -5.48840046e-01 -5.08606017e-01 1.03150594e+00
-1.44510937e+00 -1.12772572e+00 -5.44257462e-01 6.92294121e-01
4.63825554e-01 1.37790084e-01 9.20433044e-01 -3.15962166e-01
2.24528685e-01 -2.03835800e-01 3.59408945e-01 -3.89832444e-02
9.53238234e-02 -1.06276298e+00 9.85564649e-01 4.85350966e-01
3.78079891e-01 8.10055017e-01 4.12097961e-01 -5.87691605e-01
-1.79848576e+00 -1.01983166e+00 4.43816453e-01 -1.01924133e+00
3.08184743e-01 -6.42202497e-01 -8.49506319e-01 1.01961422e+00
-1.52579919e-01 4.27745551e-01 2.59210408e-01 -5.64308167e-02
-5.95018983e-01 -1.04730269e-02 -1.12862968e+00 4.77956355e-01
1.20692897e+00 -4.88869429e-01 -7.88478315e-01 6.90059483e-01
6.65962458e-01 -9.45730448e-01 -5.26797771e-01 2.98798621e-01
2.04657570e-01 -1.18750608e+00 1.34747946e+00 -2.45405838e-01
4.80049819e-01 -3.14706326e-01 -4.31912750e-01 -1.33447599e+00
-2.52859265e-01 -6.33102119e-01 -1.96235999e-02 5.40097952e-01
-1.37656212e-01 -5.81771910e-01 1.30055106e+00 6.16034329e-01
-3.78614515e-01 -8.64004195e-01 -1.20948792e+00 -5.59834599e-01
1.00184776e-01 -5.77304363e-01 7.57650197e-01 7.21406996e-01
-2.86922216e-01 1.33988976e-01 2.73988307e-01 7.87841618e-01
1.11134684e+00 4.47956532e-01 1.18491662e+00 -1.61009824e+00
-7.59915486e-02 -8.72726217e-02 -5.92696011e-01 -1.61107266e+00
-5.69885597e-03 -9.35188234e-01 -2.03705177e-01 -1.73729599e+00
9.69053730e-02 -6.70312226e-01 3.79877537e-01 -1.42663583e-01
2.68978983e-01 8.26272890e-02 1.03417970e-01 6.39415741e-01
-4.43920046e-01 6.41854882e-01 1.29488683e+00 2.11476669e-01
7.78759792e-02 1.56535417e-01 -4.81100947e-01 9.26538527e-01
3.19481283e-01 -2.97108293e-01 -4.32926625e-01 -1.11729801e+00
4.46308017e-01 4.26472723e-01 6.04025066e-01 -9.68468249e-01
2.98819333e-01 5.16176922e-03 3.59159857e-01 -1.01563966e+00
1.02876556e+00 -1.07966864e+00 3.94708604e-01 -8.21581390e-03
1.24822974e-01 2.99424618e-01 -1.49566203e-01 9.33052301e-01
-2.06919387e-03 1.29448906e-01 5.29446900e-01 -4.40944046e-01
-9.62836668e-02 6.75468981e-01 3.69411588e-01 -2.02884942e-01
1.17633855e+00 -4.10456240e-01 -3.67960986e-03 -4.25981104e-01
-4.55480397e-01 1.34725660e-01 1.27034807e+00 3.09888095e-01
1.08208525e+00 -1.30616367e+00 -6.91247225e-01 7.20289111e-01
1.93957463e-01 1.08633971e+00 -1.65088307e-02 3.56652647e-01
-8.81699085e-01 4.27205861e-01 2.88374215e-01 -1.27801824e+00
-9.57210779e-01 4.97228205e-01 3.22687984e-01 2.54201144e-01
-1.11204422e+00 9.43008065e-01 6.17992222e-01 -7.47794330e-01
8.94034430e-02 -5.23582458e-01 1.88647807e-01 -5.77261686e-01
3.63528132e-01 3.90315890e-01 2.52451271e-01 -7.01856554e-01
-4.04440552e-01 1.05098581e+00 2.36068800e-01 -2.44928822e-01
1.62614918e+00 7.30410963e-02 -1.29134446e-01 3.67142022e-01
1.05203581e+00 1.26759216e-01 -1.64547849e+00 -3.14664513e-01
-4.71864939e-01 -8.73319268e-01 -2.58440584e-01 -2.17973575e-01
-8.86502266e-01 8.93610358e-01 8.93545616e-03 1.58623427e-01
5.16064465e-01 4.58659589e-01 4.07258660e-01 3.41692746e-01
1.07688808e+00 -2.13912070e-01 -8.86265635e-02 4.31547284e-01
1.25923753e+00 -9.95164633e-01 1.89966053e-01 -6.73037231e-01
-3.66225898e-01 1.02197361e+00 1.90158963e-01 -8.05369198e-01
6.25752151e-01 1.34333283e-01 -1.65729448e-01 -5.84048152e-01
-5.91586709e-01 3.35745186e-01 7.23214597e-02 8.97614717e-01
2.37330645e-02 4.56635207e-02 5.49671471e-01 2.99496446e-02
-3.51466000e-01 -1.12514466e-01 6.59203231e-01 7.32405126e-01
-3.01722467e-01 -7.41985500e-01 -7.78348565e-01 2.72763789e-01
1.09602645e-01 1.69343352e-01 -1.80204839e-01 7.39067137e-01
-3.22146326e-01 6.25314772e-01 7.03853846e-01 4.29240391e-02
5.59355319e-01 -2.48281866e-01 6.51512384e-01 -9.39880073e-01
-1.98537502e-02 -1.41594693e-01 -5.10856807e-01 -9.35132563e-01
-3.04908544e-01 -1.03719211e+00 -1.25403214e+00 -9.97361168e-02
1.41190410e-01 1.10093005e-01 1.03562796e+00 5.67237377e-01
6.73822761e-01 -1.53522223e-01 9.94818389e-01 -1.64933836e+00
-7.55069077e-01 -7.06606865e-01 -6.12139583e-01 4.23827827e-01
6.01278424e-01 -8.27439427e-01 -5.51711500e-01 -7.97604695e-02] | [8.416385650634766, -3.298004388809204] |
ed416320-9ac8-4b37-8a5c-8e3d124699c9 | corrnet3d-unsupervised-end-to-end-learning-of | 2012.15638 | null | https://arxiv.org/abs/2012.15638v2 | https://arxiv.org/pdf/2012.15638v2.pdf | CorrNet3D: Unsupervised End-to-end Learning of Dense Correspondence for 3D Point Clouds | Motivated by the intuition that one can transform two aligned point clouds to each other more easily and meaningfully than a misaligned pair, we propose CorrNet3D -- the first unsupervised and end-to-end deep learning-based framework -- to drive the learning of dense correspondence between 3D shapes by means of deformation-like reconstruction to overcome the need for annotated data. Specifically, CorrNet3D consists of a deep feature embedding module and two novel modules called correspondence indicator and symmetric deformer. Feeding a pair of raw point clouds, our model first learns the pointwise features and passes them into the indicator to generate a learnable correspondence matrix used to permute the input pair. The symmetric deformer, with an additional regularized loss, transforms the two permuted point clouds to each other to drive the unsupervised learning of the correspondence. The extensive experiments on both synthetic and real-world datasets of rigid and non-rigid 3D shapes show our CorrNet3D outperforms state-of-the-art methods to a large extent, including those taking meshes as input. CorrNet3D is a flexible framework in that it can be easily adapted to supervised learning if annotated data are available. The source code and pre-trained model will be available at https://github.com/ZENGYIMING-EAMON/CorrNet3D.git. | ['Ying He', 'Hui Yuan', 'Junhui Hou', 'Zhiyu Zhu', 'Yue Qian', 'Yiming Zeng'] | 2020-12-31 | null | http://openaccess.thecvf.com//content/CVPR2021/html/Zeng_CorrNet3D_Unsupervised_End-to-End_Learning_of_Dense_Correspondence_for_3D_Point_CVPR_2021_paper.html | http://openaccess.thecvf.com//content/CVPR2021/papers/Zeng_CorrNet3D_Unsupervised_End-to-End_Learning_of_Dense_Correspondence_for_3D_Point_CVPR_2021_paper.pdf | cvpr-2021-1 | ['3d-dense-shape-correspondence'] | ['computer-vision'] | [-0.11587847 0.20773318 0.1501286 -0.53207564 -0.80505306 -0.6342581
0.5780835 -0.18876946 0.01973961 0.2567884 0.10867994 -0.17093429
-0.09390073 -0.9424478 -1.3113217 -0.5810451 -0.09621061 1.0601618
0.11068848 -0.14852276 0.1268735 0.7726631 -1.2876692 0.03648499
0.7282014 0.73862 0.21598937 0.15085211 0.03128261 -0.05307716
0.13569644 -0.3933429 0.8598043 -0.05479088 -0.79400736 0.19295555
0.5447789 -0.29959244 -0.2283631 0.8607678 0.64665204 -0.05447237
0.7334352 -1.1313459 -0.8741671 0.308316 -0.60659945 -0.50966907
0.33968383 0.3573267 0.9800471 -1.290457 0.7973823 1.4297945
0.8098259 0.5259209 -1.3992704 -0.6264462 -0.09793008 -0.39986843
-1.3054857 -0.21240784 1.0470605 -0.7706102 0.68847054 0.23122288
0.72405416 0.96046877 0.1084805 0.49432302 0.72863805 -0.17938125
-0.07565513 -0.2450491 -0.21093248 0.8832473 0.16719474 0.33273044
-0.09476013 -0.29883426 1.3650541 0.25639334 -0.10057542 -0.97580665
-1.4015523 0.70137763 0.73771083 0.10336342 -0.40997595 0.13239668
0.16077015 0.3401344 0.6215632 0.4288627 -0.5913672 0.11093388
-0.556392 0.5035367 0.6572556 1.1088325 1.0493966 -0.26281092
0.23093559 0.7773897 0.5644926 0.49105403 0.12382051 -0.8672084
0.5645309 0.89188987 0.01125333 -1.1579807 -0.3301734 -0.27053618
-0.93297976 0.34716928 0.19512925 -0.00828903 -0.9638475 1.5731589
0.6360494 0.33355582 -0.22840983 1.1371342 0.83665925 0.5210363
-0.5040065 0.37926546 1.0258896 -0.7771364 -0.04139482 0.05231379
0.49672914 -1.0152441 1.2566313 -0.23944461 -1.3062731 -0.6166983
-0.7683865 -0.33865625 -0.11388559 0.08273268 0.3196024 -0.26672176
-1.0186661 0.8216989 -1.1029291 -0.30274546 0.5784444 0.4687933
-0.6317995 -0.03426262 -0.7102324 0.68377954 0.08516189 0.03239463
-0.75280124 -0.9388489 -1.092385 -0.22570187 0.07791507 -1.1661189
1.0731969 -0.6577569 -1.566657 1.262745 0.11659685 -0.07309517
0.802581 -0.23021941 0.07871442 -0.19979572 0.28009796 0.84799016
0.75779825 -1.5479616 -0.08269037 -0.35951397 -0.20553033 0.16692634
0.18483132 -0.3565663 -0.618688 -0.7577518 0.4425158 -1.185393
-0.17352992 0.46248767 -0.68804854 -0.24390021 0.9993262 -0.2985765
0.4761428 -2.2438738 0.48159003 0.31099856 0.31226596 0.0550286
-0.3922643 0.5528924 -0.38189813 0.04171951 -0.55962694 -0.4488305
0.1856346 0.34654948 -0.24825175 0.70503795 0.50601834 1.0777986
-0.9986905 -0.06486333 0.4162572 0.6346446 -0.72606367 0.4989565
-0.27932307 0.86184615 -0.42070836 0.48322582 1.0430888 -0.27422178
-0.17829643 -0.39552483 -0.07009485 0.3055123 -1.3246036 1.9888264
-0.26925403 0.1096677 -0.07286175 -0.9174548 1.2956655 0.15788822
0.72994614 -0.32119703 0.18515018 0.40812021 -0.13313451 -0.43433607
0.16043028 -0.11555045 -0.03508276 0.4409722 0.04274923 -0.60627
-0.3533838 -0.08881241 0.7980968 0.40805623 -0.06561068 -0.336395
0.51283175 -0.18386269 0.60888904 0.16754365 0.39183465 0.9403026
0.24758093 -0.76390123 -1.4598763 -1.4288881 -0.25069934 0.5081534
0.31789434 -0.17207229 -0.45720893 -0.67822164 0.602067 0.40234405
-0.574934 0.00637443 -0.70344573 -0.09589399 0.12422018 0.5367671
0.37337184 -0.89900136 -0.03763606 0.06984458 0.17497604 -0.8408291
-0.71336997 -0.00632119 -1.0734386 -1.0751191 -0.5544067 -1.173768
1.1792057 0.03196391 1.0878698 0.1599417 -0.00497546 0.40340832
-0.15696245 -0.23600955 -0.37682393 0.20180458 0.07870623 0.00990492
0.10362394 -0.97030455 -0.71633387 0.53611463 -0.89252627 0.25318962
0.424149 0.73152816 1.0525813 -0.4021639 0.13719395 -0.79170454
0.28180456 -0.42100403 -0.63593227 -0.25569952 -0.10385277 0.19300635
0.4795423 -0.31889102 -0.44833192 0.33477154 -0.36782444 -0.8987187
-0.08966193 0.22083084 -0.32789364 -0.0585398 0.50462943 -0.01141821
0.3562258 -0.9141836 0.5248941 0.42984375 0.6880226 -0.9608276
1.4573048 0.6505204 0.02242924 -0.4140065 -0.6688077 -0.31155318
-1.063355 -0.00964511 0.81283736 -0.9049786 -0.43790853 0.5870652
-1.284757 -0.5826532 -0.5431689 0.34143466 -0.798718 0.10509589
-0.52450764 -0.01144488 -0.44239843 -1.3030177 1.490326 -0.07503265
-0.13602722 -1.0396495 0.21441397 0.1843157 0.07452199 0.54715997
0.9389033 -0.6538116 -0.7891243 -0.2725992 -0.06212873 0.41620454
0.43915957 0.20170067 -0.6004423 -0.55925083 -0.16356371 -0.23574999
0.5045785 0.13382858 1.4027909 -0.38591665 -0.30810842 0.92530656
1.3767815 -0.30141485 0.58948725 0.35873824 1.099935 0.19975491
0.41100845 0.29440928 0.58529854 0.67613506 0.77249944 -0.35031024
0.06721763 -0.5312128 0.04446656 1.3163072 -0.21512535 0.10294212
-0.95011526 0.5022106 -1.8401531 -0.6462866 -0.03246222 2.2547307
0.980321 0.12571046 0.09232147 -0.15888676 0.6800351 0.09185683
-0.5707568 -0.20958579 0.15135384 0.26691654 0.20378473 0.6535157
-1.0158924 0.96022004 4.610509 0.49564993 -1.4610585 -0.06736865
0.29416037 0.19253944 -0.66970485 0.08830371 -0.5278943 0.46337578
0.28925833 -0.02374086 0.2614768 0.7555612 0.28279594 0.62619317
-1.4814398 0.8650222 -0.3859652 -1.627872 0.35903114 0.120764
0.84619874 0.27309904 0.00901749 -0.03037051 0.42768016 -0.83877784
0.80042696 0.7185644 0.9596423 -0.5393552 0.53924525 0.309543
-1.2737788 0.44666147 -0.471593 0.0367662 0.10502639 0.6761102
-0.8092512 0.740666 0.6610953 1.1251842 -0.20962815 0.965383
-0.24688321 0.2815536 -0.38378662 0.5485833 0.1917747 -0.4991722
0.8584024 0.90306544 0.37478194 -0.00634485 0.3955375 1.21701
-0.4690895 -0.06700859 -0.7949044 0.3031292 0.73802257 1.2096387
-0.38525963 -0.09498518 -0.31427675 1.0102499 0.39922088 0.1486159
-0.69032454 -0.17543183 0.9351357 0.5990961 0.48698086 -0.48496965
-0.3110648 -1.065749 0.1532515 -0.5988769 -0.00909234 -0.7663153
-1.7601072 0.5981925 -0.03443233 -1.629305 0.02425195 -0.45085588
-0.86694926 0.9313902 -1.276652 -1.347318 -0.45311418 0.49996883
0.41133353 -0.10397974 0.71236587 0.24729052 -0.28544334 0.5002994
0.15571845 0.45457613 0.7118218 -1.274084 0.8860434 0.42831722
0.03889221 0.65521294 0.37035143 -0.71413875 -1.5789591 -1.3875849
0.6398473 -0.59263885 0.5758583 -0.5661443 -1.1977532 0.92922884
-0.02948504 0.46067017 0.34715486 -0.32283682 -0.34497502 -0.12914576
-1.1541139 0.58593804 1.3702816 -0.48257688 -0.83079684 0.47362515
0.9844748 -1.0448124 -1.346789 0.62260324 0.3431766 -0.6536271
1.1501282 -0.5092117 0.6172681 -0.4340171 -0.09685367 -1.4746169
-0.43070334 -0.7048446 0.1327542 1.1215084 0.30054978 -0.8289819
0.7376298 0.50150895 -0.39947653 -1.2361159 -0.86417955 -0.8369439
0.47978687 -0.14587341 1.0346106 1.1981612 -0.57516825 0.18691526
-0.00553548 0.37636352 0.67291605 0.38222343 1.2427388 -1.3566711
-0.06764115 -0.35239828 -0.6586206 -1.341692 0.19127314 -1.3006793
-0.01080782 -1.4594549 -0.11703324 -0.9404299 0.1256925 0.7461643
0.04766895 0.07644873 0.23483703 0.45554933 -0.12728642 0.86415976
1.8485112 -0.07934922 -0.3048139 0.07659301 -0.5503786 0.82556
0.57377595 -0.5185311 -0.20686412 -0.79430056 -0.09500823 -0.11353926
0.5809156 -0.73183525 0.01128166 -0.04122923 0.35236037 -0.63477784
0.15553342 -1.0901152 0.3501848 0.1971962 -0.19662236 0.34924242
0.27445638 0.36503652 0.07059088 0.06941028 0.812908 -0.09269267
-0.21382804 0.9372319 0.36913648 0.16570608 0.939508 -0.17694399
-0.10066582 -0.06947286 -0.65789807 0.24915226 0.9790363 0.5830833
0.70330113 -1.7346842 -0.94239444 0.6554653 -0.04851333 1.047977
-0.09350452 0.63628006 -0.6025412 -0.02234792 -0.23485729 -1.0835136
-0.94039655 0.23956515 0.51933056 0.01735825 -0.97904974 0.78175217
0.25497398 -1.4072994 0.08349004 -0.5741334 0.46470663 -0.47905937
-0.03873875 0.10110861 0.10531268 -0.6848658 -0.3434167 1.0493553
-0.09193323 0.26050967 1.8944647 0.18253772 -0.35756826 0.26317927
1.5685378 -0.04438313 -1.5835948 -0.4271799 -0.13768026 -0.66728926
-0.35708553 -0.38670698 -1.3513404 0.6459321 0.37471157 -0.09281121
0.75538695 0.4086868 0.9259112 0.18346912 0.3402277 -0.4219303
0.29240146 0.60789627 1.4217122 -1.045105 0.03840794 -0.49510467
-0.36491996 1.1271864 0.637086 -0.7153484 0.9313014 0.16873959
0.05380977 -0.5158587 -0.47403064 0.0756028 0.5296928 0.535356
0.28043807 0.01829589 0.08175363 0.34910503 -0.45422578 -0.03208391
0.06606782 0.8880889 -0.13204423 -1.3864077 -0.38766375 0.42870152
0.22474478 0.28204674 -0.42557013 0.8308657 0.11259821 0.12795879
0.45016816 -0.66097146 0.74162525 -0.24657977 0.4395209 -0.8154199
-0.46168393 0.12941934 -0.3198254 -0.6217282 -0.30750844 -0.7831976
-1.4854593 -0.55452275 -0.07809153 -0.09020732 0.36904064 0.7531511
0.7880679 0.17570958 0.88828766 -1.5713477 -0.63101625 -0.69400656
-0.27714533 0.82721704 0.29176912 -0.85402757 -0.36266127 0.04981881] | [8.354134559631348, -3.350670337677002] |
0d2dde20-6186-43d3-9ddc-c11f16bc6318 | vizdoom-a-doom-based-ai-research-platform-for | 1605.02097 | null | http://arxiv.org/abs/1605.02097v2 | http://arxiv.org/pdf/1605.02097v2.pdf | ViZDoom: A Doom-based AI Research Platform for Visual Reinforcement Learning | The recent advances in deep neural networks have led to effective
vision-based reinforcement learning methods that have been employed to obtain
human-level controllers in Atari 2600 games from pixel data. Atari 2600 games,
however, do not resemble real-world tasks since they involve non-realistic 2D
environments and the third-person perspective. Here, we propose a novel
test-bed platform for reinforcement learning research from raw visual
information which employs the first-person perspective in a semi-realistic 3D
world. The software, called ViZDoom, is based on the classical first-person
shooter video game, Doom. It allows developing bots that play the game using
the screen buffer. ViZDoom is lightweight, fast, and highly customizable via a
convenient mechanism of user scenarios. In the experimental part, we test the
environment by trying to learn bots for two scenarios: a basic move-and-shoot
task and a more complex maze-navigation problem. Using convolutional deep
neural networks with Q-learning and experience replay, for both scenarios, we
were able to train competent bots, which exhibit human-like behaviors. The
results confirm the utility of ViZDoom as an AI research platform and imply
that visual reinforcement learning in 3D realistic first-person perspective
environments is feasible. | ['Wojciech Jaśkowski', 'Michał Kempka', 'Marek Wydmuch', 'Jakub Toczek', 'Grzegorz Runc'] | 2016-05-06 | null | null | null | null | ['game-of-doom', 'fps-games'] | ['playing-games', 'playing-games'] | [-5.22394955e-01 -2.76572317e-01 2.98371583e-01 4.80455101e-01
1.10746667e-01 -5.20793200e-01 5.52411675e-01 -7.23036706e-01
-8.94940138e-01 8.33527625e-01 -4.67111945e-01 -4.10781592e-01
-1.46481425e-01 -7.43193328e-01 -6.60179913e-01 -5.39076805e-01
-2.44411379e-01 5.43189168e-01 6.14418149e-01 -1.02669370e+00
3.42255265e-01 3.73378009e-01 -1.88186204e+00 2.26475969e-02
5.43850780e-01 6.87624693e-01 6.28974974e-01 1.15807343e+00
4.70750213e-01 1.31040144e+00 -1.02656782e+00 -2.75471900e-02
7.30418026e-01 -3.43610257e-01 -5.83071291e-01 1.58892758e-02
-4.18148823e-02 -6.36484802e-01 -5.62653065e-01 7.46175647e-01
7.33115375e-01 2.33762413e-01 2.89145321e-01 -1.59151947e+00
-2.78745860e-01 1.17481172e-01 -3.95448178e-01 2.44230524e-01
6.09901309e-01 1.25660384e+00 5.05404949e-01 -1.31426841e-01
6.37375295e-01 1.27658451e+00 5.45294762e-01 9.07396734e-01
-8.53198171e-01 -4.94191945e-01 -3.19856107e-01 4.61739630e-01
-1.03216922e+00 2.71958202e-01 5.68074405e-01 -5.34700692e-01
1.20399117e+00 -5.52326180e-02 1.26802731e+00 1.74733937e+00
5.11757970e-01 6.97046578e-01 1.39487290e+00 -3.00184041e-01
6.98135316e-01 -5.44842742e-02 -5.09539664e-01 7.44039595e-01
1.04428135e-01 8.96210849e-01 -2.61129022e-01 1.89129859e-01
1.33289635e+00 -3.51472616e-01 -4.16793041e-02 -6.44692600e-01
-1.19969189e+00 9.17297542e-01 5.96682072e-01 8.25711265e-02
-5.86600304e-01 5.56752384e-01 4.63804811e-01 4.79264736e-01
-4.42533225e-01 8.01057041e-01 -3.11952204e-01 -5.44590950e-01
-2.83201158e-01 8.99698973e-01 7.00827658e-01 8.16688120e-01
3.96249950e-01 6.93293333e-01 -1.29891306e-01 2.21018702e-01
2.12695196e-01 4.55383629e-01 6.29605055e-01 -1.47376275e+00
2.25865796e-01 4.27300930e-01 4.26693976e-01 -6.81853294e-01
-6.23389542e-01 -2.17892006e-01 -3.51919323e-01 1.48724830e+00
7.16996193e-01 -7.06887782e-01 -9.61197257e-01 1.40832639e+00
3.74384731e-01 9.33103114e-02 1.49890125e-01 1.29705858e+00
6.78410232e-01 5.66932261e-01 -9.30423886e-02 3.04852873e-01
1.08681953e+00 -1.04505479e+00 -3.47645760e-01 -1.70419872e-01
5.46579838e-01 -1.33433014e-01 1.43190539e+00 8.53901088e-01
-1.00396430e+00 -8.47750783e-01 -9.93687809e-01 5.42858243e-01
-4.40129876e-01 -2.05385610e-01 7.08243191e-01 7.76292026e-01
-1.21069312e+00 5.64725578e-01 -6.64134800e-01 -4.91093248e-01
1.62970394e-01 4.90604162e-01 -2.71069378e-01 3.15871447e-01
-1.20798624e+00 1.15958083e+00 5.56800961e-01 -1.69450149e-01
-1.71797836e+00 -2.52874792e-01 -5.77532172e-01 -8.21936950e-02
7.90449858e-01 -6.84482336e-01 1.41035879e+00 -9.44677651e-01
-2.10375047e+00 6.67029500e-01 8.73992622e-01 -4.56900567e-01
7.96094835e-01 -1.39945984e-01 -1.41761694e-02 9.28562656e-02
-6.96460307e-02 8.47900510e-01 9.62338746e-01 -1.20405138e+00
-6.87158167e-01 -1.47608548e-01 7.08414376e-01 3.59526008e-01
1.74499571e-01 -1.36438400e-01 -3.30197737e-02 -2.44982406e-01
-7.85029709e-01 -9.44668889e-01 -4.41295266e-01 -1.44689176e-02
4.00195718e-02 -1.60147578e-01 8.25743914e-01 -1.59964219e-01
4.21200812e-01 -1.75752378e+00 1.55648142e-01 -4.73340377e-02
2.32332885e-01 7.83650041e-01 -9.24904794e-02 6.10616624e-01
1.22084372e-01 -2.56299794e-01 2.81028569e-01 3.01734447e-01
-3.80070284e-02 3.35074961e-01 9.64626223e-02 1.24000855e-01
-2.88812686e-02 9.33605194e-01 -1.14236939e+00 -2.29428306e-01
5.19865632e-01 1.43968925e-01 -6.89030290e-01 4.12164330e-01
-4.22945440e-01 8.89930964e-01 -3.13574761e-01 4.43565667e-01
3.27406168e-01 1.63989156e-01 -1.66239411e-01 5.80513656e-01
-3.28052670e-01 -3.51493984e-01 -1.08701313e+00 1.75477457e+00
-3.66735697e-01 8.31592023e-01 5.50196767e-02 -9.35994506e-01
9.55036521e-01 2.62852162e-01 4.05963778e-01 -1.08608675e+00
5.97913921e-01 1.22726122e-02 2.47046649e-01 -1.02731097e+00
4.78416920e-01 -9.51394364e-02 -1.11537158e-01 1.41685396e-01
1.84661925e-01 -6.47211611e-01 4.45931733e-01 -2.08355546e-01
1.42854619e+00 7.47023284e-01 1.38033122e-01 -7.99774379e-02
3.64749223e-01 5.61518371e-01 2.27204159e-01 1.12058985e+00
-6.11131132e-01 3.42248261e-01 5.97515821e-01 -8.10839117e-01
-1.24811411e+00 -1.15660667e+00 7.51924098e-01 8.28197122e-01
2.68461168e-01 -1.84250593e-01 -1.03050160e+00 -5.83961844e-01
-2.46512875e-01 7.07539916e-01 -5.63986599e-01 -2.05958128e-01
-5.62361956e-01 -1.07108697e-01 7.05115318e-01 2.82078356e-01
8.39103937e-01 -1.81673193e+00 -1.74984300e+00 2.30860859e-01
3.21069777e-01 -1.04770291e+00 1.72604769e-01 2.68478394e-01
-4.91379708e-01 -1.26212764e+00 -8.17542911e-01 -6.39501095e-01
-5.30097038e-02 2.37946689e-01 1.01873016e+00 1.26230922e-02
-4.19013441e-01 6.76683068e-01 -6.24555767e-01 -5.17718911e-01
-3.92762363e-01 -3.26065242e-01 1.79885253e-01 -6.63979292e-01
2.14615539e-01 -5.21691740e-01 -4.56264734e-01 3.46752644e-01
-7.39991605e-01 3.43661606e-02 5.07556081e-01 1.01805711e+00
-4.65734489e-02 1.64189562e-01 1.08416915e-01 -1.95276588e-01
9.75850999e-01 -1.19168147e-01 -1.08906066e+00 -9.20110792e-02
-8.04353133e-02 -1.50152102e-01 9.14936423e-01 -7.12685049e-01
-7.88257241e-01 1.82845294e-01 -1.85496107e-01 -5.95011592e-01
-5.59146702e-01 1.93825260e-01 1.25689954e-01 -2.79175878e-01
1.27378178e+00 1.47265658e-01 1.08247571e-01 1.70770451e-01
2.55885214e-01 7.11572945e-01 4.91223067e-01 -4.72691774e-01
9.19019401e-01 2.75825024e-01 2.63904836e-02 -9.76465523e-01
-2.17476692e-02 -4.91841026e-02 -5.02941608e-01 -7.63902366e-01
1.15100741e+00 -7.80769169e-01 -1.65011525e+00 9.46370542e-01
-1.06383741e+00 -1.14801908e+00 -4.08842176e-01 5.06021321e-01
-1.24652421e+00 2.07726538e-01 -4.18164104e-01 -8.10502052e-01
3.61030810e-02 -1.38522696e+00 6.36255205e-01 6.88648462e-01
5.62421121e-02 -7.56104887e-01 4.35933888e-01 2.89546132e-01
4.09897327e-01 2.88034379e-01 4.75432724e-01 -7.89409727e-02
-4.55965906e-01 1.61283031e-01 2.32987121e-01 1.23281956e-01
-2.05501556e-01 -8.35705176e-02 -7.89895117e-01 -3.49518865e-01
-1.94474161e-01 -9.09825742e-01 2.62675762e-01 3.11711133e-01
8.51377606e-01 2.25982293e-01 1.12718523e-01 3.76036167e-01
1.28454518e+00 6.46934628e-01 9.85413313e-01 9.84792769e-01
5.72191358e-01 3.47769469e-01 7.75435686e-01 6.39026105e-01
1.80417150e-01 8.44174385e-01 1.09092426e+00 5.76295191e-03
1.80043966e-01 -2.75744408e-01 5.93174458e-01 5.02224155e-02
-6.31253541e-01 -2.23929226e-01 -9.52858806e-01 2.31656820e-01
-1.80847371e+00 -1.19070709e+00 -3.42080444e-02 2.07194185e+00
3.71688336e-01 3.33068490e-01 7.41903841e-01 9.24117938e-02
5.45690179e-01 -2.99572170e-01 -5.90566456e-01 -5.93898356e-01
2.92814016e-01 3.94185364e-01 5.22295654e-01 2.69019157e-01
-9.65113461e-01 1.38140059e+00 6.52137804e+00 6.97402537e-01
-1.22165608e+00 -1.22053236e-01 4.91865650e-02 -9.27596837e-02
4.29841638e-01 -1.70547962e-01 -4.20203537e-01 2.86211163e-01
9.40917730e-01 1.50608897e-01 9.16620195e-01 1.10856009e+00
7.36042202e-01 -4.53591675e-01 -6.38178945e-01 1.15184927e+00
-3.08107942e-01 -1.12503600e+00 -3.03776473e-01 1.47363827e-01
4.62586135e-01 3.29881310e-02 8.96001160e-02 8.97845089e-01
8.82716179e-01 -1.23786187e+00 8.29623580e-01 2.61944175e-01
5.24280906e-01 -9.71566200e-01 6.18381143e-01 8.51611614e-01
-7.87260413e-01 -6.21078253e-01 -5.05482078e-01 -7.26189137e-01
-2.46943273e-02 -5.15842617e-01 -8.13848555e-01 2.98096448e-01
1.05897939e+00 2.54868984e-01 -5.16561687e-01 1.38417435e+00
-3.08776498e-01 2.64747828e-01 3.23577151e-02 -6.55714571e-01
7.32733965e-01 -2.66397804e-01 5.94318926e-01 5.57378173e-01
1.49523839e-01 -1.01513244e-01 1.69633955e-01 8.84238958e-01
6.73012733e-01 -4.28609222e-01 -1.03275251e+00 3.87648903e-02
-6.11909963e-02 1.13494706e+00 -7.43303001e-01 -5.04255891e-02
-1.83271453e-01 1.03947961e+00 3.60764354e-01 3.31246138e-01
-1.23103237e+00 -5.00869036e-01 6.96459413e-01 -7.68731767e-03
3.98588657e-01 -5.47404230e-01 4.95998785e-02 -8.43329787e-01
-3.15037400e-01 -1.39763045e+00 -1.52139291e-01 -1.38008094e+00
-7.12736070e-01 6.92797124e-01 8.37321803e-02 -1.40246749e+00
-5.31107903e-01 -1.23281443e+00 -7.40753889e-01 5.08078218e-01
-1.01913512e+00 -8.92007291e-01 -6.36326134e-01 1.01810098e+00
7.15032339e-01 -6.97286308e-01 6.75279975e-01 -1.52983561e-01
-3.96791577e-01 1.91139460e-01 -2.05771402e-01 -4.06088978e-02
2.48995230e-01 -1.38102031e+00 5.20074546e-01 5.12940586e-01
-2.22656764e-02 8.79164934e-02 8.33660364e-01 -5.25805414e-01
-1.48124695e+00 -5.48173726e-01 -2.86529481e-01 -4.68546778e-01
6.04571581e-01 -3.88707101e-01 -3.45843166e-01 2.87023067e-01
5.63844681e-01 -2.59876072e-01 -3.99937108e-02 -4.61654574e-01
1.36178330e-01 2.36737654e-02 -1.10381758e+00 1.15367424e+00
1.05371392e+00 -9.95516106e-02 -6.40775323e-01 4.90477905e-02
4.89737540e-01 -6.02460802e-01 -3.08941066e-01 -3.80230173e-02
4.98445570e-01 -1.31204629e+00 9.34137166e-01 -8.67395878e-01
5.63639581e-01 -4.54127371e-01 1.79109856e-01 -1.73265147e+00
-3.34188253e-01 -9.21348155e-01 6.14012666e-02 3.22477847e-01
1.02481700e-03 -2.89889306e-01 1.11853075e+00 4.56628017e-02
-4.50941883e-02 -3.69749069e-01 -8.79452050e-01 -1.03441632e+00
1.20313689e-01 -3.16525191e-01 2.62937307e-01 3.27757448e-01
1.75420940e-01 2.18923554e-01 -7.56260872e-01 -3.76370959e-02
3.97351205e-01 -6.81768239e-01 1.32043338e+00 -1.10899329e+00
-7.85373747e-01 -2.73171514e-01 -6.91775382e-01 -8.64572108e-01
-1.84481067e-03 -1.71807110e-01 2.29976296e-01 -1.49344385e+00
-3.01924109e-01 3.83170806e-02 2.07361832e-01 2.97851682e-01
4.98919077e-02 1.20752588e-01 2.75089949e-01 -4.00939584e-02
-5.43726921e-01 5.92967391e-01 1.70418155e+00 -1.43377483e-01
-4.25829679e-01 9.39348415e-02 -2.39364952e-01 6.92675591e-01
1.11873627e+00 -2.20734015e-01 -7.95219302e-01 -2.28378072e-01
2.34332010e-01 2.01827720e-01 7.45488524e-01 -1.64148152e+00
1.91044509e-01 -4.25904661e-01 3.19300324e-01 -2.59078622e-01
5.88039398e-01 -1.11183977e+00 -9.18719471e-02 1.08512902e+00
8.82632956e-02 3.86206478e-01 3.25037479e-01 4.44324434e-01
2.30542034e-01 -3.84096265e-01 6.67545021e-01 -7.58579373e-01
-1.34070837e+00 -6.24764599e-02 -1.25137210e+00 1.35666102e-01
1.67061985e+00 -6.34826779e-01 -4.29635167e-01 -6.63601279e-01
-6.27423823e-01 2.40529284e-01 4.45999324e-01 3.86474311e-01
7.72481859e-01 -1.01407766e+00 -4.80377078e-01 2.11303249e-01
-1.75497364e-02 -2.86645591e-01 1.74392760e-01 4.01027024e-01
-1.36932182e+00 1.87181994e-01 -1.25556278e+00 -6.48241937e-01
-1.07191443e+00 9.09163594e-01 7.43821323e-01 -2.27546453e-01
-7.18072534e-01 3.50923896e-01 -7.91257098e-02 -7.45605707e-01
2.38993868e-01 7.67757520e-02 -4.90243852e-01 -6.09618127e-01
4.77935821e-01 3.77363205e-01 -2.98740655e-01 -1.80440322e-01
3.49287987e-02 2.42585853e-01 3.97200853e-01 -5.19310951e-01
1.32482719e+00 2.23718405e-01 4.75872129e-01 2.14676395e-01
2.06347987e-01 -5.68108737e-01 -1.82284248e+00 4.76219654e-01
-1.69071525e-01 -5.41327477e-01 -1.60964414e-01 -7.14816093e-01
-9.41735327e-01 9.36991751e-01 9.24071968e-01 1.92035303e-01
8.30612838e-01 -4.85082328e-01 3.72566968e-01 9.31136370e-01
9.73456383e-01 -1.29114604e+00 7.88883865e-01 9.15214419e-01
8.23473811e-01 -1.10304344e+00 -5.37527740e-01 1.18486159e-01
-1.03558755e+00 1.12213838e+00 1.33259630e+00 -5.48603058e-01
1.19704463e-01 3.31677675e-01 4.41198409e-01 -3.54160935e-01
-7.48152971e-01 -5.81528068e-01 -4.02020663e-01 1.54291654e+00
-2.21926704e-01 -1.18431643e-01 8.55046287e-02 7.74472356e-02
-3.92228901e-01 2.22343817e-01 9.66201425e-01 9.77411449e-01
-6.41293049e-01 -7.79962838e-01 -6.42351210e-01 -2.66778827e-01
8.74367058e-02 4.64770436e-01 -4.22791153e-01 1.38095593e+00
2.32595935e-01 1.13181472e+00 -2.24719405e-01 -7.52296209e-01
5.45157254e-01 -4.55732077e-01 7.59723783e-01 -3.60230714e-01
-8.89499724e-01 -3.24799061e-01 -1.26187712e-01 -7.21278787e-01
-1.23533510e-01 -1.58443466e-01 -9.98773992e-01 -5.85289121e-01
8.24643970e-02 -1.06780402e-01 5.26368320e-01 8.84107411e-01
-5.96864009e-03 6.83687449e-01 4.54467773e-01 -1.40161288e+00
-6.55249119e-01 -8.82191718e-01 -6.35028899e-01 1.47498652e-01
2.46755779e-02 -9.85404253e-01 5.68535849e-02 -3.12807709e-01] | [3.752293825149536, 1.4696509838104248] |
377be5ac-c6e3-4622-9a9b-0bb74cad17f8 | toolformer-language-models-can-teach | 2302.04761 | null | https://arxiv.org/abs/2302.04761v1 | https://arxiv.org/pdf/2302.04761v1.pdf | Toolformer: Language Models Can Teach Themselves to Use Tools | Language models (LMs) exhibit remarkable abilities to solve new tasks from just a few examples or textual instructions, especially at scale. They also, paradoxically, struggle with basic functionality, such as arithmetic or factual lookup, where much simpler and smaller models excel. In this paper, we show that LMs can teach themselves to use external tools via simple APIs and achieve the best of both worlds. We introduce Toolformer, a model trained to decide which APIs to call, when to call them, what arguments to pass, and how to best incorporate the results into future token prediction. This is done in a self-supervised way, requiring nothing more than a handful of demonstrations for each API. We incorporate a range of tools, including a calculator, a Q\&A system, two different search engines, a translation system, and a calendar. Toolformer achieves substantially improved zero-shot performance across a variety of downstream tasks, often competitive with much larger models, without sacrificing its core language modeling abilities. | ['Thomas Scialom', 'Nicola Cancedda', 'Luke Zettlemoyer', 'Maria Lomeli', 'Roberta Raileanu', 'Roberto Dessì', 'Jane Dwivedi-Yu', 'Timo Schick'] | 2023-02-09 | null | null | null | null | ['math-word-problem-solving', 'math-word-problem-solving', 'math-word-problem-solving'] | ['knowledge-base', 'reasoning', 'time-series'] | [-1.05595209e-01 6.77768067e-02 -3.80030185e-01 -3.06246459e-01
-8.85649025e-01 -9.81289029e-01 7.85325050e-01 1.52143538e-01
-4.18456912e-01 4.83884424e-01 -1.76328391e-01 -9.63696122e-01
2.43359715e-01 -7.26350367e-01 -7.13945329e-01 4.90972921e-02
8.01431984e-02 6.14245117e-01 2.55243897e-01 -4.65490699e-01
3.87716055e-01 1.54882997e-01 -1.37900698e+00 3.76598537e-01
9.73813295e-01 3.92570227e-01 3.38944227e-01 8.60171139e-01
-6.07174158e-01 1.35367405e+00 -5.60222507e-01 -7.79749751e-01
-3.02048102e-02 3.89536992e-02 -1.04135978e+00 -6.44897759e-01
1.91193759e-01 -5.35258174e-01 -1.35556944e-02 6.03464544e-01
3.17512363e-01 1.09934388e-02 3.43085825e-01 -1.22942209e+00
-4.08232749e-01 1.23567200e+00 -4.72690351e-02 8.46606344e-02
6.54940724e-01 5.27514815e-01 1.07251716e+00 -6.82735562e-01
4.94937658e-01 1.00689507e+00 6.91030741e-01 4.84717399e-01
-1.17228246e+00 -7.23357260e-01 -7.97335897e-03 -1.20901220e-01
-1.05994701e+00 -5.36706269e-01 2.70123005e-01 -5.60393631e-01
1.53493536e+00 2.05018088e-01 4.27926868e-01 9.30924952e-01
2.07454860e-01 9.03934002e-01 7.39696980e-01 -5.28286934e-01
1.71396844e-02 3.26930791e-01 1.89521238e-01 1.02248645e+00
-1.05539314e-01 -2.74017543e-01 -6.18137062e-01 -2.51550704e-01
3.56797814e-01 -8.83189309e-03 8.06769952e-02 -1.40697062e-02
-1.23471296e+00 3.97584409e-01 -7.14953467e-02 4.05618548e-01
1.81209698e-01 4.64289069e-01 3.94203514e-01 6.35076463e-01
-1.05233185e-01 1.03469181e+00 -1.15383029e+00 -8.28352332e-01
-8.56882393e-01 1.65870577e-01 1.38283682e+00 1.24097419e+00
6.06914639e-01 2.34518275e-02 1.32014155e-01 5.51966131e-01
9.31053534e-02 3.48965466e-01 7.14313090e-01 -1.03936017e+00
7.23705590e-01 5.38813055e-01 1.46619663e-01 -2.95462698e-01
-3.31985831e-01 -3.93900514e-01 -1.03844039e-01 1.04934610e-01
5.92800140e-01 -1.92962736e-01 -5.70395112e-01 1.63503444e+00
-7.65231997e-03 9.58094373e-02 -1.33727580e-01 1.45523384e-01
5.18130839e-01 6.40798748e-01 2.81856120e-01 1.80558294e-01
1.12070203e+00 -1.18388164e+00 -1.49920195e-01 -5.60388803e-01
1.44319117e+00 -1.08341110e+00 1.55839872e+00 7.73516417e-01
-1.53671205e+00 -5.61014712e-01 -8.39153290e-01 -4.99009132e-01
-5.97829878e-01 3.20770293e-01 9.49758887e-01 3.93133372e-01
-1.08136928e+00 1.01125717e+00 -1.01579416e+00 -2.04681620e-01
8.78524780e-03 2.91517675e-01 -2.56280303e-01 -4.54568677e-02
-1.00330377e+00 1.19278359e+00 1.06369391e-01 -3.98710459e-01
-8.24215949e-01 -9.27400887e-01 -1.00812316e+00 3.58270347e-01
3.77368122e-01 -7.78295577e-01 1.94803762e+00 -6.20686889e-01
-1.74683714e+00 8.13456476e-01 -1.99138597e-02 -2.44215325e-01
4.93578315e-01 -5.33342957e-01 -1.29917994e-01 -4.14542437e-01
2.34502882e-01 4.75252390e-01 4.23522502e-01 -5.45840800e-01
-6.17029011e-01 4.50349487e-02 4.64170396e-01 3.01575214e-02
-3.49779516e-01 2.34168783e-01 -6.05784237e-01 -2.14798301e-01
-3.32376927e-01 -7.51486421e-01 -2.28388712e-01 -1.43686458e-01
-4.06543940e-01 -4.67540383e-01 1.15703195e-01 -6.04238212e-01
1.56711698e+00 -1.95758367e+00 1.22485518e-01 -6.25930205e-02
1.96166173e-01 3.01737219e-01 -1.95333213e-01 6.74892843e-01
2.82830279e-02 4.65063840e-01 2.15537772e-01 -1.83993414e-01
4.02648330e-01 2.75476426e-02 -3.85491222e-01 -1.72812313e-01
2.51092404e-01 9.87092972e-01 -1.10172820e+00 -4.14777994e-01
1.09680891e-01 -2.53300294e-02 -6.69043243e-01 2.73766130e-01
-6.71964467e-01 -1.59600258e-01 -4.79454190e-01 6.55952573e-01
6.14742078e-02 -6.35276675e-01 2.64929563e-01 3.44984561e-01
-2.74373859e-01 9.73976851e-01 -1.03420591e+00 1.93876839e+00
-1.03656709e+00 4.05603915e-01 3.50702740e-02 -7.60684431e-01
6.96270287e-01 2.31308520e-01 -8.71383548e-02 -4.77423668e-01
-1.62539855e-01 4.92434621e-01 2.06574112e-01 -6.44841850e-01
4.29674596e-01 1.92937970e-01 -2.42770940e-01 8.17475379e-01
1.37458026e-01 -4.90650982e-01 4.38449442e-01 4.38080996e-01
1.51514041e+00 5.71863294e-01 4.64200079e-01 7.13672265e-02
3.05969864e-01 1.83494791e-01 2.08777532e-01 1.01513541e+00
3.59251529e-01 -2.20187679e-02 6.25623465e-01 -4.55178916e-01
-9.54086900e-01 -9.54478800e-01 2.82089651e-01 1.81463146e+00
-4.37197655e-01 -8.97311330e-01 -6.22864485e-01 -4.38138574e-01
3.02806543e-03 1.18849492e+00 -8.09385702e-02 -2.08745033e-01
-8.05946231e-01 -9.90567356e-02 7.20708907e-01 6.18244052e-01
2.53761500e-01 -9.58822548e-01 -5.78444302e-01 3.79701525e-01
1.61926016e-01 -8.43254566e-01 -6.24314189e-01 6.45245373e-01
-9.62917745e-01 -8.62132311e-01 -2.65076727e-01 -8.13937604e-01
3.84205669e-01 1.43721610e-01 1.51355302e+00 3.76183778e-01
-2.46419191e-01 2.59029061e-01 1.03206508e-01 -4.74856257e-01
-6.62805200e-01 4.86002117e-01 -1.78875998e-01 -1.02400756e+00
3.98993492e-01 -7.01605678e-01 -1.93149373e-02 1.34798184e-01
-4.35618192e-01 3.14093322e-01 7.71347046e-01 7.45677650e-01
-4.98833060e-02 -2.94608265e-01 2.97987789e-01 -1.03242409e+00
6.99362099e-01 -5.12378156e-01 -6.83326840e-01 4.73755091e-01
-6.71577871e-01 3.33900750e-01 1.16846371e+00 -5.64312398e-01
-7.27786601e-01 1.16234049e-02 -1.16138615e-01 -1.06974922e-01
-2.66569480e-02 6.97606087e-01 -1.22746922e-01 1.43936768e-01
8.01731110e-01 4.83760267e-01 6.01228327e-03 -5.91187835e-01
7.23264813e-01 5.50866544e-01 4.60855246e-01 -1.25248790e+00
8.93230617e-01 -2.10916236e-01 -4.83121872e-01 -4.00245190e-01
-6.83792651e-01 -3.32752734e-01 -5.13886094e-01 2.89491415e-01
3.02381009e-01 -7.82242954e-01 -9.76943970e-01 2.16892526e-01
-1.20253360e+00 -9.76935863e-01 -1.94735229e-01 1.48529068e-01
-4.82762098e-01 1.01673141e-01 -9.01819766e-01 -5.13497055e-01
-3.83638024e-01 -1.29135370e+00 8.61129224e-01 2.02956781e-01
-6.88140273e-01 -9.26169813e-01 -1.39134333e-01 4.24469054e-01
7.04831004e-01 -4.69217807e-01 1.43060231e+00 -1.04574239e+00
-8.07559133e-01 -2.72441417e-01 5.50598577e-02 3.33414584e-01
-7.51796598e-03 2.57722855e-01 -5.71016550e-01 5.66301756e-02
-3.03109765e-01 -5.01453042e-01 3.34138960e-01 -2.99422264e-01
1.28095794e+00 -3.21943104e-01 -5.13314009e-01 7.02140570e-01
8.46246362e-01 3.14895630e-01 1.62112206e-01 4.11192924e-01
4.55845386e-01 1.36999100e-01 1.81623697e-01 9.56600010e-02
6.21560812e-01 6.04482114e-01 -4.61304858e-02 2.39152119e-01
2.72363890e-02 -4.57524776e-01 6.56785607e-01 1.09583557e+00
4.62854430e-02 4.41248789e-02 -1.33728671e+00 3.62163335e-01
-1.60654414e+00 -6.15551591e-01 1.98624566e-01 2.06954193e+00
1.36519003e+00 6.15180314e-01 -6.45729750e-02 -1.80956736e-01
-6.71072602e-02 -1.50756866e-01 -5.98015964e-01 -6.67538226e-01
4.29937124e-01 6.58949196e-01 1.63262337e-01 5.24616718e-01
-6.99846745e-01 1.36574113e+00 7.00043106e+00 7.84583390e-01
-1.24357045e+00 -4.69421372e-02 3.69102657e-01 3.97284795e-03
-3.93227428e-01 1.90023810e-01 -8.51535678e-01 4.91469920e-01
1.28662157e+00 -5.02727389e-01 8.78382862e-01 1.28786623e+00
-7.05038989e-03 2.18850207e-02 -1.65445971e+00 6.32225871e-01
-1.65717036e-01 -1.46314502e+00 2.67788377e-02 -3.23424637e-01
2.51821280e-01 2.12030545e-01 1.65505223e-02 8.13584328e-01
9.47196305e-01 -1.27770400e+00 6.86395705e-01 4.10886377e-01
6.00401402e-01 -3.90638620e-01 2.56695658e-01 7.93872535e-01
-9.93075609e-01 -2.88822390e-02 -8.99290815e-02 -4.94479209e-01
-2.41304964e-01 4.32405733e-02 -9.27392125e-01 1.92079827e-01
3.66152346e-01 5.58722794e-01 -6.48250580e-01 8.60163450e-01
-5.71661592e-01 6.67867839e-01 -4.09582406e-01 -3.91056925e-01
1.69554025e-01 1.36570647e-01 5.76155633e-02 1.41507268e+00
4.01986569e-01 1.00474291e-01 2.01892421e-01 8.87797296e-01
-1.18376501e-01 1.37265727e-01 -5.63917637e-01 -4.28120673e-01
6.45634174e-01 1.28157306e+00 -3.03078920e-01 -6.74489021e-01
-6.53063893e-01 7.18752503e-01 5.36514282e-01 2.95140803e-01
-6.54811740e-01 -7.09518969e-01 6.48867786e-01 1.90538064e-01
1.55793270e-02 -5.14495730e-01 -3.06117386e-01 -1.38189781e+00
4.21183854e-02 -1.26861680e+00 1.16571173e-01 -1.25154078e+00
-1.01672745e+00 1.66116506e-01 -6.21869639e-02 -8.92047346e-01
-5.15888929e-01 -7.66265273e-01 -9.92752910e-01 1.11375237e+00
-1.23720467e+00 -8.63532305e-01 4.77634892e-02 3.88466835e-01
5.39911985e-01 -2.44806826e-01 9.78668153e-01 3.66836041e-01
-4.06779230e-01 6.19410634e-01 -1.11348584e-01 3.21442068e-01
7.96402872e-01 -1.39297950e+00 8.23838830e-01 5.65664351e-01
1.30417705e-01 1.36802351e+00 6.37641728e-01 -4.10498679e-01
-1.88550639e+00 -7.42988229e-01 1.14515638e+00 -8.05184126e-01
1.18148589e+00 -3.43335658e-01 -7.45261192e-01 1.07031357e+00
1.10263482e-01 -1.58397093e-01 7.79132545e-01 3.73681068e-01
-5.88796496e-01 -2.13507097e-03 -6.82294250e-01 7.44037986e-01
1.03330231e+00 -6.79020405e-01 -7.85266101e-01 6.81243181e-01
8.73552501e-01 -5.71856081e-01 -7.34916747e-01 -2.03706697e-01
7.33341277e-01 -5.89950919e-01 1.00227988e+00 -1.23455632e+00
9.25841093e-01 -5.91200516e-02 5.71530126e-02 -1.25701201e+00
-9.00350958e-02 -1.06407452e+00 -2.98595399e-01 1.10479808e+00
7.55656302e-01 -6.47814572e-01 6.25266492e-01 8.30838561e-01
-3.15974772e-01 -8.57854545e-01 -3.65438700e-01 -7.95932353e-01
3.31388235e-01 -7.33005404e-01 7.26377785e-01 9.72243905e-01
4.97536331e-01 5.64911604e-01 -8.97280797e-02 -1.87706843e-01
1.81559831e-01 2.01430380e-01 1.04871559e+00 -1.00444472e+00
-8.43215704e-01 -7.17891932e-01 1.40160874e-01 -1.31945205e+00
1.03286542e-01 -1.40302753e+00 -1.65142938e-01 -1.38791955e+00
4.38908786e-02 -6.51958704e-01 -1.78722292e-01 9.80907023e-01
-7.82667696e-02 -4.69799876e-01 3.16829026e-01 2.06569731e-01
-8.39897931e-01 -8.92360061e-02 9.34573352e-01 -1.48028120e-01
-2.89228857e-01 1.50978714e-01 -8.48234296e-01 1.05794263e+00
6.85459137e-01 -3.30512971e-01 -2.72716612e-01 -9.01957810e-01
6.76292777e-01 3.49307805e-01 -8.55004266e-02 -1.10455573e+00
5.55669129e-01 -2.84075171e-01 2.15513989e-01 -3.56072038e-02
8.46750215e-02 -3.60867709e-01 -5.59471771e-02 4.36054498e-01
-7.29837716e-01 4.48858559e-01 3.77748281e-01 -5.49024902e-02
7.57334828e-02 -4.85038012e-01 3.32882315e-01 -4.89477724e-01
-5.25643766e-01 -5.49034774e-02 -5.04960835e-01 1.41939715e-01
7.54819930e-01 1.13360710e-01 -5.19754410e-01 -3.05394679e-01
-6.47463381e-01 4.67440337e-01 4.03750449e-01 3.72289330e-01
1.59271389e-01 -8.15931678e-01 -1.12090729e-01 7.28982612e-02
-7.44989421e-03 -5.36308549e-02 -4.06010002e-01 8.15103233e-01
-6.26008928e-01 5.47837496e-01 7.14261383e-02 -1.90207690e-01
-9.94022489e-01 4.87558812e-01 2.79293001e-01 -5.82780540e-01
-3.85658145e-01 9.29057002e-01 -1.80520222e-01 -9.75717843e-01
2.04662800e-01 -8.26738358e-01 1.73632249e-01 -2.42997706e-01
6.44331932e-01 1.30438849e-01 9.78356451e-02 3.15051049e-01
-2.46248111e-01 3.68077576e-01 -2.39782363e-01 1.12031780e-01
1.27247810e+00 4.59727436e-01 -2.61544138e-01 6.63281262e-01
8.31986427e-01 1.54297054e-01 -1.00494790e+00 -2.44986147e-01
2.25639567e-01 -6.97025359e-02 -2.92710692e-01 -1.29633284e+00
-4.60232764e-01 1.15323365e+00 -1.21137738e-01 2.73877352e-01
5.95497251e-01 -7.96563476e-02 8.00562739e-01 1.05160296e+00
7.67030954e-01 -9.27690208e-01 1.24719180e-02 7.48843312e-01
4.40501779e-01 -1.08620179e+00 -5.18382192e-02 -1.24060079e-01
-2.72857726e-01 1.51566911e+00 7.57754266e-01 1.42804816e-01
3.67058307e-01 8.10082436e-01 -1.54402241e-01 1.70234680e-01
-1.37530875e+00 1.08695693e-01 -2.34694883e-01 3.10484976e-01
8.65757883e-01 -5.00799972e-04 -6.19848305e-03 8.31061542e-01
-6.07964575e-01 3.88752699e-01 5.16103804e-01 1.17931068e+00
-5.69583297e-01 -1.36505330e+00 -8.14040229e-02 7.22945809e-01
-4.04734313e-01 -5.90751708e-01 -1.52203456e-01 8.55014801e-01
-1.45117762e-02 5.92421830e-01 -1.29811913e-01 -4.22508448e-01
1.85789213e-01 6.41479909e-01 4.81957197e-01 -1.03312755e+00
-9.84175205e-01 -2.76922137e-01 2.37687066e-01 -6.98036611e-01
3.86939585e-01 -4.95070219e-01 -1.21963036e+00 -4.94336158e-01
-9.88397747e-02 1.72897592e-01 7.64770150e-01 1.11099887e+00
4.12796021e-01 4.46889967e-01 9.64975879e-02 -7.06674397e-01
-1.09630287e+00 -8.47124219e-01 -5.72953373e-02 -1.46923274e-01
-1.86933968e-02 -2.52416790e-01 -3.20354640e-01 8.79811198e-02] | [8.426835060119629, 7.589385509490967] |
54885383-184a-4c1e-85a9-07e0b08c7de6 | rate-splitting-multiple-access-for-multi | 2102.08738 | null | https://arxiv.org/abs/2102.08738v1 | https://arxiv.org/pdf/2102.08738v1.pdf | Rate-Splitting Multiple Access for Multi-Antenna Broadcast Channel with Imperfect CSIT and CSIR | Rate-splitting multiple access (RSMA) has appeared as a powerful transmission and multiple access strategy for multi-user multi-antenna communications. Uniquely, this paper studies the optimization of the sum-rate of RSMA with imperfect channel state information (CSI) at the transmitter (CSIT) and the receivers (CSIR). The robustness of the RSMA approach against imperfect CSIT has been investigated in the previous studies while there has been no consideration for the effects of imperfect CSIR. This motivates us to develop a robust design relying on RSMA in the presence of both imperfect CSIT and CSIR. Since the optimization problem for the design of RSMA precoder and power allocations to maximize the sum-rate is non-convex, it is hard to solve directly. To tackle the non-convexity, we propose a novel alternating optimization algorithm based on semidefinite relaxation (SDR) and concave-convex procedure (CCCP) techniques. By comparing simulation results with conventional methods, it turns out that RSMA is quite robust to imperfect CSIR and CSIT, thereby improving the sum-rate performance. | ['Wonjae Shin', 'Bruno Clerckxz', 'Onur Dizdarz', 'Jihye An'] | 2021-02-17 | null | null | null | null | ['robust-design'] | ['miscellaneous'] | [ 2.02019915e-01 9.51122046e-02 -3.95980060e-01 1.38416858e-02
-7.12316036e-01 -3.28383803e-01 1.66729748e-01 -2.53729850e-01
-2.61220306e-01 1.08326566e+00 3.77153531e-02 -7.20525801e-01
-6.08360350e-01 -5.59200048e-01 -2.38567695e-01 -1.03385627e+00
-5.36886930e-01 -1.25261545e-01 -3.73556972e-01 -4.20313239e-01
5.59226945e-02 3.99976522e-01 -7.71674693e-01 -5.77308834e-01
7.96712935e-01 1.14088011e+00 4.62404370e-01 4.14273977e-01
1.94851592e-01 5.33263683e-01 -5.89747190e-01 -2.77166702e-02
3.19855928e-01 -5.57749212e-01 -4.53697681e-01 3.73245746e-01
-7.65747964e-01 -2.97203273e-01 -5.47257841e-01 8.58719826e-01
5.60103655e-01 1.25134941e-02 3.60578477e-01 -1.47085810e+00
-3.04797262e-01 6.18772864e-01 -9.81180251e-01 6.45962507e-02
2.83551008e-01 -7.15408742e-01 7.87462056e-01 -5.89238644e-01
8.84521008e-03 9.49145615e-01 2.56088674e-01 2.18048975e-01
-5.47989011e-01 -5.99717140e-01 -5.38955666e-02 1.35458559e-02
-1.96316254e+00 -6.77172124e-01 4.29774195e-01 2.26325512e-01
2.88290501e-01 6.99583352e-01 2.54035592e-01 2.03862861e-01
8.14011022e-02 6.46930933e-01 9.50282097e-01 -8.05938244e-01
3.33114952e-01 2.84223229e-01 -2.26477191e-01 4.12335485e-01
6.05635166e-01 -6.49422174e-03 -1.65447779e-02 -4.48319465e-01
9.95974541e-01 -2.67568648e-01 -5.36536932e-01 -2.82153159e-01
-1.04468143e+00 6.27029717e-01 2.38450512e-01 4.50617760e-01
-7.44595647e-01 -4.03842255e-02 -1.46740004e-01 5.40162086e-01
1.17120361e-02 8.54637772e-02 5.30398637e-03 1.12183429e-01
-7.49949515e-01 -8.89968872e-02 9.47945774e-01 1.26237404e+00
3.42562646e-01 3.73905629e-01 -1.46339923e-01 7.69428909e-01
5.26636541e-01 8.92015457e-01 -1.47323951e-01 -6.81818068e-01
7.98990369e-01 -1.73244122e-02 5.37286758e-01 -9.29241061e-01
-3.34595740e-01 -1.00476849e+00 -1.29449427e+00 -1.39162078e-01
4.00882751e-01 -7.26075351e-01 -3.10945928e-01 1.56356812e+00
8.26389864e-02 1.75451282e-02 5.77456474e-01 1.17713666e+00
2.58670032e-01 9.38927710e-01 -5.30026197e-01 -1.18703175e+00
8.74677062e-01 -5.14135122e-01 -1.23645365e+00 -1.69155493e-01
6.63245559e-01 -6.60480201e-01 -6.26066187e-03 4.12255496e-01
-1.33106184e+00 1.58474818e-01 -1.34705341e+00 6.42962694e-01
2.27793232e-01 4.19153392e-01 5.39607942e-01 1.21924806e+00
-9.54563558e-01 -7.96043426e-02 -3.48467946e-01 -1.17389910e-01
9.84431207e-02 7.25177526e-01 -1.02646068e-01 -3.94955844e-01
-1.11934829e+00 8.79156709e-01 2.39903927e-01 4.22243446e-01
-3.11982512e-01 -1.73165113e-01 -7.75151074e-01 1.34601176e-01
8.01442385e-01 -3.32534939e-01 1.05387688e+00 -7.42956400e-01
-1.81904089e+00 1.31403103e-01 -2.87789553e-01 -2.12871403e-01
2.29791224e-01 2.59426624e-01 -4.56217259e-01 -1.03794813e-01
-2.16996357e-01 -6.90717697e-01 3.45408827e-01 -1.34284782e+00
-4.00942951e-01 -3.14660579e-01 1.94860816e-01 3.92902970e-01
-8.82004276e-02 3.09152395e-01 -6.01405382e-01 -7.64369488e-01
4.81464148e-01 -8.71078610e-01 -6.84798002e-01 -2.32712895e-01
-5.75495899e-01 2.57520974e-01 4.14962351e-01 -5.16019583e-01
1.62819684e+00 -2.07577538e+00 3.43709022e-01 7.89716065e-01
-2.91933030e-01 3.05904865e-01 2.22511264e-03 6.04100347e-01
1.15171755e-02 -8.40129033e-02 -1.45252839e-01 -1.99038774e-01
-1.92525074e-01 3.20392013e-01 1.62521191e-03 8.39623213e-01
-5.32242775e-01 2.62290508e-01 -6.82074666e-01 -8.28003138e-02
5.32792956e-02 1.38165906e-01 -2.83823133e-01 1.80153191e-01
4.38464195e-01 2.65449107e-01 -9.55269456e-01 6.01495981e-01
1.09685767e+00 -2.20422715e-01 3.82036090e-01 -2.55322084e-02
-6.18563175e-01 -5.24242073e-02 -1.68188214e+00 1.25592279e+00
-7.37724662e-01 2.17433691e-01 8.02479267e-01 -1.21584892e+00
7.61811376e-01 7.71357059e-01 5.07356226e-01 -1.02479601e+00
4.57612365e-01 4.14945453e-01 1.03787310e-01 -1.48014024e-01
4.08145636e-01 -2.65102416e-01 -2.56147921e-01 5.73097706e-01
-4.75209087e-01 3.10569167e-01 7.66940862e-02 4.58079189e-01
9.22071040e-01 -6.11446261e-01 8.90321970e-01 -3.53022486e-01
7.81912625e-01 -6.30159140e-01 5.05031526e-01 9.00842071e-01
1.15736902e-01 2.88094252e-01 1.86310217e-01 2.96319246e-01
-8.98480952e-01 -8.56991947e-01 -1.73769459e-01 7.25380421e-01
6.26689672e-01 -6.70431852e-02 -1.82959437e-01 1.54052496e-01
-1.52704880e-01 6.89038694e-01 -3.25511366e-01 1.22799769e-01
-6.31949529e-02 -1.10014379e+00 3.16769958e-01 -1.71666846e-01
7.07239211e-01 -1.10778762e-02 4.82321307e-02 6.30157113e-01
-2.40814254e-01 -1.31634688e+00 -2.19331309e-01 7.86327869e-02
-2.80117929e-01 -5.76887131e-01 -1.23037636e+00 -2.67787188e-01
6.64788306e-01 7.73169100e-01 3.02774578e-01 8.38957950e-02
-6.20745355e-03 6.40421629e-01 -5.71571290e-01 -2.70209432e-01
-3.32268357e-01 -1.95951670e-01 2.70397335e-01 3.92687708e-01
-6.36543930e-01 -3.82831246e-01 -2.66931683e-01 6.47325933e-01
-6.54179394e-01 -1.67125054e-02 7.93731332e-01 2.99202293e-01
3.39988232e-01 3.19045454e-01 1.14942157e+00 -3.43952209e-01
6.22571886e-01 -8.60151470e-01 -7.01262116e-01 5.41191220e-01
-1.51254028e-01 8.59870762e-02 6.35623574e-01 1.00348644e-01
-1.20844209e+00 -1.11413486e-01 1.66179035e-02 1.98207453e-01
4.40620840e-01 6.98239744e-01 -4.77515757e-01 -5.36247015e-01
2.26816818e-01 4.64323342e-01 3.32533121e-02 -2.05458984e-01
6.37818649e-02 1.28806686e+00 1.26356468e-01 -3.13734204e-01
9.10977125e-01 4.40656066e-01 4.45271283e-01 -1.36348510e+00
-1.05355453e+00 -7.89122581e-01 -1.58673987e-01 -6.75590942e-03
1.98470056e-01 -1.16434443e+00 -6.79402828e-01 4.37121272e-01
-9.28132296e-01 -7.19607174e-02 5.30174494e-01 8.40459347e-01
-4.67023045e-01 8.22719932e-01 -2.03078650e-02 -1.41323280e+00
-3.59006286e-01 -9.35943305e-01 1.29168347e-01 2.28791192e-01
4.67870563e-01 -8.49178493e-01 -4.09449220e-01 1.79173455e-01
8.11687648e-01 1.73886895e-01 2.43092209e-01 -4.91724275e-02
-6.83887303e-01 -3.35843891e-01 -3.24169666e-01 6.56797960e-02
3.57211977e-01 -6.49652123e-01 -3.88732433e-01 -5.93123078e-01
-3.94033827e-02 -1.85967702e-02 1.75233588e-01 5.70316494e-01
9.07248557e-01 -6.82014465e-01 -4.38741118e-01 5.29479623e-01
1.62414038e+00 1.57836780e-01 1.01185596e+00 3.25924367e-01
1.03215836e-01 -1.15677908e-01 8.21437478e-01 1.28748298e+00
5.04948318e-01 9.99501348e-01 4.88526314e-01 -7.52433166e-02
6.10222042e-01 6.44346118e-01 1.93890810e-01 6.92245543e-01
-3.14936012e-01 -9.37869191e-01 -2.47015968e-01 3.12600642e-01
-1.77357852e+00 -8.13753664e-01 -3.92154247e-01 2.66363168e+00
6.30508959e-01 -2.68745184e-01 3.82261574e-02 4.69388276e-01
7.47937143e-01 3.48212533e-02 -1.95113704e-01 -4.57069874e-01
-3.71163458e-01 -1.89982936e-01 1.22305107e+00 3.97092491e-01
-8.56919944e-01 3.01849902e-01 5.76139164e+00 8.77914548e-01
-7.85492063e-01 -9.51745138e-02 2.46395990e-01 1.35152206e-01
-3.57265830e-01 4.51796204e-02 -4.25758958e-01 3.56478214e-01
7.17696726e-01 -4.95597571e-01 8.13104570e-01 2.41279617e-01
7.64551997e-01 -6.19053602e-01 -3.74474436e-01 1.13647485e+00
-2.09804974e-03 -1.09641492e+00 -2.53474325e-01 3.31358254e-01
7.17409372e-01 -2.77254969e-01 -2.36748755e-01 6.51712567e-02
1.49859160e-01 -7.52024531e-01 3.55462730e-01 4.58965391e-01
6.91456616e-01 -8.46467495e-01 7.52476513e-01 4.71537828e-01
-1.22717679e+00 -4.49187696e-01 -2.56394506e-01 -3.37701857e-01
5.88970840e-01 7.48116910e-01 -5.07722437e-01 1.22522032e+00
1.13879226e-01 2.48629734e-01 1.67063653e-01 1.60838854e+00
6.31433260e-03 4.91685241e-01 -6.44298613e-01 -1.20411232e-01
3.38913411e-01 -3.09661925e-01 6.96551144e-01 8.18030357e-01
7.36142337e-01 8.35245728e-01 2.29972586e-01 2.97726631e-01
1.71756849e-01 4.11829680e-01 -2.80701905e-01 1.43789619e-01
7.71146119e-01 9.67664540e-01 -2.42418259e-01 7.48900976e-03
-6.76622927e-01 7.68145382e-01 -1.96451277e-01 6.84103787e-01
-5.40280461e-01 -4.72250074e-01 5.40423155e-01 1.04708383e-02
2.57773608e-01 -7.03384399e-01 -4.33268458e-01 -7.88162529e-01
5.69130369e-02 -3.57698083e-01 3.84455115e-01 -4.53425378e-01
-6.56608641e-01 1.29337087e-01 -1.57408193e-01 -1.36443663e+00
-4.66066562e-02 -7.86663815e-02 -6.47728980e-01 9.98266816e-01
-1.86420405e+00 -6.49600029e-01 3.93326841e-02 6.37474358e-01
-1.16727576e-01 -9.97543707e-02 6.00103199e-01 3.96623403e-01
-7.64898002e-01 9.78981614e-01 7.51204610e-01 -6.70778425e-03
1.08133517e-01 -7.42073655e-01 -7.34252930e-01 8.30912232e-01
-6.38650000e-01 4.50405687e-01 9.26502168e-01 -1.20451972e-01
-1.85854053e+00 -7.75587618e-01 5.45212984e-01 6.49907768e-01
5.67054391e-01 5.52266315e-02 -5.17106056e-01 2.61592418e-01
1.89022273e-01 -1.36796668e-01 9.39702392e-01 -3.15013021e-01
2.58141279e-01 3.36278863e-02 -1.00544906e+00 4.95150268e-01
6.17921412e-01 4.77740318e-02 5.38595244e-02 1.29107028e-01
1.49397954e-01 -3.63906920e-01 -9.22743022e-01 1.85123444e-01
4.00237828e-01 -6.02120340e-01 9.19679105e-01 1.74763665e-01
-4.69374180e-01 -4.73136574e-01 -4.36605841e-01 -1.26524127e+00
-3.02191675e-01 -1.16288471e+00 1.24480873e-01 9.62723434e-01
4.87402707e-01 -4.27932531e-01 2.89587170e-01 6.18706644e-01
-2.29869615e-02 -4.65194851e-01 -1.28547192e+00 -9.61872876e-01
-2.99974889e-01 -2.00744465e-01 2.59289473e-01 7.38036692e-01
4.43488806e-01 1.93593934e-01 -9.83889163e-01 7.79125929e-01
7.62313128e-01 -9.51615199e-02 7.42104590e-01 -8.32505643e-01
-4.01180506e-01 1.02281780e-03 -1.26126468e-01 -1.31733418e+00
-2.49448761e-01 -6.24985039e-01 -1.13069348e-01 -1.75401509e+00
3.84167349e-03 -1.03098679e+00 -2.65438825e-01 3.96813184e-01
1.50320604e-01 -1.81462303e-01 1.90427005e-01 3.11396904e-02
-8.72016847e-01 7.51831174e-01 1.39213121e+00 2.84739286e-01
-4.67992157e-01 8.19034696e-01 -8.41061413e-01 2.02681601e-01
8.36388230e-01 -1.94646657e-01 -3.20312440e-01 -2.11109012e-01
3.96516442e-01 9.83842850e-01 -9.30315927e-02 -8.77001107e-01
1.26479566e-01 -4.58456606e-01 -3.74923766e-01 -5.38953543e-01
5.00212669e-01 -1.05094373e+00 2.28568614e-02 2.12227255e-01
-3.58150378e-02 -6.13663554e-01 8.03274512e-02 7.43085444e-01
7.21606687e-02 -4.01480645e-01 7.30388165e-01 2.01976508e-01
-4.43064183e-01 1.84710950e-01 -7.64680505e-01 -2.52535701e-01
1.12312710e+00 -2.27765203e-01 -4.63970331e-03 -1.02000439e+00
-6.36613131e-01 5.51553547e-01 -1.24383353e-01 1.05584793e-01
4.85998452e-01 -1.16884112e+00 -6.90179884e-01 -1.00157596e-01
-1.62005454e-01 -4.42893386e-01 3.44347864e-01 1.06296480e+00
9.97087639e-03 7.00709403e-01 1.11561559e-01 -1.38248608e-01
-1.08260059e+00 9.22370031e-02 3.85766357e-01 -6.40757680e-02
1.20445356e-01 3.86107326e-01 -3.36871743e-01 1.37630254e-01
2.17749700e-01 1.42922059e-01 -2.90652305e-01 -2.98243523e-01
6.23836040e-01 4.82407451e-01 -9.64110643e-02 -8.36012840e-01
-2.67890841e-01 3.56448829e-01 1.27336800e-01 -8.14503953e-02
1.04280162e+00 -9.58434343e-01 -1.74851850e-01 -1.81256458e-01
1.20385003e+00 2.63778001e-01 -6.13215983e-01 -6.72290921e-01
-2.66871691e-01 -5.55669785e-01 5.59051991e-01 -5.67601562e-01
-1.01436675e+00 2.16433421e-01 3.63048464e-01 2.97406942e-01
1.05072117e+00 -2.45562837e-01 5.71651757e-01 5.90849638e-01
9.83575404e-01 -1.00767708e+00 -1.97664663e-01 6.00114405e-01
8.19614708e-01 -1.07714438e+00 2.86532164e-01 -7.96815813e-01
-4.32498991e-01 1.07815421e+00 -1.78748909e-02 3.36028308e-01
9.17740822e-01 3.46852124e-01 -2.55370170e-01 2.55517989e-01
-3.56806397e-01 -6.20515108e-01 1.01024665e-01 3.00228536e-01
3.22819054e-01 3.05518150e-01 -7.32061625e-01 7.63272047e-01
1.85844630e-01 9.62259397e-02 1.07796192e+00 1.15044010e+00
-5.67142725e-01 -1.12912619e+00 -7.36212194e-01 3.64646882e-01
-7.14844108e-01 1.70409884e-02 2.33383775e-01 5.22152245e-01
-4.97683436e-01 1.52075839e+00 -3.60938489e-01 2.34602951e-02
8.98960605e-02 -7.64674366e-01 3.76195878e-01 -2.77972251e-01
3.92297119e-01 9.01103616e-02 4.81995225e-01 -1.10495768e-01
-5.91054976e-01 -4.79609787e-01 -1.36564302e+00 -1.60071522e-01
-8.95295739e-01 5.64834118e-01 7.48981595e-01 1.41795206e+00
2.22035870e-01 3.27133447e-01 1.38246787e+00 -3.23098093e-01
-6.07788920e-01 -4.77072716e-01 -1.01921797e+00 -4.03694570e-01
3.65633935e-01 -5.21036685e-01 -4.79538023e-01 -5.69923878e-01] | [6.127575397491455, 1.4756481647491455] |
538331c5-4cb1-4ed1-97ef-d3106a79d405 | public-wisdom-matters-discourse-aware | 2209.13017 | null | https://arxiv.org/abs/2209.13017v2 | https://arxiv.org/pdf/2209.13017v2.pdf | Public Wisdom Matters! Discourse-Aware Hyperbolic Fourier Co-Attention for Social-Text Classification | Social media has become the fulcrum of all forms of communication. Classifying social texts such as fake news, rumour, sarcasm, etc. has gained significant attention. The surface-level signals expressed by a social-text itself may not be adequate for such tasks; therefore, recent methods attempted to incorporate other intrinsic signals such as user behavior and the underlying graph structure. Oftentimes, the `public wisdom' expressed through the comments/replies to a social-text acts as a surrogate of crowd-sourced view and may provide us with complementary signals. State-of-the-art methods on social-text classification tend to ignore such a rich hierarchical signal. Here, we propose Hyphen, a discourse-aware hyperbolic spectral co-attention network. Hyphen is a fusion of hyperbolic graph representation learning with a novel Fourier co-attention mechanism in an attempt to generalise the social-text classification tasks by incorporating public discourse. We parse public discourse as an Abstract Meaning Representation (AMR) graph and use the powerful hyperbolic geometric representation to model graphs with hierarchical structure. Finally, we equip it with a novel Fourier co-attention mechanism to capture the correlation between the source post and public discourse. Extensive experiments on four different social-text classification tasks, namely detecting fake news, hate speech, rumour, and sarcasm, show that Hyphen generalises well, and achieves state-of-the-art results on ten benchmark datasets. We also employ a sentence-level fact-checked and annotated dataset to evaluate how Hyphen is capable of producing explanations as analogous evidence to the final prediction. | ['Tanmoy Chakraborty', 'Md. Shad Akhtar', 'S. M. Phaneendra Angara', 'Karish Grover'] | 2022-09-15 | null | null | null | null | ['rumour-detection'] | ['natural-language-processing'] | [ 1.25058249e-01 7.76423812e-01 -1.31051198e-01 -1.62889659e-01
-4.84205276e-01 -2.47312844e-01 1.11122143e+00 5.71322024e-01
1.93408340e-01 3.13198835e-01 8.49009693e-01 -3.76918465e-01
3.47319305e-01 -7.80673444e-01 -7.28996813e-01 -5.59958875e-01
-2.69304756e-02 2.33895034e-01 3.29268813e-01 -8.33311141e-01
4.03986633e-01 -1.67661324e-01 -1.06631088e+00 9.41729426e-01
8.16486120e-01 7.95885324e-01 -2.99951017e-01 7.17113495e-01
-4.45935071e-01 1.48596084e+00 -1.01510751e+00 -8.19918752e-01
-4.22382087e-01 -7.64452100e-01 -9.74001467e-01 3.23294669e-01
2.13796467e-01 1.69693485e-01 -5.75221896e-01 1.17528164e+00
2.02537075e-01 -3.02898049e-01 7.10155547e-01 -1.30841911e+00
-1.12396801e+00 1.06946743e+00 -7.50757217e-01 2.26231799e-01
6.28154278e-01 -6.89762235e-02 1.30100465e+00 -7.47981071e-01
6.43274426e-01 1.56357336e+00 1.03172576e+00 3.52498233e-01
-1.16829228e+00 -3.25077713e-01 -4.17196080e-02 2.40975082e-01
-9.71560240e-01 -7.19077373e-03 1.40583420e+00 -5.82703471e-01
6.39795482e-01 4.86590236e-01 8.03577662e-01 1.79071152e+00
3.23349297e-01 1.06713569e+00 1.00974917e+00 -1.72171310e-01
-1.61352530e-01 1.37759387e-01 3.08832824e-01 9.88375247e-01
-6.07049502e-02 -3.40831965e-01 -8.27043414e-01 -5.93111932e-01
7.34964758e-02 -1.68579891e-01 -7.61686623e-01 1.97781801e-01
-1.25801754e+00 1.37536561e+00 9.51685965e-01 5.83623528e-01
-1.26766324e-01 1.37624264e-01 5.34535885e-01 5.88319302e-01
1.27944529e+00 6.20762527e-01 3.39623131e-02 2.87401736e-01
-6.64334536e-01 1.91459745e-01 1.07760811e+00 4.57710683e-01
5.30587494e-01 -3.70276384e-02 -2.38608822e-01 7.66065419e-01
3.79493952e-01 4.42174137e-01 6.30489469e-01 -3.52385551e-01
5.16445756e-01 8.68786395e-01 -3.52750599e-01 -1.88778710e+00
-6.54969275e-01 -4.76977348e-01 -9.39913869e-01 -4.15972084e-01
4.52883869e-01 -2.07213908e-02 -2.50759989e-01 1.36893678e+00
2.40765065e-01 3.83139253e-01 -7.73228109e-02 9.20718312e-01
1.20407403e+00 7.66551673e-01 -5.12764275e-01 -2.61740297e-01
1.40547752e+00 -1.02635956e+00 -9.97425199e-01 -1.25397608e-01
9.71841812e-01 -7.29290545e-01 1.11085558e+00 1.71979278e-01
-7.07051456e-01 -1.51608258e-01 -1.09015155e+00 -2.24621475e-01
-5.74937403e-01 -3.25863928e-01 3.58369887e-01 5.76577187e-01
-7.03641057e-01 8.52878690e-01 -3.22943985e-01 -1.32578373e-01
5.00618875e-01 -3.09980899e-01 -1.42858744e-01 1.68254524e-01
-1.44322014e+00 6.24563754e-01 3.07340571e-03 5.20595722e-02
-1.40591413e-01 -4.33584005e-01 -7.88759410e-01 -2.42297173e-01
4.03658360e-01 -5.67122757e-01 8.94784272e-01 -1.31569386e+00
-1.50519848e+00 1.06630325e+00 2.34587595e-01 -5.92620730e-01
7.47898042e-01 2.78428495e-02 -5.83964169e-01 3.12758714e-01
1.40070558e-01 -4.22876440e-02 1.41341698e+00 -1.17586958e+00
1.31223917e-01 -5.02898991e-01 2.67709550e-02 -1.84181824e-01
-4.31854904e-01 1.05076343e-01 3.72041851e-01 -8.91068995e-01
2.22826108e-01 -9.97948229e-01 3.61029655e-01 -3.45811397e-01
-9.79455113e-01 -3.53110075e-01 1.13474977e+00 -6.98688745e-01
1.25656891e+00 -2.17498565e+00 2.58769363e-01 1.36967264e-02
8.82274389e-01 -3.51899973e-04 5.08618094e-02 7.45977581e-01
5.04763387e-02 5.04140913e-01 -1.74289182e-01 -6.51794255e-01
-3.24376039e-02 1.32291272e-01 -6.13584757e-01 9.48593616e-01
3.08241338e-01 1.05027139e+00 -1.21514249e+00 -3.85713160e-01
-2.78661847e-01 6.24378979e-01 -5.19770503e-01 8.65033269e-02
-3.12511653e-01 5.69885612e-01 -4.60590154e-01 1.61735326e-01
3.47767234e-01 -8.56774449e-01 6.13623597e-02 -1.86108559e-01
2.77825862e-01 5.18180668e-01 -4.05021012e-01 1.22231221e+00
-1.37414858e-01 1.13084829e+00 5.84393665e-02 -1.08891082e+00
9.01028872e-01 2.92096227e-01 1.21892028e-01 -2.97199965e-01
6.23020828e-01 2.25672975e-01 -6.71801716e-02 -6.23997509e-01
5.52317321e-01 -1.78597286e-01 -1.92884251e-01 5.55058599e-01
-8.74009058e-02 -2.90676475e-01 -3.94417197e-01 5.24860382e-01
1.14764309e+00 -4.00049508e-01 3.12714487e-01 -3.18736374e-01
5.39929450e-01 -9.95777324e-02 -1.28092468e-01 6.54355526e-01
3.77591816e-03 8.99197519e-01 1.18568158e+00 -4.54996586e-01
-7.58189559e-01 -5.12948036e-01 1.27664000e-01 1.25477147e+00
7.29412884e-02 -6.70008540e-01 -6.25409424e-01 -9.49427664e-01
2.94535123e-02 7.67731011e-01 -9.95692909e-01 -1.91365197e-01
-4.56999689e-01 -8.63146663e-01 7.18506157e-01 1.07341344e-02
3.42424512e-01 -8.02683651e-01 -1.99213043e-01 1.41006306e-01
-6.18159890e-01 -1.31059480e+00 -5.42557061e-01 -2.56937563e-01
-4.23661113e-01 -1.35252488e+00 -4.79276121e-01 -4.24132735e-01
3.07641655e-01 6.24080896e-01 1.12682915e+00 6.43835366e-01
1.42069221e-01 2.56148964e-01 -8.25069964e-01 -3.31253558e-01
-9.98254120e-01 -4.20801900e-02 -2.82662004e-01 6.95948005e-01
-4.67209518e-02 -4.52525556e-01 -3.41415077e-01 1.38022065e-01
-8.78635049e-01 2.99562693e-01 -7.18418807e-02 1.11831689e+00
-2.75740296e-01 -3.33347321e-01 6.36225641e-01 -1.34180307e+00
1.06761777e+00 -8.82578671e-01 1.87524512e-01 -2.35676959e-01
-1.71742693e-01 -1.63315255e-02 8.12284708e-01 -6.49506092e-01
-5.68206668e-01 -8.18305850e-01 6.41103610e-02 -4.89178717e-01
3.15730691e-01 8.22542906e-01 3.47476959e-01 1.40412882e-01
1.13655889e+00 4.73253578e-02 3.51514488e-01 -1.63464800e-01
4.54899639e-01 8.91254246e-01 1.77898541e-01 -1.78287476e-01
8.09318602e-01 6.84879959e-01 -6.55854195e-02 -1.32098031e+00
-1.40405762e+00 -5.22529006e-01 -1.40103787e-01 -2.39550933e-01
9.41370487e-01 -5.94975233e-01 -6.91476107e-01 3.88793409e-01
-1.73632526e+00 -1.65777504e-01 -9.01724547e-02 7.98997730e-02
-4.66255337e-01 7.19255269e-01 -8.52122307e-01 -8.26298177e-01
-1.79430962e-01 -7.86398411e-01 1.29187775e+00 -5.68453908e-01
-2.72731394e-01 -1.28010368e+00 -6.67666793e-02 7.47930706e-01
3.32544208e-01 8.05348992e-01 7.38449991e-01 -1.15374792e+00
-1.06540047e-01 -1.69019058e-01 -3.14587265e-01 4.79165018e-02
-1.27795711e-01 -1.34102762e-01 -1.20157051e+00 -1.07737221e-01
3.53599310e-01 -5.09868264e-01 9.77754533e-01 -1.32234678e-01
1.04638374e+00 -9.38373864e-01 -1.28562361e-01 2.43110895e-01
8.55778039e-01 -7.41974831e-01 4.66838241e-01 -7.84251466e-02
1.03818905e+00 8.94388676e-01 -5.88366613e-02 4.49815989e-01
4.94193763e-01 5.07105947e-01 5.92051983e-01 3.85654010e-02
-1.75925419e-01 -3.53677601e-01 6.30436599e-01 1.41560602e+00
-8.59660208e-02 -6.04273677e-01 -9.85785902e-01 1.88021928e-01
-2.03446198e+00 -1.23413634e+00 -9.88960683e-01 1.43771386e+00
4.98466432e-01 1.47868156e-01 2.45520264e-01 5.12651384e-01
7.65743911e-01 8.78811240e-01 -8.44331272e-03 -2.38170668e-01
-4.05771405e-01 -4.11602885e-01 2.41220504e-01 6.45142555e-01
-9.24005330e-01 6.14955604e-01 4.99840117e+00 7.70435989e-01
-1.22573459e+00 4.63450789e-01 5.20439029e-01 3.74436140e-01
-4.93902236e-01 -3.75545859e-01 -1.07359193e-01 5.28933823e-01
9.65944231e-01 -1.05263531e-01 4.41024125e-01 5.79696119e-01
1.12036906e-01 4.78752464e-01 -9.67777848e-01 8.23978841e-01
6.48226917e-01 -1.61697197e+00 -7.90804625e-02 1.49219483e-01
5.72050869e-01 2.56065369e-01 1.26630217e-02 2.84626275e-01
8.77455622e-02 -9.82461751e-01 9.19333518e-01 2.18513027e-01
3.95938516e-01 -2.22222403e-01 9.91405010e-01 7.03881562e-01
-8.51079345e-01 -2.24490259e-02 -1.04139920e-03 -2.15701044e-01
1.18663773e-01 5.27978420e-01 -1.01155376e+00 5.68376541e-01
3.03685039e-01 1.08941782e+00 -7.63240993e-01 1.22092217e-01
-3.61248940e-01 8.60116780e-01 -1.04510151e-01 -5.06489277e-01
3.71069074e-01 1.11545190e-01 1.02465630e+00 1.39824462e+00
-2.78982650e-02 1.77065209e-02 3.11075509e-01 1.01020586e+00
-2.19158471e-01 3.11744273e-01 -8.71890306e-01 -4.12894785e-01
-1.64126262e-01 9.24907744e-01 -7.89643943e-01 -2.46395648e-01
-5.61151922e-01 1.03973770e+00 5.23979902e-01 2.60308802e-01
-9.87807393e-01 5.08350600e-03 5.01641706e-02 5.53341925e-01
-7.20480904e-02 -4.10082079e-02 -3.02460101e-02 -1.50449395e+00
-1.82795480e-01 -9.49388444e-01 1.97674155e-01 -8.47027361e-01
-1.88498354e+00 6.98323429e-01 -3.42818171e-01 -1.02394652e+00
-7.03010187e-02 -6.29709661e-01 -7.43555009e-01 3.70794147e-01
-1.39167726e+00 -1.40395498e+00 -5.28010905e-01 5.67656517e-01
5.40012598e-01 -3.01774200e-02 6.57529593e-01 -1.38089344e-01
-2.21832156e-01 3.97166163e-01 -2.65878737e-01 3.48173678e-01
5.27340531e-01 -1.50550067e+00 2.17256039e-01 3.51287812e-01
2.54628122e-01 2.51127005e-01 1.08654940e+00 -6.32571757e-01
-1.41367781e+00 -9.98205364e-01 8.90743256e-01 -6.83440864e-01
1.51109183e+00 -4.83346254e-01 -1.31925261e+00 5.64640462e-01
3.24859589e-01 1.79063544e-01 6.35282636e-01 1.75046355e-01
-8.63588572e-01 4.57804620e-01 -8.98385644e-01 5.52745521e-01
1.12984383e+00 -8.60462964e-01 -8.38420570e-01 9.35830534e-01
1.11491847e+00 -3.43703717e-01 -5.44048309e-01 -4.60311305e-03
6.93929940e-02 -1.28113580e+00 6.26767576e-01 -6.86181307e-01
9.87373233e-01 9.68425646e-02 -8.91577750e-02 -1.55305934e+00
-1.18504860e-01 -1.08903611e+00 -2.29806915e-01 9.15956855e-01
3.27442199e-01 -9.04339731e-01 6.29228592e-01 -2.37306610e-01
-2.68715501e-01 -6.59674525e-01 -8.79314184e-01 -4.75663304e-01
7.72240683e-02 -4.93096381e-01 3.87877494e-01 1.58443499e+00
6.18950844e-01 1.05635452e+00 -5.67273915e-01 -2.21154876e-02
5.04029632e-01 -5.07091582e-02 7.98455536e-01 -1.38671458e+00
-4.46080774e-01 -7.75955558e-01 -5.55585921e-01 -9.27713215e-01
6.15260839e-01 -1.35407245e+00 -2.18993485e-01 -1.24120045e+00
1.93268042e-02 2.17248768e-01 3.22498411e-01 -1.01634420e-01
-9.11515802e-02 3.23066354e-01 3.03041577e-01 3.78603131e-01
-5.09961545e-01 7.98152626e-01 1.53670371e+00 -4.59223270e-01
1.63678005e-02 -1.41429365e-01 -6.38906419e-01 1.03114486e+00
5.11068881e-01 -4.23060894e-01 -1.89467832e-01 2.16619801e-02
7.74587989e-01 3.69093925e-01 6.44434273e-01 -5.03037155e-01
1.17019422e-01 1.79041833e-01 -2.66225696e-01 -2.49659464e-01
3.59275550e-01 -4.99074399e-01 -2.59653300e-01 4.20189172e-01
-4.86241639e-01 -2.11296789e-02 -4.31108624e-01 1.22232306e+00
-2.99978375e-01 7.30565563e-02 7.18697131e-01 -4.33051102e-02
2.91150868e-01 -1.06751509e-01 -3.71005476e-01 4.35701340e-01
5.04149914e-01 -5.85885309e-02 -9.79783952e-01 -8.18092406e-01
-6.97958589e-01 -9.30139329e-03 2.80037791e-01 5.37016153e-01
4.90971625e-01 -1.16668010e+00 -1.15248954e+00 -8.24744701e-02
2.16428190e-01 -4.26372796e-01 2.15566438e-02 1.22434294e+00
-3.12571794e-01 -4.33585644e-02 4.93337721e-01 -7.64180660e-01
-1.06178522e+00 6.54347003e-01 2.70011246e-01 -3.45075816e-01
-9.75786567e-01 5.66647351e-01 8.68510827e-02 -3.86791259e-01
-2.15877786e-01 -4.75985229e-01 -4.63351816e-01 4.43662673e-01
5.61390638e-01 3.71017039e-01 -6.23194873e-02 -1.14792895e+00
1.07284645e-02 1.93475917e-01 3.26883048e-01 1.94133267e-01
1.21106732e+00 -8.89901966e-02 -2.98637837e-01 9.84689653e-01
1.49472463e+00 3.22687835e-01 -6.49914324e-01 -3.49372715e-01
7.29980022e-02 -3.68534774e-01 8.43902305e-02 -2.52856255e-01
-7.07965434e-01 9.07306254e-01 -2.36420706e-01 1.47573853e+00
3.60137075e-01 3.28033298e-01 7.18133032e-01 1.50848955e-01
1.71575159e-01 -7.38414824e-01 9.34152901e-01 6.59835696e-01
1.54247797e+00 -1.39971960e+00 -6.62065670e-02 -9.47836459e-01
-7.34679818e-01 1.28929496e+00 5.47283515e-02 -3.25262576e-01
8.05215478e-01 -2.67252505e-01 -1.14307873e-01 -7.63952613e-01
-6.43850565e-01 4.75836061e-02 5.28102398e-01 2.23363578e-01
5.51797330e-01 2.35346735e-01 -2.51448333e-01 6.37502372e-01
-6.25388145e-01 -6.36743903e-01 1.05666685e+00 3.62515092e-01
-3.73289794e-01 -2.90222675e-01 -4.60600972e-01 3.65412414e-01
-7.31924117e-01 1.64039584e-03 -9.31907892e-01 6.70977473e-01
-2.09103033e-01 1.26230311e+00 -1.00651011e-01 -5.89617312e-01
-1.55420862e-02 1.17109768e-01 2.30683818e-01 -6.63166702e-01
-9.97994244e-01 3.47540826e-02 4.67876703e-01 -5.44460833e-01
-6.76569581e-01 -3.03957641e-01 -1.13057327e+00 -6.57310903e-01
-6.08156800e-01 1.90914348e-01 5.28987408e-01 1.01333284e+00
2.78187752e-01 6.31385505e-01 6.33718312e-01 -8.37709963e-01
-6.75215662e-01 -1.29676795e+00 -4.36190456e-01 9.29338872e-01
7.86375523e-01 -5.52638769e-01 -1.01506126e+00 -2.29873657e-01] | [8.188837051391602, 10.277271270751953] |
dd6b1dda-eee3-4210-b4b6-3a806f0f5f6d | data-driven-meta-set-based-fine-grained | 2008.02438 | null | https://arxiv.org/abs/2008.02438v1 | https://arxiv.org/pdf/2008.02438v1.pdf | Data-driven Meta-set Based Fine-Grained Visual Classification | Constructing fine-grained image datasets typically requires domain-specific expert knowledge, which is not always available for crowd-sourcing platform annotators. Accordingly, learning directly from web images becomes an alternative method for fine-grained visual recognition. However, label noise in the web training set can severely degrade the model performance. To this end, we propose a data-driven meta-set based approach to deal with noisy web images for fine-grained recognition. Specifically, guided by a small amount of clean meta-set, we train a selection net in a meta-learning manner to distinguish in- and out-of-distribution noisy images. To further boost the robustness of model, we also learn a labeling net to correct the labels of in-distribution noisy data. In this way, our proposed method can alleviate the harmful effects caused by out-of-distribution noise and properly exploit the in-distribution noisy samples for training. Extensive experiments on three commonly used fine-grained datasets demonstrate that our approach is much superior to state-of-the-art noise-robust methods. | ['Zechao Li', 'Qi Wu', 'Yazhou Yao', 'Chuanyi Zhang', 'Zhenmin Tang', 'Xiangbo Shu'] | 2020-08-06 | null | null | null | null | ['fine-grained-visual-recognition'] | ['computer-vision'] | [ 1.69523172e-02 -4.76086617e-01 -6.76905811e-02 -5.35465658e-01
-1.32515967e+00 -8.13694894e-01 3.51254284e-01 -3.99722531e-02
-3.26128811e-01 6.45024776e-01 1.37024477e-01 2.39552483e-01
-3.32949385e-02 -6.95784569e-01 -9.49554682e-01 -8.91382098e-01
7.35453188e-01 2.27977440e-01 2.20538914e-01 -4.68955189e-03
1.86880097e-01 2.48468906e-01 -1.72446609e+00 5.91735303e-01
9.92396116e-01 1.35635030e+00 3.66363078e-01 3.20450097e-01
-5.01253784e-01 7.27337003e-01 -7.99418986e-01 -3.95918638e-01
3.27912480e-01 -2.02692524e-01 -4.62533355e-01 7.27375686e-01
7.53347754e-01 -2.21119121e-01 -1.93765566e-01 1.56778085e+00
4.98077840e-01 2.08822414e-01 4.45230693e-01 -9.70782757e-01
-8.11133146e-01 3.27857584e-01 -6.05160296e-01 2.20444486e-01
-3.60452794e-02 4.65810090e-01 6.56798244e-01 -1.05920768e+00
3.46498907e-01 1.22670877e+00 5.47508776e-01 6.29627645e-01
-9.80152011e-01 -6.41864002e-01 5.30650020e-01 -2.19308697e-02
-1.50566733e+00 -4.46043879e-01 9.31012034e-01 -2.93572903e-01
9.74072292e-02 2.75309592e-01 2.59868681e-01 1.39439535e+00
-4.42774862e-01 6.05647027e-01 1.68722129e+00 -3.07665050e-01
4.45951879e-01 1.55094519e-01 1.91227987e-01 6.26726568e-01
3.01761985e-01 -2.76514366e-02 -5.57452083e-01 -2.39622727e-01
6.74514055e-01 1.20454885e-01 -2.03692257e-01 -2.81956494e-01
-9.88816857e-01 4.82470989e-01 4.04375821e-01 1.91534832e-01
-3.79174501e-01 8.82291347e-02 3.27074260e-01 2.31586091e-04
7.70710170e-01 4.61840704e-02 -4.94649589e-01 -7.25123063e-02
-8.24636877e-01 1.67571694e-01 4.73841488e-01 7.24337220e-01
8.98054540e-01 -6.19649664e-02 -3.95232290e-01 1.21690476e+00
2.88816035e-01 5.41925550e-01 5.33796906e-01 -9.25540864e-01
7.56710410e-01 6.33187473e-01 3.41347218e-01 -9.35578227e-01
1.49116859e-01 -5.96080363e-01 -1.07607543e+00 2.12037802e-01
6.86366379e-01 2.95372959e-02 -1.22122657e+00 1.55524731e+00
5.16297579e-01 1.35126948e-01 -2.71353155e-01 1.26817286e+00
7.06536829e-01 3.97874862e-01 2.57723719e-01 -5.20504415e-02
1.25970328e+00 -1.14562142e+00 -4.28032368e-01 -2.62080252e-01
5.06054461e-02 -6.62528813e-01 1.41737330e+00 4.05912280e-01
-6.65743589e-01 -6.82625949e-01 -7.37732172e-01 1.50752723e-01
-4.45794940e-01 1.89496189e-01 1.20882444e-01 7.44500816e-01
-5.73935568e-01 3.24330151e-01 -3.63359660e-01 -9.10124555e-02
8.75352919e-01 -2.06107408e-01 -4.31921482e-01 -5.86075664e-01
-7.89815545e-01 4.08700556e-01 4.13914174e-01 2.19189927e-01
-1.18225110e+00 -6.31652415e-01 -5.32792449e-01 -2.06073038e-02
7.40740240e-01 -6.11010969e-01 1.16108012e+00 -1.05570173e+00
-1.11475527e+00 8.21596742e-01 -1.27160609e-01 2.79178191e-02
6.49147153e-01 9.69588384e-02 -3.37598622e-01 -7.83613697e-02
3.18989396e-01 2.85314143e-01 1.23358572e+00 -1.78675771e+00
-7.46700287e-01 -6.41826391e-01 1.13713332e-01 6.33787364e-02
-4.17334139e-01 -2.36106008e-01 -7.83374965e-01 -9.99203205e-01
-8.50086194e-03 -7.21672118e-01 -2.80949712e-01 1.00229070e-01
-4.37076122e-01 -2.04484075e-01 6.81398153e-01 -4.86277342e-01
1.01982880e+00 -2.23580790e+00 -3.32970321e-01 2.64846981e-01
3.06623280e-01 3.84453475e-01 -4.27571505e-01 1.23220257e-01
2.84411460e-01 3.02764773e-01 -3.70886214e-02 -3.81778866e-01
1.24971390e-01 3.78236949e-01 -1.88068926e-01 2.63778031e-01
2.08826393e-01 6.53605998e-01 -9.26607013e-01 -4.10327792e-01
3.02738070e-01 3.23497206e-01 -8.75123963e-02 4.43942696e-01
-6.04855359e-01 4.78446215e-01 -7.34227657e-01 8.95265162e-01
9.10534322e-01 -4.31457251e-01 -1.45877004e-01 -5.39595902e-01
1.77649379e-01 -3.21017772e-01 -1.32061672e+00 1.60015035e+00
-4.86099780e-01 1.76770892e-02 1.54571906e-01 -6.91643238e-01
8.57972324e-01 2.42414698e-02 -8.39696229e-02 -1.01807225e+00
1.15994513e-01 2.82498062e-01 -4.72183585e-01 -6.69652581e-01
2.13242978e-01 -1.12478174e-01 -1.46186695e-01 4.19716686e-01
-5.26988395e-02 1.79750174e-01 2.45645955e-01 -1.39541492e-01
9.17372942e-01 -5.90530783e-02 8.74667764e-02 3.48024108e-02
4.15204257e-01 -1.36484131e-01 7.37351537e-01 1.04871035e+00
-4.51944649e-01 9.11577523e-01 2.59816851e-02 -5.36507308e-01
-1.16686416e+00 -7.86499441e-01 1.76632315e-01 1.42282379e+00
3.07185948e-01 -2.16075689e-01 -1.14006042e+00 -1.18676591e+00
-4.76350561e-02 1.30600885e-01 -7.73882926e-01 1.36583880e-01
-2.78432429e-01 -7.02453196e-01 5.64716935e-01 5.48920572e-01
7.72651434e-01 -8.27620804e-01 -5.41997440e-02 2.67749596e-02
-4.82006013e-01 -1.14911103e+00 -7.95271993e-01 2.33789593e-01
-4.92052913e-01 -1.12912631e+00 -9.53318059e-01 -6.64406121e-01
8.12475741e-01 7.26344407e-01 1.09370470e+00 1.75864443e-01
-2.26792037e-01 9.12153497e-02 -5.13003707e-01 -2.21877918e-01
-1.63414508e-01 -2.02669293e-01 -1.33453667e-01 4.23130423e-01
4.95477468e-01 -2.90414125e-01 -5.55374622e-01 4.87032503e-01
-1.16741550e+00 -3.09735805e-01 4.49335039e-01 1.10810673e+00
1.00291860e+00 5.03222942e-01 5.91182470e-01 -9.93620276e-01
7.44002223e-01 -4.36924547e-01 -6.25073135e-01 4.40582335e-01
-2.64937371e-01 2.03321055e-01 8.21275890e-01 -6.65953815e-01
-1.28410876e+00 8.41013119e-02 1.08183525e-03 -7.23823428e-01
-6.50870204e-01 1.93246141e-01 -5.30061185e-01 -2.88949221e-01
7.77171016e-01 2.60087937e-01 -4.03960705e-01 -9.00870562e-01
4.73676264e-01 9.81226742e-01 5.39440155e-01 -7.51783788e-01
8.44352186e-01 5.99288642e-01 -4.48187202e-01 -4.93641585e-01
-1.40055728e+00 -6.41349733e-01 -2.98887849e-01 -5.61504103e-02
5.40838838e-01 -1.03135872e+00 -3.19282264e-01 7.40285218e-01
-8.53841126e-01 -2.13404059e-01 -3.01216006e-01 -9.68870297e-02
-2.73113698e-01 3.34436864e-01 -3.20025384e-01 -8.04046512e-01
-2.56758124e-01 -1.12162471e+00 1.40072417e+00 4.34014857e-01
3.63682985e-01 -7.10947931e-01 -2.12553889e-01 8.22812617e-01
4.60086346e-01 1.16958946e-01 5.11769831e-01 -4.67441797e-01
-6.49582148e-01 -1.06394939e-01 -6.91346347e-01 6.11318767e-01
3.27514321e-01 -3.92768055e-01 -1.21687245e+00 -9.68221352e-02
-6.83139414e-02 -6.23493910e-01 9.15280700e-01 -1.62627064e-02
1.70329440e+00 -4.07891691e-01 -1.16259396e-01 4.41635698e-01
1.50128198e+00 -2.81235963e-01 2.99549341e-01 4.43841398e-01
9.92748678e-01 4.23244357e-01 8.30195487e-01 6.90462172e-01
4.85273808e-01 5.07514596e-01 4.27460819e-01 4.87190969e-02
-6.12934530e-01 -4.04259175e-01 -1.28985867e-01 3.45141768e-01
2.12116227e-01 -3.46476704e-01 -5.44741392e-01 6.52901292e-01
-1.86000049e+00 -7.02164888e-01 2.16451421e-01 1.99331737e+00
1.10469973e+00 -4.59216535e-02 1.56050310e-01 5.41570485e-02
1.04446685e+00 2.02212647e-01 -6.37827754e-01 2.84418076e-01
-3.33599418e-01 -1.22900225e-01 4.95214313e-01 7.46557415e-02
-1.29094374e+00 9.84035909e-01 5.10335159e+00 1.46880805e+00
-1.05946589e+00 3.35381716e-01 8.45269978e-01 9.59277973e-02
-4.54793632e-01 -4.11395550e-01 -7.20931292e-01 9.29919779e-01
4.04719383e-01 3.95150632e-01 7.27964520e-01 9.85147417e-01
2.64024317e-01 4.03155796e-02 -6.81328475e-01 1.40642834e+00
2.47149393e-01 -1.27571988e+00 7.02032223e-02 -1.05891768e-02
1.13174033e+00 -1.14152003e-02 1.24799768e-02 7.33589008e-02
3.62737864e-01 -8.89862180e-01 9.04754639e-01 6.47966444e-01
8.69841576e-01 -6.20392263e-01 7.55323648e-01 6.78622603e-01
-1.07319164e+00 -1.59960300e-01 -5.69549620e-01 2.73620129e-01
-8.85150135e-02 1.22764742e+00 -3.24924856e-01 2.83094406e-01
1.05153298e+00 1.99675381e-01 -8.22450638e-01 1.16182625e+00
-2.61044115e-01 6.20009720e-01 -1.83178738e-01 1.71087086e-01
2.75676906e-01 1.46182001e-01 2.23071232e-01 1.16917813e+00
2.81589665e-02 1.01360828e-01 4.69395638e-01 6.21553719e-01
-4.14329767e-01 -2.31763557e-01 -2.49275699e-01 -6.75405711e-02
5.52626312e-01 1.37678766e+00 -5.63272536e-01 -3.62116158e-01
-2.72583127e-01 1.01314235e+00 6.43567085e-01 4.73594725e-01
-5.99765122e-01 -1.86678112e-01 5.95718861e-01 9.09522362e-03
5.52392840e-01 7.35414922e-02 -3.53811473e-01 -1.47329605e+00
4.65480983e-01 -1.30129325e+00 3.69737834e-01 -7.79506922e-01
-2.00210428e+00 5.81968784e-01 -4.87911910e-01 -1.13629091e+00
7.76043013e-02 -4.05272812e-01 -1.35637254e-01 8.49293649e-01
-1.81465471e+00 -1.43124449e+00 -8.81925285e-01 8.11938167e-01
7.43738532e-01 7.27480277e-02 6.14939272e-01 4.92991716e-01
-3.72669041e-01 6.76486135e-01 1.35095820e-01 2.37561524e-01
8.31535578e-01 -1.23413360e+00 2.78122902e-01 8.79837692e-01
8.70717168e-02 4.14450705e-01 4.23138529e-01 -6.76225364e-01
-1.30513382e+00 -1.60388434e+00 4.20561999e-01 -2.57384151e-01
4.61854190e-01 -3.68927836e-01 -9.87591028e-01 1.79618195e-01
-2.17652380e-01 7.18143284e-01 4.67422128e-01 -1.37862042e-01
-9.33222651e-01 -5.07236004e-01 -1.43384504e+00 3.97590071e-01
1.13110483e+00 -6.44414365e-01 -3.68068188e-01 4.10326183e-01
6.43280983e-01 -3.59690845e-01 -5.61590731e-01 2.51046002e-01
3.64609748e-01 -7.93976128e-01 9.07995343e-01 -4.71687585e-01
1.18576877e-01 -6.16258264e-01 -4.62878287e-01 -1.55844092e+00
-4.23833966e-01 -2.26128846e-01 -1.29747123e-01 1.55800068e+00
-5.96792363e-02 -3.57678771e-01 8.26625049e-01 4.85593826e-01
1.71112698e-02 -3.64854127e-01 -7.25425065e-01 -7.28178740e-01
-1.97175339e-01 -4.03787524e-01 7.06434965e-01 6.91059530e-01
-6.14129066e-01 -4.28837799e-02 -4.96125519e-01 3.76213282e-01
1.07708526e+00 1.70965806e-01 7.53400922e-01 -1.12408435e+00
-2.24802822e-01 -1.12199754e-01 -9.11263227e-02 -9.61825669e-01
1.26784414e-01 -3.19559455e-01 4.78692055e-01 -1.45919466e+00
4.76607025e-01 -8.80482912e-01 -4.20481652e-01 5.20258486e-01
-5.16408086e-01 8.86466742e-01 2.69144684e-01 3.31393719e-01
-1.18739498e+00 4.05565202e-01 1.39414668e+00 -3.51555616e-01
3.72678816e-01 -1.90196410e-01 -1.06529200e+00 7.52487659e-01
7.60060370e-01 -5.96846879e-01 -2.71457821e-01 -4.79411215e-01
-5.86119853e-03 -3.09703529e-01 4.30208534e-01 -8.34177256e-01
1.62314430e-01 -5.21521211e-01 5.90817392e-01 -3.26324940e-01
1.87192947e-01 -9.46991503e-01 -4.69484739e-03 -1.59657881e-01
-2.78353482e-01 -4.96575356e-01 -1.02047175e-01 8.17494869e-01
-3.80320847e-01 -2.40775824e-01 9.08247769e-01 -5.74144125e-01
-8.54275584e-01 5.14675915e-01 -4.75360602e-02 3.87998194e-01
7.81573772e-01 -1.95865799e-02 -6.67344868e-01 -1.83562845e-01
-5.17353356e-01 1.91791132e-02 6.84667587e-01 4.73014951e-01
5.01531243e-01 -1.32255232e+00 -5.12676895e-01 2.27379382e-01
6.35500014e-01 1.38896435e-01 7.74353266e-01 2.48290747e-01
-6.37820289e-02 4.52371873e-02 1.25485733e-01 -4.93393272e-01
-1.07961762e+00 7.57336497e-01 2.68209130e-01 -2.45375782e-01
-1.88040957e-01 8.72028530e-01 2.87111789e-01 -6.34179354e-01
3.37620974e-01 -2.04944342e-01 1.06956638e-01 -1.09322324e-01
9.06127632e-01 2.80206263e-01 3.85357648e-01 -5.37966013e-01
-2.96776384e-01 5.84040940e-01 -1.11465693e-01 1.21839918e-01
1.14417279e+00 -5.91892898e-01 1.33007124e-01 3.52019072e-01
1.03044522e+00 7.52060488e-02 -1.64334452e+00 -5.13019145e-01
-2.54702747e-01 -8.80245686e-01 8.83800089e-02 -1.28796029e+00
-1.11001110e+00 6.49006248e-01 6.76750779e-01 1.96399987e-01
1.29001868e+00 7.13707507e-02 8.61901343e-01 2.15293989e-01
5.40392816e-01 -1.26192737e+00 3.71635169e-01 2.95077592e-01
5.46608865e-01 -1.71103144e+00 -4.92934823e-01 -4.62601602e-01
-5.20454645e-01 8.29295397e-01 6.80898547e-01 -1.57464296e-01
4.88971382e-01 2.65969038e-01 4.67937827e-01 -2.30999161e-02
-5.35649121e-01 -4.11573082e-01 3.56829524e-01 8.46169293e-01
-1.87870879e-02 1.51502326e-01 1.28834382e-01 8.10348272e-01
2.39737853e-01 1.73363760e-01 2.78936476e-01 8.28870654e-01
-5.95058084e-01 -1.03329885e+00 -7.49194324e-01 4.13144767e-01
-4.88554448e-01 -1.59467444e-01 -3.94170344e-01 2.36060023e-01
5.54193318e-01 1.27825689e+00 -2.18830377e-01 -3.11063200e-01
4.11646247e-01 -1.95455268e-01 4.38248098e-01 -5.49011171e-01
-6.14713788e-01 2.06270620e-01 4.81167762e-03 -6.84124410e-01
-4.11940634e-01 -2.04491362e-01 -7.05032885e-01 -1.85644239e-01
-3.38418096e-01 1.83504093e-02 6.88646913e-01 1.06259382e+00
5.89223623e-01 3.84356707e-01 6.66123509e-01 -8.22571039e-01
-8.42059255e-01 -9.13866162e-01 -7.34911680e-01 8.48553300e-01
5.58107674e-01 -6.36381269e-01 -4.99049336e-01 7.78443739e-02] | [9.583373069763184, 2.623800277709961] |
821c6ed4-0f40-4b85-a75f-83cbc43b1efe | gan2x-non-lambertian-inverse-rendering-of | 2206.09244 | null | https://arxiv.org/abs/2206.09244v4 | https://arxiv.org/pdf/2206.09244v4.pdf | GAN2X: Non-Lambertian Inverse Rendering of Image GANs | 2D images are observations of the 3D physical world depicted with the geometry, material, and illumination components. Recovering these underlying intrinsic components from 2D images, also known as inverse rendering, usually requires a supervised setting with paired images collected from multiple viewpoints and lighting conditions, which is resource-demanding. In this work, we present GAN2X, a new method for unsupervised inverse rendering that only uses unpaired images for training. Unlike previous Shape-from-GAN approaches that mainly focus on 3D shapes, we take the first attempt to also recover non-Lambertian material properties by exploiting the pseudo paired data generated by a GAN. To achieve precise inverse rendering, we devise a specularity-aware neural surface representation that continuously models the geometry and material properties. A shading-based refinement technique is adopted to further distill information in the target image and recover more fine details. Experiments demonstrate that GAN2X can accurately decompose 2D images to 3D shape, albedo, and specular properties for different object categories, and achieves the state-of-the-art performance for unsupervised single-view 3D face reconstruction. We also show its applications in downstream tasks including real image editing and lifting 2D GANs to decomposed 3D GANs. | ['Christian Theobalt', 'Lingjie Liu', 'Ayush Tewari', 'Xingang Pan'] | 2022-06-18 | null | null | null | null | ['3d-face-reconstruction', 'face-reconstruction'] | ['computer-vision', 'computer-vision'] | [ 6.62975848e-01 2.35162094e-01 5.12507141e-01 -5.51554680e-01
-7.02028036e-01 -5.20763338e-01 6.19388700e-01 -7.01581717e-01
3.94587994e-01 5.01856804e-01 2.06371620e-01 -5.22855259e-02
3.05839241e-01 -9.23203051e-01 -7.58723676e-01 -9.09946799e-01
6.32151723e-01 7.91584253e-01 -3.28836858e-01 -3.61336738e-01
-4.67364229e-02 7.85161674e-01 -1.60297453e+00 2.51749188e-01
7.51603961e-01 1.00160861e+00 3.93302217e-02 7.65429020e-01
-2.96912879e-01 5.88097930e-01 -4.37377334e-01 -4.85257745e-01
6.59787178e-01 -7.96172619e-01 -2.75250435e-01 5.86992919e-01
7.91017890e-01 -7.04299152e-01 -8.04024637e-02 7.58624017e-01
3.60283971e-01 -2.70141307e-02 9.65389609e-01 -9.03914988e-01
-9.00454640e-01 -3.66156191e-01 -8.58457148e-01 -6.21694326e-01
5.35925448e-01 9.69812348e-02 5.98263741e-01 -9.16275799e-01
4.02466029e-01 1.32390034e+00 6.80621386e-01 8.76373172e-01
-1.49106324e+00 -6.98183775e-01 8.72381032e-02 -3.63277465e-01
-1.07257986e+00 -5.27357221e-01 1.32001615e+00 -2.56449252e-01
5.41064024e-01 4.47892487e-01 7.88253725e-01 9.65651453e-01
-2.13587686e-01 4.63602513e-01 1.65862918e+00 -5.01467288e-01
7.16769025e-02 4.54947837e-02 -6.20257556e-01 8.18453550e-01
-7.28052929e-02 1.34022459e-01 -6.82056129e-01 -5.21094678e-03
1.18924987e+00 2.94206262e-01 -6.00924850e-01 -3.98328453e-01
-8.40094328e-01 5.30917525e-01 3.41367841e-01 -3.19716841e-01
-4.09367591e-01 -5.45616001e-02 -2.87624419e-01 1.90638915e-01
9.79779065e-01 3.49653065e-01 -3.28952223e-01 1.82333991e-01
-5.98613799e-01 5.46395704e-02 6.34607792e-01 9.84339356e-01
1.09745336e+00 3.42799574e-01 2.52463609e-01 8.18223238e-01
4.26405847e-01 1.11648035e+00 -9.25561190e-02 -1.34670508e+00
1.97902471e-01 4.11383897e-01 1.87035531e-01 -6.88543081e-01
1.79810092e-01 -2.32155755e-01 -1.02589321e+00 5.68104327e-01
1.33578017e-01 2.08277807e-01 -1.26441121e+00 1.60883856e+00
7.11958885e-01 3.09922695e-01 9.82452035e-02 1.00310731e+00
1.11473429e+00 4.87575084e-01 -5.82167387e-01 -8.50261897e-02
1.06403017e+00 -8.33174407e-01 -5.75455010e-01 -1.16118088e-01
-8.40282254e-03 -8.27261150e-01 1.16557765e+00 5.46829760e-01
-1.41320086e+00 -4.37920541e-01 -7.83331215e-01 -4.86788332e-01
1.36074498e-01 -2.14477330e-02 5.56985199e-01 7.82717168e-01
-1.11869538e+00 4.02389288e-01 -7.69143522e-01 4.04655710e-02
5.43575406e-01 1.58599690e-01 -4.10945266e-01 -4.36992347e-01
-6.16828859e-01 4.07595664e-01 -4.78118628e-01 1.78405344e-01
-9.25303638e-01 -1.03255796e+00 -9.76739705e-01 -1.59254342e-01
1.18039154e-01 -9.85963941e-01 7.99318254e-01 -8.27642441e-01
-2.05696368e+00 1.21267879e+00 -2.67986655e-01 1.79905415e-01
3.18777353e-01 -9.17522237e-02 -1.09035168e-02 2.07827345e-01
-6.36344403e-02 3.47050548e-01 1.32139575e+00 -1.89195538e+00
5.71928583e-02 -7.47904241e-01 1.51547298e-01 4.61893260e-01
7.98157156e-02 -4.40456629e-01 -5.12764096e-01 -5.76461732e-01
4.07938033e-01 -6.92950845e-01 2.23826505e-02 3.02024990e-01
-4.33116049e-01 5.43076813e-01 8.50242674e-01 -8.29119861e-01
1.37755245e-01 -1.98222625e+00 3.63217115e-01 5.09124249e-02
3.77105385e-01 6.36892468e-02 -2.55316973e-01 1.41128600e-01
2.04277337e-02 -1.70020372e-01 -5.20688593e-01 -9.67379630e-01
-1.36397392e-01 2.68693805e-01 -4.96716708e-01 4.09579039e-01
2.38855883e-01 8.98369670e-01 -7.04403222e-01 -7.84453154e-02
4.42962617e-01 1.14805543e+00 -7.50331521e-01 7.75325119e-01
-2.68002212e-01 1.32878768e+00 -2.60155618e-01 6.01279199e-01
1.20593119e+00 -2.02450961e-01 6.01715893e-02 -3.78792584e-01
2.43196309e-01 2.49212161e-01 -6.52327776e-01 1.86835444e+00
-1.08056855e+00 3.75546426e-01 5.36577165e-01 -8.82651508e-01
1.10819256e+00 1.79693669e-01 5.16977131e-01 -8.13975155e-01
-5.70040010e-03 1.16468057e-01 -6.81929052e-01 -1.50962353e-01
8.26317072e-02 -7.17748463e-01 3.32156181e-01 5.07225394e-01
-1.15341522e-01 -1.14087534e+00 -8.03635001e-01 -2.53462903e-02
7.21749008e-01 6.30419374e-01 -4.30980809e-02 1.86213940e-01
4.87438887e-01 -5.06219983e-01 4.10150379e-01 3.32428217e-01
6.57149971e-01 1.22392559e+00 1.32337674e-01 -5.01577914e-01
-1.09356940e+00 -1.40628147e+00 1.11438334e-01 5.48872411e-01
1.36892378e-01 -5.26667871e-02 -8.09008121e-01 -4.48442012e-01
-1.47294134e-01 8.06878150e-01 -7.39061654e-01 5.20149097e-02
-6.00488544e-01 -4.28560138e-01 2.36970354e-02 1.67822868e-01
7.67314672e-01 -8.01729918e-01 -4.03577238e-01 -2.12165892e-01
-1.14977829e-01 -1.29055488e+00 -6.21379077e-01 -2.76262283e-01
-7.64934778e-01 -1.03674412e+00 -8.51742387e-01 -3.87443334e-01
9.54869151e-01 4.91586268e-01 1.50369799e+00 1.82138786e-01
-2.58811027e-01 8.59871924e-01 -1.37476757e-01 -3.91474903e-01
-4.66845483e-01 -5.89085340e-01 -2.72231996e-01 5.79098046e-01
-2.07176819e-01 -1.12601817e+00 -7.49553382e-01 3.33338708e-01
-8.94614279e-01 7.65683889e-01 2.69538879e-01 8.06029320e-01
8.35864842e-01 -1.25665143e-01 6.59276620e-02 -1.25315499e+00
-2.58540362e-02 -3.58647550e-03 -5.21585763e-01 9.33228880e-02
-3.47128868e-01 -4.15011719e-02 6.82374716e-01 -3.08876038e-01
-1.78270006e+00 -3.75110619e-02 -2.98345387e-01 -7.71527469e-01
-4.63506907e-01 -2.17868313e-01 -6.69364154e-01 -1.88233316e-01
3.75774831e-01 5.86746395e-01 1.74226105e-01 -7.24508584e-01
3.56761307e-01 3.91022086e-01 5.06139517e-01 -5.79671800e-01
1.13692367e+00 1.16926765e+00 3.75870436e-01 -1.15222764e+00
-1.26306379e+00 -1.42374262e-01 -6.12785697e-01 -1.06686398e-01
8.07527423e-01 -1.06059206e+00 -7.32357025e-01 8.28708768e-01
-1.18524384e+00 -9.07139182e-01 -5.57830453e-01 1.16266474e-01
-8.19048345e-01 3.51860970e-01 -4.67617929e-01 -7.70120800e-01
-4.13790971e-01 -9.32130456e-01 1.86868942e+00 8.52277875e-02
3.71441156e-01 -1.09639585e+00 -1.19327843e-01 8.55008721e-01
3.60684574e-01 4.95972693e-01 8.19126368e-01 4.75963086e-01
-8.25638175e-01 2.59711087e-01 -2.91808605e-01 5.99791050e-01
6.17168486e-01 -4.53101784e-01 -1.27221620e+00 -1.90164939e-01
5.38094580e-01 -2.58406818e-01 6.89671159e-01 2.97328860e-01
1.45981038e+00 -3.89808804e-01 1.47812784e-01 1.46337104e+00
1.29995978e+00 -1.53825074e-01 7.05336511e-01 -3.86449844e-01
1.29702199e+00 7.30959237e-01 3.12383085e-01 4.76149768e-01
4.45624918e-01 8.16252708e-01 7.91984260e-01 -5.36884606e-01
-7.56938338e-01 -5.21539152e-01 1.98955595e-01 7.61343002e-01
-4.32810396e-01 -2.50394136e-01 -3.20978373e-01 1.16411857e-01
-1.10588562e+00 -6.22799397e-01 -2.68759318e-02 2.30476260e+00
7.74733841e-01 -4.26925719e-01 -3.23635072e-01 -5.51637188e-02
2.91776180e-01 9.60947722e-02 -8.08295369e-01 -1.44033253e-01
-2.42993087e-01 6.59583330e-01 1.86795712e-01 8.63035917e-01
-3.84919286e-01 8.37570131e-01 5.39192915e+00 5.68316579e-01
-1.23011792e+00 1.96146607e-01 7.98021793e-01 -8.94356240e-03
-1.26326454e+00 -1.35591969e-01 -4.64597970e-01 2.10746065e-01
3.32412928e-01 5.05705118e-01 9.71650720e-01 2.18382180e-01
7.36425072e-02 1.33434549e-01 -9.81592059e-01 1.33459640e+00
3.74886423e-01 -1.15389526e+00 3.24206650e-01 2.06709623e-01
1.10210443e+00 -2.38264710e-01 2.53144354e-01 -1.41758665e-01
2.64695197e-01 -1.12165761e+00 6.87987745e-01 6.92870140e-01
1.50527394e+00 -4.34787631e-01 5.22880554e-02 3.74393553e-01
-8.99191201e-01 4.42008317e-01 -1.84816450e-01 1.52699843e-01
2.98881590e-01 7.65608549e-01 -6.99446499e-01 5.73702514e-01
7.45228231e-01 9.16795909e-01 -4.10470739e-03 2.12161124e-01
-6.25932813e-01 4.95085299e-01 -3.47804189e-01 5.72520018e-01
-3.47316086e-01 -7.75127470e-01 6.22287035e-01 5.92834353e-01
4.91479129e-01 5.24758160e-01 -1.29231289e-01 1.17940414e+00
-2.26258218e-01 -2.12853909e-01 -7.76120663e-01 3.36519659e-01
-6.29991964e-02 1.18810380e+00 -5.04521489e-01 -1.50665268e-01
-2.51770377e-01 1.39781475e+00 2.30835438e-01 6.85737729e-01
-6.88694119e-01 1.80117443e-01 7.34643817e-01 4.88379449e-01
3.52785736e-01 -4.61120725e-01 -2.91655838e-01 -1.40266740e+00
9.39218700e-02 -6.91113770e-01 -2.06051230e-01 -1.33655953e+00
-1.29302180e+00 5.30294538e-01 -9.78018939e-02 -1.06213617e+00
-1.44665286e-01 -6.68018043e-01 -5.22992671e-01 1.16527247e+00
-1.86227429e+00 -1.55945635e+00 -7.37864375e-01 7.30925679e-01
5.16263306e-01 1.06141217e-01 1.00433207e+00 -6.98547661e-02
1.14577666e-01 3.60747755e-01 -8.91147330e-02 -1.34719878e-01
5.29059649e-01 -1.15020716e+00 4.88997251e-01 4.19863105e-01
9.76963863e-02 3.49013090e-01 4.97725844e-01 -4.76135492e-01
-1.86953211e+00 -9.98043537e-01 4.78020720e-02 -4.25006151e-01
-1.12542547e-01 -6.23014331e-01 -7.55688787e-01 6.35428607e-01
6.47244155e-02 2.58680582e-01 6.79134488e-01 -3.11812937e-01
-5.44984758e-01 -2.60430723e-01 -1.42831707e+00 5.06694734e-01
1.47606575e+00 -6.36684418e-01 -9.06792730e-02 3.73284787e-01
5.96160531e-01 -7.91871548e-01 -7.96657681e-01 4.23808962e-01
4.72709626e-01 -1.44303942e+00 1.19211853e+00 -2.93879926e-01
5.18808186e-01 -2.74616748e-01 -3.76122206e-01 -1.50035667e+00
8.45019296e-02 -8.49853754e-01 -1.35470897e-01 1.14729381e+00
-1.54199619e-02 -9.14576650e-01 1.02716434e+00 5.15700579e-01
-2.74070472e-01 -5.95638394e-01 -5.89848399e-01 -3.59297842e-01
-1.80918157e-01 -2.93181181e-01 1.06739020e+00 7.63436377e-01
-1.19598603e+00 2.85943389e-01 -5.74029803e-01 2.89088964e-01
1.12004733e+00 7.14496434e-01 1.17707562e+00 -1.14320862e+00
-5.44771850e-01 5.07141184e-03 -5.33787608e-02 -1.50354123e+00
4.15042400e-01 -6.87076807e-01 1.22050159e-01 -1.36508954e+00
-2.25150347e-04 -6.81921482e-01 3.67257714e-01 1.77957267e-01
8.13603625e-02 8.34375799e-01 -3.19032185e-02 1.25601649e-01
-1.21449614e-02 1.06655300e+00 1.81616557e+00 -1.42624676e-01
-1.20696537e-01 1.21841118e-01 -8.85156274e-01 8.03506553e-01
5.21777689e-01 -9.96392369e-02 -6.70823991e-01 -9.22254920e-01
1.91162422e-01 2.14409143e-01 5.33094525e-01 -5.32908022e-01
-3.52025062e-01 -3.30687165e-01 5.12113810e-01 -4.84795570e-01
1.07279670e+00 -9.95239019e-01 5.06144583e-01 -1.85213521e-01
-1.15193404e-01 -4.47906673e-01 -5.13049550e-02 5.87813914e-01
3.32443342e-02 1.78071171e-01 8.17924321e-01 -4.06639040e-01
-5.67437336e-02 7.81739354e-01 4.06222463e-01 1.05366722e-01
5.10781348e-01 -3.70065302e-01 2.95611620e-02 -7.89588392e-01
-4.34792668e-01 -3.38724047e-01 1.11818445e+00 -3.78322192e-02
1.05028033e+00 -1.29706240e+00 -8.80764127e-01 8.83145809e-01
-1.54766321e-01 6.48170352e-01 4.63282555e-01 5.06880879e-01
-6.43644869e-01 -1.65678471e-01 -8.55404884e-02 -7.60103822e-01
-1.24094379e+00 2.42540821e-01 5.17768681e-01 -5.68793267e-02
-9.82571244e-01 7.89939106e-01 1.05513394e+00 -8.60388815e-01
-3.00828278e-01 1.36491992e-02 1.92355081e-01 -5.74553549e-01
4.75486428e-01 -4.00916450e-02 1.54841781e-01 -7.87054300e-01
-9.82553605e-03 1.13078332e+00 3.96571994e-01 -1.49584413e-01
1.75705326e+00 -2.61166066e-01 -3.58270437e-01 2.73771733e-01
1.22831333e+00 4.56415027e-01 -1.69864261e+00 -2.63559371e-01
-1.08376658e+00 -9.80001509e-01 1.32631794e-01 -5.54383397e-01
-1.66683936e+00 1.17733812e+00 9.72239152e-02 -4.17250730e-02
1.59706044e+00 4.62560989e-02 8.29497159e-01 -8.57020691e-02
5.35815001e-01 -3.81788552e-01 1.69704482e-01 2.73461908e-01
1.20457053e+00 -1.06805348e+00 1.18417233e-01 -9.36726809e-01
-3.52630258e-01 9.99486208e-01 5.29533982e-01 -1.28197357e-01
7.70238638e-01 2.43580833e-01 5.01197763e-02 -3.27816546e-01
-4.08039927e-01 1.30631760e-01 5.13577461e-01 9.15944278e-01
1.83197111e-01 9.41345543e-02 6.74656332e-01 1.13008521e-01
-2.67288268e-01 -3.79349172e-01 4.03566331e-01 3.69947523e-01
2.71768361e-01 -1.04760075e+00 -4.05711889e-01 3.15751076e-01
-4.86142188e-02 -3.81767638e-02 -3.44910145e-01 3.63074034e-01
4.89532351e-02 6.44563854e-01 1.36851743e-01 -1.45164907e-01
2.84276217e-01 -3.18347961e-01 1.04668725e+00 -8.91750574e-01
4.11013663e-02 2.11963966e-01 -1.78551987e-01 -5.90322852e-01
-7.04240620e-01 -5.47572911e-01 -1.00380266e+00 -2.23013312e-01
-1.45984337e-01 -1.75038353e-01 8.42651844e-01 6.58657193e-01
3.61476153e-01 3.07909667e-01 9.80225563e-01 -1.42256844e+00
-9.68229696e-02 -6.90028787e-01 -7.94581652e-01 6.44833684e-01
7.21139669e-01 -7.64295816e-01 -7.05035985e-01 1.97271213e-01] | [12.648963928222656, -0.4665989875793457] |
08463077-762e-4fce-b3ad-74263132bdb5 | improved-beam-search-for-hallucination | 2212.02712 | null | https://arxiv.org/abs/2212.02712v1 | https://arxiv.org/pdf/2212.02712v1.pdf | Improved Beam Search for Hallucination Mitigation in Abstractive Summarization | Advancement in large pretrained language models has significantly improved their performance for conditional language generation tasks including summarization albeit with hallucinations. To reduce hallucinations, conventional methods proposed improving beam search or using a fact checker as a postprocessing step. In this paper, we investigate the use of the Natural Language Inference (NLI) entailment metric to detect and prevent hallucinations in summary generation. We propose an NLI-assisted beam re-ranking mechanism by computing entailment probability scores between the input context and summarization model-generated beams during saliency-enhanced greedy decoding. Moreover, a diversity metric is introduced to compare its effectiveness against vanilla beam search. Our proposed algorithm significantly outperforms vanilla beam decoding on XSum and CNN/DM datasets. | ['Erik Visser', 'Arvind Krishna Sridhar'] | 2022-12-06 | null | null | null | null | ['abstractive-text-summarization'] | ['natural-language-processing'] | [ 6.00195646e-01 5.32538772e-01 -2.75789708e-01 -2.66663194e-01
-1.08012021e+00 -9.33822021e-02 8.73951495e-01 5.02287626e-01
-5.52641690e-01 1.16786897e+00 1.07421637e+00 -3.13752294e-01
-1.56559888e-02 -7.71862090e-01 -5.94297826e-01 -2.05395177e-01
2.24351704e-01 3.83644044e-01 6.43566251e-02 -1.71804130e-01
7.80636191e-01 4.50622663e-02 -1.33650315e+00 7.12503791e-01
1.25819421e+00 4.88844305e-01 6.13584936e-01 6.92591965e-01
-1.96746975e-01 1.31687176e+00 -9.21216249e-01 -4.89455968e-01
-2.88804233e-01 -9.33010995e-01 -6.99903607e-01 -7.75875896e-02
3.90322983e-01 -1.88560694e-01 -6.93424717e-02 9.68606353e-01
9.46276844e-01 2.39092708e-01 6.16162539e-01 -1.08987808e+00
-6.65638626e-01 1.30486429e+00 -5.36608934e-01 5.10471046e-01
7.67315269e-01 -9.02604535e-02 1.14806151e+00 -1.20971417e+00
6.72278523e-01 1.39320743e+00 3.76195878e-01 6.78390920e-01
-9.43300366e-01 -6.52355731e-01 6.58249781e-02 5.68029761e-01
-1.28369606e+00 -5.27019560e-01 6.42711520e-01 6.48173094e-02
1.43675578e+00 3.61112475e-01 6.06207192e-01 1.10375321e+00
2.47210205e-01 1.17067420e+00 8.71954679e-01 -7.25458264e-01
2.69128293e-01 1.31809518e-01 -9.32554752e-02 8.45741093e-01
3.71065348e-01 -3.14587921e-01 -1.14738846e+00 -2.47002974e-01
2.78706878e-01 -4.70273376e-01 -3.90776694e-01 5.49273789e-01
-1.29177964e+00 8.92175794e-01 3.35059673e-01 1.76603720e-01
-5.29613495e-01 1.72988456e-02 1.98082611e-01 -1.35236576e-01
6.02089465e-01 8.12617481e-01 -1.87042847e-01 1.54773384e-01
-1.49580491e+00 6.27499163e-01 4.31365907e-01 1.01381958e+00
3.40579212e-01 2.52312630e-01 -1.13817441e+00 7.70967305e-01
3.17374557e-01 3.55386645e-01 7.57043302e-01 -5.88936985e-01
7.41688848e-01 5.70841193e-01 -1.40145868e-01 -1.00718105e+00
-3.84914547e-01 -1.05001891e+00 -1.00665689e+00 -3.66009653e-01
-3.31001192e-01 -6.01890031e-03 -9.41487014e-01 1.81795943e+00
-4.95589972e-02 1.43362373e-01 1.94213644e-01 6.25090778e-01
8.58375013e-01 7.99744189e-01 8.49153399e-02 -5.87533414e-01
1.08772111e+00 -1.01517725e+00 -9.84552205e-01 -3.79967600e-01
7.60623217e-01 -6.92270279e-01 1.25214529e+00 4.91963714e-01
-1.57051206e+00 -3.00245374e-01 -1.20317018e+00 -1.61860839e-01
6.78539276e-03 1.48179188e-01 5.12061238e-01 3.37506205e-01
-1.10473514e+00 3.70950431e-01 -3.35046679e-01 -1.45032123e-01
7.88620889e-01 -5.44972494e-02 1.45491987e-01 -7.20662400e-02
-1.28826666e+00 1.16087937e+00 6.71415508e-01 -2.52286315e-01
-1.00895798e+00 -7.66623378e-01 -9.82526243e-01 3.69849235e-01
2.67062873e-01 -1.26544416e+00 1.40179217e+00 -3.09973300e-01
-1.03634155e+00 5.53832948e-01 -7.25946963e-01 -1.03551841e+00
3.31054419e-01 -3.18759859e-01 -2.42918402e-01 7.14247674e-02
3.21145505e-01 8.89586806e-01 7.26739526e-01 -1.00094593e+00
-6.76232398e-01 -2.59996708e-02 -8.54037181e-02 5.69545686e-01
-2.61377811e-01 -4.92899828e-02 -1.54943332e-01 -9.60460842e-01
2.01458737e-01 -4.77168262e-01 -3.07381123e-01 -6.69043362e-01
-9.83770251e-01 -3.78994346e-01 1.85249507e-01 -7.35991001e-01
1.77262163e+00 -1.56055987e+00 1.29437327e-01 -5.73881976e-02
1.83378443e-01 5.88960648e-02 -1.28295675e-01 3.85068268e-01
9.68154967e-02 2.83158869e-01 -4.62495029e-01 -6.74744189e-01
-1.06114745e-01 -2.11013034e-01 -7.70559072e-01 4.00897786e-02
2.27549687e-01 9.12371099e-01 -9.09795940e-01 -8.49186718e-01
-1.44105777e-01 8.77128989e-02 -8.75148237e-01 1.06454410e-01
-5.28086066e-01 2.21457064e-01 -6.16515353e-02 6.36957884e-01
4.71443594e-01 -2.08999440e-01 -4.02667001e-02 -4.18250225e-02
5.87736145e-02 7.69179106e-01 -5.89612544e-01 1.85673177e+00
-6.04469955e-01 6.77333295e-01 -5.80941677e-01 -5.63186109e-01
7.87350893e-01 3.24220926e-01 -2.70237267e-01 -1.01001215e+00
6.45606816e-02 -9.09566041e-03 -1.09136946e-01 -2.71641225e-01
9.86913443e-01 -2.96675980e-01 -8.14449489e-02 4.97045279e-01
6.19804114e-02 -1.48106799e-01 4.05377656e-01 9.06208515e-01
1.07034814e+00 -1.80265293e-01 5.03050148e-01 -7.80833289e-02
5.08405149e-01 6.66840971e-02 4.63676035e-01 1.12537181e+00
1.54273838e-01 7.89777875e-01 4.38361764e-01 1.52799830e-01
-8.16668212e-01 -1.08866298e+00 1.68513820e-01 9.95724797e-01
-2.35456694e-02 -7.91170359e-01 -6.33310318e-01 -5.07652998e-01
-3.66932958e-01 1.84942567e+00 -4.29156810e-01 -4.78676140e-01
-4.90547091e-01 -9.93334770e-01 7.17070699e-01 5.01844287e-01
5.89298785e-01 -1.09493852e+00 -6.09385967e-01 2.04638153e-01
-6.31724298e-01 -8.81799817e-01 -2.92000830e-01 -4.04259600e-02
-8.90841484e-01 -6.41665876e-01 -6.37032211e-01 -5.97792387e-01
6.04849517e-01 5.71926311e-02 1.26190042e+00 2.20314283e-02
8.73039290e-03 -1.89049065e-01 -4.53899324e-01 -6.67176187e-01
-3.95802885e-01 8.32421780e-02 -6.01743460e-02 -4.82191473e-01
2.64365412e-02 -5.12099326e-01 -5.89394152e-01 -3.08182359e-01
-7.79660165e-01 7.27767169e-01 7.18747973e-01 6.92240417e-01
4.86404061e-01 -6.62296712e-02 8.66283953e-01 -7.27409959e-01
1.34624207e+00 -4.84705597e-01 -9.97635871e-02 3.64230663e-01
-8.06368887e-01 3.76743257e-01 3.13635826e-01 -1.59998447e-01
-1.36660933e+00 -5.36786258e-01 -2.08047077e-01 -4.73463535e-02
3.36742193e-01 8.02506626e-01 -7.30022043e-02 6.35147512e-01
7.18131781e-01 4.90429282e-01 -7.49736845e-01 -2.61175632e-01
4.29844350e-01 3.93654883e-01 6.00595891e-01 -4.03111279e-02
3.32289785e-01 2.13645935e-01 -1.00356102e-01 -5.10623157e-01
-1.21218288e+00 -5.18862642e-02 -1.57847822e-01 -2.41817072e-01
7.66705930e-01 -8.85972977e-01 -2.36287311e-01 1.25123039e-01
-1.63898432e+00 1.21924505e-01 -1.80235773e-01 4.11913723e-01
-3.95718277e-01 3.60351831e-01 -4.20611680e-01 -1.06992161e+00
-9.27372038e-01 -6.91331506e-01 9.64630961e-01 2.36007795e-01
-6.03172004e-01 -5.99217534e-01 1.95151389e-01 4.29629266e-01
4.22444850e-01 1.59873635e-01 1.09583294e+00 -6.90660357e-01
-4.97945756e-01 -1.98650926e-01 -9.90161896e-02 1.64216142e-02
-3.11434567e-01 -4.19126749e-01 -7.70065486e-01 -5.01820892e-02
2.00252682e-02 -3.23865980e-01 1.38639104e+00 4.46399033e-01
1.11044538e+00 -6.18817627e-01 -2.88709551e-01 2.87339151e-01
1.16043103e+00 3.02162971e-02 7.81372190e-01 2.88487911e-01
5.41830838e-01 3.00600827e-01 7.08528459e-01 9.01928842e-01
3.83720547e-01 3.41784120e-01 2.70871580e-01 2.05887452e-01
-3.58649433e-01 -7.13384807e-01 3.56593907e-01 7.53731310e-01
3.16116601e-01 -9.44113374e-01 -8.15702558e-01 6.82696640e-01
-1.83476806e+00 -1.12594748e+00 -1.22615866e-01 1.95026577e+00
1.02756393e+00 4.86743331e-01 -3.93864334e-01 2.25017652e-01
6.47775352e-01 2.88259685e-01 -3.43520343e-01 -4.58496153e-01
-5.71246326e-01 3.78827572e-01 1.26833752e-01 7.77252555e-01
-6.22669995e-01 1.18656707e+00 5.71192312e+00 1.03159511e+00
-7.10962951e-01 3.23290139e-01 5.35719931e-01 -4.93204713e-01
-6.84186876e-01 -4.77543883e-02 -8.18811476e-01 2.71301568e-01
6.82313979e-01 -6.05667591e-01 6.67310581e-02 6.29117668e-01
4.49642956e-01 -4.89638329e-01 -1.03444481e+00 9.41737354e-01
6.44577861e-01 -1.74969792e+00 6.70578837e-01 -4.24024850e-01
7.23634958e-01 -1.71836674e-01 1.27185091e-01 3.04140538e-01
1.33167729e-01 -1.00530899e+00 9.05811548e-01 7.65300810e-01
5.61050773e-01 -8.11942518e-01 5.20121574e-01 6.28338277e-01
-6.66677237e-01 1.02039702e-01 -2.51167446e-01 -1.75228447e-01
5.46879053e-01 7.96205878e-01 -1.62552047e+00 3.92339945e-01
1.23542557e-02 3.82765979e-01 -8.02601099e-01 1.17464411e+00
-6.89630866e-01 7.87789881e-01 2.94201411e-02 -2.70302325e-01
1.21484652e-01 4.86121535e-01 8.45625758e-01 1.46237481e+00
5.40315032e-01 -2.90600397e-02 -2.55258471e-01 9.44883227e-01
-2.20060349e-01 9.88585502e-02 -4.08168793e-01 1.28196748e-02
6.13348544e-01 8.18477213e-01 -5.43479562e-01 -6.43362641e-01
7.30143040e-02 1.13438618e+00 2.48395965e-01 1.01374239e-01
-9.09352005e-01 -2.40533501e-01 4.48358692e-02 5.90475462e-02
-6.64436594e-02 -4.10381444e-02 -7.28513718e-01 -1.18710506e+00
-2.39446357e-01 -6.00327075e-01 3.96831423e-01 -1.34973025e+00
-6.90513492e-01 6.43235326e-01 2.06607267e-01 -9.33001578e-01
-5.58090389e-01 1.75264463e-01 -9.55175877e-01 8.32773983e-01
-1.43945420e+00 -8.84674251e-01 -1.34097114e-01 3.83581519e-01
1.06981444e+00 -2.16542765e-01 6.16497219e-01 2.69088522e-02
-5.97290337e-01 7.33359516e-01 -5.53303719e-01 -3.92464697e-01
4.48088527e-01 -1.20030999e+00 2.41959229e-01 1.23078430e+00
2.22279564e-01 6.16619289e-01 1.16446674e+00 -1.09533751e+00
-8.69818747e-01 -1.31263602e+00 1.47344470e+00 -1.58640876e-01
1.90872058e-01 -2.21483663e-01 -8.39559674e-01 5.16493142e-01
7.15013206e-01 -9.82510805e-01 6.47895932e-01 -9.81295258e-02
-1.02323636e-01 3.17209214e-01 -1.11934185e+00 9.80008662e-01
1.20834529e+00 -2.44499847e-01 -9.93033469e-01 6.26795709e-01
9.42006111e-01 -2.87305057e-01 2.93630594e-03 6.26207411e-01
1.99279308e-01 -7.89811194e-01 7.72049069e-01 -4.93208826e-01
9.11255479e-01 -1.49234578e-01 -1.07467815e-01 -1.36256528e+00
-4.41110939e-01 -5.26384473e-01 -2.85865366e-01 1.16204774e+00
8.61678660e-01 -1.65860597e-02 7.32119918e-01 1.96102142e-01
-3.35024685e-01 -8.76871645e-01 -7.04362512e-01 -3.04834038e-01
-2.89804906e-01 -7.54837215e-01 4.99326169e-01 5.18199861e-01
2.69228250e-01 8.83670032e-01 -6.09560609e-01 5.12724444e-02
5.41078389e-01 -2.83543523e-02 4.48346764e-01 -6.11206353e-01
-3.11032057e-01 -4.01991844e-01 7.87547976e-02 -8.07517588e-01
2.13228598e-01 -1.28359878e+00 8.52651969e-02 -2.09547687e+00
4.82429594e-01 2.69259810e-01 -2.86868781e-01 4.99269933e-01
-4.77389485e-01 7.66512007e-02 3.16047668e-01 2.65967045e-02
-8.46315801e-01 9.25026476e-01 9.84053850e-01 -1.63266554e-01
-1.67368293e-01 -7.06673861e-02 -9.84797239e-01 7.80427754e-01
1.03714287e+00 -6.19407117e-01 -6.45634890e-01 -6.07851028e-01
7.05137610e-01 2.21592814e-01 2.94996947e-01 -1.10448205e+00
5.44875503e-01 1.77642167e-01 4.27431852e-01 -1.21197331e+00
1.95568994e-01 6.64281994e-02 -3.10483813e-01 4.85008538e-01
-9.89688516e-01 2.07585827e-01 1.61259025e-01 5.87060630e-01
-9.28862914e-02 -4.10870403e-01 5.12559950e-01 -2.82362252e-01
-4.54598695e-01 -1.40876517e-01 -6.65645838e-01 1.86495230e-01
5.79335749e-01 -3.44435498e-02 -3.74971509e-01 -6.23812377e-01
-5.23800790e-01 2.08717167e-01 -6.70398548e-02 2.96251923e-01
1.29585791e+00 -1.12016881e+00 -1.19521713e+00 3.40562239e-02
-9.85660683e-03 1.37511874e-02 1.58948615e-01 8.64302158e-01
-2.99604654e-01 5.93998134e-01 1.34146631e-01 -1.86349005e-01
-1.36754715e+00 2.19538495e-01 -1.26557902e-01 -5.78991115e-01
-3.44796240e-01 1.35895693e+00 1.24585673e-01 1.02856487e-01
2.80099154e-01 -4.69948530e-01 -3.20350885e-01 3.83881629e-02
7.75352478e-01 3.88813913e-01 2.14729920e-01 -2.14749381e-01
-4.24581409e-01 -3.13226402e-01 -3.75711471e-01 -6.61898375e-01
1.07262397e+00 -5.04541919e-02 -2.56394386e-01 1.50392413e-01
7.93037891e-01 1.16793178e-01 -5.27425647e-01 -2.31284484e-01
2.93615609e-01 -3.60161394e-01 3.40562135e-01 -1.24234521e+00
-5.32002628e-01 5.52189052e-01 1.55639261e-01 -2.22153291e-01
1.28324628e+00 1.89100370e-01 9.41437662e-01 4.93980706e-01
9.82261002e-02 -9.24400449e-01 2.20123112e-01 6.74302042e-01
1.37007284e+00 -8.02441537e-01 1.02661669e-01 -1.11052603e-01
-8.86739314e-01 6.12318039e-01 6.64317191e-01 7.06798881e-02
8.67407769e-02 2.23352179e-01 -4.84352529e-01 -9.81188118e-02
-1.20223141e+00 -1.18788771e-01 3.20101619e-01 2.86305606e-01
5.71538150e-01 -8.92040133e-02 -8.41668665e-01 7.67607212e-01
-6.58625007e-01 1.27019554e-01 7.13974893e-01 9.07525420e-01
-8.29832494e-01 -5.55203259e-01 -2.17039064e-01 7.05922365e-01
-3.37360829e-01 -8.37265193e-01 -4.94996578e-01 1.63000837e-01
1.39062911e-01 1.15357256e+00 -3.50769982e-02 -4.19499516e-01
1.47650793e-01 9.56320837e-02 5.65021157e-01 -8.95433545e-01
-4.87064302e-01 -9.48317051e-02 3.56296301e-01 -2.90017933e-01
-2.65424073e-01 -5.41314721e-01 -1.38986444e+00 -5.87468855e-02
-4.20163363e-01 1.36108786e-01 5.40368855e-01 1.14593065e+00
5.06296873e-01 7.41610110e-01 2.60260314e-01 -6.41387999e-01
-3.58812153e-01 -1.29994810e+00 -1.74158528e-01 5.61704077e-02
1.97281435e-01 -3.83032173e-01 -2.65600532e-01 -2.08976775e-01] | [12.243635177612305, 9.283980369567871] |
f7a80a5c-ee54-4afa-aa01-462b15dc212c | laplace-approximated-neural-additive-models | 2305.16905 | null | https://arxiv.org/abs/2305.16905v1 | https://arxiv.org/pdf/2305.16905v1.pdf | Laplace-Approximated Neural Additive Models: Improving Interpretability with Bayesian Inference | Deep neural networks (DNNs) have found successful applications in many fields, but their black-box nature hinders interpretability. This is addressed by the neural additive model (NAM), in which the network is divided into additive sub-networks, thus making apparent the interaction between input features and predictions. In this paper, we approach the additive structure from a Bayesian perspective and develop a practical Laplace approximation. This enhances interpretability in three primary ways: a) It provides credible intervals for the recovered feature interactions by estimating function-space uncertainty of the sub-networks; b) it yields a tractable estimate of the marginal likelihood, which can be used to perform an implicit selection of features through an empirical Bayes procedure; and c) it can be used to rank feature pairs as candidates for second-order interactions in fine-tuned interaction models. We show empirically that our proposed Laplace-approximated NAM (LA-NAM) improves performance and interpretability on tabular regression and classification datasets and challenging real-world medical tasks. | ['Vincent Fortuin', 'Gunnar Rätsch', 'Hugo Yèche', 'Alexander Immer', 'Kouroche Bouchiat'] | 2023-05-26 | null | null | null | null | ['bayesian-inference', 'additive-models'] | ['methodology', 'methodology'] | [ 3.82859141e-01 5.76292574e-01 -2.07725167e-01 -7.13917911e-01
-7.22265065e-01 -3.57250541e-01 5.39614856e-01 3.45425189e-01
-2.28421718e-01 1.02232170e+00 2.50838459e-01 -4.77876276e-01
-7.35617399e-01 -5.23315847e-01 -9.66141105e-01 -7.26623893e-01
-3.56807619e-01 5.86614370e-01 -7.32990429e-02 9.33434516e-02
-9.35960934e-02 4.68488157e-01 -1.27468181e+00 2.44701996e-01
1.32430232e+00 1.19506884e+00 -2.51753986e-01 3.26810390e-01
2.40250900e-01 8.69589329e-01 -4.89690900e-01 -5.09509444e-01
1.80333823e-01 -4.39757556e-01 -5.20877302e-01 -3.76872540e-01
1.26323581e-01 -3.93385828e-01 -1.08092539e-01 1.03264332e+00
1.84450105e-01 1.15107283e-01 1.07587957e+00 -1.04320860e+00
-5.84915817e-01 9.98691499e-01 -4.86529112e-01 -1.96595505e-01
-2.17996746e-01 -1.02685533e-01 1.26729238e+00 -7.58014679e-01
7.50752240e-02 1.57540464e+00 9.09486234e-01 2.74259597e-01
-1.42155099e+00 -5.70939779e-01 8.46105441e-02 1.39570283e-02
-1.47368133e+00 -2.70516038e-01 4.70586866e-01 -3.92758608e-01
6.58254683e-01 3.79662842e-01 3.61945093e-01 8.77172649e-01
6.06752396e-01 8.51502359e-01 8.07840824e-01 -4.72737402e-01
3.60583186e-01 2.47086093e-01 3.26052427e-01 7.87202895e-01
4.20666188e-01 3.56686622e-01 -6.10494375e-01 -5.39207995e-01
9.73286033e-01 3.90540659e-02 -2.57170439e-01 -1.23092689e-01
-8.82960439e-01 9.67186749e-01 6.87726617e-01 3.54561280e-03
-4.76076901e-01 4.05035734e-01 3.08625519e-01 6.96954951e-02
8.06747913e-01 5.65571964e-01 -5.11020303e-01 3.04990172e-01
-7.02970743e-01 3.07642817e-01 6.41132295e-01 5.24534583e-01
4.56599593e-01 -1.00867927e-01 -2.66452402e-01 1.07300496e+00
6.84458137e-01 5.78822605e-02 1.35256663e-01 -8.16806436e-01
1.91126838e-01 5.02066791e-01 1.69616903e-03 -1.01368642e+00
-6.82321072e-01 -6.88095987e-01 -1.17675626e+00 1.53384551e-01
5.87967038e-01 -1.48465127e-01 -8.32045913e-01 1.89992929e+00
1.27705589e-01 1.50053231e-02 -8.01090002e-02 7.24050105e-01
6.74334884e-01 4.99992132e-01 1.74253672e-01 -3.90403979e-02
1.44588506e+00 -5.42578340e-01 -6.81234598e-01 -1.73718750e-01
4.60516751e-01 -1.45755649e-01 7.35051692e-01 5.41901767e-01
-9.33431208e-01 -3.49590629e-01 -1.21097469e+00 -1.96720764e-01
-4.30868343e-02 2.35149160e-01 6.87274873e-01 3.83524925e-01
-1.02152658e+00 9.35898602e-01 -1.02453530e+00 2.95606524e-01
6.40821874e-01 7.43323684e-01 -1.86061069e-01 1.93923607e-01
-1.35459864e+00 8.36154401e-01 4.92426038e-01 5.66155612e-01
-5.98119080e-01 -7.97177911e-01 -8.33862484e-01 3.29832345e-01
4.39216524e-01 -9.42867935e-01 1.09846044e+00 -1.16302586e+00
-1.40537548e+00 2.04776123e-01 -2.19189659e-01 -7.28644490e-01
6.68864250e-01 -3.89221340e-01 9.86538902e-02 -6.41153827e-02
-2.06135392e-01 8.96103203e-01 8.89730871e-01 -1.12493670e+00
-4.03627217e-01 -4.60195422e-01 1.09977578e-03 1.02343857e-01
-1.58264875e-01 -4.10741925e-01 -3.14552218e-01 -8.41875792e-01
2.41306067e-01 -9.31987524e-01 -4.56162781e-01 4.05987561e-01
-7.88929820e-01 -3.10775846e-01 2.21859574e-01 -6.28982365e-01
1.09755456e+00 -1.95704317e+00 1.17421918e-01 4.45893228e-01
7.78986037e-01 -8.39151889e-02 1.22048810e-01 1.15949191e-01
-2.07504272e-01 2.79822469e-01 -4.03066456e-01 -3.90694499e-01
2.22089346e-02 2.52092749e-01 -2.65831530e-01 3.53485197e-01
5.96618414e-01 9.80829418e-01 -6.12730384e-01 -2.67992437e-01
-6.88591823e-02 6.67600870e-01 -6.61090910e-01 4.89826035e-03
-1.29479751e-01 2.64198452e-01 -5.54503918e-01 3.33689809e-01
6.67623043e-01 -5.04892766e-01 2.60931402e-01 -2.62183070e-01
2.59275824e-01 2.42272884e-01 -9.79562819e-01 7.87602603e-01
-3.99957806e-01 6.93852007e-01 -3.55875134e-01 -9.81121838e-01
8.88774335e-01 4.99936678e-02 1.73481941e-01 -2.69641310e-01
7.62049854e-02 2.29560897e-01 2.08027035e-01 -1.71471581e-01
1.62552416e-01 -1.76284879e-01 1.98179185e-02 3.22881311e-01
-1.85248200e-02 2.54044116e-01 -2.51870245e-01 -6.63780943e-02
9.51536596e-01 -1.30591560e-02 7.67698765e-01 -5.85332215e-01
2.24873379e-01 -5.06489038e-01 6.36602581e-01 1.05708826e+00
2.89932609e-01 4.80853945e-01 9.64199007e-01 -6.67278469e-01
-7.15828180e-01 -1.12040949e+00 -5.02319813e-01 7.38639355e-01
-1.60160735e-01 -8.16797167e-02 -6.95613086e-01 -6.89662576e-01
2.61210024e-01 1.02990067e+00 -1.12660205e+00 -4.43618000e-01
-2.30194882e-01 -7.14727521e-01 5.30008435e-01 8.40304673e-01
1.07265808e-01 -6.91733360e-01 -5.74647665e-01 1.12105429e-01
-1.41681328e-01 -5.98429143e-01 -3.46188128e-01 9.28771436e-01
-9.98764396e-01 -8.15855145e-01 -4.87966716e-01 -8.07903036e-02
8.29905629e-01 -2.76051879e-01 9.79647756e-01 -2.13795546e-02
-1.25093937e-01 -1.22085594e-01 -1.15491845e-01 -6.52281284e-01
-4.51181531e-01 -1.40374675e-01 2.48750195e-01 7.99040601e-04
2.31706783e-01 -4.57852453e-01 -5.84187984e-01 2.41472200e-01
-9.02769208e-01 2.37531930e-01 8.39775085e-01 1.31194055e+00
5.26183844e-01 -1.56810075e-01 7.83846021e-01 -9.93862987e-01
6.45745218e-01 -5.15385509e-01 -6.77881002e-01 3.22297901e-01
-6.59382105e-01 4.61906135e-01 6.74309552e-01 -3.82474214e-01
-1.10032475e+00 3.80315334e-02 -9.23745558e-02 -2.79768914e-01
1.55424044e-01 8.19636762e-01 -2.12631375e-02 1.63498521e-01
7.67429709e-01 -1.43672347e-01 1.33979365e-01 -4.70416099e-01
2.50106126e-01 5.89127183e-01 5.14502943e-01 -5.85876465e-01
5.22579432e-01 1.52626619e-01 3.42736363e-01 -6.15275979e-01
-1.01738024e+00 6.86351284e-02 -6.09016836e-01 -3.72388326e-02
7.06890643e-01 -6.78420722e-01 -1.00261319e+00 1.26210958e-01
-1.31534517e+00 -1.63460746e-01 -1.18122719e-01 5.05668998e-01
-4.03965473e-01 1.18799865e-01 -4.90841508e-01 -1.03758454e+00
-1.99904799e-01 -1.34341764e+00 1.06179500e+00 1.68142974e-01
-5.75527489e-01 -1.24455404e+00 -3.67530435e-01 8.50503594e-02
6.21118434e-02 2.77917176e-01 1.39931452e+00 -8.74919236e-01
-4.53338921e-01 -1.69256896e-01 -4.67882425e-01 4.73097414e-01
7.41102174e-02 5.98132461e-02 -1.25084317e+00 1.53601557e-01
-1.20859053e-02 -2.44828939e-01 1.08420002e+00 1.15795445e+00
1.59802091e+00 -6.45918608e-01 -4.77359384e-01 6.09700382e-01
9.57848132e-01 2.26307049e-01 5.55413902e-01 -1.77944332e-01
5.07413328e-01 7.92991340e-01 4.27603245e-01 6.23560429e-01
7.36259595e-02 6.19026840e-01 4.93928939e-01 -1.39776647e-01
4.50928301e-01 -1.87411964e-01 -1.90009680e-02 4.50844646e-01
-1.01997685e-02 -4.51521814e-01 -8.01994920e-01 1.77495360e-01
-2.20471692e+00 -4.74456191e-01 -8.31140652e-02 2.20201445e+00
8.43071818e-01 2.59856075e-01 -1.58569202e-01 -1.62950814e-01
5.49772680e-01 -3.29483628e-01 -9.97344494e-01 -1.87050015e-01
9.01000649e-02 -3.39917690e-02 4.24158067e-01 4.25443262e-01
-1.04450953e+00 2.32277825e-01 6.71351957e+00 1.11444294e+00
-7.28996456e-01 -5.52237183e-02 1.58544981e+00 9.48051214e-02
-4.49743390e-01 -2.13900849e-01 -6.46854877e-01 2.40801156e-01
8.48432362e-01 -9.71182808e-02 1.61433667e-01 7.36041546e-01
2.96909362e-01 -1.81038186e-01 -1.50739861e+00 7.45987713e-01
-3.91928047e-01 -1.49080312e+00 1.76491350e-01 2.45961025e-01
4.13052589e-01 -1.31381691e-01 2.91544586e-01 1.85632482e-01
3.67083073e-01 -1.37922919e+00 7.41171062e-01 7.60111511e-01
7.06664145e-01 -1.07034707e+00 8.85310769e-01 3.66811126e-01
-7.74958313e-01 -1.70410573e-01 -4.08140808e-01 8.18569586e-02
-9.20950025e-02 1.00992227e+00 -9.58911657e-01 5.13707936e-01
5.43258429e-01 3.97182465e-01 -3.18592757e-01 8.30189347e-01
-1.78656913e-02 7.54615784e-01 -5.33141434e-01 -2.10553512e-01
1.86823308e-01 -3.11888933e-01 4.71035838e-01 9.42988455e-01
1.98337018e-01 -6.30834997e-02 -2.68527716e-01 1.43793726e+00
-7.15932548e-02 -1.57628596e-01 -3.77415299e-01 -5.71707077e-02
4.41945285e-01 1.08785379e+00 -8.87683630e-01 -5.50986826e-02
1.05622314e-01 5.68414032e-01 4.73538637e-01 3.62759233e-01
-9.23519373e-01 -3.14922631e-01 5.43367743e-01 -1.56038269e-01
3.68529670e-02 2.78518826e-01 -6.09325707e-01 -7.39095151e-01
9.33203921e-02 -8.37786138e-01 3.17632854e-01 -5.99017859e-01
-1.49985218e+00 7.09621787e-01 2.90921241e-01 -1.22770321e+00
-5.38359106e-01 -8.47082496e-01 -3.71921629e-01 8.56911182e-01
-9.98550236e-01 -8.78615797e-01 1.03125907e-01 -7.99256787e-02
2.50839502e-01 9.46360081e-02 7.98165262e-01 4.86634672e-02
-5.70427716e-01 8.51918817e-01 3.51711184e-01 -1.01138599e-01
5.22266746e-01 -1.32570422e+00 2.55617917e-01 3.30743164e-01
-5.26902676e-02 9.44370329e-01 9.16140079e-01 -6.15318954e-01
-1.00054884e+00 -1.03221178e+00 6.74396157e-01 -2.89803863e-01
4.94758487e-01 -3.17117989e-01 -9.37537849e-01 5.26432276e-01
-3.23716700e-01 4.80216974e-03 7.73305237e-01 4.29392397e-01
-2.53462851e-01 -1.73979372e-01 -1.12763059e+00 5.88350952e-01
6.80196583e-01 -2.21136093e-01 -2.50154823e-01 2.86565214e-01
8.06939662e-01 -2.96365738e-01 -7.70285904e-01 6.47610068e-01
9.00549889e-01 -1.03632045e+00 7.74253309e-01 -6.07263982e-01
8.26641917e-01 -8.82927096e-04 -7.43703023e-02 -1.47830164e+00
-1.58462390e-01 -4.83884633e-01 -2.63175875e-01 9.02888894e-01
7.18076825e-01 -8.52970362e-01 3.32463682e-01 8.30911815e-01
-1.11166432e-01 -1.34907985e+00 -9.76042926e-01 -5.05238831e-01
-7.21230581e-02 -6.21093810e-01 5.01756132e-01 6.46402717e-01
-8.08674991e-02 2.54664451e-01 -3.96709740e-01 4.97180857e-02
5.42521834e-01 -4.46204066e-01 1.63436204e-01 -1.54871356e+00
-7.64125288e-01 -5.88004231e-01 -2.70691723e-01 -1.21220994e+00
8.53232890e-02 -6.52002394e-01 2.84767747e-01 -1.09633207e+00
3.75535637e-01 -4.48659509e-01 -3.51659924e-01 6.64430797e-01
-3.14540178e-01 5.98354489e-02 -1.48673221e-01 6.18733428e-02
-2.93672562e-01 6.26305759e-01 7.92480230e-01 8.23550392e-03
-9.06756744e-02 2.88123161e-01 -1.00188899e+00 1.01271605e+00
3.91051024e-01 -6.29767776e-01 -4.99123394e-01 -1.73998386e-01
5.18800497e-01 3.19605857e-01 5.07351398e-01 -5.40539861e-01
-4.44438308e-02 -1.15069775e-02 7.38218546e-01 -6.61491096e-01
2.67190874e-01 -6.84759855e-01 2.13467091e-01 5.48347175e-01
-8.42592120e-01 -2.22777054e-01 3.51661980e-01 7.26158440e-01
-1.20315626e-02 -3.90658140e-01 7.33625233e-01 3.27626407e-01
6.94881901e-02 2.38073394e-02 -4.01530534e-01 -3.53754818e-01
5.76411903e-01 2.38292161e-02 -1.32548764e-01 -6.59864545e-01
-6.42875195e-01 1.59406155e-01 -2.54910052e-01 1.32854953e-01
6.31784379e-01 -1.22651803e+00 -7.09011555e-01 6.94262013e-02
-1.81472488e-02 1.45092815e-01 2.43715212e-01 9.80651617e-01
-4.50897068e-01 3.97326052e-01 1.57413408e-01 -8.85314286e-01
-1.05985844e+00 5.70235252e-02 4.44339663e-01 -5.09893537e-01
-3.68993938e-01 9.58822072e-01 7.60794103e-01 -3.22869986e-01
4.14809048e-01 -5.05652130e-01 -1.59026057e-01 -3.52128595e-02
6.50361598e-01 2.90437520e-01 -7.66259879e-02 -1.67448789e-01
-2.79831916e-01 7.37186074e-02 -3.30476731e-01 -2.82310825e-02
1.55344045e+00 9.13338885e-02 -2.53018856e-01 5.75233400e-01
1.18896973e+00 -3.68689150e-01 -1.49208319e+00 -2.36322165e-01
6.89709634e-02 -7.92546943e-02 2.55053461e-01 -9.60359097e-01
-7.45993078e-01 7.31315017e-01 5.48009932e-01 1.85413018e-01
1.10210228e+00 9.64601189e-02 1.94144025e-01 7.12658703e-01
-4.41782475e-02 -7.47669041e-01 -1.33308604e-01 3.22677612e-01
1.03416407e+00 -1.03757691e+00 8.05170238e-02 -3.82855326e-01
-5.46596169e-01 1.41221976e+00 3.73902589e-01 -3.89877073e-02
7.73582160e-01 4.33309644e-01 -1.24440961e-01 -1.17803030e-01
-1.03493905e+00 3.33329767e-01 9.66070116e-01 2.98523575e-01
5.32232523e-01 8.59738737e-02 -1.71452731e-01 1.12621605e+00
-4.04466130e-02 -8.66782218e-02 1.05836868e-01 2.81299472e-01
-1.70902759e-01 -6.34221673e-01 -2.30427876e-01 9.34096158e-01
-4.71776545e-01 -2.50182480e-01 -4.19248611e-01 7.21190214e-01
-1.02978036e-01 6.28824711e-01 1.74566656e-01 -4.49424982e-01
-6.93495795e-02 -4.21265960e-02 2.66336054e-01 -2.97166824e-01
-3.22377682e-01 2.88689792e-01 1.86965525e-01 -4.52986598e-01
5.54360487e-02 -5.65177143e-01 -1.04108655e+00 3.08776479e-02
-6.32328868e-01 1.11228421e-01 5.67227840e-01 1.28393948e+00
5.02390504e-01 8.31406116e-01 4.37514871e-01 -9.04371917e-01
-1.15834033e+00 -1.03786218e+00 -5.79510391e-01 -1.49645448e-01
5.06879628e-01 -9.03351367e-01 -4.30855632e-01 -2.31630906e-01] | [8.485803604125977, 5.4185051918029785] |
f84c145d-af29-4bec-bb25-0d8affbac055 | chatgpt-powered-conversational-drug-editing | 2305.18090 | null | https://arxiv.org/abs/2305.18090v1 | https://arxiv.org/pdf/2305.18090v1.pdf | ChatGPT-powered Conversational Drug Editing Using Retrieval and Domain Feedback | Recent advancements in conversational large language models (LLMs), such as ChatGPT, have demonstrated remarkable promise in various domains, including drug discovery. However, existing works mainly focus on investigating the capabilities of conversational LLMs on chemical reaction and retrosynthesis. While drug editing, a critical task in the drug discovery pipeline, remains largely unexplored. To bridge this gap, we propose ChatDrug, a framework to facilitate the systematic investigation of drug editing using LLMs. ChatDrug jointly leverages a prompt module, a retrieval and domain feedback (ReDF) module, and a conversation module to streamline effective drug editing. We empirically show that ChatDrug reaches the best performance on 33 out of 39 drug editing tasks, encompassing small molecules, peptides, and proteins. We further demonstrate, through 10 case studies, that ChatDrug can successfully identify the key substructures (e.g., the molecule functional groups, peptide motifs, and protein structures) for manipulation, generating diverse and valid suggestions for drug editing. Promisingly, we also show that ChatDrug can offer insightful explanations from a domain-specific perspective, enhancing interpretability and enabling informed decision-making. This research sheds light on the potential of ChatGPT and conversational LLMs for drug editing. It paves the way for a more efficient and collaborative drug discovery pipeline, contributing to the advancement of pharmaceutical research and development. | ['Chaowei Xiao', 'Hongyu Guo', 'Ling Liu', 'Chengpeng Wang', 'Yijin Yang', 'Jiongxiao Wang', 'Shengchao Liu'] | 2023-05-29 | null | null | null | null | ['drug-discovery', 'retrosynthesis'] | ['medical', 'medical'] | [ 4.30210799e-01 1.66063905e-01 -5.43731987e-01 -6.18118495e-02
-8.53785634e-01 -9.29800630e-01 5.18911541e-01 6.58599794e-01
9.27524865e-02 9.69782650e-01 4.90859449e-01 -8.54131699e-01
4.32887264e-02 -4.49880898e-01 -7.13324189e-01 -6.96767330e-01
1.47440732e-01 3.90763700e-01 -3.29541653e-01 -1.89110830e-01
4.89514947e-01 6.19025648e-01 -7.79230475e-01 7.80509233e-01
1.13632917e+00 3.10415421e-02 2.77851224e-01 2.45588273e-01
-3.40310335e-01 7.47105658e-01 -6.02817237e-01 -4.64777768e-01
-3.56705874e-01 -6.61080062e-01 -8.58984947e-01 -2.43807048e-01
-1.96776390e-01 -2.16882542e-01 1.63774922e-01 7.44216859e-01
7.01369226e-01 -9.07004848e-02 4.33627546e-01 -6.25720382e-01
-6.82052255e-01 5.59009373e-01 -2.20037460e-01 -4.79094200e-02
7.99153507e-01 3.96487296e-01 1.06253946e+00 -1.15450215e+00
1.07726645e+00 1.27579749e+00 3.43848884e-01 7.61033475e-01
-1.03829503e+00 -7.28983819e-01 1.10098533e-01 1.94359854e-01
-1.06067467e+00 -4.89416420e-01 2.11276531e-01 -5.78029692e-01
1.33313549e+00 2.14381427e-01 4.97258455e-01 1.10419154e+00
4.67831314e-01 6.87866986e-01 7.13914752e-01 -1.91008911e-01
3.88441145e-01 -4.89390567e-02 -1.38362646e-01 5.18336236e-01
7.53904656e-02 -4.75050621e-02 -7.76947796e-01 -7.06229866e-01
4.53862041e-01 2.65521049e-01 -1.57506406e-01 1.48303017e-01
-1.13412452e+00 9.25774574e-01 8.40132532e-04 2.57164150e-01
-5.28101504e-01 -3.15662801e-01 3.84033144e-01 2.52662361e-01
5.70070744e-01 1.04213166e+00 -6.40350342e-01 -2.17560261e-01
-4.75808144e-01 1.94206193e-01 1.00490558e+00 6.47150099e-01
3.00911188e-01 -4.69290614e-01 -3.29411238e-01 7.14913726e-01
3.23904872e-01 1.17875688e-01 1.47010118e-01 -5.00578403e-01
1.99080110e-01 6.04665101e-01 1.59065485e-01 -6.52712226e-01
-3.78336132e-01 -1.67345211e-01 -2.46424466e-01 -2.21327037e-01
1.18059590e-01 -2.17519060e-01 -3.99530113e-01 1.69123447e+00
6.65094972e-01 1.37027666e-01 2.66238213e-01 6.87081337e-01
9.51532722e-01 6.20782435e-01 6.54004574e-01 -4.79467571e-01
1.41622913e+00 -9.60523069e-01 -7.59270668e-01 2.08805382e-01
1.00078928e+00 -1.14668965e+00 6.26228809e-01 5.50752103e-01
-9.75546837e-01 -2.61184629e-02 -7.73467064e-01 -1.26041606e-01
-3.34011227e-01 3.78020443e-02 9.83945608e-01 3.27648699e-01
-5.19453943e-01 9.51221824e-01 -8.04170310e-01 -1.65413439e-01
5.24826348e-01 4.74636763e-01 -3.49612474e-01 -2.77890116e-01
-1.12151241e+00 9.30317402e-01 -7.70125492e-03 -2.49252357e-02
-1.02309132e+00 -1.21609116e+00 -5.34020960e-01 -7.26336464e-02
4.55198705e-01 -7.75051296e-01 1.38806129e+00 -5.13446748e-01
-1.74076569e+00 3.87618899e-01 -3.86932701e-01 -3.35372329e-01
2.60510951e-01 -1.30520746e-01 -2.18450546e-01 8.18325300e-03
9.50468034e-02 5.19645035e-01 2.75447726e-01 -4.92142916e-01
-1.73977599e-01 -2.60490268e-01 7.25602582e-02 2.26025924e-01
-1.21282712e-01 3.91415119e-01 -1.29816055e-01 -6.15144193e-01
-5.16067624e-01 -8.29531014e-01 -5.78833044e-01 -3.83655727e-02
-4.21923935e-01 -4.62311000e-01 3.81977201e-01 -5.39325416e-01
1.24970424e+00 -1.75031614e+00 3.68319660e-01 1.14173610e-02
4.46500629e-01 6.13799214e-01 -3.37519288e-01 1.30650270e+00
-3.49887460e-01 4.62336570e-01 3.69542427e-02 1.45498201e-01
-2.96896398e-01 -1.98364884e-01 -3.79518926e-01 2.12834910e-01
3.66402626e-01 1.08405304e+00 -1.13824570e+00 -5.08151017e-03
2.24651054e-01 4.93895411e-01 -8.79031837e-01 1.16617925e-01
-7.59706616e-01 8.83820117e-01 -9.11314070e-01 6.48769140e-01
4.58077341e-01 -5.67812562e-01 7.71718919e-01 -1.54938847e-02
-3.16084355e-01 5.80566525e-01 -3.90110701e-01 1.64828229e+00
-3.12939644e-01 2.84982443e-01 -2.86735684e-01 -5.76718748e-01
7.94728994e-01 6.88168824e-01 6.91323638e-01 -3.79094779e-01
-6.86708167e-02 1.53180376e-01 1.71592906e-01 -6.10430360e-01
5.97000197e-02 -2.40385398e-01 2.42943987e-01 6.64149404e-01
-2.84391999e-01 1.28432021e-01 1.04773097e-01 3.17024559e-01
1.25418651e+00 1.32945120e-01 6.09588325e-01 6.33282885e-02
4.84469295e-01 2.27567703e-01 4.23467606e-01 3.45805913e-01
5.25184274e-02 -7.62376711e-02 6.54392898e-01 -2.15710327e-01
-9.36052680e-01 -4.50754791e-01 9.46915336e-03 1.07300258e+00
-1.70605406e-01 -1.09056365e+00 -6.89950168e-01 -6.12317026e-01
-4.60688695e-02 4.78499681e-01 -4.04504031e-01 -1.02496028e-01
-3.66895646e-01 -8.43195796e-01 5.98341703e-01 1.22263126e-01
-7.97532350e-02 -9.16142404e-01 2.41730034e-01 6.01130307e-01
-1.14944624e-03 -1.05299711e+00 -6.80677950e-01 -1.09360769e-01
-7.51315773e-01 -1.43471789e+00 -5.29750526e-01 -6.70212984e-01
3.87774289e-01 3.58265340e-01 6.84766114e-01 -7.06102252e-02
-3.76473218e-01 -1.24162510e-01 -3.00902307e-01 -6.31970823e-01
-9.68574226e-01 -1.07484356e-01 -1.63679779e-01 -5.35034895e-01
3.71554613e-01 -5.72744191e-01 -6.39030576e-01 4.34364438e-01
-7.39932597e-01 2.98020750e-01 5.88951230e-01 7.84142613e-01
5.40256321e-01 -5.99221110e-01 9.54020977e-01 -1.33412433e+00
1.01408434e+00 -7.13488281e-01 -4.43894088e-01 3.52706552e-01
-6.13746524e-01 1.11506179e-01 7.41116941e-01 -4.52643484e-01
-1.05525172e+00 4.34012488e-02 -5.83233237e-01 1.50545627e-01
4.76031080e-02 8.91694188e-01 -1.17296271e-01 -2.31953904e-01
7.07822442e-01 9.10528898e-02 2.07643166e-01 -6.28783524e-01
4.19920862e-01 6.47143185e-01 -1.68146849e-01 -4.93783325e-01
2.01857463e-01 1.27086088e-01 -1.58302546e-01 -7.95140147e-01
-4.22706723e-01 -5.03287375e-01 3.08059100e-02 2.91555017e-01
5.96161664e-01 -1.07238245e+00 -1.21914935e+00 5.27345128e-02
-1.32023811e+00 -1.52200073e-01 1.72956556e-01 4.51835573e-01
-1.79812849e-01 6.90174043e-01 -9.05577540e-01 -4.08549756e-01
-7.32473016e-01 -1.52378106e+00 1.11623049e+00 -4.78909397e-03
-6.63779974e-01 -9.47895765e-01 1.71785891e-01 7.16944158e-01
1.06205389e-01 1.22723110e-01 1.39394307e+00 -1.14889371e+00
-8.38882685e-01 1.19492814e-01 1.20961875e-01 -1.14203572e-01
3.33380222e-01 4.60812338e-02 -6.60474122e-01 3.49341929e-02
-3.81127715e-01 -2.41758913e-01 5.25812566e-01 1.07580170e-01
9.91222441e-01 -3.53837252e-01 -6.90618277e-01 1.29726067e-01
7.43194342e-01 6.43692911e-01 5.17764211e-01 -1.28786966e-01
5.96833110e-01 4.19422865e-01 8.26332748e-01 6.61979795e-01
1.19095504e-01 7.67778099e-01 1.60388932e-01 -1.75399765e-01
9.66606475e-03 -3.24252009e-01 5.52605748e-01 7.00449109e-01
6.97031096e-02 -3.16001743e-01 -8.10209692e-01 8.05135518e-02
-1.74135125e+00 -7.76696980e-01 -1.17878718e-02 1.89545584e+00
1.41697800e+00 -3.79713833e-01 -1.67479441e-01 -5.22582293e-01
5.69530606e-01 -1.54021055e-01 -9.95923102e-01 -5.60578942e-01
5.40366061e-02 5.94489336e-01 6.17194921e-02 5.98043144e-01
-6.62436545e-01 1.27762699e+00 6.58249521e+00 9.60098267e-01
-1.08587718e+00 -1.41912535e-01 4.31858838e-01 1.50823325e-01
-5.37137985e-01 5.12990765e-02 -8.40162694e-01 2.80752748e-01
9.05173481e-01 -4.45956945e-01 4.48925078e-01 7.44218946e-01
9.49746549e-01 1.69873953e-01 -1.39990723e+00 6.93082631e-01
-2.68614531e-01 -2.18156052e+00 3.73922199e-01 2.83868015e-01
5.97780168e-01 -7.25778565e-02 -2.08353639e-01 -2.49780566e-02
3.88492465e-01 -1.08278298e+00 -3.94130610e-02 2.31096134e-01
5.34923017e-01 -6.66747510e-01 4.27677900e-01 3.37455899e-01
-7.30509281e-01 2.30544612e-01 -1.65032167e-02 -6.37769047e-03
-3.10026295e-02 3.94032031e-01 -1.69274485e+00 6.40431762e-01
-2.34479427e-01 1.15780807e+00 5.94882877e-04 9.17201102e-01
-3.01555306e-01 4.84224945e-01 1.97638914e-01 -2.72964478e-01
1.02610856e-01 -1.76679105e-01 4.27850276e-01 1.19407356e+00
-1.13727920e-01 4.31231648e-01 4.39413041e-01 7.81488776e-01
-3.83007020e-01 4.83168274e-01 -2.99177200e-01 -8.95814121e-01
6.54270530e-01 9.72211480e-01 -3.85008603e-01 -2.98786640e-01
-1.64863825e-01 7.90430844e-01 -8.13605338e-02 2.27813318e-01
-7.15674937e-01 -5.85609898e-02 1.13897896e+00 -2.52466887e-01
4.87139784e-02 -1.51947424e-01 -3.10221761e-02 -1.04249191e+00
-4.66706812e-01 -1.50271308e+00 2.89197057e-01 -6.09034240e-01
-1.11267221e+00 1.56485811e-01 -3.22700977e-01 -9.55335617e-01
1.43954922e-02 -3.79956275e-01 -4.49208528e-01 7.39486456e-01
-1.33535075e+00 -8.96406770e-01 3.45974296e-01 2.24016324e-01
1.05882716e+00 -5.84821180e-02 1.11472392e+00 4.09769028e-01
-7.65996516e-01 3.35863322e-01 2.73941368e-01 -3.64683390e-01
1.05793285e+00 -7.39957154e-01 5.85866988e-01 1.81241572e-01
-1.49102435e-01 1.24858201e+00 7.41908133e-01 -8.93878222e-01
-1.77811205e+00 -1.14437222e+00 9.68114913e-01 -4.39148039e-01
7.26569891e-01 -3.04624379e-01 -8.12074482e-01 3.02143782e-01
2.81240940e-02 -7.13964462e-01 1.37174928e+00 1.85007095e-01
-1.94246128e-01 5.44338286e-01 -8.72013092e-01 8.24401557e-01
8.73529971e-01 -7.79390395e-01 -1.91760674e-01 1.25704122e+00
1.00043094e+00 -5.63234091e-01 -1.18086517e+00 6.79053441e-02
5.75924277e-01 -4.25226361e-01 9.82510448e-01 -1.19934011e+00
6.14647686e-01 -1.94086164e-01 3.26376468e-01 -1.21509433e+00
-1.20522045e-01 -1.13359559e+00 4.90117259e-02 8.05992663e-01
7.89735556e-01 -5.18606782e-01 7.28361964e-01 6.13501310e-01
-3.75252366e-01 -1.05540681e+00 -5.52508950e-01 -3.03179920e-01
4.36793789e-02 -2.26118550e-01 5.66605449e-01 9.85668123e-01
5.68395495e-01 7.27384150e-01 -4.30802345e-01 -1.29017130e-01
-1.46666095e-01 1.43028080e-01 6.32419169e-01 -9.08588469e-01
-5.11932611e-01 -2.90597528e-01 6.44063652e-02 -1.09887755e+00
2.53292709e-03 -1.03921497e+00 -2.81646937e-01 -1.43887842e+00
4.78714734e-01 -1.37124389e-01 6.67401701e-02 6.41697109e-01
-1.78757876e-01 -3.38579357e-01 -1.89442430e-02 2.30447873e-01
-6.17153406e-01 3.86369884e-01 1.54143846e+00 -1.91256702e-01
-4.63783383e-01 -5.26962765e-02 -1.22829330e+00 4.94520873e-01
6.29862309e-01 -4.61573273e-01 -3.19914967e-01 6.52223006e-02
2.67815113e-01 2.14925796e-01 -8.84249881e-02 6.21191598e-02
3.08952868e-01 -4.98897612e-01 -2.31924634e-02 -1.59167439e-01
1.95813492e-01 -2.89626658e-01 4.39399242e-01 6.67028308e-01
-6.47769094e-01 -1.96486980e-01 3.36266577e-01 7.88906455e-01
1.58446468e-02 1.54320493e-01 3.42586190e-01 -2.91111857e-01
-3.92947733e-01 4.09210593e-01 -7.26819277e-01 -2.46891364e-01
8.96746695e-01 -6.70400858e-02 -4.96110499e-01 -1.70931607e-01
-8.94109607e-01 1.70812637e-01 1.47879675e-01 5.37312925e-01
5.97738326e-01 -6.96508825e-01 -5.31976521e-01 7.04742745e-02
2.88743615e-01 -4.81593817e-01 2.54893750e-01 8.75382543e-01
-2.84303576e-01 8.68232429e-01 2.86128074e-01 -3.59833211e-01
-1.74192142e+00 3.49407941e-01 2.17872962e-01 -3.46144617e-01
-2.80870050e-01 6.55155241e-01 4.30407524e-01 -2.20994189e-01
1.07151933e-01 -1.76393673e-01 -1.23728819e-01 -1.99006088e-02
8.10472071e-01 1.22274280e-01 2.38282859e-01 2.45109033e-02
-4.92724955e-01 2.72769868e-01 -5.56992233e-01 4.27016497e-01
1.46144235e+00 1.98920548e-01 -1.96547925e-01 -2.15226218e-01
9.60990787e-01 1.98745579e-01 -8.16839695e-01 -2.26998240e-01
1.35990083e-01 -2.09401339e-01 -3.05190891e-01 -1.24623883e+00
-3.33349198e-01 7.12500274e-01 1.57273874e-01 -3.20422918e-01
5.88234365e-01 2.70840190e-02 8.78903151e-01 6.54082000e-01
2.36814469e-01 -6.51808321e-01 1.99115366e-01 4.24964488e-01
8.90910685e-01 -1.07902443e+00 1.73406556e-01 -6.23425126e-01
-7.00699687e-01 1.27127731e+00 1.63297489e-01 5.65784037e-01
2.82447070e-01 8.95906389e-02 -2.22470582e-01 -4.63604301e-01
-1.24654210e+00 2.46108636e-01 4.76249978e-02 2.90894270e-01
9.98924911e-01 1.71882249e-02 -5.68397403e-01 6.84283733e-01
3.89776468e-01 3.94423753e-01 2.83872753e-01 8.66720319e-01
-4.24690574e-01 -1.83752096e+00 3.77566874e-04 4.32392843e-02
-6.16850317e-01 -6.61208689e-01 -1.01584482e+00 3.74742448e-01
6.20525666e-02 1.17292750e+00 -7.83651412e-01 -1.47877201e-01
2.67279506e-01 2.18333751e-02 4.24915344e-01 -1.03855205e+00
-9.85709071e-01 3.10835063e-01 5.60901165e-01 -5.21884382e-01
-3.42433244e-01 -4.53153372e-01 -1.32196224e+00 -5.74367166e-01
-5.59048057e-01 5.86705625e-01 7.12260604e-01 1.11271501e+00
1.09320152e+00 4.36758012e-01 5.15186667e-01 -3.16976815e-01
-3.56407791e-01 -6.00838959e-01 -5.36581762e-02 -8.27469602e-02
4.31447178e-02 -3.45432967e-01 3.17898870e-01 9.62283015e-02] | [4.962607383728027, 5.858999252319336] |
a0654eee-aed5-4d72-b18f-ecf968e5e03b | hardvs-revisiting-human-activity-recognition | 2211.09648 | null | https://arxiv.org/abs/2211.09648v1 | https://arxiv.org/pdf/2211.09648v1.pdf | HARDVS: Revisiting Human Activity Recognition with Dynamic Vision Sensors | The main streams of human activity recognition (HAR) algorithms are developed based on RGB cameras which are suffered from illumination, fast motion, privacy-preserving, and large energy consumption. Meanwhile, the biologically inspired event cameras attracted great interest due to their unique features, such as high dynamic range, dense temporal but sparse spatial resolution, low latency, low power, etc. As it is a newly arising sensor, even there is no realistic large-scale dataset for HAR. Considering its great practical value, in this paper, we propose a large-scale benchmark dataset to bridge this gap, termed HARDVS, which contains 300 categories and more than 100K event sequences. We evaluate and report the performance of multiple popular HAR algorithms, which provide extensive baselines for future works to compare. More importantly, we propose a novel spatial-temporal feature learning and fusion framework, termed ESTF, for event stream based human activity recognition. It first projects the event streams into spatial and temporal embeddings using StemNet, then, encodes and fuses the dual-view representations using Transformer networks. Finally, the dual features are concatenated and fed into a classification head for activity prediction. Extensive experiments on multiple datasets fully validated the effectiveness of our model. Both the dataset and source code will be released on \url{https://github.com/Event-AHU/HARDVS}. | ['Yonghong Tian', 'YaoWei Wang', 'Guoqi Li', 'Lin Zhu', 'Zhimin Bao', 'Bo Jiang', 'Zongzhen Wu', 'Xiao Wang'] | 2022-11-17 | null | null | null | null | ['activity-prediction', 'activity-prediction'] | ['computer-vision', 'time-series'] | [ 8.49651322e-02 -5.39566100e-01 -1.25792980e-01 -2.29447514e-01
-4.05818880e-01 -3.06741863e-01 6.29244626e-01 3.52603570e-02
-4.49732125e-01 7.92561531e-01 7.00428963e-01 3.98620039e-01
-1.14959002e-01 -6.49118066e-01 -6.42982841e-01 -8.64403546e-01
-7.20254406e-02 -2.66712606e-01 3.41480315e-01 1.44424528e-01
-8.23411644e-02 4.03335452e-01 -1.69067085e+00 1.69455096e-01
3.39800507e-01 1.41898656e+00 -3.46980579e-02 3.69699955e-01
3.45939726e-01 1.00394642e+00 -2.96646446e-01 -1.84128150e-01
2.52947062e-02 -5.10566294e-01 -4.01320457e-01 -1.79627433e-01
-1.60793170e-01 -4.22547936e-01 -1.04251742e+00 5.86445749e-01
7.49528110e-01 3.04538608e-01 2.15570912e-01 -1.70348942e+00
-6.32406771e-01 2.76092470e-01 -3.63682300e-01 4.88807410e-01
7.54656196e-01 3.41230154e-01 6.13070488e-01 -8.80366802e-01
5.30097187e-01 8.11706960e-01 5.54001093e-01 5.84049582e-01
-8.06623042e-01 -8.94950092e-01 7.78395087e-02 6.72033131e-01
-1.46838951e+00 -6.71552658e-01 7.76664734e-01 -2.74387479e-01
1.05816758e+00 8.69361237e-02 1.17593563e+00 1.80844831e+00
2.32276425e-01 1.04498005e+00 8.79850805e-01 2.51007855e-01
3.09502810e-01 -3.01557511e-01 -1.39107943e-01 5.75940311e-01
3.65841180e-01 1.22927986e-01 -1.07825506e+00 -1.32171333e-01
6.65750027e-01 7.27365732e-01 -6.17228925e-01 -4.65139538e-01
-1.75521934e+00 5.02155900e-01 4.77753520e-01 1.53787255e-01
-4.37556654e-01 2.54472911e-01 6.06694818e-01 -1.61594555e-01
7.36160725e-02 -1.26725554e-01 -2.90959537e-01 -5.72732449e-01
-5.42901516e-01 1.02621950e-01 5.77085137e-01 8.59104991e-01
3.32160145e-01 -1.09231897e-01 -2.27539182e-01 6.78618193e-01
3.57767582e-01 5.19792974e-01 8.50958109e-01 -8.23011994e-01
3.22539330e-01 7.37999201e-01 -1.13252074e-01 -1.16422999e+00
-4.69384998e-01 -5.59646823e-02 -1.16642606e+00 -3.20207775e-01
5.60324043e-02 2.97126845e-02 -5.01156628e-01 1.77088976e+00
3.32071155e-01 8.02819431e-01 1.79668009e-01 1.14900136e+00
8.93467784e-01 8.43125284e-01 1.67570338e-01 -3.23009491e-01
1.45457828e+00 -8.86335850e-01 -6.81149483e-01 -7.92076513e-02
3.28857630e-01 -1.51585370e-01 7.53508031e-01 4.18994725e-01
-8.84115338e-01 -4.39383864e-01 -1.28196812e+00 -1.32142663e-01
-3.92047703e-01 -6.36511296e-02 6.87784970e-01 2.50283331e-01
-5.77559412e-01 4.89980072e-01 -1.18342018e+00 -7.79747069e-01
8.51906121e-01 2.13307530e-01 -5.56686044e-01 -1.68622747e-01
-1.25231612e+00 3.99023622e-01 5.14334559e-01 6.86529204e-02
-1.00097549e+00 -3.77732277e-01 -1.00904214e+00 -5.79228178e-02
2.12095782e-01 -5.72600842e-01 9.59195912e-01 -4.97969478e-01
-1.40164912e+00 5.13643682e-01 -1.19045258e-01 -5.55517316e-01
3.74991357e-01 -1.63187444e-01 -8.50559831e-01 2.55314916e-01
-3.68013531e-02 4.98878032e-01 4.27339733e-01 -6.09026134e-01
-7.46037483e-01 -5.58348417e-01 -1.53887272e-01 1.38584077e-01
-7.51262367e-01 -1.19546421e-01 -5.53038478e-01 -7.79449284e-01
-1.07971162e-01 -6.48337185e-01 3.09104063e-02 1.03580952e-01
-5.94688766e-02 -3.09988558e-01 7.58101881e-01 -4.86833930e-01
1.31781495e+00 -2.42075682e+00 1.90606907e-01 -2.20589459e-01
2.55494118e-01 -5.68479905e-03 2.37222239e-02 3.80324811e-01
1.12536848e-01 -2.35769331e-01 -2.43308663e-01 -9.38801691e-02
-5.71751334e-02 2.30767190e-01 -1.19641952e-01 7.73959160e-01
-1.19305793e-02 1.14403677e+00 -9.63483751e-01 -5.71704924e-01
4.96525198e-01 7.41926968e-01 -2.00169876e-01 1.79437026e-01
1.86967015e-01 4.01753247e-01 -5.25222123e-01 8.80267620e-01
3.57918888e-01 -3.82652938e-01 -1.03910148e-01 -3.90733898e-01
-5.30915223e-02 -6.82279840e-02 -1.20250010e+00 2.19719481e+00
-1.17999449e-01 4.81337845e-01 -4.60859388e-01 -9.72747147e-01
6.78739250e-01 3.44758213e-01 9.35319960e-01 -1.00322914e+00
4.31883603e-01 2.11741570e-02 -3.67937297e-01 -5.31887352e-01
1.96396321e-01 2.03872576e-01 -2.67290741e-01 2.63414472e-01
3.27458382e-01 7.73752987e-01 1.71246216e-01 2.03472719e-01
1.45450139e+00 3.53342682e-01 4.54889446e-01 1.68834597e-01
5.64254999e-01 -2.60650128e-01 8.89838576e-01 1.40982643e-01
-6.54330313e-01 7.22243369e-01 1.97206631e-01 -4.63640124e-01
-6.01818442e-01 -1.27598369e+00 -6.48849756e-02 9.00374234e-01
5.83743572e-01 -6.16434336e-01 -5.12363434e-01 -4.61973518e-01
-7.22145066e-02 2.32913300e-01 -6.39681756e-01 -3.98110420e-01
-5.61175942e-01 -8.47469270e-01 7.59595871e-01 9.59963262e-01
8.29609632e-01 -1.23630047e+00 -1.03483379e+00 3.57963741e-01
-3.73059124e-01 -1.28261840e+00 -4.27795202e-01 1.97460689e-02
-6.86034262e-01 -1.25584877e+00 -6.21457875e-01 -6.36072695e-01
9.15668979e-02 3.48189980e-01 7.48172998e-01 -2.84446657e-01
-4.71046388e-01 5.28851688e-01 -5.19932508e-01 -4.27421957e-01
5.47767103e-01 -2.12101877e-01 4.02068585e-01 2.91374803e-01
6.48586273e-01 -8.91222179e-01 -1.03109145e+00 3.23559314e-01
-8.70872378e-01 -5.67652546e-02 5.80861032e-01 6.23108625e-01
9.07398283e-01 -1.49718568e-01 4.36119109e-01 -9.45610628e-02
3.31758827e-01 -8.32787216e-01 -2.83032417e-01 1.27299815e-01
-2.75462270e-01 -2.51914859e-01 6.44192338e-01 -5.08453250e-01
-8.58916879e-01 2.61980802e-01 8.75425488e-02 -6.15148365e-01
-3.15916747e-01 2.23864824e-01 -4.82400328e-01 1.50819853e-01
4.62466806e-01 6.67698085e-01 -1.60529599e-01 -3.11040759e-01
1.23584040e-01 6.06879056e-01 6.97856843e-01 -2.98015863e-01
5.61978340e-01 7.37752140e-01 -2.56543696e-01 -7.61562645e-01
-5.34680247e-01 -4.60956573e-01 -3.25477928e-01 -2.16904551e-01
1.01642203e+00 -1.20965123e+00 -8.05846512e-01 7.78739691e-01
-9.83703077e-01 -6.99209869e-02 -3.67152721e-01 7.29565620e-01
-6.48608327e-01 3.70116621e-01 -5.13134539e-01 -6.59290493e-01
-3.44001323e-01 -8.58737469e-01 8.86432111e-01 5.93422771e-01
-1.88080538e-02 -6.26390994e-01 2.34691948e-01 3.84885758e-01
2.42532253e-01 5.84279835e-01 1.59595877e-01 -5.80240667e-01
-5.89844406e-01 -9.83952638e-03 -1.77098706e-01 5.13595715e-02
1.05716966e-01 -3.74470234e-01 -1.07711148e+00 -2.58059114e-01
-1.02544121e-01 -4.56386387e-01 8.33108544e-01 2.71058112e-01
1.48334908e+00 3.89954308e-04 -5.52013934e-01 9.39476788e-01
1.30166984e+00 3.09317440e-01 7.49699473e-01 3.68095517e-01
7.06280112e-01 1.44169807e-01 4.63268340e-01 9.26717460e-01
6.33427501e-01 5.60358226e-01 3.97325307e-01 2.73218572e-01
3.88871916e-02 -4.55853283e-01 6.74813867e-01 1.03268790e+00
-1.24393985e-01 -4.85174656e-01 -7.00386405e-01 6.54649734e-01
-2.09558249e+00 -1.27792823e+00 2.09307775e-01 2.00650024e+00
4.57606852e-01 -1.14924386e-01 2.96185821e-01 3.09951872e-01
4.66502875e-01 4.71588731e-01 -9.25308168e-01 9.52694193e-02
-2.60405898e-01 1.22525148e-01 3.69648576e-01 -4.04871941e-01
-1.32166910e+00 4.69003111e-01 5.02841091e+00 5.34738600e-01
-8.83438349e-01 3.21560651e-01 3.08168441e-01 -6.13632023e-01
1.78553343e-01 -4.03466582e-01 -5.98735452e-01 8.91152859e-01
1.07632983e+00 -3.15761775e-01 4.42267239e-01 7.38535464e-01
1.67599544e-01 8.29100683e-02 -1.13880229e+00 1.65925348e+00
3.96872312e-01 -1.23093605e+00 -3.68554369e-02 1.00864813e-01
1.77795798e-01 1.89693302e-01 -2.88981587e-01 2.78450161e-01
4.48195226e-02 -9.23538506e-01 6.60516798e-01 6.31404161e-01
6.78975403e-01 -7.81837225e-01 6.90770447e-01 2.64911473e-01
-1.70182228e+00 -2.78359145e-01 -2.50999451e-01 -9.96661335e-02
3.01482797e-01 3.50812227e-01 9.11084488e-02 5.58122516e-01
1.20488226e+00 1.43813217e+00 -5.84357858e-01 1.09042370e+00
5.24779633e-02 4.02748764e-01 -3.93810958e-01 -1.07914411e-01
-7.95967579e-02 -4.45051631e-03 2.26482242e-01 1.10139048e+00
5.92561364e-01 5.31337738e-01 2.27377400e-01 4.66679126e-01
-2.93171942e-01 -7.46481419e-02 -7.67933249e-01 -1.89753979e-01
5.45127749e-01 1.25048494e+00 -5.04926682e-01 -2.26021603e-01
-7.11391926e-01 1.25947535e+00 1.83058918e-01 1.97805688e-01
-1.23380554e+00 -4.66431886e-01 9.60219920e-01 -1.72159985e-01
2.37094820e-01 -1.10438339e-01 1.57900542e-01 -1.62225270e+00
3.04956675e-01 -5.32533407e-01 6.59638882e-01 -7.21961558e-01
-1.29940724e+00 3.09792757e-01 -7.37802982e-02 -1.43450963e+00
1.88582957e-01 -4.57147956e-01 -3.74344081e-01 1.45119950e-01
-1.36058712e+00 -1.06189144e+00 -8.61251831e-01 1.13466942e+00
6.16417289e-01 -2.16779873e-01 6.35452330e-01 5.90694070e-01
-9.00108397e-01 5.60444534e-01 8.07611365e-03 4.03537631e-01
5.90962172e-01 -9.00142610e-01 1.63238719e-01 7.76230216e-01
1.91539899e-01 2.99646914e-01 9.27012116e-02 -2.93775856e-01
-1.91238558e+00 -1.26791179e+00 4.96398658e-01 -5.24309039e-01
5.59956074e-01 -5.20428836e-01 -7.84182668e-01 7.45240748e-01
1.67868629e-01 5.48835576e-01 9.05724883e-01 -4.45291668e-01
-1.88957900e-01 -3.41553479e-01 -1.03210938e+00 4.01614189e-01
1.61103129e+00 -3.18168849e-01 -6.60819411e-01 1.52206793e-01
6.33993328e-01 -1.82692185e-01 -1.09982181e+00 3.43332827e-01
7.54316032e-01 -1.01194990e+00 9.98293221e-01 -3.70873868e-01
2.75291830e-01 -4.92377341e-01 -3.79548550e-01 -9.35674369e-01
-5.14842510e-01 -2.72473603e-01 -6.55755222e-01 1.22829008e+00
-1.42891929e-01 -6.73158526e-01 7.52547860e-01 1.85721472e-01
-6.58828160e-03 -7.92865574e-01 -1.13683295e+00 -7.27121532e-01
-5.43981910e-01 -5.22265077e-01 6.91718698e-01 8.83696198e-01
1.08241670e-01 2.22176835e-01 -5.49393594e-01 8.46246257e-02
7.49026120e-01 9.83851627e-02 5.44329464e-01 -1.16157484e+00
-4.64058146e-02 -1.18923545e-01 -9.36448276e-01 -9.15991485e-01
-1.70166224e-01 -7.95376003e-01 -4.64145280e-02 -1.45665038e+00
3.79746825e-01 6.38078228e-02 -9.45593119e-01 5.21046102e-01
1.76136047e-01 3.31503451e-01 3.82351875e-02 1.84903130e-01
-1.14677203e+00 1.02570868e+00 7.22470403e-01 -8.05054978e-02
-1.05625197e-01 -5.60861051e-01 -5.85270345e-01 7.67228365e-01
8.82335663e-01 -3.17537546e-01 -4.21207249e-01 -4.60195005e-01
-1.06395006e-01 9.73665863e-02 5.98803520e-01 -1.72774100e+00
4.54168439e-01 -2.04241470e-01 8.68430078e-01 -3.96411568e-01
5.60034275e-01 -9.03564095e-01 2.93845952e-01 4.57973957e-01
-5.65024912e-02 1.59501299e-01 -3.10937339e-03 1.01583719e+00
-2.92654127e-01 5.07169187e-01 6.31700158e-01 1.47134752e-03
-1.14209211e+00 7.72969007e-01 -2.77624279e-01 9.21403468e-02
1.51047301e+00 -4.06453907e-01 -1.44766495e-01 -1.14146851e-01
-5.21118462e-01 4.00780916e-01 4.92119402e-01 7.10503817e-01
9.29868519e-01 -1.82331491e+00 -4.39055651e-01 2.24380240e-01
6.09770834e-01 -1.35794640e-01 5.19146323e-01 1.00506747e+00
-2.64000088e-01 2.28061989e-01 -5.04509568e-01 -4.65949804e-01
-1.07192540e+00 6.19491935e-01 1.87471494e-01 1.10496841e-01
-8.51562083e-01 5.87926328e-01 -7.91245606e-04 -4.23455574e-02
4.35970455e-01 -2.51567006e-01 -2.50963628e-01 1.20375037e-01
8.69134367e-01 5.80135405e-01 -2.03848794e-01 -7.61629343e-01
-8.86452138e-01 5.04040241e-01 3.73939812e-01 9.56757292e-02
1.51307666e+00 -1.28132641e-01 1.81666687e-01 4.45676982e-01
1.28174531e+00 -4.79773819e-01 -1.29327476e+00 -2.65221596e-01
-2.57145137e-01 -4.32000756e-01 -1.09172761e-01 -4.92036074e-01
-1.33376539e+00 9.39672470e-01 1.03611684e+00 -2.06968144e-01
1.40940893e+00 -4.60993648e-02 1.40237451e+00 3.49769831e-01
5.14539957e-01 -1.06306791e+00 1.98868319e-01 2.09438547e-01
6.40267968e-01 -1.17589355e+00 7.58763868e-03 1.30004287e-01
-6.69186175e-01 8.58603954e-01 7.57243514e-01 -1.05036199e-01
6.52424276e-01 1.64723262e-01 -3.45855266e-01 -1.72213897e-01
-8.15538466e-01 -3.11630536e-02 -2.76920319e-01 7.13322043e-01
7.95859843e-02 -4.03820425e-02 -2.61108994e-01 1.04294813e+00
4.81002443e-02 6.60584867e-01 2.57842064e-01 1.21466219e+00
-2.44880840e-01 -6.66516960e-01 -1.38575271e-01 5.65766573e-01
-2.85554796e-01 2.90300310e-01 -1.27772048e-01 4.91098583e-01
4.94327873e-01 8.22835803e-01 3.08775287e-02 -7.32425094e-01
6.78768337e-01 2.35386584e-02 3.17179024e-01 -2.00706292e-02
-3.29925925e-01 -2.74105459e-01 -2.61968046e-01 -1.14169621e+00
-6.77313209e-01 -8.22195768e-01 -1.34227002e+00 -3.35254610e-01
2.61791497e-01 -6.12994432e-02 2.83031553e-01 7.48240888e-01
7.89280415e-01 3.86902303e-01 4.18768406e-01 -7.91940212e-01
-1.31136328e-01 -6.64240599e-01 -6.28730536e-01 6.81408942e-01
1.17081791e-01 -8.66265118e-01 -2.50616223e-01 1.84466988e-01] | [7.994204521179199, 0.7675341963768005] |
299917e0-f951-49c8-b345-5241c9964cd5 | uncertainty-inspired-underwater-image | 2207.09689 | null | https://arxiv.org/abs/2207.09689v1 | https://arxiv.org/pdf/2207.09689v1.pdf | Uncertainty Inspired Underwater Image Enhancement | A main challenge faced in the deep learning-based Underwater Image Enhancement (UIE) is that the ground truth high-quality image is unavailable. Most of the existing methods first generate approximate reference maps and then train an enhancement network with certainty. This kind of method fails to handle the ambiguity of the reference map. In this paper, we resolve UIE into distribution estimation and consensus process. We present a novel probabilistic network to learn the enhancement distribution of degraded underwater images. Specifically, we combine conditional variational autoencoder with adaptive instance normalization to construct the enhancement distribution. After that, we adopt a consensus process to predict a deterministic result based on a set of samples from the distribution. By learning the enhancement distribution, our method can cope with the bias introduced in the reference map labeling to some extent. Additionally, the consensus process is useful to capture a robust and stable result. We examined the proposed method on two widely used real-world underwater image enhancement datasets. Experimental results demonstrate that our approach enables sampling possible enhancement predictions. Meanwhile, the consensus estimate yields competitive performance compared with state-of-the-art UIE methods. Code available at https://github.com/zhenqifu/PUIE-Net. | ['Kai-Kuang Ma', 'Xinghao Ding', 'Yue Huang', 'Wu Wang', 'Zhenqi Fu'] | 2022-07-20 | null | null | null | null | ['uie'] | ['computer-vision'] | [ 1.18899181e-01 -2.93581896e-02 6.41156852e-01 -5.16100943e-01
-1.07960856e+00 -2.55003780e-01 1.91183150e-01 -2.54290223e-01
-7.40586340e-01 8.06539714e-01 1.22476645e-01 2.45000228e-01
-2.67655492e-01 -1.02702737e+00 -8.74402761e-01 -1.12817264e+00
1.81788474e-01 1.01144530e-01 1.13713540e-01 -2.30255887e-01
2.07113877e-01 1.02655664e-01 -1.63876903e+00 4.78368774e-02
1.11299574e+00 1.09550607e+00 7.43724704e-01 4.61384326e-01
7.56791607e-02 6.22861445e-01 -6.46002233e-01 -3.45366329e-01
2.20917583e-01 -4.06708866e-01 -2.75052845e-01 6.34219050e-02
1.43812984e-01 -8.11944187e-01 -4.10320431e-01 1.57657719e+00
8.40350091e-01 2.19962582e-01 7.31100798e-01 -1.18146038e+00
-6.20455861e-01 6.32644653e-01 -5.44888794e-01 -1.03027076e-01
-2.16307014e-01 -1.45567119e-01 6.15345120e-01 -9.73067522e-01
4.00058717e-01 9.66941774e-01 7.94746459e-01 4.30245250e-01
-8.36639881e-01 -7.04970837e-01 -3.44933458e-02 3.31762940e-01
-1.51766086e+00 -4.93815690e-01 8.43787551e-01 -2.98350364e-01
7.18605332e-03 -5.37485145e-02 4.79780883e-01 6.46455586e-01
3.67031902e-01 8.61983538e-01 1.42692435e+00 -1.32105246e-01
3.62006158e-01 4.09789383e-02 -2.22980395e-01 4.98127609e-01
-8.48641794e-04 1.68453008e-01 -3.61223161e-01 5.52185141e-02
5.91399908e-01 4.75179367e-02 -8.39340389e-01 7.42621571e-02
-7.99687028e-01 5.50270975e-01 4.83462363e-01 4.57105935e-02
-7.16951072e-01 -2.87569873e-02 8.76447186e-02 1.91727176e-01
4.45141405e-01 -6.39577433e-02 -3.50794107e-01 1.30683511e-01
-1.07967007e+00 1.06392011e-01 6.48666024e-01 7.64856875e-01
1.14468861e+00 2.04054430e-01 -1.10503912e-01 7.86977828e-01
5.91177285e-01 9.73959267e-01 1.45392314e-01 -1.28206229e+00
1.62108466e-01 -1.85809508e-01 5.17213583e-01 -9.76813138e-01
-9.55203474e-02 -4.98241067e-01 -1.10956442e+00 5.95831573e-01
3.70106697e-01 -3.67694914e-01 -7.89571524e-01 1.61596656e+00
2.00104013e-01 3.64643484e-01 5.63575387e-01 1.14552653e+00
7.68098950e-01 9.91441667e-01 -1.35093808e-01 -2.65668899e-01
9.53502059e-01 -5.63221514e-01 -1.10979128e+00 2.49145888e-02
-2.12271824e-01 -7.23682582e-01 5.04966557e-01 5.90294361e-01
-9.06286895e-01 -5.69015205e-01 -1.18763459e+00 1.78503886e-01
1.52944043e-01 4.18852210e-01 2.01843679e-01 3.69010478e-01
-1.24369407e+00 8.06932867e-01 -9.04915988e-01 1.70304045e-01
1.77973166e-01 1.80169106e-01 -3.63075674e-01 -3.26980948e-01
-1.36722171e+00 8.74274731e-01 5.51082134e-01 7.36051381e-01
-1.22899711e+00 -5.03640056e-01 -9.88990307e-01 2.94983648e-02
9.29538831e-02 -2.95398057e-01 1.16590726e+00 -9.47693884e-01
-1.62856841e+00 2.58175611e-01 -5.35826907e-02 -1.67310596e-01
6.64956868e-01 -1.66289479e-01 -1.29138783e-01 2.20831007e-01
1.41029051e-02 5.54211020e-01 8.17194045e-01 -1.77560377e+00
-7.42972612e-01 -2.48011187e-01 -1.15633860e-01 3.14059317e-01
-2.92002589e-01 -2.80512959e-01 -4.95720595e-01 -1.89282000e-01
4.82449949e-01 -5.76476455e-01 -2.21623540e-01 3.58252347e-01
-1.41913489e-01 9.21074823e-02 6.22581065e-01 -1.10559249e+00
8.53213906e-01 -2.13497066e+00 1.20539188e-01 2.82243602e-02
1.15988269e-01 -8.01702365e-02 -1.05562441e-01 2.52843738e-01
4.28538293e-01 -3.95778529e-02 -6.62124395e-01 -5.78903198e-01
1.43547669e-01 3.99766296e-01 -1.66984588e-01 8.51348639e-01
-9.43717733e-03 2.59369224e-01 -1.05289018e+00 -6.73912883e-01
2.80433238e-01 8.72752964e-01 -3.04879159e-01 6.27810955e-01
3.10491677e-02 5.65703213e-01 -3.81965429e-01 5.28851449e-01
1.29441226e+00 1.69569358e-01 8.92452225e-02 -6.94923818e-01
-3.85763586e-01 -5.45281053e-01 -1.47247779e+00 1.73003531e+00
-4.07095820e-01 4.94617224e-01 4.55136031e-01 -8.04541767e-01
1.22897565e+00 2.60479480e-01 3.76902908e-01 -3.87502044e-01
3.64992857e-01 3.21220577e-01 -2.52414733e-01 -5.97861648e-01
6.69654608e-01 -4.61768001e-01 2.90660203e-01 2.81772405e-01
2.41774991e-01 -1.99201927e-01 -5.20734005e-02 -8.20055455e-02
4.99317676e-01 4.53350633e-01 8.17737356e-03 -3.38854402e-01
3.38435978e-01 -3.61214966e-01 1.09592581e+00 8.15381587e-01
-3.55085552e-01 9.30961072e-01 7.50326291e-02 -1.62334904e-01
-1.16595161e+00 -1.18888652e+00 -2.38816142e-01 6.34142697e-01
6.79532349e-01 1.74743265e-01 -7.44651854e-01 -1.80590674e-01
-5.41575134e-01 5.97685635e-01 -5.93321264e-01 9.13195983e-02
-2.26949438e-01 -1.06647015e+00 3.71528774e-01 4.12639499e-01
1.03884029e+00 -8.62098932e-01 -3.02436531e-01 2.97760040e-01
-5.95713675e-01 -8.57866824e-01 -2.26293772e-01 -7.28813782e-02
-6.91665351e-01 -7.43940532e-01 -1.06532514e+00 -7.75336206e-01
6.79586589e-01 4.85379025e-02 8.71509075e-01 -2.48399097e-03
2.38078520e-01 1.92111179e-01 -4.24630851e-01 -4.28228438e-01
-4.74062949e-01 -5.41798532e-01 2.89637595e-02 1.54265404e-01
1.16650350e-01 -6.00960791e-01 -8.80784929e-01 3.67251158e-01
-1.01677752e+00 -8.74396339e-02 6.95789456e-01 1.07261097e+00
8.79482031e-01 4.78389740e-01 4.41992462e-01 -2.77515173e-01
3.60730231e-01 -6.36453986e-01 -7.70931959e-01 3.21672171e-01
-4.74690169e-01 1.48474962e-01 3.51170152e-01 -4.34970498e-01
-1.65044963e+00 5.84696196e-02 -5.31321347e-01 -4.12785083e-01
-5.16267344e-02 7.05374599e-01 -3.46295416e-01 -3.01090200e-02
3.69169176e-01 6.19833648e-01 2.05581293e-01 -4.26945388e-01
3.46706957e-02 9.03485835e-01 8.03882062e-01 -6.52471483e-01
7.82025635e-01 6.21384740e-01 -2.14604959e-01 -7.28684306e-01
-7.41802275e-01 -2.41191745e-01 -4.10322964e-01 -6.22720599e-01
8.97692919e-01 -1.27575958e+00 -6.83306277e-01 9.60945904e-01
-1.19850945e+00 -4.45667684e-01 1.85058594e-01 6.75266206e-01
-3.53133798e-01 6.40098572e-01 -8.34506810e-01 -1.17624509e+00
-4.47180390e-01 -1.18365002e+00 1.03058577e+00 7.12178349e-01
7.49597847e-01 -8.18998933e-01 9.27208923e-03 -3.85237038e-02
6.20471478e-01 3.15451264e-01 -2.42207129e-03 -2.88391765e-02
-5.90446353e-01 1.13502242e-01 -2.95310587e-01 7.38234878e-01
2.33511552e-02 8.32173005e-02 -1.10636306e+00 -2.96036065e-01
2.94516951e-01 -2.50410587e-01 9.18634832e-01 5.63676536e-01
1.04748094e+00 -1.02172293e-01 1.50403574e-01 7.89657593e-01
1.78955030e+00 1.31024390e-01 9.73879993e-01 3.24293584e-01
3.89055878e-01 6.45196259e-01 8.63977432e-01 7.99149275e-01
6.20700598e-01 2.47502699e-01 6.31497681e-01 -5.75298928e-02
1.26056939e-01 -2.34765604e-01 2.61227489e-01 8.59563589e-01
-2.68020421e-01 -4.96958017e-01 -5.92757821e-01 6.64935946e-01
-1.74918759e+00 -8.88879836e-01 1.21722668e-01 1.93957472e+00
9.16636288e-01 -3.01927388e-01 -6.41103923e-01 9.85736251e-02
1.07232594e+00 5.44516668e-02 -4.32577640e-01 3.08637768e-01
-2.56444871e-01 -1.90067708e-01 6.25839829e-01 6.22295499e-01
-1.11812055e+00 5.80001116e-01 5.07140732e+00 8.46060336e-01
-1.07063401e+00 2.10912853e-01 5.83094537e-01 6.04915500e-01
-5.16483307e-01 -2.64709771e-01 -5.91808558e-01 5.10692775e-01
5.23772359e-01 4.87544611e-02 2.45786577e-01 7.35959709e-01
3.23776871e-01 -7.20759332e-02 -5.38114071e-01 1.02614558e+00
1.38368443e-01 -1.02426529e+00 -3.03528905e-01 -1.31288901e-01
8.65416825e-01 -7.23397918e-03 -7.77902976e-02 2.13776067e-01
3.62910986e-01 -7.05375433e-01 1.02844608e+00 1.19074988e+00
8.23511064e-01 -6.50578260e-01 1.39130104e+00 3.94532114e-01
-1.07518148e+00 1.25007797e-02 -8.07672799e-01 -4.95863035e-02
4.33922946e-01 6.76038802e-01 -3.36466074e-01 7.31107652e-01
9.49644625e-01 6.01391315e-01 -4.34258357e-02 1.29166567e+00
-6.47353768e-01 5.59948564e-01 -4.16776896e-01 1.01946458e-01
-4.04379703e-02 -5.23288846e-01 4.93389457e-01 9.76241708e-01
8.23228538e-01 5.07828653e-01 5.63455895e-02 9.58216846e-01
-1.01908267e-01 -1.83548257e-01 -1.35026321e-01 4.77297127e-01
6.02527857e-01 1.26701808e+00 -4.40482199e-01 -1.97279111e-01
-6.71923235e-02 9.86440420e-01 1.82481892e-02 5.21217525e-01
-7.93951809e-01 -4.41450119e-01 3.76193613e-01 -4.80957776e-01
2.55423218e-01 -1.33412644e-01 1.30590126e-02 -1.23432374e+00
-4.97673079e-02 -5.78399420e-01 9.59267393e-02 -1.05118978e+00
-1.48128521e+00 9.68537509e-01 5.02559450e-03 -1.48455477e+00
3.37350741e-02 -4.56063360e-01 -6.38367891e-01 1.00684547e+00
-1.88190579e+00 -1.03204966e+00 -8.75234604e-01 3.98766965e-01
3.12792003e-01 1.89428180e-01 6.54160380e-01 5.04580855e-01
-2.60834515e-01 3.91154885e-01 5.02655089e-01 2.89403230e-01
7.54021049e-01 -1.39480269e+00 -3.28120619e-01 1.12258768e+00
-3.67904365e-01 2.37147510e-01 1.15423477e+00 -5.42515099e-01
-1.08685899e+00 -1.01790679e+00 2.67391324e-01 1.68480769e-01
4.56187457e-01 9.80112925e-02 -1.12313485e+00 4.04006004e-01
3.33364218e-01 1.97625980e-01 3.64964157e-01 -5.19113779e-01
1.12170786e-01 -4.60847944e-01 -1.34255433e+00 2.47106075e-01
5.08305013e-01 -2.90841192e-01 -4.38945860e-01 -9.27528664e-02
4.74011034e-01 -6.51705325e-01 -1.20731497e+00 5.83977997e-01
5.91881037e-01 -1.07481349e+00 5.99408567e-01 3.28497112e-01
6.72888100e-01 -8.75640392e-01 -5.17328680e-01 -1.51221919e+00
-5.19871227e-02 -2.01621354e-01 2.61264503e-01 1.38868666e+00
2.51468092e-01 -5.71829736e-01 4.72314894e-01 4.99543041e-01
-4.54023242e-01 -4.18129146e-01 -7.60150194e-01 -3.37514639e-01
1.85926035e-02 -2.51667351e-01 6.09475970e-01 6.56959057e-01
-5.56702495e-01 -3.56784403e-01 -8.38646412e-01 9.42104816e-01
1.38655806e+00 1.22479431e-01 5.99732697e-01 -9.22773421e-01
-4.38998431e-01 6.35378016e-03 -3.14511508e-01 -1.18044186e+00
5.82362711e-02 -2.93998361e-01 9.06597018e-01 -1.87827623e+00
3.32943261e-01 -3.46801758e-01 -3.98827821e-01 3.15973788e-01
-2.95913190e-01 4.31370676e-01 8.55338722e-02 1.29075944e-01
-5.53973556e-01 1.16261292e+00 1.27014458e+00 -2.56397933e-01
6.75101727e-02 -2.22946294e-02 -5.21999776e-01 6.10363483e-01
9.29191649e-01 -5.88716865e-01 -1.18896551e-01 -7.26652205e-01
7.83429518e-02 2.99621344e-01 3.12721461e-01 -1.03810108e+00
5.29952407e-01 -1.47965133e-01 4.83285606e-01 -5.93771696e-01
2.40019336e-01 -7.53528714e-01 2.46890277e-01 2.25885615e-01
-1.45323640e-02 -4.30650324e-01 5.00499047e-02 6.25371993e-01
-5.97103834e-01 -5.57590902e-01 9.16729748e-01 -1.01459771e-01
-1.03371084e+00 4.28028464e-01 -2.68403560e-01 -2.45784327e-01
5.29218733e-01 3.33383866e-02 -3.28026891e-01 -6.77170992e-01
-5.93925595e-01 4.00955915e-01 4.30706829e-01 -1.55676650e-02
9.38628316e-01 -1.02109766e+00 -1.17064643e+00 -1.68426380e-01
-2.88978741e-02 2.63153732e-01 8.01185131e-01 6.74700797e-01
-6.43480957e-01 -6.52971447e-01 -3.10210526e-01 -6.98894382e-01
-8.64212275e-01 1.55506194e-01 7.52924740e-01 1.50576606e-01
-4.50408340e-01 7.52324879e-01 1.16429895e-01 -5.61310530e-01
9.36314911e-02 2.54378337e-02 -4.07105386e-01 -2.30801404e-01
7.71075964e-01 2.80779898e-01 -4.14616495e-01 -6.54407084e-01
-4.24619690e-02 6.19003654e-01 4.65833902e-01 -2.82339841e-01
1.59116590e+00 -5.79214990e-01 -2.76019037e-01 1.93679839e-01
1.06441355e+00 -1.20258965e-01 -2.00360155e+00 -3.26106459e-01
-4.78273034e-01 -4.73312080e-01 4.84324813e-01 -7.00501323e-01
-1.33968091e+00 8.41571629e-01 1.01533198e+00 -2.20019683e-01
1.46284890e+00 -2.54855752e-01 6.04633272e-01 2.79972941e-01
4.95215833e-01 -9.45878685e-01 -2.32939705e-01 2.52345413e-01
9.53529358e-01 -1.67312765e+00 6.27826825e-02 -3.84030119e-02
-7.02596843e-01 1.22636533e+00 5.72616398e-01 -6.70492724e-02
7.47710347e-01 4.97250199e-01 5.15788734e-01 3.47819664e-02
-1.67309552e-01 -5.15949577e-02 1.21767847e-02 5.98447978e-01
2.26632375e-02 -8.74413550e-02 -2.42988929e-01 8.51536512e-01
3.26625071e-02 5.78531548e-02 9.19185758e-01 6.55316949e-01
-5.65810800e-01 -6.90001428e-01 -5.30842960e-01 2.36747619e-02
-5.56042016e-01 -2.07582504e-01 7.02255845e-01 4.56796348e-01
2.77589649e-01 1.11449516e+00 -7.90333375e-02 -4.57607388e-01
1.32066756e-01 -4.92263675e-01 1.82897851e-01 -2.98288260e-02
1.27440736e-01 2.63254851e-01 -2.65465826e-01 -2.43576452e-01
-8.24091315e-01 -5.96322238e-01 -1.32429159e+00 -1.77239403e-02
-4.15947407e-01 5.64033389e-01 8.15531433e-01 7.64953494e-01
-4.85806204e-02 4.38981652e-01 7.16485083e-01 -1.26828909e+00
-6.13719225e-01 -1.24303210e+00 -9.47683394e-01 1.44195050e-01
4.07873720e-01 -6.21440411e-01 -8.52112412e-01 2.06970334e-01] | [10.707134246826172, -3.523740768432617] |
493b78fd-ca58-4702-9059-d9acb8dbce2d | butterfly-robust-one-step-approach-towards | 1905.07720 | null | https://arxiv.org/abs/1905.07720v3 | https://arxiv.org/pdf/1905.07720v3.pdf | Butterfly: One-step Approach towards Wildly Unsupervised Domain Adaptation | In unsupervised domain adaptation (UDA), classifiers for the target domain (TD) are trained with clean labeled data from the source domain (SD) and unlabeled data from TD. However, in the wild, it is difficult to acquire a large amount of perfectly clean labeled data in SD given limited budget. Hence, we consider a new, more realistic and more challenging problem setting, where classifiers have to be trained with noisy labeled data from SD and unlabeled data from TD -- we name it wildly UDA (WUDA). We show that WUDA ruins all UDA methods if taking no care of label noise in SD, and to this end, we propose a Butterfly framework, a powerful and efficient solution to WUDA. Butterfly maintains four deep networks simultaneously, where two take care of all adaptations (i.e., noisy-to-clean, labeled-to-unlabeled, and SD-to-TD-distributional) and then the other two can focus on classification in TD. As a consequence, Butterfly possesses all the conceptually necessary components for solving WUDA. Experiments demonstrate that, under WUDA, Butterfly significantly outperforms existing baseline methods. | ['Bo Han', 'Masashi Sugiyama', 'Jie Lu', 'Feng Liu', 'Guangquan Zhang', 'Gang Niu'] | 2019-05-19 | butterfly-one-step-approach-towards-wildly | http://128.84.4.34/abs/1905.07720 | http://128.84.4.34/pdf/1905.07720 | null | ['wildly-unsupervised-domain-adaptation'] | ['computer-vision'] | [ 2.84320824e-02 4.45084386e-02 -1.98545661e-02 -5.28610945e-01
-1.03652692e+00 -7.46876240e-01 3.41257185e-01 -2.93312341e-01
-3.00369322e-01 9.81331825e-01 -6.70257583e-02 -2.76054770e-01
1.65154412e-01 -6.61626995e-01 -7.95966327e-01 -9.58854973e-01
3.55343848e-01 7.57861197e-01 -8.42484236e-02 -4.10948209e-02
-4.60843652e-01 4.88753542e-02 -1.30965245e+00 -1.09245442e-01
9.90778625e-01 1.11065412e+00 3.08231711e-01 2.26518035e-01
-1.37165263e-01 6.99393153e-01 -6.36599898e-01 -2.23092914e-01
5.22057414e-01 -5.53451896e-01 -9.33098376e-01 4.23638493e-01
4.50099200e-01 -4.80143040e-01 -9.77608562e-02 1.18212867e+00
6.11440778e-01 4.92162138e-01 7.51229644e-01 -1.23286319e+00
-1.05287170e+00 6.70596361e-01 -5.43498337e-01 -1.15278453e-01
-1.45732462e-01 2.54460424e-01 8.27427268e-01 -1.11461198e+00
6.97471917e-01 1.18064582e+00 5.93085647e-01 1.04749227e+00
-1.23704481e+00 -7.67343938e-01 4.85761523e-01 -1.18275262e-01
-9.97734666e-01 -6.66001558e-01 5.64769328e-01 -2.72385776e-01
6.14037037e-01 -1.08631968e-01 9.24201608e-02 1.69716847e+00
-4.59690392e-01 1.01106620e+00 1.12583911e+00 -4.57829356e-01
7.45595396e-01 1.42885763e-02 4.57191795e-01 1.77605227e-01
2.24569783e-01 -3.17344591e-02 -2.62042701e-01 -2.35921189e-01
3.16652387e-01 6.40776530e-02 -2.74515748e-01 -6.34904981e-01
-1.08675766e+00 7.22259998e-01 1.92572251e-01 4.79846187e-02
-3.02983969e-01 -3.69678885e-01 4.30073440e-01 6.27737343e-01
5.88641644e-01 4.12987739e-01 -8.26286614e-01 1.74804516e-02
-6.21738553e-01 2.90108293e-01 7.14935303e-01 1.36364186e+00
9.39750612e-01 1.69940948e-01 -1.83158275e-02 1.01886892e+00
9.46793854e-02 9.45827007e-01 7.51569450e-01 -8.37549806e-01
4.57452744e-01 3.13861549e-01 3.29192251e-01 -3.52782369e-01
-2.59124339e-01 -4.38903391e-01 -1.14646053e+00 4.28400263e-02
5.57488561e-01 -4.41868424e-01 -1.26948738e+00 2.14561439e+00
5.00676572e-01 3.14440042e-01 3.72317523e-01 8.54131997e-01
4.64933902e-01 6.85170352e-01 3.08516854e-03 -2.67911613e-01
8.97857308e-01 -1.24695003e+00 -6.59800887e-01 -6.68729484e-01
6.60924375e-01 -3.30831587e-01 1.35665774e+00 5.83210886e-01
-9.05776978e-01 -7.04731941e-01 -1.09182513e+00 -2.76744157e-01
-4.58595783e-01 1.40585348e-01 9.63141844e-02 3.32093328e-01
-1.00724494e+00 3.52530122e-01 -6.39887512e-01 -5.23550987e-01
5.14506936e-01 1.26807794e-01 -4.52537358e-01 -6.96645141e-01
-1.22083855e+00 7.43578672e-01 2.61887461e-01 -4.55472656e-02
-1.27363575e+00 -5.43233812e-01 -9.87339079e-01 -4.00477722e-02
6.04794621e-01 -7.07516432e-01 1.81550825e+00 -1.00091672e+00
-1.40134442e+00 9.27810013e-01 -2.48248667e-01 -4.92102176e-01
5.00682712e-01 -4.57484484e-01 -4.51500326e-01 -3.34147274e-01
4.23372924e-01 3.91815871e-01 1.08145249e+00 -1.37909412e+00
-7.25832582e-01 -6.10204756e-01 -5.59483878e-02 -6.68103173e-02
-3.84100020e-01 -3.17680031e-01 -2.19482690e-01 -5.74379623e-01
-1.97252855e-01 -8.19007933e-01 -7.81469569e-02 -3.06420177e-01
-4.59956646e-01 -4.11623716e-01 9.04662549e-01 -5.04662633e-01
1.22529542e+00 -2.60585284e+00 2.13727683e-01 1.17113046e-01
4.66791362e-01 5.96019566e-01 -3.74759138e-01 1.26758024e-01
-2.24783465e-01 -8.74885358e-03 -5.53323627e-01 -7.15308845e-01
2.51778126e-01 5.67438185e-01 -5.76868117e-01 3.41114372e-01
2.62514204e-01 6.25108778e-01 -1.29040861e+00 -9.80329365e-02
-7.02451169e-02 1.01169653e-01 -4.57421958e-01 4.91156369e-01
-4.24891382e-01 6.02523148e-01 -6.06006265e-01 5.96393228e-01
1.05097842e+00 -3.85329783e-01 2.57893771e-01 1.81523055e-01
1.57939076e-01 2.26111323e-01 -1.15163887e+00 1.89717960e+00
-5.76475263e-01 2.73732901e-01 2.05593973e-01 -1.21747160e+00
9.96047318e-01 9.94207263e-02 2.11059213e-01 -7.30346441e-01
1.06219105e-01 4.58448023e-01 -2.62185246e-01 -4.56609011e-01
2.34926239e-01 -1.08210348e-01 -2.35838369e-01 6.84839725e-01
3.89906734e-01 2.07489014e-01 1.26595691e-01 9.29632783e-02
1.14002275e+00 6.26095906e-02 3.40799034e-01 -1.81941181e-01
2.78643668e-01 8.21670815e-02 8.49729955e-01 8.46196413e-01
-4.77026284e-01 6.64820254e-01 2.59009629e-01 -3.87081206e-01
-1.11466026e+00 -1.12186074e+00 9.38782990e-02 1.35171127e+00
5.48515022e-02 -5.93204796e-02 -1.08702624e+00 -1.12237704e+00
-1.38447005e-02 7.97863603e-01 -7.66677201e-01 -3.01576704e-01
-4.69859481e-01 -4.14341629e-01 3.40529114e-01 5.29621899e-01
6.60490632e-01 -9.95243251e-01 -3.37314717e-02 1.78677455e-01
-1.17445856e-01 -9.29760098e-01 -6.69147670e-01 6.49591386e-01
-5.17386913e-01 -9.54844773e-01 -8.80429029e-01 -1.08356428e+00
6.89558446e-01 5.71987629e-01 1.24995482e+00 -4.84350711e-01
1.82789907e-01 1.98243946e-01 -5.28350830e-01 -3.88274431e-01
-3.87732983e-01 2.76270270e-01 3.80663455e-01 5.34107126e-02
7.45946169e-01 -7.32130527e-01 -2.42940530e-01 3.79579008e-01
-1.06960154e+00 -2.65426517e-01 5.78207850e-01 1.02513433e+00
8.32217157e-01 9.64706196e-05 9.23328876e-01 -1.37199187e+00
5.53002477e-01 -8.20896029e-01 -4.37477559e-01 3.38645220e-01
-4.65467006e-01 3.10762115e-02 1.12057602e+00 -6.56400204e-01
-9.77702141e-01 8.34495947e-02 -1.09655656e-01 -8.19461405e-01
-5.87044775e-01 4.16528314e-01 -7.51594067e-01 4.45392072e-01
9.92492974e-01 3.32354188e-01 -5.23594767e-02 -7.55565107e-01
5.43324947e-01 1.03688848e+00 7.73507655e-01 -8.24786186e-01
8.73903155e-01 3.55195820e-01 -6.10271990e-01 -5.79535067e-01
-1.36701751e+00 -4.57271427e-01 -7.32933939e-01 2.41856590e-01
6.07190549e-01 -1.04039621e+00 9.42054689e-02 8.00022542e-01
-9.30973589e-01 -8.02390575e-01 -7.19843864e-01 1.81881621e-01
-3.50229800e-01 2.81576008e-01 -2.50250757e-01 -4.81485188e-01
-2.44417265e-01 -1.05176067e+00 1.01595056e+00 2.30502725e-01
-1.63255241e-02 -1.03649163e+00 1.56281337e-01 4.18840460e-02
4.31333393e-01 1.70009315e-01 8.38974774e-01 -1.19188321e+00
-6.65663779e-02 -1.66358445e-02 -1.06068127e-01 1.05442905e+00
5.07156610e-01 -4.03278410e-01 -1.25296414e+00 -4.39031333e-01
2.56644785e-01 -7.61906981e-01 7.23812222e-01 2.61317909e-01
1.18345928e+00 -2.54755497e-01 -1.63178608e-01 6.06976032e-01
1.06363654e+00 2.08280295e-01 1.85404599e-01 3.29697579e-01
6.62401319e-01 3.26327354e-01 8.11481416e-01 4.43830311e-01
4.77132648e-01 3.36229384e-01 2.27613315e-01 -5.46923354e-02
-2.77280599e-01 -2.76848018e-01 5.83745420e-01 1.09636343e+00
4.78875101e-01 -4.29117978e-01 -9.18459535e-01 7.78724790e-01
-1.85572243e+00 -5.57483971e-01 1.71208322e-01 2.25049186e+00
1.10675073e+00 -4.88131195e-02 2.32179090e-01 -8.48444626e-02
9.05211866e-01 -4.03087866e-03 -1.16934049e+00 -1.15923427e-01
-2.47132763e-01 3.02006900e-01 1.97429493e-01 1.96768492e-01
-1.43173039e+00 7.81052172e-01 5.81106091e+00 8.25215161e-01
-1.10465431e+00 3.40811819e-01 6.48541987e-01 4.49535102e-02
-1.85509086e-01 -2.87195086e-01 -6.92857325e-01 6.37365282e-01
9.81378913e-01 -1.29670978e-01 5.66352904e-01 1.35718894e+00
-3.52056734e-02 3.11767578e-01 -1.40787077e+00 9.82719600e-01
1.86265782e-02 -7.37015545e-01 -4.25425284e-02 -8.39562714e-02
1.01594055e+00 1.96692333e-01 1.36589080e-01 7.43350506e-01
1.06495440e+00 -6.59081161e-01 6.38377368e-01 -3.69280092e-02
1.02613294e+00 -6.20392084e-01 6.58822179e-01 7.09668696e-01
-7.52363682e-01 -1.47996441e-01 -6.64844573e-01 2.42646411e-01
-8.90162736e-02 8.17330062e-01 -5.32103777e-01 6.34368539e-01
6.11297965e-01 1.01026082e+00 -2.60259777e-01 7.01230645e-01
-2.83479869e-01 7.64091551e-01 -2.13692501e-01 4.64953393e-01
3.37384909e-01 -1.83605134e-01 1.77965373e-01 1.01354742e+00
5.13623595e-01 -2.98808534e-02 4.42945719e-01 6.54722273e-01
-4.52026695e-01 -1.82065114e-01 -7.56182969e-01 1.23932302e-01
8.23110461e-01 8.80928993e-01 -1.84261143e-01 -4.16589558e-01
-5.36390662e-01 1.07060170e+00 5.84079146e-01 5.71870208e-01
-7.06817031e-01 -2.19645172e-01 7.98166752e-01 -1.93065315e-01
2.84032196e-01 3.04131620e-02 -6.29480109e-02 -1.48406625e+00
1.05180502e-01 -1.14032733e+00 4.69612598e-01 -6.66297376e-01
-2.06015396e+00 7.09289193e-01 -2.13045612e-01 -1.36786473e+00
-3.33084673e-01 -5.56754112e-01 -3.91453385e-01 8.92416716e-01
-1.73986089e+00 -1.05428636e+00 -3.28892976e-01 9.77835178e-01
7.97069013e-01 -6.03961386e-02 8.48013639e-01 4.72692430e-01
-8.97991300e-01 7.32718945e-01 6.49906456e-01 2.90889174e-01
1.31746209e+00 -1.39073527e+00 6.50285423e-01 9.50924039e-01
-1.32861897e-01 4.99369472e-01 3.12180787e-01 -4.11498040e-01
-1.25930262e+00 -1.62252009e+00 7.41197050e-01 -3.32608044e-01
6.96756065e-01 -5.07232368e-01 -1.14333594e+00 9.69013810e-01
7.39141032e-02 2.68585324e-01 6.66493118e-01 1.01752609e-01
-6.44527972e-01 -2.04876423e-01 -1.17877841e+00 2.60041803e-01
1.09863937e+00 -5.35664856e-01 -8.64916205e-01 4.16538477e-01
8.54009867e-01 -4.16163862e-01 -5.80116868e-01 3.32628302e-02
-3.50718126e-02 -6.82490885e-01 7.84109592e-01 -7.74006546e-01
2.39549473e-01 -2.77482092e-01 -2.55271584e-01 -1.89043856e+00
-2.83366829e-01 -5.40493548e-01 -1.18525371e-01 1.49031675e+00
3.63080889e-01 -8.01739335e-01 3.25391233e-01 6.50232494e-01
-4.96981889e-01 -2.69523889e-01 -8.41323853e-01 -1.15541410e+00
4.70715404e-01 -3.28993082e-01 8.49353969e-01 1.45215166e+00
-2.95144558e-01 4.65315610e-01 -4.25390601e-01 1.52974695e-01
5.79186916e-01 2.02726290e-01 8.49272251e-01 -1.30294061e+00
-1.19812220e-01 -7.35438392e-02 2.80625224e-01 -1.32103539e+00
3.24455172e-01 -8.69563103e-01 4.21173453e-01 -1.24701464e+00
8.70788172e-02 -6.87465131e-01 -4.71431762e-01 8.02253962e-01
-4.12776649e-01 -1.40109926e-01 -8.56268313e-03 4.38034952e-01
-8.39465916e-01 5.29532313e-01 1.16755140e+00 -3.12259763e-01
-1.00367330e-01 -1.16120666e-01 -9.84052956e-01 6.34659767e-01
7.75644779e-01 -5.05302250e-01 -5.68147242e-01 -9.15139318e-01
-2.68374830e-01 -3.94555748e-01 3.52633209e-03 -8.09337974e-01
7.95031711e-02 -1.86168432e-01 4.98716123e-02 -1.71970978e-01
-4.67988066e-02 -8.36438835e-01 -1.68919727e-01 -1.16775595e-01
-4.15267289e-01 -2.55194426e-01 -1.06235042e-01 5.51404059e-01
-3.65881771e-01 -1.29997596e-01 9.84730959e-01 -9.85290259e-02
-9.86337662e-01 4.85245705e-01 2.12383792e-02 6.13128960e-01
9.33257163e-01 1.81396574e-01 -6.17191136e-01 -1.94838345e-01
-8.69967818e-01 4.55529094e-01 6.16809964e-01 3.27575147e-01
2.15834066e-01 -1.50585830e+00 -7.04044700e-01 3.50052774e-01
3.00866127e-01 8.11773539e-01 3.68185848e-01 4.64177102e-01
-6.71380162e-02 1.64487123e-01 -5.93958050e-02 -4.23609048e-01
-5.06295979e-01 1.01697195e+00 9.87108871e-02 -4.06811953e-01
-5.17535388e-01 9.88110125e-01 6.01576269e-01 -8.71364832e-01
4.69000757e-01 -4.31022882e-01 1.17399238e-01 5.82080558e-02
6.18062973e-01 1.59543842e-01 1.40411466e-01 -3.60775083e-01
-1.52441204e-01 3.10518920e-01 -2.45592564e-01 3.66685688e-01
1.42042994e+00 -3.95106882e-01 6.51383102e-02 5.14383972e-01
1.06679773e+00 9.41944681e-03 -1.44881356e+00 -8.09453607e-01
-1.10926040e-01 -1.99156627e-01 -2.32002541e-01 -1.00756431e+00
-1.17264807e+00 1.09855258e+00 4.47781324e-01 2.19015956e-01
1.38174546e+00 -5.05236275e-02 9.86288965e-01 6.26317561e-01
4.02006269e-01 -1.17326450e+00 -1.19109996e-01 6.51536345e-01
5.26097059e-01 -1.40389216e+00 -5.57618082e-01 4.83943895e-03
-9.32026684e-01 7.45720744e-01 8.38049889e-01 -1.11487061e-01
5.57426214e-01 1.99880332e-01 2.74864048e-01 1.76037803e-01
-7.63193965e-01 -2.72481650e-01 -2.35525370e-01 1.09969366e+00
-1.60649881e-01 9.77379903e-02 3.89198422e-01 1.08123338e+00
3.00788861e-02 7.82002211e-02 3.25237751e-01 8.80372167e-01
-3.73714864e-01 -1.29682636e+00 -3.50125462e-01 2.59197593e-01
-2.02657640e-01 -1.30007649e-02 -3.79560828e-01 7.89454162e-01
3.52451384e-01 9.62251186e-01 -1.01633430e-01 -3.59107256e-01
4.54990804e-01 3.08261603e-01 -7.43506849e-02 -7.74851978e-01
-3.14309001e-01 4.07377370e-02 -5.86556792e-02 -3.89966786e-01
-4.14491326e-01 -6.52669549e-01 -9.89497364e-01 -2.64876217e-01
-9.54962894e-02 2.51627237e-01 2.97632098e-01 1.05758488e+00
5.00093937e-01 3.00389260e-01 9.46570516e-01 -5.52029431e-01
-9.40519929e-01 -1.10008180e+00 -9.26994503e-01 6.16658151e-01
7.60197401e-01 -7.00133264e-01 -4.40148413e-01 3.35050434e-01] | [10.442687034606934, 3.090784788131714] |
5f070ff5-8e61-436a-bd5f-0dcd5a473175 | universal-transformers | 1807.03819 | null | http://arxiv.org/abs/1807.03819v3 | http://arxiv.org/pdf/1807.03819v3.pdf | Universal Transformers | Recurrent neural networks (RNNs) sequentially process data by updating their
state with each new data point, and have long been the de facto choice for
sequence modeling tasks. However, their inherently sequential computation makes
them slow to train. Feed-forward and convolutional architectures have recently
been shown to achieve superior results on some sequence modeling tasks such as
machine translation, with the added advantage that they concurrently process
all inputs in the sequence, leading to easy parallelization and faster training
times. Despite these successes, however, popular feed-forward sequence models
like the Transformer fail to generalize in many simple tasks that recurrent
models handle with ease, e.g. copying strings or even simple logical inference
when the string or formula lengths exceed those observed at training time. We
propose the Universal Transformer (UT), a parallel-in-time self-attentive
recurrent sequence model which can be cast as a generalization of the
Transformer model and which addresses these issues. UTs combine the
parallelizability and global receptive field of feed-forward sequence models
like the Transformer with the recurrent inductive bias of RNNs. We also add a
dynamic per-position halting mechanism and find that it improves accuracy on
several tasks. In contrast to the standard Transformer, under certain
assumptions, UTs can be shown to be Turing-complete. Our experiments show that
UTs outperform standard Transformers on a wide range of algorithmic and
language understanding tasks, including the challenging LAMBADA language
modeling task where UTs achieve a new state of the art, and machine translation
where UTs achieve a 0.9 BLEU improvement over Transformers on the WMT14 En-De
dataset. | ['Łukasz Kaiser', 'Stephan Gouws', 'Mostafa Dehghani', 'Jakob Uszkoreit', 'Oriol Vinyals'] | 2018-07-10 | universal-transformers-1 | https://openreview.net/forum?id=HyzdRiR9Y7 | https://openreview.net/pdf?id=HyzdRiR9Y7 | iclr-2019-5 | ['learning-to-execute', 'lambada'] | ['computer-code', 'natural-language-processing'] | [ 6.19703829e-01 -7.45656118e-02 -3.11849684e-01 -2.27213651e-01
-6.00264251e-01 -7.52090216e-01 8.04215074e-01 -1.56496286e-01
-5.35808504e-01 6.34102762e-01 1.89586848e-01 -1.07326531e+00
3.88537765e-01 -7.49174654e-01 -1.15497243e+00 -4.94469553e-01
2.74648696e-01 6.98105276e-01 9.56594050e-02 -4.11822051e-01
1.02319643e-01 2.12188467e-01 -1.39425695e+00 5.58948696e-01
6.32336020e-01 6.01791918e-01 4.96405393e-01 8.30441713e-01
-4.38573897e-01 1.13227499e+00 -1.66806892e-01 -4.56088692e-01
1.10898435e-01 -5.85137546e-01 -1.13231206e+00 -5.69922090e-01
3.63039412e-02 -3.10528487e-01 -4.64708775e-01 7.08191037e-01
4.09125119e-01 3.07191759e-02 3.52062643e-01 -9.34836686e-01
-8.76680791e-01 1.13638389e+00 -2.83482015e-01 2.41400540e-01
2.38799810e-01 1.77095860e-01 1.21988916e+00 -9.91113305e-01
6.96721137e-01 1.11901963e+00 9.19291496e-01 8.10430050e-01
-1.39635992e+00 -5.59207559e-01 2.02686578e-01 2.90123492e-01
-9.50574875e-01 -5.61188936e-01 2.90781409e-01 2.91650333e-02
1.77269244e+00 2.10802168e-01 3.70683581e-01 1.28672493e+00
2.30248511e-01 1.22649503e+00 9.56865966e-01 -5.86253166e-01
8.75761211e-02 -3.91198486e-01 1.18305758e-01 8.11472237e-01
-1.12807594e-01 4.71817842e-03 -5.23801744e-01 -6.22141659e-02
5.84617913e-01 6.54675737e-02 -2.74832528e-02 -1.22643793e-02
-1.29254413e+00 6.09179199e-01 3.41746509e-01 4.60815340e-01
-2.28096873e-01 5.26670277e-01 6.27784729e-01 7.31109262e-01
3.72955263e-01 2.88341880e-01 -6.96099162e-01 -4.13480848e-01
-7.77609885e-01 2.02510253e-01 8.57968628e-01 9.58330989e-01
4.86261934e-01 1.21895507e-01 -1.34277090e-01 7.64524281e-01
6.38210326e-02 4.04374987e-01 8.44802260e-01 -6.69001400e-01
6.04975522e-01 4.84385371e-01 -1.87268063e-01 -3.56935799e-01
-3.74855727e-01 -6.07712209e-01 -1.00866187e+00 -3.20773751e-01
2.75465697e-01 9.97899696e-02 -1.05837083e+00 2.13756847e+00
-1.65968820e-01 2.56959707e-01 1.14085197e-01 5.74340820e-01
5.63407719e-01 1.08170068e+00 7.70679936e-02 -2.21888289e-01
1.19407940e+00 -1.03567362e+00 -5.01714051e-01 -6.18773758e-01
1.17829156e+00 -6.33845508e-01 1.20546210e+00 2.20307425e-01
-1.37179971e+00 -3.97519827e-01 -1.07641542e+00 -4.62298870e-01
-1.94937423e-01 -2.19102636e-01 8.53757143e-01 4.77527320e-01
-1.36169457e+00 6.76985621e-01 -1.04417145e+00 -3.74006927e-01
2.00438261e-01 3.54045540e-01 -1.42594084e-01 -1.12790436e-01
-1.26163805e+00 1.09299922e+00 3.78544748e-01 3.33990276e-01
-8.28772247e-01 -8.55672061e-01 -7.89084196e-01 1.73040912e-01
3.01600218e-01 -1.08521140e+00 1.82336354e+00 -1.05398226e+00
-1.83893025e+00 8.04607093e-01 -7.69550085e-01 -9.74247277e-01
4.48651880e-01 -2.35776573e-01 -1.93998054e-01 -3.15731853e-01
-3.54896247e-01 6.19992316e-01 3.88953209e-01 -6.28030002e-01
-3.94191951e-01 -1.35152906e-01 -1.08899117e-01 2.57253468e-01
-1.14324972e-01 1.70805573e-01 -2.59574324e-01 -6.01019859e-01
1.55159250e-01 -1.15281057e+00 -1.94843173e-01 -3.16036373e-01
-2.51981586e-01 -3.74180615e-01 6.09855413e-01 -5.35931408e-01
1.15238738e+00 -1.83148849e+00 4.46913868e-01 -2.21422255e-01
1.00670919e-01 4.64428723e-01 -4.00725752e-01 5.63051879e-01
-2.13510111e-01 2.66021758e-01 -4.90247190e-01 -4.69307542e-01
3.37676592e-02 5.55310071e-01 -7.98399270e-01 1.72073886e-01
2.90207863e-01 1.59729385e+00 -8.03503394e-01 2.21645255e-02
-1.18002310e-01 2.97547668e-01 -5.32085419e-01 1.23204343e-01
-6.54876292e-01 3.62550020e-02 5.17346784e-02 1.07598394e-01
1.48304075e-01 -6.28428638e-01 3.06212783e-01 4.63032871e-01
7.43332356e-02 1.03558517e+00 -5.37503660e-01 1.80980003e+00
-7.83822536e-01 6.92092538e-01 -4.20530021e-01 -1.06910396e+00
7.07792461e-01 5.34791231e-01 -5.99131845e-02 -9.32260394e-01
-7.22897723e-02 4.35987502e-01 2.84762204e-01 -1.02892198e-01
5.49710214e-01 -5.57123840e-01 6.30953386e-02 9.56477880e-01
5.93563206e-02 8.93561319e-02 1.26969963e-01 3.11637670e-01
1.26286471e+00 3.43206763e-01 2.29895487e-01 1.17533673e-02
3.94647747e-01 -2.40827546e-01 5.29083788e-01 9.11970019e-01
2.59847552e-01 4.92218196e-01 4.73641932e-01 -6.19499624e-01
-1.38139844e+00 -8.81776214e-01 3.65032852e-01 1.48167217e+00
-2.61300474e-01 -4.87406373e-01 -6.45930827e-01 -3.15294832e-01
-4.06178027e-01 9.12296355e-01 -4.98700529e-01 -3.68037134e-01
-9.95659113e-01 -8.08920681e-01 1.04762661e+00 8.33829641e-01
4.88607138e-01 -1.29094338e+00 -5.26058912e-01 4.58502144e-01
-4.79191005e-01 -1.04682469e+00 -2.90124238e-01 4.87897754e-01
-1.07730114e+00 -4.04270023e-01 -6.43459678e-01 -9.67844248e-01
4.01284099e-01 3.46859582e-02 1.25269890e+00 1.72675028e-01
1.37834549e-02 -2.71088481e-01 -1.56170920e-01 -1.82446241e-01
-8.90468776e-01 7.01914608e-01 -4.63821106e-02 -3.09677720e-01
2.87219226e-01 -6.45019591e-01 -1.38885364e-01 2.56169051e-01
-1.00323904e+00 6.64819121e-01 6.18605137e-01 1.13997459e+00
3.63817543e-01 -4.81760502e-01 5.16646087e-01 -1.13007367e+00
5.91361344e-01 -3.81101698e-01 -4.37546909e-01 3.95820230e-01
-5.33063293e-01 5.07048368e-01 9.83237088e-01 -3.93171966e-01
-9.62127388e-01 -2.65211046e-01 -3.67770284e-01 -2.78006315e-01
2.01951310e-01 7.95075655e-01 2.80078679e-01 2.99046963e-01
6.42925501e-01 8.26089442e-01 2.01377258e-01 -2.56550163e-01
2.78169274e-01 3.40818644e-01 4.51683909e-01 -6.00672901e-01
6.63330615e-01 8.30969587e-02 -6.86134100e-02 -5.12537539e-01
-7.92769313e-01 -1.15789205e-01 -2.85799146e-01 3.92120957e-01
4.25621897e-01 -7.62290597e-01 -8.98929715e-01 6.92145169e-01
-1.36134744e+00 -8.46872628e-01 -1.20523825e-01 5.50444722e-02
-9.47860837e-01 2.14152411e-01 -1.19300103e+00 -8.14980388e-01
-7.01458633e-01 -9.99106646e-01 8.23375523e-01 -1.71534851e-01
-5.00546396e-01 -1.09698999e+00 -1.32249385e-01 -8.59934986e-02
7.42702186e-01 -3.60617489e-01 1.51108205e+00 -6.07723594e-01
-6.90956891e-01 1.13307133e-01 -5.16960816e-03 2.61955082e-01
-2.10380524e-01 -2.97655344e-01 -8.43516827e-01 -2.86726624e-01
-3.23985592e-02 -4.32161123e-01 1.05866039e+00 9.54130292e-02
1.14345551e+00 -3.40679377e-01 -2.35253692e-01 7.29954064e-01
1.21551216e+00 4.65951525e-02 8.58818352e-01 2.48765454e-01
3.69704545e-01 2.11949557e-01 -3.63375992e-02 -1.00029394e-01
2.64845520e-01 5.10231078e-01 1.24035165e-01 2.07660943e-02
-1.90777078e-01 -6.81642294e-01 7.36414373e-01 1.22595179e+00
-8.82112458e-02 -3.61407131e-01 -1.05292475e+00 4.05387253e-01
-1.96474266e+00 -1.10175800e+00 4.95220125e-02 2.11462760e+00
9.96589363e-01 3.36705089e-01 1.04057388e-02 2.27227837e-01
3.94014508e-01 8.35734457e-02 -6.12582207e-01 -8.96641195e-01
-3.07929933e-01 5.39575398e-01 5.59122384e-01 5.43057382e-01
-5.16098440e-01 1.31487501e+00 6.70667076e+00 7.17032194e-01
-1.24690580e+00 1.92270786e-01 7.96204388e-01 -1.10880703e-01
-6.03337109e-01 8.72423127e-03 -8.98206949e-01 2.52635837e-01
1.50402379e+00 -1.26296148e-01 8.71288657e-01 4.24999893e-01
-1.03699423e-01 2.52295494e-01 -1.41759539e+00 8.68228078e-01
5.21834614e-03 -1.64035571e+00 1.50687397e-01 -1.40727073e-01
5.50960600e-01 6.02670968e-01 2.66226888e-01 6.46999538e-01
6.19336545e-01 -1.42025054e+00 7.68137574e-01 2.46851712e-01
9.43437994e-01 -6.03633165e-01 5.27675927e-01 7.10731566e-01
-8.82589042e-01 -2.28712887e-01 -3.54787201e-01 -4.03630286e-01
2.14397401e-01 2.42924005e-01 -9.97766197e-01 3.23480904e-01
3.84640723e-01 6.56896234e-01 -2.66089022e-01 6.07996941e-01
-4.78572994e-01 1.03192782e+00 -4.60468084e-01 -3.53309035e-01
5.39111197e-01 1.38570685e-02 3.77194971e-01 1.27434838e+00
2.34216541e-01 -2.20735762e-02 -3.00686091e-01 9.20777977e-01
-3.78118694e-01 -1.12760291e-01 -6.56841218e-01 -2.69618332e-01
3.74568194e-01 6.92149758e-01 -5.26094317e-01 -4.29203063e-01
-2.67969638e-01 1.11288071e+00 6.33920789e-01 5.31252146e-01
-7.34884679e-01 -2.15771556e-01 5.55433273e-01 -6.08244166e-02
5.10862350e-01 -4.45107698e-01 -5.66380262e-01 -1.21167934e+00
1.11936472e-01 -1.08351421e+00 2.26269901e-01 -7.86433458e-01
-8.36240411e-01 6.82256818e-01 -3.04790825e-01 -7.62619972e-01
-8.64338934e-01 -6.42296493e-01 -3.93918127e-01 1.05296636e+00
-1.43286848e+00 -1.09495699e+00 4.32567656e-01 3.09756517e-01
7.54686832e-01 -1.11157693e-01 1.05448401e+00 1.01436123e-01
-4.15684253e-01 8.33544374e-01 1.33526370e-01 7.16944784e-02
2.00251535e-01 -9.93847966e-01 1.27881408e+00 9.16028380e-01
3.90730202e-01 1.17130899e+00 6.77153826e-01 -4.10900891e-01
-1.88708603e+00 -1.00658727e+00 1.44612396e+00 -3.88728201e-01
7.94189274e-01 -7.58279145e-01 -1.13302350e+00 1.31391954e+00
1.03503890e-01 -1.25239894e-01 3.57947141e-01 4.02685881e-01
-7.73924351e-01 1.63986340e-01 -4.90488976e-01 9.25368845e-01
1.58241868e+00 -8.22060943e-01 -6.89196706e-01 2.33085752e-01
9.83014643e-01 -5.93829811e-01 -4.08326417e-01 3.07945460e-01
5.80193222e-01 -6.11988366e-01 7.27254987e-01 -9.61700082e-01
8.05668771e-01 -9.62991118e-02 -1.62753146e-02 -1.34163225e+00
-4.64801669e-01 -1.00073016e+00 -2.65042007e-01 8.75395536e-01
7.81352699e-01 -9.72731113e-01 7.51140952e-01 2.75753498e-01
-1.59063771e-01 -9.04571056e-01 -8.80004227e-01 -8.99252355e-01
3.92757922e-01 -6.05341673e-01 5.99660933e-01 6.52470767e-01
1.04266748e-01 8.71956527e-01 -3.56464893e-01 -2.64471233e-01
1.28061712e-01 1.95100948e-01 4.79143053e-01 -8.61585200e-01
-6.02102339e-01 -7.04750419e-01 -1.19507976e-01 -1.56828964e+00
4.60792571e-01 -1.41750026e+00 7.67448097e-02 -1.45028055e+00
2.84349114e-01 -4.60255027e-01 -1.95101649e-01 7.95750618e-01
-5.53365871e-02 9.15858075e-02 1.35339260e-01 2.80154496e-01
-2.98676431e-01 4.65321928e-01 9.01215911e-01 -4.19840887e-02
8.61042812e-02 3.02030388e-02 -6.48538053e-01 3.55575711e-01
7.14929223e-01 -3.45513880e-01 -4.53578770e-01 -9.53916788e-01
8.84383082e-01 1.47017494e-01 1.54294521e-01 -6.13470018e-01
3.94174606e-01 6.24034479e-02 1.18287869e-01 -4.21833605e-01
2.22068205e-01 -4.61017698e-01 2.32543141e-01 7.79355347e-01
-7.95614183e-01 4.39760923e-01 3.29195976e-01 4.51372266e-01
4.59509939e-02 -1.28959402e-01 5.20017445e-01 -2.06929818e-01
-5.83454669e-01 1.48571521e-01 -5.32469213e-01 1.93205755e-03
5.77132642e-01 -2.36279935e-01 -3.97813946e-01 -4.74134237e-01
-4.10621911e-01 1.25042517e-02 3.83082688e-01 5.78276157e-01
4.41576391e-01 -9.53949749e-01 -7.82463968e-01 2.61588216e-01
-6.81466535e-02 3.85252386e-02 -2.39022393e-02 8.89181018e-01
-4.13664877e-01 7.64284372e-01 2.90356036e-02 -5.50342321e-01
-1.07391953e+00 6.35698259e-01 4.15164411e-01 -5.78988016e-01
-7.66556561e-01 8.87643039e-01 3.39794785e-01 -8.04646671e-01
7.06428438e-02 -6.00582182e-01 3.92023265e-01 -5.23073375e-01
6.96422279e-01 3.22344974e-02 4.46024686e-01 -2.06018835e-01
-1.91719696e-01 2.04668224e-01 -4.50858802e-01 -1.91484079e-01
1.46707690e+00 1.09279513e-01 -3.76902461e-01 7.90640771e-01
1.20852387e+00 -5.72914064e-01 -8.25373650e-01 -6.84434414e-01
2.18545958e-01 2.21542224e-01 -2.70967841e-01 -9.44176495e-01
-6.06067061e-01 1.14788890e+00 -2.32531689e-03 -2.62423586e-02
9.38380122e-01 -1.17450394e-01 1.23119771e+00 7.51548886e-01
3.87427330e-01 -6.32869005e-01 -2.21403360e-01 1.30386996e+00
5.95738411e-01 -7.15896070e-01 -5.38985789e-01 -2.10711718e-01
-4.20649797e-01 1.07605767e+00 2.45475560e-01 4.60275970e-02
4.39908877e-02 6.55049980e-01 -2.61559039e-01 3.01634908e-01
-1.56717730e+00 1.45098835e-01 -3.25964019e-02 1.96623236e-01
6.53333426e-01 3.19322422e-02 -1.43259048e-01 2.15461731e-01
-4.69455928e-01 3.62840474e-01 3.70939523e-01 8.66147101e-01
-3.25579822e-01 -1.18025661e+00 -7.33321114e-03 3.69661599e-01
-5.12781024e-01 -5.38548768e-01 -2.36334458e-01 4.09114003e-01
-6.12232566e-01 5.60683191e-01 1.86124191e-01 -3.31702352e-01
-4.07028385e-02 6.06343508e-01 7.15297461e-01 -6.41896009e-01
-8.03908527e-01 -3.79697889e-01 2.20585227e-01 -3.87715489e-01
-2.77055763e-02 -5.53169191e-01 -1.43738830e+00 -6.54111326e-01
-7.40022659e-02 -4.04310599e-02 6.55253887e-01 1.24370921e+00
4.85300750e-01 4.44062591e-01 2.87085891e-01 -4.32040930e-01
-9.71311688e-01 -9.76551056e-01 2.12682672e-02 1.86506718e-01
4.07229930e-01 -6.34790882e-02 -3.42188291e-02 1.71457473e-02] | [10.808798789978027, 7.078009128570557] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.