Datasets:
J0nasW
/
paperswithcode

Dataset card Data Studio Files Community
1
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
691ad7d9-5018-4808-ba12-1eaabbc57309
compounding-the-performance-improvements-of
2001.06268
null
https://arxiv.org/abs/2001.06268v2
https://arxiv.org/pdf/2001.06268v2.pdf
Compounding the Performance Improvements of Assembled Techniques in a Convolutional Neural Network
Recent studies in image classification have demonstrated a variety of techniques for improving the performance of Convolutional Neural Networks (CNNs). However, attempts to combine existing techniques to create a practical model are still uncommon. In this study, we carry out extensive experiments to validate that carefully assembling these techniques and applying them to basic CNN models (e.g. ResNet and MobileNet) can improve the accuracy and robustness of the models while minimizing the loss of throughput. Our proposed assembled ResNet-50 shows improvements in top-1 accuracy from 76.3\% to 82.78\%, mCE from 76.0\% to 48.9\% and mFR from 57.7\% to 32.3\% on ILSVRC2012 validation set. With these improvements, inference throughput only decreases from 536 to 312. To verify the performance improvement in transfer learning, fine grained classification and image retrieval tasks were tested on several public datasets and showed that the improvement to backbone network performance boosted transfer learning performance significantly. Our approach achieved 1st place in the iFood Competition Fine-Grained Visual Recognition at CVPR 2019, and the source code and trained models are available at https://github.com/clovaai/assembled-cnn
['Jungkyu Lee', 'Hyemin Lee', 'Tae Kwan Lee', 'Kiho Hong', 'Taeryun Won', 'Geonmo Gu']
2020-01-17
null
null
null
null
['fine-grained-visual-recognition']
['computer-vision']
[ 3.56870703e-02 -3.03944796e-01 -7.08490163e-02 -4.28040653e-01 -5.77357173e-01 -5.63592315e-01 5.81680715e-01 -2.73795873e-01 -8.08459759e-01 1.04810679e+00 -2.96843708e-01 -3.86812836e-01 -1.70504317e-01 -8.95059228e-01 -1.02586818e+00 -5.19094408e-01 7.59491399e-02 -4.16880287e-03 2.56775111e-01 -2.59492993e-01 2.10555881e-01 7.16190338e-01 -1.54409552e+00 6.84309900e-01 4.82243508e-01 1.35397696e+00 9.50436890e-02 9.07017052e-01 1.41982511e-01 1.00668192e+00 -9.01951015e-01 -3.47087979e-01 3.45354408e-01 1.77297622e-01 -9.19945121e-01 -6.55797064e-01 6.82211697e-01 -3.09453428e-01 -3.55101109e-01 9.07697558e-01 7.64237225e-01 6.79857805e-02 6.64519489e-01 -1.14762342e+00 -7.48645186e-01 6.25193834e-01 -3.86480272e-01 3.73115271e-01 -2.40091622e-01 2.78675891e-02 6.87372565e-01 -8.56456876e-01 3.42967838e-01 1.05028915e+00 8.88659120e-01 7.32631147e-01 -8.52620840e-01 -1.23206031e+00 -1.29751623e-01 2.82264292e-01 -1.69827473e+00 -4.76446748e-01 1.74140744e-02 -1.85010150e-01 1.51011980e+00 3.18598360e-01 2.65362501e-01 9.67406511e-01 1.18300259e-01 4.20966655e-01 1.07215893e+00 -3.80615354e-01 -1.61164433e-01 1.77220047e-01 1.69139877e-01 8.21439147e-01 2.73446381e-01 -3.93383764e-02 -3.59184951e-01 2.25673005e-01 6.41042709e-01 -6.02471754e-02 3.30059156e-02 3.55778992e-01 -1.10507822e+00 6.66249812e-01 9.64148521e-01 5.25554001e-01 -1.77891672e-01 3.87360752e-01 6.02924228e-01 3.50148231e-01 5.70938945e-01 3.25290442e-01 -7.64810264e-01 -2.94129569e-02 -7.78017938e-01 1.79249316e-01 6.32952809e-01 9.69952703e-01 7.19386101e-01 2.27878347e-01 -2.36724824e-01 1.14266574e+00 9.84151661e-02 5.17140448e-01 6.20464206e-01 -7.57514954e-01 5.82685471e-01 4.94133502e-01 -2.28932142e-01 -9.31381345e-01 -3.61030251e-01 -7.76342392e-01 -1.31956983e+00 4.32761423e-02 3.13140124e-01 -2.03287378e-01 -1.22959530e+00 1.60615230e+00 -2.87423491e-01 1.34728983e-01 5.36353663e-02 5.93814909e-01 1.30316603e+00 7.50110149e-01 3.35843831e-01 3.71093571e-01 1.24139214e+00 -1.03371966e+00 -2.53219992e-01 -6.77642238e-04 8.47101271e-01 -8.91065240e-01 8.17732990e-01 1.91860884e-01 -9.35928822e-01 -1.08083117e+00 -1.18224752e+00 7.02601448e-02 -6.72803998e-01 3.81217241e-01 3.29098105e-01 7.98886120e-01 -1.44568467e+00 7.67746985e-01 -5.28594732e-01 -3.95986199e-01 7.36251235e-01 7.07855225e-01 -4.67769623e-01 -1.09997496e-01 -1.18124294e+00 9.27788138e-01 5.87116897e-01 1.50043011e-01 -8.02380145e-01 -7.38593459e-01 -3.90727013e-01 1.64430410e-01 -1.60038680e-01 -4.29608434e-01 1.31990409e+00 -9.07133579e-01 -1.26674259e+00 8.54743481e-01 1.10123180e-01 -8.35109055e-01 3.31344783e-01 -2.14754492e-01 -5.08299947e-01 -9.86472815e-02 -1.58090249e-01 1.24484360e+00 3.29089671e-01 -6.19037747e-01 -8.57669890e-01 -6.59877136e-02 9.82769504e-02 -2.56368537e-02 -5.12447238e-01 9.86344833e-03 -4.51198697e-01 -5.27295887e-01 -4.06281918e-01 -1.11090815e+00 3.68819200e-02 -2.52531648e-01 -2.45105147e-01 -2.40470856e-01 7.12167442e-01 -5.40671647e-01 9.16657627e-01 -2.01169395e+00 -4.48207349e-01 1.26752093e-01 1.80449665e-01 9.36495364e-01 -3.68974179e-01 1.82777256e-01 -2.01543197e-01 3.70082647e-01 1.83253884e-01 1.55677358e-02 -2.41826132e-01 -6.52190298e-02 -4.90527675e-02 2.34879151e-01 2.22897992e-01 1.16121042e+00 -2.69297570e-01 -1.18001856e-01 4.00799930e-01 8.26377988e-01 -5.12171388e-01 8.48391429e-02 2.42495045e-01 2.15436369e-01 -3.16219807e-01 5.38835466e-01 7.47018993e-01 -3.97738367e-01 -2.34615177e-01 -4.02738720e-01 -1.63334124e-02 7.99420103e-02 -9.38050508e-01 1.42205667e+00 -6.40504897e-01 9.60538447e-01 -3.68977576e-01 -1.09055388e+00 1.09193039e+00 2.60589033e-01 1.42269760e-01 -1.14203393e+00 2.84479201e-01 1.67962655e-01 2.26802081e-01 -2.85615057e-01 5.01239777e-01 3.32356840e-01 7.83989578e-02 -1.30693410e-02 3.79892796e-01 2.76263297e-01 5.93124554e-02 -6.08227449e-03 9.59832251e-01 -1.70867652e-01 1.84599191e-01 -5.07257223e-01 5.21280229e-01 8.62330850e-03 2.11281419e-01 9.94711816e-01 -1.87318712e-01 5.47176421e-01 -9.26394984e-02 -6.22427464e-01 -1.03981960e+00 -7.13319302e-01 -2.69117147e-01 1.27144086e+00 -2.43863851e-01 -4.33337927e-01 -1.02171874e+00 -5.37662745e-01 -2.56257743e-01 3.25079679e-01 -6.22994781e-01 -1.61713436e-01 -5.00943720e-01 -9.09933388e-01 1.25362146e+00 6.46401465e-01 1.25532174e+00 -1.12301314e+00 -3.63364279e-01 -7.52967000e-02 -1.51103869e-01 -1.17819762e+00 -2.09963489e-02 2.81754404e-01 -8.20420086e-01 -1.01250899e+00 -8.66111457e-01 -9.34870124e-01 4.01440859e-01 2.39528447e-01 1.05212235e+00 4.07715768e-01 -4.94950444e-01 -2.41921142e-01 -4.01820570e-01 -5.51481068e-01 -1.65136963e-01 6.78843975e-01 -1.33835793e-01 -3.31245214e-01 3.38079244e-01 -2.03919709e-01 -7.35998511e-01 4.08049941e-01 -8.27009559e-01 1.41328663e-01 6.37960911e-01 8.76785517e-01 3.91777188e-01 -8.76870677e-02 7.68977165e-01 -8.48246276e-01 4.38621700e-01 -3.95783007e-01 -4.83084500e-01 2.73033172e-01 -8.16076696e-01 -8.25568736e-02 6.14296377e-01 -3.16926271e-01 -8.67643058e-01 -9.34799612e-02 -6.16348565e-01 -3.38830173e-01 -4.88818318e-01 2.60017723e-01 1.83250695e-01 -3.59971821e-01 9.41623390e-01 1.74534425e-01 -3.34117889e-01 -4.07423943e-01 2.47317791e-01 1.00081122e+00 3.73533517e-01 -2.40454957e-01 5.37509024e-01 8.97736996e-02 -1.16853878e-01 -7.99606740e-01 -7.20500112e-01 -3.06081533e-01 -4.00364399e-01 -7.84211159e-02 7.94413626e-01 -1.29307556e+00 -8.42079699e-01 6.91185415e-01 -9.12663817e-01 -5.48312128e-01 1.37045123e-02 4.05737191e-01 -1.50694907e-01 -1.96163401e-01 -7.02195883e-01 -2.82893211e-01 -8.12096715e-01 -1.26760173e+00 8.41672063e-01 2.64838159e-01 -8.78042206e-02 -8.34498107e-01 -3.16372514e-01 3.81303221e-01 1.12991881e+00 -2.65803989e-02 5.50121188e-01 -7.34506965e-01 -3.40048850e-01 -2.94088304e-01 -7.74201393e-01 7.51445115e-01 5.09161577e-02 -1.87169034e-02 -1.42794573e+00 -3.33965600e-01 -5.02454877e-01 -4.38290805e-01 1.07067227e+00 3.30641001e-01 1.51575053e+00 -2.39409521e-01 -3.12341034e-01 7.89718449e-01 1.51513851e+00 3.63977253e-01 9.83047903e-01 5.04983366e-01 6.43603086e-01 1.76894516e-01 3.04784924e-01 1.57086566e-01 6.75626174e-02 6.56549156e-01 5.04398644e-01 -1.30496755e-01 -5.26498556e-01 8.75802618e-03 4.08839472e-02 7.52328813e-01 -4.19150114e-01 -4.34031636e-01 -9.39911902e-01 3.68582904e-01 -1.44650137e+00 -8.64064634e-01 -1.38081491e-01 1.71973348e+00 7.84764111e-01 1.58384070e-01 -1.22593224e-01 1.07530773e-01 7.16083169e-01 -7.35520199e-02 -3.27324301e-01 -4.05415207e-01 -7.78051391e-02 6.51060402e-01 8.29254270e-01 1.93857208e-01 -1.14851344e+00 1.17857075e+00 5.86784029e+00 1.24633515e+00 -1.46929502e+00 1.57751307e-01 9.76780117e-01 -8.01726431e-02 4.09189522e-01 -5.85211039e-01 -1.17084801e+00 2.97103941e-01 1.43996453e+00 2.98494995e-01 3.22953343e-01 8.12214375e-01 -2.37902075e-01 8.29771161e-02 -8.11052561e-01 1.18769526e+00 -1.72031388e-01 -1.81827414e+00 1.44009680e-01 -1.23752505e-01 8.21141839e-01 6.62001193e-01 7.40853474e-02 6.14060402e-01 2.66348898e-01 -1.55760550e+00 7.03408062e-01 4.22022849e-01 1.14367557e+00 -8.41669500e-01 1.13026500e+00 2.97719032e-01 -1.16834569e+00 1.04784459e-01 -5.85316300e-01 -1.30475372e-01 -4.79532778e-01 4.27408755e-01 -1.12964141e+00 4.17852223e-01 1.15455246e+00 6.42402291e-01 -8.65030706e-01 9.41921771e-01 1.08888343e-01 7.06858337e-01 -1.80378437e-01 -3.35250825e-01 2.92203754e-01 5.00183165e-01 -1.46263525e-01 1.59392941e+00 2.95109838e-01 -1.69146627e-01 -2.94420958e-01 3.89889419e-01 -6.28715515e-01 9.42426249e-02 -3.89572144e-01 2.08836094e-01 3.16820651e-01 1.31973040e+00 -6.97569370e-01 -4.26135421e-01 -2.52538592e-01 8.16668928e-01 4.66074079e-01 2.76938558e-01 -1.06291139e+00 -6.73790395e-01 6.32946849e-01 -1.31240457e-01 2.97342092e-01 5.55916652e-02 -2.09867880e-01 -9.51345384e-01 -1.50075644e-01 -9.34280813e-01 2.74467200e-01 -8.16534877e-01 -9.70178485e-01 1.02662504e+00 -1.22099437e-01 -1.00515473e+00 5.56514263e-02 -8.40594947e-01 -2.12213650e-01 1.09954631e+00 -1.68793726e+00 -1.07332635e+00 -6.82946920e-01 7.77093649e-01 6.47552848e-01 -5.35865068e-01 1.00067687e+00 7.11545289e-01 -5.51626623e-01 1.13431644e+00 1.87454700e-01 4.52918679e-01 7.47276306e-01 -8.07293057e-01 5.74178696e-01 5.42788267e-01 6.31977096e-02 5.28845251e-01 1.40284598e-01 -2.24113643e-01 -1.03518546e+00 -1.65643632e+00 7.51803696e-01 -1.67143181e-01 2.99910367e-01 -3.71056706e-01 -7.32417226e-01 6.21478200e-01 2.69249231e-01 2.56299555e-01 4.27851617e-01 -1.16722574e-02 -6.46564364e-01 -4.71125513e-01 -1.26523519e+00 4.72913414e-01 9.97155190e-01 -4.57495928e-01 -8.19541290e-02 2.96013743e-01 6.20338500e-01 -4.54070926e-01 -1.01514792e+00 7.08321869e-01 7.97152638e-01 -7.30720699e-01 1.05054176e+00 -5.93274474e-01 3.94537866e-01 -9.20530334e-02 -5.76448083e-01 -1.19054353e+00 -4.06015903e-01 3.02655920e-02 2.15661049e-01 1.13421404e+00 6.41104400e-01 -8.65518093e-01 7.75828362e-01 6.84691891e-02 -1.00519769e-01 -7.46700764e-01 -7.90712357e-01 -7.30685413e-01 2.73283213e-01 -5.52042127e-01 4.90997314e-01 6.54090524e-01 -7.01618910e-01 2.34541938e-01 -2.28879839e-01 4.17216159e-02 2.98134476e-01 -4.56047893e-01 6.46500707e-01 -1.08140707e+00 -1.46687731e-01 -3.37138146e-01 -6.03281438e-01 -7.74151504e-01 4.86355498e-02 -1.11479974e+00 -1.33220166e-01 -1.42321897e+00 2.62144059e-01 -7.11055100e-01 -5.95817924e-01 7.66936660e-01 1.79866895e-01 1.08904922e+00 3.11254114e-01 2.59851187e-01 -5.56515038e-01 6.74658790e-02 1.06116378e+00 -3.02377939e-01 1.78949028e-01 -7.58809447e-02 -7.41808712e-01 5.04897118e-01 1.48586500e+00 -4.16233987e-01 -4.08483684e-01 -6.70548141e-01 1.08689018e-01 -3.59880984e-01 4.52489316e-01 -1.38885665e+00 1.68374568e-01 2.82998085e-01 7.91511536e-01 -3.35956305e-01 3.36359322e-01 -5.79297781e-01 3.02407950e-01 6.19069517e-01 -5.09517133e-01 2.85272263e-02 7.60059476e-01 7.37254694e-02 -2.58979470e-01 -5.89690804e-02 9.46965933e-01 -8.99842605e-02 -9.59089339e-01 3.70865881e-01 -1.60218135e-01 -6.21821508e-02 9.37750936e-01 -7.89206326e-02 -7.12350965e-01 -6.58826437e-03 -6.17820978e-01 -2.30733141e-01 -1.48309618e-01 5.94912112e-01 4.80026066e-01 -1.21275270e+00 -8.31574142e-01 1.99369565e-01 3.40693146e-02 -3.14712733e-01 4.11342055e-01 6.53521061e-01 -8.32245588e-01 9.28990781e-01 -6.16873920e-01 -6.07010543e-01 -1.52671981e+00 6.80467859e-02 6.57100558e-01 -1.99043840e-01 -2.79167861e-01 1.11440897e+00 5.76298237e-02 -5.08717358e-01 4.96345013e-01 -5.00721753e-01 -3.33610147e-01 -1.76377624e-01 6.29203022e-01 4.56438899e-01 7.16667771e-01 -4.94373590e-01 -4.81044859e-01 5.84118187e-01 -2.77968347e-01 2.86832631e-01 1.46064401e+00 1.52012527e-01 4.92913313e-02 2.87959948e-02 1.62536883e+00 -4.26552206e-01 -9.15367067e-01 -8.01925659e-02 -3.23115379e-01 -1.23257153e-01 5.73880225e-02 -1.12616765e+00 -1.42364275e+00 9.57906246e-01 1.06122267e+00 4.34854366e-02 1.26755357e+00 -1.30047813e-01 6.11948311e-01 7.11440504e-01 3.48869085e-01 -6.40579879e-01 -2.25534067e-01 8.31581175e-01 8.32867682e-01 -1.21936429e+00 -2.11168140e-01 -2.44682580e-01 -1.97307393e-01 1.08501852e+00 7.10175633e-01 -3.42824250e-01 8.82160068e-01 3.47324610e-01 5.36507294e-02 3.21501680e-02 -7.78939784e-01 4.58158739e-02 3.62264514e-01 5.00871778e-01 6.77197039e-01 2.00265825e-01 3.67395533e-03 4.01395440e-01 -3.59964520e-01 2.49199495e-01 1.60116106e-01 7.50896335e-01 -3.86860013e-01 -8.21982086e-01 -6.71716928e-02 6.11726940e-01 -9.52583969e-01 -3.73231202e-01 -1.49598017e-01 8.92899334e-01 3.50459129e-01 9.47197437e-01 1.57538176e-01 -7.33101487e-01 3.76818091e-01 -3.86097468e-02 3.69538575e-01 -4.52562720e-01 -9.27889526e-01 -2.68397599e-01 2.74128228e-01 -5.73106408e-01 -4.91688639e-01 -3.84461917e-02 -1.02677429e+00 -7.45639920e-01 -2.48822734e-01 4.85100457e-03 9.50400472e-01 8.49119306e-01 5.02850175e-01 9.28275168e-01 2.42410958e-01 -6.92051589e-01 -3.66645247e-01 -1.20996821e+00 -3.25625956e-01 7.33003393e-02 2.32923955e-01 -3.37527275e-01 -8.57059881e-02 1.33068115e-01]
[9.289335250854492, 2.2759459018707275]
fc390fa5-6f3c-483f-8add-debc62c4eb62
don-t-lose-yourself-empathetic-response
2210.03884
null
https://arxiv.org/abs/2210.03884v2
https://arxiv.org/pdf/2210.03884v2.pdf
Don't Lose Yourself! Empathetic Response Generation via Explicit Self-Other Awareness
As a critical step to achieve human-like chatbots, empathetic response generation has attained increasing interests. Previous attempts are incomplete and not sufficient enough to elicit empathy because they only focus on the initial aspect of empathy to automatically mimic the feelings and thoughts of the user via other-awareness. However, they ignore to maintain and take the own views of the system into account, which is a crucial process to achieve the empathy called self-other awareness. To this end, we propose to generate Empathetic response with explicit Self-Other Awareness (EmpSOA). Specifically, three stages, self-other differentiation, self-other modulation and self-other generation, are devised to clearly maintain, regulate and inject the self-other aware information into the process of empathetic response generation. Both automatic and human evaluations on the benchmark dataset demonstrate the superiority of EmpSOA to generate more empathetic responses.
['Bing Qin', 'Xin Lu', 'Yanyan Zhao', 'Weixiang Zhao']
2022-10-08
null
null
null
null
['empathetic-response-generation']
['natural-language-processing']
[-3.78057450e-01 2.82819778e-01 7.97674358e-02 -5.10027766e-01 -5.07105850e-02 -3.57455492e-01 6.99176133e-01 7.70456204e-03 -1.54369295e-01 7.52756059e-01 4.55571264e-01 3.73912603e-01 1.41915917e-01 -5.91325283e-01 3.44933271e-01 -4.43265557e-01 6.98768377e-01 2.71076828e-01 -2.71011829e-01 -7.23865747e-01 6.55174851e-01 2.55448490e-01 -1.16877162e+00 4.73413736e-01 1.18013692e+00 4.43140626e-01 -1.22650668e-01 7.55013704e-01 -1.61198810e-01 1.51219749e+00 -8.26329470e-01 -4.50567484e-01 -1.92228779e-01 -1.11667240e+00 -1.13884473e+00 -1.12620100e-01 -5.92707038e-01 -4.17861700e-01 2.61905134e-01 9.68115151e-01 6.00582898e-01 2.31304422e-01 4.83464420e-01 -1.59489429e+00 -7.69807339e-01 6.00554645e-01 -3.90559524e-01 -1.20639175e-01 7.33017445e-01 4.45476085e-01 5.79415560e-01 -2.85364538e-01 2.94472843e-01 1.24821305e+00 5.73475718e-01 1.24722528e+00 -8.07204843e-01 -6.21548116e-01 -4.10752684e-01 1.35229364e-01 -7.73463547e-01 -4.39646959e-01 9.68799293e-01 -4.77967113e-01 5.58488607e-01 4.25395548e-01 8.07044506e-01 1.30298626e+00 1.29051507e-01 5.16333044e-01 1.46778703e+00 -3.78501207e-01 3.13755363e-01 8.22052836e-01 2.74697542e-01 5.42548560e-02 -3.44312847e-01 3.46857198e-02 -6.34328723e-01 -4.41851988e-02 9.59114313e-01 -3.91695201e-01 -1.39524832e-01 1.02272652e-01 -1.11541820e+00 9.35953677e-01 3.16870749e-01 5.56637883e-01 -1.00504446e+00 3.51255350e-02 4.76157129e-01 4.80631649e-01 1.47130847e-01 1.04820085e+00 2.70390123e-01 -8.08295190e-01 -6.10218942e-01 -8.72630998e-03 1.08447266e+00 5.69108486e-01 4.84033972e-01 2.40834095e-02 -4.82170433e-01 6.43953681e-01 -3.62925641e-02 1.52777165e-01 7.20911860e-01 -1.33738649e+00 -2.01936498e-01 1.14436996e+00 6.40146017e-01 -9.77486551e-01 -2.84877926e-01 -1.78971156e-01 -5.59617937e-01 4.05250460e-01 1.50549516e-01 -5.14937103e-01 -6.34312630e-04 1.75084341e+00 6.24859571e-01 -2.58031487e-01 4.93375659e-01 1.14547098e+00 9.72725093e-01 5.34824073e-01 3.92323166e-01 -3.37448120e-01 1.30704880e+00 -1.16816115e+00 -9.65105474e-01 -2.55250245e-01 7.15674698e-01 -8.35479558e-01 1.21499217e+00 3.82851481e-01 -1.28979135e+00 -5.41445076e-01 -8.17371428e-01 2.17388675e-01 1.48307294e-01 2.09098235e-01 8.31324458e-01 5.63898385e-01 -8.68272901e-01 5.55858493e-01 -1.95893556e-01 -6.02262199e-01 -3.56586836e-02 7.43792728e-02 -4.80884105e-01 7.23416924e-01 -1.61733854e+00 1.43196356e+00 1.34870589e-01 8.60267654e-02 -2.54878104e-01 -5.63159823e-01 -3.55485171e-01 3.25714797e-02 -1.70115605e-01 -7.41237104e-01 1.50257432e+00 -1.60300744e+00 -2.13042378e+00 1.03349257e+00 3.24745208e-01 -3.12031120e-01 7.45588005e-01 -1.16070509e-01 -3.10297757e-01 3.52269441e-01 -1.17401183e-01 7.36312926e-01 5.96328914e-01 -1.33626211e+00 -2.89032161e-01 2.03411561e-02 2.42476478e-01 4.87726092e-01 -4.46522206e-01 6.72724783e-01 4.27104115e-01 -1.68880120e-01 -3.74447197e-01 -6.30269825e-01 -2.47785419e-01 -2.06685498e-01 -3.22135687e-01 -2.22171903e-01 4.96648759e-01 -5.14736056e-01 1.00525570e+00 -1.96920323e+00 -3.15550208e-01 -4.13443707e-02 4.74389046e-01 6.50832713e-01 -8.04049373e-02 1.03035104e+00 -2.29363218e-01 -2.30197817e-01 3.30177486e-01 -5.57655804e-02 2.24798605e-01 -1.32078826e-01 -2.73941547e-01 1.83421537e-01 2.42923722e-01 9.80772913e-01 -9.94793177e-01 -8.02971363e-01 2.83684731e-01 4.31084424e-01 -6.50394320e-01 7.02327192e-01 1.65913105e-01 9.97409165e-01 -7.26984560e-01 -5.40832207e-02 5.67875445e-01 -1.32775590e-01 -4.42404971e-02 1.30489334e-01 -2.87247032e-01 1.99415773e-01 -7.43925631e-01 1.01539218e+00 -6.17812097e-01 1.41415820e-01 1.91955015e-01 -5.80241859e-01 1.35675871e+00 6.13151073e-01 6.11279368e-01 -6.84724331e-01 6.09445095e-01 1.83294105e-05 4.15263176e-01 -7.54872024e-01 5.51751256e-01 -6.44124031e-01 -4.89194058e-02 1.07938743e+00 -3.91057789e-01 -2.17445120e-01 -1.07791126e-01 4.39719737e-01 8.16943228e-01 1.97303981e-01 8.02316248e-01 -8.03749412e-02 8.79697025e-01 2.02627301e-01 3.68463486e-01 2.90481836e-01 -6.68459952e-01 3.24930996e-01 8.31268668e-01 -2.05873400e-01 -6.12296581e-01 -6.02001309e-01 4.54456538e-01 9.83702600e-01 3.54759187e-01 -1.05225407e-01 -1.22881019e+00 -4.20955539e-01 -4.79674280e-01 1.23882878e+00 -7.24048138e-01 -5.59153318e-01 -3.35863739e-01 -1.64685220e-01 6.03900969e-01 3.62131655e-01 7.94544280e-01 -1.80250800e+00 -1.30188429e+00 4.16473389e-01 -5.05548477e-01 -8.90556276e-01 -3.57787490e-01 -2.23528847e-01 -4.78632241e-01 -7.53337801e-01 -5.37873566e-01 -2.13410854e-01 5.42060792e-01 2.09980071e-01 7.12636769e-01 3.76967371e-01 -1.74029190e-02 3.08433652e-01 -6.77993119e-01 -3.27068776e-01 -1.08264911e+00 -3.46100122e-01 -2.10664198e-01 1.46216884e-01 4.71229643e-01 -8.06255579e-01 -7.48321354e-01 4.76175904e-01 -5.15600562e-01 5.80781639e-01 5.69117606e-01 6.07102871e-01 -4.27657127e-01 -5.57466626e-01 9.69741464e-01 -6.75951004e-01 1.37656701e+00 -4.22439039e-01 3.79753023e-01 1.22932643e-01 -4.63960588e-01 -2.90110439e-01 9.02438760e-01 -7.05709696e-01 -1.62875986e+00 -2.41407752e-01 -2.67676711e-01 -3.14068161e-02 -3.70567441e-01 1.24546558e-01 1.88285566e-03 -1.08979166e-01 8.92199576e-01 2.06717402e-01 4.58643615e-01 2.96057612e-02 4.98518646e-01 1.03140509e+00 5.26524067e-01 -7.50101089e-01 5.35837710e-01 2.70460606e-01 -5.20902455e-01 -3.15333575e-01 -7.25630522e-01 -5.14220715e-01 -4.31827933e-01 -7.06466496e-01 8.38439703e-01 -7.18213081e-01 -1.41073740e+00 4.83269572e-01 -1.46765149e+00 -3.01765829e-01 -4.23563123e-01 4.78440642e-01 -7.95901060e-01 4.79767740e-01 -7.33195364e-01 -1.09613156e+00 -9.14060414e-01 -7.01107681e-01 3.85100693e-01 7.64076471e-01 -1.17913270e+00 -8.48429084e-01 3.69183540e-01 5.31080365e-01 7.91566610e-01 2.94579595e-01 4.98991787e-01 -9.24616694e-01 1.96642235e-01 -4.02633071e-01 -2.17295706e-01 3.25429529e-01 1.96762308e-01 7.41416886e-02 -8.77527535e-01 3.13013703e-01 5.34627199e-01 -7.58947849e-01 -1.09207876e-01 -4.76497114e-01 2.43330196e-01 -6.43413782e-01 1.14160068e-01 1.95756871e-02 8.60016346e-01 2.33492047e-01 9.35692906e-01 1.61441132e-01 5.83304316e-02 1.20814192e+00 1.10062385e+00 9.04319167e-01 4.66069639e-01 4.31349963e-01 4.78105992e-01 -2.91103601e-01 1.01026632e-01 -3.15605134e-01 5.46972394e-01 7.09625065e-01 -2.47876897e-01 2.45320693e-01 -5.66540122e-01 2.83533543e-01 -1.88696384e+00 -1.38808787e+00 -5.61791658e-01 1.86302793e+00 1.14622307e+00 -3.22462291e-01 1.94839537e-01 -3.07932086e-02 9.53181684e-01 1.01357531e-02 -4.12660271e-01 -9.97607887e-01 1.61709398e-01 9.66033861e-02 -4.09320414e-01 4.07908440e-01 -2.88543522e-01 1.23577476e+00 5.67520046e+00 2.40817904e-01 -1.17651248e+00 1.25513241e-01 2.99392879e-01 1.24250226e-01 -3.15510601e-01 2.27021694e-01 -1.50652677e-01 4.05345470e-01 6.23376846e-01 -5.09654760e-01 2.93303341e-01 9.75740194e-01 6.73952222e-01 -2.55527109e-01 -9.10889864e-01 8.27254057e-01 -8.27067047e-02 -6.98943317e-01 -3.13596249e-01 -3.51364940e-01 3.39891672e-01 -9.36274409e-01 -3.38878870e-01 3.71230483e-01 5.54983675e-01 -9.62938011e-01 5.43758273e-01 7.17487395e-01 4.00322556e-01 -7.01758623e-01 8.25173676e-01 7.94808328e-01 -4.69280481e-01 9.49706808e-02 -2.50894934e-01 -6.15135789e-01 4.07746196e-01 -1.78206563e-02 -8.04209113e-01 3.08393419e-01 2.89300323e-01 2.98943728e-01 -2.05906257e-01 6.68095827e-01 -5.91591239e-01 2.97536612e-01 3.00310940e-01 -4.59847242e-01 1.86207786e-01 -4.18350399e-01 4.02805120e-01 8.17551017e-01 6.08140975e-02 4.78876799e-01 -3.84512931e-01 1.26180828e+00 3.85283828e-01 4.36706245e-01 -2.18632892e-01 -2.03810424e-01 4.85521853e-01 1.80201817e+00 -2.30479509e-01 -2.66790748e-01 2.72076190e-01 1.11851037e+00 2.90931016e-01 -1.93369631e-02 -1.26711714e+00 -3.25650871e-01 6.49480999e-01 -1.58691362e-01 -5.59126556e-01 3.07150900e-01 -3.95507514e-01 -8.11377406e-01 -3.41153413e-01 -1.20086718e+00 4.24125046e-02 -1.15801477e+00 -1.29057133e+00 5.64106226e-01 -3.66047621e-01 -1.19800758e+00 -4.54126388e-01 -1.05385117e-01 -1.46134543e+00 1.00626612e+00 -1.02753401e+00 -1.32160807e+00 -8.07542324e-01 5.93808889e-01 1.03642516e-01 1.95380941e-01 9.95259047e-01 -2.96586640e-02 -4.06915814e-01 4.83224541e-01 -8.20713401e-01 -4.03338335e-02 1.15614724e+00 -7.97689319e-01 -2.94787548e-02 4.76392478e-01 -6.16590977e-01 9.19386983e-01 1.12083292e+00 -3.59411895e-01 -8.97193313e-01 -3.62916142e-01 1.17318940e+00 -3.08736086e-01 7.11331725e-01 6.49764240e-02 -8.23804915e-01 3.19002062e-01 7.14119732e-01 -6.85711145e-01 8.05745780e-01 -2.27523506e-01 -5.98537326e-02 1.32676914e-01 -1.39342988e+00 8.74690354e-01 7.38051534e-01 -2.02529863e-01 -8.77477646e-01 7.95439854e-02 4.49354708e-01 -1.09515660e-01 -8.11299443e-01 9.52670574e-02 4.46091652e-01 -1.61866069e+00 6.12141728e-01 -7.36535609e-01 8.81699979e-01 3.80433351e-02 4.87232894e-01 -1.27574146e+00 -9.58338752e-02 -1.26343238e+00 4.66970384e-01 1.65067613e+00 -1.35643408e-01 -9.80281949e-01 7.46456444e-01 1.04225826e+00 -2.21208800e-02 -4.68692780e-01 -5.53910553e-01 -3.35476905e-01 1.73490644e-01 -7.52206966e-02 7.43259132e-01 1.31729198e+00 1.06716359e+00 5.42778492e-01 -7.93320358e-01 -3.07080090e-01 1.53910697e-01 2.21169978e-01 1.38451648e+00 -1.11844826e+00 -1.81628108e-01 -6.14176691e-01 -4.75372607e-03 -9.02507842e-01 1.33904859e-01 -4.02470767e-01 1.71292290e-01 -1.68806005e+00 3.13691825e-01 -2.67586291e-01 1.06159382e-01 5.32517433e-01 -3.83050859e-01 -1.26513109e-01 3.61569256e-01 3.99123818e-01 -5.35670877e-01 8.16637814e-01 1.63473880e+00 5.89945614e-01 -3.76705974e-01 1.97199471e-02 -1.21021712e+00 7.22441018e-01 1.09894800e+00 -3.25184166e-01 -3.93494397e-01 3.20402443e-01 8.80670473e-02 5.95384896e-01 4.97820020e-01 -9.09758151e-01 4.40667421e-01 -6.92408800e-01 -1.90978333e-01 -1.62201494e-01 5.26252747e-01 -5.66631973e-01 1.48787275e-01 7.61345804e-01 -4.88099873e-01 2.86226254e-02 -2.81960279e-01 -1.37249351e-01 -4.19566214e-01 -3.68383139e-01 1.12010968e+00 -3.08263898e-01 -4.37653273e-01 -3.13255429e-01 -6.29152119e-01 3.58482152e-02 1.47842336e+00 -5.06655574e-01 -6.08863235e-01 -1.04249871e+00 -3.86867434e-01 4.01976109e-01 5.34501076e-01 3.01075608e-01 5.28005183e-01 -1.17831314e+00 -6.67422295e-01 -2.51741946e-01 1.18828446e-01 -7.17192173e-01 7.44330704e-01 1.19429970e+00 -6.46859705e-01 1.48771331e-01 -8.63560677e-01 -9.73303691e-02 -1.28420448e+00 3.63106817e-01 4.95295525e-01 -3.58190924e-01 -4.44986910e-01 6.01267099e-01 2.07842350e-01 -5.81545651e-01 -1.60086930e-01 5.24111509e-01 -6.72324538e-01 -2.54116897e-02 5.04388869e-01 4.17284280e-01 -6.57750726e-01 -5.71443021e-01 -1.58379674e-01 1.60361543e-01 2.38334551e-01 -3.17270398e-01 9.92577016e-01 -1.13371000e-01 -3.53098273e-01 2.20167622e-01 4.06378180e-01 2.05506757e-01 -1.02304864e+00 2.43407160e-01 -7.90192708e-02 -4.49943632e-01 -6.41544878e-01 -9.98364806e-01 -5.86096346e-01 1.08514023e+00 -2.36194044e-01 3.31999302e-01 1.02533996e+00 -3.29038084e-01 9.93843019e-01 1.25654772e-01 3.07477921e-01 -1.36140156e+00 6.20754957e-01 2.71918297e-01 1.02806699e+00 -1.03297806e+00 -1.73669577e-01 -4.44260061e-01 -1.59333551e+00 1.11025190e+00 1.33516705e+00 -3.37275341e-02 -1.07361719e-01 1.24622092e-01 6.85839474e-01 -8.21775720e-02 -1.05301225e+00 -1.51536074e-02 -2.10333079e-01 7.10818589e-01 5.75430632e-01 -9.67948586e-02 -8.39171529e-01 1.08363092e+00 -5.14681160e-01 1.66618571e-01 7.97848523e-01 5.88759780e-01 -4.75451887e-01 -1.11725938e+00 -3.53494257e-01 -1.89999148e-01 -2.04118282e-01 1.74713865e-01 -1.07896519e+00 8.27957213e-01 -1.72161922e-01 1.41487980e+00 -2.75283694e-01 -4.11605000e-01 2.69213706e-01 -2.84130815e-02 1.22932717e-01 -5.46451628e-01 -1.53753090e+00 -4.36264426e-01 3.55839014e-01 -5.50217092e-01 -5.15789747e-01 -2.80785441e-01 -1.64481080e+00 -7.41268456e-01 -1.51164085e-01 5.89999557e-01 5.03742993e-01 9.32254672e-01 3.55912209e-01 1.48799002e-01 1.03124201e+00 -5.59560895e-01 -1.02731705e+00 -1.22090447e+00 -2.49570653e-01 8.41696799e-01 -3.75477999e-01 -2.38703385e-01 -6.10857546e-01 -2.75283873e-01]
[13.165223121643066, 7.603718280792236]
f26116c6-8f46-466d-995d-d60d6d6be295
modular-proximal-optimization-for
1411.0589
null
http://arxiv.org/abs/1411.0589v3
http://arxiv.org/pdf/1411.0589v3.pdf
Modular proximal optimization for multidimensional total-variation regularization
We study \emph{TV regularization}, a widely used technique for eliciting structured sparsity. In particular, we propose efficient algorithms for computing prox-operators for $\ell_p$-norm TV. The most important among these is $\ell_1$-norm TV, for whose prox-operator we present a new geometric analysis which unveils a hitherto unknown connection to taut-string methods. This connection turns out to be remarkably useful as it shows how our geometry guided implementation results in efficient weighted and unweighted 1D-TV solvers, surpassing state-of-the-art methods. Our 1D-TV solvers provide the backbone for building more complex (two or higher-dimensional) TV solvers within a modular proximal optimization approach. We review the literature for an array of methods exploiting this strategy, and illustrate the benefits of our modular design through extensive suite of experiments on (i) image denoising, (ii) image deconvolution, (iii) four variants of fused-lasso, and (iv) video denoising. To underscore our claims and permit easy reproducibility, we provide all the reviewed and our new TV solvers in an easy to use multi-threaded C++, Matlab and Python library.
['Álvaro Barbero', 'Suvrit Sra']
2014-11-03
null
null
null
null
['video-denoising', 'image-deconvolution']
['computer-vision', 'computer-vision']
[ 3.41721356e-01 9.01429206e-02 3.28413963e-01 -1.16646171e-01 -1.14432740e+00 -4.35285270e-01 -5.58513999e-02 -1.87891215e-01 -1.93018749e-01 6.51029825e-01 3.27117771e-01 -4.31644261e-01 -3.43614608e-01 -4.61639374e-01 -8.99273932e-01 -1.09110236e+00 -4.54599947e-01 1.49272025e-01 -2.42192388e-01 -3.42251688e-01 1.81669429e-01 1.16230659e-01 -1.36378109e+00 1.29399121e-01 9.00978923e-01 1.23476958e+00 4.79002185e-02 2.89262235e-01 9.87408683e-03 6.74536169e-01 5.82218915e-02 -4.44422305e-01 5.77619553e-01 -3.92588824e-01 -8.33585620e-01 9.61815864e-02 5.82209289e-01 -1.84129719e-02 -2.53297426e-02 9.84573185e-01 5.10677814e-01 2.81904697e-01 3.30128253e-01 -7.43968368e-01 -5.32258272e-01 4.47220922e-01 -7.09303856e-01 4.75751795e-02 4.06281590e-01 6.77070320e-02 1.01872742e+00 -1.25919104e+00 6.90473735e-01 9.76038516e-01 1.25635040e+00 1.94800183e-01 -1.63224185e+00 -1.10511526e-01 2.33029380e-01 -3.34681183e-01 -1.09057415e+00 -5.04259706e-01 7.03618348e-01 -6.52215183e-01 7.48281419e-01 5.40771723e-01 7.39579856e-01 8.13016832e-01 3.05698868e-02 7.83048749e-01 1.34223318e+00 -4.32591349e-01 1.11145064e-01 -1.91992447e-01 1.55182272e-01 1.00016248e+00 3.70800346e-02 8.64823535e-02 -7.46271789e-01 -5.80439627e-01 8.20802808e-01 -2.71688193e-01 -5.99363446e-01 -5.55346489e-01 -1.34123421e+00 1.07682312e+00 6.24021590e-02 1.86235473e-01 -1.76871255e-01 3.85648638e-01 6.14945889e-01 3.70677233e-01 1.00223553e+00 2.73751020e-01 -3.30484599e-01 -5.74206673e-02 -1.18426514e+00 3.62596780e-01 7.90185869e-01 8.97561371e-01 9.14074659e-01 4.73873168e-01 -8.84029716e-02 8.26970935e-01 2.03554258e-01 4.34369385e-01 -1.92261368e-01 -1.41871464e+00 2.07227379e-01 -1.56934470e-01 -1.07248127e-01 -1.21250737e+00 -2.84624159e-01 -6.01086557e-01 -9.67076540e-01 2.17661366e-01 4.28947181e-01 -2.16018230e-01 -5.66404045e-01 1.65778816e+00 3.55071962e-01 4.06203926e-01 -4.02809381e-01 1.02332819e+00 7.53538191e-01 6.20589674e-01 -2.34994575e-01 -8.15916657e-01 1.18144429e+00 -9.93680537e-01 -7.17369080e-01 -1.50595196e-02 7.74163544e-01 -9.32591498e-01 8.10897946e-01 6.40721142e-01 -1.56590068e+00 -1.54404327e-01 -5.52105010e-01 -3.18731040e-01 6.86199442e-02 -1.54865161e-01 9.52252626e-01 5.08537471e-01 -1.15978265e+00 8.18589687e-01 -7.98899710e-01 -2.76129544e-01 4.36217010e-01 2.55357891e-01 -1.47569641e-01 4.14742902e-02 -7.74016619e-01 6.28515780e-01 -3.96588176e-01 3.92143130e-01 -8.43321323e-01 -1.16053140e+00 -1.01271427e+00 -2.73924708e-01 4.43272024e-01 -9.19093192e-01 7.73495674e-01 -8.88983607e-01 -1.33302367e+00 1.35140181e+00 -3.78284067e-01 -3.39871496e-01 6.33119702e-01 -8.21666569e-02 1.54885933e-01 -1.05142146e-02 4.48325872e-02 1.76563770e-01 1.11967838e+00 -1.16836047e+00 1.07452288e-01 -3.11808228e-01 -2.57676721e-01 4.26580794e-02 -1.29799321e-01 8.92181247e-02 -1.80924624e-01 -1.21668911e+00 3.19193035e-01 -7.04803944e-01 -4.66577798e-01 3.13781798e-01 -3.58902097e-01 1.65024325e-01 5.11078119e-01 -7.22092390e-01 1.37946212e+00 -2.24438930e+00 6.71367943e-01 3.21850032e-01 4.96412575e-01 -4.40667011e-02 7.85709918e-02 5.29157162e-01 -1.94750905e-01 -5.24785072e-02 -8.07266474e-01 -6.73986852e-01 -5.57681397e-02 2.83248365e-01 -3.73224080e-01 9.02053118e-01 -2.80101061e-01 5.62161684e-01 -7.74493098e-01 -4.42673445e-01 2.07557693e-01 6.18844807e-01 -8.08865607e-01 -5.53493537e-02 -6.54198900e-02 8.24939191e-01 -5.12312949e-01 6.14456177e-01 8.48032117e-01 -1.60632357e-01 2.50462908e-02 -5.52437603e-01 -5.91008246e-01 -1.00407451e-02 -1.45851970e+00 1.98777711e+00 -1.72027171e-01 4.41542745e-01 1.06538916e+00 -1.56451082e+00 6.95643842e-01 3.40757012e-01 9.64790046e-01 -3.73113841e-01 7.00554401e-02 5.06181359e-01 -7.93093503e-01 -5.37660122e-01 2.34826043e-01 -4.94392335e-01 2.09813759e-01 4.01728362e-01 1.26902297e-01 -3.14336509e-01 2.74431467e-01 1.84279010e-01 8.97526026e-01 3.72495264e-01 3.52873206e-02 -1.00180137e+00 5.19834995e-01 1.69132859e-01 6.34412289e-01 7.73361981e-01 -2.23143384e-01 9.07184362e-01 5.66106021e-01 -7.77899623e-01 -9.92609680e-01 -8.28042448e-01 -6.11855924e-01 1.24314022e+00 -3.21162313e-01 -5.33602834e-01 -8.03588629e-01 -7.87708685e-02 1.97057687e-02 2.84115762e-01 -8.08569252e-01 4.87235248e-01 -7.90349305e-01 -9.91008818e-01 5.05056083e-01 2.91764021e-01 1.87270939e-01 -5.18990219e-01 -2.85642415e-01 2.01517358e-01 -3.12007964e-01 -9.09115434e-01 -6.64894938e-01 3.15792084e-01 -1.09411979e+00 -9.06319201e-01 -1.02999580e+00 -9.18951452e-01 4.80576873e-01 2.50044435e-01 1.16318858e+00 2.07016811e-01 -3.87333006e-01 7.59522319e-01 -1.00045808e-01 -1.62813559e-01 2.20969886e-01 -4.48672205e-01 1.26307666e-01 8.59817192e-02 -1.71917289e-01 -8.38168025e-01 -6.48305714e-01 2.62742728e-01 -8.64185750e-01 1.27066642e-01 -2.74469763e-01 9.54575896e-01 1.06567526e+00 -4.34809417e-01 9.08804834e-02 -9.28715229e-01 4.59044129e-01 -3.54692191e-01 -7.33461022e-01 4.44416814e-02 -3.65199357e-01 -1.01651326e-01 4.20010954e-01 -4.73991521e-02 -8.69303167e-01 2.50235260e-01 -4.81380194e-01 -4.54502076e-01 5.39586008e-01 7.55748332e-01 4.61550742e-01 -7.25803018e-01 7.20810115e-01 1.76998988e-01 2.43260667e-01 -5.66293597e-01 5.62138081e-01 3.54852676e-02 3.83041531e-01 -1.08982265e+00 6.84330940e-01 9.86954808e-01 2.21019819e-01 -1.23653865e+00 -9.83332455e-01 -3.27292085e-01 -2.76478499e-01 -1.40047863e-01 9.72432256e-01 -8.41194987e-01 -8.74910116e-01 3.58444810e-01 -9.32528138e-01 -5.11828423e-01 -5.52238762e-01 2.46594399e-01 -9.40353036e-01 6.34566724e-01 -6.85470343e-01 -7.83987105e-01 -4.04990792e-01 -1.34791613e+00 1.26043022e+00 -4.02261138e-01 -3.75306525e-04 -1.23396456e+00 3.29732448e-01 4.98612165e-01 7.54859090e-01 5.62694252e-01 7.77046025e-01 7.53970593e-02 -4.33748186e-01 3.06504965e-01 -1.03509285e-01 4.39200103e-01 -4.94683266e-01 -5.46081811e-02 -9.23178315e-01 -3.36913466e-01 5.07034779e-01 -3.75256956e-01 9.43963826e-01 9.60435748e-01 1.37131512e+00 -2.60006219e-01 -7.36396164e-02 1.38223147e+00 1.59033573e+00 -6.51137829e-01 5.20855188e-01 8.16881359e-02 7.02454686e-01 5.54259419e-01 1.80101618e-01 7.32456505e-01 1.07251398e-01 7.33319879e-01 3.55600566e-01 -2.74261057e-01 8.52813944e-02 3.01719844e-01 2.56563753e-01 9.49463964e-01 -5.13800740e-01 2.87799209e-01 -6.22449040e-01 4.02584672e-01 -1.96614444e+00 -8.87472808e-01 -7.80662656e-01 2.19674325e+00 7.31353760e-01 -4.21909720e-01 2.07337156e-01 3.97708341e-02 4.32056189e-01 4.69385922e-01 -1.17655285e-01 -5.24468958e-01 -4.17645425e-01 6.53523088e-01 6.03287995e-01 7.88104653e-01 -1.13559377e+00 6.89553320e-01 7.17562675e+00 8.90140235e-01 -9.17268932e-01 3.74661416e-01 5.54506540e-01 -2.12342009e-01 -5.08916080e-01 9.96231288e-03 -2.34432459e-01 2.32563525e-01 3.41691107e-01 -6.15955591e-02 7.59149492e-01 6.58404231e-01 4.83854830e-01 -9.56995506e-03 -1.08467710e+00 1.15247202e+00 -8.39002617e-03 -1.73822892e+00 -4.57268894e-01 1.04169898e-01 8.60387623e-01 2.49474898e-01 1.04218863e-01 -1.05392195e-01 1.23352669e-02 -1.14062285e+00 7.04662025e-01 3.65915269e-01 8.16624701e-01 -3.06588233e-01 1.92600057e-01 3.39295939e-02 -1.25140929e+00 1.66627035e-01 -3.58934283e-01 -1.98674947e-01 6.12043023e-01 1.10285699e+00 4.11007106e-01 6.67056382e-01 9.13322091e-01 9.80913877e-01 2.79408753e-01 9.44719017e-01 1.67179525e-01 6.85908496e-01 -6.35715425e-01 5.37575901e-01 3.76224965e-01 -7.62241960e-01 9.79857087e-01 1.31953871e+00 4.45035368e-01 5.67440748e-01 1.70297131e-01 1.00525534e+00 1.15678258e-01 1.82466596e-01 -6.54938638e-01 3.84146750e-01 -1.56338006e-01 1.10897684e+00 -4.70290005e-01 -1.23850696e-01 -5.31961262e-01 7.43342936e-01 2.09951892e-01 6.06063128e-01 -7.52546310e-01 -5.02794012e-02 8.11623991e-01 1.05338827e-01 3.85939032e-01 -5.26406050e-01 -7.16915250e-01 -1.51387477e+00 1.01314932e-02 -1.04946482e+00 4.87884015e-01 -6.57902479e-01 -1.37383509e+00 3.21689039e-01 -1.22662157e-01 -8.59301150e-01 4.15363163e-01 -7.26633370e-01 -3.97000879e-01 8.46695960e-01 -1.20696378e+00 -8.68299782e-01 -1.52301565e-01 8.28626275e-01 2.88904279e-01 4.49201643e-01 8.68589222e-01 5.18942058e-01 -6.01382315e-01 3.24943930e-01 3.65657508e-01 -6.32157102e-02 5.01848400e-01 -9.57969129e-01 -4.87679169e-02 7.53472626e-01 -1.46783277e-01 6.50278687e-01 9.01888013e-01 -4.22760904e-01 -2.10738039e+00 -7.05972373e-01 5.85660100e-01 -3.51681143e-01 7.53331363e-01 -3.44494909e-01 -9.70927298e-01 1.07088017e+00 1.46137938e-01 2.88347989e-01 6.03487611e-01 1.44538611e-01 -1.57170430e-01 -1.88650683e-01 -1.26166594e+00 3.48337680e-01 1.33559799e+00 -1.63006604e-01 -4.08930510e-01 7.04923034e-01 4.18363005e-01 -7.05500901e-01 -1.09485304e+00 3.79154474e-01 4.69877362e-01 -1.18737400e+00 1.43032992e+00 -2.94400364e-01 4.81858015e-01 -6.03244454e-02 -4.36648309e-01 -9.38317478e-01 -3.50218773e-01 -1.31309497e+00 -1.04937367e-01 6.44917905e-01 1.61995962e-01 -7.25947678e-01 5.95208585e-01 6.08202338e-01 -5.94887435e-01 -1.09395015e+00 -1.27847362e+00 -6.17283046e-01 2.76495606e-01 -7.80634940e-01 -4.20277789e-02 1.13473678e+00 1.35275468e-01 -1.71875060e-01 -7.52900660e-01 -1.51235208e-01 1.23143566e+00 -8.40051919e-02 5.21333575e-01 -9.32650983e-01 -3.20109159e-01 -4.55454677e-01 -1.62617818e-01 -1.25792372e+00 -1.15160570e-01 -1.00498319e+00 -8.13712031e-02 -1.31400228e+00 8.98627192e-03 -5.27891994e-01 1.17007971e-01 3.18757743e-01 2.99076289e-01 5.24702966e-01 -3.29576284e-02 1.31471500e-01 -3.67613316e-01 5.72236180e-01 1.35672235e+00 8.20895508e-02 -8.21913406e-02 -2.76691079e-01 -6.42436624e-01 8.91399205e-01 1.77147657e-01 -4.30964619e-01 -8.23782682e-02 -9.91086423e-01 6.30822480e-01 2.26593643e-01 4.47328657e-01 -7.22888470e-01 1.17174454e-01 -1.42506495e-01 -1.26939341e-01 -1.64043814e-01 4.01671290e-01 -7.15673864e-01 1.99318469e-01 2.97365397e-01 -1.79386348e-01 -1.29135787e-01 1.33240014e-01 3.83291662e-01 -9.19107124e-02 -2.00139388e-01 8.85279298e-01 -4.51169878e-01 -3.18160892e-01 3.83019030e-01 -4.17230755e-01 5.79974771e-01 8.26812148e-01 -3.77789766e-01 -3.11925784e-02 -3.83072138e-01 -1.07521737e+00 8.11236650e-02 3.96218121e-01 -4.81991172e-01 5.36323488e-01 -1.16044056e+00 -9.26573992e-01 1.41096905e-01 -2.43589208e-01 -2.78777815e-02 3.98561716e-01 1.61391366e+00 -7.65715301e-01 9.05750990e-02 3.48232716e-01 -9.00519311e-01 -1.09640181e+00 4.65203822e-01 5.05494177e-01 -9.02445316e-02 -9.42375839e-01 1.21166515e+00 2.35264868e-01 -2.51129240e-01 3.47441226e-01 -2.68478543e-01 4.51764375e-01 -2.81919744e-02 3.97987843e-01 8.11237931e-01 1.56486362e-01 -5.01455545e-01 -4.17932421e-01 1.11784470e+00 4.83094633e-01 -1.88612640e-02 1.79654849e+00 -2.23834500e-01 -7.73729801e-01 1.75001100e-01 1.35445273e+00 1.38832897e-01 -1.20595026e+00 -2.17350692e-01 -5.21417737e-01 -5.05236149e-01 2.57595152e-01 -2.74981171e-01 -1.37328887e+00 8.29693079e-01 1.71503246e-01 2.98657507e-01 1.33986962e+00 -1.70806706e-01 9.21948314e-01 1.02617815e-01 5.92755795e-01 -1.03449571e+00 -3.07244062e-01 5.41657627e-01 1.08572316e+00 -1.10027599e+00 3.65986139e-01 -7.87711561e-01 -3.62765431e-01 1.12050939e+00 -6.44667894e-02 -3.63048464e-01 9.74269092e-01 4.15216088e-01 -1.63474470e-01 -4.81021166e-01 -2.89805621e-01 -1.21135354e-01 2.29613870e-01 4.92367715e-01 7.13005722e-01 -2.88181335e-01 -6.05650008e-01 2.03632176e-01 -2.03821762e-03 2.23543812e-02 2.49485761e-01 9.92071688e-01 -3.25864285e-01 -9.95845556e-01 -5.68795145e-01 3.95095140e-01 -6.50329590e-01 -3.41053337e-01 1.79345161e-01 4.28098440e-01 7.55418688e-02 8.25176358e-01 -4.39445615e-01 1.41935900e-01 2.32382074e-01 -2.40143500e-02 7.70187914e-01 -3.19784403e-01 -8.31794143e-01 3.87683600e-01 2.03323767e-01 -9.32521403e-01 -6.03546560e-01 -8.11109602e-01 -8.75691593e-01 -6.58758700e-01 -4.11212146e-02 1.64899215e-01 5.12596905e-01 9.31033134e-01 2.52876878e-01 2.80214101e-01 3.70591849e-01 -1.22564745e+00 -5.30528605e-01 -4.16418970e-01 -6.60787284e-01 4.02810127e-01 4.15976882e-01 -5.22027075e-01 -4.99934524e-01 7.57170543e-02]
[6.966151237487793, 4.285747528076172]
c01a9c63-eb28-4e3c-9d41-7f4c7f37a681
self-supervised-speech-representation-1
2303.04255
null
https://arxiv.org/abs/2303.04255v1
https://arxiv.org/pdf/2303.04255v1.pdf
Self-supervised speech representation learning for keyword-spotting with light-weight transformers
Self-supervised speech representation learning (S3RL) is revolutionizing the way we leverage the ever-growing availability of data. While S3RL related studies typically use large models, we employ light-weight networks to comply with tight memory of compute-constrained devices. We demonstrate the effectiveness of S3RL on a keyword-spotting (KS) problem by using transformers with 330k parameters and propose a mechanism to enhance utterance-wise distinction, which proves crucial for improving performance on classification tasks. On the Google speech commands v2 dataset, the proposed method applied to the Auto-Regressive Predictive Coding S3RL led to a 1.2% accuracy improvement compared to training from scratch. On an in-house KS dataset with four different keywords, it provided 6% to 23.7% relative false accept improvement at fixed false reject rate. We argue this demonstrates the applicability of S3RL approaches to light-weight models for KS and confirms S3RL is a powerful alternative to traditional supervised learning for resource-constrained applications.
['Yuzong Liu', 'Francesco Caliva', 'Yue Gu', 'Chenyang Gao']
2023-03-07
null
null
null
null
['keyword-spotting']
['speech']
[ 4.00760680e-01 1.54228136e-01 -3.28581959e-01 -5.15687346e-01 -1.19319081e+00 -2.80650407e-01 3.41942519e-01 1.97026152e-02 -5.83812952e-01 3.67390543e-01 1.20429106e-01 -8.22037458e-01 -1.90486982e-01 -3.22085083e-01 -5.53964198e-01 -4.36686546e-01 -3.77777740e-02 2.49961048e-01 2.50602454e-01 -2.49613717e-01 1.55404612e-01 4.40176636e-01 -1.69575226e+00 5.50364554e-01 4.71191913e-01 1.39019513e+00 2.97061086e-01 9.34087455e-01 -1.37236387e-01 9.35690105e-01 -7.60044932e-01 -1.37059227e-01 1.83707535e-01 1.96766004e-01 -7.36576200e-01 -7.37774223e-02 3.55118155e-01 -2.02646375e-01 -4.62940902e-01 4.56887275e-01 6.84859991e-01 2.45403782e-01 3.83542866e-01 -1.21790123e+00 -4.18131053e-01 8.78562093e-01 -1.53116599e-01 4.82659668e-01 1.86685264e-01 6.10754415e-02 1.12129903e+00 -1.07494903e+00 1.02790594e-01 9.38214004e-01 6.25605524e-01 4.86196041e-01 -1.18973505e+00 -7.59177744e-01 1.98936416e-03 4.44239557e-01 -1.48450935e+00 -1.18780637e+00 5.22327840e-01 -1.72461290e-02 1.87105155e+00 4.50720131e-01 2.86572665e-01 1.25409377e+00 -2.27252454e-01 1.04882526e+00 8.19486678e-01 -7.30189979e-01 3.90518814e-01 2.95626670e-01 2.12477639e-01 4.97253597e-01 -1.59705788e-01 -1.89621840e-02 -1.23616135e+00 -2.93604702e-01 1.81572214e-01 -1.77649975e-01 2.45410949e-03 -2.95236744e-02 -9.52509403e-01 7.88740158e-01 -1.33433074e-01 3.89068753e-01 -1.79435000e-01 6.15180805e-02 5.72163820e-01 5.39646268e-01 6.84962213e-01 5.10315478e-01 -9.86365676e-01 -7.29137242e-01 -1.13634694e+00 -2.45761737e-01 7.91297734e-01 1.04216504e+00 2.93418169e-01 6.17288709e-01 1.49531970e-02 1.13745534e+00 9.44575742e-02 6.68205202e-01 9.90793347e-01 -6.26115143e-01 6.90823913e-01 3.62696916e-01 -3.85684013e-01 -5.42634308e-01 -4.99468684e-01 -6.30950332e-01 -5.27326345e-01 -2.40149006e-01 4.69985493e-02 -1.52346259e-02 -8.38970304e-01 1.45916772e+00 -6.45977855e-02 1.37093544e-01 3.73132229e-01 4.70035762e-01 6.91451788e-01 8.14405143e-01 -6.01849221e-02 -2.42093280e-01 1.15639520e+00 -8.61337304e-01 -6.13819778e-01 -2.42006108e-01 1.00863397e+00 -5.67817092e-01 1.39422321e+00 7.48159528e-01 -8.30490768e-01 -4.03908163e-01 -1.19388616e+00 5.05712219e-02 -3.30058128e-01 3.63424420e-01 7.24637210e-01 1.10994470e+00 -1.24767470e+00 3.64649504e-01 -6.82633817e-01 -3.53663027e-01 4.61282402e-01 5.22572398e-01 -4.06274498e-01 1.80247799e-01 -1.13105202e+00 8.51352811e-01 3.22488487e-01 -2.77432650e-01 -6.45687699e-01 -7.32302606e-01 -5.37341714e-01 4.17173296e-01 7.20317960e-01 1.31866513e-02 1.45806050e+00 -5.56937993e-01 -1.89021468e+00 5.87777674e-01 -2.28393048e-01 -9.61634576e-01 1.21181689e-01 -3.18979084e-01 -7.60632396e-01 4.40682620e-02 -3.71568918e-01 3.42374444e-01 1.20628679e+00 -5.72019637e-01 -6.63146734e-01 -2.11919144e-01 -3.38438541e-01 -5.44647239e-02 -1.05264461e+00 -1.46200322e-03 -3.01670790e-01 -6.76909089e-01 1.26068279e-01 -8.26639473e-01 1.57872796e-01 -4.23927128e-01 -2.32936397e-01 -3.76902610e-01 8.61640453e-01 -6.83803022e-01 1.40617979e+00 -2.31550407e+00 -2.05140457e-01 2.94318289e-01 9.56287682e-02 6.95050836e-01 -1.24505930e-01 4.33095872e-01 -1.41256347e-01 7.07975402e-02 -6.54013306e-02 -5.92079282e-01 1.12127602e-01 1.79676980e-01 -4.89202112e-01 2.59923726e-01 3.66061866e-01 6.34077907e-01 -5.42641282e-01 -1.94643572e-01 2.45903417e-01 4.64947850e-01 -6.33076727e-01 2.57067770e-01 5.48413061e-02 -3.19718659e-01 -8.83306116e-02 7.25418508e-01 1.95336655e-01 -1.39537811e-01 1.50570214e-01 -6.34265989e-02 9.86965522e-02 7.54072726e-01 -8.45287263e-01 1.72150302e+00 -8.40690494e-01 8.83131266e-01 -2.26750210e-01 -1.10469222e+00 1.14961708e+00 4.19818342e-01 2.01950312e-01 -8.84457409e-01 -2.87087839e-02 1.87033385e-01 -1.16894118e-01 -3.62296611e-01 6.50340617e-01 6.26719818e-02 5.91514818e-02 3.69959116e-01 4.11645055e-01 -1.65463001e-01 -2.63282359e-01 3.04438591e-01 1.38981020e+00 -2.79819876e-01 2.00932190e-01 -1.88839778e-01 4.82416213e-01 -2.93699086e-01 2.33648598e-01 8.23742568e-01 -5.36920726e-02 2.82426000e-01 3.25948775e-01 -3.48100394e-01 -8.10523927e-01 -7.43196130e-01 -1.46493688e-01 1.51301205e+00 -6.01592720e-01 -7.39455521e-01 -4.68855619e-01 -5.60206771e-01 5.08715063e-02 1.13034749e+00 -1.71784177e-01 -3.71989846e-01 -2.32526675e-01 -6.46907330e-01 1.09282982e+00 5.08316398e-01 7.03735650e-02 -8.06559145e-01 -6.16568267e-01 1.98902622e-01 8.66313204e-02 -1.45801830e+00 -1.15970492e-01 8.11488271e-01 -8.04649949e-01 -6.17850482e-01 -3.75617683e-01 -4.58467633e-01 3.17671001e-01 2.50833869e-01 1.04285395e+00 -4.06492576e-02 -1.38521567e-01 4.69135851e-01 -6.79045200e-01 -4.68511820e-01 -6.59806550e-01 5.06250262e-01 6.19402289e-01 -3.96768413e-02 4.06569451e-01 -6.70131266e-01 -7.71832466e-02 1.42591402e-01 -6.58616304e-01 -3.56942974e-02 6.57387197e-01 8.65885913e-01 2.37214699e-01 1.43978000e-01 9.67210591e-01 -7.40038633e-01 6.52427554e-01 -4.08062845e-01 -6.20578289e-01 2.09056959e-01 -1.13055146e+00 2.78824866e-01 6.40295565e-01 -5.26777804e-01 -8.09677899e-01 -6.03280626e-02 -3.53160948e-01 -4.85896528e-01 3.95057425e-02 3.48391235e-01 1.50014535e-01 1.30809890e-02 5.46001554e-01 4.05707031e-01 1.02882627e-02 -5.32244146e-01 3.56739968e-01 1.25182235e+00 3.88232976e-01 -4.33767080e-01 5.40241301e-01 1.79196540e-02 -3.33587021e-01 -1.08443904e+00 -3.67793292e-01 -5.10985136e-01 -3.17054480e-01 2.31684726e-02 2.78603613e-01 -9.85403180e-01 -8.86765897e-01 1.80987924e-01 -7.98221409e-01 -4.96470392e-01 -3.42810750e-01 5.00001013e-01 -4.64739114e-01 2.52812684e-01 -4.46773797e-01 -1.24252760e+00 -7.19091773e-01 -9.79982138e-01 1.06593728e+00 -3.08696181e-01 -2.15482861e-01 -5.71581483e-01 -3.92047614e-01 6.56995416e-01 7.67438650e-01 -8.18158507e-01 9.91253138e-01 -1.28302646e+00 7.96952844e-02 -4.22319025e-01 -8.08864310e-02 7.03062654e-01 4.39805910e-02 -2.11579725e-01 -1.54597068e+00 -2.52808273e-01 1.44175475e-03 -5.53129256e-01 7.14548886e-01 -5.88859953e-02 1.41642618e+00 -2.22326458e-01 4.58135689e-03 3.36612254e-01 8.55265319e-01 3.76389384e-01 3.32097113e-01 1.70473307e-01 5.75726986e-01 3.31753373e-01 4.60783929e-01 6.93882108e-01 2.40961939e-01 9.36165035e-01 1.83272257e-01 1.36895552e-01 -1.67377740e-01 -2.60931194e-01 5.67125559e-01 1.31303418e+00 2.81650722e-01 -3.09641898e-01 -1.06073785e+00 4.01812762e-01 -1.52994704e+00 -6.59015417e-01 3.37887913e-01 2.29876566e+00 8.39982092e-01 4.40875113e-01 4.02805097e-02 7.24624395e-01 2.03489929e-01 8.39855447e-02 -6.33685231e-01 -5.98704815e-01 -1.42000556e-01 5.23379028e-01 7.29840636e-01 2.09887892e-01 -8.70010972e-01 9.87247586e-01 6.26932001e+00 1.33534467e+00 -1.39638841e+00 3.57446492e-01 6.78302646e-01 -4.22129303e-01 -5.99496998e-02 -3.50905418e-01 -9.77245688e-01 3.82237762e-01 1.83547866e+00 -1.46067202e-01 6.14503682e-01 1.10946000e+00 2.27808908e-01 3.77531946e-02 -1.11567616e+00 1.30730331e+00 2.63226092e-01 -1.45877850e+00 -1.14353634e-01 -1.65734470e-01 1.69403210e-01 3.43410879e-01 1.98617101e-01 8.09862077e-01 7.54327653e-03 -1.03529477e+00 7.46752441e-01 2.24592220e-02 1.08858752e+00 -7.60521412e-01 6.23430848e-01 4.85443413e-01 -8.45041752e-01 -2.33983129e-01 -2.79850721e-01 -3.16347145e-02 -1.09960839e-01 4.82968241e-01 -1.69873190e+00 2.72426397e-01 7.66281247e-01 3.32765788e-01 -5.90835035e-01 3.34798932e-01 1.26839295e-01 1.27596700e+00 -4.75501329e-01 -2.09181294e-01 9.40816626e-02 4.23415691e-01 3.29966635e-01 1.43772542e+00 2.06865922e-01 -1.39819622e-01 -2.10582599e-01 2.46222541e-01 -1.41261667e-01 2.45456502e-01 -5.23354471e-01 -4.03186500e-01 6.92470253e-01 9.20950711e-01 -5.07049978e-01 -4.19169366e-01 -3.46805334e-01 8.51986170e-01 4.00077224e-01 1.69066891e-01 -6.81919813e-01 -2.23139510e-01 4.36062723e-01 5.65238111e-02 4.49966609e-01 -2.99156129e-01 -9.82079729e-02 -9.52499151e-01 -9.39517282e-03 -1.11692786e+00 1.28818542e-01 -6.83307827e-01 -9.77436900e-01 7.99329519e-01 -1.51461348e-01 -1.07889974e+00 -3.97423714e-01 -5.58884621e-01 -6.92317337e-02 6.36271179e-01 -1.51052415e+00 -9.51570868e-01 1.38235211e-01 4.81776983e-01 1.00700223e+00 -6.51442111e-01 1.26227772e+00 4.50762659e-01 -4.95184004e-01 1.05662525e+00 9.25225094e-02 -1.96163565e-01 5.38541794e-01 -8.78365576e-01 6.86862171e-01 7.29952753e-01 5.96357524e-01 4.99659956e-01 3.60078424e-01 -1.78691879e-01 -1.85435128e+00 -7.55907536e-01 8.25210869e-01 -2.82771677e-01 7.88620949e-01 -6.76316381e-01 -9.14486110e-01 3.92323911e-01 -2.04287603e-01 1.55814558e-01 9.08143938e-01 3.47669333e-01 -6.79123044e-01 -3.09576601e-01 -1.02580237e+00 3.83996814e-01 1.02133489e+00 -9.45916593e-01 -4.15545076e-01 3.74679774e-01 1.13732481e+00 -3.17258209e-01 -8.30508292e-01 2.93870091e-01 5.19353151e-01 -7.08510280e-01 8.79445612e-01 -7.33790815e-01 -3.17736790e-02 1.01124004e-01 -6.48536086e-01 -1.17354095e+00 2.49504652e-02 -7.63906598e-01 -3.81283224e-01 1.11330044e+00 7.88386881e-01 -6.81429625e-01 8.01152289e-01 5.54566860e-01 -3.83934736e-01 -7.82951832e-01 -1.38931668e+00 -9.22522902e-01 -3.59544307e-01 -1.26891291e+00 4.72335815e-01 7.91596234e-01 2.75646567e-01 2.85619050e-01 -4.62958097e-01 9.09503847e-02 1.70263380e-01 -4.83495384e-01 6.60968959e-01 -1.01926684e+00 -4.60152537e-01 -1.88037872e-01 -5.43323517e-01 -9.30076063e-01 3.71441275e-01 -9.78885293e-01 -4.53813188e-02 -7.27676272e-01 -1.79360420e-01 -7.62864113e-01 -5.43809593e-01 7.97380507e-01 1.60929516e-01 8.25562254e-02 1.55900657e-01 -1.63336750e-02 -5.50163686e-01 4.69298214e-01 1.98328570e-01 -3.42349082e-01 -3.17442328e-01 1.57341182e-01 -5.87967038e-01 4.26258385e-01 8.90607834e-01 -4.23836827e-01 -6.88020825e-01 -2.44690701e-01 1.10089682e-01 1.62317231e-02 -1.34029448e-01 -1.15190530e+00 3.54637802e-01 2.04527169e-01 1.04322389e-05 -3.93192112e-01 7.30097413e-01 -9.34223950e-01 -1.82068869e-01 2.57664502e-01 -6.19687140e-01 -6.83028921e-02 5.28522670e-01 5.17526090e-01 -9.97949168e-02 -1.64568976e-01 2.90071487e-01 3.72654974e-01 -9.30058539e-01 -1.35817438e-01 -5.58337748e-01 -1.36964366e-01 5.90167403e-01 -2.98147295e-02 -2.30224311e-01 -5.40764987e-01 -5.33887923e-01 -1.47554129e-01 -2.19320416e-01 7.34761238e-01 8.55894387e-01 -8.97006094e-01 -4.13035333e-01 7.36391842e-01 4.93245661e-01 -3.80566388e-01 7.75127485e-03 7.94952273e-01 -1.11362018e-01 6.79986835e-01 3.42526764e-01 -6.51967824e-01 -1.63118851e+00 2.46357203e-01 1.48475962e-02 8.81380402e-03 -6.06783986e-01 1.11022973e+00 -4.95036453e-01 -3.95482630e-01 6.50544226e-01 -6.10245943e-01 -2.36465000e-02 -2.30610017e-02 5.28554261e-01 4.09881383e-01 8.13311994e-01 -4.22436267e-01 -5.49364984e-01 -1.76961482e-01 -3.28524441e-01 -2.11079106e-01 1.55023694e+00 1.75149459e-02 4.98223990e-01 7.22968459e-01 1.06618011e+00 6.00624904e-02 -8.64296317e-01 -4.92951602e-01 3.26385498e-01 -1.62627012e-01 4.65290040e-01 -7.60044932e-01 -9.39554513e-01 7.00560510e-01 6.43957973e-01 2.00571850e-01 9.90927935e-01 -8.31880867e-02 7.54268765e-01 1.01773119e+00 5.67396998e-01 -1.19768417e+00 -5.76492250e-02 4.74615932e-01 6.37090087e-01 -1.19448876e+00 -1.96551401e-02 -1.72523901e-01 -7.36890018e-01 1.06921208e+00 1.54552922e-01 2.89078057e-01 4.73622650e-01 6.66855931e-01 2.76804976e-02 1.71490293e-02 -1.20903802e+00 7.86420703e-02 3.98401581e-02 6.02950454e-01 3.90643328e-01 1.12284757e-01 2.98329622e-01 7.02748179e-01 -3.93898785e-01 -3.02112792e-02 3.87946606e-01 9.24843550e-01 -2.57334143e-01 -1.03228843e+00 -2.37943918e-01 7.36691535e-01 -3.54012638e-01 -4.98952240e-01 -1.82995312e-02 5.30092001e-01 -3.00406247e-01 1.38180280e+00 6.32341355e-02 -9.27722275e-01 4.20303136e-01 4.73806143e-01 1.60187930e-01 -6.40154541e-01 -7.34743237e-01 -7.18241557e-03 4.03752804e-01 -5.62964737e-01 -2.41218716e-01 -4.76162136e-01 -1.18029928e+00 -2.37951547e-01 -5.58087230e-01 1.41666830e-01 1.12131810e+00 8.89830589e-01 6.07326031e-01 7.27172554e-01 6.88642144e-01 -6.22484148e-01 -8.97271276e-01 -1.15713775e+00 -5.27215838e-01 -1.07782881e-03 2.62309760e-01 -5.60197651e-01 -5.59069633e-01 -8.38496312e-02]
[14.212139129638672, 6.331079006195068]
aaac2c1a-3f44-4d99-865f-82148de63383
unbiased-methods-for-multi-goal-reinforcement
2106.08863
null
https://arxiv.org/abs/2106.08863v1
https://arxiv.org/pdf/2106.08863v1.pdf
Unbiased Methods for Multi-Goal Reinforcement Learning
In multi-goal reinforcement learning (RL) settings, the reward for each goal is sparse, and located in a small neighborhood of the goal. In large dimension, the probability of reaching a reward vanishes and the agent receives little learning signal. Methods such as Hindsight Experience Replay (HER) tackle this issue by also learning from realized but unplanned-for goals. But HER is known to introduce bias, and can converge to low-return policies by overestimating chancy outcomes. First, we vindicate HER by proving that it is actually unbiased in deterministic environments, such as many optimal control settings. Next, for stochastic environments in continuous spaces, we tackle sparse rewards by directly taking the infinitely sparse reward limit. We fully formalize the problem of multi-goal RL with infinitely sparse Dirac rewards at each goal. We introduce unbiased deep Q-learning and actor-critic algorithms that can handle such infinitely sparse rewards, and test them in toy environments.
['Yann Ollivier', 'Léonard Blier']
2021-06-16
null
null
null
null
['multi-goal-reinforcement-learning']
['methodology']
[-3.22629362e-01 6.05548561e-01 -3.46249491e-01 9.01206732e-02 -1.15249372e+00 -5.51766872e-01 2.82629728e-01 -7.88200926e-03 -8.01255763e-01 1.48077869e+00 2.93506593e-01 -1.08683072e-01 -2.72572309e-01 -6.69721365e-01 -9.86808717e-01 -8.88666034e-01 -4.59463060e-01 6.12880409e-01 -3.39111328e-01 -2.14508012e-01 1.98407799e-01 5.92054799e-03 -1.14687407e+00 -2.86099732e-01 7.79016137e-01 6.11885369e-01 2.36125901e-01 1.06092501e+00 2.30281636e-01 1.40193284e+00 -5.63150823e-01 -2.39543580e-02 3.75659972e-01 -7.28489518e-01 -5.48597574e-01 -8.19436014e-02 -4.31989953e-02 -9.45208907e-01 -1.87931746e-01 1.22959936e+00 6.72112823e-01 4.11585033e-01 3.98996800e-01 -1.33455479e+00 -7.61593223e-01 1.07878006e+00 -6.12676680e-01 -1.43497512e-01 3.14185947e-01 6.09294593e-01 1.22794712e+00 -3.67477268e-01 5.08943558e-01 1.41346824e+00 4.65451419e-01 8.58708024e-01 -1.24428403e+00 -2.32413381e-01 5.12675703e-01 -2.79814810e-01 -5.73749781e-01 -3.80559772e-01 5.49197853e-01 -1.88299477e-01 8.11239958e-01 -5.97419627e-02 8.03477943e-01 1.37613547e+00 2.09744558e-01 1.23868990e+00 1.30049181e+00 -2.28000775e-01 9.83956575e-01 -2.46383116e-01 -3.64211828e-01 4.75566834e-01 2.58858442e-01 8.33788753e-01 -3.90888005e-01 -1.85084000e-01 8.56703520e-01 2.61010468e-01 1.83369115e-01 -6.13057613e-01 -1.24719441e+00 1.10336006e+00 3.79477829e-01 4.58512157e-02 -9.01199043e-01 9.61571276e-01 3.00219029e-01 6.68414235e-01 8.80329963e-03 7.46812463e-01 -3.50029081e-01 -4.10429269e-01 -5.12683570e-01 8.63595843e-01 8.41301501e-01 9.57452297e-01 6.36008084e-01 5.45515716e-01 -5.45739472e-01 2.96679080e-01 1.15470260e-01 1.04149437e+00 4.06569183e-01 -1.68242919e+00 5.60096741e-01 -3.04640740e-01 1.14915383e+00 -3.75695527e-01 -5.32270908e-01 -5.72950125e-01 -3.62090915e-01 7.90558636e-01 8.82148921e-01 -8.55079353e-01 -4.60819125e-01 2.21564889e+00 1.97951555e-01 -1.58743292e-01 3.32384020e-01 1.18855071e+00 -8.10678154e-02 3.97604704e-01 -6.92178011e-02 -5.74516952e-01 4.46189553e-01 -9.75942075e-01 -6.81584001e-01 -5.18808663e-01 6.46400392e-01 -8.70900005e-02 1.45155859e+00 6.18176758e-01 -1.34562480e+00 -5.27937822e-02 -7.25668550e-01 3.19051147e-01 4.08706695e-01 -2.56334335e-01 7.43198097e-01 3.75747919e-01 -9.44073379e-01 9.86002207e-01 -8.22503924e-01 5.94104603e-02 5.99799216e-01 3.92030776e-02 1.01168007e-01 -9.49310064e-02 -9.02158618e-01 7.91598439e-01 1.43571213e-01 -1.50272593e-01 -1.98215771e+00 -2.54473478e-01 -6.68129742e-01 4.58480641e-02 8.59965980e-01 -4.79552120e-01 1.90253663e+00 -1.19789624e+00 -1.81882250e+00 2.90111631e-01 2.05923900e-01 -6.76983178e-01 8.55682611e-01 -4.72041875e-01 1.03829466e-01 -9.18478593e-02 4.76983070e-01 4.96611118e-01 1.12975919e+00 -1.14877963e+00 -5.01801550e-01 -4.89967108e-01 3.60518157e-01 6.88333094e-01 1.61024511e-01 -3.86370957e-01 7.02440739e-01 -8.69600549e-02 -4.13045585e-01 -7.89371431e-01 -8.48627925e-01 -2.10014418e-01 -3.69909465e-01 -1.14613503e-01 -5.71097843e-02 -4.93532792e-02 5.21956921e-01 -1.82487726e+00 -5.99950925e-02 -2.68511716e-02 3.09740901e-01 -3.46271127e-01 -4.34325635e-01 4.16871756e-01 5.16019285e-01 -2.07914218e-01 1.32462054e-01 -3.57209980e-01 5.35515547e-01 4.30166274e-01 -4.60281163e-01 7.12231815e-01 -2.23071888e-01 1.03244543e+00 -1.67282307e+00 -2.23307267e-01 -1.63121279e-02 -2.87120670e-01 -8.90893042e-01 3.43487382e-01 -5.84318638e-01 5.87709188e-01 -8.14433515e-01 3.93970937e-01 2.68284708e-01 -1.83947727e-01 2.59767920e-01 1.02207243e+00 -4.44456376e-02 1.21986851e-01 -1.32799947e+00 1.63449728e+00 -5.47278285e-01 8.72209072e-02 1.95591360e-01 -7.42526650e-01 5.78595757e-01 1.43783495e-01 3.73892725e-01 -7.76357830e-01 2.13674843e-01 2.99493641e-01 -4.73985858e-02 -4.21757817e-01 6.63022518e-01 -5.89420974e-01 -2.84581810e-01 7.31398463e-01 9.25667658e-02 -2.55127430e-01 -1.15980558e-01 1.99115500e-01 1.30931544e+00 3.35394323e-01 2.40425304e-01 2.44554393e-02 -2.55150110e-01 3.43724750e-02 5.21770537e-01 1.66973770e+00 -5.60193539e-01 1.98247001e-01 8.65481019e-01 -2.30650052e-01 -1.17031133e+00 -1.36403823e+00 5.19544601e-01 1.35064220e+00 1.93376482e-01 1.63730949e-01 -4.48210955e-01 -6.28924131e-01 3.53634357e-01 1.03875494e+00 -7.46101141e-01 -9.07067284e-02 -2.87480861e-01 -4.42388505e-01 1.73832506e-01 3.61774474e-01 9.99795347e-02 -1.46998751e+00 -1.08346045e+00 6.25621200e-01 9.74472314e-02 -5.14124513e-01 -4.45364147e-01 5.85822165e-01 -8.04806709e-01 -9.23311293e-01 -9.87078965e-01 -1.55408606e-01 3.07868093e-01 -6.10244460e-02 1.24560475e+00 -3.30025434e-01 6.97995946e-02 8.07484865e-01 -6.47958517e-02 -5.58546782e-01 -3.06132793e-01 -3.56010050e-01 3.87949258e-01 -4.55374151e-01 1.77699611e-01 -4.79298472e-01 -6.53403044e-01 -1.25507668e-01 -4.09033895e-01 -5.00151813e-01 4.65104610e-01 1.12306595e+00 5.60326874e-01 -3.42863888e-01 1.03362632e+00 -6.63454533e-01 1.11252773e+00 -7.98156977e-01 -9.74382937e-01 -1.06328048e-01 -4.27321285e-01 3.47595721e-01 1.06925786e+00 -6.11159503e-01 -8.55839133e-01 -4.77985591e-02 1.15789406e-01 -4.12942231e-01 3.32679204e-03 6.87129349e-02 2.22577944e-01 3.37203383e-01 1.09319532e+00 4.44727428e-02 2.09468409e-01 -1.38198495e-01 7.73664892e-01 1.98871732e-01 1.17767334e-01 -1.08988404e+00 3.67839932e-01 3.45243841e-01 1.11513659e-01 -3.04220140e-01 -1.21230650e+00 7.28731826e-02 2.56339282e-01 -4.90162432e-01 4.49239761e-01 -9.94126558e-01 -1.47830451e+00 1.13840260e-01 -8.20210397e-01 -1.05938876e+00 -1.23758411e+00 8.24458182e-01 -1.55207336e+00 3.89828682e-02 -4.43382174e-01 -1.61849189e+00 1.07481167e-01 -9.78821695e-01 7.88031042e-01 3.53845119e-01 7.53073990e-02 -8.87397707e-01 4.63721752e-01 -2.98453093e-01 4.46373522e-01 1.47070751e-01 2.42571726e-01 -5.85514128e-01 -6.24291778e-01 2.45841429e-01 3.65919381e-01 1.77652583e-01 -3.13003272e-01 -7.00543761e-01 -8.00364375e-01 -4.59652841e-01 2.06785649e-01 -1.28865349e+00 7.04951823e-01 9.00809169e-01 7.20861077e-01 -6.54372454e-01 1.67338967e-01 2.50745356e-01 1.41483045e+00 8.74616206e-03 5.12424447e-02 3.99417549e-01 1.64595962e-01 3.45622569e-01 9.77526307e-01 1.11873293e+00 2.67019987e-01 4.65296172e-02 1.07338190e+00 3.88435006e-01 4.02793735e-01 -7.94829249e-01 6.84859276e-01 -6.17516004e-02 4.09177318e-02 -1.55906687e-02 -3.46559674e-01 6.90769076e-01 -2.22676539e+00 -1.25021398e+00 2.26044849e-01 2.36561918e+00 9.20565188e-01 2.38871589e-01 6.30276322e-01 -3.34351301e-01 3.99454713e-01 8.33628848e-02 -1.39494109e+00 -5.49256265e-01 -1.28449336e-01 -8.27528983e-02 8.98122489e-01 7.92133152e-01 -6.10315442e-01 9.66513395e-01 7.43036318e+00 6.88917816e-01 -5.89211583e-01 3.25470597e-01 6.46727264e-01 -7.10750699e-01 -4.81411487e-01 1.01684323e-02 -6.75398111e-01 2.93847889e-01 8.00488949e-01 -3.67636442e-01 1.16528845e+00 1.51233590e+00 4.34984595e-01 -4.91032153e-01 -1.20350897e+00 7.58822441e-01 -5.18695533e-01 -9.87242758e-01 -7.40186572e-01 1.24333836e-01 9.52226639e-01 2.73871183e-01 2.84439951e-01 8.42550874e-01 1.59842455e+00 -1.15142751e+00 9.58090544e-01 5.24430513e-01 5.16925931e-01 -8.50688398e-01 3.79965991e-01 8.41730177e-01 -2.59818494e-01 -6.42507851e-01 -8.08362186e-01 -7.20442474e-01 7.25028664e-02 5.09152293e-01 -6.11108065e-01 -1.88256763e-02 2.02906772e-01 5.13149083e-01 2.95957774e-01 8.76277030e-01 -5.26940763e-01 4.36014861e-01 -4.48669434e-01 -6.64582372e-01 7.55750835e-01 -3.15547585e-01 6.07315719e-01 4.45314735e-01 4.32939500e-01 -9.50621367e-02 4.55335945e-01 1.23916447e+00 4.54085059e-02 -1.48594946e-01 -9.12265718e-01 4.97452542e-02 3.55306476e-01 1.00145578e+00 -1.81323573e-01 -1.94642052e-01 1.08669316e-02 7.36680090e-01 7.19836831e-01 5.85939169e-01 -7.09752202e-01 -6.44980371e-02 7.54058301e-01 -4.13393855e-01 1.78017274e-01 -1.58892468e-01 -1.40889093e-01 -1.09086764e+00 -6.18299693e-02 -7.09668338e-01 2.50581622e-01 -7.39652157e-01 -1.29864681e+00 -5.52091142e-03 -4.43727940e-01 -1.01739860e+00 -7.51238644e-01 -2.99116284e-01 -4.63224679e-01 7.74974167e-01 -1.58107638e+00 -5.43978751e-01 4.08946216e-01 7.27090418e-01 3.43351543e-01 -1.50042608e-01 6.31464005e-01 -4.16579187e-01 -1.25193536e-01 3.15611631e-01 6.38482571e-01 -2.74969786e-01 4.40674841e-01 -1.73145354e+00 9.67498403e-03 4.40681607e-01 -1.02486415e-02 2.08435178e-01 9.96410251e-01 -5.05221009e-01 -1.58664620e+00 -9.46776271e-01 1.87503621e-01 -4.14574921e-01 9.57806945e-01 -1.03355773e-01 -2.92604923e-01 8.39192212e-01 1.50333922e-02 3.08740735e-01 9.87249091e-02 3.32136542e-01 -5.05365506e-02 1.87110305e-01 -1.38543391e+00 8.53762627e-01 1.09797680e+00 -1.75190195e-01 -6.29544973e-01 5.53098857e-01 7.37218618e-01 -5.41616738e-01 -5.42281568e-01 -5.01624823e-01 4.19798911e-01 -1.21533704e+00 7.50676334e-01 -1.12222862e+00 4.80660111e-01 6.65673688e-02 -1.45906463e-01 -1.93085599e+00 -2.18638927e-01 -1.24245322e+00 -3.82840186e-01 4.79858249e-01 2.28884429e-01 -5.68738997e-01 9.33999062e-01 2.72856027e-01 -2.79053897e-02 -6.00370586e-01 -9.99092758e-01 -1.22446871e+00 6.20755911e-01 -3.85193497e-01 4.09740001e-01 4.23332483e-01 2.95914143e-01 2.17738077e-01 -8.70233119e-01 -2.10201979e-01 1.18391407e+00 1.18484199e-01 6.76389396e-01 -6.13149762e-01 -7.29592919e-01 -2.73034245e-01 3.65000308e-01 -1.30419898e+00 1.47086248e-01 -6.45273685e-01 4.76871759e-01 -1.30440760e+00 2.05788106e-01 -6.48423195e-01 -2.89617568e-01 2.34380394e-01 -1.66325122e-01 -4.82882589e-01 4.05909240e-01 3.93121280e-02 -1.37280416e+00 1.05159473e+00 1.79637194e+00 1.92768842e-01 -3.24873269e-01 2.68806189e-01 -9.31007147e-01 7.13495553e-01 9.58988786e-01 -7.70882547e-01 -4.56545293e-01 -1.99580267e-01 5.76336563e-01 7.69046426e-01 4.32560176e-01 -7.04109907e-01 -2.10343543e-02 -7.73514569e-01 3.60670030e-01 -2.87502944e-01 1.60022721e-01 -5.18064201e-01 -4.30408865e-01 5.92899442e-01 -9.15644228e-01 -1.60265073e-01 -4.46694195e-01 8.98337424e-01 4.27984893e-01 -5.54644108e-01 7.77144670e-01 -6.74010754e-01 -1.13130145e-01 3.13284218e-01 -4.98102635e-01 8.81730855e-01 1.01922619e+00 2.12486401e-01 -1.64159402e-01 -9.78173733e-01 -1.04395819e+00 7.16765106e-01 4.09334093e-01 -1.99673250e-01 5.97115874e-01 -1.36245763e+00 -5.94320416e-01 -2.17910752e-01 -1.38756588e-01 5.08567970e-03 2.77655244e-01 5.96781611e-01 2.15949088e-01 5.21664657e-02 -3.17178041e-01 -2.95622915e-01 -3.06376100e-01 8.40855062e-01 5.37418008e-01 -5.40756464e-01 -5.24908423e-01 7.32531428e-01 4.58149239e-02 -5.32086253e-01 4.15311366e-01 -3.83982003e-01 9.69593525e-02 -3.11875284e-01 6.45257354e-01 4.92357105e-01 -5.97495317e-01 2.23632023e-01 1.58006653e-01 -5.36529124e-02 1.05854861e-01 -7.11293817e-01 1.45922363e+00 -2.06962347e-01 4.65629250e-01 7.50894904e-01 6.88010275e-01 -8.30039158e-02 -2.12634802e+00 -1.18865512e-01 -9.82783437e-02 -4.98965710e-01 -2.15610087e-01 -7.44593740e-01 -4.71136987e-01 8.08882415e-01 3.22033554e-01 4.57552731e-01 5.10706425e-01 -1.15215272e-01 4.34439123e-01 9.22155023e-01 7.47714460e-01 -1.57936335e+00 5.50912678e-01 7.44809389e-01 7.77360439e-01 -1.36861181e+00 -3.42768312e-01 9.13559735e-01 -1.20704901e+00 7.93566525e-01 4.84828740e-01 -6.59835994e-01 1.05515242e-01 3.17080259e-01 -2.17683062e-01 1.76048372e-02 -1.01490724e+00 -5.64800918e-01 -6.53319716e-01 8.09383094e-01 -1.07658274e-01 4.37075049e-01 9.75012556e-02 6.14529312e-01 -1.39698789e-01 2.53912896e-01 9.05082762e-01 9.59402680e-01 -1.03910530e+00 -7.42850840e-01 -4.60769564e-01 6.06167674e-01 -5.73595345e-01 -6.86737895e-02 8.84508640e-02 5.01595497e-01 -5.49744725e-01 1.01122844e+00 -9.98033509e-02 1.76590770e-01 1.33383036e-01 -2.92212874e-01 8.54542673e-01 -5.33454120e-01 -4.43233460e-01 7.19326660e-02 -4.19733003e-02 -9.92375553e-01 -1.16577506e-01 -8.60807896e-01 -1.35651684e+00 -4.02043849e-01 6.82269409e-02 1.66534781e-01 4.03064191e-01 8.56557488e-01 7.25228488e-02 4.91286010e-01 9.56652820e-01 -8.25597048e-01 -1.82428384e+00 -8.25785518e-01 -1.10724926e+00 1.91170216e-01 9.39590037e-01 -6.38182342e-01 -7.08260596e-01 -6.60950661e-01]
[4.091702938079834, 2.105059862136841]
f2de0f5b-fbfe-43b4-ba3d-2ea31d4b82f0
unifying-vision-text-and-layout-for-universal
2212.02623
null
https://arxiv.org/abs/2212.02623v3
https://arxiv.org/pdf/2212.02623v3.pdf
Unifying Vision, Text, and Layout for Universal Document Processing
We propose Universal Document Processing (UDOP), a foundation Document AI model which unifies text, image, and layout modalities together with varied task formats, including document understanding and generation. UDOP leverages the spatial correlation between textual content and document image to model image, text, and layout modalities with one uniform representation. With a novel Vision-Text-Layout Transformer, UDOP unifies pretraining and multi-domain downstream tasks into a prompt-based sequence generation scheme. UDOP is pretrained on both large-scale unlabeled document corpora using innovative self-supervised objectives and diverse labeled data. UDOP also learns to generate document images from text and layout modalities via masked image reconstruction. To the best of our knowledge, this is the first time in the field of document AI that one model simultaneously achieves high-quality neural document editing and content customization. Our method sets the state-of-the-art on 8 Document AI tasks, e.g., document understanding and QA, across diverse data domains like finance reports, academic papers, and websites. UDOP ranks first on the leaderboard of the Document Understanding Benchmark.
['Mohit Bansal', 'Cha Zhang', 'Michael Zeng', 'Chenguang Zhu', 'Yang Liu', 'Yuwei Fang', 'Guoxin Wang', 'ZiYi Yang', 'Zineng Tang']
2022-12-05
null
http://openaccess.thecvf.com//content/CVPR2023/html/Tang_Unifying_Vision_Text_and_Layout_for_Universal_Document_Processing_CVPR_2023_paper.html
http://openaccess.thecvf.com//content/CVPR2023/papers/Tang_Unifying_Vision_Text_and_Layout_for_Universal_Document_Processing_CVPR_2023_paper.pdf
cvpr-2023-1
['document-ai']
['natural-language-processing']
[ 8.13710868e-01 -5.06617427e-02 -1.38429105e-01 -3.33604455e-01 -1.14545739e+00 -1.16421640e+00 1.24232364e+00 -8.01737309e-02 -1.04203500e-01 4.20776963e-01 4.25143898e-01 -3.82424533e-01 1.30165279e-01 -5.75075090e-01 -1.12181187e+00 -3.47467512e-01 7.24772155e-01 1.09984481e+00 -3.09553117e-01 -1.99740633e-01 6.88540161e-01 2.82770842e-01 -1.22469294e+00 1.10547972e+00 9.68845427e-01 7.93279231e-01 3.94362003e-01 1.32201588e+00 -6.50347888e-01 9.59409535e-01 -8.27568471e-01 -6.26601517e-01 5.93641512e-02 -3.98849308e-01 -1.03942811e+00 4.51074272e-01 8.88908029e-01 -6.30027115e-01 -3.50154787e-01 7.78160036e-01 5.42810678e-01 -8.99967179e-02 1.09880495e+00 -9.76559460e-01 -2.03944445e+00 9.75892067e-01 -6.26206517e-01 -2.52007172e-02 4.25219834e-01 1.93389773e-01 1.17532504e+00 -1.01995277e+00 1.04113722e+00 1.27936292e+00 2.98209310e-01 6.08585596e-01 -1.28041804e+00 -1.92270264e-01 3.68134826e-01 -8.02075490e-02 -9.53408301e-01 -4.42308038e-01 6.47014201e-01 -5.13576031e-01 1.12138617e+00 3.28752585e-02 1.02545105e-01 1.58876109e+00 1.58362556e-02 1.52076757e+00 6.88436747e-01 -7.47849226e-01 -1.55337140e-01 -2.06657406e-02 1.72400951e-01 7.69390404e-01 -1.57893542e-02 -3.06186914e-01 -4.90366071e-01 4.16358441e-01 6.17397189e-01 -7.05609247e-02 -3.10188264e-01 -8.56865272e-02 -1.22338581e+00 7.35650897e-01 -1.22254767e-01 3.17674100e-01 -1.32753134e-01 1.85263693e-01 5.13584316e-01 2.58240074e-01 4.15291905e-01 7.30760336e-01 -3.36121559e-01 -1.83508605e-01 -1.04046237e+00 4.56663877e-01 8.53551626e-01 1.30799961e+00 2.12731048e-01 1.66262269e-01 -7.72134304e-01 9.81228530e-01 1.23492293e-01 9.02901351e-01 6.04365587e-01 -8.10365200e-01 1.09311581e+00 5.40822506e-01 7.12230280e-02 -6.36603832e-01 -4.00039218e-02 -4.83629778e-02 -7.52972662e-01 -1.44865781e-01 2.54950672e-01 -2.59684902e-02 -1.39394093e+00 1.30052602e+00 -3.99919987e-01 -3.61436874e-01 3.18066269e-01 6.75922334e-01 8.95404398e-01 1.18286109e+00 -2.61467248e-01 1.80405721e-01 1.15043545e+00 -1.55463517e+00 -5.42071462e-01 -2.90694982e-01 6.42627656e-01 -9.00624931e-01 1.50440431e+00 6.83214486e-01 -1.44700348e+00 -7.06359982e-01 -9.61711168e-01 -7.22292900e-01 -5.75139344e-01 5.99388301e-01 4.53518808e-01 3.50889236e-01 -1.06473958e+00 1.49417520e-01 -4.68749940e-01 -3.27206939e-01 5.43706715e-01 -1.45561218e-01 -2.39998296e-01 -7.41849124e-01 -7.78462589e-01 7.41094410e-01 4.31308389e-01 -1.19107716e-01 -1.01705253e+00 -1.01869631e+00 -6.93320394e-01 1.24822482e-01 4.16262597e-01 -9.15216982e-01 1.62599885e+00 -1.16922712e+00 -1.53657913e+00 1.00627744e+00 -8.70429501e-02 -6.13623202e-01 5.41639447e-01 -4.92688596e-01 -4.29063916e-01 2.35422149e-01 1.49317250e-01 8.47270608e-01 1.16023326e+00 -1.57752848e+00 -3.50369364e-01 -4.61519450e-01 -1.31555468e-01 3.11035752e-01 -3.64772618e-01 -1.28814429e-02 -1.02539754e+00 -9.49447870e-01 -5.11727810e-01 -7.64287353e-01 2.69472808e-01 -1.26008451e-01 -7.41977334e-01 -1.33055285e-01 1.06977463e+00 -9.34560835e-01 9.12915945e-01 -1.89394450e+00 5.83774924e-01 -4.16279919e-02 -4.69052531e-02 7.32495412e-02 -7.27678895e-01 6.00270629e-01 1.99770913e-01 2.65627921e-01 -2.99249232e-01 -6.64664268e-01 5.01553059e-01 1.13014966e-01 -9.65044558e-01 -1.92478940e-01 2.84743220e-01 1.45681179e+00 -4.57497686e-01 -3.92699391e-01 -6.68284297e-02 3.51754576e-01 -6.16539180e-01 2.08380863e-01 -1.15699172e+00 6.42206371e-02 -3.91757399e-01 7.12826073e-01 5.21534443e-01 -5.38036704e-01 -1.65251326e-02 -3.63929272e-01 -3.85174900e-02 2.68834513e-02 -6.84692323e-01 2.37420392e+00 -5.32789290e-01 9.15556490e-01 -1.02269627e-01 -1.01282418e+00 7.57151783e-01 1.62550658e-01 1.19816110e-01 -9.25935805e-01 4.29604985e-02 -1.87095981e-02 -4.44865406e-01 -4.23465431e-01 1.24164665e+00 4.83100951e-01 -2.41887838e-01 1.04874432e+00 2.99888879e-01 -7.71641254e-01 5.74637651e-01 7.73322880e-01 1.00555038e+00 3.74618858e-01 -2.38859937e-01 1.08421661e-01 2.78873116e-01 1.60701767e-01 -5.17954886e-01 1.21596491e+00 4.04199690e-01 8.74438465e-01 4.60617661e-01 4.10073344e-03 -1.45935428e+00 -1.14658844e+00 2.50149071e-01 1.56596410e+00 -2.32909828e-01 -4.40963805e-01 -8.17005277e-01 -6.96899652e-01 2.02033833e-01 9.81703401e-01 -7.37091601e-01 9.41041950e-03 -4.09129500e-01 -3.74041885e-01 9.22743440e-01 8.69690180e-01 3.41143280e-01 -1.23138058e+00 1.71593919e-01 7.23389015e-02 -5.00106961e-02 -1.27794302e+00 -8.87353361e-01 -3.17827687e-02 -4.45362091e-01 -6.95274472e-01 -1.00208914e+00 -8.91531885e-01 7.60692716e-01 2.41742104e-01 1.40962529e+00 -6.54134601e-02 -4.27838057e-01 9.93454814e-01 -4.43900526e-01 -5.27355611e-01 -8.11537266e-01 1.73612610e-01 -4.74319726e-01 -1.83186740e-01 -8.18526745e-02 7.73476809e-02 -1.56399682e-01 -2.55312592e-01 -1.39730227e+00 4.09699440e-01 8.26363802e-01 9.40331101e-01 5.56899905e-01 -1.61747843e-01 2.09980413e-01 -1.20883274e+00 1.12784827e+00 -2.35448316e-01 -5.46526611e-01 9.12376463e-01 -5.62902808e-01 3.56935322e-01 8.25834215e-01 -2.92049408e-01 -1.62189710e+00 -2.72942126e-01 1.95669219e-01 -3.16089422e-01 -2.63979942e-01 5.87207675e-01 -8.06834772e-02 4.63923335e-01 6.60322666e-01 5.67871332e-01 -3.05484205e-01 -3.73811871e-01 1.15913403e+00 6.81649923e-01 1.15357065e+00 -9.89859819e-01 8.21687639e-01 4.92021829e-01 -4.15992737e-01 -6.17072046e-01 -1.05220520e+00 -1.58307910e-01 -8.25202405e-01 1.76390305e-01 9.81410921e-01 -8.27290654e-01 -2.14831382e-01 7.47783601e-01 -1.61047518e+00 -5.61461270e-01 -2.46354669e-01 -3.12457860e-01 -4.77190763e-01 3.16584110e-01 -7.80832171e-01 -3.74991775e-01 -6.90637767e-01 -1.06977558e+00 1.52834392e+00 -9.45525593e-04 -1.13103494e-01 -9.81619537e-01 -4.02677767e-02 7.40017712e-01 2.42533416e-01 -1.16000235e-01 1.06427228e+00 -4.84088123e-01 -7.98890591e-01 -4.83924858e-02 -7.68791556e-01 4.20003951e-01 -1.08699568e-01 4.64561254e-01 -9.33579803e-01 -1.41938314e-01 -6.55001581e-01 -9.42094207e-01 1.32529724e+00 2.91708320e-01 1.48715997e+00 -3.60419035e-01 -1.37219533e-01 5.97530484e-01 1.20268238e+00 3.48879665e-01 6.63018942e-01 2.96495229e-01 1.11559772e+00 4.16144520e-01 2.25561589e-01 5.31442821e-01 5.15168250e-01 3.67766529e-01 1.25192553e-01 -1.34891123e-01 -3.52274686e-01 -3.80381376e-01 4.65086132e-01 5.96730769e-01 3.25512886e-01 -1.09782112e+00 -9.72424448e-01 4.88947064e-01 -1.76585186e+00 -1.07424581e+00 2.54076481e-01 1.32831919e+00 8.76545012e-01 2.24097937e-01 -3.56105119e-01 -3.04664731e-01 4.17065293e-01 3.66880566e-01 -7.40270019e-01 -6.96014285e-01 -4.98137355e-01 2.82476664e-01 3.68584305e-01 3.46494138e-01 -1.05815852e+00 1.18347085e+00 5.84149122e+00 6.75203502e-01 -8.46781611e-01 -1.96596056e-01 6.44236147e-01 -2.35361665e-01 -6.61703765e-01 -2.52201378e-01 -9.23859656e-01 3.94821376e-01 8.70794415e-01 -1.16637915e-01 6.63619757e-01 7.73226559e-01 -1.62216425e-01 2.69354910e-01 -1.31992245e+00 9.28550661e-01 9.39468980e-01 -1.93427336e+00 9.10845399e-01 -1.99587941e-01 1.22577965e+00 -8.25827196e-03 4.76546466e-01 3.52280557e-01 7.48428226e-01 -1.31836593e+00 1.05398536e+00 6.59401298e-01 9.71807301e-01 -4.58855063e-01 2.08780035e-01 -2.69858036e-02 -7.62071431e-01 -1.39217064e-01 6.34651771e-03 5.33923090e-01 3.22644830e-01 2.95377851e-01 -7.58402944e-01 4.34697628e-01 4.30970818e-01 1.08363843e+00 -6.11168385e-01 2.43626028e-01 -1.83971465e-01 3.76217872e-01 1.04441814e-01 2.19278678e-01 5.70899367e-01 -8.37182850e-02 2.55059391e-01 1.56377959e+00 2.64959157e-01 -6.92428723e-02 1.60996020e-01 1.15051425e+00 -7.52339661e-01 -9.09832492e-02 -6.05370939e-01 -9.49189186e-01 3.43163401e-01 9.35219526e-01 -4.07359302e-01 -6.65158987e-01 -4.33877736e-01 1.57188654e+00 1.95787117e-01 5.05357802e-01 -6.65717244e-01 -3.31989110e-01 7.99080282e-02 -4.57835793e-01 6.36832058e-01 -4.52410549e-01 -5.73801458e-01 -1.33274865e+00 1.27288569e-02 -1.23118663e+00 4.00078624e-01 -1.38232911e+00 -1.24268866e+00 5.35855412e-01 -2.51885355e-01 -6.99213505e-01 -4.31187540e-01 -9.36500549e-01 -4.87635195e-01 6.65043294e-01 -1.42622876e+00 -1.64863038e+00 -2.71309167e-01 6.10208452e-01 1.18230963e+00 -5.68450391e-01 8.08564067e-01 -3.43457721e-02 -6.39459133e-01 5.86194634e-01 6.28990650e-01 2.13291839e-01 8.44216228e-01 -1.45518422e+00 9.01390672e-01 9.47360098e-01 4.81674999e-01 4.97288048e-01 2.70840257e-01 -6.29784346e-01 -1.82310736e+00 -1.05471087e+00 4.15676355e-01 -8.43451321e-01 8.12655270e-01 -7.00852096e-01 -7.62494266e-01 9.82727587e-01 6.15280569e-01 -5.56613982e-01 5.08479059e-01 -2.27172956e-01 -7.24907756e-01 3.04265618e-01 -5.97207725e-01 8.90950859e-01 9.63854432e-01 -9.40142155e-01 -5.48716545e-01 7.62768269e-01 9.58224475e-01 -7.23237872e-01 -7.50126898e-01 -1.68397620e-01 5.98564506e-01 -4.43726659e-01 1.05012238e+00 -7.85551369e-01 1.27166831e+00 -2.26827920e-03 -2.95286715e-01 -1.26556504e+00 -1.77118123e-01 -5.24215639e-01 -3.24665248e-01 1.56040096e+00 6.89487755e-01 7.80045567e-03 6.19629622e-01 4.92407054e-01 -5.83960533e-01 -3.21932346e-01 -1.34949014e-01 -5.56041539e-01 2.46726021e-01 -5.65786600e-01 6.40837550e-01 7.22501934e-01 -4.38160151e-01 5.20011067e-01 -3.30028445e-01 -1.82940811e-01 3.00933123e-01 4.17279929e-01 9.91170645e-01 -9.22385573e-01 -5.91049075e-01 -5.99499285e-01 4.34177637e-01 -1.41075790e+00 4.93888974e-01 -1.14519584e+00 7.83666596e-02 -1.97703803e+00 3.11898351e-01 1.70622338e-02 1.85705170e-01 4.88817096e-01 1.19616002e-01 1.07027832e-02 4.79426056e-01 3.30218047e-01 -8.04111004e-01 5.38728714e-01 1.45704293e+00 -8.46759856e-01 -1.82982251e-01 -7.34620333e-01 -1.14291012e+00 3.54407877e-01 4.97358978e-01 9.57654715e-02 -6.35299265e-01 -1.47298324e+00 4.40069102e-02 -8.68513212e-02 2.06233576e-01 -4.91537213e-01 3.41747344e-01 -1.45288125e-01 7.98079252e-01 -9.04542208e-01 3.27171385e-01 -3.08245927e-01 -5.86584449e-01 -1.09942228e-01 -9.28409874e-01 1.08270183e-01 4.00122404e-01 4.56812650e-01 -1.59767345e-01 -1.38038158e-01 4.35528785e-01 -3.67257923e-01 -9.86121118e-01 2.15631738e-01 -1.32311493e-01 2.91769385e-01 5.32718837e-01 -1.83981255e-01 -1.27137876e+00 -4.24207658e-01 -2.11803749e-01 1.64369777e-01 4.11603689e-01 8.90652657e-01 7.15646803e-01 -8.34949315e-01 -9.14852202e-01 1.28091186e-01 3.17992985e-01 6.09917939e-02 2.70071745e-01 2.11685926e-01 -3.72994810e-01 1.02134907e+00 -9.22101662e-02 -5.47381282e-01 -1.02253664e+00 5.06350338e-01 -3.29189971e-02 -6.01009846e-01 -5.22035599e-01 9.63590384e-01 3.49736750e-01 -5.52488148e-01 1.50057584e-01 -3.53364974e-01 1.11359224e-01 -9.28609520e-02 7.92515278e-01 9.80520435e-03 6.36007115e-02 -3.99149656e-01 1.50008753e-01 4.46891069e-01 -8.85324955e-01 -4.55439687e-01 1.18811643e+00 -1.03426717e-01 -1.02747135e-01 1.23323515e-01 1.32037580e+00 -2.16719285e-01 -1.22006929e+00 -1.61101848e-01 -2.69278244e-04 -1.20354019e-01 4.83953096e-02 -1.45537829e+00 -7.73442805e-01 8.02013218e-01 9.08266082e-02 -1.35380983e-01 9.88145649e-01 2.69092973e-02 1.02809346e+00 9.58396554e-01 -2.27891281e-01 -1.28624427e+00 8.10340941e-01 1.00560784e+00 1.47184193e+00 -1.30332136e+00 -3.52959521e-02 1.01262778e-01 -1.21881473e+00 1.11905634e+00 7.81987190e-01 5.60879568e-03 9.25701037e-02 4.41222012e-01 -1.08046122e-01 -1.59449533e-01 -9.84860241e-01 1.39539510e-01 6.20828688e-01 7.34014750e-01 5.44181764e-01 -1.33793950e-01 3.56572181e-01 6.48277879e-01 -3.02244842e-01 1.12016879e-01 4.72724438e-01 9.97527659e-01 -1.17083304e-01 -1.08139479e+00 -4.21896428e-01 7.61452317e-01 -1.56789735e-01 -5.18716693e-01 -7.88399875e-01 6.38498247e-01 -3.64402145e-01 6.17780745e-01 3.80389750e-01 1.17489845e-01 2.50066698e-01 2.16943875e-01 7.52830327e-01 -6.66619897e-01 -3.35855246e-01 -2.01234683e-01 1.16794303e-01 -2.43778422e-01 -1.86677814e-01 -5.15097439e-01 -1.26973116e+00 -1.93596721e-01 -3.53350639e-02 -1.49998710e-01 9.28149819e-01 1.02436769e+00 6.30795002e-01 7.43924618e-01 1.69049069e-01 -9.31611538e-01 -3.98994714e-01 -8.98219824e-01 -4.64073241e-01 5.97779512e-01 3.80779296e-01 -2.29950562e-01 3.10950074e-02 7.64336765e-01]
[11.458283424377441, 2.159536838531494]
2e9affce-96d6-4fdb-8342-ea9eea969635
point-voxel-transformer-an-efficient-approach
2108.06076
null
https://arxiv.org/abs/2108.06076v4
https://arxiv.org/pdf/2108.06076v4.pdf
PVT: Point-Voxel Transformer for Point Cloud Learning
The recently developed pure Transformer architectures have attained promising accuracy on point cloud learning benchmarks compared to convolutional neural networks. However, existing point cloud Transformers are computationally expensive since they waste a significant amount of time on structuring the irregular data. To solve this shortcoming, we present Sparse Window Attention (SWA) module to gather coarse-grained local features from non-empty voxels, which not only bypasses the expensive irregular data structuring and invalid empty voxel computation, but also obtains linear computational complexity with respect to voxel resolution. Meanwhile, to gather fine-grained features about the global shape, we introduce relative attention (RA) module, a more robust self-attention variant for rigid transformations of objects. Equipped with the SWA and RA, we construct our neural architecture called PVT that integrates both modules into a joint framework for point cloud learning. Compared with previous Transformer-based and attention-based models, our method attains top accuracy of 94.0% on classification benchmark and 10x inference speedup on average. Extensive experiments also valid the effectiveness of PVT on part and semantic segmentation benchmarks (86.6% and 69.2% mIoU, respectively).
['Xinyi Shen', 'Zizhao Wu', 'Haocheng Wan', 'Cheng Zhang']
2021-08-13
null
null
null
null
['3d-part-segmentation']
['computer-vision']
[ 1.54649774e-02 2.24809851e-02 8.69396254e-02 -3.25980216e-01 -8.66717219e-01 -1.68852851e-01 3.74145836e-01 1.61040798e-01 -2.74869114e-01 3.05131495e-01 -2.90525466e-01 -2.60659009e-01 -4.63167578e-02 -1.29051471e+00 -1.19601846e+00 -5.29153764e-01 8.43195394e-02 7.39618361e-01 6.60809636e-01 -7.01394826e-02 2.60306895e-01 7.40933061e-01 -1.63732958e+00 2.19775245e-01 1.18647301e+00 1.53701127e+00 2.27341712e-01 2.24021271e-01 -6.10446751e-01 3.46899688e-01 -3.50637317e-01 -3.08496803e-01 2.58967698e-01 3.57405692e-01 -8.43534172e-01 -1.14245296e-01 6.48377895e-01 -3.62541616e-01 -2.05863595e-01 9.86372471e-01 3.56259108e-01 6.03787526e-02 3.54245842e-01 -1.10646176e+00 -7.90291011e-01 6.38149321e-01 -8.73352110e-01 3.02977473e-01 -4.14090417e-02 7.02801421e-02 9.28490281e-01 -1.20690918e+00 2.75955737e-01 1.27144563e+00 8.41617644e-01 2.11560398e-01 -9.01391208e-01 -8.44337761e-01 4.52286065e-01 2.19765097e-01 -1.42199206e+00 -8.91492069e-02 8.62063348e-01 -1.83136329e-01 1.19862199e+00 5.13351381e-01 8.70005548e-01 5.60829043e-01 1.11508146e-01 8.31489027e-01 7.47180641e-01 1.09486066e-01 1.47738993e-01 -3.15384507e-01 3.20113033e-01 6.75475895e-01 2.23455042e-01 -1.13011159e-01 -1.70853019e-01 8.69247243e-02 1.13517153e+00 3.04608077e-01 -2.22178698e-01 -3.27904046e-01 -1.14627886e+00 7.48073459e-01 1.00415695e+00 1.63834408e-01 -5.07527411e-01 4.32609171e-01 3.08042049e-01 -8.37223679e-02 6.96515083e-01 2.75981307e-01 -6.08214080e-01 4.94965687e-02 -8.85863185e-01 2.37439245e-01 2.16860220e-01 1.11404943e+00 9.44772840e-01 7.02415481e-02 -4.29781735e-01 7.40699470e-01 2.39526600e-01 5.65453470e-01 3.40633929e-01 -7.52514839e-01 4.60777819e-01 1.01170993e+00 -1.43158227e-01 -1.02703214e+00 -3.85980844e-01 -7.46202528e-01 -8.70284140e-01 1.24907501e-01 2.87071317e-02 3.07674319e-01 -1.23705935e+00 1.22527564e+00 6.01969123e-01 5.55966496e-01 -3.67874205e-01 9.91331816e-01 1.18113804e+00 7.39663720e-01 1.29952282e-01 2.13954508e-01 1.47056437e+00 -1.27387404e+00 -2.26128101e-01 -1.96630806e-02 1.62749588e-01 -5.72805524e-01 1.29568231e+00 1.58302411e-01 -1.35843420e+00 -5.88673770e-01 -9.31900024e-01 -5.73517501e-01 -2.16753349e-01 -6.70171231e-02 8.62008452e-01 2.85292000e-01 -9.28850114e-01 7.81138480e-01 -1.17092836e+00 -1.51875671e-02 1.05679941e+00 6.04993820e-01 2.73005906e-02 5.25587015e-02 -6.42840385e-01 2.73880333e-01 -3.05719376e-02 2.13652715e-01 -7.33102977e-01 -1.35067439e+00 -8.29015374e-01 3.89488429e-01 3.36706787e-01 -7.68832982e-01 1.15387964e+00 -6.63425386e-01 -1.54462349e+00 6.09214306e-01 -1.58741057e-01 -4.83362734e-01 4.33264911e-01 -4.66672450e-01 2.02898756e-01 8.22575539e-02 2.48693451e-01 7.90871918e-01 8.36679101e-01 -1.20658433e+00 -6.03165507e-01 -5.70514262e-01 1.97162591e-02 1.24152601e-01 -2.13788345e-01 -2.05702201e-01 -8.34740341e-01 -6.49361849e-01 4.29187655e-01 -5.34775734e-01 -3.10099423e-01 1.65594593e-01 -3.11535120e-01 -6.44668698e-01 1.17868590e+00 -4.42866951e-01 7.93308735e-01 -2.07928729e+00 -3.47424974e-03 1.72521278e-01 4.96142447e-01 2.99349397e-01 -3.34318802e-02 -2.02133700e-01 -5.85615821e-02 1.29201889e-01 -4.50045735e-01 -5.09368420e-01 1.43680386e-02 2.59574801e-01 -5.41769385e-01 2.73426175e-01 4.36139286e-01 1.26923966e+00 -6.48904026e-01 -6.26897573e-01 5.16496837e-01 5.71369469e-01 -8.73903394e-01 -1.06792422e-02 -3.45552832e-01 3.20551723e-01 -7.39136159e-01 9.72221792e-01 1.11938500e+00 -4.35597777e-01 -5.18643558e-01 -3.83719355e-01 -1.57691926e-01 2.50838161e-01 -9.05897081e-01 1.85117018e+00 -6.01580262e-01 2.28835329e-01 1.13621362e-01 -9.05381203e-01 8.96622837e-01 -1.07255787e-01 6.71025336e-01 -7.50783324e-01 3.27914417e-01 4.87718731e-02 -3.46442789e-01 -2.30049327e-01 5.25896132e-01 8.20613727e-02 7.76834115e-02 1.65950898e-02 -1.57687381e-01 -2.17189461e-01 -2.88302392e-01 -4.50108089e-02 1.04413068e+00 3.20072442e-01 -2.46868491e-01 -4.06417549e-01 5.72799027e-01 1.23562172e-01 6.54907405e-01 4.44724828e-01 5.59111610e-02 7.93322563e-01 2.96397090e-01 -7.62666821e-01 -8.71271133e-01 -1.04809344e+00 -3.54683161e-01 1.01890779e+00 5.09200156e-01 -3.93970877e-01 -8.73116493e-01 -4.13026512e-01 1.55263498e-01 3.33031148e-01 -5.91961563e-01 5.73101304e-02 -8.47217262e-01 -7.49656856e-01 2.05786347e-01 1.07828844e+00 8.42669964e-01 -1.11631727e+00 -8.03052485e-01 2.36106038e-01 8.89358763e-03 -1.10437202e+00 -3.43304873e-01 1.57465767e-02 -1.15224349e+00 -8.74836087e-01 -5.16865194e-01 -7.35419512e-01 6.83150470e-01 3.15984815e-01 1.30934834e+00 3.74180168e-01 -1.82738379e-01 2.28548460e-02 -2.48650581e-01 -3.60644251e-01 4.43826497e-01 2.95085579e-01 -3.82963389e-01 -2.22424529e-02 1.55733332e-01 -7.81207204e-01 -6.84996247e-01 2.21426010e-01 -7.73109734e-01 1.86090261e-01 6.64496005e-01 7.06807435e-01 1.09722948e+00 -9.42146406e-02 1.59122437e-01 -7.69996941e-01 1.06895655e-01 -3.64121914e-01 -7.65308976e-01 -2.17178576e-02 -3.32330436e-01 1.00076608e-02 5.96649468e-01 -1.74042985e-01 -8.55268419e-01 3.47899497e-02 -5.16021252e-01 -9.05692339e-01 -1.77460853e-02 2.70891398e-01 -2.14242563e-01 -3.55946779e-01 1.17123023e-01 1.96709856e-01 -2.65348017e-01 -7.20249772e-01 1.43047407e-01 4.33914185e-01 6.28278971e-01 -9.02793705e-01 8.53415728e-01 7.37071335e-01 -6.64723814e-02 -6.74742162e-01 -7.10967302e-01 -2.53061503e-01 -5.80612123e-01 1.62790343e-01 9.16947603e-01 -9.35770631e-01 -9.36615646e-01 5.10231018e-01 -1.26310325e+00 -3.45991284e-01 -5.13858378e-01 1.00074783e-01 -4.00584072e-01 3.21189821e-01 -6.58680201e-01 -4.22065139e-01 -8.29013646e-01 -1.38443148e+00 1.59716070e+00 1.87629178e-01 2.09271714e-01 -4.51482862e-01 -4.32969630e-01 4.23137903e-01 5.04776478e-01 3.54481548e-01 1.01617730e+00 -3.02473485e-01 -1.14895654e+00 4.89609204e-02 -6.85185552e-01 4.46860492e-02 -2.40051299e-01 -2.80132368e-02 -9.03119802e-01 -2.11397618e-01 1.33482441e-02 -3.32254954e-02 9.55546737e-01 4.35454011e-01 1.89752483e+00 1.89389638e-03 -5.63863993e-01 1.19759011e+00 1.41584373e+00 -8.60265165e-04 5.75870156e-01 3.18876803e-01 1.09520113e+00 7.52757639e-02 4.96227562e-01 4.38922852e-01 4.80425447e-01 5.92688918e-01 8.58157635e-01 -3.13819140e-01 -1.98067278e-01 -7.31750801e-02 -1.48103744e-01 8.58297884e-01 -3.83650452e-01 8.64133760e-02 -9.61976707e-01 4.84478265e-01 -1.67379129e+00 -4.70977336e-01 -3.57659310e-01 1.93304920e+00 4.56881195e-01 2.13475645e-01 -1.31785661e-01 7.81068727e-02 4.73979026e-01 1.91741720e-01 -7.08818674e-01 -1.66340262e-01 1.17326632e-01 8.24920475e-01 6.27576828e-01 4.01326746e-01 -1.18408322e+00 1.11257815e+00 5.31549263e+00 1.09159446e+00 -1.30086446e+00 2.67512500e-01 5.36710322e-01 -4.92011122e-02 -4.27485853e-01 -2.93802500e-01 -6.95276201e-01 5.23163319e-01 5.05920470e-01 1.19068481e-01 2.61840612e-01 1.13557076e+00 -5.95779605e-02 2.50758231e-01 -9.37971950e-01 1.06363225e+00 -1.79079890e-01 -1.52871561e+00 2.70842820e-01 -7.32515156e-02 5.58113098e-01 4.12019402e-01 9.49439853e-02 4.33389753e-01 1.91341951e-01 -1.12948489e+00 8.90633464e-01 2.85406828e-01 8.27268302e-01 -9.07053232e-01 8.18095922e-01 1.90448761e-01 -1.64439428e+00 5.61961308e-02 -7.04171181e-01 1.69620104e-02 -5.29879816e-02 6.49224520e-01 -3.90383005e-01 7.75831580e-01 1.23356616e+00 8.45348775e-01 -6.09357893e-01 1.02963603e+00 9.19499695e-02 4.12055343e-01 -7.71378458e-01 2.09722593e-01 4.39870000e-01 -2.35600233e-01 4.60704952e-01 9.02690232e-01 4.28797215e-01 1.42531082e-01 1.93067074e-01 1.21008682e+00 -1.97962135e-01 6.69940487e-02 -1.10873863e-01 3.93058181e-01 4.93573189e-01 1.30074918e+00 -1.01142824e+00 -4.56827372e-01 -4.28639144e-01 8.34479690e-01 5.70993900e-01 8.56455192e-02 -1.17646539e+00 -3.68069053e-01 8.78021717e-01 2.52706856e-01 7.19503462e-01 -1.30836815e-01 -6.88241839e-01 -9.97863948e-01 2.96921194e-01 -4.82786983e-01 7.25511536e-02 -7.42554367e-01 -1.04155767e+00 8.55833292e-01 -1.65434226e-01 -1.07936549e+00 5.19778669e-01 -4.80944127e-01 -8.10637534e-01 6.54000044e-01 -1.80320883e+00 -1.44013393e+00 -7.71345615e-01 6.51133418e-01 7.65051723e-01 2.81426251e-01 4.85467196e-01 4.72521394e-01 -7.48480439e-01 5.86592913e-01 -1.50316939e-01 1.74734533e-01 1.07760191e-01 -1.24446201e+00 7.63573825e-01 5.44577956e-01 -9.86775383e-02 5.54270923e-01 1.96435124e-01 -5.72597921e-01 -1.55994654e+00 -1.53459954e+00 4.03310657e-01 -4.68741894e-01 3.71850759e-01 -3.63619775e-01 -1.22213328e+00 7.98336983e-01 -1.51431561e-01 4.95210886e-01 1.17666587e-01 4.17783111e-02 -2.94406652e-01 -2.83877045e-01 -1.01923203e+00 3.95911127e-01 1.28500152e+00 -7.63341784e-02 -6.16968870e-01 2.79570282e-01 1.20334291e+00 -8.15501332e-01 -1.07001972e+00 8.22425246e-01 2.23855078e-01 -9.97937858e-01 1.37427020e+00 -2.40836442e-01 5.41448236e-01 -3.82978886e-01 -6.58529550e-02 -8.04873228e-01 -5.35654664e-01 -3.51997256e-01 -6.93386123e-02 1.05243790e+00 1.05295293e-01 -5.88056266e-01 1.02969241e+00 4.58115369e-01 -8.01265299e-01 -1.27970362e+00 -1.12146568e+00 -6.50415838e-01 2.72548437e-01 -7.49994755e-01 1.27539742e+00 7.99198687e-01 -6.89561427e-01 1.97596923e-01 1.97164163e-01 3.72803658e-01 6.58061862e-01 6.36606216e-01 6.34203315e-01 -1.43535912e+00 7.65886679e-02 -6.74361229e-01 -4.27388251e-01 -1.23901057e+00 7.21157789e-02 -1.01291716e+00 -9.62488875e-02 -1.63861310e+00 -8.34801700e-03 -8.14977288e-01 -2.94223040e-01 5.95166326e-01 -1.96480006e-01 4.56180006e-01 9.51494575e-02 1.43743798e-01 -5.96084416e-01 8.31108391e-01 1.36486030e+00 -2.80019850e-01 -1.76696226e-01 -1.46005973e-01 -5.76808095e-01 8.87418389e-01 5.77019274e-01 -3.41448009e-01 -1.78103611e-01 -1.01560080e+00 -6.93855658e-02 -2.14583620e-01 6.60770118e-01 -1.21701288e+00 2.73477763e-01 -5.76579273e-02 4.25524890e-01 -1.16861773e+00 4.06043768e-01 -9.35241461e-01 -8.73288978e-03 3.06736559e-01 2.12422565e-01 9.41727534e-02 3.91722858e-01 4.63893354e-01 -1.76550061e-01 2.67421693e-01 8.41662705e-01 -7.86691085e-02 -6.17171824e-01 9.17718053e-01 2.09797934e-01 -1.35970816e-01 1.04923379e+00 -3.55105311e-01 -1.23141393e-01 2.88852721e-01 -5.16208827e-01 3.66236627e-01 4.45033967e-01 3.73831809e-01 7.68959761e-01 -1.37494004e+00 -4.95836526e-01 5.66949427e-01 -1.52329594e-01 1.04679000e+00 5.11732578e-01 8.32171082e-01 -7.99353480e-01 4.32046086e-01 -2.15155140e-01 -9.67385709e-01 -9.82460976e-01 6.59911871e-01 2.01443106e-01 -8.02398473e-02 -1.20913649e+00 1.21427810e+00 3.85152698e-01 -3.52182359e-01 7.76499361e-02 -9.50567245e-01 1.45515008e-02 -2.34802604e-01 2.07935259e-01 3.88249129e-01 3.74582589e-01 -4.80359733e-01 -5.15223145e-01 1.13311875e+00 -9.95144695e-02 5.18832266e-01 1.55927670e+00 1.81533128e-01 -3.74312818e-01 2.09206715e-02 1.08680737e+00 -3.95791158e-02 -1.40140831e+00 -2.70856440e-01 -2.26197645e-01 -5.37682533e-01 2.99858630e-01 -4.17383909e-01 -1.54694831e+00 1.02733135e+00 4.67646539e-01 6.90951496e-02 1.11086357e+00 1.52363837e-01 1.28760576e+00 1.41105995e-01 3.76329720e-01 -6.86620831e-01 -2.55824894e-01 4.74229604e-01 1.01382911e+00 -1.05821812e+00 9.30419490e-02 -9.09920335e-01 -7.11823404e-02 8.48081589e-01 9.06145513e-01 -6.75583959e-01 7.34497845e-01 3.81986350e-01 -3.40569317e-01 -4.27756101e-01 -5.79842806e-01 -3.12618688e-02 2.93071419e-01 4.28903252e-01 1.70509413e-01 7.94733167e-02 -2.62155458e-02 8.87143016e-01 -5.01516044e-01 -9.12714005e-02 -5.48980981e-02 7.39876509e-01 -4.29540604e-01 -6.78296864e-01 -3.45801920e-01 5.66068053e-01 -2.25724787e-01 -1.48887128e-01 1.66522746e-03 8.35961521e-01 3.00443947e-01 3.96547884e-01 7.27329493e-01 -2.85833657e-01 4.60492134e-01 -2.46213093e-01 3.22583526e-01 -5.67306757e-01 -7.92219877e-01 9.44947451e-02 -4.25059378e-01 -9.51870620e-01 -1.53030038e-01 -4.88865286e-01 -1.49783695e+00 -4.06228542e-01 -2.09698021e-01 5.27588911e-02 4.34954703e-01 8.64268363e-01 6.35328889e-01 9.68337357e-01 3.51429015e-01 -1.21930683e+00 -2.59058654e-01 -8.43648672e-01 -1.94331124e-01 2.50125885e-01 2.34127626e-01 -7.70515501e-01 -2.45823786e-02 -3.29583317e-01]
[7.93692684173584, -3.501772165298462]
dae157bc-3463-4c48-94e6-f87249c6d87f
non-generative-generalized-zero-shot-learning
2203.05335
null
https://arxiv.org/abs/2203.05335v4
https://arxiv.org/pdf/2203.05335v4.pdf
Non-generative Generalized Zero-shot Learning via Task-correlated Disentanglement and Controllable Samples Synthesis
Synthesizing pseudo samples is currently the most effective way to solve the Generalized Zero-Shot Learning (GZSL) problem. Most models achieve competitive performance but still suffer from two problems: (1) Feature confounding, the overall representations confound task-correlated and task-independent features, and existing models disentangle them in a generative way, but they are unreasonable to synthesize reliable pseudo samples with limited samples; (2) Distribution uncertainty, that massive data is needed when existing models synthesize samples from the uncertain distribution, which causes poor performance in limited samples of seen classes. In this paper, we propose a non-generative model to address these problems correspondingly in two modules: (1) Task-correlated feature disentanglement, to exclude the task-correlated features from task-independent ones by adversarial learning of domain adaption towards reasonable synthesis; (2) Controllable pseudo sample synthesis, to synthesize edge-pseudo and center-pseudo samples with certain characteristics towards more diversity generated and intuitive transfer. In addation, to describe the new scene that is the limit seen class samples in the training process, we further formulate a new ZSL task named the 'Few-shot Seen class and Zero-shot Unseen class learning' (FSZU). Extensive experiments on four benchmarks verify that the proposed method is competitive in the GZSL and the FSZU tasks.
['Jitao Sang', 'Jian Yu', 'Pengbo Yang', 'Xiaowen Huang', 'Yaogong Feng']
2022-03-10
null
http://openaccess.thecvf.com//content/CVPR2022/html/Feng_Non-Generative_Generalized_Zero-Shot_Learning_via_Task-Correlated_Disentanglement_and_Controllable_Samples_CVPR_2022_paper.html
http://openaccess.thecvf.com//content/CVPR2022/papers/Feng_Non-Generative_Generalized_Zero-Shot_Learning_via_Task-Correlated_Disentanglement_and_Controllable_Samples_CVPR_2022_paper.pdf
cvpr-2022-1
['generalized-zero-shot-learning', 'generalized-zero-shot-learning']
['computer-vision', 'methodology']
[ 4.42002058e-01 5.61301857e-02 -1.21738046e-01 -2.86644161e-01 -9.21430707e-01 -1.56275839e-01 8.15288246e-01 -4.82506692e-01 -6.97534010e-02 1.07755291e+00 1.11469857e-01 3.83409053e-01 -2.31711105e-01 -8.51111591e-01 -8.25166106e-01 -1.01284420e+00 4.13851798e-01 5.70155382e-01 3.63311708e-01 -3.85236323e-01 -9.35952552e-03 1.28967732e-01 -1.84620130e+00 3.07355791e-01 1.18403423e+00 9.95914519e-01 3.25960249e-01 1.89073920e-01 -2.39029542e-01 5.47454119e-01 -8.33338082e-01 -2.67142534e-01 2.58611441e-01 -7.04683959e-01 -2.03247353e-01 3.07685733e-01 1.41011328e-01 -2.62648702e-01 -2.29799449e-01 1.24574673e+00 6.32249594e-01 2.10519329e-01 1.13919783e+00 -1.89642954e+00 -1.16210055e+00 5.21978199e-01 -4.88036066e-01 -1.04751922e-01 1.65125534e-01 4.60563481e-01 4.91682291e-01 -1.19129634e+00 6.58900201e-01 1.53591239e+00 4.81913239e-01 8.52327943e-01 -1.26743388e+00 -9.40145314e-01 9.80038643e-02 2.41096169e-01 -1.55059314e+00 -3.60294670e-01 8.23132813e-01 -5.61018407e-01 5.16510427e-01 2.90602267e-01 4.41477448e-01 1.83334684e+00 2.32532248e-01 8.37264419e-01 1.25391912e+00 -1.33396864e-01 4.36764091e-01 4.68025386e-01 -1.01878829e-02 1.84219345e-01 3.16669255e-01 3.98302108e-01 -4.04888183e-01 5.77536412e-02 6.51665568e-01 2.89080977e-01 -4.79573339e-01 -6.41515315e-01 -1.11052144e+00 9.34220791e-01 3.92777979e-01 9.87559855e-02 -2.82944664e-02 -1.80947840e-01 2.77869076e-01 5.78023374e-01 5.30397952e-01 4.29705530e-01 -4.39676404e-01 2.41208449e-01 -5.84142864e-01 4.53184217e-01 5.24529397e-01 1.56315744e+00 9.68488514e-01 5.33927321e-01 -7.37170458e-01 8.62577140e-01 1.48530314e-02 6.97493494e-01 9.02999818e-01 -3.64475816e-01 4.74484533e-01 4.46179003e-01 1.29018426e-01 -8.47699940e-01 -2.50098333e-02 -5.72664559e-01 -1.10033953e+00 3.87232184e-01 2.16247946e-01 -1.94978774e-01 -1.17584753e+00 1.94043279e+00 2.05769092e-01 5.39521992e-01 2.43095621e-01 9.10826683e-01 1.29181063e+00 7.27945983e-01 8.90664011e-02 -3.88798863e-01 1.19087815e+00 -9.10387814e-01 -9.32747662e-01 -3.73597920e-01 8.63728896e-02 -6.16528153e-01 1.25916421e+00 2.83314228e-01 -6.85417533e-01 -9.32772219e-01 -1.30400205e+00 2.28854448e-01 -6.97500944e-01 -6.05043732e-02 3.07882398e-01 6.41691327e-01 -3.79944533e-01 6.45147979e-01 -2.35341921e-01 4.77707908e-02 7.36404300e-01 3.39251794e-02 -3.05166185e-01 -3.77328306e-01 -1.52450860e+00 7.99578190e-01 6.19484186e-01 -2.15212211e-01 -1.30206215e+00 -9.38233316e-01 -9.07974541e-01 1.59596443e-01 7.61125565e-01 -7.80392289e-01 8.89088154e-01 -1.02553129e+00 -1.33209527e+00 5.55597186e-01 1.25893161e-01 1.75305642e-03 8.15938115e-01 -1.29241690e-01 -6.27959430e-01 -4.19691414e-01 2.40114570e-01 4.63352382e-01 1.49460375e+00 -1.56090212e+00 -4.43197072e-01 -2.80327439e-01 -3.17474931e-01 3.65785331e-01 -2.63049871e-01 -6.16617262e-01 1.37583986e-02 -8.47955406e-01 -3.03379111e-02 -7.09055483e-01 5.49530759e-02 -4.83057462e-02 -3.89107287e-01 -2.55636215e-01 8.86983216e-01 -9.15787220e-02 8.71985316e-01 -2.39678574e+00 1.66963875e-01 -3.46924365e-01 1.57437161e-01 4.37164634e-01 -3.58822584e-01 2.74672925e-01 -2.34163091e-01 1.78821217e-02 -2.52043843e-01 -1.92653134e-01 -4.63551283e-02 2.80652761e-01 -5.41666806e-01 2.02515170e-01 5.17575920e-01 9.32647049e-01 -1.35617173e+00 -3.43279243e-01 2.95017719e-01 3.43032181e-01 -2.06863567e-01 3.87870461e-01 -1.59134880e-01 5.04701018e-01 -5.59031963e-01 5.45315802e-01 8.77715588e-01 1.16872400e-01 -3.71249050e-01 -3.71968120e-01 3.37522566e-01 -4.00706917e-01 -1.34378886e+00 1.63218522e+00 -2.56617397e-01 1.73039094e-01 -4.11216915e-01 -9.82241988e-01 1.14686143e+00 3.24774086e-01 -1.39437485e-02 -3.53490233e-01 3.02087635e-01 7.74092451e-02 -5.80427945e-02 -6.51060343e-01 2.04098821e-01 -5.83458304e-01 -2.79659867e-01 -6.31078612e-03 5.97588181e-01 -5.63417614e-01 -1.06702752e-01 -5.85239232e-02 6.83050394e-01 2.87593931e-01 3.56922001e-01 -1.77644566e-01 3.67935002e-01 -3.61737818e-01 8.74175370e-01 8.31725717e-01 -3.83684307e-01 1.03255522e+00 3.86767507e-01 -1.42235175e-01 -9.46216941e-01 -1.53263211e+00 -2.33053312e-01 9.01863337e-01 4.46551055e-01 5.29472716e-02 -4.37052608e-01 -7.25019157e-01 -1.27966106e-02 1.17926335e+00 -8.07176769e-01 -8.46898377e-01 4.96428683e-02 -7.50778854e-01 1.82296723e-01 4.57229406e-01 4.37026262e-01 -1.17660141e+00 -3.33957106e-01 2.28207111e-01 -1.10112742e-01 -7.20677495e-01 -3.82948875e-01 3.27735543e-01 -4.05668825e-01 -9.92856979e-01 -1.06744361e+00 -6.42448962e-01 6.44795299e-01 3.69603872e-01 9.96753395e-01 -6.05596483e-01 -3.94946486e-01 -1.05923273e-01 -5.64548612e-01 -7.42037654e-01 -3.08956593e-01 -2.47836098e-01 3.94258574e-02 3.23226362e-01 3.69218022e-01 -5.45232475e-01 -3.08782429e-01 2.99113065e-01 -1.05763662e+00 1.22943379e-01 5.99000990e-01 1.35425353e+00 4.87568170e-01 2.36242160e-01 1.08538747e+00 -1.27893102e+00 7.12725639e-01 -8.20420146e-01 -1.36051625e-01 3.50199610e-01 -5.69769263e-01 8.70744959e-02 9.88759816e-01 -8.14412236e-01 -1.33758414e+00 -1.85406536e-01 1.98493019e-01 -1.12812734e+00 -4.01100218e-01 4.27341461e-02 -7.70597279e-01 2.65970469e-01 9.90192592e-01 6.22451425e-01 5.06389178e-02 -1.62330344e-01 5.43803036e-01 5.48007309e-01 3.77105534e-01 -6.76735997e-01 1.10907567e+00 3.24545830e-01 -5.37941046e-02 -6.85923278e-01 -9.08498108e-01 -1.71687260e-01 -4.60361660e-01 -1.07265554e-01 8.31162274e-01 -1.05033970e+00 9.39548165e-02 5.38748503e-01 -8.91209006e-01 -7.74051622e-02 -7.52522528e-01 3.64013433e-01 -7.75456548e-01 1.70386657e-02 -1.38437241e-01 -8.19617867e-01 -1.19845822e-01 -1.18821383e+00 1.13980007e+00 4.20831442e-01 -4.68226075e-02 -6.55995131e-01 -1.52515233e-01 2.78168614e-03 3.88035387e-01 6.14205599e-01 9.04043317e-01 -8.23057115e-01 -4.18683797e-01 -1.67201191e-01 -2.44431034e-01 5.90540051e-01 3.12691450e-01 -2.90323317e-01 -1.19768298e+00 -3.67110252e-01 2.95482516e-01 -7.06114352e-01 7.59200394e-01 2.90415943e-01 1.19409287e+00 -1.68722332e-01 -2.46185109e-01 7.34026909e-01 1.45954072e+00 2.93355405e-01 6.95940077e-01 -8.53331834e-02 6.73885882e-01 4.66535717e-01 1.01649249e+00 4.84929562e-01 -1.12727545e-01 2.72562087e-01 3.99898589e-01 2.31100872e-01 -3.62319380e-01 -5.03163815e-01 1.42683119e-01 6.64314151e-01 -1.14120543e-02 -3.11113149e-01 -3.24593127e-01 4.86598462e-01 -1.93873882e+00 -1.02117383e+00 -5.95402457e-02 2.16909766e+00 6.90629244e-01 2.24881992e-01 -1.38753384e-01 1.67705998e-01 9.45823550e-01 3.07643831e-01 -7.97926664e-01 1.13528199e-01 -1.76464468e-01 2.25096196e-01 7.25561082e-02 8.99240375e-02 -9.75044906e-01 9.21917677e-01 4.88087940e+00 1.47084880e+00 -9.29488361e-01 2.04926223e-01 5.92848182e-01 -1.79872543e-01 -5.59502959e-01 -2.13664830e-01 -8.98570776e-01 7.68729270e-01 4.27926064e-01 -5.61459720e-01 3.77662838e-01 1.17159796e+00 -2.44765937e-01 2.65469462e-01 -1.20850122e+00 1.25746691e+00 4.32646662e-01 -8.31861615e-01 3.93992096e-01 -1.94668174e-01 8.96621764e-01 -4.41284060e-01 3.57040614e-01 9.43753421e-01 1.71772286e-01 -9.37233090e-01 7.10906982e-01 6.97061121e-01 1.23498631e+00 -8.52903485e-01 5.26922464e-01 6.17411673e-01 -9.47457135e-01 -1.50043160e-01 -8.83066654e-01 1.28868967e-01 -5.96021079e-02 6.26630604e-01 -6.19009912e-01 8.03398132e-01 3.35454822e-01 6.78141475e-01 -5.97756863e-01 8.05388391e-01 -3.27128738e-01 1.98767707e-01 2.07794085e-01 -1.61159802e-02 3.01033864e-03 -1.45199895e-01 7.29059696e-01 6.59179807e-01 6.94101512e-01 4.30775695e-02 1.42104805e-01 1.24254084e+00 1.48560077e-01 -2.69075871e-01 -8.53493273e-01 1.30865484e-01 6.16772056e-01 1.14729822e+00 -5.15263200e-01 -4.82681483e-01 -3.26535970e-01 1.07031167e+00 2.06194296e-01 5.12022555e-01 -9.53062415e-01 -5.55991113e-01 4.18434888e-01 4.16694432e-02 1.12044446e-01 4.13181573e-01 -1.66788086e-01 -1.53856432e+00 -1.14652589e-01 -9.11094546e-01 1.33125916e-01 -9.27491665e-01 -1.89932561e+00 6.71589315e-01 1.48954868e-01 -1.72484910e+00 -2.18853191e-01 -4.15879637e-01 -5.86204708e-01 1.02072370e+00 -1.35356736e+00 -1.19822967e+00 -5.37399292e-01 9.08808887e-01 1.00851643e+00 -5.84466338e-01 8.82827878e-01 1.86401024e-01 -4.47647274e-01 6.70219302e-01 2.14903951e-01 -2.98621297e-01 9.05873775e-01 -1.19311631e+00 1.86695755e-01 5.78466415e-01 -2.02153057e-01 3.06538075e-01 8.16637814e-01 -7.18200326e-01 -1.17969382e+00 -1.40277183e+00 3.95794094e-01 -4.77314830e-01 3.40056568e-01 -4.90333557e-01 -1.08828223e+00 4.86057341e-01 -1.59410432e-01 2.68901825e-01 6.93903208e-01 -1.99752942e-01 -3.72344583e-01 -1.85964659e-01 -1.35366714e+00 6.39845252e-01 1.13434243e+00 -2.50749290e-01 -1.04158235e+00 2.25098267e-01 8.13041151e-01 -2.05748081e-01 -4.62875217e-01 5.75343251e-01 3.25912982e-01 -1.01176143e+00 9.85127449e-01 -5.50119042e-01 6.48065627e-01 -2.44797274e-01 -1.95493668e-01 -1.87248230e+00 -5.83016634e-01 -2.35855624e-01 -1.39407128e-01 1.51626194e+00 1.21167945e-02 -6.73933625e-01 5.91397882e-01 4.44529474e-01 -2.67194450e-01 -6.45106137e-01 -7.90802240e-01 -1.22829056e+00 1.02518521e-01 -1.71340361e-01 8.87466133e-01 1.20926821e+00 -4.16759163e-01 6.00202262e-01 -6.73933685e-01 -1.52437434e-01 7.72275865e-01 5.43262437e-02 9.04131472e-01 -1.33477783e+00 -4.57501054e-01 -3.00185919e-01 -2.16679618e-01 -7.42926359e-01 9.25000906e-02 -8.17963004e-01 2.45779395e-01 -1.22942257e+00 3.01584482e-01 -3.87017757e-01 -2.81242907e-01 9.95238870e-02 -5.37009478e-01 -9.37602445e-02 3.04713726e-01 9.10237059e-02 -3.35505486e-01 1.14282894e+00 1.75540388e+00 -1.87758476e-01 -2.78721061e-02 1.09631218e-01 -7.74667799e-01 5.78946352e-01 4.23442602e-01 -5.16741753e-01 -1.11875939e+00 -7.77699724e-02 -2.86637872e-01 1.99101925e-01 2.97132313e-01 -1.08964229e+00 -2.14103431e-01 -4.33446646e-01 7.53326535e-01 -3.88888657e-01 5.71371853e-01 -8.63523185e-01 1.32713929e-01 3.73173982e-01 -1.30595386e-01 -7.28993952e-01 -7.48906881e-02 9.35516357e-01 -1.49118468e-01 -3.82626355e-01 9.55844581e-01 -3.15613121e-01 -7.97311604e-01 6.00947618e-01 2.98135411e-02 4.03974444e-01 1.39983904e+00 -2.79063046e-01 -4.49781388e-01 -3.65900397e-01 -8.28399241e-01 8.24596211e-02 1.76134422e-01 7.43551135e-01 7.70064116e-01 -1.62766492e+00 -8.19030344e-01 7.41067767e-01 4.14773762e-01 2.16442481e-01 6.28889978e-01 2.67781883e-01 1.55259836e-02 -9.02360827e-02 -3.93744826e-01 -4.76754010e-01 -7.19976008e-01 9.74263489e-01 -7.42154494e-02 -2.99525391e-02 -4.27211732e-01 9.18411970e-01 7.89110184e-01 -4.48847771e-01 6.42524846e-03 -3.77058238e-02 -1.93050563e-01 2.72847950e-01 5.85475743e-01 3.84361088e-01 -2.22942039e-01 -4.57924783e-01 2.65600029e-02 2.13343903e-01 -1.35150924e-03 1.21336982e-01 1.15442133e+00 -6.86368020e-03 5.04545927e-01 8.81191313e-01 1.16960001e+00 -5.51744044e-01 -1.59926701e+00 -2.60519564e-01 -4.71131414e-01 -7.33383775e-01 -2.84450352e-01 -7.32218027e-01 -8.26476216e-01 1.06421471e+00 7.79310167e-01 2.77422100e-01 1.01941144e+00 -2.13935301e-01 6.98996365e-01 -2.52435859e-02 5.39640546e-01 -9.95734274e-01 2.73890316e-01 3.28466713e-01 1.11671245e+00 -1.41387498e+00 -2.23132834e-01 -6.16808712e-01 -9.09143150e-01 8.31877232e-01 1.02507591e+00 -4.31018740e-01 7.08016992e-01 1.65965743e-02 -2.49840498e-01 1.29552156e-01 -7.18493283e-01 -1.99572161e-01 3.96113187e-01 1.10287428e+00 -5.36784865e-02 1.09344788e-01 -2.18936592e-01 1.12896812e+00 -8.64934847e-02 9.36796442e-02 3.31647992e-01 6.65428221e-01 -4.58188146e-01 -8.22754681e-01 -3.37727040e-01 6.05881631e-01 6.89084008e-02 3.23346183e-02 -7.81008303e-02 9.36656475e-01 6.28412426e-01 7.21067429e-01 -7.64606595e-02 -4.59277362e-01 4.30995524e-01 3.23826700e-01 4.16944861e-01 -9.00387049e-01 -6.63883686e-02 1.50275886e-01 -2.56590366e-01 -1.88970312e-01 6.04055785e-02 -3.54132265e-01 -7.86316574e-01 8.48397091e-02 -3.63947779e-01 -1.19087428e-01 1.55094564e-01 9.75284755e-01 1.58121243e-01 1.01861131e+00 7.49645472e-01 -9.85276282e-01 -1.04697144e+00 -1.10689747e+00 -1.08611155e+00 9.60185409e-01 8.57684612e-02 -1.15367675e+00 -6.95951700e-01 -5.58405817e-02]
[10.041853904724121, 2.7386059761047363]
3bf29711-7c69-4318-81a1-01b76cf0e567
cross-modal-implicit-relation-reasoning-and
2303.12501
null
https://arxiv.org/abs/2303.12501v1
https://arxiv.org/pdf/2303.12501v1.pdf
Cross-Modal Implicit Relation Reasoning and Aligning for Text-to-Image Person Retrieval
Text-to-image person retrieval aims to identify the target person based on a given textual description query. The primary challenge is to learn the mapping of visual and textual modalities into a common latent space. Prior works have attempted to address this challenge by leveraging separately pre-trained unimodal models to extract visual and textual features. However, these approaches lack the necessary underlying alignment capabilities required to match multimodal data effectively. Besides, these works use prior information to explore explicit part alignments, which may lead to the distortion of intra-modality information. To alleviate these issues, we present IRRA: a cross-modal Implicit Relation Reasoning and Aligning framework that learns relations between local visual-textual tokens and enhances global image-text matching without requiring additional prior supervision. Specifically, we first design an Implicit Relation Reasoning module in a masked language modeling paradigm. This achieves cross-modal interaction by integrating the visual cues into the textual tokens with a cross-modal multimodal interaction encoder. Secondly, to globally align the visual and textual embeddings, Similarity Distribution Matching is proposed to minimize the KL divergence between image-text similarity distributions and the normalized label matching distributions. The proposed method achieves new state-of-the-art results on all three public datasets, with a notable margin of about 3%-9% for Rank-1 accuracy compared to prior methods.
['Mang Ye', 'Ding Jiang']
2023-03-22
null
http://openaccess.thecvf.com//content/CVPR2023/html/Jiang_Cross-Modal_Implicit_Relation_Reasoning_and_Aligning_for_Text-to-Image_Person_Retrieval_CVPR_2023_paper.html
http://openaccess.thecvf.com//content/CVPR2023/papers/Jiang_Cross-Modal_Implicit_Relation_Reasoning_and_Aligning_for_Text-to-Image_Person_Retrieval_CVPR_2023_paper.pdf
cvpr-2023-1
['person-retrieval', 'nlp-based-person-retrival', 'text-matching']
['computer-vision', 'computer-vision', 'natural-language-processing']
[ 2.32607707e-01 -1.29165530e-01 -4.08311784e-01 -5.08486509e-01 -1.05193448e+00 -5.86498857e-01 1.16982496e+00 1.43223912e-01 -5.26031196e-01 1.64702401e-01 5.19338787e-01 3.29523504e-01 -6.60326779e-02 -2.75845140e-01 -4.99203235e-01 -6.01034760e-01 4.15846586e-01 5.19988894e-01 -2.03974694e-01 6.56818897e-02 9.01713595e-02 -2.60845944e-02 -1.62906837e+00 8.70677948e-01 7.37681448e-01 1.02465963e+00 3.58242095e-02 4.18463498e-01 -1.42055541e-01 7.12354481e-01 -1.54673204e-01 -8.54213536e-01 2.02407345e-01 -2.91730374e-01 -7.88877666e-01 3.71982068e-01 9.68154967e-01 -3.39482278e-01 -6.55881703e-01 9.81480181e-01 6.86825395e-01 4.56255451e-02 1.06561327e+00 -1.49915838e+00 -1.06249309e+00 2.25491449e-01 -7.07592368e-01 -2.93736309e-01 6.39665127e-01 1.73454121e-01 1.14805663e+00 -1.15448153e+00 4.98704433e-01 1.56873751e+00 3.52870762e-01 5.69493353e-01 -1.45147014e+00 -4.11754608e-01 6.00555725e-02 4.00905043e-01 -1.69536388e+00 -5.90160728e-01 6.24511003e-01 -6.83312535e-01 8.73777032e-01 3.51290643e-01 3.17334920e-01 1.19692481e+00 -1.61743343e-01 1.01717997e+00 9.44070995e-01 -5.09385467e-01 -4.32312757e-01 6.20821059e-01 -1.38682172e-01 7.07824826e-01 -2.14187413e-01 -1.35471240e-01 -9.78218913e-01 -1.54039130e-01 5.02835631e-01 2.53544718e-01 -1.12654559e-01 -6.68257535e-01 -1.35935307e+00 7.56458282e-01 4.38826710e-01 2.40172163e-01 -1.47737831e-01 1.47044539e-01 5.38900316e-01 3.66832651e-02 2.72881657e-01 1.54824868e-01 1.17269538e-01 1.84248880e-01 -9.66572642e-01 1.80749550e-01 1.66648149e-01 1.02200246e+00 7.07471192e-01 -4.73312825e-01 -7.71229446e-01 1.00352836e+00 7.61971354e-01 6.97698832e-01 3.82450461e-01 -7.72770941e-01 7.58194268e-01 8.26940417e-01 -1.34437317e-02 -1.23442209e+00 9.76057053e-02 -1.08098991e-01 -7.12349296e-01 -2.14762360e-01 2.71696419e-01 4.61807817e-01 -8.33459496e-01 1.85975015e+00 1.77863970e-01 -3.33260417e-01 4.37437184e-02 1.11705875e+00 9.07803595e-01 4.52190489e-01 2.78247237e-01 1.07000127e-01 1.70294166e+00 -1.17449105e+00 -8.48520458e-01 -4.12351847e-01 4.91445750e-01 -1.06586146e+00 1.28935778e+00 -1.50976568e-01 -1.18498290e+00 -5.95029473e-01 -9.26118195e-01 -5.49519300e-01 -4.07737643e-01 4.90797400e-01 3.35333049e-01 3.66610259e-01 -9.44842279e-01 -1.81694388e-01 -5.58309913e-01 -5.64369678e-01 1.11590825e-01 2.39875197e-01 -6.24636889e-01 -3.00413132e-01 -1.15174651e+00 9.09496248e-01 2.06418797e-01 5.23755029e-02 -7.65955329e-01 -4.97828573e-01 -1.06938136e+00 -1.05750553e-01 2.77755290e-01 -1.02264571e+00 9.76933420e-01 -9.18416917e-01 -1.11855936e+00 1.28000164e+00 -4.85697597e-01 -4.88268817e-03 5.51303923e-01 -3.50136817e-01 -2.28948787e-01 3.78231674e-01 3.24754179e-01 1.06213176e+00 9.11785007e-01 -1.48010671e+00 -4.24378991e-01 -4.94064033e-01 1.47284064e-02 6.17912173e-01 -8.61365259e-01 3.63696180e-02 -1.20090568e+00 -6.60538256e-01 -2.26326920e-02 -8.57384801e-01 1.89630091e-01 1.95393428e-01 -3.79927397e-01 -4.36570764e-01 6.47450745e-01 -6.22236609e-01 1.00465083e+00 -2.14297318e+00 4.86012161e-01 9.84487906e-02 2.52685368e-01 -1.58350587e-01 -3.59843642e-01 5.68284512e-01 9.08106565e-02 -9.63343605e-02 1.38279065e-01 -9.30136800e-01 4.86474544e-01 -1.58695299e-02 -3.62425834e-01 5.53053856e-01 4.29831930e-02 1.13542378e+00 -6.20507658e-01 -9.69071388e-01 3.00337821e-01 7.03774035e-01 -4.48288977e-01 4.19622838e-01 -6.61686435e-02 3.19594830e-01 -1.69278920e-01 7.69658148e-01 5.47560573e-01 -3.85841191e-01 1.33922130e-01 -7.24673271e-01 1.18860602e-01 -1.82263497e-02 -7.88349330e-01 2.07653522e+00 -3.23496073e-01 7.12721109e-01 -1.35464340e-01 -9.28215921e-01 5.79996407e-01 4.41959351e-01 6.12494826e-01 -9.39195335e-01 1.37978852e-01 -2.86366455e-02 -5.42622447e-01 -6.84442639e-01 5.27522445e-01 -1.18448101e-01 -3.38994354e-01 3.49984437e-01 1.12679318e-01 3.84120643e-01 7.39049017e-02 5.77726066e-01 5.74007213e-01 6.71627223e-02 -1.07182376e-01 2.29652494e-01 6.09322965e-01 -1.43063143e-01 1.24755286e-01 6.75201654e-01 -6.73848391e-02 7.50781596e-01 2.35108286e-01 -6.38299212e-02 -9.86957550e-01 -1.36718106e+00 -3.79130170e-02 1.32971370e+00 4.63960022e-01 -6.49148643e-01 -4.86995727e-01 -5.87582052e-01 8.00625160e-02 3.99721146e-01 -6.46631241e-01 -1.05868034e-01 -1.40677661e-01 -3.42151076e-01 7.18861580e-01 4.74659145e-01 5.70058048e-01 -6.80662453e-01 -2.23010778e-01 -4.07907605e-01 -7.53359020e-01 -1.46779418e+00 -1.02697861e+00 -2.94134468e-01 -3.36878568e-01 -7.99369574e-01 -9.03466463e-01 -1.00518477e+00 9.12701905e-01 3.71565551e-01 9.69964087e-01 -1.04578212e-01 -5.34421027e-01 1.06745481e+00 -1.75313652e-01 2.25521892e-01 2.07112376e-02 -9.51003432e-02 4.11961935e-02 3.36198777e-01 5.95765114e-01 -9.99744087e-02 -7.84106374e-01 3.88437629e-01 -1.00213861e+00 3.66705209e-01 7.59103060e-01 9.72662866e-01 4.77009624e-01 -3.43614906e-01 2.23919656e-02 -1.46931291e-01 5.23486316e-01 -2.31578588e-01 -1.83702573e-01 8.16150784e-01 -4.62170333e-01 1.59936234e-01 6.37125298e-02 -7.61215508e-01 -1.08216059e+00 2.29623064e-01 5.65714598e-01 -7.78840125e-01 -6.10564724e-02 4.75887507e-01 -3.50968391e-01 1.64253697e-01 3.27194721e-01 4.00361925e-01 -1.97404847e-02 -2.58332849e-01 8.29921067e-01 8.18866968e-01 8.08278859e-01 -7.85538018e-01 9.23864901e-01 6.05994463e-01 -2.42222309e-01 -5.18505454e-01 -7.96352863e-01 -8.39167535e-01 -9.04144406e-01 -3.43700409e-01 1.26251054e+00 -1.22140968e+00 -8.22198510e-01 2.16991290e-01 -1.20186138e+00 4.58407402e-02 1.92784742e-01 5.13977051e-01 -5.08818567e-01 7.02212572e-01 -5.70124030e-01 -6.64373696e-01 -3.70478332e-01 -1.14225507e+00 1.46011984e+00 3.77612188e-02 -2.98082709e-01 -9.04253185e-01 -2.74811611e-02 1.02984345e+00 2.65028536e-01 -3.76898386e-02 9.33810055e-01 -3.62499356e-01 -6.79845691e-01 -2.28615463e-01 -6.72509015e-01 7.56246597e-02 1.71922430e-01 -4.24431950e-01 -9.03442621e-01 -4.61676329e-01 -3.20535719e-01 -7.24765420e-01 8.59933317e-01 9.27832676e-04 8.34057212e-01 -2.76805013e-01 -4.56267238e-01 3.99427503e-01 1.16595078e+00 -2.38640979e-01 4.46460873e-01 2.88867325e-01 9.81591821e-01 9.18952882e-01 5.51449656e-01 3.14979970e-01 7.71119058e-01 1.15862560e+00 1.14410616e-01 -1.73479602e-01 -1.60943732e-01 -4.56487119e-01 5.61583519e-01 6.07071579e-01 3.99068892e-01 -2.14138076e-01 -8.78301978e-01 7.07224071e-01 -2.12346840e+00 -1.02354014e+00 1.48257658e-01 2.15515494e+00 1.06777573e+00 -3.50908548e-01 -3.78336236e-02 -3.32351297e-01 6.79680288e-01 1.14344425e-01 -3.62562180e-01 3.56899463e-02 -1.24734297e-01 -4.29105729e-01 2.74777025e-01 4.41080272e-01 -1.21678960e+00 9.13149059e-01 5.38645172e+00 8.69641185e-01 -9.26659882e-01 2.25575387e-01 1.80118173e-01 -4.18386102e-01 -4.93296415e-01 -2.73041092e-02 -6.79543138e-01 2.21159294e-01 3.90036911e-01 5.24879321e-02 3.54607761e-01 3.98068607e-01 -1.73239224e-02 3.70861813e-02 -1.59848130e+00 1.57003081e+00 6.55997455e-01 -9.45744395e-01 4.84391302e-01 1.53990552e-01 6.21658444e-01 -4.36485082e-01 6.16075158e-01 3.57826889e-01 -1.71000794e-01 -1.32406223e+00 7.67501771e-01 8.89168918e-01 9.95359838e-01 -5.80628812e-01 5.94011843e-01 8.26094151e-02 -1.38684916e+00 1.63433209e-01 -1.06943697e-01 3.64639729e-01 3.18055063e-01 1.51036605e-01 -7.45711565e-01 6.07718229e-01 7.50122666e-01 6.12482190e-01 -8.44363451e-01 6.63095891e-01 7.66595230e-02 -8.54935646e-02 -1.03609659e-01 3.83771211e-01 1.31240308e-01 1.67044550e-02 4.13494557e-01 1.29788935e+00 9.99932960e-02 -2.69697160e-01 4.28106934e-01 1.02196896e+00 -1.26534775e-01 2.23838925e-01 -5.26939452e-01 -2.88286835e-01 3.15957129e-01 1.17400897e+00 -1.58037528e-01 -3.39937091e-01 -6.69154465e-01 1.45530438e+00 3.05972606e-01 5.18576682e-01 -8.55627656e-01 -2.17974484e-01 4.79840577e-01 1.22870512e-01 -1.86454747e-02 -1.40978888e-01 -1.74561292e-01 -1.34664035e+00 2.64550298e-01 -1.01500452e+00 5.55923641e-01 -8.37495148e-01 -1.68093169e+00 3.71881723e-01 3.13626140e-01 -1.16048563e+00 -3.53688300e-01 -4.32092547e-01 -1.23239011e-01 9.83376145e-01 -1.35505950e+00 -1.92950690e+00 -4.42400813e-01 9.37502265e-01 4.84507471e-01 -3.68906260e-01 6.97055995e-01 6.09359443e-01 -3.73645574e-01 1.15812778e+00 -1.95873722e-01 2.70705134e-01 1.41778314e+00 -1.07666516e+00 -1.85408711e-01 6.03667200e-01 1.61040351e-01 9.12181139e-01 6.10406041e-01 -5.84781647e-01 -1.56022596e+00 -9.81159925e-01 1.00477433e+00 -7.50220239e-01 4.63588238e-01 -4.86309320e-01 -7.44220376e-01 6.42481863e-01 6.25198781e-01 -1.16870150e-01 8.85157347e-01 1.49741754e-01 -7.87039638e-01 2.88407821e-02 -7.75131762e-01 8.97480309e-01 9.21613216e-01 -1.21953070e+00 -6.67441428e-01 2.41948217e-01 4.72867846e-01 -2.32121333e-01 -7.96253979e-01 3.99459630e-01 7.79219985e-01 -6.21967912e-01 1.24444103e+00 -4.51264918e-01 4.84854430e-01 -4.74747896e-01 -5.19749224e-01 -6.96683705e-01 -1.51469335e-01 -1.81367084e-01 -1.26933102e-02 1.64492667e+00 1.25399411e-01 -2.41034403e-01 5.46974003e-01 8.74533296e-01 2.60563135e-01 -4.72038865e-01 -7.54336774e-01 -5.71441650e-01 -1.40864685e-01 -2.16723353e-01 6.86307400e-02 9.63332295e-01 2.38653988e-01 5.43601394e-01 -7.23527670e-01 3.33608061e-01 8.79560351e-01 1.01780467e-01 9.28924561e-01 -7.75054753e-01 -2.42215306e-01 -4.54562157e-01 -4.23808932e-01 -1.23843992e+00 3.38641047e-01 -1.24964571e+00 1.51885133e-02 -1.59278107e+00 1.02375388e+00 -2.35125683e-02 -2.73224920e-01 5.55070996e-01 -1.26523316e-01 4.44387048e-01 3.32330644e-01 4.66670573e-01 -1.14033651e+00 9.47364926e-01 9.93222833e-01 -6.04160964e-01 -6.73012622e-03 -5.86990774e-01 -5.05894125e-01 5.18578589e-01 3.82995248e-01 -2.78529018e-01 -5.57604849e-01 -7.74403691e-01 2.23857582e-01 -9.35484692e-02 7.74376392e-01 -6.95695162e-01 5.64127266e-01 -1.02027193e-01 5.65638721e-01 -8.44464123e-01 7.42983758e-01 -8.31433654e-01 -2.80408170e-02 6.99536726e-02 -8.92326117e-01 6.35968149e-02 5.03361300e-02 6.77690923e-01 -4.96812105e-01 8.45576078e-03 4.54423398e-01 1.78663358e-01 -5.78225553e-01 2.27045283e-01 -1.89527854e-01 -2.12807246e-02 9.09101367e-01 -9.74406302e-02 -3.54570985e-01 -5.42785287e-01 -6.16688609e-01 5.83819687e-01 5.43583155e-01 7.59923398e-01 7.07052886e-01 -1.86801660e+00 -6.46796167e-01 -1.17322449e-02 7.41577566e-01 -4.58246946e-01 4.68159199e-01 9.77940500e-01 1.99271012e-02 6.92077994e-01 -6.26286715e-02 -9.54051733e-01 -1.67261839e+00 6.02509916e-01 3.30337524e-01 -1.69141725e-01 -3.13392907e-01 6.77912354e-01 5.95115542e-01 -4.85613316e-01 5.75883150e-01 2.48726889e-01 -7.54564181e-02 2.44355679e-01 4.92417812e-01 -2.19200272e-02 -3.36961597e-01 -1.13538241e+00 -4.98165011e-01 8.11286151e-01 -1.85310736e-01 -5.71325779e-01 7.91956663e-01 -5.94913602e-01 -1.97720155e-01 4.54949051e-01 1.61760461e+00 -4.20047604e-02 -9.71581399e-01 -7.82066405e-01 -5.81112206e-02 -6.36917412e-01 -1.05497353e-01 -7.05577016e-01 -8.69908035e-01 1.07269597e+00 9.28946137e-01 -2.57785648e-01 9.86754239e-01 4.24473315e-01 7.69209981e-01 2.30070844e-01 -8.10591578e-02 -1.05113626e+00 7.78272510e-01 2.10680038e-01 9.89459991e-01 -1.56707799e+00 9.13753733e-02 -1.92583174e-01 -9.08632457e-01 9.34356928e-01 8.03570569e-01 5.18812001e-01 2.23508939e-01 -2.44222343e-01 1.11372009e-01 -3.05856526e-01 -8.18339050e-01 -4.28449988e-01 1.11359894e+00 4.64493722e-01 5.78828990e-01 -1.21396080e-01 -5.61274476e-02 4.82759058e-01 1.63430750e-01 -4.10107344e-01 -4.17764395e-01 6.90372825e-01 -9.20135602e-02 -1.21535432e+00 -4.70748991e-01 -4.95276079e-02 -3.71635735e-01 -1.49464101e-01 -5.66653907e-01 5.26568234e-01 1.10288553e-01 9.86661315e-01 5.12610227e-02 -3.97031069e-01 2.43039683e-01 1.67539358e-01 6.46992564e-01 -3.76999348e-01 -2.66735524e-01 3.36547047e-01 -1.53802171e-01 -4.65008438e-01 -6.02774739e-01 -6.60309732e-01 -1.06776798e+00 -1.11830324e-01 -4.87470590e-02 -1.04727387e-01 4.42929000e-01 9.50622916e-01 3.36052984e-01 1.93581283e-01 2.78639138e-01 -8.57894719e-01 -3.12587976e-01 -8.83648038e-01 -1.33392200e-01 9.12688017e-01 1.65407628e-01 -7.06596792e-01 -2.84634411e-01 3.06151599e-01]
[10.956867218017578, 1.3728755712509155]
e7800b42-9235-4d08-8cea-c166cc765de3
memory-efficient-cnn-accelerator-based-on
2110.06155
null
https://arxiv.org/abs/2110.06155v1
https://arxiv.org/pdf/2110.06155v1.pdf
Memory-Efficient CNN Accelerator Based on Interlayer Feature Map Compression
Existing deep convolutional neural networks (CNNs) generate massive interlayer feature data during network inference. To maintain real-time processing in embedded systems, large on-chip memory is required to buffer the interlayer feature maps. In this paper, we propose an efficient hardware accelerator with an interlayer feature compression technique to significantly reduce the required on-chip memory size and off-chip memory access bandwidth. The accelerator compresses interlayer feature maps through transforming the stored data into frequency domain using hardware-implemented 8x8 discrete cosine transform (DCT). The high-frequency components are removed after the DCT through quantization. Sparse matrix compression is utilized to further compress the interlayer feature maps. The on-chip memory allocation scheme is designed to support dynamic configuration of the feature map buffer size and scratch pad size according to different network-layer requirements. The hardware accelerator combines compression, decompression, and CNN acceleration into one computing stream, achieving minimal compressing and processing delay. A prototype accelerator is implemented on an FPGA platform and also synthesized in TSMC 28-nm COMS technology. It achieves 403GOPS peak throughput and 1.4x~3.3x interlayer feature map reduction by adding light hardware area overhead, making it a promising hardware accelerator for intelligent IoT devices.
['Zhongfeng Wang', 'Chenjia Xie', 'Huadong Wei', 'Wei Zhuang', 'Yuan Du', 'Lei Chen', 'Li Du', 'Xiaoliang Chen', 'Zhuang Shao']
2021-10-12
null
null
null
null
['feature-compression']
['computer-vision']
[ 2.44035274e-01 -7.62471855e-02 -4.01182950e-01 -5.10971546e-01 2.76962042e-01 3.48407812e-02 3.02992195e-01 4.73860413e-01 -9.23613310e-01 1.50622070e-01 -1.39381122e-02 -6.71015799e-01 1.05149992e-01 -1.08147216e+00 -5.79813659e-01 -5.02852380e-01 -3.06151249e-02 -4.01579410e-01 4.24958408e-01 1.79343253e-01 3.01427573e-01 4.98270392e-01 -2.01914716e+00 7.10426807e-01 4.57240283e-01 1.52275646e+00 5.49534678e-01 6.35565519e-01 -1.51954040e-01 6.25282884e-01 -5.90158045e-01 -2.21028730e-01 5.85316300e-01 2.01654941e-01 -1.56684697e-01 -2.35231057e-01 4.38592494e-01 -5.31459689e-01 -7.81478643e-01 1.20720267e+00 2.32307494e-01 -3.49646002e-01 3.11331838e-01 -1.15301669e+00 -6.17190301e-02 6.50057375e-01 -4.56295431e-01 3.24618787e-01 -2.98610270e-01 -7.97548518e-02 4.16872144e-01 -6.65852666e-01 2.41504520e-01 1.11868286e+00 5.92042685e-01 1.79694653e-01 -5.52915812e-01 -1.04948997e+00 -2.15104580e-01 6.40765071e-01 -1.36976004e+00 -4.49188828e-01 3.62211227e-01 -2.36393008e-02 1.74380922e+00 2.87692040e-01 1.07939768e+00 3.04806858e-01 9.98835325e-01 2.04874635e-01 4.57879961e-01 -4.25835103e-01 5.34481585e-01 -1.93943530e-01 2.72402793e-01 8.14453006e-01 8.66124868e-01 -1.70036674e-01 -6.18595660e-01 8.16160589e-02 6.70279860e-01 4.50687587e-01 4.88105044e-02 5.04523873e-01 -8.80237341e-01 6.12077296e-01 6.40035510e-01 6.24696352e-02 -3.70463967e-01 4.24060553e-01 9.86443877e-01 4.53445375e-01 -1.32998958e-01 -4.37737554e-01 -5.20956278e-01 -1.46363601e-01 -9.82706010e-01 1.38336420e-01 6.90862834e-01 1.44234324e+00 7.40951240e-01 3.76909971e-01 4.19013560e-01 4.21671808e-01 4.21764821e-01 3.90890777e-01 1.13896811e+00 -5.82613230e-01 7.71431744e-01 9.58714366e-01 -7.42993593e-01 -7.82795906e-01 -5.13912499e-01 -2.21986890e-01 -1.30564559e+00 1.91158369e-01 -2.92314887e-01 -3.73991467e-02 -8.21119666e-01 1.09124088e+00 1.34388864e-01 6.83336705e-02 2.86966980e-01 4.16173398e-01 7.52596557e-01 9.54860806e-01 1.74435079e-01 -1.61810666e-02 1.88955820e+00 -9.99915719e-01 -7.95962393e-01 -4.73633468e-01 9.36616242e-01 -9.62364554e-01 7.44683623e-01 1.20779373e-01 -1.10790849e+00 -8.90518963e-01 -2.02348566e+00 -5.30189276e-01 -4.73789006e-01 7.66831160e-01 5.73962569e-01 5.69770038e-01 -7.91110039e-01 6.47070289e-01 -1.20887113e+00 1.12383381e-01 3.91068190e-01 8.62173557e-01 -2.74407387e-01 5.17242737e-02 -8.24397802e-01 3.45344961e-01 8.05401921e-01 3.91177461e-02 1.24587724e-02 -1.00531435e+00 -7.63940036e-01 4.76137817e-01 -4.54787225e-01 -3.68271261e-01 1.06882191e+00 -6.09726787e-01 -1.34606516e+00 2.51731575e-01 -2.45629214e-02 -9.78876889e-01 -2.30396867e-01 1.60360724e-01 -7.09997892e-01 1.60058379e-01 -3.56415510e-01 7.35745132e-01 8.64368081e-01 2.57951021e-01 -9.75327790e-01 -3.50920469e-01 -5.79067290e-01 -8.64134729e-02 -8.77902389e-01 -2.44445130e-01 -1.20780654e-01 -5.82056403e-01 3.59738648e-01 -8.34120691e-01 -2.41440713e-01 3.65542263e-01 1.10056557e-01 -8.17918926e-02 1.56978428e+00 -9.94733199e-02 1.19449759e+00 -2.28098106e+00 -5.81018150e-01 1.18597902e-01 7.90721476e-02 7.81937480e-01 3.52789223e-01 9.99853481e-03 2.13857710e-01 -4.66175467e-01 2.22427651e-01 -1.39912978e-01 -2.45018065e-01 2.73993373e-01 -3.55955154e-01 5.18660724e-01 2.69580126e-01 5.94856262e-01 -1.89340442e-01 -3.42106819e-01 1.62800580e-01 9.12725508e-01 -8.96795154e-01 -1.91985533e-01 3.50427032e-01 -6.28346920e-01 -2.33878076e-01 5.29040575e-01 9.17343616e-01 -1.06188141e-01 2.49460265e-01 -9.47903931e-01 -5.37949204e-01 5.28397322e-01 -9.94962931e-01 1.45665646e+00 -4.78788823e-01 1.04547954e+00 -5.29560894e-02 -7.52942622e-01 1.20481682e+00 1.33733481e-01 4.12557796e-02 -9.13630366e-01 6.57582164e-01 3.93509686e-01 2.75985181e-01 -1.95238933e-01 7.94812381e-01 2.71307111e-01 -7.73121417e-02 1.18664183e-01 5.45752794e-02 3.64840388e-01 2.27391377e-01 -1.77465528e-01 1.06870174e+00 -5.51916122e-01 4.55706447e-01 -6.64163232e-01 7.72587299e-01 3.23005393e-02 6.29079223e-01 2.87838057e-02 -7.20320046e-02 -3.11832100e-01 1.44034326e-01 -7.63759255e-01 -1.30983341e+00 -7.09348083e-01 -3.30187500e-01 8.06315780e-01 -1.77671462e-01 -9.67056036e-01 -8.04484606e-01 2.10367039e-01 -1.02513954e-01 -5.40950745e-02 -2.97450498e-02 -4.56206977e-01 -9.76148605e-01 -4.77659851e-01 7.55105078e-01 7.65056789e-01 1.01710284e+00 -7.82846570e-01 -1.70239377e+00 5.48534870e-01 7.10063517e-01 -1.32608593e+00 -2.67208606e-01 5.73497176e-01 -1.30757713e+00 -6.98235869e-01 1.00069791e-02 -1.39565170e+00 8.72609258e-01 3.25132012e-01 4.83790636e-01 7.99563900e-02 -7.96156287e-01 -5.22611618e-01 -1.78133607e-01 -6.16259634e-01 -1.54229432e-01 9.89691764e-02 2.37342149e-01 -5.74998617e-01 8.42338562e-01 -8.13653171e-01 -9.44484770e-01 -1.85879663e-01 -8.61893952e-01 3.18022549e-01 1.00612533e+00 6.71026945e-01 7.28551745e-01 3.90137255e-01 2.38769308e-01 -4.79407251e-01 3.15320134e-01 -6.35971576e-02 -9.91051674e-01 -2.15693355e-01 -5.82544088e-01 1.16556644e-01 1.26923442e+00 -5.65012217e-01 -7.03121066e-01 4.82661128e-01 -2.79693425e-01 -2.67376691e-01 3.00383091e-01 3.53232801e-01 -1.40426517e-01 -2.99794137e-01 3.91599596e-01 2.68142909e-01 2.24732310e-01 -2.82737911e-01 -9.90545750e-02 8.84422302e-01 8.79798889e-01 7.10064396e-02 3.01559269e-01 3.61474484e-01 3.58375400e-01 -9.98677611e-01 -1.78950176e-01 -4.74673882e-02 -3.76733541e-01 1.04108587e-01 6.55189812e-01 -1.46991003e+00 -1.06920528e+00 2.80336678e-01 -9.80352283e-01 -1.70559198e-01 5.41908965e-02 7.32379556e-01 -9.77675095e-02 1.35971889e-01 -8.37656140e-01 -1.25306323e-01 -1.23354506e+00 -1.43265438e+00 5.96841693e-01 4.71908152e-01 -7.12401569e-02 -4.68149006e-01 -8.36299121e-01 -5.16306579e-01 6.02308095e-01 -1.19652733e-01 8.85686636e-01 -1.86928198e-01 -7.12058008e-01 -4.49924082e-01 -5.65283120e-01 1.97331846e-01 -1.25068901e-02 -1.37107503e-02 -6.08688354e-01 -3.90044332e-01 3.75632554e-01 5.82547439e-03 9.11915779e-01 2.30580598e-01 1.31979871e+00 -6.41326785e-01 -3.88219118e-01 1.11075568e+00 1.57617164e+00 4.77842629e-01 6.22277141e-01 2.72697479e-01 5.94170809e-01 5.62850013e-03 6.01211548e-01 8.03678155e-01 2.70518720e-01 2.24576786e-01 3.94445270e-01 3.39931399e-01 5.01458440e-03 -6.84201270e-02 4.00646240e-01 1.60474753e+00 4.33046192e-01 1.28524065e-01 -4.89017010e-01 9.01682973e-02 -1.36672485e+00 -5.48459053e-01 5.42778485e-02 1.81768227e+00 8.66721988e-01 4.99571562e-01 -3.44431818e-01 8.16540241e-01 2.74900645e-01 -6.38686270e-02 -5.21324754e-01 -7.86768317e-01 1.65428057e-01 5.25247216e-01 1.11949444e+00 3.20512831e-01 -7.90557325e-01 5.41976094e-01 4.47629452e+00 9.02872086e-01 -1.54194546e+00 -2.07536630e-02 4.37918603e-01 -2.83009291e-01 4.79697466e-01 -1.51492313e-01 -1.47001004e+00 5.33769846e-01 1.42508996e+00 -1.02878772e-01 -1.60037413e-01 1.32069349e+00 -1.00590222e-01 -1.07564190e-02 -9.84958053e-01 1.08650374e+00 -2.72266597e-01 -1.84870887e+00 4.90237959e-02 2.92960763e-01 2.87510067e-01 9.73380059e-02 9.74574760e-02 6.38005510e-02 -6.12728715e-01 -6.08396471e-01 7.36410558e-01 -1.33772731e-01 1.11851668e+00 -1.33836484e+00 8.08608174e-01 2.12045640e-01 -1.76808739e+00 -4.08048868e-01 -9.91097450e-01 -3.39198291e-01 -1.38450548e-01 5.60251474e-01 -8.37519705e-01 -3.77753019e-01 8.71388793e-01 5.22867560e-01 -3.47738594e-01 5.78863144e-01 5.28074265e-01 1.87353522e-01 -4.25826341e-01 -3.53462130e-01 2.92972386e-01 5.11373207e-03 -2.09417820e-01 1.47630882e+00 6.53662443e-01 2.56339818e-01 -2.17302367e-01 1.51543692e-01 -3.99337448e-02 -4.09810431e-02 -2.08785355e-01 7.80236498e-02 9.46385741e-01 1.32452178e+00 -8.85604262e-01 -6.30926251e-01 -6.14437163e-01 6.99092805e-01 3.18237357e-02 -5.69240808e-01 -6.82422042e-01 -1.25056934e+00 1.01609993e+00 1.52159050e-01 4.67241347e-01 -5.01500309e-01 -5.51049113e-01 -5.47718048e-01 7.50554949e-02 -4.66472149e-01 -8.10390711e-02 -6.64317459e-02 -2.71293342e-01 8.46807837e-01 -2.98522711e-01 -1.30145693e+00 -1.71038836e-01 -8.97201538e-01 -5.27610123e-01 5.71721315e-01 -1.47853327e+00 -7.13287294e-01 -6.75155878e-01 7.22050667e-01 6.39496326e-01 -6.18677378e-01 8.37331176e-01 6.87780976e-01 -5.17199874e-01 1.02914226e+00 -1.97363496e-01 -1.16441786e-01 1.60607308e-01 -3.41008067e-01 8.76703918e-01 7.94909596e-01 -4.89984393e-01 8.64474356e-01 4.56971645e-01 -6.82560980e-01 -2.02200603e+00 -1.55248225e+00 8.84618044e-01 1.07479346e+00 4.74464864e-01 -5.32741308e-01 -8.66086960e-01 5.05957663e-01 5.92367798e-02 5.38326085e-01 8.13937128e-01 -8.36927652e-01 -2.73846447e-01 -5.19776940e-01 -1.33176982e+00 7.97081947e-01 7.17674255e-01 -2.90080369e-01 2.06805304e-01 2.96826214e-01 8.73679221e-01 -7.05305815e-01 -1.05979240e+00 3.21811080e-01 6.13398910e-01 -6.16983771e-01 8.83704901e-01 2.34533653e-01 3.35169673e-01 -4.05643582e-01 -4.77503091e-01 -4.35136914e-01 -3.57788831e-01 -3.76949131e-01 -4.62382883e-01 5.63005805e-01 9.39476267e-02 -5.15023530e-01 1.07704902e+00 2.91374832e-01 -4.62591082e-01 -9.90652323e-01 -1.14719725e+00 -4.75710779e-01 -5.46311736e-01 -3.78691614e-01 7.86647141e-01 2.85819650e-01 2.89437234e-01 2.45654866e-01 3.64857838e-02 1.98184222e-01 4.80304182e-01 -3.45443189e-01 3.81235391e-01 -1.04586911e+00 -2.89358832e-02 -3.55830878e-01 -1.03087735e+00 -1.00083566e+00 -7.32059684e-03 -9.62474167e-01 -4.39300001e-01 -8.07205260e-01 -3.72182816e-01 -5.45317292e-01 3.24117877e-02 5.04880428e-01 5.39154291e-01 5.95537424e-01 1.48913443e-01 -1.24027610e-01 -1.43694371e-01 3.73358309e-01 8.64029884e-01 3.47932130e-02 -2.37313360e-01 -3.65382165e-01 -2.43490875e-01 7.71679401e-01 1.11091721e+00 -3.53183419e-01 -7.37770677e-01 -6.06753290e-01 -1.17279775e-02 -9.65146422e-02 1.00609019e-01 -1.73937988e+00 9.23661590e-01 3.05861086e-01 7.40754366e-01 -1.14614284e+00 5.77729702e-01 -9.35287535e-01 1.38636678e-01 1.23442864e+00 -4.40417230e-02 6.85366273e-01 6.06332064e-01 2.50259995e-01 -1.52259469e-01 -3.25149268e-01 8.84594262e-01 2.52873659e-01 -7.38444567e-01 3.29797745e-01 -7.33153820e-01 -7.02480078e-01 9.83232200e-01 -3.75438660e-01 -2.50563234e-01 4.81539160e-01 -1.80889696e-01 -3.36930931e-01 -4.34848815e-02 1.32298738e-01 1.09415913e+00 -1.45363748e+00 -2.44896948e-01 1.05809569e+00 -3.46857429e-01 1.37252107e-01 4.48690593e-01 2.98457563e-01 -1.27795351e+00 9.78966236e-01 -7.66242266e-01 -5.69925547e-01 -1.73170614e+00 1.96625412e-01 -1.74687982e-01 8.53586197e-02 -1.06558728e+00 8.66531610e-01 -4.35548604e-01 5.30631304e-01 3.16944391e-01 -8.02772582e-01 4.29050438e-02 -1.66222483e-01 1.17951393e+00 4.52439845e-01 5.40436506e-01 -3.67987454e-01 -2.69190967e-01 5.51114440e-01 -5.23091197e-01 3.31189990e-01 1.21479750e+00 1.45285338e-01 -4.07456577e-01 -3.59695911e-01 1.88351250e+00 -5.00527322e-01 -1.13516629e+00 -1.07358083e-01 5.05120829e-02 -2.25937098e-01 4.78708386e-01 2.35771790e-01 -1.38248086e+00 6.27115846e-01 8.93775225e-01 -3.59143615e-01 1.53662038e+00 -6.54165983e-01 1.29203463e+00 7.54327536e-01 2.12424189e-01 -1.25396752e+00 -2.07454134e-02 6.27089143e-01 3.76314729e-01 -6.21386647e-01 7.27065742e-01 -4.97084737e-01 2.05149576e-02 1.61296451e+00 7.89973736e-01 -7.29195356e-01 1.20001197e+00 1.24912858e+00 -4.12553906e-01 1.69786096e-01 -1.08386290e+00 5.99442840e-01 -9.91844106e-03 4.84294504e-01 3.06791663e-01 1.70122132e-01 -4.65603441e-01 5.31115294e-01 -9.70170379e-01 -5.52092195e-02 4.89507079e-01 1.35106921e+00 -7.22328722e-01 -9.71074820e-01 2.73203477e-02 6.63062513e-01 -5.33391476e-01 -1.30564883e-01 4.42241520e-01 5.60810208e-01 3.79261851e-01 5.45448303e-01 8.86305451e-01 -9.24789786e-01 -5.05676493e-02 -4.33238924e-01 4.57405746e-01 -2.14436665e-01 -7.04340875e-01 1.29456535e-01 -2.91843414e-01 -5.68339229e-01 1.02505982e-01 -1.05427310e-01 -1.70907950e+00 -6.16389811e-01 -9.52906609e-02 -3.24756086e-01 1.43127286e+00 5.45062780e-01 5.88365376e-01 8.73386145e-01 4.05516148e-01 -9.78058577e-01 -2.46858791e-01 -7.39106297e-01 -2.03349560e-01 -5.57492137e-01 4.02640104e-01 -2.26730645e-01 1.51673704e-03 1.32183760e-01]
[8.370702743530273, 2.782902956008911]
3e882648-bfa4-44c4-8982-2e9cb5e9dc6c
few-shot-image-classification-just-use-a
2101.00562
null
https://arxiv.org/abs/2101.00562v3
https://arxiv.org/pdf/2101.00562v3.pdf
Few-shot Image Classification: Just Use a Library of Pre-trained Feature Extractors and a Simple Classifier
Recent papers have suggested that transfer learning can outperform sophisticated meta-learning methods for few-shot image classification. We take this hypothesis to its logical conclusion, and suggest the use of an ensemble of high-quality, pre-trained feature extractors for few-shot image classification. We show experimentally that a library of pre-trained feature extractors combined with a simple feed-forward network learned with an L2-regularizer can be an excellent option for solving cross-domain few-shot image classification. Our experimental results suggest that this simpler sample-efficient approach far outperforms several well-established meta-learning algorithms on a variety of few-shot tasks.
['Swarat Chaudhuri', 'Chris Jermaine', 'Mingchao Jiang', 'Arkabandhu Chowdhury']
2021-01-03
null
http://openaccess.thecvf.com//content/ICCV2021/html/Chowdhury_Few-Shot_Image_Classification_Just_Use_a_Library_of_Pre-Trained_Feature_ICCV_2021_paper.html
http://openaccess.thecvf.com//content/ICCV2021/papers/Chowdhury_Few-Shot_Image_Classification_Just_Use_a_Library_of_Pre-Trained_Feature_ICCV_2021_paper.pdf
iccv-2021-1
['cross-domain-few-shot']
['computer-vision']
[ 2.78810233e-01 -3.33679944e-01 -6.94359481e-01 -4.90002722e-01 -1.08524489e+00 1.39812008e-01 7.51023293e-01 -2.32902803e-02 -6.87708676e-01 6.58704460e-01 2.17377782e-01 2.07927004e-01 -2.10247666e-01 -6.15048647e-01 -5.27416468e-01 -6.04744673e-01 -3.94573994e-02 2.53776670e-01 1.23969674e-01 -3.17469150e-01 5.12225866e-01 1.26003057e-01 -2.00845671e+00 5.40730894e-01 4.90037501e-01 1.05527151e+00 2.20719099e-01 6.58391178e-01 -1.89657450e-01 1.12503958e+00 -5.92481911e-01 -3.60547930e-01 1.75038546e-01 -7.07396626e-01 -8.49688411e-01 2.66890466e-01 4.94294614e-01 -3.84274036e-01 -3.89203399e-01 1.06219780e+00 5.12354851e-01 7.43987560e-01 1.01345789e+00 -1.19402015e+00 -7.90986359e-01 3.93740863e-01 -3.82635355e-01 6.83601260e-01 1.94365680e-01 4.34933543e-01 9.40427661e-01 -9.60261226e-01 7.85344660e-01 8.62281203e-01 8.47235382e-01 6.84956729e-01 -1.17045915e+00 -4.30585384e-01 -2.69472599e-01 5.54546058e-01 -1.11957860e+00 -5.99926472e-01 6.26580536e-01 -2.87480563e-01 1.57644856e+00 -2.07193092e-01 5.34936965e-01 1.20479572e+00 4.36503261e-01 8.73494387e-01 1.01708937e+00 -7.74664462e-01 5.33313513e-01 4.06810313e-01 3.49801183e-01 1.02679956e+00 1.55626550e-01 9.44288913e-03 -7.15923429e-01 -3.58298838e-01 3.54985446e-01 5.05491614e-01 5.50175309e-02 -5.48174918e-01 -8.00307512e-01 1.33663356e+00 4.43705916e-01 6.00317061e-01 -3.21352243e-01 2.42589176e-01 8.90160859e-01 6.73530579e-01 9.03417706e-01 7.66626656e-01 -3.88219595e-01 -2.04192951e-01 -1.07101393e+00 -8.62068236e-02 7.08668888e-01 8.87041450e-01 1.01775622e+00 2.14937761e-01 -1.98576927e-01 1.03990662e+00 -1.90556958e-01 7.76191205e-02 1.14037871e+00 -1.02793765e+00 1.28908128e-01 1.00489892e-01 1.61427036e-02 -5.23359239e-01 -1.24571584e-01 -2.18537718e-01 -6.32413328e-01 3.17020357e-01 8.53153467e-02 -1.39852673e-01 -1.10274005e+00 1.41226888e+00 -1.10539019e-01 5.07189333e-01 1.47275686e-01 5.92117786e-01 7.47249722e-01 6.63433135e-01 2.83686042e-01 -4.52294707e-01 1.20520139e+00 -1.20896268e+00 -5.87758243e-01 -2.97685325e-01 7.62586057e-01 -3.81461412e-01 1.17142332e+00 1.59417734e-01 -9.44430828e-01 -6.38413966e-01 -1.26354444e+00 1.32468184e-02 -7.70136833e-01 -1.32173836e-01 9.67915833e-01 5.92986286e-01 -6.41248643e-01 1.15859246e+00 -5.60737014e-01 -7.11040258e-01 8.89783442e-01 -1.55692426e-02 -3.57829481e-01 -3.52731884e-01 -9.16038871e-01 1.20569503e+00 4.83743459e-01 -7.16583252e-01 -9.84588265e-01 -8.71444464e-01 -1.14129448e+00 2.10111126e-01 3.76635790e-01 -8.78405571e-01 1.36111677e+00 -1.11714375e+00 -1.58197808e+00 1.04711354e+00 -9.39715207e-02 -5.26531696e-01 1.81704983e-02 -8.96275565e-02 -3.64946693e-01 4.75206941e-01 1.63718030e-01 5.28820634e-01 1.31748807e+00 -7.76959002e-01 -5.44623077e-01 -2.01186895e-01 -1.69498578e-01 1.19668946e-01 -7.14200199e-01 1.14148468e-01 7.59223551e-02 -4.48024988e-01 -4.95888680e-01 -6.00915790e-01 -3.40432435e-01 1.38771338e-02 1.77874595e-01 -3.77447367e-01 5.74183404e-01 4.79439050e-02 9.08734500e-01 -2.00567079e+00 6.39696196e-02 -2.41484404e-01 -3.05556357e-02 4.98529732e-01 -5.25986791e-01 5.05105197e-01 -1.69964969e-01 -2.08762795e-01 -1.40159607e-01 -5.49314678e-01 -1.08155258e-01 1.94543406e-01 -2.57050991e-01 5.35194516e-01 2.82531500e-01 1.11649656e+00 -1.22414470e+00 -5.88731945e-01 5.59646070e-01 2.50391632e-01 -2.44130850e-01 2.88428903e-01 -1.50296211e-01 -2.66801596e-01 -2.21245483e-01 4.38582093e-01 2.28233770e-01 -5.61616361e-01 -2.67123759e-01 3.68865952e-02 1.94691256e-01 -1.44288121e-02 -7.41617262e-01 2.11327195e+00 -5.56390703e-01 7.39960074e-01 -6.07497454e-01 -1.16749382e+00 6.51700616e-01 2.30281860e-01 5.60257316e-01 -6.55451000e-01 4.27552968e-01 3.65531519e-02 -3.92786503e-01 -7.91596055e-01 1.70814350e-01 -7.09610939e-01 4.54571936e-03 6.82095945e-01 1.19079745e+00 -2.40274221e-01 2.62395054e-01 1.88005164e-01 1.33233988e+00 -8.69319215e-02 8.33596528e-01 -7.96700865e-02 1.45688832e-01 2.25298822e-01 3.09621841e-01 1.08436382e+00 -4.98167545e-01 6.66223824e-01 -3.06803752e-02 -5.41514218e-01 -1.28689575e+00 -7.93336928e-01 -3.93188633e-02 1.62080526e+00 -1.06498599e-01 -5.96427977e-01 -6.23400867e-01 -6.68072939e-01 -8.72428864e-02 9.13492262e-01 -8.80153418e-01 -4.88493592e-01 -7.61898905e-02 -7.59399474e-01 2.17974558e-01 6.74011350e-01 3.42827320e-01 -1.24118543e+00 -9.86210346e-01 1.98856041e-01 1.98487014e-01 -8.07413161e-01 -2.31750548e-01 7.41666853e-01 -1.02999341e+00 -9.31001365e-01 -1.11091626e+00 -7.53746510e-01 3.92435312e-01 8.76679182e-01 9.62855160e-01 7.82860965e-02 -7.18658566e-01 6.81049168e-01 -6.95178509e-01 -4.13827032e-01 -1.79937303e-01 1.50892466e-01 -3.23963948e-02 -2.32199058e-01 8.85669291e-01 -4.99399930e-01 -2.66649276e-01 -1.04891159e-01 -8.14726770e-01 -4.02463436e-01 4.44209844e-01 1.36807680e+00 3.27618331e-01 -9.56045091e-02 7.39548326e-01 -1.13981438e+00 5.76502383e-01 -7.80614078e-01 -7.37549886e-02 4.26632851e-01 -5.87154508e-01 1.98912010e-01 5.78233540e-01 -6.59382582e-01 -1.22145402e+00 -8.76753591e-03 2.99043581e-02 -9.22222733e-01 -2.78663754e-01 4.16709453e-01 2.81986296e-01 -2.69214213e-01 1.21787548e+00 4.59515095e-01 7.40222782e-02 -3.62631828e-01 6.45447493e-01 5.15938461e-01 1.53863698e-01 -8.59592035e-02 6.24837220e-01 5.26773155e-01 -1.97327211e-01 -1.15068722e+00 -1.40074909e+00 -9.84609425e-01 -6.36835515e-01 -8.61403495e-02 8.09434712e-01 -9.50430691e-01 5.17895222e-02 3.56406868e-01 -8.96703064e-01 -4.64300066e-01 -6.12778187e-01 5.00398695e-01 -1.18491006e+00 2.49227405e-01 -6.95084214e-01 -6.10357761e-01 -3.37977886e-01 -6.73893690e-01 8.96011710e-01 2.16340870e-01 -2.17769504e-01 -1.08876014e+00 4.05878067e-01 3.14353146e-02 5.38250089e-01 -2.63379037e-01 7.46124029e-01 -7.86084950e-01 -8.79754722e-02 -3.33294868e-01 -1.02806665e-01 3.57557148e-01 1.80486873e-01 -3.12926441e-01 -1.23663425e+00 -3.14910769e-01 2.60580868e-01 -1.21565354e+00 1.44270790e+00 5.40545762e-01 1.13291395e+00 -1.00323306e-02 -3.23613912e-01 6.58432186e-01 1.84562147e+00 -1.57425046e-01 5.84558427e-01 2.11478978e-01 3.14199388e-01 1.69953585e-01 5.74729443e-01 5.55274010e-01 -7.20035955e-02 3.72551382e-01 -8.44842568e-02 3.53458524e-01 -2.75945306e-01 1.15914780e-04 2.01815560e-01 5.70569456e-01 -2.12653339e-01 -1.02011468e-02 -4.58620578e-01 4.65891540e-01 -1.93244612e+00 -1.69940376e+00 6.13210857e-01 1.85092235e+00 6.58338249e-01 1.35285839e-01 3.13141912e-01 -1.71966746e-01 4.97533917e-01 5.30991077e-01 -6.23873413e-01 -4.34895843e-01 8.62332806e-02 5.43494821e-01 3.91427606e-01 1.99805107e-02 -1.33072734e+00 9.51394200e-01 7.83710098e+00 1.08413124e+00 -9.44651663e-01 5.68332016e-01 3.31941098e-01 -3.36194128e-01 7.29642296e-03 -2.59699881e-01 -6.37961984e-01 2.45708346e-01 1.18921959e+00 -4.61817622e-01 3.27946186e-01 1.36482489e+00 -4.00873274e-01 -1.47490114e-01 -1.07471085e+00 1.19949055e+00 6.34032965e-01 -1.73813856e+00 -1.20194390e-01 -2.26456821e-01 1.06666517e+00 4.80673194e-01 1.87985137e-01 7.44774163e-01 1.79268122e-01 -7.99780488e-01 1.73167601e-01 4.53935802e-01 6.98660493e-01 -6.86586559e-01 6.55577779e-01 5.17372608e-01 -9.59708571e-01 -4.74969000e-01 -1.07716608e+00 -1.76305249e-01 -2.51982361e-02 3.05987507e-01 -4.67943758e-01 2.17407227e-01 4.13145900e-01 9.78170097e-01 -4.45181757e-01 1.29346621e+00 1.49488561e-02 4.44290966e-01 1.00014023e-01 -3.17881286e-01 5.90775192e-01 3.34829360e-01 3.37603718e-01 1.36319828e+00 1.76244095e-01 3.85021269e-01 1.83758754e-02 6.70038581e-01 1.62778962e-02 1.73596844e-01 -1.20285940e+00 -1.76156372e-01 1.23479061e-01 1.30479681e+00 -8.32997739e-01 -8.29888046e-01 -7.09710181e-01 1.16182935e+00 6.26488686e-01 1.66897103e-01 -5.69597363e-01 -7.43720293e-01 5.88397741e-01 -2.78612942e-01 8.42476726e-01 8.02927688e-02 -3.20207737e-02 -1.67572427e+00 -5.89153707e-01 -6.71587586e-01 7.30087757e-01 -8.24674785e-01 -1.73390961e+00 4.96503890e-01 1.20416686e-01 -1.34160423e+00 -5.39138317e-01 -7.71364331e-01 -1.03860939e+00 4.35011327e-01 -1.63696861e+00 -9.95302260e-01 -1.19370297e-01 6.81660593e-01 1.14275372e+00 -6.45464897e-01 1.23345673e+00 -3.53986360e-02 -4.36826199e-01 5.62186360e-01 2.36184537e-01 -6.63769692e-02 6.34243250e-01 -1.05292737e+00 6.81130141e-02 4.22975093e-01 6.12996876e-01 5.72555482e-01 7.75560558e-01 -4.22381669e-01 -1.31277299e+00 -9.50978279e-01 6.41204476e-01 -3.58894587e-01 5.56554496e-01 3.56905558e-03 -1.02489614e+00 7.95591474e-01 5.37652194e-01 3.97777706e-01 1.10149682e+00 2.74377763e-01 -6.08939290e-01 1.43070564e-01 -1.32332122e+00 2.41026461e-01 9.95629311e-01 -7.93056607e-01 -1.16668141e+00 4.36256558e-01 4.86714333e-01 1.72926009e-01 -6.61996722e-01 -6.52160197e-02 4.16928858e-01 -8.95168781e-01 1.00122356e+00 -1.23497057e+00 7.14828491e-01 5.42102814e-01 -3.19267422e-01 -1.78393459e+00 -5.47256470e-01 -3.56333286e-01 -3.17577690e-01 6.89443886e-01 5.81556372e-02 -3.02669168e-01 8.42914701e-01 2.83375084e-01 4.49782796e-02 -5.33199072e-01 -6.96322739e-01 -1.17415154e+00 8.33161399e-02 -3.13111782e-01 5.47501296e-02 8.88797164e-01 6.73511505e-01 5.69938540e-01 -6.53845608e-01 -6.17236376e-01 9.58061159e-01 1.28013432e-01 5.34869552e-01 -1.23964643e+00 -3.77055556e-01 -4.26757693e-01 -6.67526007e-01 -4.73127544e-01 6.29977465e-01 -9.93546665e-01 9.20019671e-02 -1.23946381e+00 6.19275987e-01 1.67637125e-01 -7.83663869e-01 5.00735760e-01 -7.98355937e-02 4.62074578e-01 1.20684482e-01 2.53109504e-02 -1.07441759e+00 6.70124054e-01 8.48275185e-01 -3.07897180e-01 7.60003403e-02 2.57718936e-03 -4.48601753e-01 8.67737889e-01 6.58651173e-01 -8.52314472e-01 -5.58640361e-01 -2.33786069e-02 -2.18458936e-01 -6.75561354e-02 1.60621405e-01 -1.12640727e+00 4.08238471e-01 -3.21571350e-01 5.52076995e-01 -1.12849139e-01 6.66758239e-01 -4.17790383e-01 -5.36934018e-01 5.24474859e-01 -5.53505599e-01 -5.24479210e-01 3.81062441e-02 6.38730645e-01 -2.37822473e-01 -9.19258296e-01 1.09919226e+00 -8.93843293e-01 -1.33443868e+00 2.84740269e-01 -5.00452995e-01 2.62061715e-01 1.21192217e+00 -3.45446110e-01 -2.61083364e-01 -3.25150520e-01 -6.79339767e-01 -2.48405114e-01 2.98582166e-01 3.15959364e-01 8.63935113e-01 -1.38752401e+00 -5.00394881e-01 1.74946934e-01 7.07782924e-01 -1.10582983e+00 3.47359329e-01 5.36330581e-01 8.98166150e-02 3.68628353e-01 -8.02259088e-01 -3.21396440e-01 -1.18934977e+00 9.20701206e-01 2.22611904e-01 5.14548272e-02 -9.34718966e-01 1.05115581e+00 -5.33528864e-01 -9.81051847e-02 1.95787340e-01 9.63842496e-02 6.56246915e-02 4.97802377e-01 1.00774944e+00 5.48189819e-01 -1.56991437e-01 -2.39058465e-01 -1.60221577e-01 4.37519878e-01 -1.55830845e-01 -5.13881724e-03 1.91712439e+00 8.04533586e-02 4.08138901e-01 9.32093978e-01 1.69522655e+00 -9.08085763e-01 -1.05623853e+00 -5.50054014e-01 -2.29103658e-02 -6.67839706e-01 2.74991900e-01 -4.21170861e-01 -7.26180077e-01 1.04234254e+00 4.33142036e-01 1.01332836e-01 9.23624218e-01 5.19237556e-02 6.73982620e-01 1.04992867e+00 6.94853485e-01 -1.35131073e+00 7.64273107e-01 4.01634604e-01 4.30656552e-01 -1.79515672e+00 1.57648683e-01 1.53331041e-01 -8.35987568e-01 1.26706362e+00 5.24834216e-01 -6.26777768e-01 8.73307109e-01 2.12624576e-03 -2.01429769e-01 -3.88989747e-01 -1.12570965e+00 -5.45029104e-01 2.84711719e-01 6.41039550e-01 2.19689935e-01 -2.81600773e-01 1.41628794e-02 3.17840099e-01 3.96454424e-01 5.84369063e-01 5.09569943e-01 1.15205634e+00 -1.07993293e+00 -7.49175668e-01 1.68396667e-01 8.31543982e-01 -2.93016255e-01 -3.97300571e-02 -5.95531836e-02 5.73387623e-01 -5.69996871e-02 6.87548339e-01 1.24468677e-01 -2.12012440e-01 -6.44277632e-02 6.09326482e-01 1.04817641e+00 -1.21937072e+00 -5.31124234e-01 -2.15523958e-01 -8.88489857e-02 -7.53847063e-01 -6.24223173e-01 -5.08950830e-01 -6.71183169e-01 -2.36561447e-01 -5.53866744e-01 3.03836334e-02 3.49970847e-01 1.36503673e+00 2.43669569e-01 2.74684280e-01 7.02424109e-01 -1.16254818e+00 -1.20963562e+00 -1.01120389e+00 -8.54340196e-01 5.61668873e-01 3.42929333e-01 -8.78689766e-01 -4.51484203e-01 -1.17204897e-02]
[9.958173751831055, 2.9812281131744385]
412722aa-3d7b-4fe1-b1d5-04888cf717d9
fully-convolutional-asr-for-less-resourced
null
null
https://aclanthology.org/2020.sltu-1.17
https://aclanthology.org/2020.sltu-1.17.pdf
Fully Convolutional ASR for Less-Resourced Endangered Languages
The application of deep learning to automatic speech recognition (ASR) has yielded dramatic accuracy increases for languages with abundant training data, but languages with limited training resources have yet to see accuracy improvements on this scale. In this paper, we compare a fully convolutional approach for acoustic modelling in ASR with a variety of established acoustic modeling approaches. We evaluate our method on Seneca, a low-resource endangered language spoken in North America. Our method yields word error rates up to 40{\%} lower than those reported using both standard GMM-HMM approaches and established deep neural methods, with a substantial reduction in training time. These results show particular promise for languages like Seneca that are both endangered and lack extensive documentation.
["Emily Prud{'}hommeaux", 'Robert Jimerson', 'Raymond Ptucha', 'Bao Thai']
2020-05-01
null
null
null
lrec-2020-5
['acoustic-modelling']
['speech']
[-5.71313798e-02 -1.50243118e-01 2.22182035e-01 -4.39014435e-01 -1.37599301e+00 -4.84130234e-01 5.95704973e-01 -1.36926815e-01 -8.16653907e-01 3.08519006e-01 3.53043228e-01 -7.63288140e-01 3.65479052e-01 -3.68598938e-01 -4.36150014e-01 -4.23496753e-01 -1.92365274e-01 5.93766689e-01 -1.88801005e-01 -4.22670960e-01 -2.16884568e-01 6.39237463e-01 -1.12793338e+00 3.19998264e-02 4.38980520e-01 5.29989421e-01 2.53643543e-01 1.10113275e+00 -8.62652287e-02 6.71782434e-01 -1.10648751e+00 -2.43248194e-01 -1.37219772e-01 9.71283242e-02 -7.33612895e-01 -1.82605714e-01 5.76883137e-01 -5.44527829e-01 -6.31357074e-01 6.76474035e-01 9.19111848e-01 1.73632577e-01 2.76617825e-01 -3.55257601e-01 -6.75602555e-01 8.34663570e-01 -2.43016168e-01 5.04471421e-01 2.17382535e-01 3.17840606e-01 8.14225197e-01 -7.61671603e-01 -4.98463064e-02 1.42702878e+00 9.60928738e-01 8.50177944e-01 -1.24662101e+00 -7.50805259e-01 -9.43942592e-02 -1.67176157e-01 -1.60213017e+00 -1.18422663e+00 4.01320696e-01 -3.63880724e-01 1.80814934e+00 1.23301484e-01 4.43819553e-01 1.07958031e+00 -4.69991833e-01 7.03138292e-01 7.39273608e-01 -4.43201721e-01 2.89196640e-01 -9.37182009e-02 -1.08577892e-01 4.28335965e-01 -2.26085305e-01 2.64566928e-01 -8.18766296e-01 -1.73118263e-01 4.97384548e-01 -5.74512720e-01 5.88124506e-02 6.05111420e-01 -9.98424709e-01 8.52425098e-01 -1.31270602e-01 5.78367174e-01 -2.87482947e-01 5.63705564e-01 5.42352140e-01 4.78019416e-02 8.63670230e-01 1.34370312e-01 -7.01315820e-01 -6.69174016e-01 -1.25627613e+00 -2.62488797e-02 8.78695428e-01 6.86236084e-01 2.34972522e-01 1.25218570e+00 4.79562193e-01 1.75004005e+00 3.96665215e-01 7.88106799e-01 5.30764937e-01 -7.58255780e-01 3.29486877e-01 -2.34132692e-01 -1.97401688e-01 -5.06161928e-01 -3.16337138e-01 -4.62272227e-01 -6.11041307e-01 1.47746913e-02 3.04019630e-01 -1.90001950e-01 -1.21934438e+00 1.81815529e+00 -1.19755641e-01 1.69579163e-01 3.76208216e-01 4.74993676e-01 8.24738383e-01 1.03592527e+00 2.29331553e-01 1.82963759e-01 1.02605450e+00 -5.02621293e-01 -5.29678643e-01 -4.28459138e-01 7.20736682e-01 -9.44452047e-01 1.12567639e+00 7.02535629e-01 -9.42863882e-01 -3.24237168e-01 -9.32254195e-01 4.16633859e-02 -4.01811004e-01 3.20017606e-01 7.17696428e-01 1.38898611e+00 -1.48066247e+00 2.87451863e-01 -1.20635247e+00 -5.58015525e-01 2.33721137e-01 4.64370370e-01 -5.13990879e-01 2.05998033e-01 -1.09544957e+00 1.05666471e+00 2.42921621e-01 2.65564620e-01 -1.40379596e+00 -7.29249239e-01 -1.15028191e+00 1.88917033e-02 9.42304209e-02 1.68068737e-01 1.69186103e+00 -3.97363842e-01 -1.75765455e+00 8.42001975e-01 -1.57331660e-01 -8.11669946e-01 2.98647187e-03 -4.85289901e-01 -7.42372692e-01 -3.06209326e-01 -3.41074914e-01 6.17808282e-01 1.25303656e-01 -9.44032013e-01 -5.31669199e-01 -2.70393848e-01 -1.86521009e-01 5.63178817e-03 -5.15700758e-01 7.05642521e-01 -1.42101616e-01 -7.88042665e-01 -2.97623962e-01 -6.57242239e-01 -2.66861469e-01 -4.44518954e-01 -1.19444206e-01 -1.96604446e-01 5.45336664e-01 -1.53531706e+00 1.16439354e+00 -1.90738332e+00 -2.63912708e-01 -3.02033126e-02 -3.19338679e-01 8.53501379e-01 -1.62341118e-01 4.71453786e-01 2.14203313e-01 3.98230255e-01 -5.42180181e-01 -4.38100755e-01 1.48052722e-02 5.17899513e-01 -3.28152984e-01 4.36812848e-01 3.17797124e-01 2.93004602e-01 -7.46119022e-01 3.84649411e-02 4.95170534e-01 9.10947740e-01 -4.69070137e-01 2.75313765e-01 4.90764380e-02 1.30292639e-01 2.37511352e-01 8.33015144e-01 4.75137770e-01 5.05110025e-01 1.92525238e-01 3.36323649e-01 -3.29068631e-01 8.67614746e-01 -7.95637727e-01 1.63622320e+00 -9.71845031e-01 9.08593893e-01 4.53808039e-01 -1.06825316e+00 1.13874638e+00 4.57735628e-01 6.83724806e-02 -2.88909137e-01 -1.70090243e-01 5.77335358e-01 2.80032903e-01 -1.84035435e-01 5.35719156e-01 -2.65073121e-01 1.64491758e-01 -4.93237711e-02 5.23347855e-01 -4.73074436e-01 -2.38062650e-01 -9.58471224e-02 1.09656513e+00 -7.95781910e-02 6.87170625e-02 -4.73392189e-01 2.84984678e-01 -1.23038642e-01 4.99829054e-01 7.32947230e-01 -1.90102458e-01 5.07495880e-01 -6.52593821e-02 -3.47092420e-01 -1.25765347e+00 -1.07495105e+00 -3.44473362e-01 1.14350593e+00 -1.05345261e+00 -3.59901249e-01 -8.76351118e-01 -6.52209343e-03 -3.18229049e-01 9.90953207e-01 -1.15256697e-01 1.23534419e-01 -8.36052537e-01 -1.01737761e+00 1.56192350e+00 6.46689653e-01 2.55985171e-01 -1.08389783e+00 -4.95203324e-02 4.49423879e-01 4.90818843e-02 -1.33326960e+00 -1.67248935e-01 4.63436604e-01 -4.62162226e-01 -2.83681452e-01 -1.01082921e+00 -8.87405276e-01 -1.10706486e-01 -2.69951940e-01 1.05391920e+00 -1.06710374e-01 -3.89791846e-01 4.51160938e-01 -1.41815990e-01 -6.03015304e-01 -1.08046150e+00 1.24550506e-01 5.54673493e-01 -4.53768432e-01 6.92391455e-01 -3.59924227e-01 -1.60596117e-01 -1.35248899e-01 -7.87990153e-01 -4.16208059e-01 2.85068363e-01 8.91751766e-01 9.59842429e-02 -3.11958969e-01 8.79323304e-01 -3.53772730e-01 2.54566103e-01 -4.19887066e-01 -6.95423186e-01 -6.28571288e-05 -2.61702001e-01 -2.73717582e-01 4.09650028e-01 -2.39663303e-01 -9.59949911e-01 1.27273545e-01 -1.22775686e+00 -9.68748182e-02 -6.94005013e-01 6.09934568e-01 -1.05871834e-01 5.11595048e-02 4.31608438e-01 4.94657040e-01 -1.25873640e-01 -8.58497858e-01 2.63521761e-01 1.28123057e+00 7.91117489e-01 -6.81522548e-01 4.70848173e-01 1.01578921e-01 -6.03753507e-01 -1.95689642e+00 -1.94861844e-01 -4.43036824e-01 -4.86392468e-01 -2.93825846e-03 8.11161637e-01 -1.06533217e+00 -4.53850299e-01 9.05019581e-01 -1.25355613e+00 -6.34099483e-01 9.01090801e-02 7.86398292e-01 -4.84576762e-01 3.85181487e-01 -9.78307068e-01 -1.32831240e+00 -4.84405726e-01 -1.05988157e+00 1.21762550e+00 -1.62748575e-01 -1.37456715e-01 -1.07157981e+00 3.75185847e-01 2.84878254e-01 7.50310302e-01 -3.96955609e-01 6.97303891e-01 -1.17366982e+00 5.27039170e-02 -1.34080261e-01 2.13936001e-01 1.02417183e+00 2.05130264e-01 1.42255366e-01 -1.58530402e+00 -3.77290100e-01 -1.98116317e-01 -4.48805332e-01 7.17230856e-01 4.16849196e-01 1.07136905e+00 -2.19500512e-01 2.00011954e-01 4.38219547e-01 8.35070431e-01 6.18743718e-01 5.27837455e-01 1.94103166e-01 4.07316238e-01 6.13234520e-01 -1.04405425e-01 1.94195449e-01 1.46734476e-01 7.65119314e-01 -2.05482319e-01 -9.33042467e-02 -3.28432441e-01 -1.89452097e-01 7.29598045e-01 1.50108898e+00 1.33806065e-01 -5.16020298e-01 -1.55676830e+00 1.14706767e+00 -1.13015997e+00 -8.06302547e-01 2.08240405e-01 2.25713468e+00 9.11402345e-01 -1.50324121e-01 1.07101753e-01 3.23028788e-02 5.98061919e-01 1.67474180e-01 -1.33656740e-01 -8.43936205e-01 -4.23649311e-01 6.26381099e-01 2.81611532e-01 7.47710705e-01 -1.05028760e+00 1.43966413e+00 8.09226227e+00 1.16977656e+00 -1.12497211e+00 2.61602849e-01 8.40397894e-01 -1.81331843e-01 -3.53986621e-02 -2.32054919e-01 -9.16393995e-01 1.32340536e-01 1.79026449e+00 2.12709755e-01 5.46964288e-01 7.03292131e-01 3.16525489e-01 1.90114826e-01 -8.17250013e-01 9.10932720e-01 8.82767141e-02 -1.12118244e+00 -2.15427831e-01 9.31775495e-02 2.77507931e-01 5.87455153e-01 1.87675655e-01 2.89768428e-01 8.82742107e-01 -1.36949861e+00 5.39771318e-01 -4.66519743e-02 8.80265951e-01 -1.02095318e+00 7.17893422e-01 2.32888758e-01 -1.03409183e+00 2.24868059e-01 -6.22742832e-01 1.04299471e-01 3.78567278e-01 4.37427014e-01 -1.03219724e+00 1.28043503e-01 8.43719184e-01 2.41319269e-01 -2.35631421e-01 7.66489804e-01 5.35916910e-02 1.67760777e+00 -7.47850716e-01 -8.66962895e-02 4.29872513e-01 9.88973603e-02 6.83562279e-01 1.90262687e+00 5.16817868e-01 -4.61476967e-02 -4.79018167e-02 5.33433199e-01 6.76723272e-02 1.88204810e-01 -7.85372138e-01 -4.22419846e-01 5.00984550e-01 8.06446373e-01 -2.08410919e-01 -2.67307788e-01 -5.36825955e-01 7.06588149e-01 2.11168110e-01 1.87154666e-01 -3.75704825e-01 -2.49677703e-01 1.01467478e+00 -2.41170958e-01 2.78673843e-02 -7.84126818e-01 -5.71043156e-02 -9.03991997e-01 -4.40771580e-01 -1.09032249e+00 1.00712225e-01 -5.61251640e-01 -1.16968656e+00 7.85347581e-01 1.00867664e-02 -4.43704456e-01 -6.56944454e-01 -8.50935102e-01 -4.58215654e-01 1.13971901e+00 -1.08865356e+00 -1.51554132e+00 4.85092819e-01 1.23395376e-01 1.03726625e+00 -8.50452840e-01 1.21083713e+00 6.07194722e-01 -6.65238738e-01 8.37713897e-01 2.92968452e-01 3.47830653e-01 1.59018978e-01 -1.20492661e+00 1.12278783e+00 1.03578103e+00 5.17435312e-01 6.51123285e-01 5.31796455e-01 -5.22635818e-01 -1.19426692e+00 -7.83696055e-01 7.91597128e-01 -3.50532651e-01 9.19172943e-01 -8.78034353e-01 -9.53183770e-01 8.41572762e-01 3.60389769e-01 -4.90532547e-01 1.00478029e+00 4.01651800e-01 -5.68858504e-01 -2.30242126e-02 -9.56113338e-01 4.58060652e-01 8.32901418e-01 -9.40570951e-01 -4.78151828e-01 1.53919712e-01 9.09248531e-01 -1.08362511e-01 -8.73106658e-01 2.97288120e-01 5.95402777e-01 -4.03993338e-01 9.63116825e-01 -7.90176988e-01 3.00434530e-02 3.61711276e-03 -8.24566066e-01 -1.58232176e+00 1.10800518e-02 -6.13402188e-01 1.17221318e-01 1.76794207e+00 5.96532643e-01 -6.01789713e-01 6.50311470e-01 3.53511661e-01 -7.46378183e-01 -2.47681946e-01 -1.33470631e+00 -1.05841863e+00 7.10210562e-01 -1.07658553e+00 5.38140655e-01 7.41818309e-01 -2.82087594e-01 -4.85232174e-02 -6.15141451e-01 4.36113298e-01 3.38952720e-01 -9.09154713e-01 5.14386952e-01 -7.80425489e-01 -3.55083823e-01 -3.99685889e-01 -7.48151958e-01 -7.42480338e-01 3.86866093e-01 -7.62191594e-01 3.28872681e-01 -1.27575314e+00 -7.27737024e-02 -4.27815616e-01 -1.62661135e-01 7.34517872e-01 3.07462394e-01 1.27676845e-01 -3.16708870e-02 -2.72791952e-01 2.86289930e-01 4.44111586e-01 1.22121938e-01 -3.32346886e-01 6.46463130e-04 -1.06708579e-01 -4.25895602e-01 9.00081217e-01 9.32325721e-01 -1.71682730e-01 -3.05471234e-02 -8.93425167e-01 -2.60395050e-01 -1.56914473e-01 1.53897107e-01 -1.15046382e+00 -1.18194707e-01 -6.15363121e-02 2.36386895e-01 -5.32415926e-01 5.62386990e-01 -3.39279801e-01 8.21194872e-02 4.75130349e-01 -4.37407732e-01 -1.91175397e-02 8.80008280e-01 1.27654716e-01 -2.16249794e-01 -2.18588337e-01 9.12463963e-01 -1.58385515e-01 -8.63687038e-01 1.58761293e-02 -1.20314801e+00 -5.73453195e-02 8.31794664e-02 2.61461079e-01 -2.82130539e-01 -5.94588757e-01 -8.39489162e-01 -2.18093246e-01 -1.11450283e-02 6.00777924e-01 6.66219652e-01 -1.04139292e+00 -1.05687940e+00 1.77277490e-01 -7.46637210e-02 -3.05319548e-01 2.54289657e-01 3.37476790e-01 -8.74759495e-01 6.60216153e-01 1.06421702e-01 -5.25581896e-01 -1.20388758e+00 1.66706219e-01 6.78519368e-01 2.29529709e-01 -4.67671841e-01 1.04593122e+00 1.43240690e-01 -1.15955162e+00 4.79957581e-01 -1.26462266e-01 -1.10188507e-01 -3.15037608e-01 6.43431485e-01 3.63910407e-01 3.97319525e-01 -1.01864207e+00 -4.89358336e-01 6.37635440e-02 6.06133603e-02 -6.41175449e-01 1.53322411e+00 9.58020389e-02 1.77703619e-01 7.50632286e-01 1.05881536e+00 1.34762242e-01 -6.96478248e-01 -2.27628946e-02 6.78042248e-02 -2.73755401e-01 4.98066038e-01 -8.52186084e-01 -7.34547138e-01 1.48592007e+00 9.86972213e-01 4.88344803e-02 7.45063841e-01 6.22742251e-02 9.17654097e-01 6.46445453e-01 3.19604099e-01 -1.12642956e+00 -3.66165668e-01 7.63359785e-01 8.48683000e-01 -9.62658048e-01 -5.75904965e-01 7.29085645e-04 -3.37354392e-01 8.64811063e-01 4.02368695e-01 1.48157671e-01 6.13798201e-01 8.51412296e-01 4.31629807e-01 1.18049800e-01 -6.02200866e-01 -3.72317344e-01 -5.22623770e-02 1.18269205e+00 8.52537334e-01 3.20929796e-01 6.30710199e-02 5.45905828e-01 -5.51940322e-01 -4.02933478e-01 5.33969879e-01 6.65035307e-01 -6.16669178e-01 -1.00940907e+00 -4.48694706e-01 3.26645166e-01 -8.35901737e-01 -6.11962795e-01 -4.76020187e-01 5.93186617e-01 -3.30645084e-01 1.22084749e+00 3.31577659e-01 -4.44856644e-01 9.96527970e-02 3.36206138e-01 9.68107283e-02 -7.63045490e-01 -5.97924232e-01 5.22431016e-01 6.66956127e-01 -8.66975263e-02 -2.40656063e-02 -7.04278648e-01 -1.02256382e+00 -4.58397627e-01 -3.98543179e-01 1.13047650e-02 1.18928289e+00 7.90597618e-01 -2.24001110e-02 2.82539666e-01 3.43544006e-01 -8.32922041e-01 -6.62001908e-01 -1.31130779e+00 -7.33673215e-01 -2.96683282e-01 4.29178715e-01 -1.73767552e-01 -3.80056262e-01 -5.76334782e-02]
[14.288917541503906, 6.744553565979004]
90eafdfa-17dd-4f4a-a8de-bd74d290d7d0
deep-gradient-learning-for-efficient
2205.12853
null
https://arxiv.org/abs/2205.12853v2
https://arxiv.org/pdf/2205.12853v2.pdf
Deep Gradient Learning for Efficient Camouflaged Object Detection
This paper introduces DGNet, a novel deep framework that exploits object gradient supervision for camouflaged object detection (COD). It decouples the task into two connected branches, i.e., a context and a texture encoder. The essential connection is the gradient-induced transition, representing a soft grouping between context and texture features. Benefiting from the simple but efficient framework, DGNet outperforms existing state-of-the-art COD models by a large margin. Notably, our efficient version, DGNet-S, runs in real-time (80 fps) and achieves comparable results to the cutting-edge model JCSOD-CVPR$_{21}$ with only 6.82% parameters. Application results also show that the proposed DGNet performs well in polyp segmentation, defect detection, and transparent object segmentation tasks. Codes will be made available at https://github.com/GewelsJI/DGNet.
['Luc van Gool', 'Alexander Liniger', 'Dengxin Dai', 'Yu-Cheng Chou', 'Deng-Ping Fan', 'Ge-Peng Ji']
2022-05-25
null
null
null
null
['defect-detection']
['computer-vision']
[ 2.67398804e-01 1.92022771e-01 -1.73185006e-01 -1.28594190e-01 -7.08858192e-01 -3.41841578e-01 5.55770770e-02 -1.88269913e-01 -1.24564156e-01 3.16685706e-01 -3.75236958e-01 -4.48174506e-01 5.31154811e-01 -5.80088556e-01 -7.89517760e-01 -6.65036798e-01 6.20993301e-02 -5.69722429e-02 1.03016686e+00 1.10493928e-01 1.45429701e-01 1.54314786e-01 -1.42639983e+00 4.07283247e-01 8.95721257e-01 1.33637989e+00 2.78701693e-01 8.61612320e-01 1.27290919e-01 7.08600163e-01 -5.57022393e-01 -6.13885224e-01 3.51347566e-01 -2.83506006e-01 -7.46524811e-01 2.98021019e-01 6.70070171e-01 -6.24271154e-01 -3.86024952e-01 1.11706054e+00 6.29698694e-01 -3.61686498e-01 1.42918542e-01 -8.87589991e-01 -7.03912199e-01 4.34010983e-01 -8.11494291e-01 3.44332397e-01 -1.01176105e-01 3.35767925e-01 8.17030728e-01 -8.19591045e-01 8.20739567e-01 1.03044629e+00 6.87794030e-01 7.28451908e-01 -1.11566246e+00 -2.31374159e-01 1.82996854e-01 -2.45768595e-02 -9.73706007e-01 -1.50539815e-01 5.27703702e-01 -3.92853886e-01 7.71119773e-01 1.02411382e-01 8.62780213e-01 8.77197564e-01 3.39815497e-01 1.21507955e+00 9.65701699e-01 -3.34168732e-01 1.21183485e-01 -1.31232381e-01 9.63818431e-02 1.26899219e+00 5.84774613e-01 2.82627791e-01 -4.01178420e-01 1.23523241e-02 9.34703469e-01 -3.13880771e-01 -4.53827232e-01 -3.85338515e-01 -9.22386348e-01 4.97645825e-01 4.70333993e-01 2.53742896e-02 2.26805374e-01 4.47006136e-01 6.22432768e-01 1.01847298e-01 6.41465604e-01 1.42624862e-02 -5.23239851e-01 -1.19170591e-01 -6.15778148e-01 -6.76093251e-02 7.47560978e-01 1.16772425e+00 4.06913877e-01 1.28029823e-01 -2.32964426e-01 7.50094891e-01 3.11818540e-01 4.20299709e-01 2.44368181e-01 -9.62182999e-01 1.09703638e-01 4.58221287e-01 -1.46442458e-01 -7.20599473e-01 -1.24019802e-01 -4.91396844e-01 -4.56854224e-01 4.71219301e-01 4.69164878e-01 -2.45765626e-01 -1.38378537e+00 1.37376380e+00 5.47337174e-01 4.74029571e-01 -3.80204916e-01 1.05493379e+00 1.09699416e+00 2.68281519e-01 1.31306872e-01 3.96145672e-01 1.40366626e+00 -1.39700234e+00 -4.54348952e-01 -2.10770458e-01 6.76874816e-01 -9.35714304e-01 1.19183946e+00 4.72168356e-01 -1.36734784e+00 -4.39211667e-01 -1.10551429e+00 -5.84879220e-01 -2.23325163e-01 7.10275769e-01 8.80629480e-01 7.67346859e-01 -1.14153361e+00 5.74643314e-01 -1.12149584e+00 -1.63955122e-01 9.03994560e-01 4.11475867e-01 4.51456122e-02 -1.34725839e-01 -5.85697293e-01 3.57916206e-01 6.76868763e-03 1.42908067e-01 -1.10713542e+00 -7.62109637e-01 -8.50463331e-01 1.34829417e-04 3.94176066e-01 -7.88329303e-01 1.53263474e+00 -8.47853959e-01 -1.72425568e+00 1.37493300e+00 -6.75266013e-02 -6.01511240e-01 6.99329913e-01 -3.63075584e-01 -2.19526574e-01 3.25013638e-01 5.52287810e-02 8.49736094e-01 1.04375899e+00 -1.14491320e+00 -9.98347759e-01 -6.55078366e-02 1.20925814e-01 -1.47998750e-01 8.13499987e-02 1.25703156e-01 -9.89047825e-01 -8.05117130e-01 -5.88581599e-02 -8.94147336e-01 -5.75294495e-02 6.34647727e-01 -7.39554584e-01 -7.89388716e-02 1.20180774e+00 -7.49031365e-01 9.70914364e-01 -2.41415906e+00 -1.70242608e-01 -2.24142402e-01 4.96868044e-01 8.19247544e-01 -5.12955710e-02 -2.07810521e-01 1.67644665e-01 5.45519702e-02 -5.06069839e-01 -5.84951818e-01 -1.10755190e-01 1.23202810e-02 4.65497039e-02 6.61914349e-01 3.15331548e-01 1.08838177e+00 -7.81887949e-01 -5.18917739e-01 4.32437241e-01 5.34589291e-01 -6.14974380e-01 -5.02680466e-02 -4.45472807e-01 2.29247481e-01 -3.27634424e-01 1.18623173e+00 9.82297540e-01 -6.06134057e-01 1.31721899e-01 -6.03212826e-02 -6.57665506e-02 2.38042682e-01 -8.19547951e-01 1.90354955e+00 -3.55854146e-02 9.03030455e-01 3.18608463e-01 -7.82167256e-01 6.04195774e-01 8.11305195e-02 2.43840247e-01 -4.81246829e-01 4.57685173e-01 4.00468022e-01 7.74696842e-02 -4.28278595e-01 3.19748133e-01 4.44583625e-01 1.27312347e-01 7.93419704e-02 2.30412617e-01 -1.79180741e-01 3.53655100e-01 1.88898399e-01 1.20377791e+00 3.13970298e-01 6.96308538e-02 -3.92753124e-01 3.76425713e-01 -5.63745946e-02 9.49642181e-01 6.86319172e-01 -5.54510295e-01 7.95837820e-01 6.83439136e-01 -4.09474909e-01 -7.08944082e-01 -1.19963861e+00 -2.08062470e-01 6.70304954e-01 6.54844344e-01 -4.06357408e-01 -8.51387322e-01 -1.01321793e+00 9.00617838e-02 1.85005233e-01 -6.55029118e-01 -5.00651747e-02 -7.02561796e-01 -6.54740512e-01 3.66223633e-01 5.01879513e-01 9.14104342e-01 -1.04167771e+00 -7.46626437e-01 7.16293529e-02 6.61025643e-02 -1.18564284e+00 -7.04203844e-01 3.68484184e-02 -8.07619810e-01 -1.26724672e+00 -7.00676978e-01 -1.15244877e+00 6.97603822e-01 3.86017114e-01 1.29651213e+00 4.76704210e-01 -8.50912452e-01 2.37710074e-01 -3.44574183e-01 -4.24639702e-01 -1.78007647e-01 -3.66594866e-02 -6.47092819e-01 -1.96761280e-01 2.24703610e-01 -2.06559300e-01 -1.05831122e+00 4.04470325e-01 -9.03593123e-01 1.07304595e-01 5.98786891e-01 9.68879580e-01 8.58779490e-01 -2.35631987e-01 1.87074497e-01 -1.00644982e+00 3.90482210e-02 -1.03720196e-01 -9.00517464e-01 1.44889548e-01 -2.90169984e-01 -4.46872681e-01 2.84146875e-01 -2.95667499e-01 -1.05817032e+00 1.88897371e-01 -2.24908501e-01 -3.13381851e-01 -2.52777249e-01 -9.08186361e-02 -5.88265024e-02 -5.38146138e-01 3.62596035e-01 -9.27329063e-03 -1.10155143e-01 -5.30739367e-01 2.92700887e-01 4.16660219e-01 5.62252760e-01 -4.02565211e-01 4.82432157e-01 9.28567529e-01 -2.34296113e-01 -7.82586455e-01 -8.47970665e-01 -4.97167796e-01 -2.99629778e-01 -6.50269985e-02 8.45882773e-01 -7.86236882e-01 -6.75805748e-01 1.01972127e+00 -1.06218672e+00 -9.35862064e-01 -7.20384657e-01 2.85625368e-01 -5.43852627e-01 5.74565232e-01 -1.24219239e+00 -2.86990553e-01 -4.11557615e-01 -1.23052549e+00 1.38285160e+00 5.29238760e-01 4.22584742e-01 -8.09168756e-01 -4.92330372e-01 2.67489165e-01 2.80371398e-01 3.00025433e-01 6.22375190e-01 -1.59862861e-01 -1.02633297e+00 -3.67108099e-02 -6.35363936e-01 6.16902709e-01 8.17145109e-02 2.42328748e-01 -9.73036528e-01 -3.83900374e-01 -9.05079991e-02 -3.28224987e-01 1.27808917e+00 8.10593665e-01 1.52254951e+00 9.07752961e-02 -5.39126635e-01 9.81250823e-01 1.39230430e+00 2.21616000e-01 6.72543287e-01 1.37383908e-01 5.93976498e-01 9.02716666e-02 6.83955848e-01 2.15268776e-01 2.29191989e-01 4.92743313e-01 6.06938183e-01 -5.84359288e-01 -9.22798097e-01 -5.51444814e-02 1.00738607e-01 5.73599160e-01 1.79982185e-01 -4.28947568e-01 -7.10385025e-01 5.61710417e-01 -1.72446954e+00 -4.55805719e-01 -3.08529586e-01 1.94752097e+00 6.06823146e-01 2.95153230e-01 3.39967832e-02 -5.75904809e-02 8.24665606e-01 6.67253733e-02 -8.50104034e-01 -3.06184500e-01 -1.73374102e-01 3.44952822e-01 5.77507079e-01 3.65572989e-01 -1.40073884e+00 1.25192165e+00 6.05105257e+00 9.80253756e-01 -1.18354917e+00 2.53584117e-01 9.37738657e-01 1.48373514e-01 1.10750899e-01 6.58172276e-03 -9.46178794e-01 6.92633510e-01 2.79720455e-01 4.17822093e-01 7.33878389e-02 9.38309610e-01 -1.55293167e-01 -3.06349277e-01 -7.21684456e-01 6.95295334e-01 -1.75490659e-02 -1.44031823e+00 -2.19354108e-01 2.00204346e-02 7.96293914e-01 2.23823503e-01 3.80629480e-01 -5.22116292e-03 4.68752503e-01 -5.74305415e-01 7.71222472e-01 -2.71281768e-02 1.04648757e+00 -1.89573213e-01 6.00809455e-01 -2.17423037e-01 -1.33766866e+00 -7.02590942e-02 -4.36498016e-01 3.57447892e-01 3.27527136e-01 9.54641104e-01 -5.48944473e-01 3.65462691e-01 1.01378512e+00 9.67038870e-01 -4.89963263e-01 1.45948243e+00 -6.00295126e-01 8.81635129e-01 -2.81597644e-01 4.46180820e-01 8.81919824e-03 6.72671571e-02 6.78739488e-01 1.32589388e+00 2.06635967e-01 -1.04554348e-01 1.85175404e-01 9.30248499e-01 -1.96077973e-01 -2.77866870e-01 -1.23110048e-01 1.51467383e-01 2.62958735e-01 1.25279558e+00 -1.12899601e+00 -3.90961409e-01 -6.87346637e-01 1.24877059e+00 1.92531571e-01 3.02454144e-01 -1.05269551e+00 -5.60711622e-01 6.76113546e-01 1.22752711e-01 1.01647651e+00 -8.14931020e-02 -3.14621449e-01 -1.22620058e+00 1.71915650e-01 -6.61966920e-01 4.36893523e-01 -6.34012878e-01 -1.14924800e+00 5.77181458e-01 -4.86122400e-01 -1.32750010e+00 4.53503400e-01 -1.09618497e+00 -5.34676969e-01 3.94152224e-01 -1.90929782e+00 -1.36681938e+00 -3.29525560e-01 4.15977687e-01 5.91267288e-01 2.50265300e-01 3.55823815e-01 2.85665959e-01 -8.04414213e-01 7.06113815e-01 -7.47149587e-02 3.06733519e-01 5.82887173e-01 -1.44625068e+00 8.62023771e-01 1.16379988e+00 -1.60032198e-01 1.86181948e-01 3.13696355e-01 -7.41502762e-01 -1.09836555e+00 -1.17347050e+00 4.83188361e-01 2.03920584e-02 5.60222566e-01 -4.13020313e-01 -7.02795863e-01 6.11666679e-01 7.14572892e-02 6.05015337e-01 2.93112338e-01 -3.42334837e-01 -2.98950881e-01 4.90286238e-02 -1.35188317e+00 5.50161660e-01 1.34278929e+00 -1.76843658e-01 -8.81586596e-02 4.06702787e-01 9.79063034e-01 -1.09384632e+00 -6.68667316e-01 5.11419713e-01 4.08863038e-01 -1.16126132e+00 9.92003620e-01 -1.45701885e-01 5.17029703e-01 -3.07671577e-01 1.54007912e-01 -9.48522866e-01 -2.50069380e-01 -8.35869014e-01 -3.99692595e-01 8.76724303e-01 4.37614560e-01 -7.79770851e-01 1.00444603e+00 2.68869221e-01 -5.78434587e-01 -1.04243886e+00 -8.70000482e-01 -7.41505861e-01 -4.85791527e-02 -2.03906357e-01 3.39067489e-01 5.50955653e-01 -3.89214545e-01 -2.33155400e-01 -1.92010492e-01 2.20264733e-01 5.94120264e-01 2.66773134e-01 4.21832234e-01 -9.08102810e-01 -5.71757972e-01 -4.33963746e-01 -5.37583947e-01 -1.52726531e+00 -2.98046798e-01 -6.91996694e-01 9.18172747e-02 -1.38977528e+00 -1.75123457e-02 -5.54399312e-01 -2.69704670e-01 4.88310665e-01 -3.27851892e-01 5.43699801e-01 1.99348763e-01 -1.53040007e-01 -7.53269672e-01 2.60316253e-01 1.56747007e+00 -1.22946307e-01 -1.81098863e-01 1.84938870e-02 -6.34053349e-01 9.46414769e-01 1.21423030e+00 -4.12903577e-01 -1.07077152e-01 -6.95694983e-01 -2.88448185e-01 -3.02152634e-01 5.72407484e-01 -9.54742610e-01 2.94143669e-02 2.96704948e-01 2.32031822e-01 -3.44926536e-01 3.20461929e-01 -4.08711106e-01 -4.18730438e-01 6.77457511e-01 -5.95885813e-02 -3.18003923e-01 3.90168786e-01 5.70798874e-01 -5.09781130e-02 -1.04879014e-01 1.09871566e+00 -9.40015316e-02 -8.89994264e-01 5.38471580e-01 -3.90800506e-01 2.51136035e-01 9.94759679e-01 -3.38900357e-01 -9.13572609e-01 1.46692127e-01 -7.69352555e-01 7.69890030e-04 5.58723569e-01 2.42595911e-01 6.36007726e-01 -9.43164587e-01 -5.39288878e-01 3.61296386e-01 -5.01618348e-02 2.19977230e-01 3.79917681e-01 1.03680933e+00 -8.22113514e-01 2.65418321e-01 7.41387457e-02 -8.69555712e-01 -1.38862860e+00 3.03293705e-01 5.30305266e-01 -2.31207088e-01 -1.07722330e+00 1.41778803e+00 3.79573733e-01 -1.06197692e-01 2.29438618e-01 -7.17483044e-01 4.08045948e-01 -6.64052844e-01 3.74011189e-01 4.58526015e-01 1.61763549e-01 -2.40565062e-01 -1.88395411e-01 6.11888468e-01 -4.13146585e-01 2.59033889e-01 1.01068485e+00 -4.74888161e-02 -8.95505250e-02 -6.69783726e-02 9.61027622e-01 -2.39854977e-02 -1.66181326e+00 -2.05129832e-01 -1.92241371e-01 -8.20548832e-01 1.94170132e-01 -1.01419532e+00 -1.56981742e+00 8.03180039e-01 8.22172344e-01 1.47163197e-01 1.31468332e+00 1.73005834e-01 1.18903625e+00 -1.22705631e-01 2.96905518e-01 -9.44247544e-01 1.35331661e-01 1.94122329e-01 5.35392642e-01 -1.20317292e+00 -9.15541276e-02 -1.19063830e+00 -3.74021620e-01 8.17457676e-01 8.70650470e-01 -4.29553717e-01 8.22683692e-01 6.64523721e-01 1.60652027e-01 -2.77612925e-01 -5.46767890e-01 -4.14509982e-01 2.40422383e-01 7.23000944e-01 4.10859257e-01 1.16731070e-01 -3.68078053e-01 2.38149747e-01 2.12222561e-01 1.45978078e-01 4.08223867e-01 1.11128628e+00 -3.28532904e-01 -1.11794674e+00 1.08836800e-01 6.07019067e-01 -7.68349469e-01 -4.33315858e-02 -1.65902138e-01 9.16084170e-01 4.37153906e-01 7.76687622e-01 2.75784116e-02 -1.80916384e-01 1.77931517e-01 -5.56295693e-01 4.19662237e-01 -6.51198924e-01 -5.55494726e-01 1.48362160e-01 5.98863252e-02 -9.83049333e-01 -2.17660144e-01 -5.45647502e-01 -1.15332389e+00 -2.12326676e-01 -5.97998559e-01 -3.10249805e-01 4.04263169e-01 3.72059196e-01 6.69180453e-01 7.18780100e-01 2.35440269e-01 -7.40412712e-01 -2.30680794e-01 -6.12714291e-01 -6.59333944e-01 9.17263925e-02 4.81004536e-01 -5.24499834e-01 -3.46368790e-01 1.96712375e-01]
[9.45703125, -0.09012100845575333]
a54e9b0b-247e-4b4f-ad40-6d755c9fb14d
robust-bayesian-inference-for-measurement
2306.01468
null
https://arxiv.org/abs/2306.01468v1
https://arxiv.org/pdf/2306.01468v1.pdf
Robust Bayesian Inference for Measurement Error Models
Measurement error occurs when a set of covariates influencing a response variable are corrupted by noise. This can lead to misleading inference outcomes, particularly in problems where accurately estimating the relationship between covariates and response variables is crucial, such as causal effect estimation. Existing methods for dealing with measurement error often rely on strong assumptions such as knowledge of the error distribution or its variance and availability of replicated measurements of the covariates. We propose a Bayesian Nonparametric Learning framework which is robust to mismeasured covariates, does not require the preceding assumptions, and is able to incorporate prior beliefs about the true error distribution. Our approach gives rise to two methods that are robust to measurement error via different loss functions: one based on the Total Least Squares objective and the other based on Maximum Mean Discrepancy (MMD). The latter allows for generalisation to non-Gaussian distributed errors and non-linear covariate-response relationships. We provide bounds on the generalisation error using the MMD-loss and showcase the effectiveness of the proposed framework versus prior art in real-world mental health and dietary datasets that contain significant measurement errors.
['Theodoros Damoulas', 'Charita Dellaporta']
2023-06-02
null
null
null
null
['bayesian-inference']
['methodology']
[ 6.58023536e-01 1.50981620e-01 -2.64208227e-01 -6.52782738e-01 -8.44458997e-01 -2.61573106e-01 5.20123065e-01 5.61678588e-01 -5.80998778e-01 1.15721321e+00 2.56512374e-01 -2.24609897e-01 -5.80899000e-01 -7.23106742e-01 -1.09187686e+00 -7.72992194e-01 -1.16412662e-01 4.61978734e-01 -1.98962521e-02 2.92164594e-01 2.15810493e-01 1.63801789e-01 -1.08432436e+00 -4.54291701e-01 1.08359754e+00 7.18679428e-01 -1.02277789e-02 2.98610359e-01 9.34499353e-02 4.55503285e-01 -5.43009460e-01 -2.11362496e-01 1.51340440e-02 -2.87967294e-01 -3.75255257e-01 -7.19271079e-02 2.31405526e-01 -1.85816705e-01 3.68962139e-01 1.08717620e+00 5.93951881e-01 7.20431507e-02 1.19895506e+00 -1.35935378e+00 -5.28925359e-01 5.17696679e-01 -9.08439994e-01 -4.22371216e-02 1.29754931e-01 -1.51220024e-01 4.56132531e-01 -5.75652480e-01 1.34974211e-01 1.54362476e+00 1.03701150e+00 9.44036990e-02 -1.78099048e+00 -7.41613388e-01 4.95134592e-02 -1.33870348e-01 -1.38208950e+00 -5.47688484e-01 4.09080654e-01 -9.11675632e-01 3.75827312e-01 6.20735139e-02 -2.56996192e-02 1.22225082e+00 4.43927795e-01 1.41334698e-01 1.46777070e+00 -4.63946551e-01 5.41442752e-01 2.65010297e-01 2.36426041e-01 1.89674839e-01 7.52451122e-01 4.08844829e-01 -1.35569826e-01 -6.67481005e-01 6.78456664e-01 5.66897877e-02 -3.54226083e-01 -4.74421710e-01 -9.01225030e-01 1.04302347e+00 -8.53227079e-03 -8.76043141e-02 -6.88823462e-01 2.23860100e-01 3.64709675e-01 1.68293133e-01 6.00493848e-01 1.57457247e-01 -4.98259097e-01 1.30270675e-01 -1.02313340e+00 3.10227066e-01 6.52460098e-01 8.36149156e-01 5.10743976e-01 -1.25845879e-01 -3.88519347e-01 7.64673591e-01 4.37232703e-01 8.78781080e-01 1.10418037e-01 -7.64269650e-01 2.77989268e-01 1.60998836e-01 7.41971493e-01 -1.14158273e+00 -5.79815149e-01 -1.84037089e-01 -1.09355569e+00 1.30190328e-01 6.86664879e-01 -4.77525532e-01 -8.68929803e-01 2.24019861e+00 5.55967867e-01 3.03717375e-01 -3.23364109e-01 7.16113508e-01 4.63649362e-01 2.58337438e-01 4.44237024e-01 -7.14204967e-01 1.13321996e+00 -1.63396195e-01 -1.09822476e+00 -7.58699775e-02 3.02340060e-01 -5.59720457e-01 8.18078756e-01 3.28892767e-01 -1.15089011e+00 -3.31206292e-01 -7.70116150e-01 1.20610908e-01 -9.13473964e-02 -1.03765890e-01 3.32837015e-01 8.80828023e-01 -6.93232119e-01 5.10113955e-01 -9.29959357e-01 -2.23982960e-01 2.22091258e-01 5.51459193e-01 -3.22155416e-01 -4.92367381e-03 -1.02403176e+00 8.95837128e-01 2.86114067e-01 2.41468832e-01 -8.38520944e-01 -1.04456484e+00 -7.57900119e-01 1.02346905e-01 5.06163001e-01 -7.51174688e-01 9.21727121e-01 -9.64308023e-01 -1.31103742e+00 4.32566822e-01 -2.13856354e-01 -2.56685734e-01 7.36208200e-01 -2.29738176e-01 -1.31684661e-01 -3.18957537e-01 8.76315609e-02 -3.84660326e-02 1.03412020e+00 -8.46646667e-01 -1.44982129e-01 -6.04585111e-01 -3.64884585e-01 -1.07559003e-01 1.36783004e-01 1.98425371e-02 3.26322854e-01 -6.68678761e-01 1.67598531e-01 -6.35176361e-01 -2.18310297e-01 7.21446723e-02 -4.43947971e-01 -7.19417483e-02 1.82519346e-01 -7.18237579e-01 1.06453383e+00 -2.06751633e+00 7.10387751e-02 2.79549479e-01 -9.13444161e-02 -2.03066558e-01 8.46483000e-03 5.88681519e-01 -1.38827890e-01 4.72128913e-02 -5.29642522e-01 -2.02263907e-01 1.06056929e-01 1.66624695e-01 5.94886653e-02 1.10732090e+00 2.46338531e-01 4.15652215e-01 -7.82363057e-01 -2.19024256e-01 5.05752042e-02 5.85625887e-01 -4.22169805e-01 3.24076116e-01 5.11966832e-02 6.52492821e-01 -3.11308891e-01 9.56033990e-02 9.32690978e-01 -1.13749772e-01 1.38949677e-01 -5.03159799e-02 -1.59410954e-01 7.80954212e-02 -1.70443630e+00 1.26445436e+00 -3.38424027e-01 -2.21086890e-02 2.55128443e-01 -1.35812771e+00 1.02392304e+00 3.95451725e-01 3.68593693e-01 -2.54598230e-01 2.06986308e-01 2.73376405e-01 -8.47842917e-02 -5.29711604e-01 -1.31772920e-01 -6.79966629e-01 -7.03746900e-02 1.14934959e-01 -6.03164956e-02 1.75602943e-01 -2.41879389e-01 -3.36827695e-01 9.68680859e-01 3.90379503e-02 8.17906916e-01 -6.50945187e-01 3.60704988e-01 -5.63985348e-01 8.67474496e-01 1.01989532e+00 5.50460629e-02 3.92534494e-01 7.34359324e-01 -1.10834464e-02 -1.03956795e+00 -1.08725333e+00 -7.35863090e-01 5.84838986e-01 -2.79716820e-01 4.33675468e-01 -3.56649548e-01 -2.58270979e-01 2.99207717e-01 8.60463083e-01 -8.70966911e-01 -3.02315623e-01 -6.85077012e-02 -1.18378615e+00 4.09272909e-01 5.04505217e-01 9.12323967e-02 -4.90150660e-01 -4.40956861e-01 3.94572377e-01 -2.14444054e-03 -6.21089816e-01 -2.56321728e-01 1.77572355e-01 -9.71091330e-01 -1.25318050e+00 -7.26225853e-01 -1.30568728e-01 5.98226428e-01 -3.04284126e-01 1.07671666e+00 -3.39022219e-01 -4.76958230e-02 3.15436333e-01 -2.32656244e-02 -6.73116028e-01 -4.42375124e-01 -5.55969119e-01 1.59119323e-01 7.22609237e-02 5.15034974e-01 -6.00644886e-01 -6.31898344e-01 1.18061550e-01 -8.72386634e-01 -4.53075081e-01 5.50229013e-01 9.77119088e-01 5.33554435e-01 9.02351961e-02 9.54148233e-01 -1.24800646e+00 4.32203382e-01 -1.01911461e+00 -7.95777082e-01 1.78839371e-01 -8.03378046e-01 1.72075957e-01 2.11782932e-01 -6.34944260e-01 -1.12361228e+00 -2.54284024e-01 1.17212527e-01 4.63354811e-02 -2.96956092e-01 6.45048201e-01 -3.10120881e-01 3.59411120e-01 6.10291362e-01 -1.75438464e-01 -9.20325294e-02 -6.60838723e-01 1.72098592e-01 6.89260244e-01 3.77660006e-01 -7.70577669e-01 2.50968128e-01 3.54034007e-01 5.53024352e-01 -7.03624070e-01 -6.27677262e-01 -4.03490186e-01 -3.89699847e-01 3.19463611e-01 8.16212535e-01 -8.50705683e-01 -8.57233286e-01 4.75240767e-01 -9.42732275e-01 -1.08702138e-01 -1.36300176e-01 9.96169627e-01 -6.80367291e-01 3.52968127e-01 -2.82357693e-01 -1.28581691e+00 -2.14836836e-01 -1.12339747e+00 9.23946857e-01 -7.48387203e-02 -2.28442505e-01 -1.35273027e+00 3.10197800e-01 -6.48911372e-02 3.85763347e-01 7.93115735e-01 1.23138797e+00 -6.83920920e-01 9.38583240e-02 -2.32346788e-01 -3.21741074e-01 2.75394708e-01 3.40656161e-01 -2.23410472e-01 -7.38241613e-01 -5.39428890e-01 2.03995258e-01 -1.78293660e-01 6.63370669e-01 1.18946147e+00 8.50609720e-01 -5.86791039e-01 -2.65794396e-01 1.91110283e-01 1.57394803e+00 6.92739263e-02 4.37129408e-01 -1.16592748e-02 3.70402873e-01 7.40843773e-01 3.33914667e-01 7.64828622e-01 3.33719939e-01 7.65124679e-01 3.31532598e-01 -4.25261743e-02 4.22237426e-01 -1.79261252e-01 6.95242211e-02 3.67996484e-01 2.84421891e-01 -2.53180563e-01 -6.51750624e-01 8.06340098e-01 -1.86701393e+00 -7.35070884e-01 -6.19114041e-01 2.90353918e+00 1.09854567e+00 -1.03038035e-01 3.36273611e-01 2.03051362e-02 8.52997541e-01 -3.19658905e-01 -7.32836545e-01 -4.76178497e-01 3.73226069e-02 1.73392758e-01 8.25693071e-01 6.43448651e-01 -8.38374138e-01 6.51818737e-02 6.90626860e+00 4.09261167e-01 -6.39009535e-01 3.65497619e-01 3.93878102e-01 2.91291267e-01 -2.09849149e-01 4.16238755e-02 -5.90997934e-01 7.29639471e-01 1.12585545e+00 -1.43413171e-01 3.48525345e-02 3.18214178e-01 8.29504550e-01 -4.35239613e-01 -1.25379431e+00 4.76371765e-01 -3.65656227e-01 -5.09603143e-01 -5.41290998e-01 1.80144161e-01 7.20270574e-01 -4.81564999e-01 -3.90945263e-02 1.48985445e-01 5.06308794e-01 -1.12207520e+00 3.97737473e-01 1.00990486e+00 7.26412416e-01 -6.78611994e-01 9.79930401e-01 5.47487617e-01 -4.62584078e-01 5.13463728e-02 -5.78731835e-01 -2.53350884e-01 1.05644293e-01 1.22485590e+00 -6.59146845e-01 2.87839681e-01 3.45480412e-01 4.02598172e-01 -2.37091668e-02 1.18080485e+00 -1.21910609e-01 9.09850657e-01 -5.65203965e-01 2.53433287e-01 -1.32842854e-01 -2.69820750e-01 4.57862407e-01 1.08148241e+00 5.47182441e-01 1.97140453e-03 1.25203952e-01 1.07613039e+00 2.35047087e-01 2.81452507e-01 -3.34170252e-01 1.98429808e-01 7.45365322e-01 6.34482265e-01 -3.40975642e-01 -1.15319908e-01 -5.69544911e-01 5.15765607e-01 2.25596175e-01 4.80308652e-01 -6.57993972e-01 2.80338433e-02 4.44051445e-01 2.24977478e-01 1.86126575e-01 1.00433350e-01 -2.75748074e-01 -7.66453743e-01 2.48112902e-02 -7.34555304e-01 4.02779669e-01 -3.64650667e-01 -1.62166345e+00 -1.69189379e-01 4.90733534e-01 -5.81778347e-01 -1.97110102e-01 -3.94469589e-01 -3.73345971e-01 1.39415991e+00 -1.38494933e+00 -8.95302713e-01 6.23219945e-02 4.05719787e-01 7.78135061e-02 3.19847763e-01 7.23219156e-01 3.00481170e-01 -4.36640620e-01 5.75591743e-01 3.72586519e-01 -4.29192990e-01 9.06628489e-01 -1.44881701e+00 -2.23611906e-01 5.50400555e-01 -6.10981822e-01 5.72788715e-01 1.18187404e+00 -9.01417017e-01 -9.44626749e-01 -9.42973137e-01 7.75404572e-01 -3.00705224e-01 5.76235592e-01 -3.20625782e-01 -1.05330586e+00 7.62315810e-01 -3.20647150e-01 -1.15572527e-01 8.30200136e-01 3.90482724e-01 -2.90425062e-01 -2.86273728e-03 -1.54933298e+00 6.82298988e-02 5.65734744e-01 -1.01309329e-01 -4.79788899e-01 2.75116950e-01 4.78552312e-01 -1.62600264e-01 -1.23950171e+00 5.69607735e-01 6.78899944e-01 -7.72124946e-01 7.82653928e-01 -9.59953070e-01 2.21305236e-01 -2.24753499e-01 -2.17690840e-01 -1.33347738e+00 -5.11998892e-01 -2.78492838e-01 1.39027566e-01 1.38757324e+00 2.72042304e-01 -5.72631180e-01 1.87375948e-01 8.63887191e-01 3.10837209e-01 -3.20684254e-01 -1.06659257e+00 -7.65947461e-01 4.63366866e-01 -4.59451914e-01 6.17346048e-01 9.84765708e-01 -3.02966714e-01 1.90096766e-01 -8.35786998e-01 5.25204003e-01 9.57827210e-01 -2.91774124e-01 5.84639013e-01 -1.42948544e+00 -6.33018553e-01 -4.43854742e-02 -4.22436088e-01 -5.79695165e-01 7.55461603e-02 -3.33930194e-01 1.49532735e-01 -1.33728838e+00 5.63768625e-01 -3.97971123e-01 -3.36408287e-01 3.13932449e-01 -5.62103748e-01 -1.20927773e-01 -3.90071690e-01 -1.44550636e-01 4.31298874e-02 5.57807684e-01 8.79052460e-01 1.85623869e-01 -2.41965055e-01 4.25865948e-01 -6.05932415e-01 7.60006130e-01 5.05559146e-01 -9.37012672e-01 -5.45708537e-01 -1.65805653e-01 3.52314413e-01 4.16490793e-01 5.07740557e-01 -4.97240841e-01 -9.89581048e-02 -4.61161286e-01 3.03319305e-01 -2.57706791e-01 8.40982124e-02 -9.83646750e-01 6.14491761e-01 3.46009970e-01 -6.12758279e-01 -5.40717319e-02 8.42796564e-02 8.87540638e-01 1.84407890e-01 -4.75973278e-01 9.32738066e-01 9.16104242e-02 1.37826383e-01 -7.02464283e-02 -1.89028099e-01 1.60113484e-01 7.16367662e-01 1.17387116e-01 -2.33231083e-01 -4.43847656e-01 -9.15772974e-01 1.74805716e-01 1.40854344e-01 1.01333879e-01 2.72722542e-01 -1.32867217e+00 -1.14576459e+00 -7.13896006e-02 -4.03783750e-04 -3.50436419e-01 2.53068775e-01 1.36611664e+00 1.69599548e-01 2.38207594e-01 1.52245834e-01 -5.74376822e-01 -9.76821601e-01 8.01386654e-01 2.40144819e-01 -1.67379573e-01 -2.63331324e-01 5.46211183e-01 4.15206134e-01 -4.56290036e-01 2.32260495e-01 -3.47990155e-01 1.87411308e-02 -1.07049108e-01 6.04160547e-01 8.29829931e-01 -1.25043347e-01 -5.40455282e-01 -2.92820722e-01 3.57748061e-01 1.44038886e-01 -2.14664340e-02 1.24972260e+00 -4.67513114e-01 -2.17182443e-01 9.28944588e-01 1.08555543e+00 -1.14665315e-01 -1.33887875e+00 -5.54397464e-01 1.72552168e-01 -4.10932124e-01 2.63448775e-01 -1.05547225e+00 -4.93665367e-01 7.12946534e-01 8.71262550e-01 9.66102406e-02 1.05542767e+00 -3.57395798e-01 4.68134396e-02 -1.52995571e-01 2.70920217e-01 -7.01871097e-01 -4.66949344e-01 -6.16627708e-02 8.25453579e-01 -1.39431226e+00 2.45106652e-01 -3.59575748e-01 -1.54512212e-01 7.59480476e-01 1.08685404e-01 -2.59461313e-01 9.31338251e-01 2.53651053e-01 -2.01737434e-01 1.01055102e-02 -5.40761530e-01 1.48454860e-01 4.97910798e-01 8.18946362e-01 6.24793172e-01 2.51738757e-01 -1.06033981e+00 6.29357934e-01 1.53358072e-01 2.73064911e-01 5.45086026e-01 5.94778776e-01 -2.33582333e-01 -9.48225796e-01 -6.04975760e-01 5.86834669e-01 -7.59339750e-01 -2.90241488e-03 2.31093690e-01 8.03523839e-01 1.34274155e-01 1.11248112e+00 -5.82411960e-02 5.65033615e-01 5.52476108e-01 8.95640254e-02 4.41421092e-01 -5.59754491e-01 -9.51708853e-02 3.86101335e-01 -5.61546125e-02 -3.77090305e-01 -6.73652709e-01 -9.84052896e-01 -7.35602856e-01 -3.42715830e-01 -6.50676191e-01 1.90395787e-01 5.98909318e-01 1.15448904e+00 9.82916281e-02 3.78949076e-01 5.22510350e-01 -3.70793998e-01 -1.15960240e+00 -1.27276433e+00 -9.03780878e-01 4.07103717e-01 5.91895700e-01 -9.68941867e-01 -4.14246410e-01 -1.28040761e-01]
[7.8248419761657715, 5.0554022789001465]
5b886e2a-de7d-4988-bc0f-81593e131f75
a-normal-form-characterization-for-efficient
2104.14098
null
https://arxiv.org/abs/2104.14098v2
https://arxiv.org/pdf/2104.14098v2.pdf
A Normal Form Characterization for Efficient Boolean Skolem Function Synthesis
Boolean Skolem function synthesis concerns synthesizing outputs as Boolean functions of inputs such that a relational specification between inputs and outputs is satisfied. This problem, also known as Boolean functional synthesis, has several applications, including design of safe controllers for autonomous systems, certified QBF solving, cryptanalysis etc. Recently, complexity theoretic hardness results have been shown for the problem, although several algorithms proposed in the literature are known to work well in practice. This dichotomy between theoretical hardness and practical efficacy has motivated the research into normal forms or representations of input specifications that permit efficient synthesis, thus explaining perhaps the efficacy of these algorithms. In this paper we go one step beyond this and ask if there exists a normal form representation that can in fact precisely characterize "efficient" synthesis. We present a normal form called SAUNF that precisely characterizes tractable synthesis in the following sense: a specification is polynomial time synthesizable iff it can be compiled to SAUNF in polynomial time. Additionally, a specification admits a polynomial-sized functional solution iff there exists a semantically equivalent polynomial-sized SAUNF representation. SAUNF is exponentially more succinct than well-established normal forms like BDDs and DNNFs, used in the context of AI problems, and strictly subsumes other more recently proposed forms like SynNNF. It enjoys compositional properties that are similar to those of DNNF. Thus, SAUNF provides the right trade-off in knowledge representation for Boolean functional synthesis.
['Supratik Chakraborty', 'S. Akshay', 'Aman Bansal', 'Preey Shah']
2021-04-29
null
null
null
null
['cryptanalysis']
['miscellaneous']
[ 4.44314837e-01 5.58660030e-01 -2.88923740e-01 -3.79427493e-01 -5.19202530e-01 -1.12997627e+00 5.40802538e-01 1.86342970e-02 2.80273616e-01 9.83291447e-01 4.26295549e-02 -7.98557699e-01 -5.02872646e-01 -1.37538826e+00 -9.65458214e-01 -6.01209164e-01 -1.31936651e-02 3.76591831e-01 3.49341482e-01 -5.63846886e-01 6.65054470e-02 7.38972604e-01 -1.80398524e+00 1.00586466e-01 5.93599617e-01 9.48261023e-01 -4.78540033e-01 6.18115425e-01 1.72519490e-01 6.74742520e-01 -5.82880497e-01 -3.37966651e-01 5.38141787e-01 -7.69708753e-01 -1.07224262e+00 -1.21131495e-01 2.59424895e-01 -1.12228148e-01 -4.93838042e-01 1.32098210e+00 -1.92912579e-01 -2.59217918e-01 2.89511621e-01 -1.88416898e+00 -3.46258640e-01 1.01284385e+00 2.89055735e-01 -4.43342090e-01 5.42970836e-01 4.44425315e-01 1.37599373e+00 2.27380678e-01 5.56121707e-01 1.54828525e+00 1.28728941e-01 8.20810199e-01 -1.70388925e+00 -6.58720553e-01 -2.00247571e-01 -1.02358855e-01 -1.37473202e+00 -3.75349551e-01 7.46934414e-02 -2.20971271e-01 1.27159274e+00 7.50909328e-01 8.29342365e-01 5.80433190e-01 6.59034550e-01 2.89252967e-01 1.29903173e+00 -4.82007444e-01 4.32884753e-01 1.68347150e-01 2.34552592e-01 7.44051695e-01 1.05346227e+00 3.57698023e-01 -2.48271674e-01 -5.91736138e-01 8.00210953e-01 -3.26191753e-01 -4.76127803e-01 -1.68279111e-01 -1.21755946e+00 8.72377753e-01 -1.93886198e-02 3.73103112e-01 2.25906298e-01 9.16487277e-01 5.20386934e-01 1.07498372e+00 -2.98936218e-01 9.46939945e-01 -3.73679966e-01 1.17375247e-01 -4.59483415e-01 7.10444391e-01 1.39896023e+00 1.11999762e+00 5.89292586e-01 3.50059450e-01 -2.57578313e-01 -4.38549697e-01 1.95637345e-01 8.74832392e-01 -1.07229948e-01 -1.21950781e+00 4.70957607e-02 7.39371657e-01 1.30814552e-01 -6.48457646e-01 -5.45276776e-02 8.02584458e-03 -5.48913300e-01 2.71794379e-01 4.07500952e-01 -4.29421365e-02 -4.98802781e-01 2.12167883e+00 1.52397547e-02 -1.86872944e-01 3.91568810e-01 6.59448266e-01 2.82150805e-01 1.00159001e+00 -5.65890372e-01 -3.83743435e-01 1.28396904e+00 -3.30596805e-01 -8.23705792e-01 1.50009200e-01 4.59349871e-01 -7.58708492e-02 7.51094103e-01 7.25312233e-01 -1.14145899e+00 5.55561520e-02 -1.48732173e+00 2.24423736e-01 -1.99144259e-01 -3.91821027e-01 1.10671639e+00 9.54359651e-01 -1.36354959e+00 2.88739264e-01 -3.77298802e-01 -5.27566299e-02 -2.55768709e-02 1.04793870e+00 -5.79900444e-01 -1.06367789e-01 -1.33898556e+00 8.65756869e-01 5.29975712e-01 -1.25039956e-02 -1.37226832e+00 -3.97467166e-01 -9.41783309e-01 2.46629670e-01 8.71300757e-01 -9.19410408e-01 1.54829228e+00 -1.02658594e+00 -1.61527777e+00 4.69091117e-01 -1.00994125e-01 -9.16764200e-01 1.60839796e-01 7.53825545e-01 -4.91232306e-01 9.29101035e-02 -3.37867975e-01 3.61685127e-01 6.32654309e-01 -8.30363512e-01 -5.30432105e-01 -3.23839098e-01 1.22632480e+00 -5.50170898e-01 1.44304022e-01 -4.52285036e-02 4.01502967e-01 1.45697489e-01 -4.61756028e-02 -8.76559198e-01 -4.06376094e-01 -7.67619535e-02 -2.80387342e-01 -4.23427194e-01 4.16263074e-01 2.21440777e-01 1.36476326e+00 -1.85106039e+00 1.80489436e-01 7.59891033e-01 4.06491786e-01 -9.55516398e-02 -1.10375714e-02 7.81572461e-01 3.56836617e-02 3.35535765e-01 -8.50236565e-02 7.88071275e-01 6.74909353e-01 4.81570393e-01 -6.09113276e-01 8.05696011e-01 4.35259491e-01 1.12829649e+00 -6.99791372e-01 -1.32397309e-01 -1.45782515e-01 -2.60096967e-01 -8.23409200e-01 1.65446405e-03 -9.45518553e-01 -1.61406934e-01 -3.19575995e-01 5.55480361e-01 5.34888268e-01 -7.34601170e-02 5.73170066e-01 3.02437320e-02 -7.42820427e-02 4.12199110e-01 -1.37312865e+00 9.84834731e-01 -2.85314411e-01 5.42433798e-01 2.38426283e-01 -7.62820363e-01 8.06730986e-01 6.61658883e-01 -1.03787765e-01 -4.77184743e-01 5.91021657e-01 4.33180094e-01 3.26306969e-01 -1.28527939e-01 2.73373008e-01 -6.28249764e-01 -5.48863411e-01 4.10048515e-01 -1.68551192e-01 -6.21263802e-01 3.87787402e-01 1.47832081e-01 1.47285509e+00 -3.24598759e-01 5.20796597e-01 -5.72798729e-01 9.31552708e-01 1.54556692e-01 6.59072220e-01 8.77816200e-01 1.60085797e-01 -7.09090978e-02 1.25175929e+00 -1.21570125e-01 -8.98193538e-01 -8.26351643e-01 2.30090335e-01 3.33768964e-01 2.14084476e-01 -9.24202800e-01 -6.71051562e-01 -2.51544654e-01 9.43921432e-02 4.83831733e-01 -3.77061933e-01 -6.58468604e-01 -4.03482854e-01 3.32594484e-01 1.17234755e+00 3.60250503e-01 1.40165314e-01 -7.46862888e-01 -7.68935680e-01 1.00593425e-01 2.90713727e-01 -8.63817215e-01 -7.86022916e-02 4.57464069e-01 -8.73596907e-01 -1.10440505e+00 3.49626929e-01 -4.44617331e-01 7.66701102e-01 8.03179368e-02 8.61682355e-01 1.20010108e-01 1.30457059e-01 2.17580900e-01 8.28145593e-02 -2.24961564e-01 -8.23601544e-01 -2.38651440e-01 1.26708940e-01 -2.50079036e-01 6.27887920e-02 -2.85964549e-01 8.28772038e-02 9.28546563e-02 -1.48877311e+00 -1.68872789e-01 3.85357827e-01 8.11139107e-01 5.55836260e-01 7.08357215e-01 5.49503684e-01 -1.06961334e+00 6.11975074e-01 -1.03035346e-01 -1.44569421e+00 3.75902504e-01 -6.53096735e-01 6.51402771e-01 1.20419550e+00 -2.81893969e-01 -4.24418062e-01 1.21955447e-01 6.02667071e-02 -1.59704596e-01 8.15232620e-02 5.02575338e-01 -7.91704416e-01 -2.43443847e-01 6.21160865e-01 1.69869393e-01 1.99368834e-01 3.84294063e-01 3.74781728e-01 4.70295072e-01 4.50318098e-01 -1.09902608e+00 1.05458796e+00 4.07085866e-01 7.39982724e-01 -2.94251472e-01 -2.12292120e-01 1.74400926e-01 2.53593534e-01 2.19619513e-01 3.20846260e-01 -5.07538974e-01 -1.59975839e+00 -8.09640512e-02 -1.09456730e+00 -1.16754383e-01 -6.94367170e-01 4.86604385e-02 -9.54749763e-01 -2.72931959e-02 -3.71740788e-01 -1.18114328e+00 -1.10719495e-01 -1.38102078e+00 7.08190143e-01 5.31545393e-02 -3.26963156e-01 -5.00595689e-01 -1.43774614e-01 -1.96133405e-01 3.55432659e-01 3.75468791e-01 1.47547126e+00 -4.80841011e-01 -9.49010193e-01 -4.07731920e-01 5.30904271e-02 4.47493225e-01 -4.97764982e-02 1.46686703e-01 -7.34157145e-01 -3.79835874e-01 -9.25024897e-02 -1.83751315e-01 2.27405980e-01 2.40923509e-01 7.15810001e-01 -8.61602306e-01 -8.05991068e-02 1.45171598e-01 1.69706964e+00 3.62804621e-01 8.07879210e-01 -9.24300402e-02 -1.23800866e-01 4.12773758e-01 4.61020350e-01 2.29840815e-01 9.88145918e-02 3.49158525e-01 5.09589195e-01 6.04557335e-01 4.18710262e-01 -2.17264488e-01 5.63261569e-01 1.55304193e-01 2.01619998e-01 -1.56003654e-01 -5.09947479e-01 3.13677698e-01 -1.63567185e+00 -1.06163847e+00 -8.70391056e-02 2.31239462e+00 1.07397556e+00 4.08517927e-01 1.45579576e-01 6.79569960e-01 4.70870703e-01 -2.98816085e-01 -1.86378151e-01 -1.09532785e+00 -2.47324958e-01 8.59550118e-01 7.59081423e-01 7.13187397e-01 -5.13900161e-01 7.95368969e-01 6.42794895e+00 7.35535502e-01 -1.15937340e+00 -5.18731365e-04 3.57499979e-02 1.97104245e-01 -9.71034348e-01 6.34953082e-01 -7.36136317e-01 -3.04048955e-02 1.57130742e+00 -7.83686936e-01 7.41306961e-01 7.40215659e-01 1.17235318e-01 -1.25409350e-01 -1.62720311e+00 5.53988278e-01 -3.72786045e-01 -1.49286699e+00 -2.88629526e-04 5.63709326e-02 7.05616772e-01 -7.79186368e-01 -6.74102902e-02 2.22365588e-01 4.15079057e-01 -1.34484613e+00 1.07668078e+00 1.31446168e-01 1.00942945e+00 -1.04885983e+00 6.91895127e-01 2.46335015e-01 -1.04037297e+00 -3.78021717e-01 -3.23925376e-01 -4.76959169e-01 -1.83590204e-01 2.94554830e-01 -7.64629543e-01 8.03460717e-01 8.96671508e-03 1.52002007e-01 -1.45135954e-01 9.27935481e-01 -2.63988107e-01 3.86512965e-01 -4.85543817e-01 -3.29235137e-01 3.75235826e-01 -1.56168282e-01 3.97439450e-01 8.60142708e-01 2.76994467e-01 2.92926401e-01 -1.98297560e-01 1.13313794e+00 5.39231561e-02 -4.44886923e-01 -9.02678311e-01 -6.56250298e-01 5.42257309e-01 8.56518209e-01 -5.17721236e-01 -3.53114694e-01 -1.37440339e-01 3.17920297e-01 -3.22682232e-01 1.38273537e-01 -1.02449930e+00 -5.65957665e-01 8.85195792e-01 1.92142427e-01 8.60819444e-02 -1.86727643e-01 -1.76339969e-01 -9.64915037e-01 -5.32836802e-02 -1.48519695e+00 2.21062154e-01 -4.85900313e-01 -9.31766212e-01 5.49167871e-01 3.26479882e-01 -1.05307031e+00 -3.27562064e-01 -5.59271932e-01 -1.39589116e-01 7.26244032e-01 -1.08182716e+00 -7.47243762e-01 1.43737018e-01 6.72399819e-01 -1.06214702e-01 2.88669974e-01 1.00026023e+00 -5.77249229e-02 -1.79979637e-01 5.29070020e-01 -4.57095385e-01 -5.92387319e-01 1.93463668e-01 -1.18653011e+00 2.57907193e-02 1.01750684e+00 -2.32560053e-01 9.65498686e-01 9.94223475e-01 -3.86955857e-01 -2.56253576e+00 -1.00896192e+00 1.02101481e+00 -4.80957255e-02 7.47407556e-01 -3.92159611e-01 -2.16582224e-01 7.84606516e-01 9.05043483e-02 -1.59246251e-01 1.35015354e-01 -5.02118826e-01 -7.00270653e-01 -5.76828063e-01 -1.32586408e+00 7.63311565e-01 9.01304901e-01 -7.88058519e-01 -4.85950798e-01 1.83670789e-01 1.15523183e+00 -3.36313754e-01 -6.45885527e-01 3.65469456e-01 6.44393325e-01 -7.90102243e-01 4.37967807e-01 -7.41863787e-01 1.17604651e-01 -9.47099507e-01 -4.68973041e-01 -7.96743691e-01 -1.44408152e-01 -1.26331353e+00 -2.49416798e-01 7.21553504e-01 3.65464598e-01 -1.09013760e+00 3.82888943e-01 5.99013805e-01 -4.69871201e-02 -7.83059299e-01 -9.66261029e-01 -1.13959217e+00 8.92505720e-02 -4.31425720e-01 1.06247151e+00 3.97555023e-01 5.73422313e-01 4.10074204e-01 -2.14556549e-02 1.96090743e-01 9.38402563e-02 4.44703072e-01 6.53540432e-01 -1.28037238e+00 -5.50893307e-01 -5.80716908e-01 -7.06781566e-01 -8.24564278e-01 5.49992859e-01 -1.10103548e+00 7.63493329e-02 -1.41598463e+00 -1.12466276e-01 -2.34924212e-01 1.80772528e-01 7.81165481e-01 7.35660136e-01 -5.20761535e-02 8.36728364e-02 -3.82807791e-01 -2.99140185e-01 5.18126860e-02 1.22793448e+00 -4.72378999e-01 2.94828713e-02 1.29557818e-01 -9.94768798e-01 3.18650343e-02 5.91105640e-01 -3.17040950e-01 -6.80073142e-01 2.20319301e-01 7.79185295e-01 5.57748079e-01 2.64587522e-01 -8.49435449e-01 1.64315939e-01 -7.67131925e-01 -6.72218740e-01 -8.73101801e-02 7.57188275e-02 -1.30857599e+00 9.96495545e-01 9.02478039e-01 -3.68325710e-01 1.89421952e-01 -3.51863541e-02 2.92392075e-01 -2.02566251e-01 -4.29488689e-01 6.22288883e-01 -1.15468381e-02 -3.31038296e-01 1.73337664e-02 -9.45675731e-01 -1.71404690e-01 1.18152809e+00 -2.57965446e-01 -5.76750398e-01 -5.65659761e-01 -2.73599744e-01 1.22493729e-01 6.48148179e-01 -1.87838510e-01 6.73311472e-01 -1.04928887e+00 -2.08894715e-01 3.27795267e-01 9.25789550e-02 -1.17863320e-01 -2.54367471e-01 9.50797796e-01 -6.59491479e-01 9.78151560e-01 -2.24909022e-01 -1.52376741e-01 -1.04319429e+00 5.35041630e-01 4.74756271e-01 -7.91905671e-02 -2.82925129e-01 4.34632063e-01 1.86911419e-01 -2.00368568e-01 2.61071473e-01 -1.19587171e+00 6.16271973e-01 -6.01971745e-01 5.38588524e-01 2.33885795e-01 7.60854036e-02 -5.03544062e-02 -5.75594008e-01 8.08900744e-02 5.29717505e-01 -2.90555209e-01 9.64871466e-01 2.62482762e-01 -9.70554888e-01 1.70424163e-01 8.80630672e-01 4.19034883e-02 -3.15094560e-01 8.97547156e-02 -1.34421717e-02 -1.67703643e-01 -2.11397275e-01 -5.62563717e-01 -8.35762024e-01 4.30996388e-01 -5.77360243e-02 5.66858351e-01 1.38467777e+00 4.34389226e-02 6.58147991e-01 1.00663280e+00 9.29388821e-01 -5.55292845e-01 -4.91546929e-01 6.91712916e-01 5.22063851e-01 -2.57160127e-01 -1.56525508e-01 -5.72737515e-01 -1.59014270e-01 1.29919219e+00 4.05981570e-01 -3.74830097e-01 -4.26766016e-02 9.99585569e-01 -6.88084781e-01 -7.38961399e-02 -1.32356405e+00 -3.06125820e-01 -9.79307815e-02 3.45523745e-01 -1.87782459e-02 2.73875922e-01 -7.15793192e-01 8.16675901e-01 -4.46235061e-01 2.83330768e-01 8.67275059e-01 1.04081643e+00 -6.34032547e-01 -1.45454800e+00 -3.91921729e-01 4.24322724e-01 -3.41151595e-01 1.12882063e-01 -5.26665032e-01 7.05880582e-01 4.94439006e-02 1.08179688e+00 -3.21709424e-01 -5.70878685e-01 3.11101973e-01 -1.36435196e-01 1.21067810e+00 -7.11219132e-01 -9.04694974e-01 -1.85663566e-01 5.03269434e-01 -7.10898936e-01 -2.01362953e-01 -4.74985875e-02 -1.65501392e+00 -7.93521285e-01 -6.34965122e-01 4.86416429e-01 2.09230155e-01 9.09866929e-01 -4.68276218e-02 2.98437029e-01 5.22431612e-01 -6.01740964e-02 -9.51302230e-01 -1.80794001e-01 -7.12044120e-01 -4.50925678e-01 3.88558716e-01 -2.99140036e-01 -4.28388089e-01 -1.49338216e-01]
[8.667452812194824, 6.837119102478027]
47f5462b-0922-4f47-976d-d4a75d6cd8e7
infinite-photorealistic-worlds-using-1
2306.0931
null
https://arxiv.org/abs/2306.09310v2
https://arxiv.org/pdf/2306.09310v2.pdf
Infinite Photorealistic Worlds using Procedural Generation
We introduce Infinigen, a procedural generator of photorealistic 3D scenes of the natural world. Infinigen is entirely procedural: every asset, from shape to texture, is generated from scratch via randomized mathematical rules, using no external source and allowing infinite variation and composition. Infinigen offers broad coverage of objects and scenes in the natural world including plants, animals, terrains, and natural phenomena such as fire, cloud, rain, and snow. Infinigen can be used to generate unlimited, diverse training data for a wide range of computer vision tasks including object detection, semantic segmentation, optical flow, and 3D reconstruction. We expect Infinigen to be a useful resource for computer vision research and beyond. Please visit https://infinigen.org for videos, code and pre-generated data.
['Jia Deng', 'Kaiyu Yang', 'Ankit Goyal', 'Hei Law', 'Alejandro Newell', 'Yihan Wang', 'Beining Han', 'Hongyu Wen', 'Karhan Kayan', 'Yiming Zuo', 'Mingzhe Wang', 'Lingjie Mei', 'Zeyu Ma', 'Lahav Lipson', 'Alexander Raistrick']
2023-06-15
infinite-photorealistic-worlds-using
http://openaccess.thecvf.com//content/CVPR2023/html/Raistrick_Infinite_Photorealistic_Worlds_Using_Procedural_Generation_CVPR_2023_paper.html
http://openaccess.thecvf.com//content/CVPR2023/papers/Raistrick_Infinite_Photorealistic_Worlds_Using_Procedural_Generation_CVPR_2023_paper.pdf
cvpr-2023-1
['3d-reconstruction']
['computer-vision']
[ 1.36502132e-01 -7.91233033e-02 2.11929366e-01 1.05268009e-01 1.81602344e-01 -9.61140037e-01 6.63528681e-01 -1.89061582e-01 6.96652234e-02 5.91612995e-01 -2.13607743e-01 -4.36649323e-01 2.39952445e-01 -1.20831716e+00 -3.56822580e-01 -3.70339483e-01 -1.02232501e-01 3.53516698e-01 4.54316974e-01 -3.26726317e-01 3.22401613e-01 8.99720728e-01 -1.48388958e+00 2.67611304e-03 8.16208541e-01 6.62540555e-01 3.68748039e-01 9.59071994e-01 -1.87600553e-01 8.30603778e-01 -3.58269811e-01 -5.37449956e-01 5.81814170e-01 -4.61820781e-01 -7.12301672e-01 1.71045542e-01 3.75259906e-01 -1.17780216e-01 -3.66603822e-01 8.87501657e-01 2.10536480e-01 1.60126761e-01 7.29898870e-01 -1.13529253e+00 -1.05333006e+00 -6.44321665e-02 -6.08408809e-01 1.20003380e-01 2.63477951e-01 7.30752349e-01 6.54194474e-01 -8.99671793e-01 1.04714262e+00 1.23328316e+00 5.57569742e-01 4.89400893e-01 -1.31296468e+00 -4.09397066e-01 -2.50714660e-01 -3.27116191e-01 -1.18319190e+00 -3.63324016e-01 5.62690735e-01 -7.30975688e-01 5.96668780e-01 5.69416463e-01 1.13943791e+00 8.22170496e-01 1.57300085e-01 6.68724000e-01 1.04137206e+00 -4.40627217e-01 4.46566164e-01 -2.30387107e-01 -4.63114053e-01 1.01858187e+00 -2.84547340e-02 3.21857393e-01 -3.46193045e-01 -1.17770866e-01 1.43721437e+00 -2.78216183e-01 -1.99937016e-01 -5.68765879e-01 -1.38842928e+00 5.69376469e-01 4.00608718e-01 -2.32659191e-01 -1.09750077e-01 2.11675808e-01 5.61375394e-02 4.14807349e-02 4.19784784e-01 7.11748302e-01 -2.75885701e-01 -1.62474096e-01 -5.18693149e-01 3.20118815e-01 8.76610100e-01 1.24034548e+00 8.05656612e-01 3.71040970e-01 2.87877917e-01 8.51296604e-01 -8.43724236e-02 8.00949395e-01 9.66959968e-02 -1.77735054e+00 4.64847721e-02 4.69521224e-01 1.66642666e-01 -1.14643812e+00 -2.59145349e-01 -1.60273239e-01 -8.12496960e-01 6.17729127e-01 3.68837655e-01 -2.02955410e-01 -1.07268751e+00 1.27023447e+00 6.11981213e-01 2.13640630e-01 -3.15581441e-01 7.27455378e-01 1.14267218e+00 6.87231481e-01 3.61950733e-02 2.84496218e-01 1.15258765e+00 -7.67155051e-01 -2.56132260e-02 -3.47502708e-01 2.34714180e-01 -8.55540335e-01 1.20014846e+00 2.30112329e-01 -1.32998097e+00 -2.28190735e-01 -6.19794488e-01 -2.89344847e-01 -4.08325106e-01 -1.97571978e-01 1.09623170e+00 4.67374265e-01 -1.01887250e+00 5.92926025e-01 -6.85284436e-01 -4.57312703e-01 7.96480179e-01 -1.14138000e-01 -3.18807662e-01 1.13827035e-01 -7.15854824e-01 7.95668244e-01 2.39219144e-01 1.85901999e-01 -6.92808807e-01 -9.49968338e-01 -9.90351915e-01 -2.09148690e-01 1.84466347e-01 -1.19884515e+00 1.31315100e+00 -8.44841182e-01 -1.45552623e+00 1.52190185e+00 -6.57970309e-02 -2.37612545e-01 6.82761133e-01 2.94894110e-02 3.43525745e-02 1.66446522e-01 9.61163864e-02 8.66059959e-01 6.87135458e-01 -1.29890001e+00 -3.84709597e-01 -1.75183177e-01 1.75004929e-01 2.65105188e-01 3.43639851e-01 -2.47723889e-02 -3.54645431e-01 -5.57382643e-01 8.69795680e-02 -8.60934138e-01 -4.61298943e-01 5.66992044e-01 -4.45891112e-01 2.81189144e-01 7.44788647e-01 -4.63303655e-01 3.49614561e-01 -2.09151745e+00 -7.64831156e-02 8.03688169e-02 2.21438959e-01 2.41593555e-01 -2.26058871e-01 2.15307102e-01 2.48325214e-01 4.15855557e-01 -5.29462159e-01 6.15270175e-02 -1.68792084e-01 -1.49228042e-02 -3.17098826e-01 4.53707010e-01 1.57346413e-01 1.18015540e+00 -1.06173337e+00 -5.42397976e-01 6.38070405e-01 3.96683723e-01 -4.05899316e-01 -3.86612229e-02 -4.76528972e-01 6.31089270e-01 -4.42403436e-01 7.40858316e-01 7.58960485e-01 -2.21193194e-01 -3.45407188e-01 2.85873950e-01 -2.51083970e-01 5.99559955e-02 -9.52572644e-01 1.60575795e+00 -5.44200301e-01 8.47378135e-01 2.32894763e-01 -3.84386420e-01 9.97049570e-01 -1.56986043e-01 4.49870795e-01 -6.84163272e-01 7.89831113e-03 1.99878234e-02 -3.21053445e-01 -4.27885264e-01 5.01717031e-01 -1.21490218e-01 9.75793377e-02 4.41867203e-01 -2.73724049e-01 -1.25808907e+00 3.75894487e-01 3.33734870e-01 1.14104486e+00 4.44070458e-01 2.87961245e-01 -4.18776482e-01 1.44288927e-01 5.60508370e-01 4.70975339e-01 6.48920715e-01 -8.63041133e-02 8.55090201e-01 2.39989385e-01 -7.06833065e-01 -1.36562371e+00 -1.68605363e+00 -2.16814086e-01 6.64385319e-01 3.56765807e-01 -3.13284062e-02 -4.89318371e-01 1.36703560e-02 4.72099148e-02 4.56998855e-01 -5.41753411e-01 8.58828947e-02 -5.52730918e-01 -3.21497411e-01 4.02738184e-01 6.62796050e-02 8.14119220e-01 -1.62800014e+00 -8.89737368e-01 -7.98623785e-02 -5.75098470e-02 -1.25606644e+00 -4.02528167e-01 -1.40504077e-01 -8.86592805e-01 -1.11006439e+00 -4.79467779e-01 -7.36510515e-01 6.09673381e-01 5.29217720e-01 1.65210843e+00 3.15610617e-01 -9.37469661e-01 4.96274769e-01 -9.14306492e-02 -5.45428693e-01 -5.47899306e-01 -1.46313652e-01 -5.03398776e-01 -3.36091936e-01 -1.77310929e-01 -7.68728852e-01 -5.71209013e-01 3.07163179e-01 -7.74803698e-01 4.71715063e-01 -2.30759270e-02 3.84269059e-01 7.70265818e-01 -1.33412227e-01 -9.99806598e-02 -9.87033069e-01 3.34124625e-01 -2.69959003e-01 -7.96813965e-01 -1.39516853e-02 2.72299409e-01 -4.34901297e-01 6.57583594e-01 -1.00284591e-02 -1.04879868e+00 5.73527440e-02 1.95845127e-01 -1.56456858e-01 -5.00977397e-01 1.00885719e-01 2.33027153e-02 -7.99290761e-02 8.98455381e-01 3.29551309e-01 -1.78686514e-01 -1.80216536e-01 6.43811524e-01 3.21581960e-01 8.85475636e-01 -6.54395640e-01 9.77246225e-01 1.01163960e+00 2.16439664e-01 -1.37252617e+00 -7.24641025e-01 -1.32596835e-01 -7.62217939e-01 -2.95484155e-01 7.75122583e-01 -7.40640104e-01 -5.89502990e-01 7.41059482e-01 -1.21082568e+00 -9.57798421e-01 -7.27888346e-01 1.36192396e-01 -7.48014390e-01 2.46188883e-02 -6.08358800e-01 -6.30890012e-01 -2.58926421e-01 -6.82964206e-01 1.04815781e+00 5.98231196e-01 -3.55828136e-01 -1.20280349e+00 1.33222908e-01 1.84822395e-01 3.22573215e-01 7.12200046e-01 8.79094839e-01 3.83807391e-01 -1.03192675e+00 1.15709856e-01 -2.25075185e-01 9.28064585e-02 4.26318377e-01 8.34393144e-01 -7.46593833e-01 1.60134733e-01 -2.65989482e-01 -2.58893043e-01 7.43586123e-01 5.20333588e-01 1.29146302e+00 -1.11048028e-01 -2.33204648e-01 9.85864162e-01 1.32312429e+00 3.17660362e-01 8.97302032e-01 3.80152822e-01 6.88230515e-01 6.28181100e-01 2.74383485e-01 4.13666606e-01 3.10992271e-01 2.61582822e-01 5.39014041e-01 -3.00278693e-01 -2.47035116e-01 -2.13083386e-01 -2.45163552e-02 2.96184361e-01 -2.41075829e-01 -1.41840339e-01 -1.02048242e+00 5.92887163e-01 -1.31843042e+00 -1.50632107e+00 -3.87522012e-01 2.07513547e+00 5.17198801e-01 -1.30380034e-01 4.04641293e-02 -1.60865337e-01 6.80803597e-01 4.49028276e-02 -5.55720747e-01 -7.03633368e-01 -5.20999730e-01 4.55589294e-01 4.23156023e-01 4.84282166e-01 -1.09164977e+00 1.27441967e+00 7.43835878e+00 6.32419348e-01 -1.21948266e+00 -2.97047257e-01 7.07718253e-01 -1.47368852e-02 -6.97568417e-01 2.07925186e-01 -2.81105042e-01 2.39128023e-01 3.30479831e-01 -3.60621274e-01 6.93095088e-01 5.33836603e-01 5.41318715e-01 -3.63053411e-01 -6.16560459e-01 8.07551086e-01 -3.72886002e-01 -1.78651500e+00 -5.13046160e-02 6.20971508e-02 9.02860641e-01 1.16788581e-01 -6.16269410e-02 -2.17283562e-01 1.07002091e+00 -1.14970136e+00 9.10913467e-01 4.62368667e-01 6.79474294e-01 -4.52240884e-01 -1.34740517e-01 3.64553869e-01 -1.06202269e+00 3.68756533e-01 -3.77983809e-01 -1.90606162e-01 2.31787831e-01 7.49946594e-01 -4.41344827e-01 1.83328673e-01 7.81127810e-01 7.90174961e-01 -5.17719209e-01 1.36395121e+00 -1.90035924e-01 4.54314530e-01 -4.93369967e-01 2.15592496e-02 5.69149060e-03 -6.66252196e-01 6.78724349e-01 1.03622854e+00 1.51999608e-01 2.04241574e-01 -6.11921586e-03 1.12051773e+00 -3.76505330e-02 -4.42673266e-03 -1.06607342e+00 9.53449309e-02 5.76937616e-01 1.23618329e+00 -1.28421938e+00 -7.07905218e-02 -3.47394720e-02 1.04805291e+00 1.28252506e-01 3.89823228e-01 -9.15056407e-01 -4.23085153e-01 6.96993411e-01 2.30287805e-01 1.43273994e-01 -8.23489904e-01 -7.29205370e-01 -1.03197718e+00 -1.85406446e-01 -5.06062269e-01 4.69691493e-02 -1.35684764e+00 -9.82058227e-01 4.12879616e-01 -1.52666822e-01 -1.05361688e+00 1.22663416e-01 -6.87070847e-01 -7.38591135e-01 6.06562197e-01 -1.09494090e+00 -1.04416358e+00 -6.08436406e-01 5.31767726e-01 4.22009140e-01 2.22046673e-02 7.52002835e-01 -2.40005851e-01 -2.85364777e-01 -2.74869651e-01 4.64060903e-02 2.36253291e-01 1.87162440e-02 -1.12634969e+00 9.20587778e-01 7.63629019e-01 3.47628474e-01 3.40046555e-01 5.70777893e-01 -5.01347959e-01 -1.24990392e+00 -1.10014319e+00 2.87610650e-01 -5.70161581e-01 6.18329525e-01 -4.97706085e-01 -4.12783384e-01 6.38845563e-01 4.25357297e-02 3.10182780e-01 3.44490290e-01 -3.24056596e-01 -2.15806201e-01 1.16460055e-01 -1.25377142e+00 1.13510442e+00 1.49988711e+00 -4.12532330e-01 -8.01684558e-02 3.62801284e-01 5.70042431e-01 -7.53618300e-01 -5.69540977e-01 2.02794164e-01 6.70910478e-01 -1.28021848e+00 1.10578561e+00 -4.22978818e-01 7.84563541e-01 -2.55523741e-01 -5.58050051e-02 -1.36398566e+00 -3.98801982e-01 -8.72283399e-01 1.27013758e-01 8.11739683e-01 4.05452460e-01 -6.82409108e-01 9.61924314e-01 7.13787198e-01 -1.31896943e-01 -3.39148283e-01 -6.66169047e-01 -8.65829289e-01 3.26706797e-01 -5.39847791e-01 6.77250147e-01 7.10255027e-01 -5.06803334e-01 1.55094787e-01 -1.15662880e-01 -2.32802048e-01 6.95129693e-01 5.78282952e-01 1.06996477e+00 -1.38439488e+00 -8.29559844e-03 -6.99671924e-01 -4.05555099e-01 -9.35215056e-01 -4.07045484e-02 -9.44256842e-01 -1.39606878e-01 -1.79725266e+00 -7.17166364e-02 -9.09833729e-01 5.59976161e-01 3.62523347e-01 1.13074787e-01 6.62794352e-01 4.00133997e-01 3.66114736e-01 -1.58080220e-01 5.56200862e-01 2.03498745e+00 -2.21773878e-01 -2.81110406e-01 -8.10028836e-02 -8.21147799e-01 9.31464016e-01 1.10770619e+00 -3.20867360e-01 -3.33326072e-01 -6.98798835e-01 1.25451922e-01 -1.06795609e-01 8.22594464e-01 -8.71696949e-01 -3.21881264e-01 -8.38853061e-01 4.28754538e-01 -1.78395912e-01 4.84115213e-01 -4.89283115e-01 4.09525990e-01 2.70451725e-01 -7.92191848e-02 -2.83171125e-02 3.49631757e-01 1.32316679e-01 1.75201505e-01 -7.51846135e-02 9.76213217e-01 -8.22948515e-01 -8.91792595e-01 5.00027418e-01 -4.11670536e-01 5.81773102e-01 1.19304550e+00 -5.76204300e-01 -5.93450665e-01 -3.66002858e-01 -6.73496842e-01 1.91582162e-02 1.06132424e+00 3.83086175e-01 7.11822391e-01 -1.07026732e+00 -8.12384427e-01 2.40207762e-01 -1.75126255e-01 4.06454861e-01 1.26093537e-01 2.91810125e-01 -1.35213542e+00 1.59369633e-02 -4.13928270e-01 -5.34504354e-01 -8.46470475e-01 2.31502771e-01 5.29820263e-01 3.12911928e-01 -8.62939894e-01 7.93456078e-01 4.75702971e-01 -5.87376177e-01 -4.66949791e-01 -2.05709860e-01 3.75890017e-01 -5.79527736e-01 3.23564708e-01 4.68120724e-01 -2.04243183e-01 -5.91881931e-01 -2.56018281e-01 6.56212449e-01 7.30041623e-01 -5.86406849e-02 1.33549547e+00 -1.70868084e-01 -3.30751508e-01 3.82837504e-01 6.83517814e-01 2.73565263e-01 -1.40099156e+00 2.61194199e-01 -4.62454438e-01 -7.08129406e-01 -2.07783833e-01 -5.74188888e-01 -1.18590760e+00 8.30913484e-01 5.03513888e-02 4.54880089e-01 8.59534740e-01 8.21614787e-02 5.83395004e-01 9.66370925e-02 6.95877731e-01 -6.79550469e-01 -2.71022152e-02 6.57114446e-01 1.06394124e+00 -8.09117556e-01 5.78019992e-02 -6.06036663e-01 -4.89637017e-01 9.62502778e-01 5.05380154e-01 -3.25323373e-01 8.16720068e-01 6.28610611e-01 1.33512944e-01 -2.37777054e-01 -5.35210669e-01 -2.29275107e-01 -1.96761549e-01 1.14065683e+00 2.86477655e-01 2.65798301e-01 1.61129192e-01 -3.03949893e-01 -6.61224484e-01 3.31982076e-02 8.66276324e-01 9.01929379e-01 -4.04246092e-01 -8.18351567e-01 -4.51503426e-01 5.20785630e-01 -8.63455534e-02 -2.45299980e-01 -4.23893958e-01 6.38673127e-01 2.11828724e-01 6.69793010e-01 2.49271140e-01 5.25478795e-02 1.73452035e-01 -3.72721881e-01 7.34760523e-01 -6.12997293e-01 -2.18665227e-01 -4.64529634e-01 6.77596359e-03 -4.49370384e-01 -3.81394535e-01 -8.02743673e-01 -1.24431002e+00 -6.80081964e-01 3.96462679e-02 -1.57343924e-01 5.17452300e-01 4.23394769e-01 4.72493291e-01 7.47911409e-02 4.95833397e-01 -1.12774599e+00 3.15091163e-01 -3.80626678e-01 -7.50436425e-01 3.31420630e-01 8.47573206e-02 -5.49272776e-01 -3.48902732e-01 4.88524973e-01]
[9.240809440612793, -2.9483330249786377]
78ef3b20-a42d-4f2a-ab13-1c7c8833c276
nl2cmd-an-updated-workflow-for-natural
2302.07845
null
https://arxiv.org/abs/2302.07845v3
https://arxiv.org/pdf/2302.07845v3.pdf
NL2CMD: An Updated Workflow for Natural Language to Bash Commands Translation
Translating natural language into Bash Commands is an emerging research field that has gained attention in recent years. Most efforts have focused on producing more accurate translation models. To the best of our knowledge, only two datasets are available, with one based on the other. Both datasets involve scraping through known data sources (through platforms like stack overflow, crowdsourcing, etc.) and hiring experts to validate and correct either the English text or Bash Commands. This paper provides two contributions to research on synthesizing Bash Commands from scratch. First, we describe a state-of-the-art translation model used to generate Bash Commands from the corresponding English text. Second, we introduce a new NL2CMD dataset that is automatically generated, involves minimal human intervention, and is over six times larger than prior datasets. Since the generation pipeline does not rely on existing Bash Commands, the distribution and types of commands can be custom adjusted. We evaluate the performance of ChatGPT on this task and discuss the potential of using it as a data generator. Our empirical results show how the scale and diversity of our dataset can offer unique opportunities for semantic parsing researchers.
['Douglas C. Schmidt', 'Jules White', 'Marco Georgaklis', 'Zhongwei Teng', 'Quchen Fu']
2023-02-15
null
null
null
null
['code-translation', 'semantic-parsing']
['computer-code', 'natural-language-processing']
[ 2.01289326e-01 4.68518764e-01 -3.73111032e-02 -7.05433130e-01 -1.33187485e+00 -1.01643133e+00 6.06045902e-01 3.72091644e-02 -2.96303719e-01 8.71551752e-01 3.06665748e-01 -6.76781774e-01 3.90549034e-01 -6.92198038e-01 -7.54737139e-01 1.32234856e-01 6.16390407e-01 9.99795616e-01 4.12849516e-01 -5.62837541e-01 2.41474569e-01 4.70492542e-02 -1.33202398e+00 4.73180145e-01 1.01020038e+00 2.27764070e-01 4.21100855e-01 7.28534818e-01 -6.26647353e-01 6.54564023e-01 -1.04873979e+00 -7.41006076e-01 3.41482311e-01 -7.67562747e-01 -1.18166459e+00 -4.47441041e-01 3.68003249e-01 -2.01584354e-01 2.53503919e-01 8.86515141e-01 5.68626642e-01 -2.88835198e-01 5.22580035e-02 -1.35555768e+00 -7.87609935e-01 1.16801810e+00 1.44361168e-01 1.04673905e-04 7.47635424e-01 2.99636722e-01 1.06099224e+00 -5.40077269e-01 1.14913607e+00 1.02368689e+00 6.79525912e-01 9.16769981e-01 -1.19972265e+00 -5.49177945e-01 -2.11292624e-01 -2.44072422e-01 -1.25280547e+00 -5.34350574e-01 2.90503621e-01 -4.42615092e-01 1.37215137e+00 3.31596643e-01 3.05597931e-01 1.30606055e+00 -2.51092821e-01 6.31641090e-01 1.16495037e+00 -8.43332708e-01 1.17603615e-01 2.76032239e-01 4.75401767e-02 5.49715400e-01 1.27313316e-01 -2.65962332e-01 -5.58260918e-01 -2.14478970e-01 5.42559266e-01 -5.79265356e-01 -9.96722281e-02 3.63673270e-02 -1.41713417e+00 7.60981500e-01 -2.10446671e-01 3.26196820e-01 4.64938208e-02 -1.75107038e-03 4.14138108e-01 4.13155168e-01 3.35659504e-01 8.92787993e-01 -8.41533601e-01 -8.17255199e-01 -9.48318303e-01 6.67214036e-01 1.26242483e+00 1.63963795e+00 7.00268745e-01 -2.67336607e-01 -4.33374234e-02 7.21700311e-01 1.81478057e-02 4.91757184e-01 6.09176099e-01 -9.13446903e-01 1.05464005e+00 6.11771524e-01 3.15660924e-01 -5.62228322e-01 -2.09237084e-01 1.71186849e-01 1.82234675e-01 -1.82704870e-02 6.63942337e-01 -4.72388536e-01 -5.81780314e-01 1.56792295e+00 2.10933223e-01 -4.59122866e-01 -2.57757306e-02 7.77262747e-01 7.14970291e-01 5.02289593e-01 3.38610215e-03 3.07230800e-01 1.13830149e+00 -8.92043948e-01 -6.53759301e-01 -3.20391089e-01 9.56061423e-01 -1.16354680e+00 1.35968685e+00 1.83917835e-01 -1.14601743e+00 -2.48664573e-01 -6.87893093e-01 -3.34175140e-01 -4.81167793e-01 -4.86385496e-03 7.09729433e-01 7.02201962e-01 -1.17559648e+00 5.21218538e-01 -9.99479711e-01 -8.38393331e-01 1.03030868e-01 -2.44088545e-02 -1.01706773e-01 -9.48012173e-02 -1.19597161e+00 9.91885483e-01 4.16302353e-01 -4.18226570e-01 -3.32707793e-01 -7.77850568e-01 -9.03739154e-01 -3.66629839e-01 4.41923171e-01 -7.65950322e-01 2.04686189e+00 -8.31994236e-01 -1.66286218e+00 8.39509606e-01 -3.77856791e-01 -2.99976587e-01 7.85767794e-01 -2.59777486e-01 -3.28233659e-01 -2.25758299e-01 5.19225240e-01 7.39683688e-01 3.90107185e-01 -7.63040900e-01 -7.15119898e-01 -3.42819899e-01 3.77566889e-02 4.99326363e-02 1.59360409e-01 7.83711374e-01 -3.75027359e-01 -7.22317278e-01 -2.35716388e-01 -9.80082691e-01 -5.84397912e-02 -3.62345845e-01 -5.35465777e-01 -1.63430661e-01 4.92071301e-01 -8.99071932e-01 1.14607430e+00 -1.76505017e+00 7.00125098e-02 -1.85402960e-01 -2.65944690e-01 1.61388710e-01 -1.88040778e-01 9.67170477e-01 2.27422953e-01 6.07113242e-01 -3.66031408e-01 -2.45654240e-01 1.98315710e-01 3.44889402e-01 -5.37320375e-01 -8.45349357e-02 5.17931044e-01 1.11804521e+00 -9.22687471e-01 -3.94768775e-01 -1.45803288e-01 2.07576901e-01 -6.68351352e-01 2.99912959e-01 -6.26483440e-01 4.40805405e-01 -3.81922722e-01 5.64394176e-01 3.70485693e-01 5.66380769e-02 1.92294851e-01 4.26302016e-01 -4.97505963e-01 1.12684155e+00 -1.08810365e+00 2.06276107e+00 -4.94734555e-01 6.13617897e-01 -1.98337939e-02 -5.10327876e-01 8.97220969e-01 3.45409960e-01 -1.15569606e-01 -5.01642823e-01 -8.82118195e-02 6.94772303e-01 -1.19272575e-01 -5.81956029e-01 7.31348991e-01 -3.56186852e-02 -3.74137282e-01 7.34748244e-01 3.21107097e-02 -5.97590327e-01 4.08725858e-01 2.02435642e-01 1.20301735e+00 5.61298251e-01 2.51082420e-01 -7.05249608e-02 -4.13543507e-02 8.36266041e-01 5.09905815e-01 8.98827314e-01 1.44145802e-01 7.67351210e-01 4.93451685e-01 -2.26657242e-01 -1.25893223e+00 -6.90568566e-01 5.63652664e-02 1.27254391e+00 -2.44616657e-01 -5.87995410e-01 -1.21202540e+00 -7.10675597e-01 -1.66466549e-01 1.21169937e+00 -2.35321790e-01 3.61099303e-01 -9.99667466e-01 -3.58941555e-01 1.18185294e+00 6.73048377e-01 3.64828616e-01 -1.08864427e+00 -6.14905417e-01 3.97778392e-01 -4.90443349e-01 -1.41233504e+00 -3.68300200e-01 -2.43216027e-02 -7.41230905e-01 -9.22514260e-01 -2.68177480e-01 -8.05938840e-01 5.21439433e-01 -8.34620297e-02 1.53016675e+00 1.37609601e-01 -1.79944858e-01 2.22732723e-01 -7.15674579e-01 -6.49017096e-01 -1.04088140e+00 5.10851800e-01 -3.93626243e-01 -7.25910664e-01 6.05925739e-01 -2.68944681e-01 8.33890885e-02 2.38424063e-01 -9.08127069e-01 4.38869834e-01 2.99876750e-01 4.52270240e-01 2.20808789e-01 -4.22696888e-01 4.04535800e-01 -1.50382423e+00 7.47949183e-01 -3.90598327e-01 -6.94043159e-01 8.87332633e-02 -4.56651747e-01 4.09686193e-02 8.42838407e-01 6.93371473e-03 -1.08832550e+00 9.18397531e-02 -3.12217832e-01 2.47101933e-01 -5.39825380e-01 4.51546043e-01 -2.37410977e-01 2.92424768e-01 6.67404294e-01 9.17287022e-02 -1.59867510e-01 -7.74009764e-01 6.28200591e-01 9.44200218e-01 5.35889447e-01 -8.25690567e-01 8.12402248e-01 7.92789645e-03 -6.17050052e-01 -5.55265069e-01 -4.86658543e-01 -6.28976896e-02 -6.48075223e-01 3.95787895e-01 8.10050488e-01 -8.06406677e-01 5.39677031e-02 5.37645936e-01 -1.50428748e+00 -6.98170841e-01 -1.56018004e-01 1.09987758e-01 -5.07013679e-01 1.68747172e-01 -7.44239807e-01 -2.70054877e-01 -4.14714277e-01 -1.26328290e+00 1.27218473e+00 -3.88552509e-02 -9.33630288e-01 -8.23616564e-01 1.11159422e-01 6.83407903e-01 6.85047209e-01 6.35125712e-02 1.04593992e+00 -1.01996398e+00 -6.89880133e-01 -1.74915001e-01 -6.43899962e-02 -4.60056253e-02 5.56841157e-02 3.75977278e-01 -6.48330212e-01 8.31924975e-02 -2.96872705e-01 -3.06622177e-01 -3.64418216e-02 -1.94559589e-01 7.74140954e-01 -4.67696488e-01 -1.70933798e-01 4.44744140e-01 1.13819873e+00 1.12593427e-01 5.19751549e-01 5.65915465e-01 7.22157776e-01 6.34850025e-01 5.21381617e-01 2.47379631e-01 8.13504100e-01 8.08138609e-01 8.79202485e-02 2.41933703e-01 -1.90387234e-01 -6.18171453e-01 3.00773025e-01 9.58244503e-01 3.34667444e-01 -2.15430677e-01 -1.19610810e+00 6.32831693e-01 -1.72013009e+00 -5.99619806e-01 -3.71706069e-01 1.99592257e+00 1.26421416e+00 -6.03996739e-02 8.71087834e-02 -3.36809397e-01 7.77034938e-01 -1.56141251e-01 -1.66526273e-01 -7.63704360e-01 -7.81115368e-02 5.94167054e-01 4.93457705e-01 6.25850856e-01 -7.88835466e-01 1.38826239e+00 6.62532139e+00 3.72436255e-01 -1.14279354e+00 6.07330576e-02 1.31947473e-01 -3.54439244e-02 -6.07155442e-01 5.18729091e-01 -1.19969082e+00 6.64702713e-01 1.16489983e+00 -2.65581042e-01 7.60260582e-01 8.55473757e-01 2.83097804e-01 7.23015517e-02 -1.40782475e+00 5.47121704e-01 6.83344603e-02 -1.42727923e+00 7.40556717e-02 -2.86091238e-01 5.57481170e-01 2.69975990e-01 -5.20268381e-01 5.23127079e-01 8.88177931e-01 -9.87198651e-01 1.03376925e+00 1.28403351e-01 8.60286117e-01 -3.76886696e-01 6.50152445e-01 5.45090854e-01 -7.04082131e-01 2.98710674e-01 -7.28412867e-02 -2.68365860e-01 3.39509934e-01 4.20278639e-01 -1.14667451e+00 5.11618912e-01 6.65012956e-01 4.62792784e-01 -9.05039966e-01 6.53038085e-01 -6.01783931e-01 7.65762866e-01 -2.84532934e-01 -3.61248255e-01 2.11353470e-02 -1.19403601e-01 5.50289810e-01 1.59561074e+00 4.24420297e-01 -5.48021868e-02 3.65645796e-01 1.18216813e+00 -1.39245465e-01 1.86322585e-01 -6.22418880e-01 -4.77644712e-01 8.02304626e-01 1.09513426e+00 -4.46996778e-01 -4.47858095e-01 -5.37132561e-01 1.06103945e+00 3.54169995e-01 1.41623348e-01 -7.20160127e-01 -8.32378030e-01 6.54696107e-01 2.36456752e-01 1.97785348e-01 -3.06982756e-01 -6.49835646e-01 -1.34480715e+00 3.52135867e-01 -1.48477113e+00 2.01709777e-01 -1.04049575e+00 -1.08746874e+00 6.84900939e-01 1.55410217e-02 -7.24401057e-01 -6.51045680e-01 -4.73218858e-01 -4.94122654e-01 1.16433585e+00 -1.25110734e+00 -1.01856399e+00 -2.91961461e-01 1.48949653e-01 7.03188002e-01 7.62154460e-02 1.09169066e+00 2.56339699e-01 -5.76583028e-01 6.06080413e-01 -2.38045067e-01 5.77089489e-01 8.69197249e-01 -1.34698117e+00 1.25051796e+00 1.03282177e+00 9.34353396e-02 9.55182135e-01 8.37129295e-01 -8.39838803e-01 -1.58801293e+00 -1.17161679e+00 1.43818104e+00 -1.03900731e+00 8.82880211e-01 -6.44587457e-01 -9.52202022e-01 9.83453870e-01 3.62653166e-01 -3.52366447e-01 6.10754371e-01 -2.07587183e-01 -3.62187207e-01 3.23921144e-01 -1.13133013e+00 6.92322612e-01 1.10340452e+00 -4.55771923e-01 -8.19806993e-01 3.48248571e-01 9.64093387e-01 -9.52930570e-01 -7.33104885e-01 5.97429983e-02 2.42971659e-01 -7.05814481e-01 2.63156086e-01 -8.93089831e-01 7.00672269e-01 -3.54404062e-01 -1.36013165e-01 -1.46631324e+00 1.15980096e-01 -1.01727748e+00 4.51716304e-01 1.70676827e+00 8.98465693e-01 -7.50965357e-01 5.43939352e-01 1.13785493e+00 -3.74190599e-01 -3.80700082e-01 -6.37145102e-01 -7.77450323e-01 3.84061426e-01 -5.62251151e-01 1.05878973e+00 1.06357205e+00 1.93562388e-01 4.31087404e-01 1.15528785e-01 -5.84953018e-02 1.11132190e-01 1.79797739e-01 1.26543963e+00 -9.95035589e-01 -5.84631741e-01 -2.59607524e-01 -1.38504580e-01 -9.47633147e-01 1.27988264e-01 -1.15873241e+00 2.13611439e-01 -1.68171084e+00 -6.36016354e-02 -6.10172093e-01 6.39122903e-01 8.75847816e-01 -1.26989260e-01 1.32918125e-02 2.61590838e-01 2.56947160e-01 -2.96303242e-01 1.10970460e-01 7.56454885e-01 3.30104858e-01 -2.66438603e-01 -2.93378383e-01 -1.07926679e+00 5.70163846e-01 9.96892571e-01 -6.76980615e-01 -1.37677774e-01 -1.10649157e+00 4.17475045e-01 -4.81449440e-02 8.10294300e-02 -7.32999623e-01 1.06087804e-01 -3.40500683e-01 -1.17103137e-01 -1.38845220e-01 -2.27090597e-01 -4.14080888e-01 2.82830238e-01 -9.32063386e-02 -5.16273379e-01 5.83564639e-01 2.69490093e-01 1.71167050e-02 -1.29453400e-02 -5.38475454e-01 4.93183762e-01 -5.59299827e-01 -6.48852348e-01 -2.64877379e-01 -5.39958715e-01 7.39182413e-01 7.44218111e-01 3.41666117e-02 -7.58833528e-01 -3.12792987e-01 -7.06129009e-03 2.77521368e-03 8.50582004e-01 7.19177902e-01 5.03177829e-02 -8.75015497e-01 -7.05898345e-01 1.87578872e-01 2.73677796e-01 1.57784551e-01 -5.73037446e-01 4.19587225e-01 -8.91213596e-01 4.48954791e-01 -4.21019271e-02 -3.46004188e-01 -9.59278345e-01 2.12425336e-01 5.26211672e-02 -5.81724010e-03 -6.45667434e-01 5.93649507e-01 -6.69102073e-01 -1.15031660e+00 -1.34777233e-01 -4.63365167e-01 2.44317606e-01 -2.81755567e-01 5.36502659e-01 2.98152357e-01 3.65341276e-01 -4.66882944e-01 -3.04751933e-01 1.82951182e-01 -9.81681719e-02 -4.88154203e-01 1.34671700e+00 4.92676534e-02 -4.26678956e-01 4.17177320e-01 8.77926409e-01 2.87185550e-01 -6.90202117e-01 -9.65350792e-02 3.34411561e-01 -4.43599552e-01 -5.13214648e-01 -1.20840740e+00 -5.26249111e-01 7.14987457e-01 -1.53343514e-01 1.75568536e-01 6.07398093e-01 6.64300025e-02 1.20713294e+00 6.03017867e-01 6.82873905e-01 -1.16384971e+00 -4.19093788e-01 8.23902011e-01 5.60552359e-01 -1.19454801e+00 -4.92910147e-01 -6.75916612e-01 -7.17456937e-01 1.14203453e+00 6.45205140e-01 1.95488006e-01 -4.18141186e-02 5.28570056e-01 4.86459225e-01 1.01979837e-01 -7.73777783e-01 2.80650370e-02 -3.24972421e-01 7.55333245e-01 8.43683660e-01 1.31056502e-01 -4.40131217e-01 7.08296180e-01 -9.37932193e-01 2.46794149e-01 9.44323003e-01 1.30671394e+00 -2.07495600e-01 -1.68104875e+00 -3.10146302e-01 3.32757711e-01 -6.68039083e-01 -5.11192918e-01 -8.87397707e-01 8.11868548e-01 -2.66457438e-01 1.18704724e+00 -1.26964912e-01 -1.43574819e-01 4.46411192e-01 5.31072557e-01 3.53394389e-01 -1.23450243e+00 -7.73777902e-01 -3.93492877e-01 5.78147054e-01 -5.51964521e-01 3.95639054e-02 -7.94522166e-01 -1.43028665e+00 -3.01304817e-01 -9.04423445e-02 2.77012229e-01 8.28009963e-01 9.91063416e-01 7.85643339e-01 1.30986005e-01 7.34162480e-02 -5.51610589e-01 -6.99817181e-01 -9.02012527e-01 3.00533213e-02 5.55266201e-01 -1.86836168e-01 -1.39323130e-01 -1.10392131e-01 3.86300951e-01]
[11.22420597076416, 9.004302978515625]
41121c8b-0060-4331-b5b3-d68d3792b285
similarity-maps-for-self-training-weakly
null
null
http://openaccess.thecvf.com//content/CVPR2023/html/Shaharabany_Similarity_Maps_for_Self-Training_Weakly-Supervised_Phrase_Grounding_CVPR_2023_paper.html
http://openaccess.thecvf.com//content/CVPR2023/papers/Shaharabany_Similarity_Maps_for_Self-Training_Weakly-Supervised_Phrase_Grounding_CVPR_2023_paper.pdf
Similarity Maps for Self-Training Weakly-Supervised Phrase Grounding
A phrase grounding model receives an input image and a text phrase and outputs a suitable localization map. We present an effective way to refine a phrase ground model by considering self-similarity maps extracted from the latent representation of the model's image encoder. Our main insights are that these maps resemble localization maps and that by combining such maps, one can obtain useful pseudo-labels for performing self-training. Our results surpass, by a large margin, the state-of-the-art in weakly supervised phrase grounding. A similar gap in performance is obtained for a recently proposed downstream task called WWbL, in which the input image is given without any text. Our code is available as supplementary.
['Lior Wolf', 'Tal Shaharabany']
2023-01-01
null
null
null
cvpr-2023-1
['phrase-grounding']
['natural-language-processing']
[ 5.03992975e-01 6.56606495e-01 -4.73344386e-01 -3.35518569e-01 -1.14509416e+00 -7.96043277e-01 9.34386790e-01 3.22146624e-01 -4.62484062e-01 5.51690340e-01 4.13847327e-01 -1.10477008e-01 -1.57239567e-02 -5.80751359e-01 -1.06832242e+00 -7.12130725e-01 3.27838928e-01 6.88908339e-01 3.49122256e-01 -1.90854333e-02 2.79646695e-01 -1.15310237e-01 -1.27236700e+00 6.38966918e-01 3.31183821e-01 6.65253520e-01 5.83998740e-01 5.67256451e-01 -1.89684823e-01 5.98382711e-01 -2.46882230e-01 -7.09611893e-01 1.25428513e-01 -7.72050560e-01 -1.05728042e+00 1.18280493e-01 7.04646766e-01 -8.67064111e-03 -3.73595893e-01 1.25201762e+00 1.02726728e-01 -4.62636240e-02 6.66655004e-01 -9.97341514e-01 -1.05792749e+00 8.04143369e-01 -3.15146685e-01 2.95425028e-01 3.97416919e-01 -9.99090001e-02 1.58031297e+00 -1.00638890e+00 9.21660960e-01 1.09974480e+00 6.26545489e-01 6.03200972e-01 -1.62349927e+00 -2.46841639e-01 1.37066975e-01 -1.69377737e-02 -1.40184712e+00 -3.39845747e-01 4.18100208e-01 -3.59333158e-01 9.47202206e-01 -1.59803331e-01 3.11204463e-01 1.19671345e+00 2.21667230e-01 8.54524970e-01 1.34043157e+00 -7.97199190e-01 8.97566751e-02 4.56718981e-01 5.33544049e-02 1.08566403e+00 -1.53219283e-01 1.94074526e-01 -1.08827698e+00 2.57466510e-02 6.88785732e-01 -2.22708017e-01 -2.66670018e-01 -4.64089632e-01 -1.57523942e+00 8.12155843e-01 7.42662191e-01 5.11484087e-01 -1.17506059e-02 3.07624787e-01 8.33875407e-03 3.29895347e-01 8.15245152e-01 5.76222718e-01 -1.59609437e-01 7.80179203e-02 -1.26520717e+00 3.56493182e-02 6.15244806e-01 1.22089243e+00 1.00854719e+00 -6.58530414e-01 -4.30440098e-01 2.90032655e-01 3.64045024e-01 5.10077953e-01 4.19476688e-01 -6.69700444e-01 5.81451952e-01 2.89332211e-01 2.65789684e-02 -8.83123279e-01 -2.37056524e-01 -7.22533762e-01 -5.55032849e-01 -9.80169028e-02 3.07417214e-01 3.31631243e-01 -9.63429630e-01 1.83507824e+00 -4.18097824e-01 4.31217045e-01 2.12140262e-01 7.43441939e-01 6.90829217e-01 6.83939159e-01 3.85248512e-02 -9.37008783e-02 1.56524003e+00 -1.31999147e+00 -6.81046069e-01 -4.77344364e-01 6.21593297e-01 -5.14532864e-01 1.21516287e+00 7.94851035e-02 -1.09447205e+00 -7.44976521e-01 -9.85088706e-01 -1.10627227e-01 -4.26072091e-01 1.70270219e-01 3.43393147e-01 1.60346106e-01 -1.67391455e+00 7.70312250e-01 -7.55003035e-01 -9.48429167e-01 3.62512767e-01 3.86818498e-01 -5.93180060e-01 4.75715064e-02 -1.00362444e+00 9.81431961e-01 4.64483947e-01 -2.43077561e-01 -7.09235191e-01 -5.15253544e-01 -8.23000968e-01 -2.97141206e-02 1.62481472e-01 -9.32910562e-01 1.35495901e+00 -7.97256827e-01 -1.21320069e+00 1.65263045e+00 -4.45461482e-01 -6.52014494e-01 2.17264518e-01 -7.84264430e-02 -5.52173257e-02 5.16537011e-01 5.35115719e-01 1.15573859e+00 1.11535156e+00 -1.37309217e+00 -7.49063611e-01 -2.91680723e-01 6.41670153e-02 3.03327590e-01 -1.52500018e-01 -7.17443153e-02 -7.36834645e-01 -6.28714025e-01 1.98504865e-01 -1.07725286e+00 -1.31592140e-01 2.48601586e-02 -4.07508969e-01 -1.46709248e-01 5.32339394e-01 -3.59351367e-01 8.80866766e-01 -2.26216817e+00 5.87970138e-01 2.11655293e-02 3.90057206e-01 -1.29596472e-01 -3.23013693e-01 4.51761663e-01 -1.67445093e-01 2.42290631e-01 -4.02206928e-01 -7.85757542e-01 8.96256417e-02 3.59635532e-01 -8.10516059e-01 4.89009380e-01 3.96226197e-01 1.35984242e+00 -1.20650196e+00 -5.13434589e-01 4.77871113e-02 9.43588614e-02 -4.18527097e-01 2.51262188e-01 -4.06123906e-01 5.83875716e-01 -2.71628469e-01 8.38702768e-02 1.38779983e-01 -7.54622161e-01 -2.09999964e-01 -2.10680410e-01 1.64400026e-01 3.95436406e-01 -5.86713076e-01 2.32534909e+00 -3.16340268e-01 8.68693948e-01 -1.79847434e-01 -9.68017995e-01 8.11420500e-01 5.34783363e-01 1.33976445e-01 -5.97067356e-01 -5.27421609e-02 1.62253261e-01 -2.66069651e-01 -3.69540930e-01 3.32378864e-01 -5.26756406e-01 -1.92695826e-01 7.29034960e-01 7.83555388e-01 -2.29884893e-01 1.13724008e-01 3.83415639e-01 1.09834266e+00 4.64535147e-01 2.54881591e-01 -5.50836265e-01 5.01979351e-01 1.34523213e-01 -7.75724575e-02 1.08241355e+00 -4.47001867e-02 8.88481081e-01 4.50773865e-01 -2.56850779e-01 -1.16452467e+00 -1.33724606e+00 -1.43523708e-01 1.23033738e+00 3.26604337e-01 -6.99687064e-01 -8.51749480e-01 -8.66928995e-01 -1.67261586e-01 3.90030026e-01 -7.88935363e-01 -1.63092747e-01 -3.85142475e-01 -4.10290837e-01 5.69857657e-01 5.82924545e-01 4.81631219e-01 -1.14659357e+00 -1.33804202e-01 3.49961184e-02 -3.58064204e-01 -1.40132654e+00 -3.74003679e-01 5.63999951e-01 -9.29400146e-01 -6.78099215e-01 -8.27049792e-01 -1.26079035e+00 1.02655780e+00 1.99327409e-01 1.38777542e+00 9.03691202e-02 1.14979811e-01 4.13399100e-01 -3.91860276e-01 -1.13304101e-01 -4.90753561e-01 5.09395778e-01 8.34585130e-02 -1.71172753e-01 4.96185184e-01 -3.51127207e-01 -3.04887503e-01 1.71885446e-01 -7.90510416e-01 4.34181899e-01 8.18747878e-01 9.89804208e-01 7.06293762e-01 -4.96545099e-02 2.15490237e-01 -9.28554058e-01 4.89698917e-01 -3.57174218e-01 -4.74366963e-01 2.39807531e-01 -5.98533809e-01 5.96888721e-01 6.08799875e-01 9.75267123e-03 -8.18283021e-01 1.34154096e-01 -1.81095734e-01 -2.42577031e-01 -4.56752181e-01 3.77119154e-01 3.06455910e-01 9.01576877e-03 7.39986837e-01 3.52584690e-01 -1.39808327e-01 -4.12554562e-01 6.86376929e-01 5.02166271e-01 8.96175206e-01 -5.90339363e-01 1.11809158e+00 6.02858961e-01 -1.82391133e-03 -3.59467417e-01 -1.50393975e+00 -7.97856808e-01 -1.06416297e+00 1.68802768e-01 1.21084249e+00 -9.71973240e-01 -2.42268518e-01 -3.17197526e-03 -1.33527315e+00 -2.97902733e-01 -4.77727443e-01 2.59208828e-01 -1.01180434e+00 3.84261236e-02 -6.27506256e-01 -2.14346975e-01 -9.73423198e-02 -9.52112973e-01 1.56600714e+00 2.39511225e-02 -2.45258629e-01 -1.21534002e+00 2.20515668e-01 4.08501297e-01 1.20238937e-01 -3.68174285e-01 8.21806490e-01 -7.09296346e-01 -7.43141234e-01 2.95829605e-02 -4.31481421e-01 4.85917963e-02 -8.13912880e-03 -5.43109238e-01 -1.12373912e+00 -3.34088385e-01 4.50110584e-02 -4.34333086e-01 1.27289581e+00 1.57823265e-01 6.43906295e-01 -2.41154850e-01 -4.84405726e-01 6.94825590e-01 1.53764939e+00 -4.77834791e-01 3.23025882e-01 3.96923065e-01 5.80294132e-01 7.30702281e-01 5.36379755e-01 -2.63226330e-01 2.08663270e-01 8.02160680e-01 2.76140302e-01 -4.22334373e-01 -2.50983328e-01 -7.10862994e-01 3.88427526e-01 7.10817218e-01 1.02932632e-01 -3.04920971e-01 -9.29848909e-01 7.10787058e-01 -2.17894173e+00 -9.67733264e-01 2.59796418e-02 1.78536820e+00 9.71713424e-01 2.89019585e-01 -6.30112737e-02 -1.19741067e-01 6.08324468e-01 2.65639931e-01 -1.92003250e-01 -1.15462504e-01 -1.60874799e-01 4.11469191e-01 6.87407911e-01 7.83912957e-01 -1.25036216e+00 1.42758334e+00 7.20682621e+00 6.33578539e-01 -7.71954417e-01 3.63037109e-01 4.47583735e-01 3.34832460e-01 -4.66136545e-01 1.99386895e-01 -8.67129445e-01 2.77889848e-01 1.00055993e+00 -3.07156742e-02 2.11344570e-01 5.98644316e-01 -7.41850957e-02 -4.51650750e-03 -1.55424118e+00 9.27085638e-01 4.38825727e-01 -1.58869982e+00 1.64430320e-01 2.91508250e-02 7.47205734e-01 2.29168802e-01 2.80086070e-01 1.27045050e-01 3.57019603e-01 -1.07916903e+00 7.60861039e-01 4.53571528e-01 7.09928513e-01 -1.80150077e-01 7.49653161e-01 4.11643237e-01 -9.80786204e-01 1.91615522e-01 -3.67371351e-01 -1.38735101e-01 1.43255934e-01 2.29101911e-01 -7.78785467e-01 4.84792024e-01 5.93994141e-01 9.56635833e-01 -8.29917967e-01 7.20289767e-01 -8.89697433e-01 6.38254881e-01 -4.42695878e-02 1.51748732e-01 5.84449351e-01 -1.07824877e-01 4.68593866e-01 1.36431670e+00 1.84053212e-01 -1.12737522e-01 1.40603319e-01 1.18753469e+00 -1.91743985e-01 4.39844131e-02 -9.16976154e-01 4.56646867e-02 -7.06569059e-03 1.19207609e+00 -9.96207416e-01 -3.04492772e-01 -5.34685254e-01 1.50540817e+00 5.20089388e-01 4.87604558e-01 -5.27531326e-01 -1.02567583e-01 1.51184335e-01 4.88885567e-02 3.77994001e-01 -2.22457409e-01 -1.85425431e-01 -1.34550428e+00 -4.15280201e-02 -3.45521003e-01 1.50375605e-01 -1.16620433e+00 -1.19758511e+00 6.48098290e-01 9.75922197e-02 -1.08734679e+00 -4.20668304e-01 -9.00542498e-01 -3.52187306e-01 8.04730833e-01 -1.50906861e+00 -1.49100876e+00 -1.77944615e-01 4.87513155e-01 4.56131399e-01 -1.69743255e-01 1.19143081e+00 -1.29136696e-01 -7.12634847e-02 3.42635393e-01 -3.26007269e-02 2.19484791e-01 7.35820472e-01 -1.59467673e+00 6.96664751e-01 9.06997144e-01 1.08039916e+00 6.97385669e-01 9.96492743e-01 -2.96068609e-01 -1.02575099e+00 -1.00195849e+00 1.21721780e+00 -9.77462471e-01 9.56024170e-01 -8.26428294e-01 -7.64771342e-01 1.07983685e+00 5.11800826e-01 1.83661118e-01 7.22450018e-01 2.15514466e-01 -4.38654482e-01 1.02801211e-01 -8.89278352e-01 5.46373487e-01 1.20610201e+00 -1.03910446e+00 -1.13169909e+00 5.29852152e-01 9.13608611e-01 -2.63856888e-01 -5.53201556e-01 -7.59868994e-02 2.50059158e-01 -7.82443404e-01 7.96095431e-01 -4.84531254e-01 5.60178041e-01 -2.62713194e-01 -1.66386530e-01 -1.28499985e+00 -6.03723884e-01 -5.53025126e-01 9.52661037e-02 1.01153421e+00 6.64734721e-01 -2.96331823e-01 1.13173962e+00 3.34879309e-02 -1.39865689e-02 -5.47542870e-01 -7.95870125e-01 -7.39777446e-01 -1.22219529e-02 -2.79905617e-01 1.12131201e-01 7.03738809e-01 4.24789906e-01 8.03263545e-01 -1.69336528e-01 2.82071233e-01 7.77547419e-01 1.13043733e-01 4.24869925e-01 -1.08281469e+00 -5.53838730e-01 -1.63621366e-01 -5.93141794e-01 -1.49691784e+00 6.07450187e-01 -1.31758070e+00 4.33028758e-01 -1.62442577e+00 5.20885408e-01 -1.55903891e-01 -5.78197777e-01 7.75771797e-01 -9.12296027e-02 8.77713203e-01 3.07974696e-01 5.88497460e-01 -9.55467522e-01 3.75391454e-01 1.02605987e+00 -1.64895296e-01 2.14414746e-01 -1.12747476e-01 -8.27129483e-01 7.51373589e-01 6.03209972e-01 -6.98597491e-01 -5.05968034e-01 -6.48221195e-01 5.27336359e-01 -9.05293077e-02 5.73558569e-01 -8.34846973e-01 4.74407673e-01 2.66860753e-01 1.79931074e-02 -5.15754998e-01 1.86872885e-01 -6.76151156e-01 -3.78804505e-01 2.48480678e-01 -8.40098083e-01 -2.42083948e-02 1.95100814e-01 5.98606706e-01 -4.80014145e-01 -5.27135670e-01 4.76348609e-01 -2.85074204e-01 -7.64048696e-01 2.22588733e-01 -3.77078354e-01 -7.46100396e-02 7.27542281e-01 -1.09815545e-01 -4.69665676e-01 -4.61146384e-01 -9.31126773e-01 -1.06090240e-01 6.16893530e-01 4.42662954e-01 5.18067896e-01 -1.30367637e+00 -6.46602869e-01 2.47258261e-01 5.23076475e-01 -5.27398065e-02 -4.25545990e-01 8.31115425e-01 -2.48681560e-01 9.02374923e-01 -1.17354006e-01 -9.65055466e-01 -1.08139277e+00 5.28995335e-01 2.18017116e-01 -3.23315144e-01 -6.97297513e-01 1.10368145e+00 5.15621006e-01 -2.85344183e-01 1.61909431e-01 -5.12481809e-01 9.94836837e-02 -1.54374897e-01 4.86420333e-01 -3.76965463e-01 5.33220451e-03 -8.74336481e-01 -2.63845742e-01 8.83740366e-01 -1.71672314e-01 -6.23741925e-01 1.17541277e+00 -2.59782791e-01 -1.77139148e-01 6.02991462e-01 1.32566965e+00 -1.12319380e-01 -1.17378092e+00 -6.69484317e-01 2.71424502e-01 -1.95305213e-01 1.02805905e-01 -5.75297713e-01 -6.32202983e-01 8.82848918e-01 4.23944890e-01 9.73996744e-02 1.00794470e+00 7.95769691e-01 3.85825902e-01 6.21378839e-01 4.59134966e-01 -6.28463507e-01 3.63395631e-01 5.81679225e-01 7.09325373e-01 -1.47262847e+00 -1.08192012e-01 -2.30327472e-01 -4.22016799e-01 8.80503595e-01 1.99754268e-01 -3.59941334e-01 5.25397003e-01 3.01260948e-01 -2.97281984e-02 -4.23805237e-01 -7.93579340e-01 -4.46730644e-01 4.86708671e-01 7.22834945e-01 4.18978572e-01 -2.87696451e-01 -8.54705721e-02 4.90242422e-01 -2.80033976e-01 -1.40162110e-01 1.84394628e-01 6.85667217e-01 -5.66725731e-01 -1.25274575e+00 -1.33622557e-01 1.62655249e-01 -4.21460718e-01 -4.75147933e-01 -5.39155841e-01 5.40481567e-01 1.69441834e-01 7.05408216e-01 2.50787258e-01 -3.11151683e-01 -6.44821748e-02 1.54016212e-01 7.07675815e-01 -1.33246696e+00 -3.72471541e-01 -4.78005931e-02 -1.68671384e-01 -6.07338130e-01 -7.88834989e-01 -4.50274020e-01 -1.26678574e+00 1.45322725e-01 -2.42585838e-01 2.42427900e-01 4.89563525e-01 1.27625775e+00 1.86512932e-01 2.53855228e-01 2.67498255e-01 -8.33665431e-01 -2.89730072e-01 -8.58522952e-01 -4.67916638e-01 6.24364495e-01 4.73632693e-01 -4.73882824e-01 -3.64378393e-01 5.59037566e-01]
[10.50529670715332, 1.5586795806884766]
f3c2cbb0-0059-40dd-b83e-4a2640a685ed
adapters-for-enhanced-modeling-of
2210.13617
null
https://arxiv.org/abs/2210.13617v2
https://arxiv.org/pdf/2210.13617v2.pdf
Adapters for Enhanced Modeling of Multilingual Knowledge and Text
Large language models appear to learn facts from the large text corpora they are trained on. Such facts are encoded implicitly within their many parameters, making it difficult to verify or manipulate what knowledge has been learned. Language models have recently been extended to multilingual language models (MLLMs), enabling knowledge to be learned across hundreds of languages. Meanwhile, knowledge graphs contain facts in an explicit triple format, which require careful and costly curation and are only available in a few high-resource languages, restricting their research and application. To address these issues, we propose to enhance MLLMs with knowledge from multilingual knowledge graphs (MLKGs) so as to tackle language and knowledge graph tasks across many languages, including low-resource ones. Specifically, we introduce a lightweight adapter set to enhance MLLMs with cross-lingual entity alignment and facts from MLKGs for many languages. Experiments on common benchmarks show that such enhancement benefits both MLLMs and MLKGs, achieving: (1) comparable or improved performance for knowledge graph completion and entity alignment relative to baselines, especially for low-resource languages (for which knowledge graphs are unavailable); and (2) improved MLLM performance on language understanding tasks that require multilingual factual knowledge; all while maintaining performance on other general language tasks.
['Mrinmaya Sachan', 'Carl Allen', 'Zhaopeng Tu', 'Meizhen Liu', 'Wenxiang Jiao', 'Yifan Hou']
2022-10-24
null
null
null
null
['entity-alignment', 'entity-alignment']
['knowledge-base', 'natural-language-processing']
[-3.92216682e-01 3.30007702e-01 -8.31236303e-01 -1.66223332e-01 -7.26378083e-01 -1.01401711e+00 7.30833054e-01 4.73692060e-01 -6.73763394e-01 9.61485982e-01 3.01165909e-01 -5.09905219e-01 8.23580250e-02 -9.65366900e-01 -1.18003011e+00 -8.38544890e-02 3.43986787e-03 6.96337938e-01 7.09070563e-02 -3.56669605e-01 -4.82055128e-01 7.07821995e-02 -8.51298034e-01 1.37880191e-01 1.21701181e+00 4.33052272e-01 1.92668721e-01 2.35408768e-01 -5.68216980e-01 1.02742755e+00 -3.08097124e-01 -1.24372709e+00 1.01936042e-01 1.21420726e-01 -1.21108484e+00 -4.72572148e-01 8.22685719e-01 1.81815717e-02 -4.09923822e-01 1.13295889e+00 1.67290166e-01 -1.54645473e-01 4.24532115e-01 -1.06029987e+00 -1.24808121e+00 1.25694203e+00 -4.15928066e-01 1.41751081e-01 5.00256300e-01 -7.23138377e-02 1.37009823e+00 -8.25343668e-01 8.69937658e-01 1.38317549e+00 6.89547122e-01 1.46395281e-01 -1.09342360e+00 -5.93714178e-01 4.76648003e-01 3.78451914e-01 -1.73459792e+00 -4.48697478e-01 3.91574442e-01 -2.93273360e-01 1.66508579e+00 -1.88586697e-01 4.68811929e-01 9.60657954e-01 -7.20792189e-02 7.59546518e-01 7.23075092e-01 -5.65457582e-01 -4.51987714e-01 3.03446263e-01 2.54144430e-01 1.07405114e+00 8.31550002e-01 -3.32389802e-01 -6.97959065e-01 -1.00393053e-02 6.08784735e-01 -5.09267032e-01 -4.43424672e-01 -3.35869879e-01 -1.54048443e+00 6.42539501e-01 3.60418499e-01 3.56819987e-01 -3.10037613e-01 -7.43526742e-02 5.41699409e-01 6.76924884e-01 5.14021337e-01 4.35525000e-01 -9.53339875e-01 1.52684137e-01 -4.55885530e-01 5.51547296e-02 1.16938698e+00 1.47380674e+00 1.16898751e+00 9.56440642e-02 1.96595624e-01 8.87188315e-01 5.52125163e-02 9.78226006e-01 4.01159197e-01 -3.14986259e-01 1.03171861e+00 9.05370355e-01 -8.40226263e-02 -1.03817320e+00 -3.92932326e-01 -5.05238235e-01 -7.98181355e-01 -6.24068201e-01 4.14228022e-01 -1.00185974e-02 -7.06035256e-01 2.01885176e+00 2.85294622e-01 3.00154001e-01 5.24031162e-01 3.44013661e-01 1.10209608e+00 4.41755921e-01 3.02405000e-01 -1.26882056e-02 1.52534151e+00 -8.97392452e-01 -7.60239005e-01 -6.41047418e-01 1.16260958e+00 -5.31262696e-01 1.33066273e+00 1.04103826e-01 -7.02707708e-01 -4.25296545e-01 -8.28773439e-01 -4.45171297e-01 -1.03824544e+00 6.63698763e-02 1.07002091e+00 5.40255487e-01 -1.17358577e+00 6.29897192e-02 -8.23952913e-01 -4.11620408e-01 2.93990552e-01 2.39365548e-01 -8.11171174e-01 -4.57607269e-01 -1.64800072e+00 1.18025744e+00 1.18833315e+00 1.02181211e-01 -5.66498280e-01 -9.49310243e-01 -1.45834887e+00 -1.03313141e-01 8.59294355e-01 -1.01872635e+00 8.36509705e-01 -6.65347457e-01 -1.06925726e+00 1.10868406e+00 -3.61281782e-02 -4.14149374e-01 8.64716843e-02 -4.49131757e-01 -6.94490790e-01 -2.76363939e-01 2.43889749e-01 3.12894851e-01 4.76963639e-01 -1.07994628e+00 -5.63157320e-01 -3.35047036e-01 5.79699934e-01 3.44960362e-01 -5.19927502e-01 -3.58072221e-02 -9.04652655e-01 -5.34198821e-01 -2.38177478e-01 -7.55258262e-01 6.77568465e-02 -7.52716124e-01 -6.07601941e-01 -3.19646686e-01 5.08948028e-01 -1.21066546e+00 1.27994573e+00 -1.48332214e+00 2.71849424e-01 9.18120295e-02 2.32070312e-01 5.92100322e-01 -4.18296367e-01 5.28015733e-01 1.43965825e-01 1.75463244e-01 -8.94111916e-02 -1.42596513e-01 6.43766895e-02 6.60847485e-01 -3.64894569e-01 8.23044702e-02 2.03754619e-01 1.58846724e+00 -1.15955472e+00 -5.47798455e-01 -4.79424037e-02 3.31174374e-01 -4.44886327e-01 3.32303606e-02 -4.15803641e-01 2.08148509e-01 -4.65034217e-01 7.84213781e-01 2.89616734e-01 -4.39726919e-01 6.02073908e-01 -4.57593143e-01 3.08464259e-01 4.82702225e-01 -1.17630541e+00 1.97506976e+00 -8.03505540e-01 1.87166303e-01 -1.61225438e-01 -7.19196737e-01 5.67280233e-01 4.11958575e-01 4.13387567e-02 -5.70594668e-01 -2.82142758e-01 2.70158321e-01 -8.88435170e-02 -5.36598027e-01 5.75058579e-01 -1.07926399e-01 -1.58590451e-01 3.22733283e-01 6.14967108e-01 -1.25641808e-01 5.92533231e-01 7.67755687e-01 8.61745954e-01 1.42532364e-01 5.98443806e-01 -1.66496873e-01 7.91502059e-01 1.17342293e-01 6.19972229e-01 5.91133893e-01 3.21717769e-01 -4.09593403e-01 1.81899175e-01 -3.19735736e-01 -7.13748693e-01 -1.03398728e+00 1.22771524e-01 1.15412128e+00 -3.11786053e-03 -8.55787277e-01 -4.22997236e-01 -1.07484949e+00 2.49544412e-01 8.05017531e-01 -2.92990804e-01 -6.55592978e-02 -7.78748453e-01 -7.38148987e-01 8.73312414e-01 5.67043543e-01 3.62548709e-01 -8.54661584e-01 4.50450420e-01 2.48813674e-01 -4.34278190e-01 -1.97716308e+00 -2.63270169e-01 -1.80379227e-01 -4.75289941e-01 -1.38691509e+00 -1.92344189e-01 -8.85833383e-01 7.27356672e-01 9.75615717e-03 1.64659083e+00 1.38991633e-02 9.35413539e-02 7.57193327e-01 -3.61115783e-01 -4.62836713e-01 -4.73819971e-01 4.42797691e-01 3.36374074e-01 -2.44780764e-01 5.79951167e-01 -5.22704303e-01 8.93895254e-02 2.88693104e-02 -9.03515875e-01 1.32596135e-01 6.67974174e-01 6.20318711e-01 6.21033549e-01 4.70333546e-02 7.46975005e-01 -1.38982499e+00 3.83265734e-01 -4.55638617e-01 -6.55007482e-01 9.12624419e-01 -4.84711736e-01 3.19049329e-01 8.45880151e-01 -2.53797203e-01 -1.12370336e+00 -3.72031450e-01 -3.61523102e-03 -1.14949532e-01 1.91478059e-01 1.08723032e+00 -4.60268170e-01 -3.54161531e-01 5.48530400e-01 2.19379470e-01 -5.24807036e-01 -5.38027823e-01 1.06689322e+00 2.73582190e-01 8.05165827e-01 -1.12229288e+00 1.06203687e+00 1.77264616e-01 -3.87033492e-01 -6.72578275e-01 -1.30644894e+00 -3.54849070e-01 -1.07161951e+00 2.87737668e-01 5.84404051e-01 -1.55110168e+00 -3.39945465e-01 3.80008727e-01 -1.10843575e+00 -5.15584826e-01 -1.22330844e-01 7.12196410e-01 -1.67966336e-01 4.78582591e-01 -7.56890833e-01 -1.35960966e-01 -4.82432246e-01 -6.58604860e-01 9.03160930e-01 -1.64466165e-03 3.40671837e-02 -1.76481700e+00 1.35311693e-01 5.26158929e-01 3.30568671e-01 -1.61223207e-02 1.49485087e+00 -8.23533118e-01 -6.60593569e-01 -6.15621060e-02 -2.95004189e-01 4.15789872e-01 3.60553890e-01 -2.78968304e-01 -6.62703514e-01 -5.07032037e-01 -5.72871685e-01 -7.02760875e-01 6.36152983e-01 -1.24796025e-01 5.92063189e-01 -5.01317978e-01 -4.50334102e-01 7.19438195e-01 1.47046924e+00 -5.42205215e-01 2.39244416e-01 3.36622357e-01 1.43460798e+00 4.84220922e-01 2.86282808e-01 -3.18212539e-01 1.26409316e+00 6.49965227e-01 -9.22540296e-03 -2.17873991e-01 -4.27650303e-01 -6.99928045e-01 4.62555796e-01 1.83648849e+00 -2.35998690e-01 -6.86489716e-02 -1.23880291e+00 8.87037635e-01 -1.82174993e+00 -4.61542934e-01 -2.78491437e-01 2.19753981e+00 1.43558383e+00 -9.83871520e-02 -2.31157601e-01 -7.09044278e-01 4.72614646e-01 2.19048470e-01 -4.92069095e-01 1.25222029e-02 -6.81301415e-01 2.46807918e-01 7.31053650e-01 7.53542066e-01 -9.74316478e-01 1.72076786e+00 5.50208092e+00 9.59846675e-01 -8.86445045e-01 2.60843486e-01 -8.32483992e-02 3.66584659e-01 -6.10141397e-01 4.26791757e-01 -1.09777045e+00 7.26655424e-02 7.46397555e-01 -5.74442327e-01 4.54499900e-01 5.45855165e-01 -3.75075072e-01 1.95285872e-01 -1.14096844e+00 9.75695133e-01 1.12083457e-01 -1.14655674e+00 6.76989436e-01 -1.38889045e-01 9.23295259e-01 4.63103294e-01 -4.20001805e-01 9.00958955e-01 1.00732839e+00 -1.01464176e+00 3.43079656e-01 3.38835984e-01 9.42762613e-01 -5.95269024e-01 6.82240009e-01 4.53984052e-01 -1.44223011e+00 3.42806607e-01 -5.75072765e-01 2.67104357e-01 4.06488538e-01 7.11626112e-01 -8.58379304e-01 1.38845515e+00 3.58031124e-01 9.74898279e-01 -9.07102406e-01 4.42321599e-01 -9.94328856e-01 5.91887832e-01 -3.15515220e-01 4.39403743e-01 2.04888821e-01 -1.82208270e-01 3.03914249e-01 1.47208047e+00 -2.32645515e-02 -1.35565490e-01 5.16173303e-01 7.37563193e-01 -5.91168284e-01 5.32795012e-01 -7.91265011e-01 -4.43816483e-01 4.95989203e-01 1.23751485e+00 -1.89934880e-01 -6.23926580e-01 -1.10303521e+00 7.89970040e-01 1.04525006e+00 5.94932318e-01 -4.38224554e-01 -2.07923293e-01 5.80579519e-01 -2.14290202e-01 -4.55191694e-02 -4.91119623e-01 3.27805370e-01 -1.82953179e+00 2.20935047e-01 -1.04830694e+00 8.92799616e-01 -4.43964571e-01 -1.54567730e+00 4.88837302e-01 1.12034678e-01 -5.16392112e-01 -3.89753729e-01 -7.90997148e-01 6.60100877e-02 8.83107901e-01 -2.24492669e+00 -1.91340494e+00 7.65578672e-02 1.11437333e+00 9.78625193e-02 -3.24462652e-01 9.24642444e-01 6.56047642e-01 -5.02612650e-01 8.21453691e-01 -1.66973680e-01 5.62363863e-01 1.04844093e+00 -1.37769914e+00 7.40186453e-01 9.35055077e-01 8.43971074e-01 9.10847187e-01 1.89577341e-01 -1.01406264e+00 -1.82227647e+00 -1.25866306e+00 1.45125473e+00 -8.88787031e-01 1.23883986e+00 -4.39802349e-01 -1.42151690e+00 1.49411750e+00 1.00054197e-01 -2.33980361e-02 7.19936013e-01 8.22210133e-01 -9.25898373e-01 1.12166911e-01 -6.05018079e-01 5.60885012e-01 1.25467443e+00 -9.76313710e-01 -8.00645173e-01 6.37423337e-01 9.07700002e-01 -6.16630256e-01 -1.32602680e+00 4.67683524e-01 1.86033353e-01 -9.90416631e-02 1.03026211e+00 -1.07210600e+00 1.23107545e-02 -3.96702856e-01 -4.51288410e-02 -1.52398407e+00 -1.33514464e-01 -4.93431926e-01 -4.23953623e-01 1.42903745e+00 8.19111824e-01 -9.63499427e-01 3.52087915e-01 4.98345494e-01 9.65158828e-03 -2.87599474e-01 -6.43000185e-01 -1.12771475e+00 2.72514880e-01 -5.38070440e-01 5.71789324e-01 1.75563776e+00 8.78560320e-02 6.89308703e-01 -4.35499966e-01 5.25424659e-01 4.27991152e-01 2.06856325e-01 9.81744945e-01 -9.96325135e-01 -2.45917261e-01 -4.80230972e-02 -4.03337181e-01 -9.28368270e-01 8.27225447e-01 -1.60591877e+00 -4.63368148e-01 -1.64738560e+00 3.51210147e-01 -6.38565600e-01 -2.99966514e-01 1.16775203e+00 -6.95283651e-01 -1.37184173e-01 -1.11834042e-01 7.33793527e-02 -7.62856901e-01 5.54837763e-01 1.04652917e+00 -1.39598340e-01 6.67181090e-02 -5.61955929e-01 -7.93346226e-01 8.47347140e-01 3.98171306e-01 -4.10486370e-01 -4.88354117e-01 -1.00245166e+00 7.10376143e-01 -3.59308332e-01 3.14418793e-01 -5.36422551e-01 5.14491856e-01 -1.46329269e-01 -1.33389398e-01 -4.80493344e-02 4.19466943e-03 -6.78827584e-01 1.39842657e-02 -2.40623448e-02 7.14883730e-02 9.79496390e-02 5.11126578e-01 6.31359696e-01 -6.05280638e-01 4.03202251e-02 1.54081240e-01 -2.09673479e-01 -1.19222617e+00 5.38725317e-01 3.96073401e-01 6.79758906e-01 6.04090512e-01 2.99954265e-01 -6.79479003e-01 -3.04087311e-01 -4.91436988e-01 5.36935151e-01 4.98177648e-01 7.14987755e-01 8.87767076e-02 -1.41369402e+00 -9.30328071e-01 -1.41733065e-02 5.72133720e-01 1.74591124e-01 1.73308581e-01 8.24279368e-01 -2.54888803e-01 6.95861161e-01 1.36756197e-01 -1.99386150e-01 -9.51740742e-01 7.06376195e-01 1.12615131e-01 -7.65815437e-01 -6.50335014e-01 7.75001228e-01 2.48439133e-01 -1.16036010e+00 -1.28673129e-02 -3.22453022e-01 -1.54931441e-01 -4.88418192e-02 3.52895349e-01 -5.75169288e-02 3.34381729e-01 -8.50257337e-01 -3.30271840e-01 6.91080749e-01 -2.72053361e-01 3.52600217e-01 1.24991667e+00 -2.15738386e-01 -4.18329537e-01 2.21158430e-01 7.88109839e-01 5.46746016e-01 -5.91178060e-01 -1.13302076e+00 3.82249743e-01 -2.16446131e-01 -1.76546127e-01 -9.39652443e-01 -1.01921654e+00 5.26895463e-01 -2.52871275e-01 -2.20493123e-01 8.13923895e-01 2.33255267e-01 9.04406190e-01 9.46346760e-01 8.18343103e-01 -9.54472244e-01 -3.78503889e-01 9.50617492e-01 6.60637915e-01 -1.40085423e+00 1.34210005e-01 -7.74793267e-01 -6.71044230e-01 7.19188392e-01 6.01307452e-01 4.21908379e-01 5.18973827e-01 1.89310223e-01 1.97760180e-01 -4.91739899e-01 -7.32079268e-01 -5.33040702e-01 7.01666832e-01 6.50114000e-01 4.73007113e-01 3.10206771e-01 -9.40907970e-02 7.60242701e-01 -3.82690936e-01 -1.96445197e-01 2.86144279e-02 6.88359857e-01 -2.66947187e-02 -1.47461200e+00 7.34054297e-02 3.53312969e-01 -5.06073833e-01 -7.10061908e-01 -4.50273186e-01 1.22447896e+00 6.90984502e-02 7.65232682e-01 -5.17423093e-01 2.12902315e-02 4.60887283e-01 3.24044317e-01 5.82346141e-01 -9.01379168e-01 -4.37605917e-01 -6.51870370e-01 5.75801671e-01 -5.41270792e-01 -4.22989458e-01 -2.30137229e-01 -1.13025427e+00 -2.78786480e-01 -3.48506808e-01 1.82773307e-01 4.62119460e-01 1.22397459e+00 3.00309330e-01 3.06016296e-01 -1.78173751e-01 -8.49859789e-03 -1.54227227e-01 -8.19821894e-01 -6.21678710e-01 4.75243628e-01 -2.01636955e-01 -4.81268644e-01 5.44121899e-02 3.49324346e-01]
[9.409282684326172, 8.531822204589844]
86478e2b-bd55-40b9-a428-7f9cecd1d154
randomrooms-unsupervised-pre-training-from
2108.07794
null
https://arxiv.org/abs/2108.07794v1
https://arxiv.org/pdf/2108.07794v1.pdf
RandomRooms: Unsupervised Pre-training from Synthetic Shapes and Randomized Layouts for 3D Object Detection
3D point cloud understanding has made great progress in recent years. However, one major bottleneck is the scarcity of annotated real datasets, especially compared to 2D object detection tasks, since a large amount of labor is involved in annotating the real scans of a scene. A promising solution to this problem is to make better use of the synthetic dataset, which consists of CAD object models, to boost the learning on real datasets. This can be achieved by the pre-training and fine-tuning procedure. However, recent work on 3D pre-training exhibits failure when transfer features learned on synthetic objects to other real-world applications. In this work, we put forward a new method called RandomRooms to accomplish this objective. In particular, we propose to generate random layouts of a scene by making use of the objects in the synthetic CAD dataset and learn the 3D scene representation by applying object-level contrastive learning on two random scenes generated from the same set of synthetic objects. The model pre-trained in this way can serve as a better initialization when later fine-tuning on the 3D object detection task. Empirically, we show consistent improvement in downstream 3D detection tasks on several base models, especially when less training data are used, which strongly demonstrates the effectiveness and generalization of our method. Benefiting from the rich semantic knowledge and diverse objects from synthetic data, our method establishes the new state-of-the-art on widely-used 3D detection benchmarks ScanNetV2 and SUN RGB-D. We expect our attempt to provide a new perspective for bridging object and scene-level 3D understanding.
['Jie zhou', 'Cho-Jui Hsieh', 'Jiwen Lu', 'Yi Wei', 'Benlin Liu', 'Yongming Rao']
2021-08-17
null
http://openaccess.thecvf.com//content/ICCV2021/html/Rao_RandomRooms_Unsupervised_Pre-Training_From_Synthetic_Shapes_and_Randomized_Layouts_for_ICCV_2021_paper.html
http://openaccess.thecvf.com//content/ICCV2021/papers/Rao_RandomRooms_Unsupervised_Pre-Training_From_Synthetic_Shapes_and_Randomized_Layouts_for_ICCV_2021_paper.pdf
iccv-2021-1
['unsupervised-pre-training']
['methodology']
[ 4.23914462e-01 5.28129227e-02 7.54387155e-02 -4.84281570e-01 -6.77585483e-01 -6.79717720e-01 7.93563068e-01 2.91882977e-02 -2.73820609e-01 2.60377616e-01 -3.41240644e-01 -2.97896713e-01 1.55514270e-01 -9.60954428e-01 -1.01132858e+00 -4.08508301e-01 1.33542627e-01 9.60014999e-01 7.69732177e-01 -3.99279088e-01 2.94453740e-01 9.28505242e-01 -1.77687359e+00 1.91329956e-01 7.56078362e-01 9.32775259e-01 7.36842990e-01 2.97582030e-01 -4.96546984e-01 3.36807311e-01 -3.92636061e-01 -8.64891484e-02 6.81285262e-01 -1.79894373e-01 -6.99470341e-01 5.51157534e-01 4.72726703e-01 -3.01382303e-01 9.49195027e-02 8.18641365e-01 4.84569222e-01 -2.61785220e-02 5.99331260e-01 -1.33313334e+00 -4.64388020e-02 2.77819932e-01 -7.92564929e-01 -8.76507014e-02 2.80443460e-01 4.97003913e-01 6.95723355e-01 -9.68148410e-01 7.57899582e-01 1.40669107e+00 5.83175659e-01 5.85497558e-01 -1.34493601e+00 -6.66576922e-01 1.26421675e-01 -6.21428806e-03 -1.31129718e+00 -1.18840791e-01 9.71816838e-01 -4.87277597e-01 8.07555020e-01 4.40522432e-02 6.77744865e-01 1.06984532e+00 -4.42952842e-01 9.13461745e-01 1.13218725e+00 -6.38867915e-01 3.55061144e-01 2.93422192e-01 -1.44547448e-01 6.18352234e-01 2.38800734e-01 2.10166931e-01 -2.47377232e-01 1.83761552e-01 8.99864197e-01 -1.57707497e-01 -1.10950552e-01 -1.05224240e+00 -1.25219750e+00 6.81590259e-01 6.99042082e-01 3.84665728e-02 -3.37319732e-01 1.11038812e-01 2.10718840e-01 5.56169488e-02 6.09631240e-01 6.68821275e-01 -4.94594425e-01 1.66607156e-01 -8.74441803e-01 3.26353520e-01 5.13862431e-01 1.01368058e+00 9.62093771e-01 -9.53380764e-02 4.01809253e-02 6.88503683e-01 1.74543008e-01 5.93587458e-01 -4.43813577e-02 -8.55494261e-01 5.63673615e-01 9.56264436e-01 6.42431751e-02 -5.76480925e-01 -4.29118127e-01 -4.31143880e-01 -5.53051770e-01 7.32963562e-01 5.88202178e-01 2.69020021e-01 -1.35438251e+00 1.43655062e+00 6.01311982e-01 8.15527812e-02 -1.41250893e-01 9.51439559e-01 6.04390204e-01 5.90098917e-01 -7.01528564e-02 3.98248464e-01 1.09520996e+00 -7.83552051e-01 1.72884122e-01 -5.10391891e-01 5.44046938e-01 -7.52715170e-01 1.26798558e+00 2.44564816e-01 -9.45862949e-01 -8.43247116e-01 -9.82957363e-01 1.23442709e-01 -4.66501117e-01 1.81604907e-01 7.21879542e-01 6.76578462e-01 -6.39895141e-01 3.18111360e-01 -8.45577598e-01 -5.48192263e-01 8.44047725e-01 2.86789656e-01 -3.99698615e-01 -4.93005544e-01 -6.76143527e-01 9.24409866e-01 6.55177772e-01 -2.39484161e-02 -1.13100326e+00 -8.93715382e-01 -7.21315861e-01 -1.72003388e-01 7.36969650e-01 -9.25650537e-01 1.08985519e+00 -6.69217527e-01 -1.19411778e+00 1.32817006e+00 3.86185706e-01 -2.22959206e-01 8.00089896e-01 -7.24919662e-02 1.69369593e-01 -2.77372841e-02 1.88790068e-01 1.03514433e+00 8.10399830e-01 -1.64940751e+00 -6.54422641e-01 -4.32404220e-01 2.69172698e-01 1.06173642e-01 1.26016125e-01 -2.74139643e-01 -6.58906281e-01 -2.84779280e-01 3.21038187e-01 -9.74973917e-01 -4.92639601e-01 1.91091508e-01 -4.68398780e-01 -1.19479924e-01 8.45214546e-01 9.02696103e-02 3.86956006e-01 -2.05294394e+00 -4.07977663e-02 1.95656180e-01 -9.09698606e-02 3.89714837e-01 -2.58438975e-01 3.46872211e-01 -1.47502095e-01 9.93158817e-02 -3.45103592e-01 -4.16297078e-01 -6.23880476e-02 2.35838339e-01 -3.78446966e-01 3.26426953e-01 6.68454289e-01 8.39391232e-01 -9.97997344e-01 -4.20192301e-01 5.97018600e-01 2.66977012e-01 -6.85882688e-01 3.39715302e-01 -7.03789115e-01 6.47165656e-01 -5.93684912e-01 6.31606698e-01 1.03857613e+00 -2.86788464e-01 -2.49149352e-01 -2.24140823e-01 -1.06507547e-01 2.90729582e-01 -1.48087072e+00 2.08719420e+00 -5.60481071e-01 4.59409416e-01 -2.14061975e-01 -1.06468034e+00 9.48037505e-01 -1.61878705e-01 3.10390234e-01 -5.97362876e-01 8.95620808e-02 2.86781460e-01 -1.06319264e-01 -4.98572886e-01 1.98162124e-01 -1.60857424e-01 -1.90736130e-02 3.56613785e-01 -2.14372128e-02 -1.14194524e+00 2.48867109e-01 1.75190955e-01 9.93423104e-01 5.89520276e-01 1.92615092e-01 -8.27672053e-03 4.05861557e-01 4.81883258e-01 2.09310502e-01 9.10993099e-01 2.13667050e-01 8.83605003e-01 3.72095883e-01 -5.08326769e-01 -1.29889858e+00 -1.09551954e+00 -1.75216600e-01 6.63810253e-01 3.35104376e-01 -1.35176435e-01 -5.93480885e-01 -9.77785587e-01 1.77517638e-01 7.60186851e-01 -5.60445547e-01 -7.58473948e-02 -6.36785269e-01 -7.06599474e-01 3.43562514e-01 5.23760855e-01 5.87255716e-01 -9.96943176e-01 -9.82092023e-01 6.69642836e-02 1.57232493e-01 -1.29588878e+00 9.86576155e-02 4.51103598e-01 -8.61802816e-01 -1.04963195e+00 -4.90955204e-01 -5.33364356e-01 7.18136191e-01 6.65375590e-01 1.42101312e+00 9.82441753e-02 -5.22996426e-01 2.38483906e-01 -4.13175374e-01 -6.07953131e-01 -5.99253237e-01 2.32001781e-01 -1.37519702e-01 -1.62675887e-01 9.19132680e-02 -4.91458386e-01 -5.60311377e-01 5.19782186e-01 -1.05253994e+00 4.82799441e-01 8.19786549e-01 6.92375839e-01 6.53839409e-01 1.13292988e-02 2.96683222e-01 -8.96927357e-01 -4.49878387e-02 -2.74017394e-01 -8.40112746e-01 1.18242331e-01 -3.21771711e-01 2.10726932e-01 3.24621141e-01 -3.26017886e-01 -1.10715008e+00 5.44823170e-01 -2.25272447e-01 -5.79219341e-01 -4.95004267e-01 3.00250165e-02 -3.32112104e-01 -1.22611456e-01 8.67812753e-01 6.10017851e-02 -1.76742509e-01 -5.65234840e-01 5.68268657e-01 4.46761250e-01 3.50640535e-01 -6.83295667e-01 1.17740059e+00 6.68130875e-01 2.40074635e-01 -8.07986617e-01 -9.54738498e-01 -5.44483066e-01 -8.89822900e-01 -1.47645786e-01 7.56906033e-01 -8.37443233e-01 -2.33300388e-01 3.53022754e-01 -1.21227324e+00 -4.93379086e-01 -5.25522768e-01 3.05072308e-01 -6.57171547e-01 9.72476676e-02 5.44550866e-02 -7.38062441e-01 1.02621011e-01 -1.19805706e+00 1.60203803e+00 -1.07376285e-01 1.67873472e-01 -5.92486322e-01 -1.36414543e-01 2.50345767e-01 2.10221425e-01 4.60268885e-01 1.04806876e+00 -3.67484689e-01 -1.15472019e+00 -2.51567245e-01 -4.59536940e-01 4.47356611e-01 7.15741441e-02 -1.61873281e-01 -1.08389580e+00 -1.03765195e-02 -9.25601795e-02 -5.24922252e-01 8.40316772e-01 1.01426855e-01 1.30363667e+00 5.17217875e-01 -5.27300656e-01 5.90546012e-01 1.53769827e+00 -1.94416672e-01 3.98677677e-01 3.81333351e-01 6.95906281e-01 6.75314367e-01 8.97156298e-01 3.04927230e-01 1.39459103e-01 8.86408865e-01 8.71167958e-01 -4.13795948e-01 -5.12151599e-01 -4.18696314e-01 -1.68286219e-01 4.48491378e-03 -1.56702250e-01 -2.27309421e-01 -1.31409216e+00 4.07089382e-01 -1.52880847e+00 -8.06989968e-01 -2.69985765e-01 2.20907855e+00 6.51461005e-01 5.85512280e-01 1.22854985e-01 2.32349277e-01 4.41453218e-01 -8.10459405e-02 -5.79025626e-01 3.43364805e-01 -1.45338535e-01 3.48929971e-01 4.99904901e-01 1.07400328e-01 -1.04389262e+00 1.04304206e+00 5.09771442e+00 8.72130573e-01 -1.18703699e+00 -1.07231282e-01 5.31852007e-01 9.73543823e-02 -2.19373897e-01 8.97145271e-02 -9.82155979e-01 2.71371961e-01 3.56486976e-01 1.88474819e-01 1.11661240e-01 1.10927081e+00 2.22718209e-01 -1.92774415e-01 -1.40604353e+00 1.06993341e+00 -8.94164592e-02 -1.47008348e+00 1.77635878e-01 3.46019096e-03 7.92439580e-01 3.21282268e-01 -2.43807897e-01 2.56432384e-01 2.00480804e-01 -7.19487250e-01 9.86602962e-01 1.85604885e-01 7.55778909e-01 -4.30907726e-01 5.80391169e-01 7.87170112e-01 -1.09688163e+00 6.68435842e-02 -4.94440138e-01 8.94456953e-02 2.46646926e-01 7.41382241e-01 -1.46185243e+00 5.64233363e-01 5.87513983e-01 5.44921041e-01 -8.14754546e-01 1.27693939e+00 -3.74351978e-01 3.60730320e-01 -6.30343258e-01 9.58681926e-02 2.86391258e-01 2.15782747e-02 4.20474678e-01 1.13386226e+00 2.18370125e-01 -6.13009259e-02 3.65813017e-01 1.15823042e+00 -8.60803295e-03 -1.44320369e-01 -8.27610552e-01 2.36555696e-01 2.65212387e-01 1.19989109e+00 -1.12263179e+00 -1.82555124e-01 -2.94674069e-01 8.22902322e-01 3.66706520e-01 4.34473455e-02 -8.57300341e-01 -1.51558459e-01 3.60236049e-01 4.96864706e-01 4.01180893e-01 -3.85117799e-01 -3.65127325e-01 -9.98923421e-01 7.95560831e-04 -5.67020595e-01 -2.32089963e-03 -8.94986570e-01 -1.21975911e+00 5.58499277e-01 3.22135806e-01 -1.50359607e+00 -2.76456654e-01 -7.41310894e-01 -3.24108064e-01 7.60605037e-01 -1.66095626e+00 -1.38917339e+00 -7.90079296e-01 3.54434848e-01 8.30921113e-01 1.59154281e-01 5.39162815e-01 1.64936334e-01 -1.18514650e-01 6.49363101e-02 -2.98599869e-01 -1.53306201e-01 5.35790503e-01 -1.27629066e+00 7.55594134e-01 6.93098068e-01 3.64557236e-01 8.58677998e-02 5.31554937e-01 -4.46283132e-01 -1.38377810e+00 -1.16797400e+00 3.14966649e-01 -8.65514696e-01 3.37236673e-01 -8.15222442e-01 -8.22786570e-01 3.45933437e-01 -2.52724022e-01 2.37575293e-01 1.54056758e-01 -2.11806446e-01 -2.67978221e-01 -8.85033756e-02 -1.13192976e+00 4.00553286e-01 1.50911784e+00 -1.77099109e-01 -6.21622086e-01 4.99083012e-01 7.22132444e-01 -6.26786232e-01 -4.64874744e-01 7.27361977e-01 2.89272130e-01 -1.04024971e+00 1.40960407e+00 -5.64202309e-01 4.87102509e-01 -6.15385294e-01 -2.47326463e-01 -1.39232421e+00 -9.27903596e-03 -1.20807081e-01 3.19502860e-01 1.14638197e+00 3.45566899e-01 -2.47194648e-01 1.08520830e+00 3.19567114e-01 -3.18602532e-01 -6.25961304e-01 -6.90177619e-01 -9.28387403e-01 -1.41328290e-01 -7.48426080e-01 6.77899837e-01 7.46255755e-01 -8.02531481e-01 4.32225674e-01 5.94078712e-02 1.44014001e-01 6.71124101e-01 4.26417321e-01 1.52732396e+00 -1.48151004e+00 -2.70215482e-01 -3.31551790e-01 -5.14862597e-01 -1.22915351e+00 3.56107876e-02 -9.08051074e-01 1.98819429e-01 -1.47167540e+00 4.32079285e-02 -1.09419405e+00 9.32268426e-02 4.40058202e-01 -1.38513982e-01 4.34314460e-01 3.04072142e-01 9.76212323e-02 -4.02808964e-01 5.53996921e-01 1.47447860e+00 -2.30047390e-01 -2.92356938e-01 2.58690119e-01 -3.08672547e-01 7.81415939e-01 5.20889401e-01 -5.18998504e-01 -4.22784835e-01 -5.43019593e-01 9.13506523e-02 -2.39502162e-01 6.49948955e-01 -1.12424135e+00 -1.36608118e-02 -2.13011593e-01 4.95265543e-01 -9.85170543e-01 5.99711597e-01 -1.00732028e+00 1.38956411e-02 3.59551758e-01 3.61341611e-02 -4.15485442e-01 3.09286147e-01 4.28925455e-01 4.37282100e-02 -2.76021570e-01 9.12078977e-01 -5.41392386e-01 -1.09620345e+00 2.72827327e-01 2.93958157e-01 -9.23316553e-03 1.08651698e+00 -4.83839899e-01 -1.43557921e-01 1.01566359e-01 -5.30868530e-01 1.90536991e-01 7.48960197e-01 5.16355634e-01 5.58701575e-01 -1.02761817e+00 -7.54390717e-01 5.51221371e-01 5.18333316e-01 6.57147348e-01 5.16223982e-02 4.55052733e-01 -5.99039137e-01 2.54296303e-01 -2.92720795e-01 -1.13893688e+00 -9.33638573e-01 6.28093362e-01 3.45442444e-01 -5.77744842e-02 -6.52106881e-01 8.21149945e-01 5.02907455e-01 -5.25263071e-01 1.19656727e-01 -5.15201688e-01 2.94076413e-01 -1.34650812e-01 1.49063364e-01 1.45726785e-01 4.15625244e-01 -2.58288711e-01 -4.64079112e-01 7.50634074e-01 8.34816620e-02 2.79533602e-02 1.60546494e+00 8.03183839e-02 2.41843313e-01 3.44935745e-01 9.80519533e-01 -9.55040827e-02 -1.42206788e+00 -2.06939414e-01 9.53540206e-02 -8.47847283e-01 -4.16661426e-02 -8.19111466e-01 -8.83855760e-01 9.46407855e-01 7.22774625e-01 1.61974967e-01 9.59655404e-01 5.09135425e-01 2.39850923e-01 4.37525719e-01 8.27141106e-01 -7.09500670e-01 2.73619026e-01 2.83693284e-01 8.33048224e-01 -1.66183078e+00 -3.36544365e-02 -7.34016538e-01 -2.89224774e-01 1.07109106e+00 7.44868040e-01 -2.60056198e-01 3.67652386e-01 1.53276771e-01 -4.12376560e-02 -4.22670066e-01 -3.24149460e-01 -5.16036212e-01 2.59858787e-01 7.41214871e-01 7.45410621e-02 -1.15489751e-01 2.10941240e-01 7.56184608e-02 -1.20989338e-01 -9.29229259e-02 3.95010322e-01 9.25966620e-01 -5.10858297e-01 -1.43703043e+00 -3.61299306e-01 2.02748433e-01 1.25076056e-01 2.57349849e-01 -4.13339406e-01 1.16813147e+00 4.40723747e-01 5.73769629e-01 -1.12926429e-02 -1.58962503e-01 7.53724813e-01 -7.50775486e-02 8.13558877e-01 -1.10283327e+00 -2.29934692e-01 -1.65459067e-01 -3.79947424e-02 -6.12252474e-01 -5.05999267e-01 -6.54301763e-01 -1.12723565e+00 2.10764751e-01 -5.20240486e-01 -2.20646217e-01 1.05505931e+00 7.74951577e-01 2.53291279e-01 5.03992617e-01 7.07877755e-01 -1.48354876e+00 -5.33665419e-01 -7.90159643e-01 -3.04561108e-01 6.62454963e-01 1.88125014e-01 -9.90715384e-01 -1.79825947e-01 -2.45854887e-03]
[8.13074016571045, -2.8414418697357178]
bd8906f8-c7e8-400c-93fb-67607a2a4292
news-clustering-approach-based-on-discourse
null
null
https://aclanthology.org/W15-4503
https://aclanthology.org/W15-4503.pdf
News clustering approach based on discourse text structure
null
['Tatyana Makhalova', 'Boris Galitsky', 'Dmitry Ilvovsky']
2015-07-01
null
null
null
ws-2015-7
['text-clustering']
['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]
[-7.395345211029053, 3.6547932624816895]
a39a46c5-9c60-422c-9027-8f51d67c5be5
spcolor-semantic-prior-guided-exemplar-based
2304.06255
null
https://arxiv.org/abs/2304.06255v2
https://arxiv.org/pdf/2304.06255v2.pdf
SPColor: Semantic Prior Guided Exemplar-based Image Colorization
Exemplar-based image colorization aims to colorize a target grayscale image based on a color reference image, and the key is to establish accurate pixel-level semantic correspondence between these two images. Previous methods search for correspondence across the entire reference image, and this type of global matching is easy to get mismatch. We summarize the difficulties in two aspects: (1) When the reference image only contains a part of objects related to target image, improper correspondence will be established in unrelated regions. (2) It is prone to get mismatch in regions where the shape or texture of the object is easily confused. To overcome these issues, we propose SPColor, a semantic prior guided exemplar-based image colorization framework. Different from previous methods, SPColor first coarsely classifies pixels of the reference and target images to several pseudo-classes under the guidance of semantic prior, then the correspondences are only established locally between the pixels in the same class via the newly designed semantic prior guided correspondence network. In this way, improper correspondence between different semantic classes is explicitly excluded, and the mismatch is obviously alleviated. Besides, to better reserve the color from reference, a similarity masked perceptual loss is designed. Noting that the carefully designed SPColor utilizes the semantic prior provided by an unsupervised segmentation model, which is free for additional manual semantic annotations. Experiments demonstrate that our model outperforms recent state-of-the-art methods both quantitatively and qualitatively on public dataset.
['Yue Zhang', 'Yu Zhang', 'Mingdao Wang', 'Xianlin Zhang', 'Xueming Li', 'Siqi Chen']
2023-04-13
null
null
null
null
['colorization', 'semantic-correspondence']
['computer-vision', 'computer-vision']
[ 4.62743312e-01 -1.07098907e-01 -1.25531286e-01 -3.12102705e-01 -6.13967776e-01 -5.55119932e-01 3.20686877e-01 -1.42277882e-01 -3.49420577e-01 5.54967880e-01 -2.79665262e-01 4.60228138e-02 9.88789946e-02 -9.58769500e-01 -5.47128558e-01 -8.04402709e-01 7.24275172e-01 1.79998234e-01 6.30601108e-01 -2.59813637e-01 4.45457041e-01 4.92381573e-01 -1.38405478e+00 8.15728307e-02 1.12798178e+00 1.01457429e+00 3.81852567e-01 -2.51245424e-02 -5.37394285e-01 1.79722384e-01 -6.07900202e-01 -2.39889979e-01 4.82795507e-01 -6.45607293e-01 -8.96836817e-01 5.02960980e-01 6.52547002e-01 -4.82889004e-02 -4.43449691e-02 1.71159756e+00 -4.08862792e-02 1.97967350e-01 4.68076825e-01 -1.45840919e+00 -9.39315379e-01 1.32836252e-01 -9.49331403e-01 -1.89452469e-01 1.61186904e-01 9.84554663e-02 8.31651390e-01 -8.12034845e-01 4.63216811e-01 1.18876851e+00 3.85749698e-01 4.26228881e-01 -1.27021205e+00 -8.74523461e-01 5.21241426e-01 4.79742959e-02 -1.61342311e+00 -4.18581404e-02 1.12975061e+00 -1.09645568e-01 2.89184358e-02 2.31724054e-01 7.84001946e-01 6.27035141e-01 -2.79207915e-01 6.05995417e-01 1.44337773e+00 -3.03982794e-01 3.34531903e-01 1.36442289e-01 3.86378095e-02 7.51036942e-01 2.62546659e-01 1.70360819e-01 -2.88698107e-01 1.64139256e-01 9.53548491e-01 3.19729209e-01 -4.81582224e-01 -6.01478875e-01 -1.31037676e+00 3.94350141e-01 7.36956239e-01 4.82130796e-01 3.09499558e-02 3.03526055e-02 -4.49856780e-02 -8.33335221e-02 2.30140895e-01 3.26566696e-01 -3.48829716e-01 2.31731802e-01 -1.06154263e+00 -3.16295296e-01 3.30908477e-01 1.12351429e+00 1.41710687e+00 -3.99822928e-02 -5.85075021e-02 8.99578750e-01 3.17706347e-01 4.78560239e-01 2.46022657e-01 -1.09324348e+00 2.95672268e-01 9.29236114e-01 2.19901860e-01 -1.25592220e+00 -1.46829626e-02 -3.97505790e-01 -8.92083168e-01 4.76252258e-01 6.33988976e-01 3.08729738e-01 -1.17948020e+00 1.55974174e+00 3.70528668e-01 2.73172945e-01 2.48732194e-02 1.12082410e+00 8.41516554e-01 4.94161725e-01 9.92456973e-02 -2.20153984e-02 1.28462577e+00 -1.19402826e+00 -6.02542579e-01 -4.59830493e-01 -6.90211803e-02 -8.87815833e-01 1.39179945e+00 2.88939267e-01 -8.54320347e-01 -7.83670902e-01 -1.19215000e+00 3.57079022e-02 -5.81570745e-01 1.80728570e-01 6.18644476e-01 7.17863441e-01 -1.03464186e+00 4.09274310e-01 -3.78994673e-01 -4.30436641e-01 2.30891317e-01 -3.18345428e-02 -3.05573136e-01 -2.98646837e-01 -1.11748433e+00 6.12328827e-01 6.00179911e-01 3.02838534e-01 -4.68836844e-01 -4.97260571e-01 -6.81881845e-01 -8.14441890e-02 3.74837816e-01 -4.93811667e-01 5.69504797e-01 -1.56944501e+00 -1.46094942e+00 1.04013562e+00 -1.27050832e-01 2.50923008e-01 5.97052336e-01 1.08497441e-01 -6.24487996e-01 2.69839138e-01 2.81404376e-01 7.89926589e-01 9.30108249e-01 -1.86631000e+00 -7.86565483e-01 -1.94441602e-01 1.74717471e-01 3.11181009e-01 -1.48846000e-01 -8.20629150e-02 -1.37055552e+00 -8.88508499e-01 6.66876256e-01 -6.80413544e-01 -1.44139692e-01 3.19129169e-01 -5.69932938e-01 9.16884318e-02 9.43826675e-01 -3.67499650e-01 9.91165578e-01 -2.30440521e+00 -2.57710338e-01 5.37118852e-01 2.07718369e-02 1.31507292e-01 -3.53218734e-01 -2.40414776e-02 -2.35191509e-01 1.53286546e-01 -6.48483992e-01 8.88041034e-02 -2.64424533e-01 3.12314212e-01 -1.33793637e-01 3.90498579e-01 5.16344942e-02 7.13601470e-01 -1.12594640e+00 -7.44021297e-01 2.52316177e-01 2.34662727e-01 -2.48641595e-01 8.71169195e-02 -2.57276539e-02 5.00991523e-01 -5.54437399e-01 8.92052412e-01 1.18152630e+00 -2.05954731e-01 -7.61870518e-02 -6.89968050e-01 -1.13918502e-02 -2.78602719e-01 -1.50127113e+00 1.88424897e+00 -1.51875228e-01 2.26969242e-01 4.97392379e-02 -9.73593414e-01 1.08958709e+00 -1.64262086e-01 4.05123681e-01 -7.98158705e-01 -7.83454552e-02 9.99065787e-02 -2.52817333e-01 -1.12015188e-01 4.33185279e-01 2.94857286e-02 9.70735177e-02 4.61014897e-01 -3.41613024e-01 -4.96887058e-01 1.17708012e-01 1.57639384e-01 3.86555344e-01 4.35370743e-01 -3.23032476e-02 -2.46854916e-01 6.84696674e-01 1.38783097e-01 8.95719945e-01 6.96892917e-01 -3.66023600e-01 1.11855650e+00 1.30221903e-01 -2.01797724e-01 -6.89959824e-01 -1.20994699e+00 -1.96557164e-01 9.27171469e-01 1.19208372e+00 -6.35422440e-03 -8.12300801e-01 -7.88382828e-01 -1.96435452e-01 4.01610106e-01 -6.79475784e-01 -4.29931097e-02 -4.48441535e-01 -7.67468810e-01 1.45577982e-01 4.29297209e-01 1.00643647e+00 -9.95016456e-01 -1.54116124e-01 -5.40761948e-02 -2.17276379e-01 -9.46303248e-01 -8.93743873e-01 -1.44658029e-01 -7.99332082e-01 -1.38791740e+00 -8.60879421e-01 -9.42370594e-01 1.27573216e+00 7.70473540e-01 1.10096216e+00 4.40541059e-01 -1.92613542e-01 2.80889541e-01 -4.10299659e-01 2.56802708e-01 -1.68191597e-01 -5.07070482e-01 -4.49228108e-01 3.65192652e-01 2.81075507e-01 -2.07518935e-01 -9.69694972e-01 6.59855962e-01 -1.15087140e+00 4.09300596e-01 6.09427273e-01 8.77517819e-01 9.72559571e-01 2.26421252e-01 1.35395736e-01 -9.54022110e-01 2.08523169e-01 -1.88502353e-02 -5.59877396e-01 7.10111737e-01 -6.78667903e-01 -6.76528290e-02 4.35095638e-01 -3.06404322e-01 -1.17133451e+00 2.71617353e-01 1.95539266e-01 -2.54285365e-01 -4.61962879e-01 7.14791864e-02 -5.14344335e-01 -3.53574485e-01 2.78892726e-01 3.77479017e-01 -1.00360848e-01 -4.04613078e-01 7.34650254e-01 4.66702700e-01 7.91219234e-01 -6.94192827e-01 1.00750256e+00 7.64936328e-01 -2.35843763e-01 -3.76066208e-01 -7.55949318e-01 -5.63688397e-01 -8.31537664e-01 -2.63580889e-01 9.01825905e-01 -6.52184784e-01 -2.09027112e-01 5.13618410e-01 -9.73043680e-01 -2.65688270e-01 -1.19767122e-01 1.98009729e-01 -3.40430439e-01 7.16046691e-01 -3.75444531e-01 -3.56193066e-01 -9.18795690e-02 -1.18439186e+00 1.06538463e+00 6.62141681e-01 1.67842329e-01 -7.97133863e-01 -3.50708216e-01 3.17250699e-01 2.70861477e-01 1.16714269e-01 8.70483279e-01 -1.29719466e-01 -7.32713282e-01 -5.83208585e-03 -9.20857489e-01 4.06141579e-01 6.63059175e-01 3.14942390e-01 -6.92546546e-01 -1.38175473e-01 -2.40687847e-01 1.15982369e-01 8.45871270e-01 1.23243622e-01 1.32716668e+00 -4.74330932e-02 -3.31339270e-01 7.85747468e-01 1.66260791e+00 3.24367851e-01 8.04480791e-01 4.14943904e-01 9.15417075e-01 6.23840094e-01 8.75621438e-01 -4.53335866e-02 1.71455771e-01 5.71763337e-01 3.35429877e-01 -8.77671003e-01 -3.95003229e-01 -2.12505162e-01 1.36525109e-02 3.29931796e-01 7.93522745e-02 1.51136309e-01 -4.94891435e-01 4.04802293e-01 -1.74565351e+00 -6.15258455e-01 -1.37532011e-01 2.37907243e+00 1.01091480e+00 -4.12273817e-02 -4.53615859e-02 -3.81164290e-02 1.11859858e+00 8.92680511e-02 -6.79909468e-01 6.73485994e-02 -3.79807919e-01 1.76910430e-01 5.84779143e-01 2.47082695e-01 -1.04492164e+00 1.13115406e+00 5.91663647e+00 1.15691650e+00 -1.13614142e+00 5.69335781e-02 1.00193501e+00 4.67204988e-01 -4.01959807e-01 2.84988403e-01 -3.97988886e-01 6.42984271e-01 -2.18073070e-01 1.25355989e-01 4.51096326e-01 6.58666730e-01 3.07645928e-02 -4.75560606e-01 -8.32671046e-01 1.08418369e+00 9.04071257e-02 -9.61865783e-01 2.09347904e-01 -4.00780678e-01 1.00821090e+00 -4.62016284e-01 1.82793602e-01 -1.08521171e-01 2.60694295e-01 -7.90306330e-01 8.15433443e-01 5.49405515e-01 9.76189375e-01 -6.74711227e-01 5.36237955e-01 -1.40240088e-01 -1.43100369e+00 3.34444344e-01 -4.57806259e-01 5.06632686e-01 -6.47765994e-02 5.01366913e-01 -6.44326508e-02 6.53092146e-01 8.72165442e-01 7.60695219e-01 -7.91573703e-01 1.10535252e+00 -6.09755814e-01 2.01040447e-01 -1.73617586e-01 3.81999701e-01 2.12378293e-01 -7.85118222e-01 1.38454124e-01 1.04380655e+00 1.50112420e-01 1.19183972e-01 3.91829491e-01 1.39889741e+00 1.27080679e-01 3.94478105e-02 -5.23861535e-02 2.45245323e-01 5.40036082e-01 1.37492394e+00 -1.21468616e+00 -3.63705158e-01 -5.16184151e-01 1.53534579e+00 -4.05434100e-03 9.45849776e-01 -8.14131021e-01 -5.66250622e-01 3.62349927e-01 -1.40211806e-01 1.57002389e-01 -9.34058726e-02 -4.89917487e-01 -1.07752585e+00 -5.18147275e-02 -5.79201818e-01 3.66892397e-01 -1.09223735e+00 -1.44810355e+00 5.77307701e-01 -1.64554447e-01 -1.59192204e+00 4.09414649e-01 -4.36984539e-01 -8.09191048e-01 1.08132935e+00 -1.76381898e+00 -1.20082533e+00 -7.12784231e-01 7.85119057e-01 3.30775648e-01 2.60139197e-01 5.94265223e-01 2.19653249e-01 -6.20158792e-01 5.34348190e-01 1.19082488e-01 3.46556097e-01 9.36306238e-01 -1.22971189e+00 -2.56510475e-03 1.09526134e+00 3.78567576e-02 7.50379741e-01 4.03810292e-01 -6.33521736e-01 -9.26439881e-01 -1.16816795e+00 2.53377616e-01 -1.13342129e-01 4.51022536e-01 -8.68732110e-02 -1.06117260e+00 3.40896472e-02 1.33361548e-01 2.81293958e-01 3.09515893e-01 -3.10065120e-01 -4.88106728e-01 -4.57300514e-01 -1.06514204e+00 6.92368627e-01 8.69650126e-01 -5.44265568e-01 -4.77022052e-01 1.78742215e-01 5.19111216e-01 -2.76311278e-01 -5.38153589e-01 3.70115161e-01 4.23673838e-01 -1.16950500e+00 1.14833665e+00 -8.84649903e-02 2.58415848e-01 -1.05461586e+00 1.75307915e-02 -1.17490625e+00 -1.44383594e-01 -2.90900618e-01 5.84508598e-01 1.41530657e+00 2.27407590e-01 -6.76211119e-01 8.21798980e-01 8.19458306e-01 -2.65496355e-02 -4.64676738e-01 -6.34757698e-01 -8.59699905e-01 -2.83802506e-02 -8.44600797e-02 7.96086550e-01 1.07973564e+00 -4.19265062e-01 -1.93323553e-01 -4.22906205e-02 2.27877781e-01 6.82572901e-01 6.98601186e-01 6.43867910e-01 -1.10534024e+00 -9.86136571e-02 -8.13483894e-01 -1.84728578e-01 -1.03226578e+00 -5.61109651e-03 -5.87623417e-01 2.87885636e-01 -1.57974112e+00 4.61296022e-01 -9.51173842e-01 -6.59323156e-01 6.02873921e-01 -5.40214241e-01 7.24461854e-01 1.23558082e-01 4.04492170e-01 -6.39202178e-01 5.29845893e-01 1.61268115e+00 -4.23605710e-01 -1.72150552e-01 -1.95446074e-01 -8.69912446e-01 7.09394693e-01 7.36457348e-01 -2.59845734e-01 -3.06091219e-01 -2.01879337e-01 -8.91227871e-02 -2.60586768e-01 5.24892151e-01 -8.45570922e-01 2.11828962e-01 -6.04228616e-01 5.14246285e-01 -4.72443163e-01 8.39408487e-02 -1.05211842e+00 1.94790363e-01 1.65289417e-01 -1.12687059e-01 -3.07523400e-01 -8.78657587e-03 7.48374403e-01 -4.27132070e-01 -2.56435543e-01 1.09326279e+00 -2.33116642e-01 -1.17683601e+00 4.00221407e-01 1.13453627e-01 7.70992413e-02 9.80468214e-01 -8.18415523e-01 -3.49120617e-01 -1.53960451e-01 -4.37161028e-01 4.76680174e-02 9.95925486e-01 3.39449108e-01 7.17960954e-01 -1.34920037e+00 -3.20044339e-01 3.43987614e-01 4.72783297e-01 7.56858811e-02 3.29377919e-01 7.03328371e-01 -5.18571615e-01 3.59126925e-02 -3.32705468e-01 -4.95986819e-01 -1.03565753e+00 5.82117915e-01 4.77233559e-01 3.38290751e-01 -6.17727876e-01 7.61819422e-01 7.54557133e-01 -2.43861884e-01 1.17921695e-01 -3.97361964e-01 -1.67452142e-01 -3.42473179e-01 2.91139930e-01 4.20549214e-02 -1.24928333e-01 -8.01347733e-01 -2.80012012e-01 1.06250143e+00 8.29569027e-02 -8.30491781e-02 8.39621544e-01 -4.25995976e-01 -3.99528921e-01 1.37151092e-01 1.14080453e+00 1.82175919e-01 -1.50246811e+00 -4.81506675e-01 -2.63706744e-01 -8.43152523e-01 8.02866071e-02 -9.25354719e-01 -1.69151521e+00 7.97184289e-01 7.45872617e-01 -3.84236574e-02 1.51767945e+00 -5.92571162e-02 6.73821628e-01 -1.44781083e-01 4.68984097e-01 -1.35485423e+00 2.02622265e-01 2.10425436e-01 6.57677650e-01 -1.27475524e+00 7.39658698e-02 -8.20096135e-01 -6.98538363e-01 1.16401052e+00 9.12544906e-01 -7.42153376e-02 4.53602612e-01 -1.84979975e-01 2.33183712e-01 -1.24174200e-01 2.39039361e-01 -4.08125252e-01 6.27082169e-01 7.09023774e-01 1.72539681e-01 -7.31185004e-02 -2.19316855e-01 5.05093694e-01 2.10603163e-01 -4.57510322e-01 2.60842949e-01 5.89433432e-01 -3.58349323e-01 -1.20563555e+00 -5.63516259e-01 -3.29054631e-02 -4.86714691e-02 -2.98966378e-01 -4.51734871e-01 8.03009868e-01 3.19662333e-01 1.00709748e+00 2.24496916e-01 -1.46177456e-01 1.21509045e-01 -3.88652056e-01 3.02676737e-01 -4.74557489e-01 -1.94725513e-01 3.18890750e-01 -4.48953807e-01 -7.24099278e-01 -7.43697703e-01 -2.02679753e-01 -1.52114868e+00 1.01529188e-01 -4.07019615e-01 1.22200094e-01 4.05631036e-01 7.47487843e-01 1.55561283e-01 4.04135406e-01 6.70504093e-01 -6.87173903e-01 1.77988574e-01 -4.85915869e-01 -9.48958695e-01 8.06926131e-01 4.00511064e-02 -7.37806618e-01 -3.77011657e-01 2.50238210e-01]
[11.166094779968262, -1.229921579360962]
9f879f2c-2b2c-47a4-ab1a-e23996a63271
multiview-based-3d-scene-understanding-on
1812.01712
null
http://arxiv.org/abs/1812.01712v1
http://arxiv.org/pdf/1812.01712v1.pdf
Multiview Based 3D Scene Understanding On Partial Point Sets
Deep learning within the context of point clouds has gained much research interest in recent years mostly due to the promising results that have been achieved on a number of challenging benchmarks, such as 3D shape recognition and scene semantic segmentation. In many realistic settings however, snapshots of the environment are often taken from a single view, which only contains a partial set of the scene due to the field of view restriction of commodity cameras. 3D scene semantic understanding on partial point clouds is considered as a challenging task. In this work, we propose a processing approach for 3D point cloud data based on a multiview representation of the existing 360{\deg} point clouds. By fusing the original 360{\deg} point clouds and their corresponding 3D multiview representations as input data, a neural network is able to recognize partial point sets while improving the general performance on complete point sets, resulting in an overall increase of 31.9% and 4.3% in segmentation accuracy for partial and complete scene semantic understanding, respectively. This method can also be applied in a wider 3D recognition context such as 3D part segmentation.
['Miklas Strøm Kristoffersen', 'Pablo Martínez-Nuevo', 'Zhuang Fu', 'Sven Ewan Shepstone', 'Fabien Moutarde', 'Ye Zhu']
2018-11-30
null
null
null
null
['3d-shape-recognition', '3d-part-segmentation']
['computer-vision', 'computer-vision']
[ 3.07933420e-01 -2.01235209e-02 1.05648808e-01 -5.88877380e-01 -6.35611236e-01 -6.63159847e-01 5.21234453e-01 4.09639150e-01 -2.41639271e-01 -1.22261755e-01 -5.04280329e-01 1.88240819e-02 -3.15565728e-02 -8.36510539e-01 -1.08577859e+00 -3.97584379e-01 3.43979299e-01 9.37524140e-01 3.69273871e-01 4.03154315e-03 3.07950824e-01 1.01196849e+00 -1.88153994e+00 1.39549330e-01 7.98850358e-01 1.19153845e+00 5.76415896e-01 2.05860585e-01 -6.27779841e-01 -1.34089842e-01 -3.44405383e-01 -1.19846068e-01 5.90501487e-01 3.88159752e-01 -5.24267316e-01 6.85638249e-01 7.80889094e-01 -3.30951691e-01 1.99653041e-02 1.13533437e+00 1.22163400e-01 1.44751474e-01 3.99869114e-01 -1.19106102e+00 4.34220359e-02 1.88502111e-02 -6.54952347e-01 -3.09416920e-01 3.84798139e-01 -5.04358001e-02 8.51799488e-01 -9.31257665e-01 4.56875235e-01 1.34064376e+00 3.02006662e-01 1.75428644e-01 -1.14350688e+00 -4.86453295e-01 4.46629316e-01 1.61655515e-01 -1.27148223e+00 2.76076533e-02 1.07058394e+00 -6.06543243e-01 9.92974937e-01 1.14560425e-01 8.05410385e-01 5.71333051e-01 -2.77684361e-01 8.83546174e-01 1.04301012e+00 -1.09824635e-01 3.37087899e-01 2.61386726e-02 2.79254049e-01 2.64952809e-01 2.01160237e-01 -4.43169624e-01 -1.02883957e-01 1.49859145e-01 9.29922581e-01 5.73357463e-01 -1.81081533e-01 -9.45104718e-01 -1.16192973e+00 6.48809910e-01 7.42288470e-01 1.49489745e-01 -4.98483390e-01 -1.38101920e-01 2.19944030e-01 -9.69994292e-02 7.67009914e-01 8.75014067e-02 -5.02077878e-01 1.95765018e-01 -9.08506513e-01 3.27485025e-01 4.57503647e-01 1.06925642e+00 9.59476531e-01 -3.52325700e-02 4.52521890e-01 8.32719922e-01 2.39895970e-01 8.12684119e-01 -2.30580375e-01 -1.16530776e+00 6.80604339e-01 1.17895699e+00 1.24638148e-01 -9.35586631e-01 -4.73900646e-01 -3.65661532e-01 -7.37110794e-01 3.58106524e-01 3.63158941e-01 3.78051609e-01 -1.16802132e+00 1.18462396e+00 6.89741313e-01 9.09320787e-02 -1.25759706e-01 1.06575167e+00 8.59465599e-01 8.03550184e-01 -4.11132544e-01 1.68523327e-01 1.34798300e+00 -5.91882110e-01 -8.21807161e-02 -5.18888295e-01 3.66484106e-01 -7.43409038e-01 8.20175707e-01 4.22300100e-01 -9.44302559e-01 -6.50945961e-01 -9.49173868e-01 -1.09245978e-01 -3.61844689e-01 -1.01659246e-01 4.22229767e-01 2.77151108e-01 -8.92302811e-01 4.12786245e-01 -8.13053429e-01 -3.65869075e-01 7.37361133e-01 5.11070549e-01 -5.57137907e-01 -6.13608062e-01 -4.36985344e-01 5.46499431e-01 4.39260066e-01 3.69959287e-02 -6.91098511e-01 -8.46114576e-01 -8.17474961e-01 2.41243973e-01 6.02472603e-01 -7.68653333e-01 9.08817708e-01 -6.09105408e-01 -1.05047369e+00 1.10868084e+00 -1.29677489e-01 -1.43026590e-01 5.51716506e-01 -4.23465997e-01 2.67981380e-01 2.36127242e-01 8.48149136e-02 7.24374592e-01 6.63080752e-01 -1.58012497e+00 -5.88876843e-01 -1.17920709e+00 1.89093858e-01 4.83006328e-01 1.23606168e-01 -3.12537521e-01 -7.36678958e-01 -9.88185257e-02 7.69936264e-01 -1.03230190e+00 -3.38886023e-01 -6.20017983e-02 -4.33738768e-01 -2.11938858e-01 1.07982814e+00 -4.74307507e-01 6.73972443e-02 -2.05997825e+00 3.78273785e-01 6.87963841e-03 1.10851213e-01 2.85299927e-01 6.40486041e-03 1.95783004e-01 -7.86220506e-02 2.62842644e-02 -4.52516288e-01 -5.94360888e-01 -1.52474359e-01 2.80591190e-01 -3.17730695e-01 4.59941387e-01 4.24110666e-02 6.54723287e-01 -5.31593740e-01 -7.90564343e-02 7.88324416e-01 5.48454762e-01 -4.33496892e-01 2.88139343e-01 -6.20724022e-01 6.06719971e-01 -5.88201046e-01 7.11827099e-01 1.08549988e+00 -2.22293451e-01 -1.91519991e-01 -9.44147930e-02 -1.76519990e-01 -1.39240131e-01 -1.20620632e+00 2.12648416e+00 -4.31698889e-01 4.30516779e-01 2.44729921e-01 -1.01459849e+00 1.20184934e+00 2.18556628e-01 7.66992986e-01 -5.13346016e-01 7.33581185e-02 2.46910810e-01 -3.33729923e-01 -1.60117254e-01 4.14911419e-01 -1.36433586e-01 -7.14680329e-02 2.64833540e-01 -1.41194835e-01 -8.71800601e-01 -2.43290186e-01 -7.32226893e-02 6.75811589e-01 7.95693323e-02 -2.69966014e-02 1.38005733e-01 5.50590694e-01 2.40273401e-01 3.90245289e-01 3.22731704e-01 2.19490170e-01 8.51288497e-01 2.74773985e-01 -4.48020726e-01 -1.23411763e+00 -1.03502297e+00 -1.43948644e-01 2.61274725e-01 5.74484289e-01 -2.31839102e-02 -6.29358292e-01 -5.09804368e-01 1.87949330e-01 6.97341919e-01 -9.84589383e-02 1.36710420e-01 -5.40016711e-01 -3.63394946e-01 -1.68746322e-01 5.05677104e-01 4.94049877e-01 -7.75983810e-01 -8.14526141e-01 -2.44723465e-02 -9.36248451e-02 -1.64800680e+00 6.97026774e-02 -2.07536994e-03 -1.30972004e+00 -9.99135315e-01 -7.32975483e-01 -5.23143291e-01 6.80041790e-01 8.54147792e-01 9.60929871e-01 -1.63870737e-01 -8.91643390e-02 6.18969381e-01 -3.19569677e-01 -4.90156859e-01 -1.22369729e-01 -1.05919376e-01 -7.24288374e-02 2.74333060e-01 3.02312523e-01 -7.35524178e-01 -3.82605880e-01 3.38787764e-01 -9.64295089e-01 2.28229180e-01 4.85573769e-01 3.58192235e-01 9.25020874e-01 -1.56006545e-01 8.83963257e-02 -7.31432378e-01 -1.90939456e-01 -2.87995011e-01 -9.33874726e-01 -2.00770736e-01 -1.38826758e-01 -3.20428520e-01 4.03120995e-01 2.05134712e-02 -9.00821030e-01 2.91129321e-01 -1.25693932e-01 -1.18053937e+00 -7.93584704e-01 1.45606205e-01 -4.79481846e-01 3.57379802e-02 1.35673434e-01 2.70261228e-01 2.13551782e-02 -7.98885345e-01 2.87655860e-01 4.85436887e-01 2.21060276e-01 -3.30904424e-01 8.18043411e-01 7.16341674e-01 1.95654154e-01 -1.06442952e+00 -7.38886416e-01 -9.13299501e-01 -1.07732546e+00 -3.41812879e-01 1.02297139e+00 -1.06174815e+00 -6.87383950e-01 5.04200339e-01 -1.42302024e+00 -1.46189287e-01 -1.93379119e-01 4.34656143e-01 -6.59174621e-01 5.74682415e-01 -9.04818252e-02 -6.32982910e-01 -2.83009738e-01 -1.53295577e+00 1.58755338e+00 1.12662807e-01 1.82035685e-01 -6.81481063e-01 -4.08536345e-01 8.18878651e-01 -1.61037818e-01 4.59479481e-01 1.16476190e+00 -6.38276637e-01 -1.12446618e+00 -3.23721856e-01 -3.31382275e-01 4.66880620e-01 -5.71059249e-02 -1.85805589e-01 -1.04282546e+00 -6.58407584e-02 2.76360750e-01 -5.67070656e-02 6.09759212e-01 4.26478088e-01 1.09381115e+00 4.10448104e-01 -4.56333846e-01 6.46207392e-01 1.57901978e+00 2.52902687e-01 1.69299632e-01 4.96023707e-02 1.12803638e+00 8.23483348e-01 7.61645973e-01 4.03676033e-01 3.71137947e-01 9.45272565e-01 1.17114651e+00 -4.29466702e-02 1.50170490e-01 -8.19358155e-02 -2.05227226e-01 6.24903202e-01 -2.90573724e-02 -9.00103375e-02 -1.22695076e+00 5.84908307e-01 -1.66809905e+00 -5.64661622e-01 -3.89476836e-01 2.35749865e+00 9.10674874e-03 6.35889247e-02 -1.71107985e-02 2.48901218e-01 8.34554136e-01 3.47588480e-01 -8.95334244e-01 -1.86832264e-01 1.03978381e-01 -8.19014311e-02 3.54522735e-01 2.32423887e-01 -9.88214910e-01 8.09663415e-01 4.37570858e+00 6.72600627e-01 -1.14998996e+00 -2.00342819e-01 5.27490675e-01 2.36073714e-02 -2.39139870e-01 -1.11055970e-02 -7.58936524e-01 1.71834797e-01 3.19628447e-01 2.35416248e-01 1.28718823e-01 9.61971581e-01 1.37271643e-01 -1.28607661e-01 -1.28814507e+00 1.46179116e+00 1.34693384e-01 -1.14509416e+00 3.01350027e-01 3.82153898e-01 7.20272481e-01 3.05856705e-01 -8.79624113e-02 8.22778046e-03 -1.47035703e-01 -9.30579305e-01 6.54406250e-01 4.84010518e-01 4.75068241e-01 -8.60686004e-01 5.55373847e-01 1.03254509e+00 -1.04714775e+00 1.29388317e-01 -4.36774760e-01 3.16855758e-02 4.18051720e-01 9.01317477e-01 -8.94742191e-01 8.80299270e-01 8.08328450e-01 9.27243114e-01 -3.06172222e-01 1.17673171e+00 2.58505289e-02 1.79621354e-01 -7.23487437e-01 3.42889309e-01 2.90528774e-01 -6.56128466e-01 8.46720576e-01 5.96214294e-01 4.71347332e-01 2.05397993e-01 3.26816231e-01 1.06837571e+00 -1.14501171e-01 1.33795396e-03 -8.72188747e-01 3.10472578e-01 2.97963470e-01 1.29306912e+00 -9.60838437e-01 -3.30734462e-01 -4.90534633e-01 8.08611453e-01 1.73985079e-01 2.13651210e-01 -5.29207587e-01 3.99990939e-02 7.53426135e-01 9.33672860e-02 5.84792495e-01 -5.28281331e-01 -5.08044779e-01 -1.14594185e+00 3.33598405e-01 -4.63461071e-01 -9.37526748e-02 -9.88019407e-01 -9.88663852e-01 4.47475016e-01 2.47545585e-01 -1.37182462e+00 -2.58082245e-02 -5.51394522e-01 -3.76094550e-01 8.05079162e-01 -1.38109231e+00 -1.05890429e+00 -6.91819668e-01 4.39416409e-01 9.56264496e-01 1.79363698e-01 5.69622517e-01 1.10348798e-01 -2.68270016e-01 -2.07859129e-01 6.31037951e-02 -1.87265873e-01 1.30141318e-01 -1.06830978e+00 5.39842784e-01 5.87114394e-01 2.73180872e-01 1.64534405e-01 4.29756582e-01 -4.57571298e-01 -1.60524857e+00 -1.15187395e+00 4.60060328e-01 -5.97188890e-01 2.38766638e-03 -5.60274243e-01 -1.19444978e+00 6.33580863e-01 -2.33947486e-01 7.42671117e-02 2.98369884e-01 -1.12117574e-01 -2.19750896e-01 -1.84636027e-01 -1.11203229e+00 3.84201229e-01 1.16094983e+00 -2.65883625e-01 -4.77019876e-01 4.14226949e-01 7.77512133e-01 -7.35376358e-01 -9.66156602e-01 8.26377749e-01 2.01755568e-01 -1.07176161e+00 1.34244049e+00 -1.24245100e-01 3.38727474e-01 -3.17175597e-01 -4.64902401e-01 -1.21752465e+00 1.56058237e-01 1.37089401e-01 2.15026721e-01 1.05298221e+00 -3.97890806e-02 -5.68573833e-01 9.72278476e-01 5.74726462e-01 -3.89654309e-01 -6.48845136e-01 -1.07087052e+00 -5.45435131e-01 -8.38808566e-02 -8.71197522e-01 6.32884145e-01 7.46299624e-01 -8.68775487e-01 3.06992382e-01 1.49154902e-01 5.14094949e-01 7.86552608e-01 8.93604934e-01 1.12194312e+00 -1.66997266e+00 3.55422422e-02 -4.00246292e-01 -6.42214656e-01 -1.38785613e+00 2.59884030e-01 -1.04737413e+00 -1.75289646e-01 -1.72867227e+00 -3.70521359e-02 -5.51160574e-01 1.94950312e-01 1.57673448e-01 1.32427841e-01 1.27844915e-01 6.48235202e-01 2.60199010e-01 -4.46525365e-01 6.55913174e-01 1.22226489e+00 -2.56587565e-01 -2.44022474e-01 3.29916716e-01 -1.96397856e-01 9.56349492e-01 5.70214629e-01 -2.31123343e-01 -3.84099633e-01 -7.91735947e-01 -1.14903487e-01 4.57851738e-01 4.80577379e-01 -1.03199565e+00 1.47011941e-02 -5.24860695e-02 3.99595588e-01 -1.17591512e+00 1.03247523e+00 -1.22410679e+00 3.32735002e-01 8.08987990e-02 1.72469065e-01 -2.22040847e-01 3.24238449e-01 7.67275214e-01 -3.42485487e-01 -7.84743875e-02 6.17306054e-01 -4.25234765e-01 -8.34541738e-01 4.84473705e-01 1.70981601e-01 -2.76888877e-01 1.24368155e+00 -5.24784863e-01 5.81231751e-02 -1.42538935e-01 -7.92289615e-01 3.58490616e-01 8.24741244e-01 5.54370284e-01 8.65860641e-01 -1.00981462e+00 -5.61175704e-01 3.32833022e-01 1.88684106e-01 9.92129803e-01 6.56359375e-01 6.07441604e-01 -5.55803835e-01 6.88853681e-01 -1.78522065e-01 -1.46389890e+00 -1.32742095e+00 4.94202167e-01 1.90679297e-01 1.83862939e-01 -9.71770406e-01 5.14860153e-01 6.99598312e-01 -6.55714571e-01 -4.48917272e-03 -6.21515274e-01 -2.04735994e-01 -3.49415950e-02 7.69376382e-02 3.56428772e-01 3.22849691e-01 -9.45426404e-01 -3.87187064e-01 1.23689711e+00 6.37304336e-02 1.90931529e-01 1.52585196e+00 -8.25736374e-02 -4.08088639e-02 6.67248905e-01 1.20789504e+00 -2.75050014e-01 -1.40293372e+00 -3.90426934e-01 -2.04056591e-01 -6.74112022e-01 -2.16511521e-03 -4.46169853e-01 -1.21478593e+00 1.35984552e+00 5.38625598e-01 7.50599653e-02 9.40196514e-01 3.97469789e-01 6.89170539e-01 2.83102065e-01 7.09050000e-01 -5.06312668e-01 -2.54025996e-01 5.88160634e-01 8.07522416e-01 -1.31121409e+00 -1.45780314e-02 -7.67205894e-01 -3.87910306e-01 1.13608921e+00 3.21982414e-01 -2.81893790e-01 3.64267081e-01 -2.68311560e-01 -1.24263816e-01 -4.13258225e-01 -3.60783517e-01 -1.69933349e-01 3.23155642e-01 4.47898537e-01 -1.58431128e-01 1.49638653e-01 3.74735415e-01 2.91325718e-01 -5.35647906e-02 -3.89598370e-01 3.95690590e-01 5.88863671e-01 -4.47011322e-01 -9.21339512e-01 -6.34300113e-01 4.19286102e-01 -6.43343180e-02 3.66256744e-01 -2.32288152e-01 8.47950816e-01 3.10550295e-02 7.94196069e-01 3.77759039e-01 -1.56973958e-01 7.63625026e-01 1.30689830e-01 4.50631380e-01 -9.25488412e-01 -2.28979260e-01 2.84198195e-01 -2.98215181e-01 -6.44596934e-01 -3.92974228e-01 -8.60310555e-01 -1.50497246e+00 5.08795977e-02 -4.64326739e-02 -2.14555874e-01 1.05333912e+00 1.03641295e+00 4.19647217e-01 2.84443319e-01 6.25682950e-01 -1.53901613e+00 -1.22785166e-01 -5.51568747e-01 -6.49609745e-01 5.11053383e-01 2.29132071e-01 -7.33156085e-01 -1.64667174e-01 -2.25347593e-01]
[8.070469856262207, -3.098529100418091]
836fb890-8052-4c82-9af9-4bc3ad319e12
aligning-step-by-step-instructional-diagrams
2303.138
null
https://arxiv.org/abs/2303.13800v2
https://arxiv.org/pdf/2303.13800v2.pdf
Aligning Step-by-Step Instructional Diagrams to Video Demonstrations
Multimodal alignment facilitates the retrieval of instances from one modality when queried using another. In this paper, we consider a novel setting where such an alignment is between (i) instruction steps that are depicted as assembly diagrams (commonly seen in Ikea assembly manuals) and (ii) video segments from in-the-wild videos; these videos comprising an enactment of the assembly actions in the real world. To learn this alignment, we introduce a novel supervised contrastive learning method that learns to align videos with the subtle details in the assembly diagrams, guided by a set of novel losses. To study this problem and demonstrate the effectiveness of our method, we introduce a novel dataset: IAW for Ikea assembly in the wild consisting of 183 hours of videos from diverse furniture assembly collections and nearly 8,300 illustrations from their associated instruction manuals and annotated for their ground truth alignments. We define two tasks on this dataset: First, nearest neighbor retrieval between video segments and illustrations, and, second, alignment of instruction steps and the segments for each video. Extensive experiments on IAW demonstrate superior performances of our approach against alternatives.
['Stephen Gould', 'Cristian Rodriguez', 'Yizhak Ben-Shabat', 'Yanbin Liu', 'Anoop Cherian', 'Jiahao Zhang']
2023-03-24
null
http://openaccess.thecvf.com//content/CVPR2023/html/Zhang_Aligning_Step-by-Step_Instructional_Diagrams_to_Video_Demonstrations_CVPR_2023_paper.html
http://openaccess.thecvf.com//content/CVPR2023/papers/Zhang_Aligning_Step-by-Step_Instructional_Diagrams_to_Video_Demonstrations_CVPR_2023_paper.pdf
cvpr-2023-1
['video-retrieval', 'video-alignment']
['computer-vision', 'computer-vision']
[ 5.09320498e-01 -1.02800645e-01 -1.32827312e-01 -4.36993986e-01 -1.45338261e+00 -9.74370062e-01 5.13436556e-01 -1.04774654e-01 -2.88369775e-01 1.90613195e-01 3.19941670e-01 1.15848035e-01 -4.03185874e-01 -1.14782624e-01 -1.40854013e+00 -5.55043697e-01 -7.34860450e-02 6.92607284e-01 -1.41632393e-01 -1.76695034e-01 5.63673198e-01 3.63015562e-01 -1.75869799e+00 6.99266672e-01 5.09631574e-01 9.51733232e-01 3.00348759e-01 9.78281379e-01 1.41467601e-01 9.78172660e-01 -7.93307483e-01 -6.48887753e-01 4.27082449e-01 -3.24997336e-01 -8.93985033e-01 4.46375519e-01 1.57099402e+00 -6.59355819e-01 -6.42659128e-01 8.21429193e-01 3.04145873e-01 5.51940560e-01 8.50046098e-01 -1.63165188e+00 -5.57472765e-01 5.95678389e-01 -5.77347577e-01 4.87774462e-02 1.30477536e+00 2.86181629e-01 1.18345451e+00 -7.67196894e-01 1.30909145e+00 1.17653954e+00 5.75269997e-01 2.73123115e-01 -1.26046395e+00 -5.34741461e-01 1.75257146e-01 5.99426746e-01 -1.50788736e+00 -5.88878274e-01 9.20157433e-01 -5.52396655e-01 6.95294976e-01 3.63933951e-01 7.59813607e-01 1.43966770e+00 -1.16874672e-01 1.22944510e+00 4.71139580e-01 -5.77960551e-01 -8.40048864e-02 -1.51047021e-01 -1.15072601e-01 9.10909474e-01 -2.68317819e-01 -3.01205546e-01 -1.02400386e+00 9.63566452e-02 5.13314843e-01 1.86250955e-01 -6.70085192e-01 -7.86696196e-01 -1.50786734e+00 5.12585759e-01 2.46278569e-01 2.01830745e-01 -1.87255532e-01 8.09453651e-02 4.30842102e-01 5.62741995e-01 -8.59605372e-02 7.48783469e-01 -1.43099025e-01 -3.61860842e-01 -9.71360981e-01 5.33235669e-01 9.42083776e-01 1.54939604e+00 6.51248157e-01 -5.09057105e-01 -4.84333374e-02 5.69022655e-01 4.69556786e-02 5.96497715e-01 4.22590584e-01 -1.38390160e+00 9.68230188e-01 3.28170240e-01 1.85650274e-01 -1.35979927e+00 3.78004387e-02 3.27521622e-01 -2.58932471e-01 -3.35739791e-01 6.39888823e-01 5.06357074e-01 -6.60432756e-01 1.54567754e+00 2.07155421e-01 1.57303452e-01 -2.76720524e-03 9.52545047e-01 8.77259195e-01 7.76936114e-01 -3.32029521e-01 1.90718248e-02 1.14889085e+00 -1.41135514e+00 -7.95395851e-01 -1.13182319e-02 6.19611442e-01 -1.19409764e+00 1.40221059e+00 5.00413656e-01 -1.37167156e+00 -6.89954698e-01 -1.10673952e+00 -6.13367200e-01 -4.02950943e-01 3.06009561e-01 4.83131669e-02 -1.12615317e-01 -9.35506165e-01 6.17005706e-01 -5.20424664e-01 -4.93022084e-01 2.04949290e-01 1.80350050e-01 -8.21644604e-01 -5.64629078e-01 -5.92086256e-01 6.14731252e-01 2.09465146e-01 2.40592554e-01 -1.09072506e+00 -7.00467646e-01 -1.27043581e+00 3.92962620e-02 7.65755653e-01 -4.60042387e-01 1.31938648e+00 -1.19194996e+00 -1.24629211e+00 1.20707810e+00 -2.93103140e-02 2.13256273e-02 6.41255438e-01 -7.21963286e-01 -2.06312120e-01 5.90539575e-01 1.01666048e-01 8.58925879e-01 1.20352030e+00 -1.43097007e+00 -5.74258924e-01 -3.46571684e-01 5.68744719e-01 3.48454773e-01 -1.11823432e-01 -5.06890655e-01 -9.99363005e-01 -6.35245919e-01 7.29599893e-02 -1.23082483e+00 3.63902450e-01 1.54888675e-01 -6.14357412e-01 -2.15798065e-01 7.17894614e-01 -9.93134677e-01 1.05405283e+00 -2.36145544e+00 8.74071598e-01 2.63971448e-01 9.33335572e-02 -1.76561221e-01 -6.79214597e-01 7.19491363e-01 -1.88740879e-01 -3.42248619e-01 6.27563745e-02 -5.17603874e-01 3.35100621e-01 2.81521857e-01 -4.68772352e-01 6.54926300e-01 -7.94839337e-02 8.41286659e-01 -1.11264753e+00 -7.01903224e-01 1.41103685e-01 2.92893440e-01 -5.62344491e-01 8.70452762e-01 -7.71754310e-02 2.57086813e-01 -2.11756751e-01 9.69668865e-01 2.73845464e-01 3.37142833e-02 1.49588332e-01 -1.00632405e+00 2.32340395e-01 -1.79768249e-01 -8.98163497e-01 2.52027178e+00 -3.52435321e-01 9.59617853e-01 -1.02478147e-01 -1.04995036e+00 4.59080130e-01 1.26801074e-01 4.12539959e-01 -5.18355489e-01 -7.29580373e-02 3.43603380e-02 -4.17636991e-01 -1.20092320e+00 7.62897670e-01 5.07294357e-01 -2.56053448e-01 3.66369009e-01 3.72389555e-01 -6.19992733e-01 5.02713740e-01 5.55341780e-01 1.06533742e+00 5.81211269e-01 2.31976643e-01 9.91106480e-02 3.59752655e-01 2.28183344e-01 -9.06558111e-02 8.86367142e-01 -1.33901983e-01 7.37013996e-01 3.75217825e-01 -6.76795065e-01 -1.13127649e+00 -1.14762676e+00 4.38036531e-01 1.27273440e+00 3.23646009e-01 -7.43935168e-01 -7.34847307e-01 -8.09061468e-01 -1.20713733e-01 3.63728374e-01 -5.67005098e-01 -7.80405700e-02 -9.69420791e-01 3.91587317e-01 3.02057207e-01 4.09433663e-01 2.66665965e-01 -8.72929156e-01 -4.78076279e-01 -5.10277689e-01 -6.96341753e-01 -1.32142675e+00 -1.16519904e+00 -2.46059656e-01 -5.08824348e-01 -1.43952990e+00 -5.80411971e-01 -1.05426264e+00 9.60511744e-01 3.05300832e-01 1.58399487e+00 2.54727185e-01 -4.36330825e-01 1.15463376e+00 -3.60206634e-01 1.03176817e-01 -4.61117387e-01 1.72563657e-01 -8.94435942e-02 -9.34648663e-02 1.72855139e-01 -3.32774192e-01 -5.41972101e-01 3.97135109e-01 -1.24344921e+00 1.20929249e-01 7.33453333e-01 8.63474727e-01 7.37360358e-01 -4.38728631e-01 -1.87139511e-01 -7.25881457e-01 8.47928897e-02 -3.04955542e-01 -4.48418856e-01 7.02198386e-01 6.13285042e-02 -1.27985120e-01 5.42574883e-01 -6.36841834e-01 -7.11708486e-01 4.28930581e-01 3.29000801e-01 -1.04224467e+00 -2.20521539e-01 3.90661538e-01 -3.92933279e-01 3.20820021e-03 2.53192663e-01 5.08005172e-03 -7.10665286e-02 -2.14221254e-01 6.72814012e-01 5.02853274e-01 1.26733208e+00 -9.04985666e-01 9.47398424e-01 2.19944403e-01 -1.78531691e-01 -7.58208156e-01 -8.62115502e-01 -7.94132471e-01 -1.04445612e+00 -5.84444821e-01 8.20009470e-01 -7.91776061e-01 -8.12134743e-01 6.66622519e-02 -1.12232614e+00 -3.53657365e-01 -3.35054636e-01 3.50133121e-01 -1.23595572e+00 3.52433532e-01 -5.99831820e-01 -2.89535522e-01 1.33731663e-01 -1.40905869e+00 1.65543818e+00 1.05366018e-02 -5.18520594e-01 -6.85699344e-01 6.45977110e-02 8.26433420e-01 -2.43998617e-01 4.25177723e-01 1.01621103e+00 -7.19400227e-01 -9.74353313e-01 -4.73785937e-01 1.03434376e-01 1.29695117e-01 1.13828018e-01 3.35934967e-01 -7.16021478e-01 -4.78999555e-01 -1.87849417e-01 -6.89592957e-01 3.44596058e-01 1.65367741e-02 1.24324572e+00 -5.65751553e-01 -2.88891762e-01 6.34592295e-01 1.29899228e+00 1.88730940e-01 5.88671386e-01 2.22189441e-01 8.62232625e-01 9.19914067e-01 1.13517249e+00 4.20686305e-02 1.55534714e-01 9.46393311e-01 5.20184994e-01 2.38865450e-01 -5.98128103e-02 -5.48936009e-01 5.56513488e-01 1.31261539e+00 -4.74124886e-02 -4.20882314e-01 -6.97564125e-01 6.22616172e-01 -1.93575823e+00 -1.07848060e+00 4.23530042e-01 2.11760044e+00 7.79975116e-01 -2.46024162e-01 -2.67665014e-02 -2.09199101e-01 5.58427453e-01 2.15766579e-01 -5.32935083e-01 -3.13299857e-02 7.60338008e-02 -2.43779838e-01 -6.38069259e-03 3.12487274e-01 -1.31961405e+00 4.24222410e-01 6.06797075e+00 8.97717416e-01 -5.87573051e-01 -2.67782629e-01 5.27130306e-01 -4.96788889e-01 -3.74796800e-02 -3.35173547e-01 -4.40368384e-01 3.54171962e-01 6.25091732e-01 1.64210573e-01 6.71788812e-01 7.58015931e-01 -1.99874043e-01 -5.23169339e-02 -2.03681016e+00 1.31916070e+00 8.43086898e-01 -1.31223011e+00 3.94507468e-01 -3.80704671e-01 9.03678119e-01 -5.52303731e-01 2.12513506e-01 3.67843240e-01 -3.58924642e-02 -9.62417006e-01 9.30376887e-01 5.80419600e-01 8.04903686e-01 -4.91438895e-01 5.47778904e-01 4.57038395e-02 -9.99066710e-01 4.46280977e-03 -4.73104790e-02 3.41415375e-01 1.07704028e-01 -2.56690800e-01 -5.89922190e-01 6.29326761e-01 8.98089051e-01 1.14129603e+00 -6.79198086e-01 8.62351656e-01 -1.37243513e-02 4.19257209e-03 -1.04393743e-01 2.39059642e-01 3.54750127e-01 -3.42607319e-01 6.56216681e-01 1.22736073e+00 3.65445793e-01 -1.32168487e-01 2.51169503e-01 5.30383945e-01 -3.86083245e-01 -9.38487872e-02 -1.21534693e+00 -6.19917326e-02 5.49834311e-01 1.14644182e+00 -5.07813215e-01 -3.57915014e-01 -3.31271917e-01 1.27028513e+00 3.61492217e-01 6.32657290e-01 -1.00362921e+00 -3.18909496e-01 5.18075705e-01 -7.72730410e-02 4.23446059e-01 -8.96214843e-02 8.02142620e-01 -1.20359445e+00 2.71659374e-01 -1.39537466e+00 5.19697011e-01 -1.29748440e+00 -1.32564294e+00 5.05214930e-01 4.19690996e-01 -1.61740339e+00 -5.47569036e-01 -6.28919661e-01 -2.24422514e-01 1.09601371e-01 -9.67203379e-01 -1.26357460e+00 -6.27526283e-01 6.96906626e-01 1.23448503e+00 -7.56393895e-02 5.35043597e-01 4.31887001e-01 -4.83150005e-01 5.17025232e-01 2.58730978e-01 1.30096376e-01 1.23335183e+00 -1.31906712e+00 -6.09796196e-02 6.23697937e-01 7.16033816e-01 6.72293603e-01 7.86405802e-01 -3.22537720e-01 -1.88894010e+00 -7.78337955e-01 3.66813779e-01 -8.34763587e-01 6.34804010e-01 -4.26030546e-01 -7.04104483e-01 1.27851319e+00 5.04134536e-01 -2.63475358e-01 6.43285930e-01 -3.39703083e-01 -4.42731112e-01 -7.52357021e-02 -7.53726304e-01 7.88988233e-01 1.21342945e+00 -8.77490819e-01 -8.76390338e-01 7.63755798e-01 6.48726881e-01 -9.53859389e-01 -9.53476667e-01 1.17782317e-01 1.02327001e+00 -6.69722080e-01 1.33663464e+00 -9.35139716e-01 1.07628357e+00 -2.64322490e-01 -5.98374188e-01 -1.25553942e+00 2.60032028e-01 -8.24668109e-01 -1.98866025e-01 1.17194724e+00 1.37579124e-02 2.78725535e-01 5.13949931e-01 6.35534227e-01 -2.33033746e-01 -7.23571539e-01 -7.93719351e-01 -7.67766237e-01 -4.90671724e-01 -7.72585124e-02 4.00357187e-01 9.70048845e-01 -2.63562314e-02 2.72734702e-01 -4.64891434e-01 5.99255785e-03 4.01353925e-01 3.77264649e-01 1.18350768e+00 -5.43440104e-01 -3.20084184e-01 -3.52895141e-01 -7.04877615e-01 -1.36791265e+00 4.72067505e-01 -5.60601473e-01 4.44068164e-01 -8.67899954e-01 4.03225869e-01 1.26846597e-01 1.41304344e-01 1.47466287e-01 8.57175444e-04 3.86839956e-01 4.95023161e-01 5.06946385e-01 -1.19332516e+00 3.14745933e-01 1.10310352e+00 -7.26762116e-01 1.54572099e-01 -4.32720095e-01 2.75458917e-02 7.57240534e-01 -6.58369903e-03 -2.50593811e-01 -5.21717429e-01 -7.22505927e-01 1.46560669e-01 2.24396706e-01 3.07010263e-01 -8.64469945e-01 3.42318147e-01 7.20620826e-02 2.66833782e-01 -8.07920098e-01 5.30230224e-01 -1.14096224e+00 3.56346250e-01 1.20341733e-01 -7.75726080e-01 6.16565585e-01 6.43104091e-02 8.09408069e-01 -4.87526715e-01 -5.26362777e-01 8.21518973e-02 5.37566133e-02 -9.21455622e-01 -2.10205205e-02 -1.93596169e-01 8.74535516e-02 1.01591301e+00 -2.65920341e-01 -5.12835205e-01 -6.60342038e-01 -7.38702714e-01 2.54785627e-01 7.24873006e-01 5.51766336e-01 7.72935390e-01 -1.43536377e+00 -3.64720702e-01 1.06078900e-01 4.56793368e-01 2.10516527e-01 3.68606120e-01 9.63537812e-01 -8.84110987e-01 3.82712573e-01 -3.11937034e-01 -8.90892446e-01 -1.51944292e+00 7.26749182e-01 3.43084335e-03 -1.82281092e-01 -5.91850102e-01 7.70792127e-01 4.95642215e-01 -3.48154038e-01 6.88630104e-01 -5.01923978e-01 2.62873452e-02 7.83699974e-02 4.93340194e-01 3.27336758e-01 -5.85252196e-02 -7.29639471e-01 -8.89374837e-02 7.96868563e-01 -1.90208599e-01 1.18208736e-01 1.08161092e+00 -3.20830524e-01 1.90245826e-02 8.22257757e-01 1.65727675e+00 8.37550387e-02 -1.41623914e+00 -7.99817964e-02 -5.12016863e-02 -8.53295445e-01 -6.45069838e-01 -6.04797781e-01 -9.43125665e-01 5.82521200e-01 4.52491492e-01 -8.97863805e-02 1.11775923e+00 3.08296859e-01 8.64727259e-01 1.02843881e+00 2.35273391e-01 -8.46727073e-01 6.51654363e-01 -1.53122609e-03 1.21271849e+00 -1.32495785e+00 -6.60044840e-03 -2.79916734e-01 -4.54242945e-01 1.24627137e+00 6.57435715e-01 -1.51329353e-01 1.95968766e-02 -8.44914168e-02 -1.55439402e-03 -4.06737089e-01 -6.43070936e-01 2.28357151e-01 5.77613235e-01 5.58057487e-01 2.15878502e-01 -2.72359073e-01 4.30605143e-01 2.71640867e-01 -9.88355428e-02 -3.45262289e-01 5.04358113e-01 1.08464348e+00 1.93462357e-01 -6.21604264e-01 -4.36649382e-01 3.65045220e-01 4.91947774e-03 8.16552043e-02 -6.50203764e-01 1.28921926e+00 -1.70822382e-01 5.96374452e-01 3.47412437e-01 -2.31203392e-01 7.55654275e-01 1.23858480e-02 8.66464734e-01 -3.05344820e-01 -3.69660497e-01 -1.35971680e-01 -2.30019242e-02 -1.10998034e+00 -7.79788554e-01 -7.49051034e-01 -5.65371990e-01 -2.31668532e-01 -8.47341120e-02 1.47683114e-01 4.48437750e-01 9.71320331e-01 9.22660604e-02 4.31868583e-01 3.85024101e-01 -1.43932486e+00 -4.37277198e-01 -7.30041742e-01 -4.70269978e-01 1.12692356e+00 5.03548622e-01 -7.40784764e-01 -3.89332741e-01 7.55288064e-01]
[10.224395751953125, 0.8496160507202148]
beef27c7-06e3-45ea-aafe-1accaa4658e8
millie-modular-iterative-multilingual-open
null
null
https://openreview.net/forum?id=KNqKOUnl_3F
https://openreview.net/pdf?id=KNqKOUnl_3F
MILLIE: Modular & Iterative Multilingual Open Information Extraction
Open Information Extraction (OpenIE) is the task of extracting $(subject, predicate, object)$ triples from natural language sentences. Current OpenIE systems extract all triple slots independently. In contrast, we investigate the hypothesis that it may be beneficial to extract triple slots iteratively: first extract easy slots, followed by the difficult ones by conditioning on the easy slots, and therefore achieve a better overall extraction. Based on this hypothesis, we propose a neural OpenIE system, MILLIE, that operates in an iterative fashion. Due to the iterative nature, the system is also modular: it is possible to seamlessly integrate rule based extraction systems with a neural end-to-end system, thereby allowing rule based systems to supply extraction slots which MILLIE can leverage for extracting the remaining slots. We confirm our hypothesis empirically: MILLIE outperforms SOTA systems on multiple languages ranging from Chinese to Arabic. Additionally, we are the first to provide an OpenIE test dataset for Arabic.
['Anonymous']
2021-11-16
null
null
null
acl-arr-november-2021-11
['open-information-extraction']
['natural-language-processing']
[ 8.01536739e-02 8.50348592e-01 -3.52497280e-01 -2.33441040e-01 -7.99727142e-01 -7.03092456e-01 4.30408329e-01 9.48085859e-02 -4.35005456e-01 9.29582179e-01 1.08478740e-01 -7.02202559e-01 -7.82919526e-02 -1.07891595e+00 -7.31033742e-01 1.17589578e-01 -1.87393442e-01 6.79876268e-01 2.04660594e-01 -4.46843714e-01 -2.08035167e-02 9.48633850e-02 -1.58234835e+00 5.44455826e-01 1.02418709e+00 8.55176270e-01 -6.94907904e-02 4.14832532e-01 -3.59022468e-01 8.34108889e-01 -4.52661924e-02 -7.81198502e-01 4.02002960e-01 -1.57554418e-01 -1.23863316e+00 -3.47430378e-01 1.89551115e-01 -1.58654675e-01 2.29721397e-01 6.73704982e-01 2.18705833e-01 -1.14881724e-01 6.03057802e-01 -1.21591735e+00 -5.89907944e-01 1.39408755e+00 -4.21639591e-01 -2.50988692e-01 5.70590198e-01 -5.47450744e-02 1.82146657e+00 -1.13624036e+00 1.06051970e+00 1.03421772e+00 6.03903651e-01 4.96836454e-01 -1.01320529e+00 -6.03479445e-01 2.54369020e-01 2.03322321e-01 -1.40136373e+00 -6.32032812e-01 6.02135599e-01 1.26389060e-02 1.56937253e+00 2.37713978e-01 4.54856783e-01 7.30708480e-01 1.30453750e-01 1.27955163e+00 9.94222939e-01 -9.16697323e-01 -3.60478051e-02 6.92903250e-02 4.75213081e-01 9.48746741e-01 3.82030070e-01 -4.46145976e-04 -6.13415718e-01 1.56792045e-01 -7.17734471e-02 -4.66559857e-01 -4.53849994e-02 -3.29838902e-01 -1.00335884e+00 7.21009254e-01 2.86728889e-01 3.39275509e-01 -4.50233489e-01 -3.08575481e-01 4.13774848e-01 3.37444484e-01 2.68029243e-01 5.82064867e-01 -1.06665301e+00 -6.28148839e-02 -1.03549600e+00 6.09864295e-01 1.34996772e+00 1.07327056e+00 1.05749893e+00 -4.46710914e-01 2.72671618e-02 8.09241951e-01 4.33629185e-01 3.53984177e-01 2.95255661e-01 -6.48639441e-01 7.27786183e-01 9.24383938e-01 1.66899964e-01 -7.11170554e-01 -7.13757217e-01 -1.44983202e-01 -2.57522494e-01 -5.79747334e-02 5.49000740e-01 -5.37897050e-01 -9.21280563e-01 1.67161238e+00 2.45599508e-01 -6.04036927e-01 3.53108495e-01 4.93616790e-01 8.94796133e-01 4.97016311e-01 2.57975072e-01 -1.36053264e-01 1.69665146e+00 -6.42391205e-01 -6.25800133e-01 -5.80142200e-01 8.59767199e-01 -6.12371266e-01 9.55324829e-01 4.39642191e-01 -1.16774857e+00 -8.99251271e-03 -1.17174351e+00 -3.32069516e-01 -9.50407505e-01 3.45956892e-01 8.52891326e-01 4.82694745e-01 -6.67786181e-01 4.41299170e-01 -8.07806551e-01 -3.31175923e-01 2.20607772e-01 5.66335678e-01 -3.34173352e-01 3.11256088e-02 -1.43303251e+00 1.06188571e+00 8.35509419e-01 1.97606087e-01 -3.28141376e-02 -5.98259270e-01 -1.25091588e+00 2.03010172e-01 9.64124560e-01 -7.12720871e-01 1.25299466e+00 -7.61754990e-01 -1.22624099e+00 7.96853900e-01 -4.80150551e-01 -6.02661669e-01 5.09392284e-02 -2.46321067e-01 -3.50988209e-01 1.94488410e-02 1.63939014e-01 9.34993267e-01 2.74095029e-01 -1.20711696e+00 -8.45661640e-01 -3.73690635e-01 2.75085866e-01 -1.39975147e-02 -2.07890168e-01 1.18497096e-01 -2.94428945e-01 -1.08564593e-01 1.58809036e-01 -1.00754917e+00 9.06909034e-02 -6.03311181e-01 -7.38313317e-01 -6.27613842e-01 5.47974110e-01 -7.93195844e-01 1.54719079e+00 -1.86047697e+00 -1.52743682e-02 5.40760934e-01 4.08513695e-01 2.92113930e-01 -4.02250402e-02 4.46140975e-01 -1.35599032e-01 3.50998253e-01 -3.07030827e-01 -9.92121547e-02 5.20915329e-01 1.97594702e-01 -1.88923925e-01 -3.02956849e-01 7.03905940e-01 1.31136584e+00 -6.06102228e-01 -8.03848207e-01 -2.92600065e-01 -7.49054477e-02 -8.39608312e-01 -1.68547064e-01 -7.12209165e-01 -5.28903842e-01 -2.63545126e-01 8.69792819e-01 7.61920154e-01 -1.95719004e-01 5.84892929e-01 -2.58921772e-01 -2.96778798e-01 7.32227206e-01 -1.33323038e+00 1.22172415e+00 -4.64190513e-01 4.56635237e-01 7.51494542e-02 -7.13527799e-01 7.05716968e-01 2.70196259e-01 5.56051970e-01 -7.17505753e-01 2.62961715e-01 5.09254873e-01 1.16242073e-01 -4.15540963e-01 9.23326492e-01 -8.30333605e-02 -4.74106938e-01 7.89687693e-01 4.17148769e-01 8.39951262e-02 9.11868513e-01 4.08678412e-01 9.98860061e-01 5.22174537e-01 5.40096164e-01 -1.42662793e-01 3.35685700e-01 2.00150669e-01 6.90500140e-01 5.45173943e-01 -1.48400351e-01 1.19485766e-01 5.25883555e-01 -3.92827690e-01 -9.22074199e-01 -1.13606501e+00 -1.32158265e-01 1.15131497e+00 -1.96445405e-01 -9.17675316e-01 -7.91824102e-01 -1.01679647e+00 -4.23233025e-02 6.19962871e-01 -5.39655328e-01 2.82315135e-01 -8.03353965e-01 -5.73250532e-01 6.33113921e-01 5.23894072e-01 4.20350999e-01 -1.30984175e+00 -6.82547212e-01 4.05761749e-01 -6.61264598e-01 -1.20883310e+00 -7.37244710e-02 7.30141222e-01 -2.43291378e-01 -1.20923340e+00 6.22995198e-02 -6.57268345e-01 2.58573323e-01 -3.08256686e-01 1.56522119e+00 1.62779167e-01 5.83654344e-02 6.25897944e-02 -5.14206648e-01 -7.10542977e-01 -3.09634507e-01 7.04012334e-01 -1.50912166e-01 -3.71340364e-01 1.10686803e+00 -2.68009514e-01 -1.80857420e-01 -2.47770902e-02 -9.56211030e-01 1.54948592e-01 6.69347703e-01 7.15736926e-01 2.21353307e-01 1.53128086e-02 6.40738368e-01 -1.14307022e+00 7.89725065e-01 -4.69386101e-01 -5.24627030e-01 6.51494026e-01 -7.43122876e-01 3.61992508e-01 5.08646250e-01 7.09929913e-02 -1.06914401e+00 1.69096768e-01 -3.90717655e-01 3.20870012e-01 -1.22395001e-01 9.59823310e-01 -3.23772460e-01 2.73020625e-01 6.57214224e-01 4.64719068e-03 -1.69176653e-01 -1.68224663e-01 6.15071356e-01 7.09216833e-01 3.81032646e-01 -9.52205896e-01 6.03828371e-01 -5.29266009e-03 -1.82202458e-01 -6.43201709e-01 -8.44285965e-01 -1.27478451e-01 -7.74076819e-01 9.09740180e-02 6.31153047e-01 -7.66643047e-01 -8.72450471e-01 7.67976046e-02 -1.04266357e+00 -2.39590615e-01 -1.97961420e-01 -1.20870268e-03 -3.75464499e-01 2.70340383e-01 -7.45793760e-01 -6.85124159e-01 -6.01247430e-01 -8.84071946e-01 9.77243721e-01 2.26484954e-01 -6.43124878e-01 -8.45883250e-01 -1.95189446e-01 1.98361963e-01 3.82465869e-02 -7.96016380e-02 1.12653136e+00 -9.69850183e-01 -6.35976017e-01 1.19286209e-01 -3.92035782e-01 -1.60041656e-02 -1.75985754e-01 1.82131454e-01 -7.21525133e-01 2.01044008e-01 -6.74545705e-01 -6.12684011e-01 8.38391423e-01 9.61612388e-02 4.28343862e-01 -3.92310679e-01 -4.71055478e-01 1.32527560e-01 1.33853400e+00 -6.36868253e-02 6.23110473e-01 6.29386306e-01 3.10380250e-01 6.89801037e-01 6.12632215e-01 2.52793938e-01 1.07814765e+00 3.21208715e-01 6.77395612e-02 9.61334258e-02 9.54628587e-02 -3.47289324e-01 4.07130659e-01 3.62579882e-01 2.51769781e-01 -1.65720537e-01 -1.12244236e+00 6.90989017e-01 -1.76370680e+00 -7.53059030e-01 9.50377211e-02 1.67272532e+00 1.26742899e+00 3.67094159e-01 1.48993015e-01 3.68104815e-01 6.53135031e-02 -6.63014352e-02 -5.55612482e-02 -6.80524766e-01 -7.22377375e-02 6.20967984e-01 3.37032765e-01 4.95935947e-01 -1.04570365e+00 1.36847925e+00 6.33044863e+00 7.52034247e-01 -9.17949736e-01 -2.31424570e-01 2.87789434e-01 2.85919383e-03 -5.08534849e-01 3.52308393e-01 -1.43376982e+00 4.59942780e-02 1.01049745e+00 1.13746608e-02 4.65556741e-01 4.98206705e-01 -2.72969633e-01 -3.76384795e-01 -1.36409724e+00 3.13341022e-01 -1.94703013e-01 -1.28234601e+00 3.53787914e-02 5.48398048e-02 2.58058041e-01 1.06426682e-02 -3.24321151e-01 8.32927108e-01 8.08383286e-01 -8.17942917e-01 7.51697540e-01 3.48307282e-01 3.59131455e-01 -8.00650418e-01 5.71697176e-01 4.58219171e-01 -1.08324611e+00 -2.27017865e-01 4.72713858e-02 -1.77776754e-01 1.36157855e-01 6.17645919e-01 -1.12097275e+00 6.83711529e-01 5.68534315e-01 4.34607983e-01 -6.93902850e-01 5.72138071e-01 -4.34081048e-01 5.40183306e-01 -6.89298451e-01 -1.88843668e-01 2.58481592e-01 -5.56634702e-02 3.51990908e-01 1.38528240e+00 2.47700494e-02 6.26291409e-02 3.26844543e-01 9.87248421e-01 -7.37857223e-02 2.30659857e-01 -7.17852473e-01 -1.92165896e-01 5.07444382e-01 1.17985702e+00 -6.23530924e-01 -3.00925255e-01 -6.20816410e-01 4.74662185e-01 7.76359856e-01 2.33286574e-01 -3.82978320e-01 -7.46131241e-01 1.38177887e-01 -2.17954352e-01 6.56897426e-01 -2.75763482e-01 -5.33845603e-01 -1.24028146e+00 2.40536749e-01 -1.06287062e+00 7.53553033e-01 -7.99196839e-01 -8.66918087e-01 6.03540480e-01 -7.58942822e-03 -7.64298558e-01 -7.90716231e-01 -7.47189105e-01 -6.15372546e-02 6.82881057e-01 -1.76142418e+00 -1.39536583e+00 4.56448287e-01 4.85094637e-01 2.71591544e-01 8.62227008e-02 9.35125709e-01 2.97230303e-01 -6.47007942e-01 7.79629886e-01 -5.61383784e-01 6.18647397e-01 3.93067271e-01 -1.34459352e+00 2.19847560e-01 1.08546579e+00 4.46838111e-01 1.14257896e+00 6.03212893e-01 -7.29258895e-01 -1.34197652e+00 -7.35504687e-01 1.72327220e+00 -5.29404163e-01 7.09328651e-01 -4.47184473e-01 -5.84857643e-01 1.09726119e+00 4.68178540e-01 -3.30346555e-01 7.84941077e-01 8.45675290e-01 -4.69644278e-01 8.28502178e-02 -9.47477758e-01 7.62995422e-01 8.87242854e-01 -4.56360489e-01 -8.98561895e-01 -1.15429386e-01 6.22180283e-01 -4.60672140e-01 -1.01806438e+00 5.32267630e-01 7.05358446e-01 -7.40090072e-01 8.48685503e-01 -5.28710783e-01 8.59752953e-01 -1.71732351e-01 -1.13896139e-01 -1.34396279e+00 -1.56461596e-02 -4.82314527e-01 -3.86308849e-01 1.41289818e+00 1.29061222e+00 -4.78673548e-01 6.96196795e-01 9.36421812e-01 1.51970740e-02 -8.87711108e-01 -8.20921421e-01 -4.63848531e-01 2.99606681e-01 -7.08521426e-01 8.27467918e-01 6.17926240e-01 7.71974564e-01 6.76441550e-01 -2.19263002e-01 3.26540805e-02 2.35129938e-01 3.87880534e-01 6.78395510e-01 -1.18067849e+00 -1.88274205e-01 -3.73236477e-01 2.90558755e-01 -9.53684390e-01 5.52044928e-01 -1.07986867e+00 1.90926731e-01 -1.39499915e+00 1.52651310e-01 -8.04724514e-01 -2.10855305e-01 1.16623676e+00 -2.08740935e-01 1.88425779e-01 3.69899273e-01 -1.46635085e-01 -7.97461808e-01 2.98015058e-01 1.01267731e+00 -8.55024904e-02 -3.41919214e-01 -2.51115948e-01 -1.32304549e+00 6.81323528e-01 4.67978179e-01 -2.18068093e-01 -6.55658841e-02 -3.93981427e-01 7.24675596e-01 -5.76772504e-02 -8.66760835e-02 -7.33572125e-01 2.07408786e-01 -3.22446227e-02 2.04691797e-01 -6.54117763e-01 -6.16932940e-03 -9.12620485e-01 -4.96924132e-01 4.23859246e-02 -1.89775482e-01 -2.35177338e-01 3.90746742e-01 -1.65105969e-01 -1.24758899e-01 -3.59721661e-01 2.68548906e-01 -1.13081947e-01 -7.80638456e-01 1.06747024e-01 -4.69559312e-01 1.54775396e-01 5.66695988e-01 -3.27514894e-02 -2.37855300e-01 7.59334788e-02 -7.55851805e-01 4.17045921e-01 7.98540562e-02 4.10149217e-01 4.81675774e-01 -9.96762156e-01 -6.48365438e-01 3.71396571e-01 2.62278795e-01 1.10962540e-02 -4.40860599e-01 6.45819366e-01 -2.04597652e-01 6.83524728e-01 4.76589054e-02 -3.11532795e-01 -1.09671712e+00 4.91656184e-01 2.48619080e-01 -8.78720403e-01 -3.76385570e-01 5.26414156e-01 -4.95316803e-01 -8.91533256e-01 -9.56551451e-03 -6.32077098e-01 -2.84853041e-01 2.30075702e-01 4.09163386e-01 -1.26293406e-01 3.96863192e-01 -4.46066201e-01 -4.21324432e-01 1.29200473e-01 -4.20733809e-01 -2.38518029e-01 1.35152400e+00 -9.03867558e-02 -1.64566308e-01 2.95620970e-02 8.30753326e-01 3.45831335e-01 -6.50230587e-01 -5.03606021e-01 5.19583344e-01 -1.00207761e-01 -1.92870110e-01 -1.14031923e+00 -7.13187039e-01 4.04980004e-01 -2.12241337e-02 2.69620270e-01 1.10415697e+00 4.50462066e-02 1.01560032e+00 7.87170887e-01 4.10956472e-01 -1.25028563e+00 -5.91786444e-01 1.02394283e+00 4.91596192e-01 -1.21803892e+00 1.18369516e-02 -5.51754415e-01 -4.51929420e-01 1.04526901e+00 6.07600808e-01 1.65033117e-01 8.07614028e-01 5.96552849e-01 5.44082336e-02 -3.79442722e-01 -1.01156080e+00 -6.40614331e-01 1.75532088e-01 3.36322576e-01 8.43108058e-01 7.00065121e-02 -7.55210996e-01 9.71686423e-01 -6.50508344e-01 2.90666431e-01 4.21605855e-01 1.13327336e+00 -4.10533577e-01 -1.63893449e+00 -2.39795074e-01 6.07589304e-01 -5.12368500e-01 -5.20016551e-01 -5.83142221e-01 1.04353106e+00 2.85488456e-01 1.14978671e+00 -1.86712250e-01 -2.57887453e-01 4.02517587e-01 6.11172497e-01 5.87458074e-01 -5.76030493e-01 -6.04581296e-01 -1.45249709e-01 7.01972544e-01 -4.39870983e-01 -3.08568031e-01 -8.12717319e-01 -1.57553518e+00 -2.48307273e-01 -4.12326038e-01 3.47478062e-01 4.34022248e-01 1.46815121e+00 4.52706039e-01 3.19496304e-01 2.16533095e-01 -4.22559142e-01 -2.68904060e-01 -9.13942456e-01 -4.73245710e-01 3.14599693e-01 8.93020257e-02 -6.65982664e-01 1.01726413e-01 -2.39063017e-02]
[9.577545166015625, 8.619068145751953]
c14bd148-95bc-4c28-afe5-207f0f2a1fdc
multi-source-transformer-architectures-for
2210.10212
null
https://arxiv.org/abs/2210.10212v1
https://arxiv.org/pdf/2210.10212v1.pdf
Multi-Source Transformer Architectures for Audiovisual Scene Classification
In this technical report, the systems we submitted for subtask 1B of the DCASE 2021 challenge, regarding audiovisual scene classification, are described in detail. They are essentially multi-source transformers employing a combination of auditory and visual features to make predictions. These models are evaluated utilizing the macro-averaged multi-class cross-entropy and accuracy metrics. In terms of the macro-averaged multi-class cross-entropy, our best model achieved a score of 0.620 on the validation data. This is slightly better than the performance of the baseline system (0.658). With regard to the accuracy measure, our best model achieved a score of 77.1\% on the validation data, which is about the same as the performance obtained by the baseline system (77.0\%).
['Hugo Van hamme', 'Wim Boes']
2022-10-18
null
null
null
null
['scene-classification']
['computer-vision']
[ 5.21642109e-03 -1.36971667e-01 -3.05730700e-02 -1.63636193e-01 -1.36639738e+00 -5.41126549e-01 5.64703584e-01 4.23907816e-01 -3.37635726e-01 3.97603869e-01 3.11203182e-01 -6.51952475e-02 2.57362753e-01 -2.62510926e-01 -4.93603915e-01 -6.02447450e-01 -2.12008357e-01 -2.35243484e-01 2.64991581e-01 -2.35478252e-01 2.29617044e-01 1.78279117e-01 -2.04101110e+00 1.05502772e+00 3.81332994e-01 1.76582611e+00 -1.63553059e-01 1.14579475e+00 2.77175844e-01 1.00305104e+00 -8.41164827e-01 -2.98810273e-01 -1.30067140e-01 -3.60504597e-01 -7.70061970e-01 -3.75095487e-01 6.73573494e-01 4.36978042e-02 -3.71180385e-01 8.46674919e-01 7.50179052e-01 7.07294047e-02 8.81691694e-01 -1.39193308e+00 -2.00797111e-01 3.18355501e-01 -4.06546861e-01 4.89180684e-01 8.75465870e-01 -2.98923478e-02 1.31353688e+00 -1.20687950e+00 2.78258830e-01 1.16367626e+00 7.48585522e-01 3.21292639e-01 -1.23803067e+00 -7.32479930e-01 -9.05288011e-03 5.39806128e-01 -1.70859480e+00 -8.70399773e-01 5.97599328e-01 -6.70968354e-01 1.09348011e+00 5.91674924e-01 5.76925576e-01 1.13001072e+00 2.31026933e-01 8.58120561e-01 1.11885655e+00 -4.68947798e-01 2.33973473e-01 4.39942867e-01 -3.15329768e-02 3.17359924e-01 -2.55513400e-01 7.52974972e-02 -7.97214448e-01 -1.08394258e-01 2.32819188e-03 -8.75828564e-01 -3.15309644e-01 1.75685026e-02 -9.71257210e-01 7.17224956e-01 3.37618917e-01 3.51687700e-01 -2.21484810e-01 -1.14961110e-01 6.23498619e-01 1.57582596e-01 4.86386448e-01 3.90476167e-01 -2.92009830e-01 -4.16730672e-01 -1.13823414e+00 1.20203204e-01 7.20737755e-01 7.42323697e-01 -1.59970485e-02 3.91927600e-01 -5.00824094e-01 1.08487189e+00 2.48278171e-01 4.93738979e-01 4.31235343e-01 -9.83841598e-01 5.34456074e-01 1.54766813e-01 -8.44730362e-02 -9.64189708e-01 -3.08071882e-01 -6.50223136e-01 -6.83368504e-01 2.45599017e-01 2.75889158e-01 -1.79851148e-02 -8.24350178e-01 1.58919537e+00 -2.23057941e-01 -2.58603715e-04 2.34796271e-01 7.04044938e-01 1.29704845e+00 9.94466603e-01 4.33180153e-01 -2.27533177e-01 1.24559975e+00 -7.66751707e-01 -6.97827399e-01 -6.72482997e-02 3.35399270e-01 -9.59458411e-01 1.12705493e+00 6.99860334e-01 -1.09781897e+00 -7.77265549e-01 -1.21307826e+00 4.12399471e-01 -3.20975065e-01 2.47545928e-01 6.23376109e-02 7.22639084e-01 -1.23195028e+00 1.44511551e-01 -1.68048903e-01 -3.17481637e-01 1.98624983e-01 -1.28049970e-01 -3.17978591e-01 2.94675946e-01 -1.10303211e+00 8.11734319e-01 3.92297000e-01 -3.22494417e-01 -1.17988551e+00 -7.83624470e-01 -7.80351043e-01 1.39581516e-01 4.03803214e-02 6.54673725e-02 1.54356110e+00 -5.62339067e-01 -1.33397770e+00 7.63236046e-01 -1.41346708e-01 -5.37697375e-01 4.97839689e-01 -1.61399111e-01 -8.51711154e-01 3.39011312e-01 -1.14336126e-01 7.37097979e-01 5.93959749e-01 -1.46472335e+00 -9.68496680e-01 1.67361483e-01 -4.01324965e-02 3.47760409e-01 -2.09314466e-01 2.48248681e-01 -2.75554240e-01 -6.48786008e-01 -2.28907987e-01 -7.55329132e-01 1.13142706e-01 -3.90663743e-01 -5.85972071e-01 -1.04020983e-01 6.57393217e-01 -8.98053944e-01 1.55267560e+00 -2.48660898e+00 -2.93423474e-01 2.17322841e-01 1.42040163e-01 2.63393790e-01 -2.20379710e-01 1.73548982e-01 -5.57461500e-01 1.97681814e-01 4.01069829e-03 -3.68005872e-01 -4.03719060e-02 -7.03033626e-01 -4.15156275e-01 1.79057568e-01 1.89767450e-01 4.91675764e-01 -6.00750566e-01 -4.75791365e-01 3.50753158e-01 4.30167139e-01 -5.97444534e-01 3.16553980e-01 2.03534141e-01 9.24555287e-02 8.82598087e-02 6.44809961e-01 3.93893421e-01 1.60530042e-02 -1.17228083e-01 -3.72367144e-01 -5.17246500e-03 2.06374243e-01 -9.44332838e-01 1.37101769e+00 -4.66597855e-01 1.24301326e+00 -8.81919339e-02 -8.40398669e-01 8.73903394e-01 7.03516006e-01 6.84248269e-01 -6.55521989e-01 -9.60730612e-02 9.32799727e-02 -3.25655825e-02 -1.77895665e-01 5.10459900e-01 6.58453256e-02 -3.58359665e-01 -2.48481631e-02 2.28770316e-01 -2.61852443e-01 5.59592135e-02 2.79146343e-01 9.80146766e-01 -2.95897424e-01 3.33808988e-01 -1.86841503e-01 6.33259416e-01 -2.52825707e-01 1.42758682e-01 5.91063440e-01 -4.86307859e-01 9.05831754e-01 5.08990407e-01 -2.70248055e-02 -9.45601702e-01 -1.28579366e+00 -3.73540789e-01 8.86033833e-01 -4.55585606e-02 -8.96731555e-01 -7.79442847e-01 -3.81051540e-01 -1.39154971e-01 7.74854839e-01 -5.18610656e-01 -3.37171435e-01 1.36685655e-01 -4.71276611e-01 9.79061604e-01 6.24562025e-01 3.26598942e-01 -9.42830265e-01 -4.78471279e-01 3.12940404e-02 -7.54687905e-01 -1.38323367e+00 -1.76533476e-01 2.56194651e-01 -2.77693510e-01 -8.89436185e-01 -5.84745228e-01 -4.69769210e-01 -2.27098167e-01 -4.32492271e-02 1.10957944e+00 -2.96972692e-01 -1.76683009e-01 3.52879137e-01 -2.46424392e-01 -6.07425749e-01 -3.48896444e-01 -1.20183781e-01 8.16653818e-02 8.79547894e-02 2.08502948e-01 -2.44193956e-01 -4.53677684e-01 2.58296490e-01 -4.60065365e-01 -2.93435127e-01 2.08381310e-01 5.26519299e-01 6.72384501e-01 -3.30974683e-02 6.83243096e-01 9.15269367e-03 6.25269055e-01 -5.25122285e-01 -4.00007963e-01 2.90392935e-01 -6.22157633e-01 -4.56076294e-01 5.43012023e-01 -4.02467936e-01 -8.07265937e-01 -2.43449248e-02 -3.27785909e-01 -4.34864193e-01 -4.11048234e-01 2.92713910e-01 -1.88698340e-02 5.90147413e-02 6.17791772e-01 1.10163882e-01 -3.81256670e-01 -5.34242272e-01 8.26007873e-02 1.17424512e+00 6.50567651e-01 -1.59801915e-01 2.76713729e-01 -5.63432723e-02 -2.46341199e-01 -1.01848173e+00 -8.52645040e-01 -4.79611397e-01 -2.47570768e-01 -5.57804108e-01 9.72095251e-01 -1.21318340e+00 -8.44354570e-01 6.09939396e-01 -1.00581157e+00 -1.92197517e-01 -3.15937608e-01 6.07694685e-01 -8.02032292e-01 4.40939032e-02 -3.27280015e-01 -1.16280186e+00 -2.56756753e-01 -9.53951895e-01 1.06954384e+00 -6.69015869e-02 -5.69569588e-01 -6.89267874e-01 1.59103855e-01 3.83241147e-01 4.61349159e-01 2.95418262e-01 7.63956785e-01 -8.59101355e-01 2.12032497e-01 -3.41144562e-01 -2.04830885e-01 5.90196192e-01 -1.84069350e-01 1.77722111e-01 -1.66559422e+00 -3.01361442e-01 -2.57929385e-01 -5.23060739e-01 1.02470672e+00 5.06655037e-01 1.58104110e+00 9.53386948e-02 -8.23684558e-02 3.06186676e-01 1.17538500e+00 5.51584542e-01 5.19989133e-01 2.19393030e-01 1.91917658e-01 3.72501701e-01 5.56654096e-01 7.72361279e-01 5.67205429e-01 1.07436836e+00 4.19341803e-01 9.37443078e-02 -4.74698275e-01 -4.16896313e-01 5.47668338e-01 6.99881971e-01 -8.37577041e-03 -4.11952466e-01 -1.19352829e+00 5.71000874e-01 -1.39503658e+00 -1.20158553e+00 -5.55716120e-02 2.16222668e+00 5.26821434e-01 2.43712246e-01 5.15125394e-01 8.01336110e-01 6.82856202e-01 3.42615813e-01 -7.36390054e-02 -8.24914575e-01 -3.48104626e-01 1.00035369e-01 -5.74594587e-02 4.82725680e-01 -1.38817227e+00 5.56226313e-01 8.12673092e+00 1.20612800e+00 -1.01466608e+00 -6.68303445e-02 7.78253138e-01 -3.91374141e-01 1.65474609e-01 -4.44640279e-01 -6.05520368e-01 6.71132028e-01 1.78584027e+00 -3.01586390e-01 3.31718504e-01 7.09559917e-01 -6.30751923e-02 -3.56729388e-01 -1.05962908e+00 1.29934549e+00 3.06856692e-01 -9.49799776e-01 -1.99547887e-01 -6.48534158e-03 5.19783854e-01 3.42688449e-02 3.54337156e-01 5.07467508e-01 -1.31768972e-01 -1.20238268e+00 1.25110257e+00 4.18846458e-01 1.10658622e+00 -8.85674894e-01 7.37281024e-01 1.19600974e-01 -1.52506936e+00 -2.57835448e-01 3.53923216e-02 6.61478415e-02 1.31595835e-01 4.38417464e-01 -7.11459696e-01 3.10581267e-01 1.13708234e+00 5.03877819e-01 -7.25473523e-01 1.31986940e+00 8.61404017e-02 9.22319174e-01 -3.34773988e-01 9.51006915e-03 -6.34090975e-02 5.20827591e-01 6.62044227e-01 1.68294668e+00 4.14797217e-01 -8.24448541e-02 1.47953059e-03 1.73282459e-01 7.54114613e-02 3.23422849e-01 -5.89866221e-01 1.84534550e-01 5.36814451e-01 9.07172143e-01 -1.82351142e-01 -5.25686622e-01 -2.18227968e-01 5.52404404e-01 4.94702645e-02 3.40595663e-01 -1.02887797e+00 -8.05236816e-01 6.26573384e-01 -1.54306576e-01 3.57524902e-01 3.11654180e-01 -3.41835588e-01 -9.65730429e-01 1.07046530e-01 -9.41406071e-01 5.19020200e-01 -1.04141057e+00 -9.34324801e-01 1.04639935e+00 1.71627328e-01 -1.63573277e+00 -3.71585160e-01 -4.41452771e-01 -4.35817391e-01 7.18279660e-01 -1.30377138e+00 -7.05710232e-01 -3.13070744e-01 6.57004595e-01 4.65289205e-01 -5.59180081e-01 1.09776080e+00 4.29178357e-01 -2.92660326e-01 1.13186109e+00 -6.41189814e-02 4.75149415e-02 6.59761548e-01 -1.01668394e+00 1.55590158e-02 6.05744958e-01 3.52571696e-01 -1.06055476e-01 8.00179243e-01 -6.37831464e-02 -7.90301561e-01 -8.55297148e-01 9.70502555e-01 -5.03030777e-01 5.84449291e-01 -1.49911895e-01 -6.96304858e-01 2.47116417e-01 3.02896649e-01 -1.22871511e-02 9.71836030e-01 1.39215901e-01 -6.31085336e-01 -2.50576645e-01 -1.32438195e+00 2.23244548e-01 6.43516719e-01 -9.40012395e-01 -3.64697874e-01 1.06121086e-01 5.59159338e-01 -3.35975498e-01 -1.16561353e+00 5.92041314e-01 8.10839355e-01 -1.06511879e+00 1.08703959e+00 -6.81961179e-01 5.06662369e-01 -3.81110273e-02 -9.48620737e-01 -1.30073988e+00 -2.33724818e-01 -1.99722975e-01 -2.16316551e-01 1.20070267e+00 6.17622197e-01 -2.18731761e-01 4.60644096e-01 2.13380992e-01 -1.28960580e-01 -7.26303399e-01 -1.27843976e+00 -9.67077613e-01 8.94517004e-02 -1.18131340e+00 3.68457943e-01 6.22690856e-01 2.39901900e-01 4.28038538e-01 -4.41276997e-01 -1.29741624e-01 5.89220226e-01 -1.16593197e-01 5.32735825e-01 -1.12347877e+00 -5.94595186e-02 -5.50250888e-01 -7.88558781e-01 -6.05042338e-01 1.08075015e-01 -7.90637493e-01 5.21096252e-02 -1.33918023e+00 3.42048287e-01 -9.03062150e-02 -8.24190259e-01 4.57362533e-01 1.45417735e-01 7.00763106e-01 6.44726515e-01 7.94516876e-02 -6.77010298e-01 4.86886591e-01 6.48143649e-01 -3.05802822e-01 -7.45047480e-02 1.91857681e-01 -7.44488478e-01 7.28372991e-01 1.04836702e+00 -2.79711157e-01 -1.72093019e-01 -3.12110111e-02 -2.49513000e-01 1.54524237e-01 3.09487939e-01 -1.49060297e+00 -7.85897970e-02 -6.62197471e-02 3.82749349e-01 -7.02745736e-01 1.01503420e+00 -6.51397347e-01 3.50385644e-02 4.28901762e-01 -5.79010665e-01 -3.48847397e-02 5.94461620e-01 4.24116582e-01 -7.43641794e-01 2.63525337e-01 9.41892326e-01 4.47459400e-01 -8.39460790e-01 -2.18058944e-01 -6.68594360e-01 3.64571325e-02 1.10642433e+00 -3.33953798e-01 -3.82340997e-01 -9.05107975e-01 -9.43427503e-01 -1.27146497e-01 3.42493169e-02 6.82457089e-01 7.78984249e-01 -1.66141975e+00 -7.95480192e-01 1.18380845e-01 5.28742433e-01 -8.38777363e-01 1.93253741e-01 7.41550863e-01 -1.08072728e-01 8.54295552e-01 -2.65753984e-01 -6.85270727e-01 -1.55750179e+00 1.79486200e-01 4.37594324e-01 -7.48019814e-02 3.64370877e-03 8.80571187e-01 1.23839192e-02 -1.17843613e-01 4.04624939e-01 -1.98461011e-01 -4.50193971e-01 2.80195445e-01 6.61500692e-01 5.21309793e-01 3.39014888e-01 -1.08636701e+00 -8.08031678e-01 5.40402889e-01 3.62957954e-01 -3.67633998e-01 1.10986435e+00 4.74562719e-02 3.16336900e-01 7.22267866e-01 1.46703434e+00 1.33822514e-02 -8.30242574e-01 -2.17743609e-02 -1.01860300e-01 -4.21951532e-01 1.85606256e-01 -1.25467956e+00 -8.13963950e-01 1.05170977e+00 1.04162776e+00 4.60228324e-01 1.30458748e+00 1.87800094e-01 2.37462789e-01 2.01504767e-01 1.31510958e-01 -1.09904587e+00 2.26039425e-01 6.67886555e-01 1.18391347e+00 -1.24375296e+00 -1.58507764e-01 -2.52673239e-01 -1.02828944e+00 8.13457370e-01 6.26349747e-01 1.07333183e-01 8.18244934e-01 1.65342033e-01 1.12726845e-01 1.37985855e-01 -1.21692741e+00 -3.25428359e-02 9.12992001e-01 6.87187135e-01 6.03008509e-01 2.10469484e-01 9.39033702e-02 8.43852580e-01 -4.62964982e-01 -1.92656323e-01 3.41979623e-01 5.40975571e-01 -5.57560205e-01 -4.39077497e-01 -3.25233281e-01 3.87164831e-01 -6.82975590e-01 -3.11803594e-02 -4.62212473e-01 4.71611261e-01 -2.23212972e-01 1.52401984e+00 1.83122262e-01 -1.10266459e+00 6.31503642e-01 2.53762007e-01 1.32919222e-01 -3.98642749e-01 -4.89845186e-01 7.82944262e-03 4.22453642e-01 -7.92687476e-01 -2.39391923e-01 -6.16289437e-01 -8.83468449e-01 -2.91695744e-01 -1.99261591e-01 2.09085926e-01 8.57392251e-01 5.04249454e-01 2.73760229e-01 5.76409519e-01 8.06650341e-01 -7.50894189e-01 -4.19667214e-01 -1.10852993e+00 -5.29548228e-01 2.74424076e-01 5.41129351e-01 -5.83040237e-01 -6.23455405e-01 3.06236167e-02]
[15.0498046875, 5.091681480407715]
60b1bc5b-347d-44d7-bfdf-1a7bf43749bc
learning-advisor-networks-for-noisy-image-1
2211.04177
null
https://arxiv.org/abs/2211.04177v1
https://arxiv.org/pdf/2211.04177v1.pdf
Learning advisor networks for noisy image classification
In this paper, we introduced the novel concept of advisor network to address the problem of noisy labels in image classification. Deep neural networks (DNN) are prone to performance reduction and overfitting problems on training data with noisy annotations. Weighting loss methods aim to mitigate the influence of noisy labels during the training, completely removing their contribution. This discarding process prevents DNNs from learning wrong associations between images and their correct labels but reduces the amount of data used, especially when most of the samples have noisy labels. Differently, our method weighs the feature extracted directly from the classifier without altering the loss value of each data. The advisor helps to focus only on some part of the information present in mislabeled examples, allowing the classifier to leverage that data as well. We trained it with a meta-learning strategy so that it can adapt throughout the training of the main model. We tested our method on CIFAR10 and CIFAR100 with synthetic noise, and on Clothing1M which contains real-world noise, reporting state-of-the-art results.
['Alberto del Bimbo', 'Tiberio Uricchio', 'Simone Ricci']
2022-11-08
learning-advisor-networks-for-noisy-image
https://link.springer.com/chapter/10.1007/978-3-031-06430-2_37
https://link.springer.com/chapter/10.1007/978-3-031-06430-2_37
iciap-2022-5
['learning-with-noisy-labels', 'learning-with-noisy-labels']
['computer-vision', 'natural-language-processing']
[ 2.92799413e-01 2.41298020e-01 1.29518136e-01 -6.96877122e-01 -5.44812799e-01 -4.62667078e-01 2.57809579e-01 1.79122388e-01 -9.87450063e-01 7.15500534e-01 -6.66805729e-02 -2.65216958e-02 1.94860678e-02 -6.69265270e-01 -8.08908820e-01 -8.35567176e-01 1.51435316e-01 3.29080671e-01 1.80539042e-01 8.27918127e-02 -1.80458739e-01 2.40141407e-01 -1.58900416e+00 6.01223111e-01 5.28068781e-01 1.23634028e+00 1.87592581e-01 3.88127506e-01 7.48788491e-02 1.24853587e+00 -9.86287951e-01 -6.24642670e-01 2.26051867e-01 -1.85521498e-01 -7.54009426e-01 1.76833853e-01 7.34971166e-01 -1.71210393e-01 -2.93949276e-01 1.28661823e+00 5.39297640e-01 4.46867980e-02 4.69250888e-01 -1.13036466e+00 -3.88074517e-01 8.78371000e-01 -1.90886125e-01 1.02821857e-01 -6.02805495e-01 1.47320077e-01 9.86113906e-01 -9.73354578e-01 5.49498081e-01 1.08501852e+00 7.52052605e-01 8.37029159e-01 -1.40847754e+00 -6.80458367e-01 4.31686282e-01 1.79712027e-01 -1.14753282e+00 -5.75382769e-01 5.57196319e-01 -3.37915033e-01 5.55909216e-01 7.72990063e-02 2.74615735e-01 1.37273622e+00 -2.85705030e-01 9.07896936e-01 9.05215263e-01 -4.41316903e-01 3.03373754e-01 4.35613930e-01 5.05801141e-01 4.40802395e-01 1.66294068e-01 1.25938684e-01 -3.85838628e-01 -6.20215610e-02 7.16785789e-02 3.03460937e-02 -3.01660836e-01 -2.28686348e-01 -8.80068958e-01 7.40715802e-01 6.18642449e-01 3.04494172e-01 -3.45300287e-01 1.95384726e-01 4.56037879e-01 4.77474183e-01 6.34977400e-01 4.45750892e-01 -7.12644637e-01 5.67187481e-02 -8.37924540e-01 6.38547838e-02 6.24385953e-01 5.72442293e-01 7.92695165e-01 -2.08554119e-02 -4.24352825e-01 1.26640737e+00 6.40823394e-02 3.18968683e-01 4.84681666e-01 -1.00283301e+00 4.19936717e-01 6.44721031e-01 -1.42692208e-01 -5.73429227e-01 -4.97396141e-01 -1.07153404e+00 -9.41724062e-01 3.48910600e-01 5.69280624e-01 -3.67268682e-01 -1.19381917e+00 1.88936877e+00 2.73753814e-02 1.01075850e-01 -3.32772844e-02 8.92020643e-01 7.90801287e-01 3.29183966e-01 2.71662623e-01 2.32898414e-01 9.76856053e-01 -1.03726578e+00 -5.29124260e-01 -4.27678734e-01 8.96569490e-01 -5.17057598e-01 9.17931855e-01 6.44615591e-01 -7.36379504e-01 -6.38652205e-01 -8.29014063e-01 4.60978188e-02 -3.45701426e-01 3.85249019e-01 1.53785601e-01 4.72489536e-01 -8.04572999e-01 1.04405200e+00 -5.57344079e-01 1.29043326e-01 8.90668213e-01 3.71713221e-01 -3.39117020e-01 -3.72505575e-01 -1.07595050e+00 8.28051090e-01 4.83034849e-01 2.06804216e-01 -8.92380834e-01 -6.48053944e-01 -5.00094175e-01 2.55834639e-01 5.21789789e-01 -1.92571849e-01 1.29968882e+00 -1.51739740e+00 -9.38769877e-01 8.07789862e-01 2.23046914e-01 -7.23618925e-01 6.43222392e-01 -3.15884978e-01 -3.12879920e-01 -2.31194824e-01 -5.00814244e-02 9.82576430e-01 8.89257312e-01 -1.44560337e+00 -7.44890213e-01 -2.47895002e-01 3.34321670e-02 -1.82298884e-01 -3.99730384e-01 -4.16146368e-01 -8.39707851e-02 -8.41327190e-01 6.60363585e-02 -1.05008459e+00 -1.56994924e-01 -9.26608592e-03 -4.46472377e-01 -1.73537657e-01 7.87979066e-01 -2.67434627e-01 8.76320720e-01 -2.34969735e+00 -1.16503827e-01 2.77775764e-01 3.53223264e-01 6.93832755e-01 -6.31352186e-01 -7.12861717e-02 -3.03735912e-01 2.78461754e-01 -2.31383085e-01 -7.09487677e-01 -2.72226870e-01 6.37891829e-01 -1.72574133e-01 2.51229793e-01 4.77364033e-01 5.30801296e-01 -9.27044272e-01 2.82882713e-03 -3.47435102e-02 4.29368138e-01 -4.29240942e-01 1.48452282e-01 -2.69383848e-01 3.24700445e-01 -4.46389653e-02 3.04939121e-01 7.55324125e-01 -1.41943827e-01 3.02868579e-02 -2.91424245e-01 3.32267731e-01 2.58191288e-01 -1.18687499e+00 1.21375167e+00 -5.83198309e-01 5.08340478e-01 2.67384965e-02 -1.12269437e+00 8.42019141e-01 2.42039010e-01 2.78220296e-01 -7.11039722e-01 2.57035762e-01 1.97353274e-01 2.28137866e-01 -5.57483435e-01 6.21690713e-02 1.28156334e-01 3.00548792e-01 3.06623250e-01 4.26937580e-01 4.43960071e-01 3.27050209e-01 4.88083623e-02 1.16396129e+00 -1.74959511e-01 -1.97434902e-01 -5.27777076e-02 3.75047475e-01 -1.38137996e-01 6.89663172e-01 1.03706408e+00 -9.87117887e-02 6.49993896e-01 5.34999132e-01 -5.54704905e-01 -9.91900444e-01 -6.55879021e-01 -1.87496454e-01 1.32143652e+00 -3.38552088e-01 -2.38554940e-01 -6.86150014e-01 -1.24420738e+00 -4.52504084e-02 9.35686827e-01 -8.53295982e-01 -4.64103669e-01 -4.80597645e-01 -9.36962485e-01 5.66667378e-01 3.84149820e-01 4.51744586e-01 -1.11179900e+00 -3.62697363e-01 3.47783029e-01 -2.35520869e-01 -9.76528585e-01 -2.21774936e-01 9.37094331e-01 -8.04967463e-01 -1.21400857e+00 -4.12192762e-01 -7.24314690e-01 9.86348450e-01 1.26548916e-01 1.29252744e+00 4.10988301e-01 -1.37464300e-01 1.14677117e-04 -5.16599834e-01 -5.21842718e-01 -6.68982208e-01 1.90898061e-01 -2.39774451e-01 1.82261825e-01 4.29174274e-01 -3.49540472e-01 -2.60520071e-01 4.01680768e-01 -9.63764429e-01 -2.15409756e-01 5.11856854e-01 1.33868885e+00 5.07984519e-01 1.90053642e-01 7.40956724e-01 -1.34525573e+00 2.49949858e-01 -5.51185787e-01 -5.88351727e-01 1.62465647e-01 -5.58249116e-01 3.36455733e-01 8.28682005e-01 -7.03177989e-01 -9.12784934e-01 1.25368088e-01 -2.05872327e-01 -4.71038133e-01 -1.64636046e-01 2.55520821e-01 -1.68945059e-01 8.27110037e-02 8.93864274e-01 -2.17672706e-01 -2.41931692e-01 -8.50020111e-01 2.43054703e-01 5.99945128e-01 4.41708565e-01 -3.55078459e-01 4.08387005e-01 2.74692327e-01 -8.05501342e-02 -4.61291552e-01 -1.39822328e+00 -3.58928889e-01 -5.46281338e-01 -7.28501379e-02 3.73930842e-01 -8.12573612e-01 -5.12481272e-01 6.25692725e-01 -1.09008145e+00 -4.05818909e-01 -6.35329723e-01 4.00579900e-01 7.97479376e-02 -1.77426353e-01 -5.68405807e-01 -7.31094420e-01 -4.56353463e-02 -1.26661289e+00 6.07577860e-01 1.21495714e-02 -6.98607638e-02 -7.76506484e-01 -1.73638076e-01 1.79390699e-01 4.22055840e-01 -6.97896928e-02 8.61389041e-01 -1.12025654e+00 -2.69771427e-01 -2.45839447e-01 -2.97690004e-01 1.20479226e+00 -1.35765389e-01 -1.72609627e-01 -1.68483973e+00 -2.35862762e-01 2.12774158e-01 -6.76385999e-01 1.45681322e+00 1.72107786e-01 1.48647630e+00 -4.28630531e-01 -4.96596098e-02 3.89176577e-01 1.38933849e+00 -5.60978502e-02 4.14959371e-01 3.58853012e-01 6.42696440e-01 6.93842709e-01 3.06163579e-01 1.32586554e-01 7.59008005e-02 4.04839188e-01 7.74826705e-01 -9.14699584e-02 -3.71357471e-01 -6.19090274e-02 1.94210678e-01 3.33534747e-01 3.01543564e-01 -4.22778219e-01 -7.31123686e-01 4.79155570e-01 -1.86167789e+00 -5.57833910e-01 -8.82267207e-02 2.13385463e+00 8.91521633e-01 3.79312724e-01 -2.63899297e-01 3.71643692e-01 5.71022093e-01 -4.30457555e-02 -5.28032839e-01 -9.19931605e-02 -2.27815881e-01 1.86799765e-01 7.57309735e-01 3.91933650e-01 -1.46913838e+00 7.59927571e-01 5.46304274e+00 8.62774968e-01 -1.09569788e+00 2.37973213e-01 1.03397751e+00 -4.02121544e-01 3.96778211e-02 -2.27786064e-01 -8.09363842e-01 5.56145310e-01 9.32049274e-01 6.00878179e-01 2.88107961e-01 1.02725542e+00 4.20317724e-02 -2.81595528e-01 -1.24721074e+00 8.59829843e-01 -6.49525598e-02 -1.07329965e+00 -8.93993303e-02 -1.45458281e-01 7.76375055e-01 4.01975274e-01 3.53591107e-02 4.22458500e-01 4.52365100e-01 -1.00264978e+00 9.24576163e-01 5.43678224e-01 2.81094551e-01 -7.94553757e-01 1.31510711e+00 5.08779287e-01 -3.49717975e-01 -3.65580082e-01 -6.17997110e-01 -1.73466410e-02 -3.27948779e-01 1.29092228e+00 -8.04795086e-01 2.66927537e-02 7.91767299e-01 5.48198760e-01 -9.33586001e-01 1.09921551e+00 -4.16799009e-01 9.41067576e-01 -2.45264217e-01 1.68553069e-01 2.61107445e-01 3.64937000e-02 3.06653798e-01 1.10005939e+00 -2.53286716e-02 -1.81383401e-01 2.85676003e-01 6.79624319e-01 -6.41401768e-01 -2.32887939e-01 -4.00253713e-01 2.81408310e-01 3.83434147e-01 1.26034474e+00 -5.66786647e-01 -4.23074931e-01 -2.30199978e-01 6.68028653e-01 5.94101131e-01 4.06210423e-01 -5.26984036e-01 -9.37639102e-02 4.38614011e-01 1.28881544e-01 2.86849648e-01 2.16696471e-01 -4.21659589e-01 -8.90453935e-01 1.00892551e-01 -8.68023992e-01 3.33345205e-01 -5.46768308e-01 -1.53046823e+00 8.62981915e-01 -3.52824032e-01 -1.09229326e+00 1.21892773e-01 -6.08654261e-01 -2.54654467e-01 7.86625445e-01 -1.65355194e+00 -8.40822935e-01 -1.45604461e-01 1.79382443e-01 4.17623430e-01 -2.15842932e-01 6.73714161e-01 6.23634100e-01 -5.75618029e-01 7.11777925e-01 2.65389085e-01 5.31704366e-01 9.14531171e-01 -1.22892630e+00 1.61874607e-01 7.58889973e-01 2.85888642e-01 2.28672251e-01 4.96492893e-01 -3.80747527e-01 -4.55247462e-01 -1.42558050e+00 8.89499962e-01 -2.82299846e-01 3.90350282e-01 -4.48498458e-01 -1.15520966e+00 4.33406323e-01 -1.69746295e-01 4.63453650e-01 4.28942591e-01 5.34018837e-02 -5.58680773e-01 -3.72184783e-01 -1.22285116e+00 3.41132432e-01 9.88490403e-01 -2.85409182e-01 -2.72296131e-01 4.35640782e-01 5.35925031e-01 -2.05486268e-01 -4.03159916e-01 2.88657516e-01 2.86163926e-01 -9.91846859e-01 6.48719192e-01 -6.07152760e-01 2.14677706e-01 -1.53658852e-01 -1.70594037e-01 -1.70522916e+00 -2.67649621e-01 2.63959039e-02 2.82043088e-02 1.35141242e+00 5.87142587e-01 -3.77692610e-01 7.35610604e-01 4.38294053e-01 -1.31765470e-01 -6.37809694e-01 -8.41573596e-01 -7.28408456e-01 -1.76888064e-01 -6.34790063e-01 4.24653590e-01 8.35191011e-01 -7.16660321e-01 3.02378982e-01 -4.50619400e-01 9.25110187e-03 6.37380600e-01 -6.09567761e-01 3.97643507e-01 -1.39143634e+00 -1.50977314e-01 -2.42141813e-01 -2.32631564e-01 -5.60681283e-01 2.75540650e-01 -9.97038186e-01 2.07097501e-01 -9.71275270e-01 6.82561472e-02 -7.79394686e-01 -4.63918924e-01 1.00190055e+00 -1.45232588e-01 5.06263554e-01 4.33662176e-01 1.70804262e-01 -6.12397790e-01 4.42145884e-01 1.09230161e+00 -3.74381393e-01 1.23865694e-01 2.41258308e-01 -5.89568019e-01 1.02615476e+00 6.94758236e-01 -1.23581028e+00 -4.30391580e-01 -7.31990397e-01 2.68658757e-01 -6.29751503e-01 6.22978032e-01 -1.19620466e+00 1.24076299e-01 3.30857873e-01 6.00335896e-01 -3.16924930e-01 1.43977627e-01 -1.25728536e+00 -1.50128141e-01 4.15275604e-01 -8.17500055e-01 -2.83381790e-01 6.85531506e-03 4.08180624e-01 -2.05900922e-01 -7.90277243e-01 1.19357574e+00 -2.21101165e-01 -5.36862731e-01 1.49669528e-01 -1.90470561e-01 2.77392238e-01 5.35007417e-01 1.78055197e-01 -3.46490026e-01 -2.38296434e-01 -1.15746093e+00 3.53980124e-01 2.32014090e-01 4.34077382e-01 2.64736027e-01 -1.16451192e+00 -6.96372151e-01 2.89502919e-01 1.72933608e-01 2.40411654e-01 2.45047137e-01 4.69428629e-01 -6.05713502e-02 4.60532270e-02 6.59669414e-02 -6.31209433e-01 -1.17871773e+00 4.25693542e-01 6.59157395e-01 -4.15351868e-01 -4.69282776e-01 1.00457454e+00 -9.58295073e-03 -5.62985361e-01 8.01350594e-01 -4.30472374e-01 -3.43710303e-01 3.05471838e-01 6.33915305e-01 3.17301840e-01 6.07013762e-01 -3.64809185e-01 -1.95945486e-01 1.40995592e-01 -3.27093869e-01 1.14164837e-01 1.48724878e+00 -5.18269278e-02 3.23447399e-02 4.97799963e-01 1.32567155e+00 -3.22479874e-01 -1.55345511e+00 -6.46171987e-01 2.56546736e-01 -2.49915183e-01 3.52327168e-01 -1.27055299e+00 -1.46606684e+00 9.17283356e-01 9.30336297e-01 3.17388140e-02 1.02546477e+00 -2.25478292e-01 5.76034904e-01 6.92200720e-01 2.45339796e-02 -1.35226619e+00 2.03431919e-01 6.98020041e-01 5.16414285e-01 -1.39336514e+00 -2.24863037e-01 -5.69418259e-02 -6.24141276e-01 1.08737779e+00 6.19586468e-01 -1.35568410e-01 7.65630245e-01 3.07713091e-01 2.92304873e-01 -2.03732699e-02 -8.36675942e-01 -9.62908044e-02 1.50673106e-01 5.96405447e-01 1.30350292e-01 -9.75951329e-02 7.09922239e-03 7.98936129e-01 1.68210536e-01 2.84197126e-02 4.96428639e-01 8.81067872e-01 -4.89015162e-01 -1.14711916e+00 -3.00008088e-01 6.51153326e-01 -5.31021059e-01 -2.28477299e-01 -1.97225317e-01 5.15578449e-01 8.32489371e-01 9.89432812e-01 1.75641969e-01 -3.08283240e-01 5.41476011e-01 2.94553876e-01 1.39965400e-01 -7.36458838e-01 -7.61629879e-01 -3.64578404e-02 1.31209508e-01 -4.43954259e-01 -4.52010900e-01 -3.13618928e-01 -8.64475846e-01 -1.54257184e-02 -3.92584503e-01 1.44332558e-01 7.13055730e-01 9.73504364e-01 2.30195940e-01 8.95053625e-01 6.86344445e-01 -5.84291101e-01 -9.00427461e-01 -1.20832860e+00 -4.94892836e-01 6.85572088e-01 5.70047021e-01 -6.47394180e-01 -4.48503256e-01 4.35741395e-02]
[9.389491081237793, 3.8142952919006348]
7787eb27-7967-4e8a-a351-c87519b17474
vrdformer-end-to-end-video-visual-relation
null
null
http://openaccess.thecvf.com//content/CVPR2022/html/Zheng_VRDFormer_End-to-End_Video_Visual_Relation_Detection_With_Transformers_CVPR_2022_paper.html
http://openaccess.thecvf.com//content/CVPR2022/papers/Zheng_VRDFormer_End-to-End_Video_Visual_Relation_Detection_With_Transformers_CVPR_2022_paper.pdf
VRDFormer: End-to-End Video Visual Relation Detection With Transformers
Visual relation understanding plays an essential role for holistic video understanding. Most previous works adopt a multi-stage framework for video visual relation detection (VidVRD), which cannot capture long-term spatiotemporal contexts in different stages and also suffers from inefficiency. In this paper, we propose a transformerbased framework called VRDFormer to unify these decoupling stages. Our model exploits a query-based approach to autoregressively generate relation instances. We specifically design static queries and recurrent queries to enable efficient object pair tracking with spatio-temporal contexts. The model is jointly trained with object pair detection and relation classification. Extensive experiments on two benchmark datasets, ImageNet-VidVRD and VidOR, demonstrate the effectiveness of the proposed VRDFormer, which achieves the state-of-the-art performance on both relation detection and relation tagging tasks.
['Qin Jin', 'ShiZhe Chen', 'Sipeng Zheng']
2022-01-01
null
null
null
cvpr-2022-1
['video-visual-relation-detection', 'relation-classification']
['computer-vision', 'natural-language-processing']
[-5.57097159e-02 -1.41808450e-01 -7.12336957e-01 -1.79153487e-01 -5.28659284e-01 -4.43801463e-01 8.16267312e-01 -9.99164432e-02 -1.17662273e-01 2.77828097e-01 2.08413959e-01 -5.35539865e-01 8.04653242e-02 -6.18132353e-01 -6.46140516e-01 -2.51252443e-01 1.89397801e-02 2.64086604e-01 8.07111144e-01 -1.14032373e-01 -1.47912279e-01 2.78572083e-01 -1.34834063e+00 6.60964966e-01 5.51931858e-01 1.09915900e+00 2.22008377e-01 7.83538103e-01 -1.93453096e-02 1.47841704e+00 -2.88297921e-01 -6.55703962e-01 -3.07333190e-03 -3.38791668e-01 -9.27912652e-01 9.45738181e-02 4.22274232e-01 -4.42824870e-01 -9.89886999e-01 7.85874307e-01 1.84695944e-01 1.89018995e-01 5.38631201e-01 -1.54131401e+00 -9.88541722e-01 4.75631028e-01 -7.62386918e-01 7.22857118e-01 7.62859762e-01 5.11112139e-02 1.13043058e+00 -1.05466282e+00 9.29505646e-01 1.47362649e+00 3.50001574e-01 3.34514380e-01 -1.11683881e+00 -6.61503255e-01 6.94870293e-01 7.39803016e-01 -1.69465005e+00 -3.86239976e-01 7.19486415e-01 -5.02004147e-01 1.25805759e+00 1.60797730e-01 7.84393132e-01 1.06862497e+00 4.29127477e-02 1.06778371e+00 5.69858372e-01 -2.02614129e-01 -1.93488017e-01 -1.79527581e-01 9.47874412e-02 7.74566174e-01 3.57080363e-02 1.91483527e-01 -8.41130495e-01 7.73384422e-02 1.04049575e+00 1.34454206e-01 -3.17852020e-01 -5.15228331e-01 -1.29360080e+00 4.79977548e-01 4.74287868e-01 1.26201704e-01 -2.59167492e-01 1.32033661e-01 5.87235391e-01 3.03571761e-01 5.52935898e-01 -9.94401351e-02 -9.56662819e-02 2.71399412e-02 -4.11738813e-01 1.89621016e-01 4.56656516e-01 1.51251423e+00 3.11363071e-01 -3.72502297e-01 -7.07091868e-01 6.97054744e-01 4.05888557e-01 2.61839867e-01 -3.64415236e-02 -9.02734578e-01 6.89658880e-01 8.61550391e-01 -1.35952070e-01 -1.16588223e+00 -1.69521756e-02 -2.44639099e-01 -7.74630666e-01 -2.24878699e-01 -1.15602985e-01 4.22670424e-01 -9.47961688e-01 1.55484116e+00 5.63189864e-01 6.61900342e-01 1.85217485e-01 1.00256610e+00 1.40774393e+00 6.74516737e-01 5.83630264e-01 -2.64645666e-01 1.66667485e+00 -1.30409825e+00 -9.44468617e-01 -1.06489561e-01 5.49945414e-01 -5.83256185e-01 8.60551834e-01 2.05809367e-03 -1.00984979e+00 -7.71822393e-01 -8.62131715e-01 -4.36437637e-01 -2.23841280e-01 4.54531685e-02 8.83359671e-01 2.02844813e-02 -9.63843822e-01 -1.46338522e-01 -9.32310998e-01 -3.90773088e-01 6.97914422e-01 5.44579327e-02 -5.72602510e-01 -1.21886872e-01 -1.15830970e+00 6.00380301e-01 5.17273724e-01 2.33876169e-01 -1.07606721e+00 -5.62129200e-01 -1.02660465e+00 -7.42392167e-02 9.25226808e-01 -9.15682912e-01 1.26675904e+00 -3.93307775e-01 -1.01020145e+00 9.73828197e-01 -5.47814250e-01 -5.31508565e-01 3.39679241e-01 -4.18723285e-01 -4.52537894e-01 4.05314058e-01 1.69156324e-02 5.62938750e-01 6.75139666e-01 -1.30992186e+00 -8.78243148e-01 -4.85420041e-02 4.22069371e-01 4.30279374e-01 -8.40434581e-02 4.87824261e-01 -1.34016097e+00 -8.58756006e-01 2.15619653e-01 -6.74333274e-01 7.12112337e-02 6.72347844e-02 -3.04273546e-01 -6.72617733e-01 1.13504374e+00 -4.84934896e-01 1.37600780e+00 -2.05372238e+00 1.89280450e-01 -1.94068775e-01 5.23049712e-01 5.06537080e-01 -1.61519840e-01 2.87351012e-01 -2.09609360e-01 1.39656961e-02 2.81465292e-01 -3.59720916e-01 -2.56516278e-01 2.96041936e-01 -3.75297964e-01 1.19137764e-01 4.28581327e-01 1.28976560e+00 -1.18251610e+00 -8.93446624e-01 1.96478784e-01 6.35527134e-01 -2.66087085e-01 5.48765004e-01 -4.02483404e-01 2.81409889e-01 -5.53980291e-01 9.35172319e-01 4.89093363e-01 -6.13985062e-01 4.53900874e-01 -6.47230983e-01 2.12132037e-01 2.24499404e-01 -8.30889881e-01 1.68407714e+00 -2.00829446e-01 7.55316138e-01 -3.80466104e-01 -8.66414666e-01 7.04453528e-01 4.42438006e-01 3.63349646e-01 -8.92074585e-01 -5.72391860e-02 -3.65827322e-01 -2.91477919e-01 -8.64575028e-01 5.17997563e-01 3.21549416e-01 2.16375232e-01 1.21205002e-01 1.81790948e-01 5.52043080e-01 1.40294686e-01 7.03753769e-01 1.09298110e+00 7.78564572e-01 5.61853051e-01 2.70288855e-01 7.43878186e-01 -1.11630455e-01 6.68490529e-01 6.72536552e-01 -3.87443274e-01 4.74473029e-01 8.08162570e-01 -4.44271326e-01 -5.89682877e-01 -1.15328979e+00 3.30732822e-01 1.15786278e+00 8.14095557e-01 -8.65812063e-01 -1.37835383e-01 -1.06734550e+00 -2.31357485e-01 3.82939905e-01 -5.96124530e-01 -3.36258620e-01 -7.36646235e-01 -3.94176155e-01 3.59312832e-01 9.22303677e-01 6.33861959e-01 -9.17747617e-01 -4.43074852e-01 6.48967922e-02 -6.14127636e-01 -1.77506840e+00 -4.77849215e-01 -3.23805094e-01 -5.47325611e-01 -1.13324392e+00 -3.68368864e-01 -9.95034814e-01 3.47265065e-01 7.71636963e-01 1.52050424e+00 1.61476895e-01 -2.66612396e-02 3.12995404e-01 -5.87275803e-01 4.52553034e-02 -1.17014445e-01 9.34156775e-03 -3.51214558e-01 4.46304604e-02 4.93978620e-01 -4.60798383e-01 -6.29717529e-01 4.65373397e-01 -6.53108776e-01 3.63496929e-01 5.24899244e-01 7.59130836e-01 9.36917901e-01 -9.87405777e-02 4.02759194e-01 -7.88825154e-01 2.13039666e-01 -4.96395826e-01 -5.19111037e-01 7.58364499e-01 -3.64130795e-01 -1.64867744e-01 1.62706241e-01 -5.53918958e-01 -1.30778670e+00 4.35053669e-02 1.77263662e-01 -1.05408096e+00 -1.34414047e-01 2.33330935e-01 -4.10735905e-01 7.85603281e-03 2.13448986e-01 1.83561698e-01 -4.52110380e-01 -3.65956157e-01 5.56675553e-01 2.77043343e-01 9.30402577e-01 -2.43572354e-01 7.55729496e-01 4.98074353e-01 2.97479369e-02 -3.24409038e-01 -9.78542805e-01 -6.81884527e-01 -6.09489560e-01 -3.42410266e-01 1.20459557e+00 -1.41093123e+00 -1.00230026e+00 1.91366404e-01 -1.37995839e+00 -3.29703599e-01 -2.88184006e-02 4.07730162e-01 -3.35208207e-01 2.73225278e-01 -7.08955944e-01 -8.37495863e-01 -2.62105614e-01 -1.00327063e+00 1.17177355e+00 2.65272617e-01 1.10069990e-01 -8.43799829e-01 -1.44320771e-01 3.74446660e-01 -1.40134141e-01 1.84423625e-01 6.21052086e-01 -3.36343169e-01 -1.23388648e+00 2.20974952e-01 -7.57474363e-01 -2.18227610e-01 9.06495824e-02 -2.54590511e-02 -8.56072307e-01 2.11987466e-01 -4.87422496e-01 -3.46200466e-01 1.06065810e+00 9.46406201e-02 1.04507160e+00 -5.87982610e-02 -7.99571395e-01 7.25303054e-01 1.24856496e+00 4.17179167e-01 7.40926504e-01 1.53680101e-01 1.21411407e+00 4.48003262e-01 1.32026255e+00 1.42478138e-01 9.61493552e-01 1.09232318e+00 4.24077332e-01 -2.08953276e-01 -5.21758497e-01 -4.68239367e-01 1.70497760e-01 5.18785596e-01 -3.59469026e-01 -5.08248568e-01 -8.64146709e-01 7.35092461e-01 -2.39183211e+00 -1.07133055e+00 -2.89120227e-01 1.69563425e+00 5.88567436e-01 1.48331404e-01 2.35392734e-01 -1.76069036e-01 7.00089812e-01 3.83781761e-01 -2.85437316e-01 1.08470872e-01 -1.29842699e-01 -1.41292334e-01 8.99265930e-02 3.93584341e-01 -1.44910681e+00 1.39378631e+00 5.87126684e+00 7.33470261e-01 -6.77542984e-01 2.82182485e-01 4.98389214e-01 1.85620248e-01 -5.41018210e-02 1.62055448e-01 -7.11202741e-01 -3.60683980e-03 4.44802463e-01 1.67314708e-01 1.08702637e-01 6.67892933e-01 6.72620013e-02 -2.27765292e-01 -1.18911135e+00 1.31787753e+00 1.22702584e-01 -1.39149857e+00 2.85602957e-01 -2.69432485e-01 3.54260087e-01 -3.62116873e-01 -8.14425051e-02 2.70153314e-01 2.52481580e-01 -8.03841293e-01 7.98983395e-01 5.78417659e-01 9.23294067e-01 -6.29306734e-01 6.66965902e-01 -1.38328850e-01 -2.05650473e+00 2.39756063e-01 -1.06265284e-01 1.30708637e-02 4.92525399e-01 1.53927967e-01 -7.29621112e-01 1.03800511e+00 8.80651593e-01 1.24180222e+00 -6.69981122e-01 7.49654174e-01 -4.27479595e-01 4.91258949e-01 -8.70572478e-02 3.04274827e-01 -1.05512105e-01 6.93607703e-02 5.70887566e-01 1.20501971e+00 -1.48251042e-01 4.30447668e-01 3.35454732e-01 5.93255460e-01 -7.91382715e-02 -2.44457692e-01 -6.80236161e-01 3.15118358e-02 7.05989420e-01 1.14965522e+00 -7.75648832e-01 -5.63364446e-01 -7.31401265e-01 9.77739930e-01 4.37977165e-01 6.14588499e-01 -1.22511995e+00 -1.71285197e-02 7.62029231e-01 -1.62026197e-01 7.27525830e-01 -2.67182231e-01 -3.38203809e-03 -1.46240282e+00 1.41336307e-01 -6.99967206e-01 8.47678840e-01 -8.66746366e-01 -1.08943319e+00 7.50554502e-01 3.58297020e-01 -1.27393544e+00 -6.93317577e-02 -2.60591358e-01 -2.82111734e-01 5.01197815e-01 -1.55812645e+00 -1.58500779e+00 -6.28324986e-01 8.26995969e-01 6.74488544e-01 1.95323810e-01 4.39063489e-01 5.60513377e-01 -7.79699266e-01 5.52938759e-01 -8.23412001e-01 2.76901096e-01 5.92955232e-01 -1.02358282e+00 5.78514457e-01 1.04568541e+00 5.13872862e-01 4.70381528e-01 4.37121660e-01 -8.26701522e-01 -1.35389507e+00 -1.45038044e+00 9.83956754e-01 -6.48473859e-01 4.64683264e-01 -5.41458011e-01 -1.05902171e+00 9.83138502e-01 3.05066645e-01 4.64697987e-01 5.36955059e-01 1.11250915e-01 -7.24871457e-01 -1.86875641e-01 -5.05094886e-01 7.25664914e-01 1.88546503e+00 -8.28774691e-01 -6.49579048e-01 2.79589415e-01 1.16239119e+00 -5.40662408e-01 -9.87580240e-01 6.85793281e-01 5.97174704e-01 -9.27443206e-01 1.30993760e+00 -6.45387471e-01 4.35962528e-01 -5.73489666e-01 -9.28653851e-02 -4.77340609e-01 -4.27332610e-01 -6.04586363e-01 -9.82013464e-01 1.64339495e+00 1.46063685e-01 -3.57999235e-01 5.41359484e-01 2.70809948e-01 1.48502886e-01 -9.69984531e-01 -5.14743507e-01 -8.44687402e-01 -7.66492784e-01 -5.24947524e-01 5.30054808e-01 7.90496528e-01 -2.53812373e-01 8.08495760e-01 -4.40362245e-01 3.48717868e-01 5.10318577e-01 3.68524224e-01 8.29099476e-01 -7.22060382e-01 -3.53334457e-01 -1.30712852e-01 -7.50789046e-01 -1.47370934e+00 1.10972829e-01 -5.95853031e-01 -5.53252287e-02 -1.68548334e+00 5.98071039e-01 -2.76774168e-01 -3.78471702e-01 5.61572611e-01 -6.04420900e-01 3.45100403e-01 2.65332669e-01 3.60907823e-01 -1.38057148e+00 6.55616403e-01 1.19516146e+00 -1.13767229e-01 -2.53184944e-01 -3.39269221e-01 -5.28185248e-01 6.30754590e-01 3.08629125e-01 -3.55657339e-01 -9.91464257e-01 -5.59687376e-01 5.45900352e-02 2.12869689e-01 7.42806733e-01 -7.44261444e-01 2.02421159e-01 -1.97961465e-01 1.91398576e-01 -9.92003024e-01 4.54542845e-01 -6.56014860e-01 3.48178536e-01 5.32128736e-02 -3.61141294e-01 8.92747343e-02 -5.33510894e-02 9.44366455e-01 -4.13196981e-01 6.19176626e-01 2.58784562e-01 1.51659101e-01 -1.19308627e+00 6.30774081e-01 -7.58696869e-02 1.39861807e-01 1.35965872e+00 -9.99544337e-02 -5.37499487e-01 -4.11174417e-01 -5.39127767e-01 4.78149503e-01 2.77696848e-01 6.85038865e-01 9.34673190e-01 -1.62294173e+00 -5.29607713e-01 -1.50839016e-01 5.65874100e-01 1.94535866e-01 3.23319912e-01 9.62974250e-01 -2.80413538e-01 3.25502813e-01 2.95858979e-01 -8.66346121e-01 -1.64471221e+00 1.04781461e+00 1.28296688e-01 -4.51244295e-01 -9.77486670e-01 1.11736310e+00 7.01821566e-01 2.79669493e-01 3.08230460e-01 -3.63082856e-01 -5.03416359e-01 -3.10172159e-02 6.26433969e-01 -8.53215158e-02 -2.66162604e-01 -8.76276612e-01 -5.81622422e-01 5.41339993e-01 -3.06021124e-01 1.27827451e-01 8.69991779e-01 -4.24448878e-01 1.07422672e-01 3.05695713e-01 1.09049642e+00 -4.20231909e-01 -1.43475211e+00 -5.63244879e-01 2.03412864e-02 -7.05346346e-01 -1.02760203e-01 -3.70273620e-01 -1.18157434e+00 6.33174777e-01 3.44710976e-01 1.06563345e-01 1.36456394e+00 6.54984236e-01 8.55726182e-01 1.19052485e-01 2.80954182e-01 -4.94391054e-01 2.69576579e-01 3.11654449e-01 7.86339104e-01 -1.21807420e+00 7.57710263e-02 -1.20238411e+00 -7.94176877e-01 7.38544881e-01 9.69866395e-01 9.37023759e-02 5.26818693e-01 1.82248458e-01 3.59625705e-02 -3.37870568e-01 -1.16431534e+00 -5.64283550e-01 5.75037956e-01 8.24967682e-01 4.21140522e-01 -1.40391245e-01 -1.68239340e-01 4.81677860e-01 5.58449984e-01 1.55216128e-01 -1.42318428e-01 1.05421317e+00 9.87643749e-02 -1.12772059e+00 -1.10699058e-01 6.07903153e-02 -3.48600388e-01 -1.79216430e-01 -4.13723856e-01 8.05283248e-01 2.53944665e-01 1.02283096e+00 1.30583376e-01 -7.35965967e-01 3.39091301e-01 -2.23471567e-01 5.02302110e-01 -4.47993338e-01 -2.96089023e-01 1.44874081e-01 4.24008638e-01 -9.89765942e-01 -1.02209520e+00 -5.97994030e-01 -1.19938993e+00 6.93067238e-02 -3.90372366e-01 -1.94224238e-01 -1.85933277e-01 8.54369044e-01 5.58168352e-01 8.93456697e-01 3.05286407e-01 -4.09211457e-01 2.72649974e-01 -6.48028851e-01 -7.48307407e-02 4.51097071e-01 2.85002768e-01 -1.13416588e+00 4.50139850e-01 4.61434931e-01]
[9.435338973999023, 0.776303768157959]
fe60c03b-e4db-470c-ae75-eb0ff8ced870
towards-automatic-manipulation-of-intra
2009.05859
null
https://arxiv.org/abs/2009.05859v3
https://arxiv.org/pdf/2009.05859v3.pdf
Towards Automatic Manipulation of Intra-cardiac Echocardiography Catheter
Intra-cardiac Echocardiography (ICE) is a powerful imaging modality for guiding electrophysiology and structural heart interventions. ICE provides real-time observation of anatomy, catheters, and emergent complications. However, this increased reliance on intraprocedural imaging creates a high cognitive demand on physicians who can often serve as interventionalist and imager. We present a robotic manipulator for ICE catheters to assist physicians with imaging and serve as a platform for developing processes for procedural automation. Herein, we introduce two application modules towards these goals: (1) a view recovery process that allows physicians to save views during intervention and automatically return with the push of a button and (2) a data-driven approach to compensate kinematic model errors that result from non-linear behaviors in catheter bending, providing more precise control of the catheter tip. View recovery is validated by repeated catheter positioning in cardiac phantom and animal experiments with position- and image-based analysis. We present a simplified calibration approach for error compensation and verify with complex rotation of the catheter in benchtop and phantom experiments under varying realistic curvature conditions. Results support that a robotic manipulator for ICE can provide an efficient and reproducible tool, potentially reducing execution time and promoting greater utilization of ICE imaging.
['C. Huie Lin', 'Ankur Kapoor', 'Ponraj Chinnadurai', 'Zhongyu Li', 'Young-Ho Kim', 'Tommaso Mansi', 'Jarrod Collins']
2020-09-12
null
null
null
null
['non-linear-elasticity']
['miscellaneous']
[-8.54112282e-02 -1.45757720e-01 2.67243564e-01 5.45246266e-02 -2.87255883e-01 -1.35229158e+00 -2.46454686e-01 1.16855644e-01 -1.99230537e-01 3.88801247e-01 -2.88219750e-01 -1.11512518e+00 -3.90114814e-01 -1.43137261e-01 -5.04594445e-01 -2.98549742e-01 -3.07580173e-01 5.78414083e-01 -8.53314027e-02 2.04163760e-01 3.01587999e-01 1.18438566e+00 -4.08921719e-01 -2.32476637e-01 8.83752167e-01 4.76032197e-01 5.21912575e-01 1.00126851e+00 5.50422847e-01 6.27961934e-01 -5.00831664e-01 2.04594463e-01 4.35779929e-01 -4.13215667e-01 -4.67387736e-01 8.61807317e-02 -7.26241916e-02 -8.02960396e-01 9.55620110e-02 2.55425900e-01 1.12003887e+00 -3.58544916e-01 3.50141406e-01 -6.14082992e-01 1.28854513e-01 3.65329444e-01 -2.71084338e-01 2.03131720e-01 2.34346494e-01 4.86661851e-01 6.01444542e-02 -6.73346281e-01 1.00160825e+00 4.93198246e-01 8.84076416e-01 1.38050362e-01 -1.29477537e+00 -1.29690751e-01 -6.17187738e-01 -4.33318555e-01 -8.54195893e-01 -7.90467709e-02 6.86569095e-01 -9.37289894e-01 4.71225768e-01 3.22926819e-01 9.91446853e-01 4.39387888e-01 8.75988662e-01 -2.70405173e-01 8.42227519e-01 -4.58568990e-01 3.04377768e-02 4.08112764e-01 -1.61921196e-02 6.02195799e-01 4.88589853e-01 4.27437812e-01 2.07673922e-01 -3.33337426e-01 1.69215047e+00 1.69736937e-01 -8.42074692e-01 -8.64452600e-01 -1.59983969e+00 3.39137405e-01 6.41348511e-02 2.36662760e-01 -6.98573887e-01 3.11792493e-02 4.17812705e-01 2.87636906e-01 -5.61247289e-01 9.94713187e-01 -6.33588076e-01 -6.49381340e-01 -4.92425144e-01 2.26748347e-01 8.56147587e-01 9.17543888e-01 -3.62460554e-01 3.42329517e-02 -3.53691936e-01 6.21304870e-01 3.71608347e-01 5.91343105e-01 1.07793994e-01 -1.43330777e+00 2.96054989e-01 3.52631807e-01 6.39151037e-01 -7.09943056e-01 -9.09997165e-01 -7.40917265e-01 -5.72153151e-01 7.14210212e-01 3.76611203e-01 -5.67873359e-01 -5.93127012e-01 1.01986897e+00 3.56239498e-01 -2.92797685e-01 -3.13511819e-01 1.17601383e+00 4.05164480e-01 4.10491452e-02 -8.50179344e-02 -5.61740637e-01 1.47645175e+00 -5.95292270e-01 -5.58214784e-01 1.31403819e-01 9.62694168e-01 -8.98044288e-01 1.00295961e+00 4.52527642e-01 -1.23689640e+00 -6.78220749e-01 -9.62550998e-01 5.38355649e-01 3.88961554e-01 4.82243121e-01 3.27470899e-01 8.67649913e-01 -8.40014994e-01 1.03055358e+00 -1.17086720e+00 -1.77943692e-01 2.30724081e-01 2.82232553e-01 -2.27351487e-01 2.47098088e-01 -4.06592041e-01 1.26522350e+00 -9.99072343e-02 3.78576159e-01 -4.97273147e-01 -1.29820299e+00 -5.37504435e-01 -3.55496979e-03 8.35041702e-03 -9.93038058e-01 1.06418526e+00 2.17953056e-01 -1.80304587e+00 7.14585721e-01 1.53914750e-01 -1.99782073e-01 1.01425397e+00 -5.35266578e-01 -1.29321113e-01 5.33152699e-01 9.78180990e-02 1.66004300e-01 5.56914508e-01 -1.01600385e+00 1.29404992e-01 -4.40704972e-01 -2.46477857e-01 7.42666721e-02 3.45875382e-01 -6.29052371e-02 -3.12123388e-01 -5.24168193e-01 4.71322060e-01 -1.37772429e+00 -4.22839135e-01 4.09468204e-01 -1.40909523e-01 8.47721457e-01 8.99687648e-01 -8.04115057e-01 1.11967885e+00 -1.94214451e+00 8.23196769e-02 3.67291451e-01 3.94156754e-01 5.92550933e-01 6.23245358e-01 5.63332558e-01 -2.58950531e-01 1.68449450e-02 1.20862555e-02 3.98126841e-01 -5.64017653e-01 -2.50291735e-01 6.04869332e-03 3.95798892e-01 -3.62091660e-01 9.60360646e-01 -5.56538999e-01 -3.17429334e-01 8.29812765e-01 1.86003760e-01 -3.61721337e-01 4.21222150e-01 5.57957172e-01 1.53412080e+00 -5.06409585e-01 4.81968611e-01 7.84220695e-01 -4.33520943e-01 6.49307370e-01 -2.31468290e-01 -4.72266525e-01 -5.98142073e-02 -9.70692813e-01 1.67878127e+00 -6.45463169e-01 7.28964508e-02 5.66291630e-01 -4.24754709e-01 9.69669938e-01 5.92972577e-01 8.03642452e-01 -2.50293523e-01 3.22128743e-01 3.13471973e-01 3.77031446e-01 -9.82146919e-01 -1.11426175e-01 -1.69472918e-01 3.84725511e-01 6.26349688e-01 -4.95065808e-01 -6.63944960e-01 -2.80122310e-01 1.36820227e-01 1.05098736e+00 4.57214594e-01 1.41693160e-01 -6.84076607e-01 4.54937071e-01 2.05458835e-01 3.49655360e-01 8.48186076e-01 -4.45017040e-01 7.62149751e-01 3.48266423e-01 -8.00507605e-01 -1.10439658e+00 -1.22490191e+00 -3.19481403e-01 2.67677736e-02 1.36271328e-01 -2.08401173e-01 -3.79976243e-01 -1.83966070e-01 2.88469672e-01 3.15852374e-01 5.19029517e-03 7.20916614e-02 -8.25165927e-01 -1.30759537e-01 1.97577953e-01 5.99727094e-01 -6.69867173e-02 -7.89669335e-01 -1.70107222e+00 6.91870630e-01 -3.10287494e-02 -1.00277102e+00 -2.28195533e-01 3.62565853e-02 -1.45285964e+00 -1.25691199e+00 -6.56568646e-01 -5.97698629e-01 7.64831603e-01 9.04127955e-02 8.95437956e-01 7.38211721e-02 -9.57230926e-01 5.59140682e-01 -2.33603626e-01 -5.05762219e-01 -7.79412508e-01 -2.20486253e-01 -1.74052536e-01 -9.09936666e-01 -7.82544434e-01 -7.52065063e-01 -1.21437919e+00 6.05792880e-01 -1.91119641e-01 8.82646516e-02 3.83853316e-01 5.08986056e-01 2.64605135e-01 -7.97672212e-01 5.68578482e-01 -9.40839887e-01 8.34751248e-01 4.64939252e-02 -9.50162292e-01 3.45570743e-02 -5.73225379e-01 -3.55385840e-01 8.05078149e-01 -8.55456218e-02 -8.38713765e-01 1.62356138e-01 1.47649765e-01 -6.75377727e-01 1.10590728e-02 4.13632870e-01 5.48276484e-01 -2.73449659e-01 8.64336133e-01 -3.03240329e-01 6.50635242e-01 -1.36393964e-01 1.01349227e-01 3.79918814e-01 5.29382348e-01 -6.88874602e-01 3.70962858e-01 3.32590908e-01 5.37858427e-01 -6.11853123e-01 2.01018989e-01 -2.64485478e-01 -8.41365457e-01 -5.01039326e-01 5.49619555e-01 -3.20559889e-01 -1.15070462e+00 -1.57723725e-02 -1.00117826e+00 -2.43343800e-01 -2.57270992e-01 1.03145063e+00 -4.37109500e-01 5.44298112e-01 -7.08210945e-01 -6.28269017e-01 -6.40504599e-01 -1.56324625e+00 7.00617492e-01 4.26763073e-02 -5.58018863e-01 -9.47767615e-01 -1.43018872e-01 7.80901238e-02 6.69412434e-01 9.11093175e-01 9.84143198e-01 1.41251951e-01 -6.58726096e-01 -4.25004691e-01 1.55658841e-01 1.08858481e-01 1.43555447e-01 7.65618235e-02 -5.73337018e-01 -5.21342039e-01 3.25112402e-01 1.52188480e-01 -3.74255240e-01 1.00753140e+00 1.09739339e+00 3.12137693e-01 -5.83724618e-01 6.26771450e-01 1.22506118e+00 6.33881807e-01 5.70679724e-01 7.51685118e-03 5.23377955e-01 3.18373382e-01 7.59166539e-01 5.85757613e-01 -1.71162799e-01 5.72415173e-01 1.56554937e-01 -2.30421975e-01 2.56354839e-01 1.83444768e-01 -4.96049225e-01 6.35717869e-01 -5.00682890e-01 2.66750723e-01 -1.05040038e+00 2.65784208e-02 -1.51709974e+00 -3.32027078e-01 -5.19307077e-01 2.66809416e+00 1.50307015e-01 8.95138085e-03 -3.96708921e-02 -1.46072045e-01 5.73012471e-01 -6.11778975e-01 -4.30591226e-01 -4.64132905e-01 7.46859610e-01 2.86800593e-01 7.23462641e-01 4.51120794e-01 -6.76709831e-01 1.12437122e-01 6.97097254e+00 -6.24597907e-01 -1.44717097e+00 -1.35150373e-01 4.37253892e-01 3.50780576e-01 6.08515926e-02 2.08658010e-01 -9.29032341e-02 3.48273158e-01 5.23062110e-01 -1.29235834e-01 9.37171131e-02 7.03460574e-01 7.64738977e-01 -2.52330333e-01 -1.15611804e+00 1.09835839e+00 -4.23172385e-01 -1.73928118e+00 -5.05495489e-01 -3.53446277e-03 2.03389585e-01 -2.67650366e-01 -4.05570745e-01 -1.46042854e-01 -5.50182760e-01 -5.95040083e-01 4.06539321e-01 6.61074579e-01 1.05052197e+00 -2.06068337e-01 4.80737656e-01 5.15569627e-01 -8.47223043e-01 6.06474653e-03 -5.39137796e-02 3.28771211e-02 6.87875271e-01 5.57171285e-01 -1.27453411e+00 6.98037028e-01 3.81663799e-01 3.19784492e-01 -4.76356782e-02 1.13341153e+00 5.55385575e-02 3.42828423e-01 -2.02841848e-01 4.04593915e-01 -3.37544709e-01 -5.79068244e-01 9.39645171e-01 6.80505693e-01 3.38457674e-01 9.64589417e-02 5.92902005e-02 1.12793469e+00 4.83955383e-01 -8.63791555e-02 -6.00642204e-01 1.58562049e-01 7.16901183e-01 1.44008136e+00 -9.82681215e-01 -1.00866742e-01 1.28051177e-01 6.32849634e-01 -2.58169830e-01 2.17715308e-01 -9.93241668e-01 -4.61889714e-01 1.53272739e-02 7.39471257e-01 3.57016474e-02 -5.63099325e-01 -7.62095273e-01 -7.41053581e-01 3.84897172e-01 -4.36194420e-01 -1.25333024e-02 -9.80635583e-01 -3.23280215e-01 2.65620142e-01 -6.39956295e-02 -1.61894262e+00 -4.59586710e-01 -6.87125444e-01 -7.39355683e-01 1.16929841e+00 -9.92770433e-01 -7.59582996e-01 -5.60020208e-01 2.57278699e-02 1.62641674e-01 1.37012646e-01 1.08459449e+00 3.73679399e-01 4.12195846e-02 6.02802355e-03 -7.05056712e-02 -2.42319882e-01 7.87529945e-01 -1.08575189e+00 1.05258403e-02 4.86372083e-01 -8.35222960e-01 1.17380559e+00 4.98666883e-01 -7.58269846e-01 -1.82303822e+00 -7.57752836e-01 2.79856427e-03 -6.21522725e-01 2.93943044e-02 2.18953714e-01 -5.00908554e-01 5.43690860e-01 -1.97289646e-01 2.89716661e-01 5.21173298e-01 -2.05007091e-01 5.43413520e-01 -8.68324563e-02 -1.25834668e+00 6.17355824e-01 7.86042511e-01 -9.91144925e-02 -3.03734660e-01 2.40652576e-01 1.41016230e-01 -1.17425692e+00 -1.34052587e+00 5.88836551e-01 1.14696634e+00 -8.31311703e-01 7.69465685e-01 -2.70548165e-01 2.51708835e-01 -4.78991479e-01 8.72336626e-01 -1.08117723e+00 -3.77111405e-01 -1.11866593e+00 2.73885310e-01 4.73707587e-01 3.98385972e-01 -8.69221210e-01 4.95556414e-01 8.29161942e-01 -4.20092672e-01 -9.62983847e-01 -5.74517727e-01 -5.52310169e-01 -9.42375064e-02 -1.85917377e-01 -4.61427607e-02 7.41313696e-01 4.27404314e-01 -1.02765216e-02 1.28196245e-02 4.20139074e-01 3.61597985e-01 1.31342798e-01 7.63887763e-01 -1.16015041e+00 -4.28123921e-01 -5.18558128e-03 -4.36259151e-01 -6.52230561e-01 -7.87190855e-01 -6.40096724e-01 -2.80722111e-01 -1.52494240e+00 -2.34350204e-01 -8.45706284e-01 4.27623302e-01 -1.69923469e-01 1.71407834e-02 -2.50176936e-01 1.63210809e-01 5.29544413e-01 2.16963708e-01 -3.14360470e-01 1.73727369e+00 7.93325901e-01 -6.23736441e-01 2.35535190e-01 -3.36496532e-01 4.36615676e-01 7.61802495e-01 -3.73627275e-01 -6.35460973e-01 -2.11927101e-01 2.77174134e-02 8.56070220e-01 6.07846498e-01 -1.11726522e+00 1.71574578e-01 3.51170182e-01 5.50594211e-01 -6.07907236e-01 -1.64038181e-01 -1.15385520e+00 5.66968203e-01 1.11442220e+00 1.37201354e-01 3.78790915e-01 4.03341502e-01 1.47973403e-01 1.89696819e-01 -9.82795209e-02 9.67405975e-01 -3.49138707e-01 2.96963751e-01 -1.03372760e-01 -6.43105865e-01 -1.50186671e-02 1.35627258e+00 -5.97223222e-01 -1.83665052e-01 -2.03780368e-01 -1.26857030e+00 2.38353923e-01 2.88100064e-01 4.75871041e-02 5.40513158e-01 -6.14066899e-01 -6.09395504e-01 5.20300865e-01 -3.15765262e-01 9.10379216e-02 6.90311015e-01 1.36301446e+00 -1.66857123e+00 4.18009430e-01 -3.17662865e-01 -1.27545917e+00 -1.09950483e+00 3.94450933e-01 8.94115627e-01 -2.54209489e-02 -1.14451098e+00 2.78859019e-01 1.76921114e-02 -5.64831972e-01 -2.12782368e-01 -4.84964311e-01 2.10250407e-01 -6.81541145e-01 1.30043760e-01 5.11366606e-01 2.99175739e-01 2.82982051e-01 -3.58707368e-01 6.83484137e-01 8.18782076e-02 -8.11464861e-02 1.02471471e+00 -2.96600401e-01 4.22593355e-02 2.09391981e-01 4.37127799e-01 2.31612414e-01 -1.14753580e+00 5.89320123e-01 -4.85866725e-01 -7.17392027e-01 -2.71020532e-01 -1.22052634e+00 -7.73548245e-01 7.80100644e-01 8.08444679e-01 -2.54628807e-02 7.99596965e-01 -2.07713366e-01 1.44814983e-01 1.39620394e-01 4.45755422e-01 -9.20762956e-01 -2.61969119e-01 -1.69618040e-01 1.12950385e+00 -7.38147438e-01 1.70829460e-01 -1.04592466e+00 -6.57129467e-01 1.45637977e+00 3.44377160e-01 -3.15453820e-02 8.74656856e-01 5.86591125e-01 6.21013939e-01 -2.56452203e-01 -7.87795708e-02 7.96551287e-01 -1.11601226e-01 5.23955584e-01 7.97413826e-01 1.89077616e-01 -6.79460406e-01 1.73119664e-01 1.26451030e-01 5.54469824e-01 7.28592992e-01 1.60666120e+00 -2.68459201e-01 -1.03272712e+00 -4.95339096e-01 6.18899524e-01 -5.37262142e-01 2.40414545e-01 2.83723146e-01 7.34265327e-01 -1.21868670e-01 6.29960239e-01 -5.47412634e-01 2.67925680e-01 7.66517818e-01 -2.65166853e-02 7.70831764e-01 -5.43680251e-01 -9.28702116e-01 3.65148872e-01 7.96935111e-02 -5.66868365e-01 2.02419043e-01 -4.67534870e-01 -1.30768955e+00 1.43069506e-01 -2.84176528e-01 5.52123301e-02 9.44092572e-01 4.79212940e-01 9.42249179e-01 1.04203355e+00 5.80803335e-01 -7.31470287e-01 -6.98300064e-01 -8.33244264e-01 -8.08244944e-01 1.84628800e-01 7.55843520e-02 -5.36553621e-01 7.48188868e-02 2.62465328e-01]
[13.891419410705566, -2.7652714252471924]
b12e69db-2e40-48aa-bfe7-8859240e3454
mutual-information-regularization-for-weakly
2306.0363
null
https://arxiv.org/abs/2306.03630v1
https://arxiv.org/pdf/2306.03630v1.pdf
Mutual Information Regularization for Weakly-supervised RGB-D Salient Object Detection
In this paper, we present a weakly-supervised RGB-D salient object detection model via scribble supervision. Specifically, as a multimodal learning task, we focus on effective multimodal representation learning via inter-modal mutual information regularization. In particular, following the principle of disentangled representation learning, we introduce a mutual information upper bound with a mutual information minimization regularizer to encourage the disentangled representation of each modality for salient object detection. Based on our multimodal representation learning framework, we introduce an asymmetric feature extractor for our multimodal data, which is proven more effective than the conventional symmetric backbone setting. We also introduce multimodal variational auto-encoder as stochastic prediction refinement techniques, which takes pseudo labels from the first training stage as supervision and generates refined prediction. Experimental results on benchmark RGB-D salient object detection datasets verify both effectiveness of our explicit multimodal disentangled representation learning method and the stochastic prediction refinement strategy, achieving comparable performance with the state-of-the-art fully supervised models. Our code and data are available at: https://github.com/baneitixiaomai/MIRV.
['Yuchao Dai', 'Jing Zhang', 'Yuxin Mao', 'Aixuan Li']
2023-06-06
null
null
null
null
['rgb-d-salient-object-detection', 'salient-object-detection-1']
['computer-vision', 'computer-vision']
[ 3.93359333e-01 5.31228542e-01 -5.45467675e-01 -2.30528980e-01 -1.08036613e+00 -3.80702138e-01 6.08205497e-01 -1.04307368e-01 -1.47990733e-01 4.76029724e-01 5.39993167e-01 1.64135583e-02 -1.02114119e-01 -3.83137345e-01 -8.35814834e-01 -1.06145859e+00 2.23092332e-01 3.83791625e-01 -1.93012193e-01 -7.66336396e-02 -3.68610471e-02 2.71199998e-02 -1.57020950e+00 3.29786777e-01 7.16543078e-01 8.46098661e-01 2.96335101e-01 5.85188568e-01 3.08539242e-01 1.03611648e+00 2.06905007e-01 -5.34964383e-01 1.87220648e-01 -2.98691034e-01 -8.44880641e-01 3.16935360e-01 2.88754851e-01 -3.57253343e-01 -5.50603688e-01 9.43067491e-01 4.36450571e-01 -4.44786586e-02 7.93552756e-01 -1.37256980e+00 -9.22571778e-01 8.15880954e-01 -9.68753755e-01 -2.46489033e-01 2.91101545e-01 -8.26556012e-02 1.67472148e+00 -1.17593443e+00 5.50502241e-01 1.15739834e+00 2.03314766e-01 7.71321177e-01 -1.61590767e+00 -5.32360911e-01 2.97132522e-01 6.46027550e-02 -1.18233299e+00 -4.03800994e-01 1.28613722e+00 -5.96131921e-01 2.94430852e-01 2.76075214e-01 3.99186522e-01 1.27071893e+00 -2.54329264e-01 1.57064354e+00 8.52228820e-01 -4.85630214e-01 -2.31130630e-01 2.37887070e-01 1.15967840e-01 1.19923329e+00 1.60879254e-01 1.16745800e-01 -8.72869134e-01 -3.60840648e-01 6.47986710e-01 4.26178604e-01 -4.43745494e-01 -1.26262760e+00 -1.60125649e+00 1.17165720e+00 5.79222381e-01 -7.73135200e-02 -2.14410439e-01 1.43607333e-01 2.61883765e-01 -1.23298116e-01 5.54085493e-01 1.20679542e-01 -2.41144419e-01 2.33558416e-01 -5.82728326e-01 -1.25275203e-03 4.21652764e-01 8.97796214e-01 9.84216988e-01 -2.59381771e-01 -1.80786371e-01 6.59024060e-01 8.63453925e-01 6.63282335e-01 8.40120688e-02 -1.02078605e+00 5.17047048e-01 8.30788374e-01 1.56055985e-03 -7.25787938e-01 -1.84551090e-01 -2.38529518e-01 -9.01748359e-01 1.89842880e-01 2.54071563e-01 -2.07821429e-02 -5.84975839e-01 1.89311147e+00 2.69611448e-01 1.42331839e-01 2.84276217e-01 1.26720870e+00 9.47241843e-01 4.13958043e-01 -4.77758907e-02 6.60872087e-02 1.34893239e+00 -8.62017035e-01 -6.40068173e-01 7.48746172e-02 5.13491809e-01 -6.26894653e-01 1.03615534e+00 9.11828130e-02 -1.03154564e+00 -2.49694243e-01 -9.52367485e-01 -4.55166489e-01 6.23450503e-02 5.00700593e-01 9.76107538e-01 1.92409098e-01 -6.11467779e-01 2.43940309e-01 -1.19385946e+00 -1.58131003e-01 5.59593081e-01 2.01342881e-01 -6.36371076e-01 8.66228938e-02 -8.20043325e-01 6.43437743e-01 1.21870756e-01 7.13296682e-02 -1.05459261e+00 -7.30227053e-01 -1.25380516e+00 -2.40033269e-01 3.46253008e-01 -1.01070821e+00 1.18330538e+00 -8.32047880e-01 -1.52033448e+00 1.20677280e+00 -3.11590582e-01 -1.08935572e-01 2.32039839e-01 -3.26091290e-01 1.56510353e-01 3.50590855e-01 3.31980921e-02 7.70262718e-01 1.07023084e+00 -1.79100275e+00 -2.91722029e-01 -4.90547895e-01 7.00203106e-02 3.58739376e-01 -2.40025789e-01 -2.65208960e-01 -2.83116072e-01 -5.03819346e-01 3.08875054e-01 -9.67467427e-01 -1.90317899e-01 2.41334662e-01 -7.08323181e-01 -2.10170656e-01 6.07586086e-01 -5.33024967e-01 7.86868453e-01 -2.27421832e+00 9.69898760e-01 1.26438633e-01 7.12826848e-01 -3.76052856e-01 -1.81904897e-01 3.40140253e-01 -1.51338533e-01 -8.45911950e-02 -6.23216271e-01 -1.03380299e+00 3.17724407e-01 1.41521007e-01 -5.25623620e-01 7.84115911e-01 5.40751159e-01 9.76984143e-01 -1.02131355e+00 -4.30896968e-01 2.06315562e-01 7.91896343e-01 -6.30753994e-01 4.28858548e-01 -1.60779968e-01 6.80970252e-01 -6.49919629e-01 8.20118904e-01 4.37823176e-01 -7.24780321e-01 1.23817831e-01 -5.87126076e-01 4.28809561e-02 2.02424243e-01 -7.47645915e-01 2.35172176e+00 -3.41278017e-01 5.09406865e-01 -1.20773981e-03 -9.92108345e-01 8.10097694e-01 1.17672101e-01 5.09695053e-01 -2.49250963e-01 9.53257978e-02 -1.25920204e-02 -4.25027579e-01 -4.71878856e-01 4.17862415e-01 -9.15600285e-02 -1.15852669e-01 5.50895751e-01 3.28308791e-01 -2.31063142e-02 -2.43359774e-01 6.78170919e-01 5.95682979e-01 5.67421138e-01 6.08433560e-02 -7.68583491e-02 4.91161197e-01 -3.68251503e-01 4.67214227e-01 4.83971417e-01 -1.03423588e-01 9.16854560e-01 6.92216933e-01 -3.10407765e-02 -8.65752101e-01 -1.39104676e+00 -7.29145110e-02 1.23301792e+00 4.88525540e-01 -4.05745327e-01 -2.44592771e-01 -7.32072055e-01 5.98226190e-02 4.94204760e-01 -8.56269479e-01 -2.00474292e-01 -1.42815471e-01 -6.64872229e-01 2.30556905e-01 3.44504148e-01 1.34376884e-01 -5.96926868e-01 -5.09340644e-01 -4.44291890e-01 -3.05638582e-01 -9.88010645e-01 -4.05224860e-01 1.64821759e-01 -7.88186371e-01 -1.07543635e+00 -8.33021104e-01 -7.38587618e-01 7.67138124e-01 5.42301536e-01 9.29833829e-01 -1.41363367e-01 -2.44900942e-01 9.38842773e-01 -3.90121460e-01 -3.02763246e-02 -7.75921196e-02 8.63764808e-03 -6.10578582e-02 1.78740934e-01 2.25505546e-01 -4.93725479e-01 -7.20426500e-01 7.31785893e-02 -7.58727074e-01 6.51820660e-01 7.76316285e-01 1.03079081e+00 6.90415382e-01 -8.72812390e-01 1.20956615e-01 -6.08981490e-01 1.00284763e-01 -6.14921927e-01 -5.28075218e-01 2.82827705e-01 -1.71226114e-01 4.36270505e-01 -9.32323635e-02 -3.25670749e-01 -1.18603218e+00 4.27089512e-01 3.26893687e-01 -7.25178301e-01 -7.31983408e-02 4.42767859e-01 -2.28219405e-01 4.03443053e-02 3.29469860e-01 2.29999423e-01 3.73884767e-01 -4.18134123e-01 8.14343035e-01 4.45311099e-01 3.53423864e-01 -7.25781322e-01 9.78887618e-01 7.94179142e-01 -1.20599285e-01 -6.74380183e-01 -1.08136344e+00 -4.84084189e-01 -7.26356864e-01 -1.69463411e-01 9.47435379e-01 -1.38701296e+00 -9.61008847e-01 7.66472220e-02 -1.21022499e+00 -1.47126719e-01 -2.24703908e-01 6.13452137e-01 -7.37833142e-01 4.80283290e-01 -6.18190348e-01 -8.80671024e-01 -1.90449789e-01 -1.08849955e+00 1.77198243e+00 5.96038699e-02 3.29027474e-02 -9.04017568e-01 2.97875732e-01 7.63204873e-01 -2.02589229e-01 4.08603549e-01 7.16390610e-01 -2.65237242e-01 -8.80913019e-01 7.12722093e-02 -4.23276991e-01 9.61272642e-02 4.62569520e-02 -1.85952246e-01 -1.15933990e+00 -2.14732960e-01 -2.91500956e-01 -6.88498974e-01 1.37047911e+00 1.16296083e-01 9.17033672e-01 -1.61338165e-01 -2.98240185e-01 6.69642389e-01 1.12530494e+00 -7.41865635e-01 2.27977097e-01 -1.05072157e-02 1.04456341e+00 8.04507256e-01 4.38065261e-01 6.41947031e-01 7.42450953e-01 4.80199456e-01 7.89050400e-01 -1.31178677e-01 8.85006413e-02 -6.55809939e-01 4.57436591e-01 8.16095948e-01 -2.64359176e-01 7.16464221e-02 -7.07785964e-01 4.26067382e-01 -2.18044400e+00 -1.00446081e+00 1.10527482e-02 2.03520441e+00 8.06036055e-01 -3.50962490e-01 2.76845306e-01 -1.20175086e-01 4.83550161e-01 3.49297076e-01 -5.16150236e-01 3.47649336e-01 -3.11945140e-01 -3.25655401e-01 2.75956064e-01 7.32925236e-01 -1.24147594e+00 7.44916618e-01 4.55408955e+00 4.49000835e-01 -6.97349787e-01 2.55314946e-01 3.64027798e-01 -3.19372982e-01 -7.97245920e-01 2.44486239e-02 -5.52978218e-01 9.31718275e-02 3.37624371e-01 1.51535407e-01 3.56854767e-01 5.69594383e-01 -9.96347703e-03 2.10387744e-02 -1.22353292e+00 1.24299085e+00 2.93210357e-01 -1.33249402e+00 1.44754156e-01 3.91335450e-02 6.35453463e-01 1.67108685e-01 4.08927143e-01 2.84458082e-02 3.45170438e-01 -7.44432807e-01 8.11671436e-01 7.92518020e-01 3.89171004e-01 -6.81854665e-01 4.76724088e-01 2.36906171e-01 -1.08187306e+00 1.43069163e-01 -1.85067400e-01 8.66828859e-02 9.93277505e-02 4.19214696e-01 -2.70544797e-01 6.39494956e-01 5.58100283e-01 1.28149199e+00 -5.27277648e-01 5.83721757e-01 -5.27967930e-01 3.88045549e-01 -1.14402711e-01 -8.24755728e-02 -5.72199561e-02 -3.30027521e-01 8.80768716e-01 9.50069964e-01 -7.62151703e-02 1.90508232e-01 2.05353685e-02 1.23909116e+00 -1.43631786e-01 -5.35310581e-02 -6.59395039e-01 -6.91538006e-02 5.15012816e-02 1.35527360e+00 -2.02090517e-01 1.35652095e-01 -6.00726008e-01 1.16782594e+00 6.43887699e-01 6.87288940e-01 -7.80176640e-01 1.09380156e-01 7.93189228e-01 -3.84693593e-01 3.78086060e-01 -1.30265206e-01 -2.01930299e-01 -1.73067725e+00 -5.34980232e-03 -5.07138014e-01 3.61580133e-01 -7.36290693e-01 -1.60267484e+00 2.42615908e-01 4.57587093e-02 -1.40276110e+00 9.50051546e-02 -7.40736127e-01 -4.00104403e-01 6.65424228e-01 -1.60234976e+00 -1.81432080e+00 -1.36125326e-01 6.56130493e-01 1.36571378e-01 -1.17991216e-01 8.12972963e-01 -1.70291625e-02 -6.62941277e-01 5.17263353e-01 -2.17638966e-02 1.07951634e-01 5.45730352e-01 -1.33000684e+00 -4.56542552e-01 5.50857067e-01 3.32080662e-01 7.08666027e-01 6.27261400e-01 -2.34197766e-01 -1.87806356e+00 -8.69194686e-01 4.28335369e-01 -6.76089764e-01 1.08005726e+00 -6.78763568e-01 -7.53876925e-01 8.85529578e-01 5.10954201e-01 7.37259760e-02 1.20312417e+00 2.48454213e-01 -6.26511872e-01 1.66757762e-01 -6.97288096e-01 5.86971819e-01 1.04616106e+00 -8.78563404e-01 -7.08226204e-01 4.15938526e-01 1.02628291e+00 -3.47002000e-01 -9.17333901e-01 6.08513057e-01 5.81121504e-01 -8.77512991e-01 1.05435681e+00 -6.06450379e-01 7.44292378e-01 -3.15520644e-01 -5.25778532e-01 -9.25434649e-01 -2.08270460e-01 -6.99308276e-01 -6.41829491e-01 1.13101900e+00 6.36240304e-01 -3.95465493e-01 7.58266389e-01 5.42695463e-01 5.21645769e-02 -9.93075073e-01 -8.38384032e-01 -1.15939125e-01 -9.73367393e-02 -4.15266007e-01 1.05361700e-01 8.87175500e-01 4.13549155e-01 5.46989143e-01 -7.04153121e-01 4.53280181e-01 1.26545906e+00 6.94481850e-01 7.59178519e-01 -1.02182627e+00 -6.33791625e-01 -3.87467444e-01 -3.21970463e-01 -1.34650600e+00 4.85085547e-01 -1.12509632e+00 -1.03703983e-01 -1.21351373e+00 8.62235725e-01 3.07338554e-02 -4.11421865e-01 5.54017365e-01 -2.52260774e-01 2.38947824e-01 4.35113370e-01 3.22810829e-01 -9.58584487e-01 1.35658598e+00 1.30261993e+00 -4.08624500e-01 -4.00879569e-02 -5.04882969e-02 -8.83264244e-01 7.27767467e-01 5.94963729e-01 -2.95639366e-01 -3.40033323e-01 -5.49329638e-01 3.78736794e-01 5.38186580e-02 9.69848931e-01 -2.61072248e-01 9.35597122e-02 1.00982614e-01 1.80754825e-01 -6.47520900e-01 5.91204524e-01 -8.23322177e-01 -2.81756938e-01 1.99191689e-01 -6.75982237e-01 -5.57831466e-01 5.97474873e-02 8.12693119e-01 -8.13176855e-02 1.01345137e-01 5.70408046e-01 3.47422421e-01 -5.07328033e-01 4.85319406e-01 -6.15457296e-02 -1.35481715e-01 8.54145467e-01 1.90330714e-01 -3.46175671e-01 -2.97758669e-01 -8.99153411e-01 4.71670806e-01 5.12815952e-01 4.66257900e-01 1.02418423e+00 -1.48050749e+00 -7.82818675e-01 1.30736515e-01 6.08094692e-01 -2.69170292e-02 2.34983370e-01 1.04552996e+00 1.09413780e-01 8.72181430e-02 1.02878958e-01 -8.63815844e-01 -1.11962175e+00 3.57969910e-01 2.35318795e-01 1.77702755e-02 -5.24100304e-01 9.17882442e-01 4.98012930e-01 -7.49098539e-01 2.37776205e-01 -1.00938112e-01 -2.12100893e-01 1.18242269e-02 3.18032473e-01 1.43874481e-01 -6.27551258e-01 -8.44842017e-01 -3.14247638e-01 5.98090529e-01 -2.08039172e-02 -1.95521519e-01 1.39812279e+00 -3.94693792e-01 -2.08444521e-01 7.02509522e-01 1.37735045e+00 -2.03221571e-02 -1.62989247e+00 -5.41898370e-01 -2.21295863e-01 -3.57661903e-01 9.29135904e-02 -2.85793453e-01 -9.86451387e-01 9.95996833e-01 6.35216773e-01 -8.75504687e-02 9.41632092e-01 7.32779622e-01 3.65118116e-01 3.53521407e-01 2.13135555e-01 -3.59876662e-01 2.70460129e-01 2.41360232e-01 1.11755168e+00 -1.82674778e+00 -1.87952202e-02 -4.51939374e-01 -9.90527034e-01 8.56442094e-01 4.82880384e-01 -1.85521737e-01 6.66155040e-01 -1.27319649e-01 -2.17676073e-01 -2.96282381e-01 -7.51125932e-01 -4.45967674e-01 6.41419530e-01 4.21162784e-01 5.46942830e-01 1.80209666e-01 1.94788694e-01 7.51286268e-01 1.59674451e-01 -4.76467252e-01 1.95055470e-01 6.69948041e-01 -2.46949583e-01 -8.11702669e-01 -1.84899390e-01 1.75255373e-01 -9.70882326e-02 -1.29041910e-01 -4.31578100e-01 6.41792595e-01 -1.23659767e-01 8.23946357e-01 -1.71521232e-01 -2.75793076e-01 7.53817111e-02 -1.22406207e-01 7.05335975e-01 -5.73930979e-01 9.40331668e-02 1.94746554e-01 -2.50740647e-01 -6.34650290e-01 -1.00429690e+00 -1.00952160e+00 -1.30177283e+00 9.02567953e-02 -4.36262190e-01 8.43195841e-02 3.75604928e-01 9.08304870e-01 4.08760518e-01 2.46014729e-01 6.59648001e-01 -1.30231225e+00 -3.96910876e-01 -7.57009625e-01 -4.37318474e-01 4.65369582e-01 8.15774202e-01 -9.18412268e-01 -5.22305906e-01 2.06364065e-01]
[10.75829029083252, 1.1855634450912476]
7edfa1e7-bab6-4ee6-8a5c-3cd143eaf024
alleviating-the-sample-selection-bias-in-few
2210.16834
null
https://arxiv.org/abs/2210.16834v1
https://arxiv.org/pdf/2210.16834v1.pdf
Alleviating the Sample Selection Bias in Few-shot Learning by Removing Projection to the Centroid
Few-shot learning (FSL) targets at generalization of vision models towards unseen tasks without sufficient annotations. Despite the emergence of a number of few-shot learning methods, the sample selection bias problem, i.e., the sensitivity to the limited amount of support data, has not been well understood. In this paper, we find that this problem usually occurs when the positions of support samples are in the vicinity of task centroid -- the mean of all class centroids in the task. This motivates us to propose an extremely simple feature transformation to alleviate this problem, dubbed Task Centroid Projection Removing (TCPR). TCPR is applied directly to all image features in a given task, aiming at removing the dimension of features along the direction of the task centroid. While the exact task centroid cannot be accurately obtained from limited data, we estimate it using base features that are each similar to one of the support features. Our method effectively prevents features from being too close to the task centroid. Extensive experiments over ten datasets from different domains show that TCPR can reliably improve classification accuracy across various feature extractors, training algorithms and datasets. The code has been made available at https://github.com/KikimorMay/FSL-TCBR.
['Zenglin Xu', 'Yanan Li', 'Wenjie Pei', 'Xinglin Pan', 'Xu Luo', 'Jing Xu']
2022-10-30
null
null
null
null
['selection-bias']
['natural-language-processing']
[ 3.72085333e-01 -2.59777904e-01 -1.33504361e-01 -3.11756581e-01 -6.21629477e-01 -4.88813490e-01 7.30105281e-01 -1.41741022e-01 -5.28938949e-01 6.49098337e-01 1.53554454e-01 1.05863787e-01 -5.30548453e-01 -2.74033725e-01 -4.50707495e-01 -9.23786581e-01 3.42626661e-01 2.40961641e-01 5.13642907e-01 -1.13304360e-02 4.65055585e-01 3.32691073e-01 -1.79072726e+00 2.53855903e-02 7.48637736e-01 9.92897689e-01 3.77753943e-01 4.92852151e-01 1.85469314e-02 4.59209740e-01 -7.27881134e-01 -2.17796024e-02 4.50673670e-01 -2.68810034e-01 -5.18892109e-01 1.77878380e-01 4.83126700e-01 -2.14153230e-01 -2.95058340e-01 1.10997343e+00 5.24465859e-01 4.54129428e-01 8.72485876e-01 -1.44425571e+00 -6.28427744e-01 -8.64925012e-02 -6.05006695e-01 4.08365399e-01 5.48818782e-02 9.34123099e-02 8.68817985e-01 -1.22821546e+00 7.39183009e-01 6.81102872e-01 6.65486991e-01 6.11288965e-01 -1.03608263e+00 -3.96536261e-01 -7.14803189e-02 4.12218869e-01 -1.45445943e+00 -7.43353963e-01 8.84153962e-01 -6.68467164e-01 7.20178604e-01 1.79733902e-01 4.05432820e-01 1.26479077e+00 1.34165153e-01 5.55437505e-01 1.03517401e+00 -5.69495797e-01 5.20413280e-01 2.94836909e-01 5.93649685e-01 4.06535059e-01 3.38669300e-01 -2.16114260e-02 -7.03157306e-01 -3.23907435e-01 4.46651757e-01 2.23293722e-01 -4.59812135e-01 -8.17440271e-01 -1.14714992e+00 8.89814496e-01 2.65194595e-01 3.51188898e-01 -1.74133003e-01 -1.86574742e-01 3.78503352e-01 1.33033574e-01 4.19923663e-01 3.72885793e-01 -3.33746195e-01 -1.93547592e-01 -6.85512602e-01 1.85644999e-01 5.94411314e-01 1.14150345e+00 7.69351304e-01 -7.78594539e-02 -3.37760895e-01 9.89604890e-01 -2.93891221e-01 2.60835052e-01 7.59478152e-01 -8.42788517e-01 2.07813159e-01 5.24891198e-01 2.59361655e-01 -8.05640459e-01 -3.66587430e-01 -2.97385544e-01 -5.90995371e-01 1.92462310e-01 7.50524759e-01 -3.24319512e-01 -9.44788694e-01 1.49205434e+00 4.30609643e-01 2.70724416e-01 -1.72827896e-02 1.15360510e+00 7.15132892e-01 3.00661147e-01 -2.54095018e-01 -2.88803428e-01 1.36231816e+00 -6.67745054e-01 -5.46304703e-01 -4.37402695e-01 6.88472211e-01 -6.12462521e-01 1.23332310e+00 2.08635882e-01 -3.55399162e-01 -5.13755262e-01 -1.02247334e+00 7.26193264e-02 -5.04550457e-01 7.74693415e-02 6.01166368e-01 5.33179402e-01 -6.64391696e-01 4.13403988e-01 -5.22910774e-01 -5.72070539e-01 4.87532079e-01 4.65316176e-02 -5.19691765e-01 -2.58336097e-01 -9.42354381e-01 8.27107310e-01 2.59141147e-01 -6.59914911e-02 -5.35322607e-01 -5.97728133e-01 -8.16105723e-01 2.93430015e-02 8.24718654e-01 -4.62973952e-01 1.18536448e+00 -8.49657238e-01 -9.34005499e-01 5.04797161e-01 -4.51722145e-01 -3.77364576e-01 3.63313168e-01 -3.46183956e-01 -2.04136014e-01 -1.34479487e-02 2.90798932e-01 4.56461221e-01 1.31204259e+00 -1.02068472e+00 -5.47556579e-01 -5.07935286e-01 -2.98617244e-01 1.92417890e-01 -5.26342332e-01 3.26638333e-02 -2.34790638e-01 -5.00063419e-01 1.65737301e-01 -9.05901849e-01 -6.59328997e-02 5.32652177e-02 -3.38198006e-01 -4.66123164e-01 9.65756655e-01 -1.28611967e-01 9.52960074e-01 -2.35394883e+00 -1.57138899e-01 -3.35635617e-02 2.65937686e-01 3.86703253e-01 1.35569528e-01 3.76804799e-01 -2.14978419e-02 -2.29707599e-01 -2.78671861e-01 -1.09186791e-01 -1.43180460e-01 1.45148098e-01 -5.20038724e-01 6.77733958e-01 1.30114898e-01 8.02857697e-01 -1.06792867e+00 -4.47641701e-01 4.32075322e-01 3.05799067e-01 -1.29136503e-01 -1.36899456e-01 5.52333519e-02 1.31976143e-01 -5.08847594e-01 5.26475310e-01 5.10865986e-01 -1.67022049e-01 -1.93097413e-01 -1.10892214e-01 -1.45468369e-01 -1.10899851e-01 -1.21549356e+00 1.41196001e+00 -4.44375444e-03 8.70107472e-01 -4.50693399e-01 -9.44799960e-01 1.02828979e+00 2.19192103e-01 5.55676997e-01 -4.62609768e-01 7.48700872e-02 7.05074593e-02 1.34093285e-01 -6.20209634e-01 4.35722232e-01 -9.09620002e-02 -6.10178784e-02 3.39594990e-01 2.58177459e-01 6.10690862e-02 2.68178552e-01 3.78633440e-02 1.07400298e+00 -1.01825401e-01 6.51663959e-01 -1.31460011e-01 1.46576837e-01 1.43383637e-01 7.19488978e-01 9.83235061e-01 -8.02827001e-01 8.82198751e-01 3.57568890e-01 -4.46764529e-01 -9.82304931e-01 -8.47124279e-01 -3.30430567e-01 1.24492073e+00 7.90800825e-02 -4.58416581e-01 -6.82217956e-01 -7.70416260e-01 1.48360118e-01 8.27964723e-01 -8.11557531e-01 -4.97497350e-01 -2.17425808e-01 -5.91439486e-01 1.75783351e-01 3.59867573e-01 3.28133732e-01 -9.03164566e-01 -9.45020378e-01 2.78960317e-02 -3.34169939e-02 -1.08605552e+00 -5.32168031e-01 3.69393319e-01 -6.82971299e-01 -1.31114662e+00 -7.02618420e-01 -6.22718692e-01 6.71575546e-01 8.29573452e-01 5.24833798e-01 -2.88107902e-01 -6.30417347e-01 4.08915877e-01 -4.78977084e-01 -7.19721675e-01 2.47100994e-01 -9.84431803e-02 2.28051260e-01 2.17843801e-01 8.40030849e-01 -3.03796470e-01 -5.46673000e-01 5.39378583e-01 -4.67934936e-01 -2.32284397e-01 2.28846088e-01 9.63601768e-01 4.96701539e-01 4.02363613e-02 7.07296789e-01 -8.53556335e-01 6.06487930e-01 -3.78489852e-01 -2.63303608e-01 1.68941379e-01 -4.11797374e-01 -1.39021933e-01 6.09188855e-01 -6.00083530e-01 -8.25162888e-01 2.52511591e-01 4.47049022e-01 -7.99731791e-01 -5.00632823e-01 1.36852756e-01 6.15474470e-02 -2.32613944e-02 9.85830963e-01 4.17704582e-01 3.52734029e-02 -3.93868655e-01 1.62218288e-01 8.82804811e-01 4.25199866e-01 -2.07634270e-01 7.00754523e-01 6.30731285e-01 -6.88965768e-02 -1.23581100e+00 -1.03565502e+00 -8.77359569e-01 -8.10824275e-01 -3.32606047e-01 5.88946640e-01 -5.73193192e-01 -2.78177768e-01 3.71148080e-01 -7.99424827e-01 -1.09828308e-01 -4.51038390e-01 6.59561336e-01 -6.17274284e-01 3.16427052e-01 -1.66212227e-02 -8.30583692e-01 -3.12974244e-01 -8.35286975e-01 8.87375951e-01 2.86698490e-01 -4.49201584e-01 -7.01761186e-01 -1.71997044e-02 1.88402697e-01 2.85864890e-01 6.64464831e-02 7.24276960e-01 -9.68690634e-01 -1.69093505e-01 -5.40864348e-01 -1.17892556e-01 2.90948421e-01 3.87447417e-01 1.88600123e-02 -1.28210890e+00 -3.45148951e-01 5.16667187e-01 -2.78132766e-01 9.84936535e-01 6.32796586e-01 1.05030644e+00 1.97540503e-02 -4.56054658e-01 4.90095228e-01 1.10789871e+00 8.56303498e-02 3.52055192e-01 2.57184714e-01 6.50977612e-01 5.40103793e-01 9.54386175e-01 6.33788466e-01 -5.21899350e-02 6.72117412e-01 2.55807072e-01 3.95964116e-01 -1.51628584e-01 1.34953901e-01 1.22241810e-01 2.42284402e-01 -8.10578689e-02 1.20231926e-01 -1.00972283e+00 6.12036228e-01 -1.96511352e+00 -9.09367263e-01 -9.96300206e-02 2.35712147e+00 4.74037856e-01 2.98579484e-01 4.29646596e-02 1.62983924e-01 8.36283088e-01 2.11410955e-01 -8.24605227e-01 -1.02025606e-01 1.38438717e-01 -2.20421359e-01 4.47596192e-01 2.08256736e-01 -1.21834540e+00 8.47169161e-01 5.51899338e+00 8.86566043e-01 -9.55528498e-01 6.96596131e-02 1.89866394e-01 -3.84340763e-01 3.48250449e-01 9.76087991e-03 -1.10548258e+00 5.54042459e-01 5.59319615e-01 -5.07930934e-01 3.85204047e-01 1.09649682e+00 1.77403077e-01 -3.17919791e-01 -1.13459444e+00 1.07290983e+00 4.14041609e-01 -9.69050527e-01 -1.36670932e-01 -5.35622388e-02 5.54536521e-01 4.51086387e-02 8.81611109e-02 3.63456279e-01 -1.50230929e-01 -6.91560745e-01 5.33124506e-01 5.24671555e-01 7.32434869e-01 -6.33848071e-01 5.82738757e-01 6.01922631e-01 -9.52663898e-01 -4.19507444e-01 -8.04070234e-01 -1.64511636e-01 -2.24729612e-01 7.69171476e-01 -1.17496538e+00 2.87492394e-01 6.48005903e-01 7.87414551e-01 -7.30040014e-01 1.31656063e+00 3.51014845e-02 3.76319587e-01 -1.20018780e-01 -7.26880357e-02 9.84561443e-02 -4.39419076e-02 8.34049940e-01 8.24112475e-01 3.04379165e-01 -1.15861781e-01 8.29835609e-02 5.38503647e-01 2.16980219e-01 -5.93052804e-02 -1.08120072e+00 1.87344119e-01 7.14043617e-01 1.33811402e+00 -7.02418566e-01 -2.12006390e-01 -5.32732308e-01 7.43195474e-01 3.63271981e-01 5.07318497e-01 -6.43105805e-01 -7.13716626e-01 6.71051383e-01 7.91138113e-02 4.32498306e-01 -2.13652790e-01 -3.05957258e-01 -9.73336041e-01 2.68493265e-01 -5.32632411e-01 3.86978567e-01 -7.17554927e-01 -1.41763031e+00 3.79442573e-01 4.80297506e-02 -1.52499855e+00 -2.14701638e-01 -6.36334956e-01 -7.89085984e-01 7.15923429e-01 -1.18842494e+00 -7.23812819e-01 -4.00175571e-01 7.34374940e-01 8.01746666e-01 -3.56387168e-01 7.56023467e-01 -1.65162802e-01 -5.52279115e-01 4.06546950e-01 3.37036043e-01 1.19447179e-01 1.08395791e+00 -1.02423954e+00 2.28769317e-01 7.28697538e-01 1.31474271e-01 7.32418835e-01 7.69560933e-01 -6.30611658e-01 -1.32273066e+00 -1.00833821e+00 7.60451674e-01 -5.89453459e-01 7.24894881e-01 -4.50140625e-01 -1.01822531e+00 6.11145675e-01 -2.70311445e-01 4.30365592e-01 6.05693638e-01 1.32715508e-01 -3.32028300e-01 -1.62142411e-01 -9.29868579e-01 5.87302983e-01 1.05170059e+00 -6.09131038e-01 -7.89539218e-01 3.80709350e-01 3.90283138e-01 -4.43305597e-02 -4.48118031e-01 3.19264293e-01 3.64875823e-01 -8.69762719e-01 8.50525141e-01 -5.44018507e-01 5.71861751e-02 -2.33162537e-01 -2.54807591e-01 -1.32462656e+00 -4.24303234e-01 -2.03101367e-01 -2.83941239e-01 1.04731858e+00 3.02039608e-02 -6.61708474e-01 6.86515152e-01 6.23023391e-01 -2.01384813e-01 -7.43061781e-01 -1.26809740e+00 -1.02159882e+00 -3.38337690e-01 -3.55957568e-01 1.33472070e-01 1.03837776e+00 3.19228470e-02 4.18554962e-01 -3.21043015e-01 -1.21964775e-01 8.39484513e-01 1.63976684e-01 7.58099854e-01 -1.55056787e+00 -2.73345020e-02 -1.67479917e-01 -3.93286675e-01 -5.62326074e-01 7.93168172e-02 -7.81564832e-01 2.20764220e-01 -1.44707918e+00 2.88476050e-01 -3.06476653e-01 -2.33997300e-01 6.07612610e-01 -2.03609705e-01 -1.61860976e-03 1.30840465e-01 4.33670223e-01 -5.78243792e-01 5.22500455e-01 1.25089657e+00 3.45162414e-02 -1.82419643e-01 1.85447440e-01 -6.23403251e-01 1.02262807e+00 1.11930764e+00 -6.19588792e-01 -6.00490689e-01 -1.14612028e-01 -2.87591040e-01 -3.22332263e-01 4.60585624e-01 -1.27246284e+00 3.91812861e-01 -2.14648321e-01 6.53613150e-01 -4.51584399e-01 5.66238344e-01 -8.11780870e-01 -1.76796481e-01 2.89151281e-01 -1.62703097e-01 -4.84890580e-01 -3.57592069e-02 8.23295414e-01 1.55980572e-01 -6.97864830e-01 6.16086602e-01 -6.84392154e-02 -1.05179131e+00 1.29761308e-01 -2.59509921e-01 1.08342044e-01 1.30038369e+00 -6.11398458e-01 -5.88582993e-01 -5.00207730e-02 -5.93232155e-01 1.16064824e-01 4.97044951e-01 5.85975349e-01 7.83225715e-01 -1.19081104e+00 -4.06085789e-01 4.08242643e-01 5.24749279e-01 -1.69159859e-01 2.68044382e-01 9.82611775e-01 2.14083403e-01 5.08723617e-01 -2.16374204e-01 -6.45254433e-01 -1.45794833e+00 7.31739759e-01 1.86087623e-01 2.84376383e-01 -9.62988496e-01 6.24628186e-01 2.47158214e-01 -1.28062025e-01 2.94683814e-01 -2.55639702e-02 -3.41636986e-01 3.64200175e-01 7.82503426e-01 4.99233425e-01 8.48119333e-02 -4.27535146e-01 -3.95695150e-01 5.57173431e-01 -4.45715040e-01 1.83580279e-01 1.25770509e+00 -7.23314434e-02 3.19391221e-01 8.41019094e-01 1.05676961e+00 -3.58404130e-01 -1.59033573e+00 -4.62331027e-01 1.87307075e-01 -8.41335297e-01 -1.56455953e-02 -4.86834049e-01 -6.59955561e-01 7.77595043e-01 6.71131432e-01 2.53759056e-01 7.57498682e-01 8.55119899e-02 4.53981400e-01 5.91867030e-01 5.37083268e-01 -1.35068333e+00 8.82399827e-02 6.80350959e-01 7.32145786e-01 -1.45082128e+00 -7.62064978e-02 -3.35906953e-01 -8.87782097e-01 9.46866870e-01 6.76042080e-01 -2.14957863e-01 6.20758414e-01 3.78073789e-02 -1.85943067e-01 -1.83729157e-01 -7.20651329e-01 -3.47351193e-01 2.73629695e-01 8.14226687e-01 1.31811589e-01 -4.34476836e-03 -1.15601875e-01 6.02869034e-01 -9.09681339e-03 2.00195387e-01 5.95990598e-01 1.20406711e+00 -8.73649955e-01 -5.67330241e-01 -4.62006211e-01 8.88257384e-01 -1.43104613e-01 2.79641803e-03 -3.86229873e-01 5.92124224e-01 4.13874611e-02 9.28227782e-01 1.31762996e-01 -3.75117987e-01 3.20296943e-01 4.59256321e-01 3.27643186e-01 -7.93209612e-01 1.45904748e-02 -1.23127662e-01 -3.39290425e-02 -4.03211415e-01 -9.51814875e-02 -7.25000143e-01 -1.09789944e+00 -9.84138623e-02 -4.14859593e-01 -8.54916126e-03 4.27579522e-01 8.27253044e-01 3.61686468e-01 3.18358898e-01 6.00500226e-01 -9.37007964e-01 -8.53157938e-01 -1.05081761e+00 -8.63707066e-01 4.21981484e-01 4.01517093e-01 -1.11958122e+00 -5.70437312e-01 1.29571959e-01]
[9.924725532531738, 2.9500880241394043]
054f5caa-4fb2-432d-a59b-025935581b04
semantic-source-code-search-a-study-of-the
1908.06738
null
https://arxiv.org/abs/1908.06738v2
https://arxiv.org/pdf/1908.06738v2.pdf
Semantic Source Code Search: A Study of the Past and a Glimpse at the Future
With the recent explosion in the size and complexity of source codebases and software projects, the need for efficient source code search engines has increased dramatically. Unfortunately, existing information retrieval-based methods fail to capture the query semantics and perform well only when the query contains syntax-based keywords. Consequently, such methods will perform poorly when given high-level natural language queries. In this paper, we review existing methods for building code search engines. We also outline the open research directions and the various obstacles that stand in the way of having a universal source code search engine.
['Muhammad Khalifa']
2019-08-15
null
null
null
null
['code-search', 'code-search']
['computer-code', 'computer-vision']
[-3.13547671e-01 -3.52255136e-01 -5.57965457e-01 -2.17508271e-01 -9.77268934e-01 -8.25903237e-01 3.51672769e-01 5.64390779e-01 -1.26209125e-01 2.57492840e-01 1.20119013e-01 -6.34392321e-01 -3.58318716e-01 -6.31196260e-01 -1.16547585e-01 2.67468184e-01 -2.18282882e-02 6.29534274e-02 7.54017234e-01 -3.95205855e-01 7.61544347e-01 1.34091705e-01 -1.88907945e+00 1.53166175e-01 1.11268842e+00 4.46951032e-01 5.23506820e-01 6.45829737e-01 -9.02310669e-01 8.26036394e-01 -7.02929497e-01 -6.31485224e-01 -7.26984590e-02 -4.18090850e-01 -1.05270207e+00 -6.78485096e-01 2.24430203e-01 -1.03108108e-01 -3.79353404e-01 1.51115263e+00 2.89713502e-01 -2.15742171e-01 2.34847203e-01 -1.46655595e+00 -7.78272688e-01 5.81206739e-01 -2.46349767e-01 3.10825527e-01 1.05416512e+00 -4.57643688e-01 1.11551034e+00 -8.89856517e-01 1.06922603e+00 8.08322191e-01 5.93405604e-01 4.24367338e-01 -8.33993256e-01 -2.50361323e-01 -1.78046122e-01 2.29218245e-01 -1.56092036e+00 -3.26067269e-01 6.66551232e-01 -6.68506384e-01 1.50880206e+00 4.29328173e-01 3.54101092e-01 2.84414858e-01 2.55305320e-01 5.04229009e-01 2.91205704e-01 -7.21083522e-01 -4.86382954e-02 1.89261347e-01 3.63954872e-01 8.72471213e-01 5.43263972e-01 -1.94096312e-01 -1.69548765e-01 -9.64632928e-01 3.71935785e-01 6.99706152e-02 -9.70269889e-02 -5.59603155e-01 -8.78455460e-01 7.59552121e-01 8.97857696e-02 7.55674779e-01 6.23213910e-02 2.13864237e-01 6.84048116e-01 6.39707983e-01 -7.96066150e-02 7.10267425e-01 -5.35526216e-01 -5.32199621e-01 -1.10220420e+00 5.04409671e-01 1.30298150e+00 1.52172279e+00 9.96703863e-01 -3.50393385e-01 2.32533738e-01 1.00209081e+00 5.49903452e-01 3.19217682e-01 4.29251432e-01 -9.43533838e-01 4.68264788e-01 1.16260672e+00 1.97273284e-01 -1.28435886e+00 -1.42543605e-02 -1.58518106e-01 2.37137660e-01 -1.50117697e-02 9.73975584e-02 4.13514644e-01 -4.54137683e-01 1.16595376e+00 7.90468231e-02 -8.40075314e-01 -7.75257766e-04 3.53077531e-01 8.04012537e-01 4.40354973e-01 -1.45466521e-01 -1.93042159e-01 1.18788791e+00 -7.70141065e-01 -6.94560111e-01 -3.85058016e-01 8.42000186e-01 -1.25284648e+00 1.09629047e+00 -5.47143519e-02 -9.85130727e-01 -1.39718771e-01 -9.64703202e-01 -1.92190573e-01 -5.94740808e-01 -1.92933336e-01 6.66798234e-01 8.65284145e-01 -1.15799403e+00 8.47117454e-02 -6.98628366e-01 -6.36132479e-01 -1.36408716e-01 -2.53410498e-03 -8.29420576e-04 -4.36157048e-01 -8.84644508e-01 9.18859780e-01 4.48770612e-01 -3.79745871e-01 -6.59666836e-01 -6.85753107e-01 -6.82545841e-01 2.69289970e-01 4.26292896e-01 -3.42538178e-01 1.47847760e+00 -7.74830580e-01 -6.38481021e-01 8.01064789e-01 -2.29741052e-01 2.22737327e-01 -1.27379864e-01 -5.14347255e-02 -6.28702223e-01 6.00699633e-02 4.34087634e-01 9.88264233e-02 1.78240925e-01 -1.12841773e+00 -7.43917644e-01 -1.85469404e-01 4.84707087e-01 -4.25084345e-02 -5.13028324e-01 9.15107191e-01 -8.33350182e-01 -5.14767289e-01 -8.15291144e-03 -6.43904150e-01 -2.84259766e-01 1.20138183e-01 1.48560807e-01 -5.81687152e-01 7.11584568e-01 -3.93577248e-01 2.04248023e+00 -2.23952723e+00 -2.03738939e-02 1.57025188e-01 2.48807177e-01 3.49544026e-02 -1.57793894e-01 1.15648842e+00 7.48030022e-02 5.25291681e-01 -4.82514035e-03 5.68267822e-01 1.80767521e-01 8.72960165e-02 -3.16209137e-01 -1.70576870e-02 -2.96191603e-01 7.02238142e-01 -1.16029227e+00 -9.49569106e-01 -3.00903797e-01 1.17154859e-01 -7.95732856e-01 1.53285012e-01 -4.50745702e-01 -5.49148321e-01 -1.04440069e+00 1.02850676e+00 1.18432105e-01 -4.33743596e-01 1.86268345e-01 4.69723612e-01 -5.57567239e-01 5.07146716e-01 -9.95342135e-01 2.15004778e+00 -4.81041342e-01 5.84678411e-01 1.99910149e-01 -7.04228699e-01 8.72612655e-01 4.02158499e-01 5.42490423e-01 -6.40443444e-01 -4.51106936e-01 7.57810295e-01 -1.38213322e-01 -9.31268930e-01 5.32314837e-01 2.53642350e-01 -4.54576671e-01 5.14482558e-01 -2.78546244e-01 -4.08582777e-01 6.91288352e-01 4.31483358e-01 1.52848816e+00 5.13819642e-02 4.43194777e-01 -4.38380480e-01 6.63697004e-01 7.44116962e-01 3.24859679e-01 7.89135337e-01 -2.54000984e-02 1.70506120e-01 4.99868900e-01 -4.24697518e-01 -7.79242694e-01 -6.68153584e-01 -1.41139865e-01 1.34829140e+00 2.16893479e-01 -1.19735909e+00 -7.00867236e-01 -6.10889077e-01 -2.54457146e-02 4.00216550e-01 -1.81950014e-02 7.90770631e-03 -5.63405812e-01 -2.87604749e-01 8.20753276e-01 2.22542405e-01 5.40109128e-02 -8.44351053e-01 -7.83928394e-01 3.30643177e-01 -2.94292569e-01 -5.01629353e-01 -3.47150236e-01 -1.96544513e-01 -8.40324163e-01 -1.35737026e+00 -4.91581708e-01 -9.55876350e-01 6.17927492e-01 3.91402900e-01 1.52437603e+00 1.04742908e+00 -5.37115872e-01 6.39443099e-01 -6.90123081e-01 -2.32921347e-01 -6.53042853e-01 2.73164272e-01 -6.12576962e-01 -1.29006064e+00 5.83367407e-01 -3.24031293e-01 -3.48344088e-01 2.45118767e-01 -1.10699010e+00 -4.57294643e-01 4.08193260e-01 5.68446457e-01 -5.23204729e-02 1.46407917e-01 6.33854628e-01 -6.19944394e-01 9.37585473e-01 -7.67090917e-01 -8.60581040e-01 7.44584620e-01 -1.11568463e+00 1.68373436e-01 3.81515235e-01 7.44148865e-02 -1.06506407e+00 -2.05903240e-02 -3.33935648e-01 1.60822585e-01 4.00636308e-02 1.30141985e+00 2.65805423e-01 -2.88064986e-01 9.89300430e-01 2.81690866e-01 -3.32867295e-01 -8.14074814e-01 1.94292709e-01 8.50348413e-01 1.56375229e-01 -8.07345629e-01 8.27190042e-01 7.49693811e-02 -6.41387165e-01 -5.80177665e-01 -2.26586893e-01 -8.79241168e-01 -3.43689293e-01 -4.22310680e-02 4.10725474e-01 -5.92220366e-01 -5.45197465e-02 -2.44007111e-02 -1.40233183e+00 3.14350665e-01 8.76436383e-02 1.03152588e-01 -3.46759200e-01 7.33155072e-01 -4.57071662e-01 -5.34609914e-01 -1.29590496e-01 -1.36826503e+00 8.74788642e-01 7.88146704e-02 -3.49080443e-01 -7.84987986e-01 5.93741357e-01 6.67387918e-02 8.97571564e-01 -1.99787244e-01 1.43332195e+00 -5.27698100e-01 -6.93515837e-01 -5.28916001e-01 -2.56369472e-01 -3.42609465e-01 9.12131369e-02 2.34286934e-01 -3.16405863e-01 -1.55126378e-01 -1.98836654e-01 -1.40368238e-01 8.21699202e-02 -2.68860191e-01 7.59209633e-01 -2.89674252e-01 -6.73488200e-01 2.81610727e-01 2.00215149e+00 6.53034687e-01 4.23201621e-01 5.58441281e-01 1.70766637e-01 5.65087676e-01 5.07533371e-01 3.91152799e-01 4.20394361e-01 6.35116160e-01 2.55197585e-01 4.06078607e-01 -1.74609087e-02 -2.51525074e-01 4.68205623e-02 1.20539224e+00 2.69107759e-01 -5.65436520e-02 -1.60115266e+00 8.85409057e-01 -1.72814727e+00 -7.43818521e-01 -1.56683072e-01 2.05317450e+00 1.17643261e+00 -2.23143861e-01 -1.36788040e-01 -2.68383473e-01 4.29982632e-01 -1.81161314e-01 -4.17910904e-01 -2.58607179e-01 3.10122252e-01 -1.64516076e-01 1.13666803e-01 2.75394142e-01 -5.42710781e-01 5.85303187e-01 8.24285507e+00 5.53138614e-01 -7.60146022e-01 1.06355369e-01 -4.01980311e-01 3.78534019e-01 -7.25813031e-01 5.96132398e-01 -6.39603436e-01 2.47148573e-01 1.01069629e+00 -1.03955734e+00 4.43371475e-01 1.46379411e+00 -3.25363964e-01 -1.07004263e-01 -9.97968853e-01 9.08263743e-01 1.35064527e-01 -1.40447628e+00 -8.79652947e-02 -3.05409878e-01 5.73145390e-01 3.41527432e-01 -5.90548158e-01 4.53076422e-01 3.69917154e-01 -6.43858373e-01 6.04575157e-01 3.25733066e-01 7.07630157e-01 -3.38223517e-01 4.56631780e-01 4.90271777e-01 -1.40325177e+00 -2.03110948e-01 -4.89544183e-01 -8.88670236e-02 -2.08501786e-01 3.89623016e-01 -3.55697960e-01 6.31606400e-01 9.05764580e-01 3.63919050e-01 -8.02502871e-01 1.53425920e+00 1.55128598e-01 1.28487051e-01 -3.44143450e-01 -4.56831753e-01 -1.23530656e-01 2.11758628e-01 4.60609913e-01 1.44283223e+00 5.00116289e-01 -1.62012860e-01 4.59200501e-01 1.02916539e+00 4.21780720e-02 4.13838327e-01 -9.55702662e-01 -4.51846689e-01 7.19318509e-01 8.46972167e-01 -7.18200326e-01 -4.28229958e-01 -1.16713560e+00 4.59939897e-01 1.27915055e-01 4.36961740e-01 -4.70719516e-01 -1.25101566e+00 5.79293907e-01 5.83286174e-02 8.29318836e-02 -3.89199704e-01 1.84928998e-01 -1.38216305e+00 4.35178816e-01 -1.18168128e+00 6.11799181e-01 -8.37497771e-01 -1.01339197e+00 6.40408874e-01 3.79194289e-01 -9.95492518e-01 -5.90728641e-01 -3.32354426e-01 -3.39078724e-01 6.97640717e-01 -1.32349002e+00 -5.40178359e-01 -2.61991378e-02 3.31707865e-01 5.70886016e-01 -4.16101739e-02 9.67658043e-01 7.05928862e-01 -1.49532095e-01 2.81411558e-01 2.97037572e-01 2.18196690e-01 4.33603585e-01 -9.08202708e-01 1.96532398e-01 1.05017054e+00 7.48228133e-02 1.48665297e+00 7.61285484e-01 -7.27305114e-01 -1.94872916e+00 -3.91355693e-01 1.25416720e+00 -5.77166796e-01 1.03324425e+00 -2.49101341e-01 -9.92242992e-01 4.54674572e-01 2.55661786e-01 -6.43957183e-02 6.07599378e-01 1.75691918e-01 -7.68721998e-01 -4.54863422e-02 -9.04778719e-01 3.51416886e-01 8.75258327e-01 -1.02724218e+00 -8.30902755e-01 4.75814432e-01 6.72088504e-01 -2.60373235e-01 -1.10766447e+00 1.15812853e-01 5.55817008e-01 -7.07538545e-01 8.97181988e-01 -4.67698544e-01 2.90851265e-01 -4.48333204e-01 -1.70618743e-01 -6.26774251e-01 -5.79497218e-02 -6.40517414e-01 9.45593715e-02 1.03450680e+00 5.48112094e-01 -4.35855597e-01 5.95045805e-01 1.08136749e+00 1.29011227e-03 -4.09998119e-01 -8.14697385e-01 -9.21166539e-01 1.21842332e-01 -5.67311823e-01 6.16096914e-01 1.01386452e+00 9.32904422e-01 8.15238729e-02 3.22281867e-01 -2.46565402e-01 3.00623506e-01 5.03539264e-01 4.65237319e-01 -1.34932590e+00 -6.42478839e-02 -7.37488270e-01 -4.52529609e-01 -1.10303831e+00 -3.25910561e-02 -9.56179917e-01 1.95249021e-01 -1.56498051e+00 5.69214046e-01 -5.06934226e-01 -3.14838104e-02 5.79303861e-01 -7.85374194e-02 -2.69120097e-01 -2.58788746e-02 7.03548968e-01 -7.90051818e-01 -1.04351878e-01 4.26000714e-01 -1.11457691e-01 7.92523399e-02 -1.91164911e-01 -7.13521302e-01 6.30294561e-01 6.26036942e-01 -9.51414585e-01 -5.96482933e-01 -7.22316504e-01 1.10305619e+00 4.04236674e-01 9.23120230e-02 -7.26516724e-01 7.07109571e-01 -3.72197062e-01 -5.71042955e-01 -4.33186829e-01 -3.37189347e-01 -9.38443482e-01 2.85794854e-01 6.13564730e-01 -3.97300661e-01 6.96535945e-01 8.52542147e-02 3.18040341e-01 -5.47572851e-01 -1.06789374e+00 3.65040362e-01 -3.70970219e-01 -8.87641788e-01 1.01880040e-02 -5.95427394e-01 3.14837664e-01 7.23068118e-01 -1.76475093e-01 -5.01886606e-01 1.26465216e-01 -3.50863524e-02 1.90701976e-01 9.97742355e-01 8.72883856e-01 4.88742709e-01 -1.12727547e+00 -3.47052395e-01 -2.89499927e-02 6.62061334e-01 -5.57123840e-01 -4.04656589e-01 3.55422765e-01 -8.23481023e-01 7.54901290e-01 -3.46449378e-04 -1.08637817e-01 -1.26509655e+00 8.21075499e-01 4.92211133e-02 -5.42040095e-02 -4.41600829e-01 4.38244611e-01 -1.63595110e-01 -4.75626975e-01 2.91780651e-01 -1.71898857e-01 -1.07269704e-01 -3.66628081e-01 7.48799920e-01 3.32009166e-01 2.68704206e-01 -2.50297487e-01 -7.50946522e-01 4.06084567e-01 7.56170899e-02 3.27035822e-02 1.20741963e+00 -1.35898918e-01 -8.09581816e-01 4.02799070e-01 1.33455992e+00 2.62973040e-01 -2.19290957e-01 -1.65480837e-01 9.90683317e-01 -7.96629548e-01 -5.61757982e-02 -7.27100194e-01 -7.89207339e-01 5.09076178e-01 3.48239958e-01 7.63475537e-01 9.14455235e-01 3.28375846e-01 4.42036271e-01 9.04545069e-01 8.39393735e-01 -9.43134785e-01 -2.97226727e-01 6.24920964e-01 7.91964591e-01 -8.58129203e-01 8.61318931e-02 -4.36731517e-01 1.31758228e-01 1.41954005e+00 4.67824101e-01 2.51059949e-01 8.17761421e-01 5.98844945e-01 1.97963685e-01 -7.40858912e-01 -8.45574439e-01 2.08831411e-02 8.50076452e-02 5.23987234e-01 9.85440433e-01 -5.75351477e-01 -7.16044903e-01 1.82941303e-01 1.06345274e-01 1.97325677e-01 5.77412903e-01 1.68447316e+00 -8.08976769e-01 -1.56455791e+00 -4.29396868e-01 5.32599866e-01 -7.61181653e-01 -3.99267375e-01 -4.35226113e-01 7.95102835e-01 -3.92467529e-01 9.62863207e-01 -3.44785184e-01 -7.42117912e-02 4.06269044e-01 2.45167419e-01 7.38010406e-02 -1.00796700e+00 -6.76069140e-01 -2.51062214e-01 4.09002416e-02 -7.00284481e-01 -3.84907782e-01 -6.08946264e-01 -1.18926442e+00 -1.39252782e-01 -6.38492167e-01 7.81281471e-01 7.70207345e-01 5.54824650e-01 5.38397789e-01 -3.48999575e-02 1.20047487e-01 3.36527713e-02 -6.41126931e-01 -4.47202027e-01 -9.49371755e-02 3.24031003e-02 1.11235917e-01 -5.41877389e-01 -2.55241752e-01 3.42071056e-01]
[7.548091888427734, 8.059141159057617]
7537d8e4-f091-42d5-a085-acde789b6660
integration-of-radiomics-and-tumor-biomarkers
2303.11177
null
https://arxiv.org/abs/2303.11177v1
https://arxiv.org/pdf/2303.11177v1.pdf
Integration of Radiomics and Tumor Biomarkers in Interpretable Machine Learning Models
Despite the unprecedented performance of deep neural networks (DNNs) in computer vision, their practical application in the diagnosis and prognosis of cancer using medical imaging has been limited. One of the critical challenges for integrating diagnostic DNNs into radiological and oncological applications is their lack of interpretability, preventing clinicians from understanding the model predictions. Therefore, we study and propose the integration of expert-derived radiomics and DNN-predicted biomarkers in interpretable classifiers which we call ConRad, for computerized tomography (CT) scans of lung cancer. Importantly, the tumor biomarkers are predicted from a concept bottleneck model (CBM) such that once trained, our ConRad models do not require labor-intensive and time-consuming biomarkers. In our evaluation and practical application, the only input to ConRad is a segmented CT scan. The proposed model is compared to convolutional neural networks (CNNs) which act as a black box classifier. We further investigated and evaluated all combinations of radiomics, predicted biomarkers and CNN features in five different classifiers. We found the ConRad models using non-linear SVM and the logistic regression with the Lasso outperform others in five-fold cross-validation, although we highlight that interpretability of ConRad is its primary advantage. The Lasso is used for feature selection, which substantially reduces the number of non-zero weights while increasing the accuracy. Overall, the proposed ConRad model combines CBM-derived biomarkers and radiomics features in an interpretable ML model which perform excellently for the lung nodule malignancy classification.
['Neo Christopher Chung', 'Lennart Brocki']
2023-03-20
null
null
null
null
['interpretable-machine-learning']
['methodology']
[ 2.06084803e-01 3.38095933e-01 -5.07754803e-01 -2.85789490e-01 -4.64428902e-01 -2.76414603e-01 3.72866571e-01 2.44502902e-01 -4.46737438e-01 7.34838068e-01 -7.58580072e-03 -7.54302800e-01 -5.51429212e-01 -6.98307931e-01 -3.48045260e-01 -9.01268244e-01 -1.26118017e-02 7.37863779e-01 6.11232640e-03 1.16248079e-01 -2.58044094e-01 8.49427938e-01 -1.18173873e+00 4.29540485e-01 9.48075712e-01 1.49628162e+00 3.23163986e-01 7.64810503e-01 -1.39520481e-01 1.05914223e+00 -2.16692775e-01 -3.70009840e-02 1.44026607e-01 -3.95916164e-01 -7.24884808e-01 -1.47504136e-01 1.99607134e-01 -2.56306529e-01 -3.65247726e-01 6.58529580e-01 3.26131076e-01 -4.05566514e-01 9.12540853e-01 -1.12267792e+00 -4.72057402e-01 4.92373556e-01 -3.92499939e-02 1.66372582e-01 -2.40839258e-01 2.12990984e-01 7.92517126e-01 -9.23348308e-01 4.37034756e-01 8.53418946e-01 1.04426289e+00 6.56050920e-01 -8.49292994e-01 -3.34755957e-01 -2.87359536e-01 3.14874463e-02 -9.45737898e-01 -3.20787728e-02 7.29976743e-02 -5.81214249e-01 6.31841958e-01 8.11841309e-01 9.51742828e-01 1.03399444e+00 7.38767505e-01 5.67391992e-01 1.10975480e+00 -2.98748940e-01 2.48687357e-01 3.33741039e-01 9.92117971e-02 1.33398175e+00 4.89641398e-01 4.93618667e-01 -2.05896422e-01 -3.04279387e-01 8.82823110e-01 6.19276702e-01 -2.53882706e-01 -1.18139476e-01 -1.40343189e+00 1.00540996e+00 9.20305014e-01 4.98958647e-01 -4.41938818e-01 3.88502866e-01 4.28366423e-01 -4.13201600e-02 3.89497608e-01 6.35098040e-01 -4.03666139e-01 4.63164091e-01 -1.04864502e+00 -3.67461711e-01 8.62939835e-01 5.54045975e-01 2.34663934e-01 7.51281753e-02 -4.54927742e-01 6.81814730e-01 1.40139818e-01 4.72196132e-01 9.97514069e-01 -5.00205815e-01 -1.25378892e-01 8.48508000e-01 -3.43847811e-01 -7.47992337e-01 -9.67375338e-01 -9.31914508e-01 -1.51460469e+00 1.63533941e-01 1.46636084e-01 -5.34821767e-03 -1.33550727e+00 1.12591064e+00 1.35479961e-02 -1.47531498e-02 6.67395592e-02 9.73851442e-01 1.18730414e+00 3.33172977e-02 3.25963020e-01 -6.91801906e-02 1.40368736e+00 -1.19237494e+00 -5.03314972e-01 -2.40940284e-02 9.92184162e-01 -4.35428560e-01 6.36793673e-01 1.63348943e-01 -5.86595893e-01 -4.17493522e-01 -1.04197228e+00 5.45033738e-02 -3.86315018e-01 6.69563890e-01 1.06420028e+00 4.91148859e-01 -9.22807038e-01 7.05827177e-01 -1.16545832e+00 -5.43147743e-01 7.39988148e-01 7.11739540e-01 -4.23924059e-01 -1.08634494e-01 -9.26761329e-01 1.11282015e+00 4.07605320e-01 4.07825172e-01 -1.13575363e+00 -8.23321223e-01 -5.98275840e-01 1.75694749e-01 2.01638296e-01 -1.19408631e+00 1.15551972e+00 -1.05255163e+00 -1.27196920e+00 7.17441678e-01 3.63027938e-02 -5.96567452e-01 7.97928751e-01 -7.89635070e-03 -1.54285952e-01 1.80606216e-01 5.94857056e-03 6.31435096e-01 6.13305390e-01 -1.02321279e+00 -5.85363686e-01 -8.27752724e-02 -2.79104650e-01 -1.19235769e-01 -4.21373636e-01 -4.19836819e-01 5.89040518e-02 -5.88647068e-01 4.14034843e-01 -1.03284776e+00 -5.85491955e-01 4.70933586e-01 -5.98878086e-01 2.48069093e-02 6.49243653e-01 -4.19574797e-01 9.71316040e-01 -1.83844149e+00 -6.32740781e-02 3.38933200e-01 7.96086967e-01 2.07878336e-01 1.67701751e-01 -7.67441317e-02 -1.93261221e-01 4.64984119e-01 -2.37338915e-01 8.47446695e-02 -4.54661995e-01 3.83553177e-01 3.49236757e-01 4.55847502e-01 5.26346505e-01 1.22646093e+00 -9.12885904e-01 -7.40404665e-01 3.60196352e-01 2.30003402e-01 -2.76902467e-01 2.64173359e-01 -1.38262212e-01 5.52955985e-01 -7.36172736e-01 9.68050659e-01 3.94947588e-01 -6.20204985e-01 1.08509086e-01 -3.01263303e-01 2.12379485e-01 -1.16078936e-01 -5.18502414e-01 1.26961005e+00 -6.13721073e-01 5.85484266e-01 -1.98528171e-01 -1.01688325e+00 8.18511844e-01 5.27144074e-01 7.77876318e-01 -4.84967828e-01 2.02908725e-01 6.27763093e-01 2.78042465e-01 -7.99213767e-01 -2.56462634e-01 -4.64374483e-01 2.70688713e-01 1.73166826e-01 4.60375138e-02 -9.31992102e-03 -2.26796791e-01 -1.48616329e-01 1.42891061e+00 -3.38911682e-01 5.99198937e-01 -3.92519146e-01 6.79098010e-01 4.37676102e-01 4.15561050e-01 8.60985935e-01 -1.08159184e-01 6.82059228e-01 5.03789902e-01 -7.27319062e-01 -9.45026815e-01 -9.01081622e-01 -4.32539821e-01 7.98378348e-01 -2.31638804e-01 1.46159694e-01 -3.05648983e-01 -1.06116176e+00 2.20579043e-01 3.06242406e-01 -8.60007703e-01 -8.18497315e-02 -6.52584195e-01 -1.01928043e+00 6.73171759e-01 7.12815583e-01 1.56693697e-01 -9.25808012e-01 -7.37192035e-01 1.93538427e-01 1.26820467e-02 -7.59677589e-01 -3.43995802e-02 8.79843533e-01 -1.29862976e+00 -1.42386115e+00 -6.66608334e-01 -6.16898835e-01 1.02199042e+00 2.81170756e-02 1.04649854e+00 5.41710496e-01 -6.93753541e-01 1.49459615e-01 -1.77583918e-01 -6.62880659e-01 -7.56798863e-01 2.59166777e-01 -4.75124419e-02 -2.96844721e-01 1.06334195e-01 -1.91804871e-01 -8.01120281e-01 2.70747215e-01 -9.02655840e-01 3.82371485e-01 1.26040113e+00 1.39888334e+00 7.81362414e-01 -2.87588537e-01 6.29619479e-01 -1.24873352e+00 2.84066021e-01 -7.35455155e-01 -2.11774990e-01 2.72666931e-01 -7.68184304e-01 1.14638090e-01 9.07718956e-01 -3.10789347e-01 -6.66178167e-01 1.87854096e-01 -8.73629376e-02 -3.79991323e-01 8.33128542e-02 7.84270167e-01 4.94355440e-01 -2.72587538e-01 9.31647003e-01 -5.96558005e-02 2.12091878e-01 -1.46972209e-01 -7.92403668e-02 5.48777342e-01 2.26699069e-01 -2.25133434e-01 6.34659708e-01 5.07982016e-01 6.96177065e-01 -4.90345120e-01 -1.13883603e+00 -5.22881210e-01 -7.05639243e-01 -1.72832742e-01 8.38239193e-01 -7.54181504e-01 -6.24593437e-01 7.29183927e-02 -9.18228865e-01 -2.78685298e-02 -4.56535280e-01 7.28612185e-01 -4.79933709e-01 5.10038882e-02 -6.84081733e-01 -6.18407488e-01 -7.24407077e-01 -1.19741940e+00 8.25286090e-01 1.58154219e-01 -2.30026782e-01 -1.14476871e+00 -3.91517013e-01 2.86732942e-01 7.01228559e-01 5.67283273e-01 1.25525510e+00 -1.15282047e+00 -5.72925508e-01 -4.48620319e-01 -5.71883380e-01 4.38067317e-01 2.81373143e-01 -1.23681717e-01 -1.09746778e+00 -1.72972202e-01 1.87060937e-01 -4.05275345e-01 1.02633762e+00 5.28580427e-01 1.48344886e+00 -3.59044462e-01 -7.47897744e-01 8.01191270e-01 1.60269630e+00 1.05109900e-01 1.64027214e-01 3.68537992e-01 7.28880107e-01 2.95506716e-01 4.13796097e-01 2.30509356e-01 -9.30966884e-02 5.24037406e-02 7.50780284e-01 -8.14826846e-01 -2.60483444e-01 1.65181234e-03 -7.85358548e-02 8.46979439e-01 -2.54327744e-01 -2.80552119e-01 -1.15369105e+00 3.10085833e-01 -1.85858965e+00 -4.42481458e-01 -4.40346509e-01 1.75918186e+00 5.79273045e-01 6.06235303e-02 -4.24887568e-01 4.22424600e-02 3.55201781e-01 -3.61569166e-01 -6.61688089e-01 -1.30147353e-01 -5.35160564e-02 4.71132725e-01 7.39423573e-01 8.55251402e-02 -1.21019089e+00 3.17561984e-01 6.49468231e+00 7.46958494e-01 -1.44743359e+00 4.11250293e-01 1.01815021e+00 1.08652718e-01 -1.22783266e-01 -3.13540310e-01 -3.47246855e-01 1.27064466e-01 1.00639415e+00 1.86197326e-01 -1.22933932e-01 9.75853920e-01 4.02584553e-01 -2.27193028e-01 -1.58272612e+00 6.07395709e-01 -1.72503531e-01 -1.60629606e+00 5.15814088e-02 2.25242809e-03 6.58295274e-01 3.04874986e-01 -2.21441016e-02 1.15575925e-01 1.09286994e-01 -1.66200423e+00 3.23340237e-01 6.89689457e-01 9.84094858e-01 -2.55220681e-01 1.40209031e+00 2.81095445e-01 -7.69182444e-01 -3.60933304e-01 -3.52633983e-01 3.24014813e-01 -4.19919759e-01 7.47960567e-01 -1.51834452e+00 7.38138080e-01 4.45878416e-01 5.24823666e-01 -6.75795257e-01 1.10892034e+00 9.73324478e-02 7.19484508e-01 -3.29387844e-01 -2.07505450e-01 4.97239530e-01 2.88206905e-01 2.29238763e-01 1.41143847e+00 5.15416205e-01 3.27774361e-02 2.86256999e-01 9.13655162e-01 7.91563317e-02 2.86965650e-02 -3.06007326e-01 4.16107662e-03 7.79958665e-02 1.55591941e+00 -9.68927205e-01 -3.19115937e-01 -1.70637593e-01 5.93639910e-01 8.71548206e-02 4.74903872e-03 -8.85836422e-01 1.64586589e-01 9.64210480e-02 2.77002633e-01 1.16476920e-02 2.21515119e-01 -7.09655941e-01 -9.04843450e-01 -2.02069655e-01 -6.48550391e-01 4.06693876e-01 -5.55349171e-01 -1.46948957e+00 8.74871850e-01 -3.27868402e-01 -1.41433048e+00 -8.16635415e-02 -1.08900452e+00 -6.28137708e-01 7.45411336e-01 -1.67319870e+00 -1.36070812e+00 -7.86613524e-01 2.26362899e-01 4.17656541e-01 -2.03410715e-01 1.01496637e+00 1.44273102e-01 -4.88573343e-01 4.20855612e-01 3.22034985e-01 1.51377946e-01 4.76216555e-01 -1.24113202e+00 -2.76382387e-01 1.42814144e-01 -3.11140925e-01 2.34150156e-01 2.42582873e-01 -5.38887084e-01 -1.15612543e+00 -1.47185683e+00 6.54710114e-01 -3.87306660e-01 5.49169064e-01 1.65782254e-02 -7.06395388e-01 4.53484267e-01 -4.09550697e-01 5.26467681e-01 8.31545711e-01 -3.47831666e-01 1.27156675e-01 -1.08272821e-01 -1.22369087e+00 1.30235031e-01 7.28963912e-01 -1.15621336e-01 -1.72770336e-01 7.69587278e-01 5.29428780e-01 -3.35147917e-01 -1.08704090e+00 7.64956176e-01 6.88212454e-01 -8.90139699e-01 9.55256104e-01 -8.13973904e-01 6.99949205e-01 -8.87670442e-02 9.96927395e-02 -1.06999290e+00 -4.38408792e-01 2.01116130e-01 1.04732811e-01 5.08708119e-01 6.43715262e-01 -6.13838077e-01 1.01910818e+00 4.26740676e-01 -3.24869305e-01 -1.43335700e+00 -1.10442173e+00 -6.75762177e-01 1.19442130e-02 -2.72990614e-01 1.43951371e-01 8.15056503e-01 -3.21390390e-01 -1.04345717e-01 1.67612564e-02 -2.78362520e-02 3.42127889e-01 -2.02373669e-01 6.60400540e-02 -1.44049942e+00 -2.33717501e-01 -5.20871341e-01 -5.69431484e-01 -4.21837360e-01 -1.78426966e-01 -1.38367367e+00 1.63970385e-02 -1.62330639e+00 5.97423136e-01 -8.92890573e-01 -6.27838731e-01 5.83122671e-01 -9.66805518e-02 1.36508524e-01 1.16549253e-01 5.09025693e-01 -1.72675416e-01 3.06258410e-01 1.40887487e+00 -4.88690436e-01 6.88760057e-02 2.68602461e-01 -4.29824054e-01 9.28405344e-01 6.14034474e-01 -7.98872352e-01 -1.41867429e-01 -1.83775455e-01 -4.69064079e-02 2.63386846e-01 7.15765715e-01 -1.16052365e+00 3.24785590e-01 -2.71730542e-01 9.55393553e-01 -5.11084020e-01 1.02106251e-01 -1.02728570e+00 3.15750182e-01 1.25711429e+00 -2.43129015e-01 -1.39468610e-01 6.77196160e-02 5.70879757e-01 -2.97299147e-01 -4.39879477e-01 8.08073878e-01 -5.42893708e-01 -2.41968244e-01 4.88294542e-01 -4.73246425e-01 -3.84524345e-01 1.10439527e+00 -3.50170910e-01 -2.10981339e-01 2.92605795e-02 -8.28410685e-01 1.40574813e-01 -9.25998166e-02 -2.62864828e-02 7.78224826e-01 -1.16756749e+00 -7.24960923e-01 1.95836206e-03 3.73930149e-02 4.44947809e-01 7.91748315e-02 1.21854722e+00 -9.44017887e-01 8.41267347e-01 -2.29113668e-01 -1.00054073e+00 -9.89917755e-01 4.31875795e-01 8.51640642e-01 -7.69877553e-01 -3.91512811e-01 7.66568184e-01 3.61828655e-01 -5.30288577e-01 1.95832506e-01 -7.43731380e-01 -1.93880960e-01 -2.30471551e-01 -1.10461675e-01 3.71207923e-01 2.99396783e-01 -1.21666767e-01 -3.27466905e-01 2.53179550e-01 -3.10415328e-02 4.77358162e-01 1.57403564e+00 3.60073030e-01 -3.91874075e-01 3.44099730e-01 1.02857780e+00 -5.83957434e-01 -6.77853107e-01 -2.94848774e-02 2.09843561e-01 6.78241551e-02 1.85945019e-01 -1.13325644e+00 -1.20151365e+00 7.30552733e-01 9.11019802e-01 1.20527968e-01 1.16216445e+00 8.30574930e-02 6.21139109e-01 6.17717206e-01 -1.31864976e-02 -6.06132388e-01 1.16441004e-01 2.12608874e-01 8.43488753e-01 -1.55625331e+00 3.03697258e-01 -6.05407238e-01 -4.03757930e-01 1.71442711e+00 7.89696753e-01 -3.41590680e-02 7.03369856e-01 2.30844215e-01 1.21603109e-01 -3.41805577e-01 -7.67932951e-01 8.66732746e-03 3.77579719e-01 4.60316628e-01 5.52165508e-01 3.43936056e-01 -3.19925785e-01 8.89459014e-01 -1.79185554e-01 3.36943299e-01 3.46101820e-01 7.67608345e-01 -4.25846308e-01 -7.00482965e-01 -3.39629501e-01 1.04112673e+00 -5.33632338e-01 -2.03681692e-01 -2.49195620e-01 1.05715764e+00 4.09303874e-01 5.25442302e-01 -1.44358605e-01 -3.38102937e-01 7.03239441e-02 -4.68548201e-02 1.01396538e-01 -6.55117035e-01 -9.40060496e-01 -2.07656860e-01 5.14159650e-02 -3.47042412e-01 -3.71542960e-01 -2.26547807e-01 -1.26391327e+00 -6.10971190e-02 -5.79813778e-01 -1.71385929e-01 7.52769053e-01 1.05908930e+00 1.07998163e-01 9.69846487e-01 6.06278718e-01 -3.94430310e-01 -7.19022095e-01 -1.06855440e+00 -2.52024114e-01 1.45379290e-01 5.33227265e-01 -5.59334815e-01 -3.29416990e-01 -4.21597585e-02]
[15.307098388671875, -2.221406936645508]
b71653a8-ef32-4215-8766-a299253367cb
modality-aware-negative-sampling-for-multi
2304.11618
null
https://arxiv.org/abs/2304.11618v1
https://arxiv.org/pdf/2304.11618v1.pdf
Modality-Aware Negative Sampling for Multi-modal Knowledge Graph Embedding
Negative sampling (NS) is widely used in knowledge graph embedding (KGE), which aims to generate negative triples to make a positive-negative contrast during training. However, existing NS methods are unsuitable when multi-modal information is considered in KGE models. They are also inefficient due to their complex design. In this paper, we propose Modality-Aware Negative Sampling (MANS) for multi-modal knowledge graph embedding (MMKGE) to address the mentioned problems. MANS could align structural and visual embeddings for entities in KGs and learn meaningful embeddings to perform better in multi-modal KGE while keeping lightweight and efficient. Empirical results on two benchmarks demonstrate that MANS outperforms existing NS methods. Meanwhile, we make further explorations about MANS to confirm its effectiveness.
['Wen Zhang', 'Mingyang Chen', 'Yichi Zhang']
2023-04-23
null
null
null
null
['graph-embedding', 'knowledge-graph-embedding', 'multi-modal-knowledge-graph']
['graphs', 'graphs', 'knowledge-base']
[-3.33085209e-01 1.85308680e-01 -6.71140671e-01 -8.78894255e-02 -3.01521361e-01 -3.02156240e-01 5.20739973e-01 3.90501954e-02 -3.79447848e-01 5.67271888e-01 5.24872780e-01 -1.56956464e-01 -1.99650884e-01 -1.17386580e+00 -5.53968191e-01 -5.85699022e-01 -1.41289234e-01 1.35923356e-01 1.02273889e-01 -4.18825209e-01 -1.56218693e-01 3.57335538e-01 -1.29031956e+00 1.14155030e-02 1.08080053e+00 5.38971424e-01 7.17757493e-02 1.30686224e-01 -3.50338876e-01 6.26243293e-01 -3.56488824e-01 -7.62614846e-01 -2.16654055e-02 -9.46244448e-02 -5.36356330e-01 -2.52652735e-01 3.46282423e-01 -8.83409604e-02 -5.60967624e-01 1.00375044e+00 7.84784198e-01 2.41582975e-01 7.82267690e-01 -1.87673116e+00 -1.50254941e+00 7.62855113e-01 -7.22847044e-01 7.01199174e-02 2.50676304e-01 4.34665717e-02 1.25676513e+00 -1.36368549e+00 7.49497771e-01 1.31796765e+00 6.63812637e-01 8.16442132e-01 -1.09243631e+00 -6.78300977e-01 1.16058595e-01 6.72607601e-01 -1.54139566e+00 -7.03068897e-02 1.03091717e+00 -1.23110339e-01 9.17448759e-01 3.33530128e-01 1.13072217e+00 1.05352604e+00 -2.06568107e-01 1.04012108e+00 6.37552321e-01 -3.96273881e-01 2.19959300e-02 3.24326098e-01 -1.65229589e-01 8.45390022e-01 6.12745762e-01 -1.04193878e-03 -8.44244421e-01 -7.97793418e-02 5.90887129e-01 -8.64740461e-02 -4.84911025e-01 -9.03415918e-01 -1.39285481e+00 9.68324363e-01 9.49560583e-01 2.77469784e-01 -4.22844231e-01 1.98491648e-01 5.14325738e-01 6.01176843e-02 4.35358554e-01 4.83545631e-01 -1.50073811e-01 1.68153584e-01 -4.01910126e-01 2.27051377e-01 2.90476650e-01 9.07276571e-01 7.06166446e-01 2.36867726e-01 -3.62113744e-01 8.80761504e-01 5.52890837e-01 5.50376177e-01 4.81748253e-01 -2.64125139e-01 5.51927745e-01 1.09457338e+00 -1.22087792e-01 -1.25523007e+00 -4.14349973e-01 -3.54916513e-01 -8.94747436e-01 -2.94351548e-01 -2.23630413e-01 1.78158477e-01 -9.76824582e-01 1.63188279e+00 5.43800473e-01 3.28853399e-01 2.99341708e-01 9.69737351e-01 1.39520359e+00 6.42689228e-01 5.18182993e-01 7.36732483e-02 1.26875782e+00 -1.03350258e+00 -9.44904089e-01 -3.03238481e-01 1.09630418e+00 -4.44901049e-01 1.34714878e+00 -2.42751583e-01 -6.43960893e-01 -1.54796407e-01 -9.33666468e-01 -5.81014417e-02 -8.56971920e-01 4.53445226e-01 1.08865047e+00 5.95371604e-01 -9.78322983e-01 7.24648163e-02 -4.07887012e-01 -3.56788337e-01 6.12161934e-01 2.54793495e-01 -5.67882657e-01 -2.95145988e-01 -1.67768109e+00 8.25532079e-01 8.13325822e-01 2.43745327e-01 -4.30443197e-01 -9.80550468e-01 -1.24961042e+00 3.63655128e-02 3.59136015e-01 -8.69682789e-01 4.53779280e-01 -4.32715088e-01 -1.03845012e+00 7.00775445e-01 -2.38236133e-03 -7.21713603e-02 2.59931386e-01 2.94699911e-02 -7.48983204e-01 1.57203153e-01 4.71312590e-02 9.80439842e-01 6.04150653e-01 -1.48869956e+00 -8.34463015e-02 -4.05119449e-01 3.31604838e-01 4.48940814e-01 -1.16222060e+00 -4.38645393e-01 -6.33154452e-01 -8.19440901e-01 3.23150419e-02 -8.06353211e-01 -1.26584604e-01 1.26547199e-02 -6.00460887e-01 -5.77069283e-01 1.00271463e+00 -4.66784626e-01 1.50916517e+00 -2.13153362e+00 3.29629034e-01 3.50958079e-01 4.06057179e-01 5.83171606e-01 -2.41389140e-01 6.07816339e-01 -1.74420148e-01 2.58889437e-01 2.71075994e-01 -2.01916903e-01 3.34099114e-01 2.59645551e-01 -8.81073549e-02 2.29591176e-01 2.76962966e-01 1.61392629e+00 -1.15128887e+00 -8.89479220e-01 3.61751854e-01 6.04963541e-01 -6.08558774e-01 8.92884657e-02 -4.27635349e-02 -2.39395663e-01 -5.21493793e-01 9.65816438e-01 7.53534079e-01 -4.38745767e-01 3.78788710e-01 -8.46241951e-01 3.53681356e-01 -3.17172468e-01 -1.08522248e+00 1.29711819e+00 -2.92310238e-01 3.29173744e-01 -4.90521878e-01 -9.85420525e-01 7.52103209e-01 1.32873505e-01 1.94147527e-01 -8.68743777e-01 2.33942404e-01 6.89372495e-02 -2.96969593e-01 -6.25380397e-01 1.06326556e+00 -4.26307380e-01 1.33317754e-01 5.87751018e-03 4.79472652e-02 2.46130064e-01 3.08857769e-01 7.27837980e-01 7.18601286e-01 -1.08466715e-01 6.29904270e-02 -2.27782018e-02 2.85978109e-01 -1.30931526e-01 4.84671950e-01 9.19206366e-02 -3.32980752e-01 2.24694729e-01 3.88542682e-01 -9.58718657e-02 -6.79871321e-01 -1.17368329e+00 1.87180653e-01 7.94399321e-01 5.39212584e-01 -8.07987452e-01 -2.14484453e-01 -1.10420287e+00 1.24141358e-01 6.12136841e-01 -8.82825553e-01 -6.34939909e-01 -2.48649329e-01 -9.62009549e-01 4.37782645e-01 7.50600755e-01 3.65627289e-01 -1.07339275e+00 2.79310554e-01 -7.70865977e-02 -2.85251439e-01 -8.98911297e-01 -4.21611428e-01 -3.56646210e-01 -5.87148666e-01 -1.27263629e+00 -9.06819940e-01 -1.07869792e+00 1.10025167e+00 5.24322391e-01 1.01578724e+00 -6.96716011e-02 -1.39016911e-01 6.83962643e-01 -6.81159616e-01 -3.11601967e-01 9.35372710e-02 5.36638498e-02 7.37349167e-02 -1.59663752e-01 5.90094149e-01 -3.68909627e-01 -6.59475446e-01 1.48231909e-01 -1.25253212e+00 1.65738508e-01 7.21307993e-01 1.01472104e+00 5.72941840e-01 3.59259509e-02 8.57832253e-01 -8.48618329e-01 8.53068948e-01 -5.99533737e-01 -6.92424830e-04 7.51660585e-01 -7.92354167e-01 -5.03637567e-02 6.48534477e-01 -5.63817680e-01 -8.06552351e-01 -5.24605870e-01 3.53201889e-02 -7.81101584e-01 5.60331225e-01 8.21759939e-01 -2.42899358e-01 -3.33730757e-01 4.83641624e-01 2.38470346e-01 -2.03075334e-01 -1.58746317e-01 6.67295337e-01 2.88320392e-01 7.31956810e-02 -4.61181849e-01 1.08161569e+00 3.60939533e-01 2.64815073e-02 -6.65243149e-01 -6.18857384e-01 -3.18280667e-01 -8.10292810e-02 -2.22714409e-01 6.07697368e-01 -9.40131843e-01 -6.51049912e-01 8.11153352e-02 -6.85088634e-01 -7.72711486e-02 -4.09033686e-01 7.68380225e-01 -8.18180516e-02 4.62949872e-01 -2.99850553e-01 -6.34777427e-01 -4.07915145e-01 -6.01744175e-01 8.87780130e-01 5.02860546e-01 3.99304107e-02 -1.34153295e+00 1.82077903e-02 3.88743013e-01 3.87981623e-01 1.85917154e-01 1.05295753e+00 -2.48614028e-01 -5.64166605e-01 -1.37106135e-01 -4.57283378e-01 1.61957189e-01 3.94649804e-03 1.30582839e-01 -6.57895505e-01 -3.19588959e-01 -9.53852654e-01 -5.13770044e-01 8.35772157e-01 1.44785389e-01 1.14228451e+00 -2.91941643e-01 -5.16727924e-01 5.04899323e-01 1.68079150e+00 -3.06108177e-01 7.37889111e-01 6.06929362e-01 1.18312013e+00 4.11274523e-01 1.09621346e+00 3.89484018e-01 8.38694811e-01 5.28882682e-01 6.68844759e-01 -3.60721976e-01 -1.37224376e-01 -6.22854054e-01 3.44309151e-01 1.08529902e+00 -3.09119560e-02 -3.72117817e-01 -6.94980741e-01 1.18446875e+00 -1.74984598e+00 -8.32458675e-01 -2.38535702e-01 1.69451392e+00 8.72295022e-01 -2.65065461e-01 7.39433914e-02 1.66257083e-01 6.71835184e-01 4.34883714e-01 -3.00168306e-01 -1.01197936e-01 -3.43764812e-01 1.25789225e-01 3.77110779e-01 -1.46847889e-02 -9.14008200e-01 1.00310600e+00 5.33547688e+00 1.25645304e+00 -9.49330509e-01 1.18995793e-01 3.82933579e-02 -1.58219397e-01 -1.18366611e+00 -7.63156414e-02 -6.81116700e-01 3.96409720e-01 4.44715619e-01 -2.80620039e-01 2.62511551e-01 8.27680290e-01 -1.61404386e-01 2.10641086e-01 -6.11088157e-01 1.20033407e+00 1.39444903e-01 -1.51453602e+00 4.52160478e-01 -2.08913207e-01 8.58617485e-01 -2.07818642e-01 1.79155171e-01 6.76024616e-01 1.45835340e-01 -9.43176687e-01 3.70063394e-01 2.50632852e-01 8.48877132e-01 -9.41608548e-01 7.04784274e-01 4.11540363e-03 -1.45132792e+00 1.38445318e-01 -4.78235424e-01 5.11713326e-01 3.65493685e-01 4.50308353e-01 -8.45463455e-01 1.01164901e+00 5.65170705e-01 8.60704720e-01 -8.73233378e-01 8.62057447e-01 -4.17208493e-01 4.41216797e-01 -2.65757572e-02 -2.44211674e-01 2.39397541e-01 -1.54062957e-01 3.91090333e-01 9.82878864e-01 3.86611551e-01 -2.57610261e-01 -6.33709058e-02 8.14990282e-01 -3.07028264e-01 5.03648818e-01 -7.58724630e-01 -5.47442853e-01 6.37358427e-01 1.35176361e+00 -5.15537083e-01 -2.55811870e-01 -4.66819376e-01 7.85176575e-01 6.03497803e-01 5.43605447e-01 -1.03685677e+00 -5.55465400e-01 4.52889383e-01 -8.42986442e-03 3.53115290e-01 4.22274023e-02 5.01880087e-02 -1.31886315e+00 2.72405714e-01 -4.47671026e-01 5.68953156e-01 -9.03291881e-01 -1.51228178e+00 3.00887078e-01 -5.01714088e-02 -1.40742695e+00 2.88400233e-01 -4.60530370e-01 -3.34181726e-01 5.83389521e-01 -1.93640196e+00 -1.74675679e+00 -2.99623132e-01 7.15698063e-01 -1.70669332e-01 6.38057524e-03 8.44761610e-01 7.18158245e-01 -7.47668564e-01 1.00424290e+00 -4.86666290e-03 8.14964697e-02 5.76525629e-01 -1.31012928e+00 4.98950891e-02 6.66255414e-01 1.74966633e-01 8.58139217e-01 4.23505515e-01 -6.87790930e-01 -1.74110985e+00 -1.36827195e+00 9.57474649e-01 -1.01431450e-02 7.42683828e-01 -3.33453319e-03 -8.81284535e-01 6.85272098e-01 2.98311953e-02 3.54840904e-01 1.02368140e+00 2.84645230e-01 -4.85277683e-01 -5.14924452e-02 -1.08405507e+00 9.18012142e-01 1.09839118e+00 -5.81998467e-01 -4.56986666e-01 2.70294279e-01 6.77960873e-01 -2.49000952e-01 -1.29147983e+00 6.15297914e-01 4.28770781e-01 -6.48813546e-01 1.26166511e+00 -6.21724308e-01 3.94074559e-01 -5.12416244e-01 5.99379055e-02 -1.70461297e+00 -3.44566584e-01 -7.84582272e-02 -5.65232337e-01 1.39919066e+00 2.24578366e-01 -8.57464671e-01 9.22297597e-01 4.37871814e-01 -2.11090725e-02 -1.10583818e+00 -7.49638736e-01 -9.46176827e-01 -2.01012596e-01 -1.21999219e-01 8.23642612e-01 1.59241462e+00 1.22786112e-01 4.47066456e-01 -4.66214180e-01 1.88965917e-01 5.20335913e-01 -3.97304632e-02 7.70915091e-01 -8.64105225e-01 2.05049798e-01 -2.12371126e-01 -7.28597045e-01 -6.85865164e-01 1.95102796e-01 -1.11623001e+00 -5.62879145e-01 -2.01274514e+00 4.13981646e-01 -6.30934954e-01 -4.54041213e-01 5.86980760e-01 -5.52149773e-01 5.76514423e-01 1.49337813e-01 -1.69490606e-01 -7.80524313e-01 1.29822707e+00 1.53500736e+00 -3.63172621e-01 9.19690058e-02 -6.91503823e-01 -9.07720208e-01 3.72573018e-01 8.97788227e-01 -2.12039247e-01 -8.59678864e-01 -2.15162352e-01 6.44629419e-01 -4.05380696e-01 5.00218809e-01 -4.71439362e-01 5.06496690e-02 -2.34331757e-01 3.26188982e-01 -7.67783761e-01 3.57750475e-01 -8.01915407e-01 7.35607594e-02 1.27716109e-01 7.59441927e-02 -3.10927741e-02 2.41194189e-01 7.74957240e-01 -4.67809111e-01 -5.18422760e-02 3.75463963e-01 2.85368979e-01 -1.29178476e+00 5.81928551e-01 2.25924328e-01 3.31195533e-01 1.23245537e+00 -3.88732702e-01 -6.20370269e-01 -8.80986527e-02 -6.07010305e-01 5.93786895e-01 5.15731692e-01 5.89398026e-01 1.08437872e+00 -2.25410175e+00 -3.60200107e-01 9.24252719e-02 8.48495483e-01 -6.37201294e-02 7.99898207e-01 9.44855332e-01 -3.74009132e-01 2.55482465e-01 3.12745348e-02 -1.97703764e-01 -1.44705021e+00 5.91080070e-01 -3.01535688e-02 -3.82919073e-01 -4.56931084e-01 1.04398966e+00 1.25747427e-01 -7.76151299e-01 1.58142865e-01 1.16030544e-01 -5.03326118e-01 2.50135690e-01 3.49581718e-01 3.44607204e-01 -1.74734965e-01 -6.94581985e-01 -4.75282699e-01 3.43160152e-01 6.30614087e-02 3.74971300e-01 1.41482759e+00 4.51787412e-02 -1.32799268e-01 2.25533023e-01 1.31770325e+00 -1.30129913e-02 -6.15686417e-01 -3.81339073e-01 -3.11682761e-01 -7.42636919e-01 9.65168104e-02 -5.09438157e-01 -1.33532691e+00 7.21223593e-01 4.76040334e-01 1.23439923e-01 1.04870963e+00 5.83730750e-02 9.93191421e-01 1.88840464e-01 2.61994720e-01 -1.17050028e+00 2.51307309e-01 2.56309975e-02 7.94972003e-01 -1.17181349e+00 1.29645139e-01 -5.78344762e-01 -9.19650197e-01 8.27447593e-01 1.05847025e+00 2.46393904e-01 6.63035333e-01 -4.87199962e-01 -1.30619302e-01 -6.70482516e-01 -5.37560225e-01 -3.64061475e-01 4.64224786e-01 8.21097910e-01 2.86327809e-01 2.09236622e-01 -6.01977110e-01 4.25258547e-01 5.65489605e-02 -6.77296077e-04 1.75257072e-01 9.18325126e-01 4.50118780e-02 -1.20304203e+00 -8.24683458e-02 6.44275427e-01 -3.41370218e-02 -1.81969255e-01 -3.65990788e-01 1.13731849e+00 1.57478675e-01 6.29166842e-01 -3.66001219e-01 -7.16939330e-01 4.28325355e-01 -1.66169181e-01 6.17867172e-01 -4.48065072e-01 -1.71881124e-01 -5.06070614e-01 1.99259996e-01 -3.92149180e-01 -7.11748660e-01 -1.31124958e-01 -1.28709531e+00 -5.73476911e-01 -6.37242734e-01 2.18644381e-01 3.68161142e-01 4.53906327e-01 5.38255394e-01 6.67879462e-01 4.91413355e-01 -5.90647042e-01 -3.53864044e-01 -7.95978665e-01 -1.00013232e+00 7.24267721e-01 -2.51196861e-01 -1.09124136e+00 -2.93469608e-01 -3.74127060e-01]
[8.6871976852417, 7.7601399421691895]
b016a235-4f10-477a-9acb-a9aa675388bf
uwspeech-speech-to-speech-translation-for
2006.07926
null
https://arxiv.org/abs/2006.07926v2
https://arxiv.org/pdf/2006.07926v2.pdf
UWSpeech: Speech to Speech Translation for Unwritten Languages
Existing speech to speech translation systems heavily rely on the text of target language: they usually translate source language either to target text and then synthesize target speech from text, or directly to target speech with target text for auxiliary training. However, those methods cannot be applied to unwritten target languages, which have no written text or phoneme available. In this paper, we develop a translation system for unwritten languages, named as UWSpeech, which converts target unwritten speech into discrete tokens with a converter, and then translates source-language speech into target discrete tokens with a translator, and finally synthesizes target speech from target discrete tokens with an inverter. We propose a method called XL-VAE, which enhances vector quantized variational autoencoder (VQ-VAE) with cross-lingual (XL) speech recognition, to train the converter and inverter of UWSpeech jointly. Experiments on Fisher Spanish-English conversation translation dataset show that UWSpeech outperforms direct translation and VQ-VAE baseline by about 16 and 10 BLEU points respectively, which demonstrate the advantages and potentials of UWSpeech.
['Ke-jun Zhang', 'Tie-Yan Liu', 'Yi Ren', 'Chen Zhang', 'Xu Tan', 'Tao Qin']
2020-06-14
null
null
null
null
['speech-to-speech-translation']
['speech']
[ 2.58030146e-01 1.51789337e-01 -2.22921386e-01 -3.07169229e-01 -1.50792575e+00 -9.01007354e-01 6.89606607e-01 -6.72771573e-01 -1.15682021e-01 9.27354217e-01 3.38435680e-01 -9.56861675e-01 7.20449865e-01 -5.90435922e-01 -1.00010133e+00 -5.48712671e-01 8.29403639e-01 8.19426596e-01 -1.04215100e-01 -5.44779718e-01 -5.64786494e-01 -5.65538295e-02 -7.22882271e-01 3.57237309e-01 9.80097353e-01 5.93548179e-01 4.78392690e-01 7.07116127e-01 -3.61378312e-01 5.12543917e-01 -6.77088559e-01 -4.35675144e-01 9.12374854e-02 -1.02679718e+00 -6.81904972e-01 6.47002980e-02 -2.35765167e-02 -3.63867104e-01 -4.85409468e-01 1.06289697e+00 5.74321806e-01 6.05982542e-02 8.47539902e-01 -1.22188795e+00 -1.18782449e+00 1.10719192e+00 -1.20703414e-01 -5.37853874e-02 2.54307896e-01 2.31212210e-02 9.26293492e-01 -1.26632726e+00 5.55795491e-01 1.47492659e+00 2.45796204e-01 8.85702848e-01 -1.13347125e+00 -8.78698945e-01 -2.19634444e-01 -7.32020289e-02 -1.35603368e+00 -1.00813770e+00 5.60105145e-01 -1.10365964e-01 1.20344079e+00 2.73759425e-01 2.49772429e-01 1.67249060e+00 1.15665309e-01 9.06599700e-01 9.18879986e-01 -4.94280040e-01 5.01798019e-02 5.28368711e-01 -5.35978556e-01 4.27524954e-01 -4.53818262e-01 2.57164985e-01 -4.47200537e-01 8.18025395e-02 5.57726920e-01 -3.08729500e-01 -4.94003683e-01 3.06539267e-01 -1.60515058e+00 8.87777627e-01 -3.15588899e-02 1.76411718e-01 -2.59697765e-01 -2.37948716e-01 6.44113719e-01 7.84902215e-01 4.74104702e-01 -8.36019814e-02 -4.41492289e-01 -2.55644143e-01 -8.23765397e-01 -2.84394711e-01 8.71615469e-01 1.54804432e+00 6.63981259e-01 6.51725113e-01 -3.22329819e-01 1.13524413e+00 3.80840987e-01 1.48815584e+00 9.37774539e-01 -6.45513117e-01 9.23143923e-01 -6.05329359e-03 2.56176125e-02 -1.81307331e-01 4.51615930e-01 -1.49395525e-01 -7.60851026e-01 -2.03428686e-01 -2.20081627e-01 -4.53000426e-01 -8.06697845e-01 1.63842082e+00 2.86362499e-01 -7.19595030e-02 1.02330041e+00 8.18814278e-01 9.86243606e-01 1.62808728e+00 -2.67853737e-01 -4.73059416e-01 1.07703531e+00 -1.57032835e+00 -9.54132318e-01 -3.73329908e-01 4.45769131e-01 -1.12251163e+00 1.43619180e+00 -1.66025814e-02 -1.23943448e+00 -7.56883323e-01 -8.27796698e-01 -2.49416560e-01 -1.57467633e-01 5.73057353e-01 -2.92966098e-01 6.09629631e-01 -1.13310993e+00 6.88984841e-02 -8.45774353e-01 -2.10257500e-01 -2.63439208e-01 3.17106843e-01 -3.35520774e-01 1.28464296e-01 -1.63087285e+00 8.92478466e-01 1.78368852e-01 2.46053492e-03 -1.28540623e+00 -2.92460650e-01 -1.14874578e+00 1.43634990e-01 8.91500711e-02 -3.84628564e-01 1.68606222e+00 -1.21485734e+00 -2.36502767e+00 4.55785692e-01 -5.07766843e-01 -2.93019921e-01 4.68380094e-01 2.81922370e-01 -6.94357336e-01 -1.87990159e-01 3.50853443e-01 3.69075358e-01 9.21390533e-01 -1.08419764e+00 -7.19917893e-01 1.33287190e-02 -4.38699841e-01 3.87801200e-01 -4.10367280e-01 1.08203225e-01 -5.45450628e-01 -5.42358220e-01 -1.99053392e-01 -9.10823882e-01 3.21715146e-01 -6.05510592e-01 -4.48854268e-01 -4.03053701e-01 1.16307294e+00 -1.00195181e+00 1.02533865e+00 -2.07783198e+00 4.23266798e-01 -1.21734872e-01 -3.72905433e-01 4.61532533e-01 -5.11029243e-01 7.60279536e-01 2.20247120e-01 -1.81325749e-01 -2.77082026e-01 -7.05457330e-01 1.53918192e-01 4.73322600e-01 -9.39203799e-01 1.14145011e-01 2.39955232e-01 1.22236514e+00 -9.73641694e-01 -4.40281779e-01 2.49563485e-01 7.02969730e-01 -1.26468360e-01 5.05116343e-01 -1.17373429e-01 4.62620169e-01 -3.68029386e-01 6.13661468e-01 3.38221222e-01 1.24435514e-01 1.11576840e-01 1.47785559e-01 -1.98333412e-01 9.03579175e-01 -4.94430333e-01 1.54110217e+00 -9.82861459e-01 7.13187993e-01 3.12020212e-01 -8.84850621e-01 1.16173136e+00 8.33319426e-01 -1.05603479e-01 -6.21693969e-01 1.83173224e-01 6.46924078e-01 -2.05066398e-01 -2.86619931e-01 4.46622431e-01 -3.87518078e-01 -1.84541598e-01 5.04317343e-01 2.78455883e-01 -5.98687887e-01 -2.41082177e-01 3.67164775e-03 6.14460528e-01 2.68853456e-01 1.32080525e-01 8.55522677e-02 7.17958748e-01 5.23094609e-02 4.16114748e-01 2.96497732e-01 -6.31627887e-02 5.48719823e-01 1.91669121e-01 4.00222629e-01 -1.20219946e+00 -1.48974717e+00 2.35419616e-01 1.21402013e+00 -8.99679810e-02 -2.01726973e-01 -1.03597236e+00 -6.85239017e-01 -4.92290795e-01 1.09831703e+00 -1.31311268e-01 -2.22559810e-01 -7.33163118e-01 -4.04684581e-02 9.72544670e-01 3.93067628e-01 2.91352153e-01 -1.20206642e+00 3.40859026e-01 5.63072324e-01 -8.03969085e-01 -1.36452746e+00 -1.34517205e+00 -5.85960317e-03 -4.43547100e-01 -3.30706209e-01 -1.13268888e+00 -1.38210344e+00 2.22663254e-01 2.10505649e-01 8.46076369e-01 -5.43149829e-01 6.57828689e-01 -2.80613992e-02 -5.19501925e-01 -2.99200088e-01 -1.41519904e+00 3.43406975e-01 4.41464305e-01 1.68795466e-01 3.27634573e-01 -3.39948535e-01 8.35239813e-02 3.54808927e-01 -5.85421443e-01 9.78483632e-02 6.86480761e-01 1.14555931e+00 7.15252161e-01 -5.32572627e-01 7.51270592e-01 -5.57378650e-01 7.58979559e-01 -4.07500118e-01 -5.38546324e-01 3.75457644e-01 -4.07663733e-01 -1.01406220e-02 1.27169573e+00 -7.77712107e-01 -1.01133478e+00 -1.47896335e-01 -5.10321736e-01 -9.17806447e-01 3.19051385e-01 3.55999500e-01 -5.68202913e-01 4.34315860e-01 5.13030767e-01 1.05881786e+00 -3.91029231e-02 -1.65521309e-01 5.24930835e-01 1.70053601e+00 7.40237772e-01 -4.05313253e-01 8.17305803e-01 -1.29639566e-01 -7.85562754e-01 -9.02279854e-01 -2.57157534e-01 -1.36993125e-01 -2.81339079e-01 1.66754365e-01 9.49130177e-01 -1.12382138e+00 -3.83771777e-01 3.97762537e-01 -1.54555690e+00 -5.74427128e-01 -8.86449963e-02 7.58830488e-01 -7.24756658e-01 1.81185424e-01 -7.55976140e-01 -6.61838233e-01 -6.37279212e-01 -1.59704697e+00 1.47137606e+00 -2.00094417e-01 1.06726401e-03 -9.87363398e-01 -3.04948892e-02 4.39137697e-01 4.52779382e-01 -4.85606164e-01 6.99072421e-01 -7.05156386e-01 -3.77882093e-01 5.00730649e-02 1.00801073e-01 9.22965348e-01 5.28703153e-01 -1.86310440e-01 -8.02603364e-01 -4.09347862e-01 -8.64676759e-02 -4.57051724e-01 2.94561952e-01 1.52236715e-01 3.01432282e-01 -5.79063356e-01 -1.73774302e-01 6.31420016e-01 8.70283782e-01 4.50154513e-01 3.15066457e-01 -1.91355243e-01 6.19262099e-01 3.43210578e-01 4.21409935e-01 -1.00571230e-01 6.96006119e-01 6.69613719e-01 -6.79349601e-02 -1.93275779e-01 -3.49680990e-01 -6.38851345e-01 1.25535631e+00 2.04130888e+00 3.50471914e-01 -5.95884979e-01 -6.43982470e-01 5.87693274e-01 -1.51193583e+00 -5.83902717e-01 7.86798373e-02 1.84120679e+00 1.16683972e+00 -1.94683611e-01 -2.19580039e-01 -2.75641203e-01 9.11192954e-01 7.21700191e-02 -5.94425559e-01 -6.32159054e-01 -2.88606770e-02 2.23153874e-01 4.34866279e-01 1.02436316e+00 -5.72831929e-01 1.61857784e+00 5.32987833e+00 1.05061615e+00 -1.50061500e+00 6.22135997e-01 3.50953490e-01 2.43225753e-01 -5.92925131e-01 -8.79978761e-02 -8.47745180e-01 5.28124392e-01 1.52172089e+00 -4.05012369e-01 8.80287826e-01 8.45752239e-01 4.35399413e-01 8.93803418e-01 -1.28878987e+00 1.00542223e+00 7.54918009e-02 -1.07072830e+00 2.60054767e-01 -2.96075702e-01 7.10118175e-01 2.61928946e-01 8.19898695e-02 9.38701689e-01 4.74525809e-01 -1.01520479e+00 9.22840238e-01 -6.01980910e-02 1.53311265e+00 -7.43130445e-01 5.46773732e-01 5.80010593e-01 -1.27367890e+00 4.65323269e-01 -5.61682522e-01 2.83311278e-01 3.24083686e-01 -1.47525862e-01 -1.23617935e+00 6.29092455e-01 2.96228766e-01 5.27099907e-01 2.81000912e-01 -7.04234689e-02 -4.31613028e-01 1.10949230e+00 -1.45520359e-01 -5.02773583e-01 5.30129373e-01 -4.27178890e-01 5.69629192e-01 1.31117213e+00 8.26327443e-01 -1.00760952e-01 2.25647435e-01 9.28496540e-01 -2.61754870e-01 4.08751249e-01 -5.46422303e-01 -4.86459851e-01 8.89009058e-01 7.76842892e-01 -1.25988975e-01 -7.39866734e-01 -5.31197906e-01 1.60936546e+00 2.50432957e-02 7.96721816e-01 -8.88750315e-01 -7.60387361e-01 7.64875054e-01 -3.24032903e-01 2.32130826e-01 -2.22669616e-01 1.43672466e-01 -1.43565357e+00 5.20451441e-02 -1.31111181e+00 -3.30528557e-01 -9.87561822e-01 -1.13846374e+00 1.25429952e+00 -4.03658748e-01 -1.42202175e+00 -6.96086764e-01 -2.31432974e-01 -5.11757135e-01 1.47998333e+00 -1.35664213e+00 -1.51813781e+00 2.75837421e-01 8.81615698e-01 1.28955245e+00 -7.07848489e-01 9.25918460e-01 2.21937373e-01 -3.80429864e-01 8.81444156e-01 6.30284548e-01 4.56735969e-01 8.26440096e-01 -9.41683352e-01 9.92915630e-01 7.73947597e-01 2.44201466e-01 3.74572694e-01 7.09732771e-01 -4.60388780e-01 -1.55937350e+00 -1.53859580e+00 1.33249199e+00 -4.63913918e-01 5.78829885e-01 -7.52046108e-01 -7.79803216e-01 1.11019719e+00 6.70755506e-01 -2.44166851e-01 4.22069788e-01 -6.15718782e-01 -1.35996714e-01 -1.64206419e-02 -8.05554509e-01 7.70555913e-01 7.53896892e-01 -1.09720039e+00 -8.88295412e-01 3.82150203e-01 1.42898679e+00 -6.80926800e-01 -6.28082395e-01 5.29529899e-02 3.08177978e-01 -2.14547530e-01 6.94886088e-01 -4.49174851e-01 3.26713085e-01 -2.79493928e-01 -6.00517690e-01 -1.96430957e+00 1.89347953e-01 -8.71231794e-01 2.54728287e-01 1.46596813e+00 8.67027402e-01 -9.23644066e-01 1.91677868e-01 -1.43040136e-01 -6.67368770e-01 -3.92263442e-01 -1.15548325e+00 -9.28108990e-01 5.39019227e-01 -3.49848658e-01 8.23705792e-01 9.49690938e-01 -9.82521772e-02 8.31370473e-01 -5.99924088e-01 3.28592420e-01 2.79096961e-01 -1.45602738e-04 8.35134029e-01 -3.26766461e-01 -4.38942879e-01 -2.41790116e-01 1.29639566e-01 -1.52750659e+00 6.45157516e-01 -1.24239528e+00 4.70581770e-01 -1.59864473e+00 -3.59484017e-01 -6.82700649e-02 2.02534333e-01 4.61245865e-01 -7.16356933e-02 2.30641529e-01 -5.30946665e-02 2.57003844e-01 1.45246729e-01 1.13822770e+00 1.46783209e+00 -4.83530372e-01 -3.33571285e-01 7.80970156e-02 -3.73171538e-01 1.29978091e-01 4.22321588e-01 -4.60159242e-01 -6.34695888e-01 -6.53392911e-01 -5.31677425e-01 7.33269215e-01 -2.30454624e-01 -5.14955878e-01 3.05098705e-02 -2.93909460e-01 7.09080719e-04 -3.98942798e-01 4.25214469e-01 -5.20586133e-01 -7.88365975e-02 2.34255850e-01 -3.24174136e-01 9.91239324e-02 2.15003025e-02 1.79296583e-01 -5.62446415e-01 -6.68599531e-02 6.62159264e-01 5.82887232e-02 -3.36243063e-01 3.78260374e-01 -6.65709019e-01 6.57786950e-02 7.78971851e-01 4.68780436e-02 -1.75667718e-01 -6.76550865e-01 -7.05705047e-01 1.54901341e-01 2.26812303e-01 8.23452771e-01 6.61780536e-01 -1.79054928e+00 -1.12416553e+00 5.92887640e-01 3.21467826e-03 -2.05818847e-01 -7.74187744e-02 5.82360983e-01 -2.90984005e-01 6.14752233e-01 1.52917460e-01 -6.39311016e-01 -1.05924642e+00 4.78863329e-01 3.78704280e-01 2.32780173e-01 -4.50023532e-01 8.50213826e-01 2.92079657e-01 -1.13456905e+00 4.07157205e-02 -4.43665147e-01 2.72504628e-01 -2.57920235e-01 3.66101861e-01 -2.13063117e-02 1.73928976e-01 -1.13016498e+00 -2.89900690e-01 3.64434808e-01 1.68472812e-01 -6.91746891e-01 7.71469295e-01 -5.00067055e-01 -5.76343276e-02 6.18313551e-01 1.47767472e+00 2.97011346e-01 -1.00115013e+00 -4.79917705e-01 -5.96529961e-01 -1.27439934e-03 -1.37241319e-01 -4.94902194e-01 -7.80037701e-01 1.17781258e+00 1.64132535e-01 -4.17930931e-02 9.47728992e-01 1.22565851e-01 1.52175868e+00 5.90031505e-01 3.36642325e-01 -1.01014721e+00 -1.69569209e-01 8.89367044e-01 1.05021656e+00 -1.19105947e+00 -1.06348395e+00 -1.48808584e-01 -1.12977755e+00 1.11651635e+00 3.77198488e-01 1.81399718e-01 4.68264133e-01 3.64628285e-01 8.30867708e-01 6.61035359e-01 -8.86014998e-01 -3.38900858e-03 1.71551988e-01 7.04097450e-01 3.99631113e-01 2.87215382e-01 4.47576232e-02 7.85091579e-01 -8.27934206e-01 -2.09338814e-01 5.02922773e-01 3.26012611e-01 -3.51612508e-01 -1.29984832e+00 -5.40415168e-01 -6.15166463e-02 -2.05104575e-01 -4.78022397e-01 -4.21722502e-01 5.21282852e-01 -1.91556543e-01 1.28858125e+00 5.12277940e-03 -6.40427113e-01 3.79987150e-01 2.66376793e-01 3.47592384e-02 -7.60636210e-01 -4.19063151e-01 4.79335159e-01 1.04411110e-01 -1.26114607e-01 -6.04772754e-02 -4.55745041e-01 -1.37893724e+00 -2.13289410e-01 -2.96438932e-01 6.49909079e-01 7.15133965e-01 1.05254304e+00 1.91800743e-02 5.47564805e-01 9.87611055e-01 -6.27566874e-01 -1.02380824e+00 -1.30733335e+00 -4.23540801e-01 -8.74429345e-02 7.94801891e-01 7.59741012e-03 -3.46222967e-01 3.75394851e-01]
[14.5374755859375, 7.108080863952637]
ce974d44-c0b2-4448-a03d-2679d7f9f3e2
gispy-a-tool-for-measuring-gist-inference
2205.12484
null
https://arxiv.org/abs/2205.12484v1
https://arxiv.org/pdf/2205.12484v1.pdf
GisPy: A Tool for Measuring Gist Inference Score in Text
Decision making theories such as Fuzzy-Trace Theory (FTT) suggest that individuals tend to rely on gist, or bottom-line meaning, in the text when making decisions. In this work, we delineate the process of developing GisPy, an open-source tool in Python for measuring the Gist Inference Score (GIS) in text. Evaluation of GisPy on documents in three benchmarks from the news and scientific text domains demonstrates that scores generated by our tool significantly distinguish low vs. high gist documents. Our tool is publicly available to use at: https://github.com/phosseini/GisPy.
['David A. Broniatowski', 'Mona Diab', 'Christopher R. Wolfe', 'Pedram Hosseini']
2022-05-25
null
https://aclanthology.org/2022.wnu-1.5
https://aclanthology.org/2022.wnu-1.5.pdf
naacl-wnu-2022-7
['coherence-evaluation']
['natural-language-processing']
[-2.89942116e-01 8.25092345e-02 -2.86274523e-01 -7.17625737e-01 -2.17207730e-01 -5.99814117e-01 9.36157167e-01 6.23417735e-01 -1.80731177e-01 4.95669127e-01 5.95991075e-01 -8.44379485e-01 -6.20015502e-01 -1.15482426e+00 -1.71022505e-01 1.56536803e-01 4.12884951e-01 4.45068151e-01 -2.27095246e-01 -1.56170309e-01 1.11457562e+00 -2.08895266e-01 -1.42940247e+00 5.14351726e-01 1.34436202e+00 8.51919770e-01 -3.70103747e-01 4.47286636e-01 -4.69569027e-01 1.07458770e+00 -5.55772305e-01 -8.06348860e-01 1.57995239e-01 -3.19670171e-01 -7.44548082e-01 -5.09623647e-01 4.26348269e-01 -3.89789492e-01 -2.19886482e-01 1.09493887e+00 1.92957178e-01 1.42313808e-01 9.46340919e-01 -1.11880302e+00 -8.74291718e-01 1.32341671e+00 -4.34768379e-01 3.29466164e-01 9.44283247e-01 1.76691249e-01 1.04998386e+00 -9.19463933e-01 7.33967662e-01 1.53817582e+00 9.57628191e-01 -4.58929315e-02 -1.28842890e+00 -6.41171336e-01 1.26842503e-02 1.66497231e-01 -1.60023999e+00 -2.04488039e-01 2.99041480e-01 -9.23770607e-01 1.06581259e+00 6.01816833e-01 5.86146951e-01 1.01519752e+00 6.08831882e-01 5.53766727e-01 1.87073970e+00 -4.41101730e-01 4.82100010e-01 6.51792288e-02 8.15878570e-01 4.98722225e-01 6.94768667e-01 6.00272007e-02 -7.15424836e-01 -2.22157910e-01 4.32342350e-01 7.33533502e-02 2.99950510e-01 9.82521355e-01 -1.02742016e+00 8.71628046e-01 5.66047765e-02 5.19405723e-01 -5.32693088e-01 -2.63733491e-02 -4.26840447e-02 4.40266997e-01 7.37223923e-01 2.61448562e-01 2.75387838e-02 -5.91696560e-01 -1.17272973e+00 6.53529525e-01 1.27495146e+00 6.98453009e-01 4.65905160e-01 -4.94749635e-01 -7.40657628e-01 5.83760619e-01 5.70575595e-01 3.43336254e-01 2.31089756e-01 -1.21620822e+00 3.88508111e-01 9.56220210e-01 1.00617170e-01 -1.46107662e+00 -1.98559389e-01 -1.62723854e-01 -3.97324443e-01 2.48069495e-01 6.42688572e-01 -3.98555130e-01 -4.55534309e-01 9.58554804e-01 8.80229920e-02 -2.47892365e-01 -2.16323063e-01 6.64338291e-01 8.95309567e-01 6.04618371e-01 2.49402165e-01 4.82537031e-01 1.24435079e+00 -2.20113024e-01 -9.58261132e-01 -1.17053591e-01 5.84912062e-01 -8.51730168e-01 1.00291419e+00 6.75515711e-01 -1.07432377e+00 -2.27414265e-01 -8.78885031e-01 -2.26523101e-01 -6.59829438e-01 -5.53290583e-02 6.58767462e-01 7.61606574e-01 -1.05913305e+00 8.89656603e-01 -5.17855763e-01 -5.42308807e-01 5.92033565e-01 -1.79095879e-01 2.61388510e-01 -9.57602113e-02 -1.23989058e+00 9.31812406e-01 3.50466669e-01 -3.62211853e-01 -2.30082735e-01 -7.14631200e-01 -5.33309937e-01 3.91675718e-02 4.50175405e-01 -7.02512443e-01 1.40499485e+00 -6.23063564e-01 -1.22805154e+00 7.00852275e-01 1.15176879e-01 -3.92915815e-01 9.89331663e-01 -1.91938162e-01 -4.45616931e-01 -1.13585768e-02 4.83329594e-01 1.94165871e-01 2.74659067e-01 -6.69829905e-01 -8.43247354e-01 -3.99457216e-01 -1.45178260e-02 9.53931212e-02 -9.74593982e-02 3.56840163e-01 1.02328777e-01 -7.72879720e-01 -2.12170273e-01 -4.71658826e-01 1.40800074e-01 -1.58240795e-02 -6.73259676e-01 -8.77884448e-01 1.31003082e-01 -9.30832386e-01 1.82962644e+00 -1.76515484e+00 -6.08382225e-01 7.02701688e-01 4.19619948e-01 -2.54056573e-01 7.05852091e-01 9.71427679e-01 4.60257620e-01 8.05126548e-01 -1.04218908e-02 1.49889901e-01 7.40399063e-01 -1.67805180e-02 -1.54634163e-01 2.21771657e-01 -1.54779717e-01 6.10217452e-01 -8.49953830e-01 -7.80746758e-01 2.60508239e-01 2.66024768e-01 -2.83991992e-01 -5.68740845e-01 -1.98088512e-01 2.93691233e-02 -7.71579683e-01 7.46549010e-01 5.67126215e-01 -3.49235445e-01 1.79287329e-01 6.64353728e-01 -8.03810060e-01 8.58082891e-01 -1.08879220e+00 1.42203522e+00 1.38898522e-01 7.59603679e-01 -3.14019591e-01 -5.31087756e-01 9.02704477e-01 3.18292230e-02 -1.89376101e-02 -6.33142829e-01 4.56672698e-01 4.47563827e-02 4.52946424e-02 -4.42421645e-01 4.98766363e-01 1.38767272e-01 -2.54718661e-01 9.59201634e-01 -3.02931577e-01 -1.85517058e-01 5.77694476e-01 4.58166867e-01 1.02252960e+00 5.76572565e-05 4.50294077e-01 -9.15548325e-01 1.72589317e-01 3.53230715e-01 2.88669318e-01 9.30712879e-01 7.94405043e-02 4.34427261e-02 8.10281754e-01 -1.96076453e-01 -8.18598449e-01 -9.38539565e-01 -7.56541193e-01 1.15765333e+00 -3.09399426e-01 -9.20876980e-01 -9.84577417e-01 -9.45950821e-02 5.85701168e-01 1.65429831e+00 -7.29361951e-01 3.55363756e-01 1.52543813e-01 -4.74981040e-01 7.39865720e-01 4.36299026e-01 6.60420179e-01 -5.71178377e-01 -6.80125356e-01 1.24256283e-01 1.68536566e-02 -6.57752931e-01 -2.13244632e-01 -4.18166578e-01 -5.86716175e-01 -7.97302723e-01 -2.03132793e-01 -1.37326075e-02 4.95257556e-01 -8.40094015e-02 1.14121532e+00 3.91210318e-01 -1.00698769e-01 5.70222735e-01 -4.26810145e-01 -8.75271320e-01 -2.78485715e-01 -1.62329525e-01 -2.17829525e-01 -4.54379141e-01 1.06815481e+00 -1.20028742e-01 -4.98180717e-01 2.78453499e-01 -6.67266607e-01 2.91758299e-01 -2.19364818e-02 1.92120641e-01 2.01205090e-01 3.27317387e-01 1.87085390e-01 -9.20430481e-01 1.28892183e+00 -8.06821883e-01 -5.35776377e-01 1.60524204e-01 -1.09301114e+00 -2.54396439e-01 1.09175611e-02 2.14548036e-01 -1.06363237e+00 -7.55529046e-01 -1.61032155e-01 5.23573577e-01 -1.71333462e-01 1.16878808e+00 6.17954731e-01 1.27439663e-01 9.01700318e-01 -2.23989785e-01 5.49476929e-02 -2.31904551e-01 3.79357897e-02 1.02980316e+00 5.26555553e-02 -9.14318621e-01 3.52096409e-01 2.62926280e-01 -4.84542131e-01 -4.84924585e-01 -8.32106352e-01 -1.07373923e-01 -2.13865891e-01 -6.29162073e-01 6.90892756e-01 -6.51943207e-01 -1.00847042e+00 2.14552492e-01 -6.85783923e-01 -4.92369920e-01 -9.25039202e-02 3.57212722e-01 -1.86592758e-01 2.56556761e-03 -5.43221533e-01 -9.57773447e-01 -3.93082470e-01 -5.44522226e-01 4.01264757e-01 5.43160737e-01 -1.04664552e+00 -1.35820043e+00 -8.61315727e-02 6.06300652e-01 3.78011525e-01 5.01604974e-01 4.57793832e-01 -8.70581210e-01 -1.98448047e-01 -3.24954659e-01 -1.36406049e-01 -4.98639978e-02 -3.10454577e-01 6.95131004e-01 -7.50260592e-01 3.02174509e-01 -7.09450468e-02 3.91567945e-02 7.03490138e-01 5.15008509e-01 9.40359116e-01 -5.44845104e-01 -4.38088685e-01 1.87534243e-01 1.24118912e+00 2.67991181e-02 4.67268378e-01 6.10000372e-01 1.10785961e-01 1.03180492e+00 6.77442193e-01 1.00781715e+00 8.75055969e-01 1.62806377e-01 -3.86481375e-01 5.02511978e-01 2.04246461e-01 -5.18175066e-01 3.32234085e-01 5.08534431e-01 -2.78285146e-01 -2.31086522e-01 -1.84482062e+00 2.61669993e-01 -1.94728565e+00 -9.86364722e-01 -4.87615794e-01 1.89556944e+00 8.03602397e-01 5.07345974e-01 8.93423110e-02 2.29220405e-01 7.37025321e-01 -4.25332814e-01 -3.88716906e-01 -6.59556627e-01 -1.69150636e-01 1.04564011e-01 3.14137429e-01 5.75562596e-01 -6.62240326e-01 6.65110588e-01 6.77022409e+00 7.21183002e-01 -8.03239524e-01 -1.18101332e-02 8.33407581e-01 -4.37057726e-02 -7.27592289e-01 9.94114876e-02 -7.54128754e-01 7.76340127e-01 1.28244007e+00 -1.13895524e+00 1.67330444e-01 5.82141459e-01 4.57011968e-01 -4.33386236e-01 -9.14275765e-01 5.29233932e-01 -2.84927368e-01 -1.40372252e+00 -1.79272145e-02 3.20485353e-01 7.50494719e-01 -4.09011543e-02 3.00451219e-01 1.07098907e-01 9.73345816e-01 -1.16901076e+00 1.17368722e+00 8.87117982e-01 6.82949960e-01 -4.75921571e-01 6.36183739e-01 1.93806022e-01 -5.98583996e-01 -8.10533464e-02 -2.79755384e-01 -7.48366117e-01 -3.81820112e-01 9.65822041e-01 -9.87359881e-01 5.20489037e-01 9.10014033e-01 1.05244517e+00 -7.71251917e-01 7.82876968e-01 -3.55271012e-01 9.69758987e-01 -3.09302658e-01 -3.89138252e-01 8.08518305e-02 -4.81636792e-01 4.40348327e-01 1.46459389e+00 5.51046491e-01 2.94364184e-01 -1.11085502e-02 1.39469457e+00 2.22058982e-01 1.55614227e-01 -5.15256107e-01 -6.62483633e-01 9.07713890e-01 9.82721925e-01 -9.94259655e-01 -6.07139647e-01 -5.45167625e-01 2.42258638e-01 -1.90098971e-01 2.43423596e-01 -4.12952751e-01 -4.74997371e-01 6.31383896e-01 4.90114629e-01 -1.15552448e-01 -1.89241156e-01 -1.44579947e+00 -9.27126110e-01 -3.24712962e-01 -7.78814912e-01 6.24919891e-01 -7.87537575e-01 -1.52267802e+00 5.00639491e-02 1.81059450e-01 -9.02851939e-01 -3.46267730e-01 -6.65497780e-01 -8.11824083e-01 9.68592763e-01 -9.49511528e-01 -6.59861743e-01 -5.47440827e-01 5.59675634e-01 1.38183951e-01 8.01648200e-02 4.90189642e-01 4.82845232e-02 -5.85744858e-01 3.42710525e-01 5.48752360e-02 2.25827530e-01 6.88076794e-01 -1.49222505e+00 7.29779303e-01 5.24953187e-01 -3.64708781e-01 1.18417072e+00 9.39225197e-01 -1.01917279e+00 -1.11244726e+00 -5.35260558e-01 1.09990776e+00 -8.96904647e-01 9.81809199e-01 -8.57217051e-03 -8.53565216e-01 8.65673780e-01 6.03867829e-01 -1.03989232e+00 1.35821557e+00 2.94798166e-01 -1.93317682e-01 2.16454253e-01 -1.35709369e+00 6.46279693e-01 8.77247810e-01 -3.60629857e-01 -1.10221779e+00 2.15469956e-01 3.48586924e-02 -2.19298065e-01 -1.35062456e+00 -2.76881963e-01 7.36590803e-01 -1.09171081e+00 6.30146742e-01 6.85598925e-02 6.20736361e-01 -1.68930635e-01 5.80447987e-02 -1.11086643e+00 -5.33425927e-01 -7.11657703e-01 4.51702803e-01 1.29021335e+00 5.65677524e-01 -1.17323816e+00 1.73917174e-01 1.12990475e+00 -1.14408448e-01 -2.75438786e-01 -7.43591845e-01 -4.51213568e-01 4.45123732e-01 -7.45421171e-01 8.77565503e-01 1.24300003e+00 5.02450347e-01 -9.29713920e-02 4.57030773e-01 -3.47609431e-01 7.06169546e-01 -1.01419687e-01 3.73337418e-01 -1.89679015e+00 1.38448209e-01 -9.44182158e-01 -1.34633914e-01 -4.13878769e-01 -6.75199851e-02 -1.21467090e+00 -4.83952194e-01 -2.14595151e+00 1.84633341e-02 -2.51538336e-01 -2.46271729e-01 7.16672063e-01 -9.86568723e-03 -5.74480705e-02 1.69183329e-01 3.25677723e-01 -4.15277749e-01 -1.24948516e-01 1.07866478e+00 -8.39668214e-02 4.46857624e-02 -3.55389655e-01 -1.41711080e+00 8.38984489e-01 1.36440349e+00 -5.09530127e-01 -9.60770547e-02 -3.28162283e-01 8.00813079e-01 -1.87267870e-01 5.04525125e-01 -1.12344348e+00 4.50270057e-01 -4.75303441e-01 6.80449486e-01 -5.50918639e-01 -1.94925025e-01 -3.12707990e-01 1.44083560e-01 5.33427775e-01 -5.11393666e-01 1.61854282e-01 2.17467159e-01 2.43342482e-02 1.44320130e-01 -3.82137269e-01 7.12587163e-02 -1.26167014e-01 -3.49633604e-01 -2.19294146e-01 -7.59474576e-01 1.69944495e-03 8.63281369e-01 -4.05791342e-01 -8.69065046e-01 -3.76542062e-01 -3.80187452e-01 5.23294270e-01 5.86009622e-01 2.41261050e-01 4.86731231e-01 -1.02203822e+00 -1.13674462e+00 -1.70728303e-02 1.45391189e-02 -3.99285465e-01 7.43787810e-02 1.03030908e+00 -8.93973649e-01 5.42007983e-01 -4.15300429e-01 -1.21451750e-01 -8.41947138e-01 3.31390053e-02 6.27876371e-02 1.76868811e-01 -7.11931884e-01 5.78419924e-01 -2.91049510e-01 -3.12412411e-01 -5.48619218e-02 -6.70854747e-01 -2.90754437e-01 2.85768747e-01 9.90314305e-01 9.51977193e-01 -2.12236568e-01 -2.28124470e-01 -2.61927783e-01 2.61654168e-01 -4.93317842e-02 -5.26286244e-01 1.11559129e+00 -1.24376737e-01 -4.99386162e-01 8.48272979e-01 2.82682270e-01 -1.52998520e-02 -8.54789197e-01 -8.59693065e-02 4.88868713e-01 -7.67836571e-01 4.03517455e-01 -1.32589209e+00 -7.11149871e-02 4.55112129e-01 9.66382921e-02 7.74932921e-01 6.15872502e-01 -1.88369051e-01 3.32376391e-01 3.17098379e-01 2.57534593e-01 -1.74803472e+00 -4.78686959e-01 4.64612126e-01 1.07975984e+00 -7.82374859e-01 2.25727096e-01 -3.51687342e-01 -9.55154300e-01 1.15440059e+00 2.26874784e-01 -1.56660736e-01 9.75144684e-01 3.85861188e-01 1.71278119e-02 -2.61941761e-01 -1.00742710e+00 -3.61928828e-02 3.51859123e-01 3.66290957e-01 8.67541254e-01 5.46634495e-01 -9.17573392e-01 8.66550446e-01 -9.37866628e-01 6.53758109e-01 8.25741291e-01 1.09523833e+00 -4.05217886e-01 -8.18711221e-01 -7.47557342e-01 9.45804179e-01 -7.70854831e-01 -3.81231338e-01 -1.00504637e+00 4.64598209e-01 -1.15263080e-02 1.46030104e+00 4.25205737e-01 -5.79670072e-01 9.61653665e-02 7.87110925e-02 1.68515310e-01 -3.76913577e-01 -9.23119545e-01 2.78743859e-02 4.71749127e-01 -3.86391580e-01 -1.42575383e-01 -1.10346210e+00 -1.42844427e+00 -1.37108517e+00 4.44442272e-01 1.40992314e-01 6.34875894e-01 7.48616576e-01 4.99570191e-01 3.50844860e-01 -1.77053407e-01 2.80522816e-02 -3.33151102e-01 -1.23736477e+00 -6.50962532e-01 5.54932700e-03 -2.05875695e-01 -6.95733368e-01 -3.11294317e-01 -2.40498587e-01]
[9.755837440490723, 7.871340751647949]
6a2eb5cc-3b38-4968-b65b-814dde2cf01c
6-dof-pose-estimation-of-household-objects
2203.05701
null
https://arxiv.org/abs/2203.05701v2
https://arxiv.org/pdf/2203.05701v2.pdf
6-DoF Pose Estimation of Household Objects for Robotic Manipulation: An Accessible Dataset and Benchmark
We present a new dataset for 6-DoF pose estimation of known objects, with a focus on robotic manipulation research. We propose a set of toy grocery objects, whose physical instantiations are readily available for purchase and are appropriately sized for robotic grasping and manipulation. We provide 3D scanned textured models of these objects, suitable for generating synthetic training data, as well as RGBD images of the objects in challenging, cluttered scenes exhibiting partial occlusion, extreme lighting variations, multiple instances per image, and a large variety of poses. Using semi-automated RGBD-to-model texture correspondences, the images are annotated with ground truth poses accurate within a few millimeters. We also propose a new pose evaluation metric called ADD-H based on the Hungarian assignment algorithm that is robust to symmetries in object geometry without requiring their explicit enumeration. We share pre-trained pose estimators for all the toy grocery objects, along with their baseline performance on both validation and test sets. We offer this dataset to the community to help connect the efforts of computer vision researchers with the needs of roboticists.
['Stan Birchfield', 'Jeffrey Smith', 'Terry Mosier', 'Jia Cheng', 'Thang To', 'Jonathan Tremblay', 'Stephen Tyree']
2022-03-11
null
null
null
null
['robotic-grasping']
['robots']
[-4.92957607e-02 1.16667233e-01 1.56364255e-02 -4.81185108e-01 -7.92170644e-01 -8.68193448e-01 2.36616313e-01 -3.26876253e-01 6.66293800e-02 3.20354074e-01 -3.29365104e-01 1.38335332e-01 -3.96024019e-01 -3.81211907e-01 -1.03890479e+00 -4.92065012e-01 -1.99212059e-01 1.31382120e+00 3.28045726e-01 -1.95425421e-01 3.02024454e-01 9.76852417e-01 -1.50173771e+00 1.62524760e-01 2.55858153e-01 1.29707968e+00 6.00686967e-01 4.17278707e-01 4.37199175e-01 4.52739179e-01 -3.63622487e-01 -3.69547188e-01 8.96125913e-01 2.49650180e-01 -6.45244062e-01 5.82357049e-01 5.50766885e-01 -7.50503600e-01 -6.48515105e-01 5.79850376e-01 1.85259447e-01 1.22400016e-01 6.99881077e-01 -1.49647295e+00 -4.08008903e-01 3.23089808e-01 -4.85003799e-01 -6.31714582e-01 6.30495310e-01 2.14259312e-01 8.04070354e-01 -1.02349389e+00 1.10437918e+00 1.50606799e+00 6.17971063e-01 4.60478008e-01 -8.48468781e-01 -3.34626496e-01 -1.77603126e-01 1.41505823e-01 -1.17431414e+00 -2.96010107e-01 8.71927202e-01 -5.56325376e-01 6.04318082e-01 1.57964066e-01 5.97858131e-01 1.28276289e+00 6.08143434e-02 6.48035228e-01 9.83640790e-01 -4.04811114e-01 2.81681865e-01 -9.19823274e-02 -4.05028731e-01 7.53286660e-01 2.67561048e-01 -1.09415479e-01 -5.45176983e-01 -1.52854353e-01 1.39794469e+00 1.88002557e-01 8.04866105e-02 -1.62260067e+00 -1.72991502e+00 3.99815768e-01 5.36135495e-01 -4.49738055e-01 -3.76717657e-01 4.76259261e-01 -2.44897529e-02 1.51759917e-02 1.99469253e-02 7.10592330e-01 -6.17034853e-01 -8.96381810e-02 6.85299933e-02 6.71005189e-01 9.31366265e-01 2.09994984e+00 5.67785800e-01 -3.65961492e-01 2.28741482e-01 7.13241696e-01 3.00894320e-01 8.70408416e-01 -1.23438932e-01 -1.57868099e+00 6.52683318e-01 3.95271540e-01 7.57838845e-01 -1.03046608e+00 -4.56800491e-01 1.70868263e-01 -1.42281666e-01 1.32718295e-01 6.55040085e-01 4.44829285e-01 -8.23073149e-01 1.15793931e+00 6.97570205e-01 -6.08237386e-01 -2.81111211e-01 1.16767168e+00 5.24586797e-01 3.59826535e-02 -5.96692920e-01 2.73866624e-01 1.16865003e+00 -9.99884844e-01 -3.18478912e-01 -1.65779099e-01 2.32136071e-01 -9.99268115e-01 1.25533104e+00 6.21332288e-01 -9.82191980e-01 -1.95311755e-01 -1.02120352e+00 -2.77425230e-01 -1.53402403e-01 3.96581203e-01 9.99022126e-01 1.68299302e-01 -4.83074754e-01 7.14724779e-01 -1.17945707e+00 -4.16916907e-01 3.43059242e-01 3.08865368e-01 -6.94868863e-01 -4.59339231e-01 -2.12883547e-01 1.37535739e+00 2.57567644e-01 2.92739511e-01 -1.11472070e+00 -3.31544131e-01 -7.36171544e-01 -5.93966305e-01 6.46648407e-01 -2.94309169e-01 1.51670611e+00 -1.75715327e-01 -1.51675379e+00 1.15543699e+00 4.17575479e-01 1.76599532e-01 7.47954190e-01 -3.25766236e-01 3.49820524e-01 5.14354527e-01 4.84939143e-02 8.37117612e-01 9.36799228e-01 -1.60303509e+00 4.70872410e-02 -7.39423633e-01 3.29871893e-01 2.86866099e-01 1.51525095e-01 -1.38091087e-01 -4.83609498e-01 -5.18257797e-01 1.02858329e+00 -1.25748837e+00 -2.11001337e-01 8.23574007e-01 -4.15653586e-01 7.13721514e-02 9.90182459e-01 -7.21291542e-01 -2.22226888e-01 -1.89872098e+00 4.90772098e-01 2.59363472e-01 -9.17883962e-02 -5.17874658e-01 -3.12161654e-01 4.18059468e-01 3.79179239e-01 -5.16182721e-01 1.77365765e-01 -1.61964729e-01 4.07000512e-01 4.58385140e-01 -3.23988050e-01 7.48786628e-01 1.39461964e-01 8.67920637e-01 -8.84838104e-01 -2.75314629e-01 2.51723111e-01 3.53955142e-02 -5.58019519e-01 4.26442772e-01 -5.10520577e-01 5.45805097e-01 -6.85972810e-01 1.32917488e+00 6.62175655e-01 8.83694887e-02 1.09022737e-01 -3.50799739e-01 2.51952738e-01 1.68964475e-01 -1.25009477e+00 2.13376594e+00 -3.05517882e-01 1.31730512e-01 2.57186800e-01 -6.68402433e-01 1.18878233e+00 4.04112786e-02 5.98714828e-01 -1.36508167e-01 3.01350862e-01 4.63479608e-01 -1.16766416e-01 -7.07994580e-01 4.86885041e-01 2.40549251e-01 -2.75167078e-01 4.09248114e-01 2.24145085e-01 -1.29307652e+00 3.38991247e-02 -5.13365045e-02 9.71504152e-01 8.10346782e-01 -7.97357932e-02 -1.98663563e-01 -3.81126374e-01 4.36070174e-01 1.88263357e-01 4.57615405e-01 1.28265068e-01 9.36288118e-01 2.65802383e-01 -7.05758750e-01 -1.78763056e+00 -1.25973761e+00 -1.50281429e-01 7.18876481e-01 5.58388948e-01 3.32011352e-03 -6.42511427e-01 -1.31853580e-01 5.47759831e-01 1.17344484e-01 -5.38120687e-01 4.17575501e-02 -5.66610336e-01 -2.20962971e-01 -2.10728254e-02 6.64614975e-01 3.10698211e-01 -1.01646924e+00 -9.72266972e-01 -1.65424477e-02 -1.50112197e-01 -1.44704032e+00 -1.13648489e-01 1.02006167e-01 -9.39304829e-01 -1.34786236e+00 -6.23358846e-01 -8.19719851e-01 8.90787542e-01 4.86491889e-01 8.72097492e-01 -2.27130130e-01 -8.15761566e-01 9.05785084e-01 -7.01267958e-01 -5.17597795e-01 -3.19802195e-01 -1.99996531e-01 4.66396362e-01 -5.57928622e-01 -9.34370384e-02 -4.25662011e-01 -4.46885049e-01 8.95363867e-01 -5.55216372e-01 3.18779573e-02 5.31179190e-01 6.05626464e-01 7.50521958e-01 -5.67535937e-01 4.35335060e-05 -5.17245643e-02 2.86940448e-02 -1.08507685e-01 -7.64203250e-01 2.68872738e-01 2.44729474e-01 -2.34337360e-01 -1.32743090e-01 -7.84212112e-01 -7.08837748e-01 4.85390127e-01 5.39186358e-01 -9.55207109e-01 4.18108888e-02 -9.75534618e-02 -2.63999015e-01 -5.38596272e-01 7.36317635e-01 -4.12256032e-01 1.06272854e-01 -6.52708590e-01 3.88276249e-01 6.37842715e-01 7.80389488e-01 -1.18113756e+00 7.53304958e-01 4.77440327e-01 1.83487058e-01 -5.69962084e-01 -5.04376292e-01 -1.98259443e-01 -1.13565922e+00 -2.40575492e-01 4.10550565e-01 -7.70177484e-01 -8.73205125e-01 5.86069465e-01 -1.27377474e+00 -5.99528790e-01 -1.43023178e-01 7.02204227e-01 -1.33706939e+00 2.64969230e-01 -5.07103205e-01 -6.23446226e-01 1.11995213e-01 -1.37297797e+00 1.60102177e+00 -2.44657636e-01 -1.78368956e-01 -1.70423001e-01 -5.89304388e-01 6.76578939e-01 9.63671580e-02 6.52614534e-01 7.56753206e-01 -1.72815293e-01 -1.14184940e+00 -4.08820242e-01 -4.04807776e-02 2.53885221e-02 5.47743663e-02 4.98010032e-02 -7.46435106e-01 -3.43694508e-01 -3.57581489e-02 -8.83532405e-01 1.09658286e-01 3.92867029e-02 1.27990544e+00 -1.70949250e-01 -2.29050905e-01 4.14179921e-01 9.45547521e-01 -3.44366208e-02 4.21822876e-01 2.74540633e-01 6.99752092e-01 8.59695017e-01 1.20579863e+00 6.18150234e-01 2.81544954e-01 1.02337885e+00 1.00151432e+00 4.09574062e-01 6.02511503e-02 -2.00513393e-01 -4.88229319e-02 5.10181248e-01 -3.34899276e-01 4.40952592e-02 -9.88064170e-01 3.01958591e-01 -1.71539509e+00 -2.24612176e-01 -6.53001145e-02 2.22207832e+00 5.50655007e-01 -5.74626736e-02 4.27458622e-03 1.23329507e-03 3.73125196e-01 -4.18399394e-01 -8.21932554e-01 2.93907039e-02 4.74606566e-02 1.54674545e-01 7.21700251e-01 1.16074637e-01 -1.02627826e+00 9.23523068e-01 6.27145195e+00 4.42611337e-01 -7.41074502e-01 -5.89827895e-02 6.00976385e-02 -6.84140250e-02 -4.02065776e-02 -6.76508248e-02 -3.84092033e-01 -3.32340673e-02 7.69865187e-03 3.41027111e-01 8.29538584e-01 1.28990662e+00 -2.30208397e-01 -3.84036183e-01 -1.47998297e+00 1.14401877e+00 1.19428590e-01 -9.30065036e-01 -2.39242584e-01 -1.42788678e-01 8.53663445e-01 1.08728714e-01 -8.62584412e-02 -4.23645079e-02 1.91389814e-01 -7.42005408e-01 1.31899297e+00 4.71549928e-01 7.98938036e-01 -1.68829918e-01 3.99244756e-01 3.92749846e-01 -6.84872746e-01 -1.60132423e-01 -6.99302316e-01 -7.61309564e-02 -8.64214301e-02 8.56913775e-02 -1.01190150e+00 2.76107848e-01 8.96514893e-01 2.94855952e-01 -2.18762726e-01 9.42685068e-01 -1.73121825e-01 -2.66567349e-01 -6.65199816e-01 -2.44616166e-01 -2.74933428e-02 -1.78796396e-01 4.67652261e-01 2.99805343e-01 4.09451038e-01 2.27451235e-01 2.43804693e-01 9.83071029e-01 -2.06566285e-02 -1.63764670e-01 -6.05587065e-01 2.27617640e-02 6.14100039e-01 1.36468351e+00 -1.00877798e+00 5.25579266e-02 1.54724911e-01 8.58165383e-01 2.63460934e-01 6.96341842e-02 -6.23866558e-01 -3.28804076e-01 3.82986695e-01 2.06692055e-01 4.02486742e-01 -8.68404388e-01 -2.37188488e-01 -1.05467927e+00 7.05558419e-01 -7.51860738e-01 -4.79143500e-01 -1.42819226e+00 -1.14365530e+00 2.42217600e-01 6.02913439e-01 -1.39932871e+00 -3.22412103e-01 -1.32484686e+00 1.04017332e-01 5.01869917e-01 -6.91865385e-01 -1.38600183e+00 -7.54319251e-01 2.38245800e-01 4.72220927e-01 1.62908509e-02 9.29567754e-01 -1.97698712e-01 2.65556186e-01 1.32054582e-01 4.38754112e-02 -5.22765331e-02 5.78977644e-01 -8.00737977e-01 2.58763313e-01 6.68957978e-02 -1.39318937e-02 4.39268082e-01 7.55428255e-01 -4.05890405e-01 -2.29128408e+00 -7.42214441e-01 1.14008069e-01 -1.05769098e+00 5.04295409e-01 -9.44126487e-01 -4.41516757e-01 1.04679549e+00 -5.73698342e-01 3.11140507e-01 -3.11665803e-01 -2.26786613e-01 -2.85706311e-01 5.67886643e-02 -1.54494715e+00 3.12947214e-01 1.47181451e+00 -2.37036228e-01 -6.61575615e-01 9.89796162e-01 5.61569929e-01 -1.38784242e+00 -1.05931425e+00 7.51285970e-01 1.02681661e+00 -5.45947433e-01 1.07951498e+00 -5.35654068e-01 5.06545544e-01 -2.61896908e-01 -6.17831588e-01 -9.97180343e-01 -4.52966690e-02 -4.99601156e-01 4.06287909e-02 5.40082872e-01 1.09321684e-01 -2.66279131e-01 8.66909862e-01 9.02643681e-01 -2.80142605e-01 -7.54161358e-01 -8.99110377e-01 -1.15251517e+00 -2.80289382e-01 -2.96624452e-01 6.05364740e-01 6.78687334e-01 2.85569653e-02 -3.70884866e-01 -2.47675866e-01 2.38875061e-01 6.67120755e-01 4.92993742e-01 1.39251685e+00 -1.13748443e+00 -2.36569986e-01 6.23417012e-02 -7.63729990e-01 -1.10522914e+00 5.53719074e-05 -6.40279889e-01 5.00087142e-01 -1.27869689e+00 1.83239862e-01 -8.86623263e-01 5.89223206e-01 6.48661196e-01 5.99863470e-01 4.35025245e-01 1.57765284e-01 3.48487347e-01 -5.46013236e-01 6.38455927e-01 1.73102272e+00 -3.82203003e-03 2.11006999e-01 -1.58661410e-01 -7.13540427e-03 8.40852737e-01 4.73710567e-01 -2.87612081e-01 -1.37581378e-01 -8.34035218e-01 2.37031914e-02 2.05476835e-01 6.90072536e-01 -9.61790621e-01 -1.96014389e-01 -4.45583433e-01 4.81234461e-01 -6.53842032e-01 1.00212121e+00 -1.09290946e+00 3.82150650e-01 2.58313030e-01 -2.98573911e-01 -2.29943115e-02 4.55530472e-02 3.55598837e-01 3.25613648e-01 -2.19477326e-01 4.75939661e-01 -5.57561994e-01 -5.36396384e-01 4.89825189e-01 3.10441464e-01 -2.41579637e-01 1.32696629e+00 -2.36368656e-01 -5.67581356e-02 -4.65232551e-01 -7.64019072e-01 1.30409345e-01 9.94715929e-01 6.88395798e-01 7.08047092e-01 -1.45411134e+00 -4.52951491e-01 1.74279749e-01 4.31699127e-01 5.54183066e-01 -7.71471933e-02 5.66305995e-01 -1.09938335e+00 2.54584819e-01 -4.78008628e-01 -9.72244203e-01 -8.65944266e-01 4.78928864e-01 6.81324154e-02 6.21013045e-01 -5.52767396e-01 8.27716649e-01 -1.72228768e-01 -1.06246054e+00 3.95080775e-01 -8.13024938e-01 7.77127802e-01 -6.64199769e-01 4.75420356e-02 5.11198580e-01 1.62155628e-01 -5.05962908e-01 -2.03789383e-01 6.05884254e-01 3.75614464e-01 -3.08902556e-04 1.45280933e+00 1.38314366e-01 -1.80604398e-01 3.60927224e-01 9.34219658e-01 -3.20954531e-01 -1.64895785e+00 -1.38692901e-01 -1.23945624e-01 -8.71380925e-01 -6.49536550e-01 -6.53548002e-01 -7.96491504e-01 8.11837316e-01 2.80370891e-01 -4.04815257e-01 4.42407370e-01 5.47876060e-01 4.75619555e-01 1.21444988e+00 1.34775674e+00 -1.05852282e+00 5.24271131e-01 5.52086473e-01 1.71668005e+00 -1.18820083e+00 2.18413338e-01 -8.35573077e-01 -3.17768127e-01 1.29098833e+00 6.98606610e-01 -4.66678739e-01 1.99094653e-01 2.09675059e-01 -4.13219668e-02 -1.95450142e-01 -2.75601894e-01 3.57727855e-01 1.73638389e-01 8.40089381e-01 -3.06227297e-01 2.32419252e-01 2.55410850e-01 2.92440772e-01 -5.16046107e-01 -3.08709115e-01 3.11217844e-01 1.37771273e+00 -4.16902214e-01 -8.29890549e-01 -6.35519683e-01 2.53587961e-01 1.86126336e-01 5.16264141e-01 -4.97791976e-01 9.88680184e-01 -1.34999394e-01 5.86280465e-01 4.51301374e-02 -3.38906199e-01 5.52066386e-01 -2.21361190e-01 1.61588919e+00 -5.51609218e-01 1.45243213e-01 -1.25768244e-01 8.50622803e-02 -8.26200962e-01 -3.61891299e-01 -8.13121557e-01 -1.05533254e+00 4.47499938e-02 -4.69037324e-01 -3.91964793e-01 1.30304420e+00 7.60199070e-01 1.48324668e-01 -1.37969241e-01 6.34243906e-01 -1.90995383e+00 -1.07930458e+00 -1.21176147e+00 -7.49632299e-01 5.52105427e-01 -1.16513498e-01 -1.40561318e+00 -7.94686452e-02 -2.26221066e-02]
[5.934572219848633, -0.9830020070075989]
7375f39a-ed04-4d73-be43-f12418a73215
region-of-interest-detection-in-melanocytic
2210.16457
null
https://arxiv.org/abs/2210.16457v1
https://arxiv.org/pdf/2210.16457v1.pdf
Region of Interest Detection in Melanocytic Skin Tumor Whole Slide Images
Automated region of interest detection in histopathological image analysis is a challenging and important topic with tremendous potential impact on clinical practice. The deep-learning methods used in computational pathology help us to reduce costs and increase the speed and accuracy of regions of interest detection and cancer diagnosis. In this work, we propose a patch-based region of interest detection method for melanocytic skin tumor whole-slide images. We work with a dataset that contains 165 primary melanomas and nevi Hematoxylin and Eosin whole-slide images and build a deep-learning method. The proposed method performs well on a hold-out test data set including five TCGA-SKCM slides (accuracy of 93.94\% in slide classification task and intersection over union rate of 41.27\% in the region of interest detection task), showing the outstanding performance of our model on melanocytic skin tumor. Even though we test the experiments on the skin tumor dataset, our work could also be extended to other medical image detection problems, such as various tumors' classification and prediction, to help and benefit the clinical evaluation and diagnosis of different tumors.
['Nancy E. Thomas', 'J. S. Marron', 'Sherif Farag', 'Jayson R. Miedema', 'Yao Li', 'Yi Cui']
2022-10-29
null
null
null
null
['medical-image-detection']
['computer-vision']
[ 3.42276424e-01 -4.31163125e-02 -2.50196248e-01 -3.01898848e-02 -1.22725272e+00 -3.03452790e-01 2.47922704e-01 4.88578796e-01 -6.55428112e-01 5.81562400e-01 -3.11645269e-01 -4.80635434e-01 2.91759539e-02 -7.90302634e-01 -1.51484430e-01 -1.43230546e+00 4.88254279e-02 4.10426944e-01 3.53448749e-01 4.16148342e-02 2.84090579e-01 7.80338645e-01 -9.30554092e-01 3.95224899e-01 8.13907027e-01 9.61674273e-01 2.83359081e-01 8.03784549e-01 4.99983411e-03 8.18861961e-01 -4.03234661e-01 -1.22534089e-01 -2.91356212e-03 -1.82271391e-01 -7.52770364e-01 1.19149059e-01 3.90245825e-01 -1.83039755e-01 -3.11255977e-02 1.09036791e+00 7.04885483e-01 -4.15515661e-01 8.33273947e-01 -6.96657181e-01 9.60310549e-02 2.65716076e-01 -1.36597967e+00 3.83593082e-01 -3.43769252e-01 8.46458226e-03 7.42371321e-01 -6.56702697e-01 9.15213704e-01 3.31592798e-01 8.71815741e-01 3.55901986e-01 -7.68582582e-01 -6.31058276e-01 -4.86966729e-01 3.38212788e-01 -1.64841974e+00 -1.11110568e-01 3.76498818e-01 -4.32905108e-01 5.82803965e-01 4.06899422e-01 6.98471785e-01 4.59666520e-01 6.89905941e-01 1.10284984e+00 1.12796199e+00 -5.86696744e-01 1.96954533e-01 2.08176211e-01 1.49154495e-02 1.00457466e+00 1.12698153e-01 -3.91527295e-01 2.36391854e-02 -1.07128076e-01 8.81693959e-01 1.98708400e-01 -9.16641280e-02 -4.50949557e-02 -1.05277836e+00 8.31815898e-01 7.00628698e-01 5.89405239e-01 -1.44568339e-01 3.72326449e-02 5.63589573e-01 -8.10076892e-02 5.52559793e-01 -2.00277362e-02 9.83841717e-02 2.17293799e-01 -9.85489845e-01 -8.21687877e-02 2.61124760e-01 2.48718001e-02 2.69085556e-01 -4.45856303e-01 -3.15763839e-02 7.99269557e-01 1.24096140e-01 2.91831762e-01 9.49320436e-01 -2.11511418e-01 -7.43839368e-02 8.95559072e-01 -2.71401763e-01 -5.81712902e-01 -8.69905710e-01 -6.48152053e-01 -1.20763695e+00 4.87282008e-01 9.06794906e-01 -4.37301509e-02 -1.01587927e+00 9.62837934e-01 6.47015631e-01 1.26527160e-01 -8.83105770e-02 9.47794855e-01 7.68897116e-01 3.12535524e-01 2.10695297e-01 -9.90001857e-02 1.70329452e+00 -7.50248551e-01 -4.14820820e-01 4.02375638e-01 1.39154124e+00 -6.65104866e-01 7.87229955e-01 2.27504313e-01 -6.96022391e-01 -1.73090219e-01 -7.13421226e-01 -7.19113201e-02 -3.62283230e-01 7.26742983e-01 8.24673891e-01 3.73420447e-01 -1.23227215e+00 7.76683018e-02 -8.97784531e-01 -8.32345068e-01 8.31073642e-01 5.74245036e-01 -6.30872309e-01 3.48998159e-02 -6.40398741e-01 7.02587366e-01 1.73832372e-01 -8.02706107e-02 -9.27660584e-01 -7.97290266e-01 -4.36041951e-01 -8.60721394e-02 1.19504541e-01 -5.64724326e-01 1.05864787e+00 -9.39961493e-01 -1.06894159e+00 1.33107710e+00 -1.93559662e-01 -5.03029644e-01 3.50444347e-01 6.60532773e-01 -2.65346080e-01 4.59060848e-01 -3.16036819e-03 6.67965770e-01 3.09596837e-01 -6.95718825e-01 -1.12114990e+00 -5.79246879e-01 -2.34057859e-01 1.98247939e-01 -3.46491069e-01 -2.80017763e-01 -1.81038097e-01 -4.11835194e-01 -1.53335750e-01 -9.14419889e-01 -5.57725489e-01 4.12999779e-01 -5.35314322e-01 -2.56300241e-01 7.81024814e-01 -8.28720391e-01 9.66172040e-01 -2.16460609e+00 -2.89514631e-01 4.18096125e-01 3.17706466e-01 3.63209605e-01 -3.74998748e-02 1.07716054e-01 2.49815583e-02 1.89699978e-01 -8.30547139e-02 1.04160309e-02 -3.80666435e-01 -3.69684547e-01 3.99587244e-01 9.57691073e-01 1.79800510e-01 9.62874532e-01 -5.33501267e-01 -1.08894157e+00 2.08255187e-01 5.15935779e-01 -1.00125208e-01 -3.72649170e-02 1.71002358e-01 1.61268666e-01 -4.86866176e-01 1.00743175e+00 5.92725873e-01 -5.05305886e-01 5.32926284e-02 -1.80693373e-01 6.09638058e-02 -4.83268470e-01 -6.76740468e-01 1.39038098e+00 -3.01752180e-01 1.03886163e+00 2.34879643e-01 -5.85631013e-01 4.73253191e-01 2.24316671e-01 7.56925762e-01 -6.45905316e-01 3.06429654e-01 1.04638964e-01 2.11581156e-01 -8.49873900e-01 1.67401254e-01 -4.00723130e-01 3.50571573e-01 3.49293381e-01 -3.67223233e-01 9.19321328e-02 1.70719475e-01 9.82219428e-02 1.35369492e+00 -5.26334047e-01 8.89827132e-01 -2.92759657e-01 6.99494004e-01 4.14606601e-01 3.10291260e-01 1.39353648e-01 -4.67715591e-01 4.63408470e-01 5.60329735e-01 -5.80491483e-01 -8.66721928e-01 -7.21085548e-01 -6.41628206e-01 8.37649345e-01 -8.96379352e-02 2.91495651e-01 -5.79106569e-01 -9.58449483e-01 5.91214672e-02 1.33838952e-01 -1.02069461e+00 3.48929614e-01 -3.59225243e-01 -1.21517622e+00 5.75359046e-01 5.18736720e-01 4.27495450e-01 -8.37809324e-01 -4.61285353e-01 1.45205930e-01 1.16608432e-02 -7.40218043e-01 -1.09188884e-01 1.58645675e-01 -7.40499794e-01 -1.60654461e+00 -1.07002044e+00 -1.21671414e+00 1.20153463e+00 3.10334772e-01 6.98478580e-01 8.09784979e-02 -1.29731357e+00 -4.78161089e-02 -1.99251577e-01 -5.75082481e-01 -6.25899196e-01 -1.09787896e-01 -5.16720414e-01 4.03539799e-02 5.40192306e-01 -3.08433315e-03 -8.84591162e-01 2.50434071e-01 -8.18597317e-01 8.20647478e-02 1.01529956e+00 1.11375165e+00 9.28226054e-01 2.94262081e-01 5.25645196e-01 -1.04375029e+00 3.29467833e-01 -3.72943401e-01 -4.85551864e-01 3.93991739e-01 -5.44135198e-02 -5.47140002e-01 5.11983931e-01 -1.67126253e-01 -8.32331538e-01 2.38535866e-01 -3.36066812e-01 1.39433786e-01 -2.27175534e-01 5.33048570e-01 3.33969533e-01 -5.57669699e-01 7.15224802e-01 2.84441292e-01 4.71849054e-01 4.79086973e-02 -2.58404821e-01 6.98363721e-01 1.90394700e-01 2.99647391e-01 3.41981024e-01 9.51762140e-01 4.94316190e-01 -8.78910124e-01 -3.77847701e-01 -9.36843991e-01 -4.20520127e-01 -2.19502524e-01 7.59070873e-01 -1.01042950e+00 -8.28782618e-01 5.87361634e-01 -5.52772462e-01 -3.09906363e-01 -1.21374190e-01 3.09493959e-01 -1.64196953e-01 4.21315730e-01 -1.03103125e+00 -5.56141973e-01 -6.67286754e-01 -1.06553078e+00 1.43101501e+00 5.65367222e-01 -1.12927184e-02 -1.44926167e+00 3.10781091e-01 2.69475579e-01 4.01192546e-01 5.94269812e-01 8.61438632e-01 -6.37092113e-01 -1.56428710e-01 -7.41082191e-01 -3.94313097e-01 -1.08818104e-03 2.53734976e-01 5.79068422e-01 -1.09409249e+00 -5.25132656e-01 -3.55341464e-01 -3.22916120e-01 1.15399921e+00 7.07131028e-01 1.29979718e+00 1.41357649e-02 -1.09000635e+00 5.90643823e-01 1.94284022e+00 2.60426342e-01 5.40325999e-01 2.78853953e-01 5.91081381e-01 6.15309596e-01 7.39631534e-01 3.51286054e-01 1.08633332e-01 2.52653033e-01 6.80631340e-01 -8.78771961e-01 -2.19018131e-01 1.64759353e-01 -1.72541425e-01 2.68436015e-01 -1.07869923e-01 -3.75239044e-01 -1.08660674e+00 9.15851772e-01 -1.38320148e+00 -7.66815245e-01 -2.91557074e-01 1.85640824e+00 7.11555481e-01 -2.89341420e-01 6.46910220e-02 3.10185015e-01 9.27921355e-01 -2.31769741e-01 -6.26626253e-01 -7.61412904e-02 -3.89309190e-02 1.66483223e-01 4.19774354e-01 1.50274977e-01 -1.32825768e+00 4.96510655e-01 6.12738419e+00 1.34846556e+00 -1.44048083e+00 -6.81003109e-02 1.35940361e+00 -1.82173233e-02 2.75642693e-01 -6.02131963e-01 -8.67863774e-01 1.54933378e-01 4.89141881e-01 3.20620462e-02 -3.94251525e-01 6.35526061e-01 3.15344393e-01 -5.93745232e-01 -9.48925257e-01 8.89932036e-01 -1.14736900e-01 -1.65499735e+00 -1.00716248e-01 5.52955091e-01 6.51468277e-01 -1.19257100e-01 1.67655736e-01 -1.10103497e-02 7.77691677e-02 -1.14716852e+00 -2.76997626e-01 3.54255855e-01 1.01122212e+00 -8.47360969e-01 1.36127615e+00 3.99242371e-01 -9.68429208e-01 5.62792681e-02 -6.13638103e-01 3.53300065e-01 -4.11238432e-01 6.26310110e-01 -1.81566632e+00 2.71122575e-01 2.64224499e-01 6.97960496e-01 -8.13508868e-01 1.29996443e+00 3.51401031e-01 7.44622350e-01 -2.33160794e-01 -4.59247559e-01 2.73675948e-01 2.86435336e-01 9.89803299e-02 1.40628111e+00 3.75166118e-01 3.17663513e-02 -1.44383892e-01 3.55518520e-01 1.68781579e-01 5.16230404e-01 -3.02171320e-01 1.18251614e-01 2.36682922e-01 1.98605692e+00 -1.10033035e+00 -2.13645369e-01 -5.96680753e-02 4.84292865e-01 3.17025989e-01 9.76124182e-02 -9.37098205e-01 -4.46298689e-01 2.61814535e-01 4.74410594e-01 6.50488287e-02 3.78900558e-01 -2.78855801e-01 -7.96263158e-01 -4.61565107e-01 -6.12750411e-01 6.47264242e-01 -3.80943298e-01 -1.37405276e+00 3.58374596e-01 -4.29070055e-01 -1.37963235e+00 5.84889837e-02 -7.89206564e-01 -1.15548277e+00 4.17970717e-01 -1.73745465e+00 -1.48003352e+00 -5.38035393e-01 5.26387513e-01 3.98134381e-01 -1.73715100e-01 9.54326868e-01 -4.31903787e-02 -6.86016917e-01 8.79455149e-01 4.18607742e-01 4.12192941e-01 8.66106033e-01 -1.37496030e+00 -2.88876235e-01 4.36840594e-01 -4.20039952e-01 1.73425704e-01 1.51583239e-01 -1.45466879e-01 -1.20920622e+00 -1.26710486e+00 3.46012294e-01 5.55976760e-03 8.26057971e-01 5.12337238e-02 -5.79496682e-01 2.92735189e-01 3.25061172e-01 2.78147548e-01 1.33440387e+00 -2.97625154e-01 2.49911681e-01 -2.87970006e-01 -1.55113053e+00 5.39520323e-01 1.64575920e-01 -1.52645975e-01 2.47259095e-01 7.92504966e-01 5.62499017e-02 -4.33903545e-01 -1.14530194e+00 4.25575495e-01 5.79119146e-01 -8.08546484e-01 7.52392173e-01 -3.38284820e-01 3.81319970e-01 -1.80206642e-01 1.99461892e-01 -1.15457678e+00 -4.64657366e-01 1.24251306e-01 4.43494588e-01 9.36472118e-01 4.76859510e-01 -5.08249104e-01 1.20438719e+00 1.12257071e-01 -8.84490088e-02 -1.30529463e+00 -1.06280053e+00 -1.17267989e-01 2.01313093e-01 1.62932441e-01 2.79742151e-01 9.11858797e-01 1.70077309e-01 -1.00842550e-01 4.25411135e-01 5.45628443e-02 6.64583504e-01 6.47902042e-02 6.01910830e-01 -8.73602211e-01 -1.62776094e-02 -6.49239957e-01 -8.04905474e-01 -2.11131081e-01 -1.75600514e-01 -8.34889591e-01 -3.09858650e-01 -1.55915916e+00 6.88682437e-01 -5.93443334e-01 -5.26147783e-01 7.37738848e-01 -2.55125821e-01 7.44930923e-01 -3.75876188e-01 1.60715207e-01 -6.30284488e-01 -9.17219594e-02 1.60123312e+00 -5.70615172e-01 1.32848695e-01 6.13342933e-02 -5.22769213e-01 7.04407692e-01 8.46164048e-01 -8.67322460e-02 -9.30107683e-02 4.94162999e-02 -1.52961522e-01 1.23389713e-01 4.41095918e-01 -1.20207155e+00 4.35409606e-01 -2.47836471e-01 8.17113638e-01 -7.47544944e-01 2.51145095e-01 -5.99046528e-01 -2.63594836e-02 8.09979200e-01 -2.79911906e-01 -3.96337301e-01 2.58192688e-01 4.67439502e-01 -3.88715774e-01 -9.82135907e-02 1.12391675e+00 -2.22003415e-01 -7.54043162e-01 2.61750102e-01 -5.91770589e-01 -6.24022067e-01 1.84655905e+00 -5.39894223e-01 -7.23545790e-01 1.32342920e-01 -7.28868783e-01 8.70497301e-02 4.97039646e-01 -3.67281437e-01 6.29606485e-01 -1.05778718e+00 -1.10638332e+00 6.72482774e-02 4.71359611e-01 9.94756743e-02 6.35562956e-01 1.24152851e+00 -1.13301897e+00 4.22575176e-01 -2.70898759e-01 -8.37242484e-01 -1.69322824e+00 2.35057756e-01 5.46322584e-01 -7.87089765e-01 -3.56273502e-01 1.31657112e+00 5.18446624e-01 5.41797206e-02 1.46693528e-01 -4.11380619e-01 -5.58651268e-01 -5.01049310e-02 6.43507540e-01 2.95732588e-01 1.14910260e-01 -4.58639055e-01 -2.97407955e-01 6.23642981e-01 -5.95678508e-01 5.21287084e-01 1.14259541e+00 2.29260311e-01 -3.73936504e-01 1.07644521e-01 1.27225614e+00 1.30044386e-01 -9.39139128e-01 4.67415415e-02 -4.06439364e-01 -1.68081716e-01 1.87703669e-01 -9.73635554e-01 -1.18751955e+00 9.06790018e-01 1.07621837e+00 4.72927094e-02 1.28905725e+00 -3.59929278e-02 6.16076469e-01 7.37794936e-02 1.80394500e-01 -7.98245013e-01 -1.89318746e-01 -7.62660578e-02 3.50111812e-01 -1.66685069e+00 3.32172930e-01 -5.54569662e-01 -5.26998758e-01 1.36684740e+00 5.69914877e-01 -2.38643274e-01 6.43833578e-01 4.35374558e-01 3.74804854e-01 -2.65544534e-01 -9.63424087e-01 -2.72671133e-01 -5.13184741e-02 6.18125975e-01 6.73848271e-01 4.67956275e-01 -3.54858547e-01 3.10573369e-01 2.64588118e-01 -2.12713927e-02 5.68038046e-01 8.28410745e-01 -8.06943059e-01 -8.39660466e-01 -2.61662692e-01 8.10129523e-01 -7.47326612e-01 8.77142921e-02 -6.20536208e-01 1.37690699e+00 2.44586095e-02 5.14947295e-01 1.10806182e-01 -2.38635495e-01 -3.04254472e-01 -3.98472518e-01 2.48349130e-01 -3.34941149e-01 -7.95940220e-01 3.63626003e-01 -7.42309988e-02 -5.13334526e-03 -2.97779113e-01 -4.85195905e-01 -1.39400804e+00 -1.82337552e-01 -5.84116757e-01 9.78682414e-02 8.55947971e-01 5.83334625e-01 3.43625173e-02 5.53155601e-01 7.33234227e-01 -2.99382120e-01 -2.62302369e-01 -1.10203207e+00 -1.04427254e+00 7.15706423e-02 2.59353280e-01 -2.58395106e-01 -1.53902248e-01 1.18454777e-01]
[15.172819137573242, -3.027350664138794]
8621e8bd-0900-4c5e-9150-ff54ac7ec819
continuous-landmark-detection-with-3d-queries
null
null
http://openaccess.thecvf.com//content/CVPR2023/html/Chandran_Continuous_Landmark_Detection_With_3D_Queries_CVPR_2023_paper.html
http://openaccess.thecvf.com//content/CVPR2023/papers/Chandran_Continuous_Landmark_Detection_With_3D_Queries_CVPR_2023_paper.pdf
Continuous Landmark Detection With 3D Queries
Neural networks for facial landmark detection are notoriously limited to a fixed set of landmarks in a dedicated layout, which must be specified at training time. Dedicated datasets must also be hand-annotated with the corresponding landmark configuration for training. We propose the first facial landmark detection network that can predict continuous, unlimited landmarks, allowing to specify the number and location of the desired landmarks at inference time. Our method combines a simple image feature extractor with a queried landmark predictor, and the user can specify any continuous query points relative to a 3D template face mesh as input. As it is not tied to a fixed set of landmarks, our method is able to leverage all pre-existing 2D landmark datasets for training, even if they have inconsistent landmark configurations. As a result, we present a very powerful facial landmark detector that can be trained once, and can be used readily for numerous applications like 3D face reconstruction, arbitrary face segmentation, and is even compatible with helmeted mounted cameras, and therefore could vastly simplify face tracking workflows for media and entertainment applications.
['Derek Bradley', 'Paulo Gotardo', 'Gaspard Zoss', 'Prashanth Chandran']
2023-01-01
null
null
null
cvpr-2023-1
['3d-face-reconstruction', 'facial-landmark-detection', 'face-reconstruction']
['computer-vision', 'computer-vision', 'computer-vision']
[-1.24379650e-01 -3.15430947e-02 -2.78202653e-01 -4.35156107e-01 -7.77951539e-01 -7.48513997e-01 4.67438549e-01 -7.32641816e-02 -3.79506767e-01 1.65184408e-01 -5.40776432e-01 -3.29938233e-01 9.27948430e-02 -5.59172273e-01 -6.40443504e-01 -5.00051618e-01 -1.03100941e-01 1.05711079e+00 2.57410318e-01 -5.49061922e-03 4.81781885e-02 1.28575253e+00 -1.86320662e+00 -4.10085887e-01 2.23886326e-01 1.11696327e+00 1.12327985e-01 4.36726809e-01 -1.04371682e-01 -2.76819497e-01 -3.55135828e-01 -9.79405791e-02 3.88935149e-01 -3.95844579e-02 -5.08057714e-01 2.30680272e-01 8.98776174e-01 -3.84200931e-01 2.56369203e-01 6.89142585e-01 6.40791655e-01 1.08664803e-01 6.17185652e-01 -1.19038928e+00 -1.34044632e-01 -2.18118131e-02 -6.08427763e-01 -7.82837570e-02 4.01871711e-01 9.24612358e-02 7.42739856e-01 -1.25776768e+00 9.55175221e-01 9.44533229e-01 8.95777643e-01 8.49716127e-01 -1.27439475e+00 -7.30857372e-01 2.33704716e-01 -2.58061260e-01 -1.68663144e+00 -9.28345859e-01 8.77973378e-01 -4.23958212e-01 6.77838683e-01 2.23373964e-01 8.56555164e-01 6.91583991e-01 -3.50181192e-01 2.31636778e-01 4.56072867e-01 -3.90663385e-01 3.54374349e-01 -4.30779941e-02 -3.06875139e-01 1.30508363e+00 -1.91155568e-01 -1.76888898e-01 -4.69735980e-01 -9.08131972e-02 1.13832200e+00 2.70858612e-02 -2.76191309e-02 -7.86094785e-01 -9.96476769e-01 5.70996642e-01 2.81853497e-01 9.55434516e-02 -5.86897992e-02 1.28085926e-01 1.13701567e-01 -1.80580653e-02 3.39275062e-01 2.74486035e-01 -7.16484129e-01 1.08162217e-01 -1.20467269e+00 7.81178847e-02 7.67804265e-01 1.08139348e+00 1.09176314e+00 -2.83188391e-02 1.62439331e-01 6.53012753e-01 3.42274845e-01 5.72096407e-01 -4.01037745e-02 -1.25832772e+00 -1.30459562e-01 6.36369586e-01 2.94876453e-02 -9.70432281e-01 -7.14937270e-01 -5.72816022e-02 -5.44303477e-01 6.52290821e-01 5.47300816e-01 -6.07613549e-02 -1.06087661e+00 1.60573804e+00 9.19635594e-01 3.80564481e-01 -5.63820004e-01 8.80729437e-01 8.79912198e-01 1.04110390e-01 -1.41788796e-01 7.72450268e-02 1.42878377e+00 -3.81594032e-01 -1.24074943e-01 -1.86610773e-01 5.50399065e-01 -6.94226444e-01 9.85930562e-01 1.56743124e-01 -1.04619300e+00 -3.96236807e-01 -8.96096289e-01 -3.32235992e-01 -4.92525637e-01 1.56032905e-01 7.50864863e-01 7.61478424e-01 -1.57536948e+00 5.66019058e-01 -1.07762527e+00 -4.92022872e-01 6.12412274e-01 1.01206613e+00 -6.94065392e-01 1.50199816e-01 -4.99999881e-01 8.28072786e-01 -1.00607552e-01 3.96872699e-01 -7.94601679e-01 -7.11681843e-01 -9.74958956e-01 -1.14975609e-01 4.61294621e-01 -8.06114316e-01 1.13919795e+00 -5.67867815e-01 -1.52604258e+00 1.31751502e+00 -4.47741836e-01 2.46858090e-01 4.83617604e-01 2.15577766e-01 -1.08845234e-01 2.47129858e-01 1.49623513e-01 1.02046633e+00 9.83948231e-01 -1.14667332e+00 -3.39276552e-01 -6.37662709e-01 8.83620977e-02 8.11719671e-02 -9.56892669e-02 9.68063176e-02 -1.08831167e+00 -6.44247830e-02 4.24358815e-01 -8.33907187e-01 -2.62241453e-01 7.51543105e-01 -4.30636644e-01 -2.86112458e-01 1.04206240e+00 -2.49987841e-01 5.84747732e-01 -2.18298316e+00 -1.24831423e-01 6.31117702e-01 1.78496882e-01 4.46892269e-02 -1.15575992e-01 7.56397843e-03 1.36392415e-01 1.83584109e-01 4.31131758e-03 -8.88607979e-01 2.32241116e-02 1.72084153e-01 1.87528595e-01 8.65042746e-01 2.51973242e-01 7.54094541e-01 -7.77071655e-01 -8.46540630e-01 4.76132423e-01 8.51096451e-01 -6.67738616e-01 -2.55042072e-02 -3.17027807e-01 6.19306147e-01 -5.70656478e-01 1.04817772e+00 6.48328483e-01 -2.43413806e-01 8.55806768e-02 -5.51329032e-02 -2.67719865e-01 9.73192379e-02 -1.51610076e+00 1.96129572e+00 -5.41397810e-01 4.75130528e-01 3.66819590e-01 -4.77074593e-01 1.12702656e+00 3.37642640e-01 8.03892791e-01 -2.72602499e-01 1.72429457e-01 2.67605454e-01 -5.60103595e-01 -1.72740832e-01 2.59966999e-01 8.59855190e-02 2.57190764e-01 6.19982362e-01 1.37860239e-01 -2.36595586e-01 -8.80104154e-02 -2.19655693e-01 8.19318891e-01 3.61309260e-01 3.41859013e-02 -2.15277687e-01 4.35329974e-01 -1.87532902e-01 5.65328956e-01 4.37025428e-01 -3.42393070e-02 8.90212953e-01 5.33937156e-01 -7.15387166e-01 -1.04215384e+00 -9.04891849e-01 -5.43678939e-01 1.20204341e+00 7.77464435e-02 -3.06005776e-01 -7.23537445e-01 -5.78467309e-01 -2.08387338e-02 2.37989146e-02 -4.94905561e-01 4.20865476e-01 -6.49100423e-01 -8.68644193e-02 4.55786139e-01 4.10496175e-01 8.03042129e-02 -9.42103386e-01 -7.11485088e-01 -1.14622034e-01 5.72725356e-01 -9.84915137e-01 -7.39687562e-01 3.46147746e-01 -7.60670841e-01 -1.24580348e+00 -5.19149005e-01 -1.11695468e+00 1.30152762e+00 9.11895931e-02 9.79486227e-01 6.30639672e-01 -3.31110388e-01 5.25319159e-01 1.48443207e-01 -1.50139213e-01 1.30979046e-01 2.10521400e-01 1.37178749e-01 -8.66696760e-02 3.71232986e-01 -6.68258727e-01 -7.01164186e-01 5.31134784e-01 -4.63328660e-01 -2.26951882e-01 -2.79904739e-03 6.84845269e-01 9.99899507e-01 -2.07929462e-01 1.94920972e-01 -7.58640051e-01 3.22051868e-02 -1.31193489e-01 -1.08153272e+00 1.68585464e-01 -4.33857977e-01 -8.75650346e-02 5.43883502e-01 -2.92909503e-01 -4.31360930e-01 7.62940168e-01 -2.72798866e-01 -6.40079081e-01 -4.04706419e-01 2.21991107e-01 -4.25791472e-01 -4.16204244e-01 5.24698496e-01 -2.27097616e-01 1.72512218e-01 -6.53389275e-01 5.12351751e-01 3.05972755e-01 5.41809857e-01 -7.21241891e-01 7.94905543e-01 5.25988638e-01 4.32831526e-01 -7.63158619e-01 -2.84659535e-01 -3.32786053e-01 -1.39324784e+00 -2.35251039e-01 6.10616028e-01 -6.02037430e-01 -1.14628148e+00 2.27577895e-01 -1.22107840e+00 -3.82563263e-01 -5.44606335e-02 9.85488147e-02 -4.63473141e-01 1.25508070e-01 -2.53142625e-01 -5.56884766e-01 -2.13371739e-01 -1.34520769e+00 1.45926881e+00 2.46264473e-01 -2.20689893e-01 -1.09686077e+00 -3.07303578e-01 -1.90417752e-01 9.01759267e-02 4.34455365e-01 6.75864220e-01 -4.30035174e-01 -7.94346154e-01 -4.61335301e-01 3.56202498e-02 -2.97592610e-01 1.46202803e-01 5.07349133e-01 -9.59822476e-01 -2.17018872e-01 -4.09769952e-01 -1.76972806e-01 3.20548266e-01 3.77280086e-01 1.32395387e+00 -1.50888413e-01 -5.82493544e-01 1.13166034e+00 1.05941474e+00 -1.30581753e-02 1.94108382e-01 2.05920488e-01 6.21001780e-01 4.31340963e-01 1.39294446e-01 2.33844742e-01 4.21906948e-01 1.02059388e+00 3.99412364e-01 -2.74026662e-01 6.56684414e-02 -3.71368706e-01 -1.97014034e-01 1.82645783e-01 -2.74782479e-01 2.15712160e-01 -9.83476698e-01 1.65764719e-01 -1.44367146e+00 -8.05760145e-01 2.53832936e-01 2.34207535e+00 7.38785028e-01 -1.96359366e-01 2.29515970e-01 -6.77891523e-02 6.23852551e-01 -3.94360535e-02 -7.92666137e-01 -9.68625173e-02 3.02735269e-01 5.24542570e-01 4.39268112e-01 5.81203222e-01 -1.08986211e+00 1.17968929e+00 7.13521433e+00 5.25329649e-01 -1.30796731e+00 -2.96330415e-02 5.74301064e-01 -2.17548549e-01 -2.77603418e-01 -1.17993228e-01 -1.05386472e+00 8.96434411e-02 5.56505442e-01 1.54917911e-01 5.37489057e-01 1.06853068e+00 2.10260034e-01 -1.55643830e-02 -1.34808373e+00 1.18176138e+00 -1.49530366e-01 -1.32728326e+00 -2.32675284e-01 1.82900444e-01 2.93756753e-01 -7.87794217e-02 1.65942848e-01 -1.49818167e-01 5.51530570e-02 -1.30990946e+00 5.41530430e-01 3.72605652e-01 1.36689591e+00 -8.02375853e-01 2.03799233e-01 3.30540091e-01 -1.12410784e+00 4.27223742e-01 -5.14952838e-01 2.84918368e-01 7.42561668e-02 2.21626371e-01 -1.09087622e+00 6.69208914e-02 5.46100795e-01 4.29839134e-01 -4.62087095e-01 1.07156670e+00 -9.52005014e-02 2.03642204e-01 -9.76804793e-01 6.28336743e-02 -7.65645579e-02 -9.55378786e-02 1.21694081e-01 8.38497639e-01 5.72387576e-01 1.03277666e-03 3.08842897e-01 7.83539295e-01 -1.58741057e-01 2.60956764e-01 -7.09654331e-01 5.21753788e-01 9.28939581e-01 1.45877361e+00 -1.17938292e+00 1.65116563e-01 -3.41364354e-01 7.40837216e-01 5.06217539e-01 2.17871964e-01 -6.11118555e-01 -1.67626902e-01 6.60964131e-01 4.03066248e-01 3.86132181e-01 -5.60771286e-01 -2.83728242e-01 -8.38095844e-01 7.82669485e-02 -4.21151161e-01 2.03192487e-01 -6.39040351e-01 -8.15852940e-01 7.00375438e-01 -1.92904979e-01 -1.09769154e+00 -4.24801260e-01 -8.21474493e-01 -6.42112434e-01 9.08423722e-01 -1.38291967e+00 -1.39591956e+00 -2.99731880e-01 8.92087162e-01 3.24838385e-02 -8.32108110e-02 1.08565760e+00 3.89156967e-01 -7.69520164e-01 8.34052920e-01 -3.12533885e-01 3.96182418e-01 4.83793437e-01 -9.83091354e-01 4.69023734e-01 5.40922225e-01 6.19362712e-01 8.63466918e-01 3.70683193e-01 -3.32020760e-01 -1.66020596e+00 -8.98845851e-01 7.61487901e-01 -8.49388719e-01 2.95558721e-01 -6.45123005e-01 -6.17896914e-01 9.36594009e-01 -4.18448806e-01 5.98882258e-01 6.31635606e-01 4.08588290e-01 -2.67257661e-01 -2.72256374e-01 -1.20240366e+00 6.09452486e-01 1.25447142e+00 -5.60615778e-01 -1.61822401e-02 5.16563594e-01 2.68275261e-01 -9.91211951e-01 -6.82747662e-01 1.47041723e-01 6.60385907e-01 -7.91371346e-01 1.12207961e+00 -5.05844474e-01 -1.55444384e-01 -6.10523462e-01 1.52978405e-01 -7.49621570e-01 -8.34427252e-02 -7.86681533e-01 1.22244209e-01 1.18888676e+00 5.94669163e-01 -4.54454720e-01 1.12114203e+00 1.15278018e+00 -1.90473586e-01 -8.33215356e-01 -1.43213928e+00 -4.91538256e-01 -2.96640098e-01 -5.14734864e-01 1.04168272e+00 7.29170680e-01 -1.70681089e-01 1.26408134e-02 -1.55773357e-01 4.31963056e-01 4.45744246e-01 1.50293499e-01 1.03464723e+00 -1.35099840e+00 1.32679835e-01 -6.15785539e-01 -6.13977134e-01 -1.23608327e+00 2.93578923e-01 -1.08386159e+00 1.22228421e-01 -1.17611969e+00 -3.25699985e-01 -1.01551640e+00 3.29777971e-02 9.76480007e-01 2.19771415e-01 6.00823402e-01 -2.10310936e-01 1.07019313e-01 -4.25196648e-01 2.39731133e-01 1.10022843e+00 1.75828233e-01 -4.47601825e-01 1.50996760e-01 -5.12799799e-01 8.02304864e-01 5.56432664e-01 -3.72730047e-01 -1.98992684e-01 -4.81073290e-01 3.29764262e-02 -5.18426038e-02 5.33897161e-01 -8.52087379e-01 5.41178167e-01 -2.03858063e-01 7.60219812e-01 -5.16276360e-01 5.85895061e-01 -1.07674360e+00 3.02699536e-01 -1.87482044e-01 -4.22392748e-02 6.88811466e-02 1.97622940e-01 3.28348801e-02 2.46613652e-01 -2.41758615e-01 7.61857450e-01 -1.83398515e-01 -7.69661546e-01 8.84476423e-01 1.69643432e-01 -3.87703747e-01 1.12098002e+00 -5.16882479e-01 1.06027037e-01 -1.02191284e-01 -8.58270526e-01 2.32196704e-01 9.32539105e-01 2.01646790e-01 6.45073593e-01 -1.25129485e+00 -2.41584986e-01 6.83520794e-01 -1.25741020e-01 4.81005341e-01 -1.17687091e-01 5.32716453e-01 -5.83607733e-01 2.14236766e-01 -4.25488092e-02 -9.15305793e-01 -1.19978976e+00 4.41375345e-01 5.49224973e-01 4.76585567e-01 -7.85602212e-01 8.00121546e-01 -4.00189519e-01 -5.79360604e-01 2.94068336e-01 -2.06116751e-01 8.13684054e-03 6.45728856e-02 5.72883785e-01 -8.76280305e-04 2.17843398e-01 -7.82894671e-01 -5.28976798e-01 1.09051335e+00 2.43744537e-01 -4.39526588e-02 1.41148221e+00 -2.32965164e-02 -2.33145013e-01 1.86939776e-01 1.12511301e+00 1.35537207e-01 -1.54692197e+00 4.19363305e-02 -8.07754174e-02 -5.24716496e-01 -4.06649075e-02 -4.14063483e-01 -1.43401146e+00 7.60265529e-01 4.57071424e-01 -1.71174362e-01 8.78508091e-01 2.49138236e-01 4.63807344e-01 3.42522055e-01 7.50850379e-01 -7.65983820e-01 -2.90279329e-01 3.05188775e-01 6.41348779e-01 -1.11487055e+00 -5.19141927e-02 -4.42511618e-01 -9.72381756e-02 1.36372066e+00 5.33830941e-01 6.20581966e-04 8.84744167e-01 4.98357773e-01 3.49433869e-01 -4.81225103e-01 -4.27559227e-01 -2.91610152e-01 4.36751455e-01 8.76250565e-01 3.78278285e-01 -3.37871820e-01 2.82530189e-01 6.84302971e-02 -3.98557842e-01 -1.26766235e-01 -9.28850565e-03 7.02477157e-01 -3.60241562e-01 -1.25637269e+00 -3.74237627e-01 3.59193772e-01 -1.83838561e-01 1.27048433e-01 -5.94118871e-02 9.87285256e-01 2.00728878e-01 5.43423116e-01 3.53454143e-01 -1.20416827e-01 3.39028358e-01 2.87044346e-01 5.53122222e-01 -7.98421621e-01 -2.13809133e-01 9.60137621e-02 -2.76086926e-01 -7.07462966e-01 -2.80579805e-01 -8.92862141e-01 -1.40743637e+00 -3.18116665e-01 -3.16305697e-01 -8.95894095e-02 8.54513168e-01 8.27381849e-01 6.36211395e-01 -2.18488291e-01 5.06108105e-01 -1.39650118e+00 6.12830855e-02 -4.46247548e-01 -5.43514848e-01 6.09863549e-02 5.47939122e-01 -8.93095195e-01 -2.45090336e-01 1.26254737e-01]
[13.448878288269043, 0.2824179530143738]
15587263-acc2-41e0-a569-3f468625786e
solving-the-undirected-feedback-vertex-set
1405.0446
null
http://arxiv.org/abs/1405.0446v1
http://arxiv.org/pdf/1405.0446v1.pdf
Solving the undirected feedback vertex set problem by local search
An undirected graph consists of a set of vertices and a set of undirected edges between vertices. Such a graph may contain an abundant number of cycles, then a feedback vertex set (FVS) is a set of vertices intersecting with each of these cycles. Constructing a FVS of cardinality approaching the global minimum value is a optimization problem in the nondeterministic polynomial-complete complexity class, therefore it might be extremely difficult for some large graph instances. In this paper we develop a simulated annealing local search algorithm for the undirected FVS problem. By defining an order for the vertices outside the FVS, we replace the global cycle constraints by a set of local vertex constraints on this order. Under these local constraints the cardinality of the focal FVS is then gradually reduced by the simulated annealing dynamical process. We test this heuristic algorithm on large instances of Er\"odos-Renyi random graph and regular random graph, and find that this algorithm is comparable in performance to the belief propagation-guided decimation algorithm.
['Hai-Jun Zhou', 'Shao-Meng Qin']
2014-05-01
null
null
null
null
['feedback-vertex-set-fvs']
['graphs']
[ 4.49253976e-01 8.07814777e-01 -3.48334283e-01 -1.06950425e-01 -2.46590555e-01 -6.71565413e-01 7.52849817e-01 2.10470840e-01 -1.13979369e-01 9.18052971e-01 -2.05307961e-01 -4.75900352e-01 -3.66410464e-01 -1.25891209e+00 -7.08719671e-01 -1.03272271e+00 -4.95747745e-01 1.16369939e+00 5.58832943e-01 -2.12582707e-01 3.56066167e-01 6.64354861e-01 -1.27210951e+00 -3.08389783e-01 4.85139698e-01 3.40882868e-01 2.19022974e-01 8.48266125e-01 -6.43359795e-02 3.75510931e-01 -3.82304728e-01 -1.40006796e-01 2.66114563e-01 -6.59465849e-01 -1.02392507e+00 5.38063407e-01 -3.51042569e-01 2.59534955e-01 -4.53838110e-01 1.44361484e+00 1.94367900e-01 2.16445997e-01 6.04745626e-01 -1.31722200e+00 -2.02460125e-01 8.28658164e-01 -6.90667987e-01 2.67052710e-01 4.12467420e-01 -1.66583255e-01 1.17908907e+00 -2.41172642e-01 8.71995389e-01 9.87924635e-01 2.61971354e-01 3.30408782e-01 -1.49988186e+00 -2.47707322e-01 2.73035228e-01 -3.35454708e-03 -1.51076400e+00 -1.11352839e-01 6.97555363e-01 -2.69274443e-01 7.11324632e-01 5.37916362e-01 9.22530890e-01 5.24708092e-01 1.91454276e-01 1.91193730e-01 1.01450682e+00 -8.97054255e-01 5.83109677e-01 -8.45412239e-02 4.21583086e-01 1.19671965e+00 6.38249815e-01 4.35452342e-01 -2.02261552e-01 -4.33544010e-01 6.68461323e-01 -3.36403787e-01 -3.04687768e-01 -5.23116887e-01 -7.94114113e-01 1.09745133e+00 4.02605325e-01 2.44894400e-01 -1.36710554e-01 4.60649788e-01 -8.55725780e-02 3.93386155e-01 1.91848159e-01 3.81199270e-01 -1.40151739e-01 4.15770411e-01 -7.44349241e-01 1.43298015e-01 1.13998890e+00 7.77923107e-01 9.38615620e-01 -2.82636166e-01 1.80955172e-01 3.77966017e-01 4.51992303e-01 3.69625628e-01 -2.68092394e-01 -8.18984747e-01 1.45608321e-01 6.10930324e-01 5.43869250e-02 -1.12258065e+00 -6.32515430e-01 -3.69435102e-01 -9.74608898e-01 2.63223231e-01 4.33990508e-01 -3.39105427e-02 -1.33751082e+00 1.88326609e+00 4.92338955e-01 7.90700242e-02 -1.75544322e-01 7.96970665e-01 4.12850112e-01 8.83158684e-01 -4.25666958e-01 -7.52684653e-01 1.08256793e+00 -8.52918863e-01 -5.12117863e-01 -2.15018839e-01 6.35270059e-01 -4.62960482e-01 3.17959130e-01 3.50968540e-01 -1.11147547e+00 1.87627926e-01 -1.13138211e+00 5.46317697e-01 -1.10332280e-01 -4.23303157e-01 7.48579741e-01 7.99453914e-01 -1.36354530e+00 5.78094721e-01 -8.99489224e-01 -1.00666925e-01 -2.53936946e-01 7.21305609e-01 -2.27602690e-01 -1.09136812e-01 -1.09323299e+00 7.79630542e-01 6.33515954e-01 2.42302105e-01 -1.04544008e+00 2.26999998e-01 -8.71838152e-01 -4.31987494e-02 7.64124274e-01 -7.06239939e-01 8.90920758e-01 -9.98473823e-01 -1.23110652e+00 8.18873763e-01 -3.40364307e-01 -3.82369548e-01 2.32498735e-01 9.30017531e-01 -2.53782004e-01 2.24705458e-01 -2.11360097e-01 2.43882105e-01 9.13400590e-01 -1.53530848e+00 -3.98777723e-01 -3.43571752e-01 3.09281141e-01 2.46176586e-01 4.50611003e-02 -1.58785239e-01 -6.90929472e-01 -1.89470589e-01 6.79820836e-01 -1.53223729e+00 -6.94483757e-01 -6.56221569e-01 -7.54779577e-01 -2.96244740e-01 3.51321995e-01 -4.78427075e-02 1.51230562e+00 -1.76969552e+00 4.79969770e-01 1.06798041e+00 4.89767790e-01 -1.82084277e-01 -5.24542667e-03 5.57573855e-01 -1.13391258e-01 2.94993281e-01 -3.97528410e-01 8.11579525e-02 -3.30239922e-01 6.24090850e-01 6.86525404e-02 9.05725658e-01 -3.89406353e-01 3.84156764e-01 -1.08639979e+00 -3.65593702e-01 -1.66225657e-01 -2.66022772e-01 -6.99100852e-01 -1.21286318e-01 -5.17711699e-01 1.31030232e-01 -6.40889347e-01 4.45122004e-01 6.19951010e-01 -6.55467987e-01 6.15859747e-01 5.36919653e-01 9.27388109e-03 -1.67810414e-02 -1.65858364e+00 1.08718777e+00 1.04255207e-01 3.07040036e-01 1.81296706e-01 -9.39812481e-01 9.72953141e-01 1.95868358e-01 3.87511641e-01 -9.14573595e-02 2.45195821e-01 2.17995830e-02 2.32272148e-01 -7.49322474e-02 4.88037407e-01 -1.19909324e-01 -6.46915659e-02 5.72361052e-01 -2.11589769e-01 -2.68707305e-01 4.27428812e-01 5.50505042e-01 1.50028157e+00 -5.40483713e-01 1.58753008e-01 -5.49705982e-01 5.02080560e-01 -5.37991188e-02 6.44978464e-01 1.01869774e+00 1.25101758e-02 4.89536017e-01 9.02970731e-01 -2.86342621e-01 -9.41974699e-01 -9.70132709e-01 -1.14539288e-01 6.36458755e-01 5.24325669e-01 -7.14559734e-01 -8.73931408e-01 -3.55843216e-01 -1.65033862e-01 4.64471102e-01 -7.89154112e-01 -2.76823282e-01 -4.25414175e-01 -9.69355941e-01 2.89747007e-02 -1.11954898e-01 2.83077866e-01 -8.28123748e-01 -1.18841752e-01 3.63482445e-01 7.96314776e-02 -6.60943270e-01 -4.50260341e-01 4.37761724e-01 -8.93881559e-01 -1.25552332e+00 -1.86306566e-01 -1.04669452e+00 1.23650610e+00 2.23664001e-01 1.04882705e+00 4.78330582e-01 1.11789986e-01 7.39920586e-02 -2.00876683e-01 8.05746689e-02 -6.94718897e-01 1.71443328e-01 -1.08690806e-01 -1.85250357e-01 -4.13284414e-02 -3.69515151e-01 -2.64571846e-01 3.60065997e-01 -9.22128260e-01 1.19223513e-01 -8.07873607e-02 9.34508204e-01 8.89981270e-01 6.62858546e-01 9.31433290e-02 -1.01067138e+00 4.86913502e-01 -6.58694565e-01 -1.13453257e+00 2.32499406e-01 -6.96485639e-01 5.38083494e-01 4.30248559e-01 -1.11478299e-01 -6.26317739e-01 2.03924611e-01 2.62331367e-01 1.29568264e-01 4.50740486e-01 8.78376663e-01 -1.37878433e-01 -3.83668602e-01 5.29320717e-01 -4.66259681e-02 -8.08077231e-02 -6.40844852e-02 2.51286954e-01 3.26781482e-01 2.07748860e-01 -4.40522581e-01 7.63134360e-01 3.39810878e-01 7.16703892e-01 -7.82168508e-01 -2.76276916e-01 -7.36122951e-02 -2.72214562e-01 -5.03382206e-01 5.22004426e-01 -2.78626353e-01 -7.60714710e-01 2.54804015e-01 -9.99620199e-01 -2.44000718e-01 -2.35749289e-01 3.01879406e-01 -5.43749332e-01 4.55479115e-01 -6.80675805e-01 -8.30202341e-01 7.68824518e-02 -1.29364407e+00 3.83458853e-01 2.51172632e-01 -1.89439103e-01 -9.74621058e-01 4.93132383e-01 -3.19191851e-02 -1.78158104e-01 4.47553307e-01 1.05581439e+00 -5.88913739e-01 -9.01733339e-01 -4.37772423e-01 2.17568979e-01 -1.14681058e-01 -1.56000689e-01 3.16527188e-01 -1.84740379e-01 -5.08235335e-01 -1.33094490e-01 1.15590356e-01 8.97637606e-01 5.37643552e-01 7.59041607e-01 -2.66912311e-01 -7.87356675e-01 4.16941553e-01 1.71261406e+00 5.51870465e-01 5.66897273e-01 3.94203037e-01 4.03316200e-01 2.77840227e-01 2.25308999e-01 1.96673930e-01 2.37155318e-01 2.77450979e-01 6.43526256e-01 1.75280035e-01 2.44167209e-01 -1.47051424e-01 9.02179927e-02 9.94107187e-01 -2.56189227e-01 -8.97599161e-01 -9.69347358e-01 6.04601145e-01 -1.79847360e+00 -7.36390233e-01 -4.50306714e-01 2.36909795e+00 7.12260127e-01 4.61181015e-01 -7.72069395e-02 3.93260360e-01 1.36305511e+00 3.20692986e-01 -3.18932563e-01 -8.18237305e-01 -1.18172832e-01 4.06573489e-02 8.96469295e-01 1.05868185e+00 -5.83614588e-01 8.86754513e-01 6.82686615e+00 5.36536455e-01 -4.32050139e-01 -1.30699471e-01 6.06117129e-01 1.43747538e-01 -8.08870494e-01 5.68590224e-01 -7.18952358e-01 4.72208023e-01 1.00382018e+00 -1.88110337e-01 9.25300896e-01 3.23894531e-01 -1.35835305e-01 -5.66689849e-01 -8.78103733e-01 4.01910543e-01 -1.39605522e-01 -1.34689999e+00 -2.35493496e-01 4.15998578e-01 1.21870065e+00 -1.09675616e-01 -2.25918025e-01 -2.49964654e-01 1.15459383e+00 -1.02271783e+00 4.54993904e-01 2.06746832e-01 5.27072966e-01 -9.64201152e-01 3.10093135e-01 4.29375827e-01 -1.17385006e+00 1.58507433e-02 -6.47878274e-02 6.34134933e-02 2.76102394e-01 7.01470017e-01 -5.54572880e-01 4.80305851e-01 3.23505312e-01 -3.45474184e-02 -1.03618488e-01 1.08882129e+00 -3.06578070e-01 7.03950167e-01 -8.61858606e-01 -3.69108200e-01 4.44545597e-01 -8.04572225e-01 9.45405602e-01 7.40617394e-01 1.52580515e-01 5.63587427e-01 1.81632787e-01 5.65483570e-01 -7.91715905e-02 -2.46510938e-01 -7.32072294e-01 -2.77243685e-02 6.12385571e-01 1.00297308e+00 -1.48062944e+00 -3.37372661e-01 -7.09316805e-02 6.46857619e-01 3.23631376e-01 4.00435507e-01 -7.37729251e-01 -2.31495634e-01 2.43375331e-01 2.03737065e-01 2.81626523e-01 -1.56015173e-01 -2.80314684e-01 -8.12935650e-01 -2.84043193e-01 -7.03719497e-01 6.95351243e-01 -4.95380253e-01 -5.64021707e-01 7.47316241e-01 -9.61801484e-02 -6.40823543e-01 -2.69665897e-01 -1.34392664e-01 -4.47151423e-01 6.29630983e-01 -9.38304126e-01 -4.64652717e-01 4.60231639e-02 5.96690595e-01 -8.02121609e-02 1.09925330e-01 7.40579367e-01 -2.79455096e-01 -6.19624257e-01 1.76709935e-01 3.03031355e-01 -3.47633123e-01 -1.16991445e-01 -1.40359199e+00 2.71801561e-01 1.14951611e+00 9.73447189e-02 4.37996775e-01 1.15942943e+00 -9.56154764e-01 -1.58724976e+00 -7.37947941e-01 9.99671340e-01 -3.19669768e-02 6.03320181e-01 -3.69310051e-01 -6.75792634e-01 9.98775780e-01 1.85718596e-01 -4.29020412e-02 1.47526860e-01 9.46331024e-02 7.53812306e-03 1.71274856e-01 -1.13453639e+00 8.17840755e-01 1.22399449e+00 -3.62316072e-02 -3.65742743e-01 8.58047187e-01 4.27210420e-01 -4.45069045e-01 -5.63537598e-01 3.23833913e-01 -7.13978196e-03 -6.48517847e-01 6.70806229e-01 -3.62694860e-01 1.77956209e-01 -5.84207594e-01 1.42831057e-01 -1.42703199e+00 -7.12103307e-01 -1.08541870e+00 1.08503863e-01 8.10767889e-01 4.57703233e-01 -7.27410316e-01 1.08275092e+00 3.42299521e-01 5.05875833e-02 -5.98145366e-01 -1.15088499e+00 -6.37971759e-01 -1.59536168e-01 -8.02370161e-03 4.83872443e-01 7.63588846e-01 2.78679371e-01 5.24306118e-01 -2.38366872e-01 4.23867315e-01 9.15750623e-01 3.39684874e-01 3.77961248e-01 -1.36201954e+00 -5.36169171e-01 -3.99489015e-01 -3.99631113e-01 -7.44037569e-01 6.82322383e-02 -1.17978561e+00 5.62546588e-02 -1.55199790e+00 2.18678460e-01 -6.96098149e-01 7.21756518e-02 2.62997001e-01 1.57802865e-01 -1.90909818e-01 -1.03841480e-02 2.06430610e-02 -3.67526621e-01 1.80091217e-01 1.23678505e+00 2.93647200e-02 -4.48335826e-01 9.48102698e-02 -3.49880308e-01 7.83316731e-01 7.32668400e-01 -7.57855356e-01 -4.09772575e-01 7.26912841e-02 5.03588378e-01 7.70902157e-01 -1.47054806e-01 -4.68528926e-01 3.07944238e-01 -3.27305675e-01 -1.82768255e-01 -5.61352551e-01 8.40332732e-02 -6.91610157e-01 7.89328158e-01 8.68007362e-01 -2.77488828e-01 4.57585715e-02 -3.27314496e-01 8.88310611e-01 1.62349969e-01 -6.14225447e-01 8.76301169e-01 -1.59615204e-01 -4.02844965e-01 1.83931544e-01 -6.85151935e-01 -1.29372671e-01 1.37248266e+00 -3.30315500e-01 -2.22259536e-01 -3.88297588e-01 -1.03030407e+00 3.80716741e-01 5.50382376e-01 -5.31560555e-03 4.41730052e-01 -1.09920728e+00 -3.68868053e-01 1.80105627e-01 -6.56437874e-02 1.45610198e-01 -9.16171744e-02 4.99423593e-01 -7.68181682e-01 1.83735162e-01 1.94545105e-01 -3.99890125e-01 -1.41327858e+00 7.82557905e-01 4.01215941e-01 -4.17553931e-01 -6.79219544e-01 1.05486715e+00 -8.23006183e-02 -1.75326273e-01 1.28890611e-02 -7.78557286e-02 -1.51122317e-01 -1.64842442e-01 9.22973230e-02 4.81665730e-01 -3.58938463e-02 -6.65336668e-01 -2.41375342e-01 2.47964427e-01 -1.40219808e-01 -4.02349323e-01 1.08191848e+00 -2.48105556e-01 -5.57655454e-01 1.01258650e-01 1.12198222e+00 4.58934717e-02 -8.19655418e-01 -1.63225904e-01 1.04424283e-01 -3.09881300e-01 1.22572891e-01 -2.78238744e-01 -1.31842291e+00 -2.54196879e-02 1.01521708e-01 7.94058263e-01 1.09857833e+00 2.99066454e-01 3.44215542e-01 3.06433290e-01 7.37383246e-01 -1.01262498e+00 -4.20656830e-01 5.57566285e-01 4.95111167e-01 -7.58689106e-01 2.32955683e-02 -7.06269085e-01 -2.13558912e-01 1.18593872e+00 1.33833259e-01 -5.33278286e-01 7.69796491e-01 4.64503616e-01 -6.74372196e-01 -4.28294599e-01 -9.96638954e-01 -9.15204883e-02 -1.38837665e-01 2.15839386e-01 -2.79006243e-01 2.81506032e-01 -8.98630559e-01 1.62782311e-01 -2.84246325e-01 -9.56645831e-02 9.40875232e-01 9.87884879e-01 -8.15935493e-01 -1.18401027e+00 -5.85347414e-01 5.50678730e-01 -7.44174793e-02 -5.25395498e-02 -3.82509977e-01 7.40863740e-01 -7.69780353e-02 1.03630614e+00 9.76900756e-02 -3.93721402e-01 -1.40264437e-01 -4.06858593e-01 7.25614190e-01 -5.28271556e-01 -3.46543640e-01 5.37230559e-02 3.98928881e-01 -2.01662108e-01 -1.81843683e-01 -7.25822270e-01 -1.41009247e+00 -7.00161040e-01 -8.34262908e-01 4.79152679e-01 5.28786182e-01 8.81586194e-01 -1.20177865e-01 3.06681812e-01 9.04486835e-01 -3.85892898e-01 -3.24808836e-01 -4.15783256e-01 -1.11356699e+00 -2.73261845e-01 1.03234544e-01 -6.33976042e-01 -8.52444649e-01 -2.28364870e-01]
[6.893631935119629, 5.372917652130127]
839d2fae-c6c9-4c38-9225-d2aff2a8a7fc
css-a-large-scale-cross-schema-chinese-text
2305.15891
null
https://arxiv.org/abs/2305.15891v1
https://arxiv.org/pdf/2305.15891v1.pdf
CSS: A Large-scale Cross-schema Chinese Text-to-SQL Medical Dataset
The cross-domain text-to-SQL task aims to build a system that can parse user questions into SQL on complete unseen databases, and the single-domain text-to-SQL task evaluates the performance on identical databases. Both of these setups confront unavoidable difficulties in real-world applications. To this end, we introduce the cross-schema text-to-SQL task, where the databases of evaluation data are different from that in the training data but come from the same domain. Furthermore, we present CSS, a large-scale CrosS-Schema Chinese text-to-SQL dataset, to carry on corresponding studies. CSS originally consisted of 4,340 question/SQL pairs across 2 databases. In order to generalize models to different medical systems, we extend CSS and create 19 new databases along with 29,280 corresponding dataset examples. Moreover, CSS is also a large corpus for single-domain Chinese text-to-SQL studies. We present the data collection approach and a series of analyses of the data statistics. To show the potential and usefulness of CSS, benchmarking baselines have been conducted and reported. Our dataset is publicly available at \url{https://huggingface.co/datasets/zhanghanchong/css}.
['Kai Yu', 'Yefeng Zheng', 'Yu Huang', 'Yunyan Zhang', 'Ruisheng Cao', 'Lu Chen', 'Jieyu Li', 'Hanchong Zhang']
2023-05-25
null
null
null
null
['text-to-sql']
['computer-code']
[-1.63716041e-02 -1.12266026e-01 1.60662215e-02 -8.80925596e-01 -1.46290278e+00 -7.93531597e-01 3.03812176e-01 3.29428315e-01 -4.49049383e-01 6.47875011e-01 1.87697217e-01 -4.40286398e-01 3.87109630e-02 -9.00318027e-01 -8.61803651e-01 -2.00469211e-01 4.40563291e-01 8.01033497e-01 3.45310897e-01 -5.03650844e-01 5.35622165e-02 -8.01149383e-02 -1.25800610e+00 1.04072034e+00 1.03834784e+00 1.07990241e+00 2.31139392e-01 5.83237231e-01 -4.12904829e-01 6.63194537e-01 -8.24356794e-01 -9.07469749e-01 2.15150546e-02 -3.08429509e-01 -9.43609476e-01 -3.26341271e-01 5.02240062e-01 -3.38201597e-02 -9.06746462e-02 8.66168261e-01 8.51257861e-01 -3.81677181e-01 2.10781097e-01 -1.02597570e+00 -9.49485004e-01 5.52240431e-01 8.86165574e-02 -1.06756520e-02 7.26433754e-01 -8.65655243e-02 9.97655034e-01 -1.07330024e+00 8.85013461e-01 1.06117189e+00 6.37944043e-01 5.67770481e-01 -8.62681150e-01 -6.81634605e-01 -2.87942231e-01 -7.25131854e-02 -1.24988949e+00 -3.72341156e-01 4.56805915e-01 -1.62139326e-01 8.18025172e-01 6.05240941e-01 2.07365919e-02 1.48608422e+00 -1.06532030e-01 9.35679495e-01 9.81840253e-01 -4.49250698e-01 8.57823640e-02 6.42066360e-01 3.46205324e-01 5.00222445e-01 2.17670873e-01 -1.30421981e-01 -4.12489831e-01 -3.15640897e-01 2.62183011e-01 5.39547578e-02 -2.64561385e-01 -1.81591302e-01 -1.39554989e+00 8.00415218e-01 7.14191273e-02 3.45234782e-01 9.85382274e-02 -4.95955139e-01 5.87280035e-01 5.89536071e-01 7.43882060e-02 5.38747847e-01 -1.04227924e+00 -1.04893982e-01 -4.27807182e-01 6.29797041e-01 1.16388226e+00 1.91507006e+00 5.44316590e-01 -6.42113030e-01 -1.79202214e-01 1.00345004e+00 -1.10094897e-01 6.47340119e-01 7.00856149e-01 -6.17695570e-01 1.19133019e+00 9.17129815e-01 -1.51818246e-01 -7.53570735e-01 -3.81328374e-01 -7.38274008e-02 -7.97607422e-01 -8.20909381e-01 4.87962782e-01 -2.94879436e-01 -8.11111212e-01 1.51101732e+00 1.57178596e-01 -3.88667017e-01 3.84280056e-01 5.22150934e-01 1.45843244e+00 4.34017628e-01 3.89070213e-02 1.57626241e-01 1.76727080e+00 -6.84515774e-01 -8.56450498e-01 4.85337451e-02 8.56864333e-01 -8.85619402e-01 1.68634665e+00 4.24616873e-01 -8.45852494e-01 -7.04315960e-01 -6.15662217e-01 -4.21010584e-01 -1.00984395e+00 3.32478315e-01 1.40158400e-01 8.53977859e-01 -6.88336372e-01 3.36656510e-03 -6.07385874e-01 -6.82073414e-01 1.01743132e-01 -1.81378648e-01 -3.65617901e-01 -3.43175858e-01 -1.53673708e+00 6.74854457e-01 7.43055880e-01 -1.83591157e-01 -5.27846992e-01 -7.72341251e-01 -7.86991298e-01 -2.00608954e-01 8.56163442e-01 -5.09197056e-01 1.27707922e+00 -2.53496110e-01 -8.35440576e-01 1.10323524e+00 -2.59490252e-01 -1.73017144e-01 5.54242730e-01 -3.17677408e-01 -9.24424171e-01 1.16853174e-02 3.17927241e-01 1.68973312e-01 2.50204653e-01 -9.57531452e-01 -4.28217918e-01 -4.88564998e-01 -1.98081464e-01 -1.73101664e-01 -3.01668704e-01 1.50566608e-01 -1.13169444e+00 -9.44156826e-01 -9.58624855e-03 -7.76461422e-01 1.52036054e-02 -4.00710464e-01 -9.33023512e-01 -2.05753922e-01 4.58431274e-01 -7.25399315e-01 1.62792230e+00 -2.12955785e+00 -4.76486266e-01 -9.64006931e-02 8.96379724e-02 1.06867716e-01 -2.02999920e-01 8.29024255e-01 -2.88008869e-01 4.49559718e-01 -4.09293175e-01 -2.06215099e-01 -6.20402358e-02 3.44520926e-01 -4.33852226e-01 -2.42554545e-01 5.36059797e-01 1.05579722e+00 -4.88771588e-01 -8.45431030e-01 -4.43466753e-01 -1.34918734e-01 -6.75279617e-01 6.16915405e-01 -4.45730776e-01 6.35568649e-02 -7.89145291e-01 1.06055558e+00 6.96398973e-01 -4.20149475e-01 3.02693486e-01 -2.13358209e-01 2.64112353e-01 3.81057382e-01 -9.63652968e-01 1.79043460e+00 -7.06626698e-02 5.66256344e-02 -3.67377162e-01 -8.88101280e-01 1.02567661e+00 3.20140749e-01 2.96475023e-01 -1.03862131e+00 -1.17747821e-01 3.31741095e-01 -5.22145569e-01 -1.15101039e+00 6.13438964e-01 -2.15296634e-02 -7.53705382e-01 1.21199541e-01 -5.21329418e-03 -3.50978404e-01 4.29589272e-01 3.46912354e-01 1.07592642e+00 -2.63808697e-01 2.24632099e-01 -6.96130991e-02 6.17086589e-01 4.48566765e-01 5.57280064e-01 9.24811184e-01 -2.99381930e-02 9.85756099e-01 7.83650279e-01 -3.37390840e-01 -9.24598515e-01 -1.38460672e+00 -4.88585949e-01 1.04799891e+00 -1.74156204e-01 -6.58582211e-01 -6.81409001e-01 -1.01499593e+00 1.70384929e-01 8.51483464e-01 -7.20231056e-01 9.12963226e-02 -6.23962283e-01 -7.46827960e-01 1.03700078e+00 4.37491238e-01 5.46611547e-01 -1.09190583e+00 -2.22950950e-01 9.20228586e-02 -6.68087482e-01 -1.35001194e+00 -6.25825167e-01 1.49556607e-01 -6.72547281e-01 -1.48920548e+00 -5.22998333e-01 -1.04122078e+00 2.77287096e-01 -1.46564335e-01 1.69652677e+00 1.61344297e-02 -4.54953432e-01 3.66467655e-01 -6.42820954e-01 -7.19008327e-01 -5.92278302e-01 4.92349774e-01 -5.18560946e-01 -4.08950418e-01 8.35397422e-01 1.13836996e-01 -2.91172564e-01 5.19048035e-01 -1.31849265e+00 -1.87506914e-01 5.60400903e-01 9.83250022e-01 7.99048126e-01 -1.33050969e-02 7.57151544e-01 -1.60085750e+00 8.48090112e-01 -7.89003491e-01 -5.87454498e-01 7.86800325e-01 -5.95023572e-01 1.39994740e-01 6.02066994e-01 -2.01880991e-01 -1.04393458e+00 -1.05797231e-01 -3.91379476e-01 -9.93703380e-02 -5.18591464e-01 8.18590939e-01 -5.00619233e-01 8.61094296e-01 8.55533004e-01 3.76814246e-01 -4.12100628e-02 -9.51483846e-01 4.64782536e-01 9.50480223e-01 7.81683683e-01 -7.63139367e-01 5.44202924e-01 9.61882994e-02 -7.51803935e-01 -6.12964809e-01 -8.98910046e-01 -5.13900757e-01 -6.59273386e-01 4.42802191e-01 9.52359736e-01 -9.59393024e-01 -5.37586033e-01 4.25993145e-01 -9.52488661e-01 -9.23773050e-02 -1.00879610e-01 6.16313182e-02 -3.65355432e-01 2.57287085e-01 -6.60167277e-01 -3.70729029e-01 -2.05078229e-01 -9.68020618e-01 1.12547243e+00 -1.19535908e-01 1.05002411e-01 -8.89275193e-01 1.26180366e-01 6.63781345e-01 3.03450763e-01 3.79878312e-01 1.19450235e+00 -1.34986985e+00 -4.12080288e-01 -2.98769027e-01 -1.93419009e-01 4.29304153e-01 1.36297315e-01 -1.33041024e-01 -7.96729863e-01 -2.11321250e-01 2.99552977e-01 -7.45512366e-01 5.80133677e-01 -3.20895761e-01 1.71003735e+00 -2.06387743e-01 -1.65780514e-01 4.46452707e-01 1.55977976e+00 2.32926682e-01 6.45294607e-01 2.37084121e-01 3.47560853e-01 7.43555963e-01 8.06088805e-01 3.47549856e-01 7.98811674e-01 4.87471223e-01 7.69022405e-02 -1.35810718e-01 1.91869721e-01 -5.11022866e-01 -2.25228593e-02 1.02325606e+00 7.21703827e-01 -3.68172199e-01 -1.22906613e+00 5.09430707e-01 -1.34447742e+00 -6.50190949e-01 -3.04042906e-01 2.09266901e+00 1.31240487e+00 1.81947216e-01 -1.58723700e-03 -2.98490256e-01 5.20009935e-01 -2.00854808e-01 -5.80011845e-01 -1.32713430e-02 -4.62962747e-01 3.50958586e-01 1.48392633e-01 2.97241062e-02 -1.13712347e+00 8.08500051e-01 6.16251230e+00 8.31530750e-01 -9.26806629e-01 5.73340505e-02 7.37053692e-01 -1.16282338e-02 -3.92832845e-01 -1.20767653e-01 -1.09180510e+00 6.25016809e-01 1.16874731e+00 -1.93226337e-01 -2.89181489e-02 8.95978630e-01 -2.89832085e-01 2.27908660e-02 -1.26207983e+00 8.30834210e-01 1.43919602e-01 -1.21768022e+00 1.35847792e-01 -3.38404715e-01 3.16199541e-01 1.66982040e-01 -3.50560471e-02 8.07667434e-01 1.19055383e-01 -8.63400340e-01 4.87406760e-01 4.00277913e-01 1.21972203e+00 -3.22142839e-01 9.00962770e-01 4.77339596e-01 -9.89461124e-01 9.27375257e-02 -4.79157388e-01 6.99909329e-01 -5.07087931e-02 5.53694487e-01 -9.35097158e-01 1.00837910e+00 1.13868332e+00 5.11251509e-01 -1.27621758e+00 6.29535615e-01 2.35992178e-01 3.79522443e-01 -1.83395296e-01 -2.57140458e-01 -8.50697756e-02 1.51706748e-02 -4.00078930e-02 1.55022359e+00 2.87023038e-01 2.12946862e-01 -1.32271901e-01 1.11827338e+00 -2.16784000e-01 3.40561211e-01 -6.38453066e-01 -1.44701898e-01 6.25581861e-01 9.51209605e-01 -1.56676441e-01 -6.87038839e-01 -8.12699556e-01 6.54412210e-01 3.40472192e-01 4.46033269e-01 -9.42685008e-01 -9.29373205e-01 4.89869446e-01 1.07833847e-01 2.71385521e-01 1.28542364e-01 -2.36995816e-01 -1.57681370e+00 5.52338839e-01 -1.43781579e+00 1.00430405e+00 -6.58371508e-01 -1.83261263e+00 8.75788927e-01 8.53449330e-02 -1.19021273e+00 -1.87224865e-01 -6.18160844e-01 -1.52887881e-01 8.65285814e-01 -1.52869320e+00 -8.95841002e-01 -6.03564620e-01 1.13629270e+00 4.98142332e-01 -3.22574675e-01 8.84599686e-01 7.69657731e-01 -7.09916651e-01 1.03237462e+00 2.90135950e-01 8.20890069e-01 1.28601718e+00 -1.33619606e+00 7.02149928e-01 4.86574203e-01 -7.44566098e-02 1.07372057e+00 2.80393034e-01 -8.38166952e-01 -1.58376908e+00 -1.30957973e+00 1.18175101e+00 -1.14729893e+00 5.01229227e-01 -9.05889690e-01 -1.31940973e+00 7.41945505e-01 1.40083298e-01 -1.21105246e-01 8.48707199e-01 -6.40763268e-02 -3.92455846e-01 -1.41284510e-01 -1.21663904e+00 3.69538903e-01 8.10514569e-01 -7.51298964e-01 -1.00935960e+00 3.89502496e-01 1.16697729e+00 -6.84869051e-01 -1.21666706e+00 4.92199749e-01 2.36779973e-01 -1.01559138e+00 9.10638452e-01 -1.20376742e+00 4.95203465e-01 -1.29364096e-02 -5.33318579e-01 -6.33172154e-01 3.17544580e-01 -3.47949326e-01 3.13838691e-01 1.38821840e+00 8.75631154e-01 -7.75243223e-01 8.67452681e-01 8.03657353e-01 -2.13556647e-01 -7.88518727e-01 -8.30357492e-01 -6.95472836e-01 5.04073322e-01 -5.94046831e-01 9.89469886e-01 1.14529836e+00 -1.10581331e-01 2.16732055e-01 1.05028726e-01 -2.29104348e-02 2.87512839e-01 2.62980968e-01 9.33009207e-01 -7.94218063e-01 -2.06805795e-01 5.56234755e-02 3.01570207e-01 -1.15158105e+00 -3.84895355e-01 -9.84132409e-01 -6.05707243e-02 -9.85615253e-01 2.75610954e-01 -7.33581424e-01 -2.94424612e-02 3.87362748e-01 -4.04795527e-01 -4.08155531e-01 5.53112254e-02 8.46518949e-02 -5.94743133e-01 3.94348264e-01 1.06257331e+00 8.91275853e-02 -1.01569658e-02 9.90388468e-02 -9.09342051e-01 8.61543566e-02 6.96987629e-01 -6.33297086e-01 -3.75462055e-01 -5.74748099e-01 1.19854636e-01 3.94267887e-01 4.51317336e-03 -7.46381462e-01 3.30169976e-01 -7.36279711e-02 2.37581685e-01 -1.02259195e+00 8.28351453e-02 -7.38569021e-01 -1.51436761e-01 3.88415903e-01 -7.76321173e-01 6.06464088e-01 1.43504292e-01 4.71052617e-01 -5.77673852e-01 -1.46694094e-01 5.20755649e-01 -3.51942599e-01 -4.76521581e-01 1.42895460e-01 5.61989099e-02 1.05691791e+00 7.69115388e-01 2.17999101e-01 -6.59002006e-01 -2.40128428e-01 -7.02099502e-01 5.52690148e-01 1.73739910e-01 8.60407650e-01 7.31523693e-01 -1.27645504e+00 -7.59557188e-01 4.65103716e-01 7.89742410e-01 1.74990952e-01 1.33248806e-01 5.10677218e-01 -6.73453689e-01 9.53618586e-01 1.42816037e-01 -7.00868845e-01 -1.02981508e+00 9.49752748e-01 2.04686120e-01 -3.26993912e-01 -2.00593755e-01 5.91100216e-01 6.37165084e-02 -1.28206992e+00 3.72709036e-01 -8.14268947e-01 1.04361009e-02 -8.81205052e-02 4.23458695e-01 1.00143448e-01 4.25512224e-01 1.31004304e-01 -3.12233508e-01 2.78410345e-01 -2.21414849e-01 3.31411362e-02 1.26236749e+00 -1.33428976e-01 -1.86156839e-01 4.24163252e-01 1.63844240e+00 9.30661429e-03 -4.03171539e-01 -6.19519770e-01 5.47361195e-01 -3.19606364e-01 -7.91960120e-01 -1.07371390e+00 -9.87762928e-01 8.00048470e-01 5.18673182e-01 2.47592896e-01 1.32377172e+00 1.95257619e-01 9.52278316e-01 6.93351865e-01 2.23210096e-01 -1.11551332e+00 2.28900313e-01 5.04708588e-01 1.03313637e+00 -1.71532488e+00 -3.19950283e-01 -4.42515224e-01 -9.89787400e-01 1.07957757e+00 8.33836198e-01 4.40877974e-01 8.11135828e-01 3.60184997e-01 3.95259172e-01 -3.73089194e-01 -9.71698344e-01 2.71172710e-02 1.94658235e-01 4.33832616e-01 6.17773056e-01 -1.51815429e-01 -8.37175697e-02 1.13115811e+00 -5.05460739e-01 1.15297124e-01 4.70370770e-01 1.15913725e+00 4.26127724e-02 -1.36281908e+00 -2.76934206e-01 6.49940431e-01 -6.16159141e-01 -3.03129852e-01 -4.60465491e-01 1.19292879e+00 -4.42518368e-02 1.08525300e+00 -1.29157022e-01 -3.28876346e-01 9.09222364e-01 2.71551967e-01 -2.32143641e-01 -7.31163442e-01 -8.40721905e-01 -4.16219197e-02 2.70672113e-01 -6.02133334e-01 1.52715564e-01 -6.72048092e-01 -1.12266874e+00 -1.69436067e-01 -1.00333080e-01 2.49913722e-01 5.15687466e-01 2.98239887e-01 6.32947445e-01 2.72042632e-01 4.23942685e-01 5.59708178e-01 -8.15291464e-01 -9.11126316e-01 -3.99618000e-01 8.07109177e-01 2.11593360e-01 -7.66278133e-02 -1.89868398e-02 1.28687486e-01]
[9.885321617126465, 7.834498882293701]
6edd797e-81db-43d3-87d6-93ad5bc6bebb
graph-edit-distance-computation-via-graph
1808.05689
null
https://arxiv.org/abs/1808.05689v4
https://arxiv.org/pdf/1808.05689v4.pdf
SimGNN: A Neural Network Approach to Fast Graph Similarity Computation
Graph similarity search is among the most important graph-based applications, e.g. finding the chemical compounds that are most similar to a query compound. Graph similarity computation, such as Graph Edit Distance (GED) and Maximum Common Subgraph (MCS), is the core operation of graph similarity search and many other applications, but very costly to compute in practice. Inspired by the recent success of neural network approaches to several graph applications, such as node or graph classification, we propose a novel neural network based approach to address this classic yet challenging graph problem, aiming to alleviate the computational burden while preserving a good performance. The proposed approach, called SimGNN, combines two strategies. First, we design a learnable embedding function that maps every graph into a vector, which provides a global summary of a graph. A novel attention mechanism is proposed to emphasize the important nodes with respect to a specific similarity metric. Second, we design a pairwise node comparison method to supplement the graph-level embeddings with fine-grained node-level information. Our model achieves better generalization on unseen graphs, and in the worst case runs in quadratic time with respect to the number of nodes in two graphs. Taking GED computation as an example, experimental results on three real graph datasets demonstrate the effectiveness and efficiency of our approach. Specifically, our model achieves smaller error rate and great time reduction compared against a series of baselines, including several approximation algorithms on GED computation, and many existing graph neural network based models. To the best of our knowledge, we are among the first to adopt neural networks to explicitly model the similarity between two graphs, and provide a new direction for future research on graph similarity computation and graph similarity search.
['Yunsheng Bai', 'Ting Chen', 'Hao Ding', 'Yizhou Sun', 'Wei Wang', 'Song Bian']
2018-08-16
simgnn-a-neural-network-approach-to-fast
https://doi.org/10.1145/3289600.3290967
http://web.cs.ucla.edu/~yzsun/papers/2019_WSDM_SimGNN.pdf
wsdm-19-proceedings-of-the-twelfth-acm
['graph-similarity']
['graphs']
[ 2.27837667e-01 -4.10041213e-02 -2.22343788e-01 -3.63509983e-01 -2.74227887e-01 -5.16706705e-01 3.94336611e-01 9.54741418e-01 -3.85703087e-01 3.22747439e-01 -1.13040328e-01 -4.55637753e-01 -4.49003696e-01 -1.30666900e+00 -5.80054343e-01 -5.60877681e-01 -1.80147409e-01 2.89816380e-01 1.41775131e-01 -1.86145440e-01 3.36907089e-01 5.69388986e-01 -1.00053132e+00 -4.47621852e-01 1.00708818e+00 9.73004162e-01 3.28595877e-01 2.51439810e-01 -1.75139308e-01 4.22512442e-01 -2.19378129e-01 -5.96626878e-01 1.49938926e-01 -2.93061703e-01 -7.72821367e-01 -1.61803707e-01 4.61818993e-01 5.11271283e-02 -7.53566146e-01 1.52646458e+00 6.73273683e-01 4.87874210e-01 4.50285703e-01 -1.31887317e+00 -1.12165189e+00 6.04611278e-01 -6.01150751e-01 2.88640976e-01 4.29167986e-01 -1.02502974e-02 1.31266296e+00 -6.52972460e-01 5.56315899e-01 1.14224327e+00 6.35350585e-01 2.13487834e-01 -1.13785481e+00 -6.19424939e-01 4.22960758e-01 4.91068065e-01 -1.53299642e+00 1.41959488e-01 1.02527964e+00 -2.22770214e-01 8.92652512e-01 2.51254827e-01 5.80547690e-01 6.10862017e-01 2.37981707e-01 3.11274379e-01 5.37807524e-01 -1.86848804e-01 3.01614672e-01 -3.11468035e-01 3.47856760e-01 9.36530113e-01 5.18830597e-01 -2.02620879e-01 -5.25686480e-02 -2.61688054e-01 4.95310277e-01 4.51219618e-01 -5.27557373e-01 -5.89682281e-01 -1.03082037e+00 1.07661498e+00 1.06372738e+00 5.15091956e-01 -3.15902859e-01 3.22370172e-01 4.77533042e-01 3.78537685e-01 4.60909694e-01 6.79205179e-01 -7.20665902e-02 3.89112502e-01 -4.30576921e-01 1.30104087e-02 8.63674641e-01 9.04264629e-01 8.98173749e-01 -8.40725824e-02 -1.14504993e-01 6.50368869e-01 2.51617253e-01 9.15726274e-02 6.30560517e-01 -4.01054084e-01 4.77554381e-01 1.03139174e+00 -5.32956600e-01 -1.71782005e+00 -3.75797093e-01 -4.28743064e-01 -1.10759485e+00 -3.09553802e-01 9.31886584e-02 2.21348241e-01 -7.21859574e-01 1.79300928e+00 4.70974892e-01 3.64792496e-01 -2.14338541e-01 7.12907195e-01 1.16054058e+00 5.51499844e-01 -1.45854391e-02 -6.60248175e-02 1.18968499e+00 -9.67721224e-01 -5.03045857e-01 -1.07566655e-01 8.60018194e-01 -4.84520167e-01 8.94970357e-01 -7.32549429e-02 -8.82722139e-01 -4.15961981e-01 -1.11678433e+00 -1.02279708e-01 -7.85884380e-01 -3.75028133e-01 8.78571928e-01 5.09380937e-01 -1.09625304e+00 1.19396222e+00 -6.39448345e-01 -6.16598427e-01 3.24128479e-01 5.60283124e-01 -5.60670137e-01 -3.09288532e-01 -1.35559142e+00 6.12288296e-01 7.39398181e-01 1.98014840e-01 -4.51673657e-01 -5.96025765e-01 -1.23364198e+00 3.30378741e-01 6.51824236e-01 -6.18778467e-01 7.22213626e-01 -5.96354902e-01 -9.88119543e-01 5.20429313e-01 -7.34261125e-02 -4.01708901e-01 -1.34602979e-01 3.14084023e-01 -5.35781622e-01 6.90108016e-02 -9.73031111e-03 3.59822273e-01 4.60202783e-01 -7.88165331e-01 -2.40351811e-01 -5.59694469e-01 3.01754922e-01 1.92630053e-01 -8.42932940e-01 -1.51424661e-01 -6.26513302e-01 -7.12390363e-01 1.84288859e-01 -8.85142207e-01 -5.08159697e-01 9.92227867e-02 -4.11398232e-01 -6.68183565e-01 5.05015492e-01 -5.67312479e-01 1.46052217e+00 -1.85563457e+00 3.32750380e-01 5.30233800e-01 8.19052279e-01 3.48960280e-01 -5.11740804e-01 7.18909025e-01 -4.28762883e-01 1.98947653e-01 -4.35249120e-01 9.28591117e-02 6.11476265e-02 2.59426199e-02 1.00447752e-01 5.31533301e-01 1.64418355e-01 1.27947819e+00 -1.22032773e+00 -2.27467179e-01 4.37070988e-03 3.88945907e-01 -3.70672673e-01 1.62648916e-01 -2.87287775e-02 -4.01311107e-02 -5.92392743e-01 4.76556778e-01 7.54876077e-01 -6.38912320e-01 4.34874237e-01 -4.67064947e-01 3.86562109e-01 1.07000411e-01 -1.17340648e+00 1.76032019e+00 -2.91905493e-01 2.84944117e-01 -2.99728304e-01 -1.48199546e+00 1.08675289e+00 -7.89748579e-02 3.23634744e-01 -7.11995780e-01 1.77661762e-01 2.06363752e-01 2.34749049e-01 -1.90741301e-01 4.66519475e-01 2.89315969e-01 -8.48437324e-02 4.77432787e-01 5.61391423e-03 7.51417950e-02 2.40876868e-01 5.39342761e-01 1.21928859e+00 -3.91054541e-01 6.56560361e-01 -2.04797924e-01 8.09150815e-01 -4.13955450e-01 3.74676287e-01 3.92447680e-01 -1.89048812e-01 2.08946139e-01 5.05215466e-01 -5.96716642e-01 -6.23024464e-01 -6.12217844e-01 4.34729278e-01 8.22505891e-01 5.62649786e-01 -7.00741231e-01 -7.33778894e-01 -6.75733149e-01 1.58644974e-01 2.70232528e-01 -7.00622380e-01 -7.14007914e-01 -5.76462269e-01 -5.49139857e-01 1.97442383e-01 4.77198154e-01 3.84881318e-01 -8.90171170e-01 1.92519218e-01 3.22631806e-01 2.30400741e-01 -1.02106524e+00 -1.10291195e+00 9.11060348e-02 -8.16140234e-01 -1.26277614e+00 -5.83325684e-01 -1.24793720e+00 7.54780710e-01 6.82168961e-01 1.10243559e+00 4.74409610e-01 -3.04093659e-01 2.14906767e-01 -3.30201745e-01 3.72519530e-02 -1.40890762e-01 1.36203185e-01 -4.47730683e-02 8.31879154e-02 4.36098397e-01 -7.72675931e-01 -6.35422766e-01 1.50823742e-01 -1.05165648e+00 -3.76763225e-01 5.06033719e-01 8.13038588e-01 7.17982769e-01 2.56395698e-01 4.72958624e-01 -9.50934052e-01 1.15697241e+00 -5.93822539e-01 -6.92073703e-01 4.69209731e-01 -9.52323616e-01 2.78920978e-01 1.06102729e+00 -3.35815459e-01 -3.47390085e-01 -1.05907686e-01 4.86800782e-02 -4.69504356e-01 1.73313096e-01 9.56959069e-01 -3.25928688e-01 -6.49017334e-01 4.26724672e-01 4.87864316e-01 -9.79846716e-02 -3.66009742e-01 5.25245488e-01 3.49586964e-01 4.05522406e-01 -3.20277572e-01 9.79670882e-01 1.81273639e-01 4.57585871e-01 -5.97771347e-01 -3.15151781e-01 -5.96588850e-01 -4.79140967e-01 2.10997269e-01 6.22805953e-01 -6.55409992e-01 -1.01280892e+00 1.82488754e-01 -1.21701670e+00 2.53103003e-02 1.62340835e-01 4.10984486e-01 -1.68435410e-01 1.11570752e+00 -4.64188159e-01 -3.17468882e-01 -6.92217290e-01 -1.05596423e+00 8.13615024e-01 2.26307198e-01 -8.30747001e-03 -1.32782257e+00 1.57721624e-01 1.73939280e-02 1.87325567e-01 4.43911225e-01 1.10359812e+00 -1.04846668e+00 -6.75259769e-01 -4.10952628e-01 -5.38838565e-01 5.36647998e-02 4.17125136e-01 -3.95996153e-01 -4.29832697e-01 -7.14629114e-01 -3.26254874e-01 -6.32534027e-02 9.10400033e-01 9.72938538e-02 1.46286249e+00 -2.58810937e-01 -6.27743423e-01 7.38081396e-01 1.62106264e+00 3.38123232e-01 3.37955654e-01 1.94021109e-02 1.18247139e+00 3.41161400e-01 3.67356747e-01 1.51628152e-01 4.58673567e-01 6.43797696e-01 6.56024456e-01 -1.58745185e-01 -4.55352142e-02 -4.80345964e-01 -4.90437895e-02 1.16529846e+00 3.54057513e-02 -5.40758550e-01 -5.68803012e-01 4.04092163e-01 -1.81299138e+00 -7.27143526e-01 -4.33379114e-02 2.29799962e+00 5.21866500e-01 -1.66469827e-01 -6.41950071e-02 3.88881341e-02 1.03478801e+00 3.87290031e-01 -7.13042080e-01 -5.22174597e-01 1.91164762e-01 4.46206778e-01 5.15537739e-01 4.05651480e-01 -9.74158168e-01 8.60658169e-01 4.91089678e+00 1.02842331e+00 -1.02937829e+00 -1.81094915e-01 4.45694774e-01 3.84335309e-01 -5.10140002e-01 -5.51956380e-03 -3.34473789e-01 5.50505698e-01 7.24986553e-01 -6.18561685e-01 7.63445020e-01 8.21478665e-01 -6.44485876e-02 4.95612860e-01 -1.33264971e+00 1.13801289e+00 3.73781174e-01 -1.40402925e+00 2.38399163e-01 9.16811600e-02 4.73549545e-01 -1.09137431e-01 -2.07252055e-01 1.77309409e-01 1.97480559e-01 -1.03454280e+00 1.94022879e-02 2.08931848e-01 6.01543486e-01 -9.53861833e-01 6.77041113e-01 3.65899578e-02 -1.87275767e+00 1.35932654e-01 -5.86387813e-01 1.54803902e-01 -8.81206244e-02 4.21491146e-01 -7.28822529e-01 9.90391552e-01 4.28989828e-01 1.08908248e+00 -6.68658793e-01 1.20170212e+00 -1.94704130e-01 1.32291600e-01 -1.40725235e-02 -4.76914138e-01 4.81336832e-01 -5.19977927e-01 4.81958836e-01 8.20184290e-01 3.37413788e-01 3.89478542e-03 3.97517413e-01 8.36364031e-01 -5.95173061e-01 3.90308022e-01 -9.25847352e-01 -3.87906671e-01 6.54462039e-01 1.45727050e+00 -7.67855406e-01 -2.30947360e-01 -4.03846681e-01 1.22508037e+00 8.06255162e-01 1.91987991e-01 -7.89674103e-01 -1.14622664e+00 5.16923726e-01 -1.50122359e-01 3.95050466e-01 -2.50253469e-01 2.24742040e-01 -9.39201772e-01 2.24395350e-01 -7.97011733e-01 6.57639205e-01 -4.66251820e-01 -1.47022617e+00 8.01510096e-01 -3.58126938e-01 -1.12277591e+00 1.03093654e-01 -6.77798748e-01 -8.97960484e-01 7.47740507e-01 -1.59180665e+00 -9.90110874e-01 -5.59165061e-01 6.35160387e-01 1.92851201e-01 -9.16301161e-02 8.51928890e-01 4.74291652e-01 -6.47014618e-01 9.12938416e-01 1.77054822e-01 1.94035426e-01 4.44524884e-01 -1.20513165e+00 9.38219726e-01 7.29524255e-01 4.04933363e-01 9.20463026e-01 2.26506501e-01 -6.10496163e-01 -1.79415417e+00 -1.35671723e+00 1.00578737e+00 -2.79717259e-02 9.15955663e-01 -3.12162161e-01 -1.17931128e+00 4.34662491e-01 8.16630647e-02 1.70460656e-01 6.63039327e-01 -4.72047180e-02 -4.20412213e-01 -2.37846255e-01 -9.33219612e-01 6.80592477e-01 1.41564381e+00 -7.09920049e-01 -3.13294888e-01 6.19534194e-01 1.21976352e+00 -2.01282933e-01 -1.24965501e+00 3.16478431e-01 2.81659603e-01 -5.86473107e-01 1.05766237e+00 -8.61592412e-01 8.35275799e-02 -4.09144670e-01 -1.00599281e-01 -1.52728760e+00 -6.53128862e-01 -7.64303625e-01 -1.06668927e-01 9.86240447e-01 2.57658869e-01 -9.97340679e-01 7.94542491e-01 4.14395809e-01 -1.14548407e-01 -9.80339050e-01 -7.05730975e-01 -1.07425094e+00 -1.05283082e-01 6.98482916e-02 1.02019882e+00 1.27953720e+00 8.54424760e-02 4.13185507e-01 -1.81232393e-01 1.71645477e-01 5.24024487e-01 4.73881543e-01 6.69947803e-01 -1.34231973e+00 -2.71404088e-01 -6.43719137e-01 -9.26594138e-01 -9.74287987e-01 3.69768709e-01 -1.48102808e+00 -2.39035890e-01 -1.65195668e+00 3.00268203e-01 -1.26447216e-01 -5.54384768e-01 3.73594910e-01 -4.40591186e-01 -2.71088968e-04 1.08927175e-01 -1.66981921e-01 -5.35560071e-01 7.56069064e-01 1.22824955e+00 -5.59567809e-01 -1.46738529e-01 -2.08221078e-01 -1.09911048e+00 5.28537929e-01 8.37709248e-01 -4.91271883e-01 -6.17893636e-01 -4.68415260e-01 1.34654790e-01 -1.73497409e-01 2.01519862e-01 -6.82165980e-01 5.93458533e-01 -1.35885417e-01 -7.30269700e-02 -2.67096460e-01 9.37166885e-02 -8.30736458e-01 1.90929070e-01 7.60646105e-01 -2.24399760e-01 3.93854916e-01 -4.13402691e-02 1.05452645e+00 -3.55683893e-01 -1.69316679e-01 5.84915698e-01 -6.31188676e-02 -8.74723971e-01 1.10041308e+00 3.20873201e-01 -4.62633856e-02 9.89512920e-01 -2.77809381e-01 -3.23987395e-01 -3.20724070e-01 -4.25385416e-01 3.03329438e-01 3.47292781e-01 4.90584642e-01 8.16598356e-01 -1.67229688e+00 -4.22480136e-01 -9.66505557e-02 4.05037791e-01 -1.47537425e-01 2.51072310e-02 6.36840463e-01 -4.23748940e-01 4.47549254e-01 5.23202717e-02 -2.08078936e-01 -1.36092532e+00 9.42405879e-01 1.89233452e-01 -4.17245001e-01 -4.02439177e-01 9.00074303e-01 3.04822862e-01 -5.06381214e-01 2.11570904e-01 -2.93922037e-01 -2.52003431e-01 -1.53430223e-01 3.96280110e-01 4.20303851e-01 1.60468116e-01 -5.44663846e-01 -5.52872181e-01 8.56941879e-01 -2.12744832e-01 7.50816166e-01 1.24301553e+00 1.85751274e-01 -3.62397671e-01 -1.28215030e-01 1.70231259e+00 -3.19177240e-01 -5.02887130e-01 -4.94298071e-01 9.69917923e-02 -4.83167589e-01 5.78773618e-02 -1.90664247e-01 -1.32418275e+00 8.33530366e-01 4.77881879e-01 3.23650360e-01 1.10531676e+00 -6.31742701e-02 1.04636312e+00 8.17816973e-01 1.99425876e-01 -7.76683092e-01 1.17269777e-01 2.02494130e-01 8.36965442e-01 -1.32001007e+00 1.48130491e-01 -6.09202623e-01 -2.28103429e-01 1.11836863e+00 6.56982541e-01 -3.04630518e-01 6.49538040e-01 -4.57373977e-01 -5.75487912e-01 -4.93595809e-01 -4.42814976e-01 -1.90736309e-01 5.55593729e-01 5.49008608e-01 3.67010474e-01 4.66260947e-02 -5.56268632e-01 3.78435612e-01 1.80548191e-01 -2.04440773e-01 1.97901145e-01 7.49632418e-01 -4.48199838e-01 -1.20271349e+00 2.81065822e-01 6.29042029e-01 -2.39343978e-02 -4.24323916e-01 -6.41621232e-01 6.99421465e-01 -1.37610137e-01 7.68366635e-01 -1.08123794e-01 -6.02770090e-01 3.82596523e-01 -4.21106309e-01 3.18317354e-01 -5.87229431e-01 -5.16504288e-01 -4.81351495e-01 -1.28210008e-01 -6.50232077e-01 -1.69195652e-01 -1.40403956e-01 -1.20300937e+00 -4.86071259e-01 -6.11094713e-01 2.48173937e-01 3.71786684e-01 6.14065886e-01 5.98592699e-01 3.87211174e-01 9.15010691e-01 -4.92761582e-01 -5.52412808e-01 -6.63323343e-01 -7.65637159e-01 6.86771333e-01 -1.20727457e-02 -4.84800816e-01 -3.20278436e-01 -4.61359710e-01]
[7.168111801147461, 6.259625434875488]
23cbd5fe-79e9-4e0e-9e94-38c2b38b774f
injecting-numerical-reasoning-skills-into-1
2112.06109
null
https://arxiv.org/abs/2112.06109v2
https://arxiv.org/pdf/2112.06109v2.pdf
Injecting Numerical Reasoning Skills into Knowledge Base Question Answering Models
Embedding-based methods are popular for Knowledge Base Question Answering (KBQA), but few current models have numerical reasoning skills and thus struggle to answer ordinal constrained questions. This paper proposes a new embedding-based KBQA framework which particularly takes numerical reasoning into account. We present NumericalTransformer on top of NSM, a state-of-the-art embedding-based KBQA model, to create NT-NSM. To enable better training, we propose two pre-training tasks with explicit numerical-oriented loss functions on two generated training datasets and a template-based data augmentation method for enriching ordinal constrained QA dataset. Extensive experiments on KBQA benchmarks demonstrate that with the help of our training algorithm, NT-NSM is empowered with numerical reasoning skills and substantially outperforms the baselines in answering ordinal constrained questions.
['Hong Chen', 'Cuiping Li', 'Lemao Liu', 'Xiaokang Zhang', 'Jing Zhang', 'Yu Feng']
2021-12-12
null
null
null
null
['knowledge-base-question-answering']
['natural-language-processing']
[-2.94569939e-01 4.74015713e-01 -1.49604008e-01 -4.94475991e-01 -1.38956892e+00 -5.57812572e-01 2.86508322e-01 1.81138575e-01 -6.63840353e-01 8.03345084e-01 4.66036677e-01 -7.19692647e-01 -4.50013340e-01 -1.29395413e+00 -6.54538274e-01 5.85613074e-03 1.24252550e-01 1.25205493e+00 2.01378688e-01 -9.33695912e-01 6.81692809e-02 1.08324476e-01 -9.99198318e-01 5.69024205e-01 1.40569675e+00 1.28469360e+00 -4.21422988e-01 8.47774863e-01 -5.17573476e-01 1.28665650e+00 -6.44767284e-01 -1.40211737e+00 1.93341166e-01 -8.89910311e-02 -1.50516355e+00 -7.88493633e-01 8.48139882e-01 -8.04030001e-01 -4.87417072e-01 4.84490871e-01 6.36170566e-01 4.41020817e-01 6.68840170e-01 -1.21971381e+00 -1.41016746e+00 6.05819404e-01 4.19553280e-01 1.88922867e-01 6.71182692e-01 3.45576465e-01 1.72750723e+00 -8.66882861e-01 3.44278157e-01 1.33050561e+00 8.77701104e-01 8.96941602e-01 -1.00600088e+00 -1.37205586e-01 -3.09360921e-01 8.95553887e-01 -8.34577918e-01 -3.66770066e-02 9.58914936e-01 8.29473287e-02 1.32302320e+00 4.03122693e-01 5.74463189e-01 8.76772583e-01 -3.79844338e-01 9.86092031e-01 8.45434487e-01 -3.88184518e-01 2.71729320e-01 -1.32506415e-01 2.79734462e-01 9.07255232e-01 -1.00808769e-01 -3.24401796e-01 -4.68575478e-01 -4.80835319e-01 4.93949950e-01 -6.33601606e-01 -8.71142745e-02 -5.59495091e-01 -1.15624237e+00 1.29903066e+00 6.73429370e-01 -9.46891084e-02 -2.97450185e-01 5.42027354e-01 5.90471625e-01 7.32950747e-01 6.80845007e-02 1.13801515e+00 -9.21607316e-01 -4.12608802e-01 -4.65469122e-01 8.81323874e-01 9.76527870e-01 5.70945621e-01 5.88381529e-01 -1.13406710e-01 -7.97145903e-01 9.13242757e-01 6.84277117e-02 4.67412531e-01 4.11168158e-01 -1.50002706e+00 9.38307643e-01 1.00427854e+00 2.11478189e-01 -6.02390468e-01 -1.94040373e-01 -2.11393964e-02 -4.62095380e-01 -1.31084561e-01 3.88076872e-01 1.33050367e-01 -7.33722150e-01 1.47542334e+00 4.37099069e-01 -1.44886225e-02 5.93881547e-01 9.26780403e-01 1.08974826e+00 6.91681623e-01 1.44953042e-01 2.84414977e-01 1.55869734e+00 -1.29865026e+00 -9.07999754e-01 -4.62209843e-02 9.84362304e-01 -1.00687407e-01 1.89539599e+00 4.61203754e-01 -1.41020775e+00 -3.76766831e-01 -8.38569164e-01 -9.41160262e-01 -7.21349359e-01 -9.90336090e-02 9.94274139e-01 7.85448432e-01 -1.11395395e+00 2.05753282e-01 -3.51546407e-01 3.26009274e-01 4.79332209e-01 3.30707073e-01 -3.44474554e-01 -2.93546677e-01 -1.94442260e+00 1.37976289e+00 5.14025450e-01 -4.82646562e-02 -6.48230851e-01 -1.10459566e+00 -1.18948317e+00 2.64343232e-01 4.33086932e-01 -1.33831346e+00 1.65439296e+00 3.13889608e-02 -1.70411074e+00 4.02338803e-01 1.35040805e-01 -6.33455813e-01 3.98361474e-01 -4.09798622e-01 -1.77624315e-01 4.65581834e-01 3.36138159e-02 8.87426138e-01 5.05085826e-01 -8.51746619e-01 -1.57634765e-01 -1.19135648e-01 8.08404267e-01 4.85626489e-01 -6.66028500e-01 -4.09699678e-01 -3.92957956e-01 -3.82385910e-01 -3.79285723e-01 -2.46578053e-01 -2.06163689e-01 1.48050115e-01 1.95832878e-01 -9.10207987e-01 5.17958403e-01 -1.05654705e+00 1.39819431e+00 -1.44887865e+00 3.09281588e-01 3.08733247e-02 1.45917209e-02 3.78561825e-01 -5.58118343e-01 4.75511938e-01 5.18391788e-01 -2.46825907e-02 -4.24909979e-01 -3.02805692e-01 5.98179996e-01 6.94137216e-01 -5.55718362e-01 -3.83741170e-01 7.74093091e-01 1.63519073e+00 -1.26708651e+00 -8.14361572e-01 4.61553670e-02 1.43821791e-01 -1.09499085e+00 4.09671545e-01 -9.40867007e-01 -9.45309997e-02 -3.07995945e-01 7.31423914e-01 5.71572483e-01 -1.22104242e-01 2.45615058e-02 -2.85912871e-01 6.10954046e-01 7.41734445e-01 -8.19662690e-01 2.17479229e+00 -8.19696307e-01 2.60221899e-01 -4.06794667e-01 -9.07733858e-01 6.67790234e-01 3.17942679e-01 1.84883119e-03 -9.72557008e-01 -1.71402961e-01 3.58981013e-01 -1.08401977e-01 -6.15185678e-01 1.06773925e+00 -2.29981896e-02 -3.73924404e-01 3.21736634e-01 3.51594448e-01 -9.98262525e-01 3.52521420e-01 6.14968598e-01 1.23020339e+00 -6.00126460e-02 -1.67296022e-01 1.79255754e-01 7.58323312e-01 1.21361315e-01 1.06145822e-01 6.93252206e-01 -1.96442217e-01 1.23968340e-01 4.93616700e-01 -4.68699068e-01 -9.44843292e-01 -1.29196548e+00 2.31456496e-02 1.29627264e+00 -8.47574323e-02 -3.85499775e-01 -4.02010053e-01 -1.05797887e+00 2.85696566e-01 9.73560810e-01 -7.49474883e-01 -3.57677311e-01 -7.68425465e-01 -3.21886182e-01 1.06998432e+00 8.80695820e-01 6.40964687e-01 -1.27767432e+00 -4.41012502e-01 3.16392094e-01 -5.01315176e-01 -1.22608495e+00 -6.35315999e-02 -1.01815470e-01 -9.52979088e-01 -8.71878803e-01 -5.66367090e-01 -8.03938150e-01 2.41087317e-01 -4.75747585e-01 1.71644235e+00 1.97816357e-01 4.03149873e-02 7.74837315e-01 -4.83373314e-01 -3.17674041e-01 -2.98698276e-01 2.51227319e-01 -2.59371549e-01 -7.48167336e-01 3.11736733e-01 -2.49839917e-01 -6.03752553e-01 -1.40381530e-01 -9.93548274e-01 -3.70649189e-01 6.24727786e-01 1.08629823e+00 1.30507961e-01 -3.70750219e-01 7.89304793e-01 -5.43229878e-01 1.26121712e+00 -2.85298586e-01 -4.10127729e-01 6.16208732e-01 -5.43750584e-01 3.42455626e-01 8.24698269e-01 -3.54259431e-01 -1.00195229e+00 -6.52258515e-01 -5.61835647e-01 -1.88412920e-01 3.62814844e-01 7.77330518e-01 -1.47431031e-01 -4.70545664e-02 8.47812474e-01 1.78901404e-01 -5.39028533e-02 -2.52175510e-01 1.14661944e+00 5.75223207e-01 7.92437911e-01 -1.19843602e+00 9.03968930e-01 2.01131895e-01 -2.01824278e-01 -2.69666612e-01 -1.07075107e+00 -5.03650427e-01 -3.06486487e-01 2.20585689e-01 7.20797420e-01 -7.57355094e-01 -9.42850173e-01 1.40106127e-01 -1.31617618e+00 -6.86821222e-01 -6.31513596e-01 1.82526425e-01 -8.19837272e-01 5.73277175e-01 -8.32709610e-01 -7.77390540e-01 -8.85122597e-01 -7.98072398e-01 1.00695908e+00 -2.86568906e-02 -1.37060508e-01 -1.31681967e+00 3.59516054e-01 1.31897604e+00 7.42546678e-01 -2.41562888e-01 1.43431926e+00 -6.07394278e-01 -4.71306264e-01 -8.00006688e-02 -3.74917805e-01 6.58795416e-01 -2.24748254e-01 -5.32690465e-01 -7.20962644e-01 2.21330076e-01 -3.60847026e-01 -1.29836345e+00 8.21678638e-01 -1.77496642e-01 1.36251521e+00 -5.80483437e-01 5.87073207e-01 3.50673437e-01 1.14535034e+00 -3.44709963e-01 9.89143610e-01 6.69785261e-01 4.85045016e-01 3.36079806e-01 6.73105955e-01 2.30683818e-01 1.12157822e+00 5.83002627e-01 6.80038154e-01 2.59941369e-01 3.16794105e-02 -1.95411220e-01 1.42742440e-01 1.03542101e+00 1.71980727e-02 -2.94271167e-02 -9.91814315e-01 9.74706411e-01 -1.72633338e+00 -7.38368332e-01 1.38332441e-01 1.48549891e+00 1.75028217e+00 -8.79524127e-02 -3.77378240e-02 2.04861373e-01 -4.37652916e-01 3.87532986e-03 -6.21291280e-01 -9.75177944e-01 1.22151030e-02 9.28405762e-01 2.26017330e-02 6.37090743e-01 -8.55985522e-01 1.08711565e+00 6.14561605e+00 9.14137125e-01 -2.69495904e-01 9.17976424e-02 2.54351765e-01 3.73643160e-01 -8.82572770e-01 -2.19562009e-01 -4.72897559e-01 5.61457127e-02 1.01579261e+00 1.33931875e-01 5.80763817e-01 8.01375687e-01 -6.26145482e-01 1.97483152e-01 -1.00469708e+00 7.22270787e-01 3.84298549e-03 -1.55352271e+00 7.55033731e-01 -4.76964056e-01 7.03738391e-01 -4.56187367e-01 3.87842476e-01 1.23041630e+00 4.38555598e-01 -1.29772544e+00 3.59526992e-01 5.07264197e-01 7.77533233e-01 -6.81382954e-01 1.08504534e+00 1.78280354e-01 -8.05091560e-01 -3.27389181e-01 -5.69054902e-01 -2.65885442e-01 1.66575596e-01 1.77163258e-01 -9.51402485e-01 9.83892024e-01 5.62729597e-01 -5.62865846e-02 -7.06363440e-01 4.91998285e-01 -5.34988880e-01 6.77212238e-01 -9.83569920e-02 -3.21718574e-01 4.84129936e-01 -1.78888440e-01 -9.14708748e-02 8.80907357e-01 -7.67499432e-02 3.95031035e-01 -3.67828697e-01 9.04534459e-01 -4.17080820e-01 8.74073207e-02 4.11352739e-02 -2.36509785e-01 4.55993831e-01 1.18177271e+00 3.71134520e-01 -5.87918758e-01 -5.03354728e-01 1.15072501e+00 9.34809864e-01 5.95495030e-02 -9.27871048e-01 -4.70610708e-01 5.63455045e-01 -3.82896543e-01 3.24885488e-01 -1.57915473e-01 -4.03034598e-01 -1.23932409e+00 2.29167461e-01 -1.36470914e+00 7.90731490e-01 -1.03785074e+00 -1.61576712e+00 1.85294151e-01 8.07037950e-02 -3.91065866e-01 -3.85491878e-01 -9.70976114e-01 -4.95650053e-01 8.87887001e-01 -2.07324815e+00 -1.45807469e+00 -4.00763035e-01 6.56305075e-01 3.01182926e-01 -4.85267229e-02 1.15654075e+00 2.83152014e-01 1.16398975e-01 9.62277532e-01 -1.44320026e-01 5.21641791e-01 4.91785020e-01 -1.78305292e+00 5.96904755e-01 7.02598393e-02 1.76638648e-01 6.91340625e-01 4.14248794e-01 -2.54640073e-01 -1.66565526e+00 -8.58464122e-01 1.02594328e+00 -1.15949249e+00 1.02218688e+00 -1.73768148e-01 -1.06063354e+00 4.48043972e-01 2.02900022e-01 -5.42662144e-02 8.82187247e-01 2.85716563e-01 -7.10594594e-01 -5.24261855e-02 -1.24198663e+00 6.43616199e-01 7.97194242e-01 -9.18239415e-01 -1.55591989e+00 5.05787849e-01 1.27343273e+00 -5.32860518e-01 -1.27025688e+00 6.88510299e-01 4.31916296e-01 -5.50841451e-01 1.53308618e+00 -1.09505546e+00 6.59687161e-01 -5.09028286e-02 -1.37038186e-01 -1.26829243e+00 3.76671143e-02 -2.81845331e-01 -8.51430118e-01 7.44749844e-01 3.50355893e-01 -5.63973546e-01 1.05076385e+00 6.93508565e-01 -1.88433543e-01 -1.32941151e+00 -1.25203335e+00 -9.32985067e-01 7.71964252e-01 -4.72632736e-01 8.70906353e-01 9.81203914e-01 5.40439859e-02 3.43760818e-01 5.58273345e-02 5.17846793e-02 2.39350736e-01 1.94928497e-02 9.82716262e-01 -9.19021130e-01 -3.81074727e-01 -3.45623076e-01 -3.88737023e-01 -1.04620028e+00 2.94489235e-01 -7.03861058e-01 -3.59098822e-01 -2.15720034e+00 -1.85072258e-01 -3.99720013e-01 -2.29402736e-01 6.31722689e-01 -6.04832768e-01 2.46768430e-01 1.40357716e-02 -4.22729135e-01 -9.09509897e-01 1.01366353e+00 1.35560822e+00 -3.98050219e-01 3.31881642e-02 -5.37406147e-01 -6.02892458e-01 2.24983692e-01 8.19985032e-01 1.83327626e-02 -6.63367808e-01 -6.10946774e-01 9.11975980e-01 -3.11291404e-02 5.34231722e-01 -7.58677661e-01 4.32405695e-02 -6.71580285e-02 2.52857674e-02 -5.33185244e-01 8.70632946e-01 -5.93489707e-01 -7.80693948e-01 3.60116214e-01 -4.46040869e-01 1.55965835e-01 3.75924200e-01 2.15941444e-01 -3.49642485e-01 -4.20676559e-01 2.11312801e-01 -5.73797040e-02 -8.15507650e-01 1.95790499e-01 1.06571108e-01 6.78034782e-01 4.64211345e-01 1.09075978e-01 -7.12717831e-01 -5.75541675e-01 -3.36354911e-01 8.45980525e-01 -9.91306677e-02 2.59778704e-02 8.44004929e-01 -1.56871831e+00 -7.13087976e-01 -3.58190238e-01 3.08928221e-01 3.83835375e-01 2.43698388e-01 5.16464114e-01 -8.50172937e-01 7.94470608e-01 -9.51030180e-02 6.21385761e-02 -7.85985053e-01 4.88940746e-01 2.65602976e-01 -1.03073800e+00 3.25484872e-02 1.24494720e+00 -5.70679307e-01 -1.44611168e+00 1.78953826e-01 -8.40919018e-01 -3.05171132e-01 3.27161141e-02 2.25478172e-01 5.36564291e-01 3.23865533e-01 8.94366577e-02 -1.67520449e-01 1.73018843e-01 3.74287553e-02 -3.70935738e-01 1.15152478e+00 4.92451876e-01 -2.15915501e-01 9.44481269e-02 1.08509612e+00 -1.10307366e-01 -8.34603131e-01 -2.75026739e-01 5.80634847e-02 -2.86756516e-01 -2.19345137e-01 -1.29300392e+00 -4.15174276e-01 1.00982189e+00 1.01022020e-01 -9.46746990e-02 1.03976953e+00 -3.70587036e-02 1.34549987e+00 1.19781315e+00 2.14532405e-01 -1.08161366e+00 8.66645217e-01 9.15519834e-01 1.13102818e+00 -1.26780140e+00 -9.92418155e-02 -1.25682861e-01 -6.69856250e-01 1.14124656e+00 9.75939512e-01 7.09439963e-02 3.65685448e-02 -4.18086976e-01 2.73849338e-01 -2.82636046e-01 -9.46551323e-01 -3.56107771e-01 7.41403162e-01 6.82575047e-01 1.25978500e-01 -1.38357192e-01 -2.60845125e-01 8.67111325e-01 -7.41216242e-01 1.56253707e-02 1.52885646e-01 9.33832884e-01 -3.04258049e-01 -1.19612646e+00 -1.94544300e-01 5.93392491e-01 -1.65663198e-01 -5.33056557e-01 -4.41511422e-01 8.01348031e-01 -1.09475002e-01 7.51779139e-01 -4.51929234e-02 -1.79074541e-01 5.38923681e-01 5.03296375e-01 8.19630802e-01 -4.38357770e-01 -8.31327140e-01 -1.25452578e+00 3.86860460e-01 -4.59614128e-01 -1.10588774e-01 -1.41287118e-01 -1.42418957e+00 -2.02882349e-01 -4.86260325e-01 5.21539092e-01 4.19555247e-01 8.35257053e-01 3.00651282e-01 4.29029644e-01 -3.31240110e-02 -5.10407761e-02 -1.37107480e+00 -1.07685578e+00 -5.60125075e-02 5.41561365e-01 2.84736186e-01 -4.24105942e-01 -3.55930090e-01 -3.71934533e-01]
[10.55698299407959, 7.92397403717041]
4b9443c6-f57b-423f-a8db-ef8f703aa169
extending-automatic-discourse-segmentation
1703.04718
null
http://arxiv.org/abs/1703.04718v1
http://arxiv.org/pdf/1703.04718v1.pdf
Extending Automatic Discourse Segmentation for Texts in Spanish to Catalan
At present, automatic discourse analysis is a relevant research topic in the field of NLP. However, discourse is one of the phenomena most difficult to process. Although discourse parsers have been already developed for several languages, this tool does not exist for Catalan. In order to implement this kind of parser, the first step is to develop a discourse segmenter. In this article we present the first discourse segmenter for texts in Catalan. This segmenter is based on Rhetorical Structure Theory (RST) for Spanish, and uses lexical and syntactic information to translate rules valid for Spanish into rules for Catalan. We have evaluated the system by using a gold standard corpus including manually segmented texts and results are promising.
['Irene Castellón', 'Juan-Manuel Torres-Moreno', 'Iria da Cunha', 'Eric SanJuan']
2017-03-11
null
null
null
null
['discourse-segmentation']
['natural-language-processing']
[ 1.58923224e-01 9.03812647e-01 -1.70603782e-01 -1.07008919e-01 -4.84863341e-01 -7.06758976e-01 1.00856781e+00 6.89072371e-01 -4.08836514e-01 1.11269557e+00 6.08470976e-01 -7.33151197e-01 5.08516170e-02 -7.45412171e-01 -3.14263463e-01 -3.22440565e-01 3.39829504e-01 6.44489944e-01 7.02929735e-01 -4.82812762e-01 4.10083562e-01 3.06547612e-01 -1.51059008e+00 7.25446641e-01 9.13036585e-01 9.99686196e-02 5.15495718e-01 5.90109289e-01 -5.71386218e-01 1.43418932e+00 -1.10721374e+00 -6.94217682e-02 -4.09190774e-01 -1.00482154e+00 -1.47398710e+00 -2.77077854e-02 -2.52132386e-01 4.13451940e-02 2.75578499e-01 8.81858110e-01 -2.17030514e-02 1.01397298e-02 7.62766898e-01 -5.66093326e-01 2.93477178e-02 1.22724962e+00 1.29577324e-01 9.79158729e-02 6.89276695e-01 -4.49344307e-01 8.56770515e-01 -3.31632465e-01 1.26511383e+00 1.46242416e+00 1.54311061e-01 4.15243506e-01 -8.92016947e-01 -1.24192275e-02 -1.47106066e-01 3.70092332e-01 -7.16734588e-01 -3.23830754e-01 8.79101098e-01 -6.72319949e-01 1.08689201e+00 3.37800801e-01 7.23036110e-01 1.08123446e+00 1.39449432e-01 6.78220510e-01 1.27468431e+00 -1.24375141e+00 4.02430117e-01 5.32189906e-01 4.13790196e-01 1.33142555e-02 2.73399621e-01 -4.69346315e-01 -6.47904426e-02 1.75566837e-01 3.50948542e-01 -1.02233636e+00 -1.00976698e-01 1.71936810e-01 -8.80766094e-01 1.27884340e+00 -2.20794573e-01 1.32885659e+00 -2.95889139e-01 -4.45690989e-01 9.43875849e-01 2.94705421e-01 5.30045748e-01 6.70257211e-01 9.12552029e-02 -4.09602225e-01 -8.12546074e-01 5.45943618e-01 1.46750903e+00 5.38828611e-01 1.49038374e-01 -1.71799198e-01 -1.82715908e-01 6.78241432e-01 4.52218264e-01 3.62230152e-01 5.25462806e-01 -9.56197977e-01 5.49526930e-01 9.77232635e-01 7.79711306e-02 -1.02360880e+00 -4.75435674e-01 3.65062356e-01 -2.46340334e-01 1.27922788e-01 6.85198665e-01 -4.22929883e-01 -2.50781566e-01 1.26974738e+00 3.54373038e-01 -9.85998392e-01 6.26859546e-01 6.03504300e-01 1.18517220e+00 1.03762484e+00 4.23566513e-02 -8.75425875e-01 1.54506350e+00 -7.06806183e-01 -1.13197494e+00 3.29850777e-03 1.15535951e+00 -1.06349921e+00 7.22802818e-01 3.48405719e-01 -1.23632729e+00 -2.53649354e-01 -9.77327228e-01 -3.27581674e-01 -3.14773798e-01 2.38479301e-01 2.11363137e-01 7.63080716e-01 -7.29428113e-01 7.22597539e-02 -6.22059524e-01 -6.51235819e-01 -7.73294717e-02 -6.38561249e-02 -1.95594132e-01 4.13622707e-01 -1.29889071e+00 1.12828016e+00 1.01442742e+00 -2.05194637e-01 -1.11503929e-01 1.66612491e-01 -8.36553991e-01 -9.15042982e-02 6.50471807e-01 -1.30885923e-02 1.54011238e+00 -1.22851562e+00 -1.89737809e+00 1.13517928e+00 -2.53853858e-01 -8.01898837e-01 5.37048459e-01 -4.30310853e-02 -2.62639999e-01 5.49875975e-01 3.83563668e-01 5.27227931e-02 2.65726715e-01 -1.08699965e+00 -8.04760635e-01 -9.33670700e-02 4.44063544e-01 2.30306312e-01 -1.14907008e-02 7.60740697e-01 -1.01807527e-01 -4.27882701e-01 -2.53780991e-01 -8.21101665e-01 1.43665314e-01 -8.69591057e-01 -2.60575414e-01 -6.97029889e-01 7.11079717e-01 -1.00887752e+00 1.52677369e+00 -1.83721077e+00 6.34275526e-02 3.69577073e-02 -1.47784889e-01 5.97579539e-01 5.30257761e-01 6.50406361e-01 9.86005962e-02 -1.92782450e-02 -1.27523184e-01 2.55182117e-01 1.62623171e-02 4.62061882e-01 -2.08048314e-01 3.37830521e-02 -3.32515500e-02 6.05795205e-01 -8.43588829e-01 -8.83452177e-01 3.93368542e-01 1.43854201e-01 -3.58581364e-01 -4.91095185e-02 -8.75497818e-01 4.85865921e-01 -6.13207698e-01 4.02035415e-02 2.57749885e-01 4.65504080e-02 9.63660359e-01 4.00528967e-01 -8.27747583e-01 8.45568717e-01 -9.79123533e-01 1.26542056e+00 -2.91571289e-01 9.49308634e-01 2.80129984e-02 -1.28866673e+00 1.07178390e+00 6.78883731e-01 3.52202296e-01 -7.38637924e-01 5.50182641e-01 4.35453951e-01 3.59954298e-01 -9.54219520e-01 7.42850482e-01 -8.47215801e-02 -2.62519568e-01 5.38020968e-01 -2.51112133e-01 -8.64526927e-02 1.02639008e+00 -1.98517330e-02 8.54886711e-01 3.10073406e-01 9.03218269e-01 -3.81424308e-01 1.17940891e+00 8.91768038e-01 3.95128787e-01 4.93289754e-02 1.62597582e-01 3.18745881e-01 1.00684464e+00 -2.12742597e-01 -1.05362594e+00 -1.48484081e-01 -1.33578286e-01 9.56805527e-01 -2.32758626e-01 -5.66464305e-01 -1.35885131e+00 -5.67272961e-01 -7.09579468e-01 1.14778519e+00 -3.84223253e-01 8.11473370e-01 -1.18020141e+00 -3.43328327e-01 3.91168803e-01 -2.91674864e-02 5.71748078e-01 -1.56591070e+00 -9.89881754e-01 5.87265670e-01 -4.27852064e-01 -1.24013221e+00 4.12260801e-01 -1.01061158e-01 -4.03586239e-01 -1.59227347e+00 -4.41600204e-01 -9.78735507e-01 1.36074677e-01 -1.04111947e-01 8.44068766e-01 1.36323243e-01 2.23236620e-01 3.60928923e-02 -8.96665812e-01 -9.94626999e-01 -1.55891824e+00 4.22531635e-01 -7.17244983e-01 -5.78625262e-01 3.76902670e-01 7.43569285e-02 1.76506609e-01 3.04378644e-02 -9.44889069e-01 1.20595045e-01 2.48440444e-01 3.97734582e-01 2.00406447e-01 -1.22527704e-01 4.47947919e-01 -1.45521843e+00 9.80348170e-01 -9.92460921e-02 -7.17674494e-01 3.55253547e-01 -1.78871572e-01 -3.72336917e-02 6.94230616e-01 -4.81904931e-02 -1.54393220e+00 -2.60644436e-01 -6.45329714e-01 6.29884720e-01 -5.45780182e-01 7.29686439e-01 -2.90923387e-01 5.07285833e-01 9.00646448e-01 -2.97663808e-01 1.11100413e-01 -3.82848889e-01 1.18049949e-01 7.13203311e-01 1.90632552e-01 -5.77593565e-01 1.48725525e-01 1.61934510e-01 -5.63141629e-02 -1.36428213e+00 -6.61886513e-01 -4.09824252e-01 -6.75680995e-01 -5.03666162e-01 1.17495489e+00 -6.76989973e-01 -6.14008665e-01 -9.04807076e-02 -1.60791922e+00 -4.08446908e-01 -6.18924975e-01 5.99634588e-01 -5.75554430e-01 3.76568168e-01 -4.40501809e-01 -8.56582642e-01 -3.06054920e-01 -1.01641631e+00 5.11835337e-01 2.39727840e-01 -7.82013834e-01 -1.27366960e+00 4.61554736e-01 3.21462601e-01 1.24760978e-02 6.28058851e-01 9.88112688e-01 -9.12281156e-01 -7.18383491e-02 1.88942224e-01 2.56677605e-02 2.74030089e-01 -7.47686327e-02 2.18921766e-01 -6.09086156e-01 2.54276901e-01 3.60069633e-01 -1.93410024e-01 3.82231295e-01 3.00466180e-01 1.36198372e-01 -4.45545793e-01 -4.08360213e-01 -3.73529196e-01 1.15848935e+00 5.84450126e-01 8.14938903e-01 8.20136309e-01 1.56364545e-01 1.08561051e+00 1.05130219e+00 1.49935633e-01 6.23229206e-01 5.63491940e-01 -1.10553600e-01 3.55309159e-01 -2.94047087e-01 3.31165045e-02 6.66700780e-01 1.08404756e+00 -2.83168912e-01 -2.45272875e-01 -1.12104177e+00 6.01457059e-01 -1.94254851e+00 -1.14534414e+00 -7.93077588e-01 1.56001604e+00 7.95538485e-01 3.61683428e-01 3.96140099e-01 5.31607151e-01 6.45115614e-01 1.80038273e-01 4.48956162e-01 -9.13853228e-01 -3.87685835e-01 3.04302752e-01 -1.53349265e-01 9.40469742e-01 -1.03232169e+00 1.00316858e+00 5.80587435e+00 6.21679425e-01 -1.07735884e+00 1.33021578e-01 -3.19160484e-02 4.27840978e-01 -2.33219564e-01 4.03947949e-01 -7.81494379e-01 3.32097322e-01 1.12452722e+00 -3.41199517e-01 -1.18598908e-01 7.91505098e-01 5.56008279e-01 -6.76515341e-01 -7.18971014e-01 3.83713365e-01 2.27726832e-01 -1.72241890e+00 -1.25246346e-01 1.84491798e-02 6.77191496e-01 -4.08358157e-01 -8.45670581e-01 3.05731237e-01 1.05116755e-01 -7.22573698e-01 8.06732416e-01 1.63298741e-01 2.22509548e-01 -5.36612630e-01 1.05118799e+00 5.64498603e-01 -6.96341991e-01 1.48216471e-01 -2.29562119e-01 -3.21354270e-01 4.12960887e-01 4.29073483e-01 -1.27740788e+00 6.63044512e-01 2.29147673e-01 3.26206207e-01 -3.68767083e-01 5.99935949e-01 -6.39778137e-01 8.51266623e-01 -2.43824422e-01 -6.97754323e-01 2.45416105e-01 -3.64696085e-01 8.58783603e-01 1.44903803e+00 1.78076550e-01 1.51749313e-01 9.15561616e-02 4.21755016e-01 2.29556054e-01 9.12028193e-01 -5.66897213e-01 -6.47388026e-02 2.89017469e-01 7.22791791e-01 -1.07809174e+00 -4.07794774e-01 -3.20596725e-01 5.27336240e-01 2.96447743e-02 -1.03623986e-01 -5.28692484e-01 -3.03691745e-01 -1.23988546e-01 3.15044075e-01 2.71838252e-03 -2.17070207e-01 -2.33882427e-01 -7.25139916e-01 2.19975901e-03 -1.11884761e+00 3.27389717e-01 -4.15834159e-01 -4.89954144e-01 7.35196292e-01 4.63222712e-01 -7.64490783e-01 -9.59150970e-01 -7.69998789e-01 -6.29296601e-01 7.03317225e-01 -1.26810265e+00 -1.12465537e+00 3.90099436e-02 1.92378819e-01 7.78088987e-01 -1.36733800e-01 1.16111803e+00 -6.87416829e-03 -2.95224041e-01 -2.72299975e-01 1.49791554e-01 2.41780072e-01 6.05597079e-01 -1.18568051e+00 -7.89674819e-02 7.49796748e-01 -1.25026375e-01 2.87683606e-01 1.18748069e+00 -6.64687037e-01 -7.52760053e-01 -5.41949391e-01 1.64113414e+00 -7.23262802e-02 8.50027323e-01 -1.03630677e-01 -9.90142524e-01 5.12332380e-01 8.34799528e-01 -1.03408349e+00 7.53005803e-01 -5.91905564e-02 2.58014083e-01 4.98072624e-01 -8.62460554e-01 3.89152557e-01 4.55228597e-01 -1.81708843e-01 -1.41430914e+00 4.88532275e-01 4.27996695e-01 -7.02905416e-01 -9.23009634e-01 7.86178708e-02 1.64352715e-01 -1.07077777e+00 3.07633877e-01 -1.20722584e-01 4.94432241e-01 -2.96419144e-01 1.17160134e-01 -8.98091853e-01 3.75882447e-01 -7.36529589e-01 3.02040905e-01 1.54706073e+00 5.89841008e-01 -6.26257479e-01 6.79200411e-01 3.36846203e-01 -2.47518092e-01 -3.64051014e-02 -8.06119263e-01 -6.28160477e-01 4.12301481e-01 -5.19765973e-01 3.59746516e-01 7.61410296e-01 6.66840434e-01 8.74833405e-01 2.82242060e-01 -5.25475323e-01 -1.40559869e-02 3.56710643e-01 1.02895868e+00 -1.54828954e+00 -1.19435586e-01 -6.39860392e-01 1.14117436e-01 -5.84203184e-01 4.60040838e-01 -8.08805466e-01 4.96044233e-02 -2.02069092e+00 -3.02106321e-01 -1.34702682e-01 9.78732109e-01 1.14592604e-01 2.75174975e-01 -3.69388521e-01 3.67409378e-01 2.24917516e-01 -3.74975950e-01 2.28849888e-01 1.08838713e+00 7.76459789e-03 -7.77667463e-01 -4.30931635e-02 -4.47876304e-01 9.98978198e-01 1.07486999e+00 -4.20825869e-01 -2.90319920e-01 9.83661041e-02 2.97620624e-01 -3.57182398e-02 -1.96037292e-01 -7.40345895e-01 1.70329437e-01 -3.32596093e-01 -2.81209201e-01 -8.00544202e-01 2.90911160e-02 -6.12285078e-01 1.49368346e-01 3.79157096e-01 -3.07132661e-01 -1.63782656e-01 2.91623533e-01 -1.47960901e-01 -6.10977829e-01 -1.03910005e+00 6.00203753e-01 -2.26886213e-01 -6.60183430e-01 -8.12649310e-01 -9.54448402e-01 2.79619008e-01 1.50353348e+00 -2.25574076e-01 -6.62909091e-01 -9.96419713e-02 -6.55549228e-01 -4.30506505e-02 5.47099292e-01 1.71581581e-02 1.30577505e-01 -7.86861181e-01 -8.05527925e-01 -3.31069052e-01 -7.29947463e-02 1.52695000e-01 -5.05315997e-02 6.94740415e-01 -1.28281951e+00 9.93547201e-01 -3.22507977e-01 -3.92574012e-01 -1.59743786e+00 5.46919823e-01 -1.01604365e-01 -7.44569540e-01 -6.70117915e-01 -4.98776138e-02 -3.08685660e-01 -5.68946870e-03 -2.07519025e-01 -4.29570735e-01 -1.26364732e+00 5.49758911e-01 5.93800306e-01 4.43976998e-01 -1.36643210e-02 -1.16222048e+00 -1.02371827e-01 2.91545331e-01 2.24032134e-01 -4.90049005e-01 1.25404525e+00 -1.96513399e-01 -5.41328967e-01 6.21696234e-01 7.30541229e-01 3.93742293e-01 -2.60854870e-01 2.18916625e-01 6.20792150e-01 2.37143203e-03 -3.41129035e-01 -3.77130806e-01 -5.00451773e-02 6.48801565e-01 -1.36678740e-01 1.07675004e+00 7.84388661e-01 6.66949600e-02 4.26129520e-01 4.09980655e-01 1.53880730e-01 -1.59316933e+00 -3.58072162e-01 1.09955835e+00 1.03041887e+00 -8.52172017e-01 -4.05193307e-02 -8.47759008e-01 -6.89970970e-01 1.36651635e+00 2.22233936e-01 4.60065901e-04 6.10464633e-01 1.99614123e-01 2.31295049e-01 -3.45098913e-01 -4.92762357e-01 -5.57833433e-01 -5.89918941e-02 6.67045593e-01 9.33962107e-01 1.82638407e-01 -1.62749565e+00 5.25222063e-01 -5.39481163e-01 7.74405673e-02 8.11822832e-01 1.00269449e+00 -7.82386363e-01 -1.57528985e+00 -9.30941284e-01 -1.71283931e-02 -7.69705296e-01 3.79900604e-01 -8.13764036e-01 1.42497683e+00 1.09842062e-01 1.26465333e+00 -2.85743643e-02 2.00274259e-01 4.80313092e-01 8.34229216e-02 5.51114500e-01 -8.51904333e-01 -9.10247147e-01 3.78545612e-01 1.01023018e+00 -1.06356531e-01 -1.14268315e+00 -1.05896807e+00 -1.56316018e+00 -3.06557119e-01 -5.37927933e-02 6.48125112e-01 7.94806540e-01 1.27756357e+00 -3.12599868e-01 6.29241526e-01 2.27720261e-01 -5.04616976e-01 7.39274696e-02 -1.22220087e+00 -2.75600463e-01 2.40565747e-01 -2.06128538e-01 -4.49226171e-01 -2.21422110e-02 3.31282675e-01]
[10.777650833129883, 9.463423728942871]
b2531109-919f-4cd2-8f6f-15e36f351072
is-neural-language-acquisition-similar-to
2207.0056
null
https://arxiv.org/abs/2207.00560v1
https://arxiv.org/pdf/2207.00560v1.pdf
Is neural language acquisition similar to natural? A chronological probing study
The probing methodology allows one to obtain a partial representation of linguistic phenomena stored in the inner layers of the neural network, using external classifiers and statistical analysis. Pre-trained transformer-based language models are widely used both for natural language understanding (NLU) and natural language generation (NLG) tasks making them most commonly used for downstream applications. However, little analysis was carried out, whether the models were pre-trained enough or contained knowledge correlated with linguistic theory. We are presenting the chronological probing study of transformer English models such as MultiBERT and T5. We sequentially compare the information about the language learned by the models in the process of training on corpora. The results show that 1) linguistic information is acquired in the early stages of training 2) both language models demonstrate capabilities to capture various features from various levels of language, including morphology, syntax, and even discourse, while they also can inconsistently fail on tasks that are perceived as easy. We also introduce the open-source framework for chronological probing research, compatible with other transformer-based models. https://github.com/EkaterinaVoloshina/chronological_probing
['Tatiana Shavrina', 'Oleg Serikov', 'Ekaterina Voloshina']
2022-07-01
null
null
null
null
['language-acquisition']
['natural-language-processing']
[-5.56175448e-02 5.73187292e-01 -3.49677235e-01 -2.85721362e-01 -5.02060354e-01 -7.57603049e-01 1.12383556e+00 2.38934740e-01 -1.66316435e-01 5.75157404e-01 4.70159233e-01 -8.27815711e-01 4.59497944e-02 -9.62874591e-01 -7.97293246e-01 -2.53380448e-01 2.63271462e-02 7.49386489e-01 2.42418930e-01 -6.05632961e-01 2.69229323e-01 2.63275594e-01 -1.48036718e+00 7.16181815e-01 7.94629335e-01 6.47846460e-01 5.49631476e-01 3.84450942e-01 -5.52989721e-01 1.11154151e+00 -3.69148463e-01 -3.23276639e-01 -1.01240829e-01 -3.80461365e-01 -1.23178089e+00 -3.85581255e-01 1.37117669e-01 -9.84453037e-02 6.19039778e-03 7.91855156e-01 5.57402223e-02 -2.92549670e-01 4.54373419e-01 -7.71610141e-01 -7.38105774e-01 1.43668020e+00 4.06626910e-02 4.61438179e-01 4.58024383e-01 -1.01257060e-02 1.24656487e+00 -1.11790013e+00 8.29620063e-01 1.39211667e+00 6.58254325e-01 6.03972256e-01 -1.20562267e+00 -4.53350484e-01 1.64722025e-01 3.46610874e-01 -8.49942327e-01 -6.26220942e-01 8.06800425e-01 -6.35299563e-01 1.21368110e+00 4.74334955e-02 7.00170934e-01 1.27424896e+00 1.31951034e-01 8.91921759e-01 1.32419550e+00 -9.19902861e-01 -1.43103123e-01 5.84267378e-01 3.15056175e-01 8.98902059e-01 -1.60850007e-02 3.31169784e-01 -8.50710869e-01 2.97750801e-01 6.14210606e-01 -7.38896430e-01 -1.88729644e-01 1.07459791e-01 -1.07743907e+00 8.04500997e-01 4.34045255e-01 1.00265920e+00 -2.62327343e-01 -9.49836895e-02 4.66509640e-01 4.14481521e-01 6.33245468e-01 4.62151051e-01 -8.59073937e-01 -1.64873779e-01 -7.55437553e-01 7.10587278e-02 6.15977466e-01 8.53568733e-01 6.71238124e-01 1.12489335e-01 1.87764671e-02 8.26376319e-01 4.78718549e-01 4.72041294e-02 8.54727447e-01 -7.58806944e-01 6.29746556e-01 8.45837951e-01 -1.75122172e-01 -6.91858888e-01 -3.42384636e-01 -2.45040715e-01 -3.35458219e-01 -2.36886054e-01 6.78382397e-01 2.08250046e-01 -6.57887101e-01 2.03479433e+00 -8.11885297e-03 -2.65448958e-01 1.96310967e-01 3.07070524e-01 1.01058710e+00 8.76443923e-01 4.38113004e-01 -1.45157740e-01 1.55877280e+00 -7.75155246e-01 -4.86697078e-01 -7.99792647e-01 9.76003230e-01 -5.69061816e-01 1.41932511e+00 2.48211592e-01 -1.36851251e+00 -6.65885448e-01 -8.80508423e-01 -3.75261575e-01 -6.92393780e-01 2.70719081e-01 7.09349751e-01 6.18850291e-01 -1.44069731e+00 6.27234876e-01 -8.43452573e-01 -4.76668119e-01 8.97390917e-02 5.78037761e-02 -3.74647290e-01 1.41405419e-01 -1.42176592e+00 1.45615482e+00 6.89930499e-01 2.52317697e-01 -8.12574208e-01 -6.58317685e-01 -9.83078778e-01 -4.03046682e-02 -3.32084484e-04 -5.24930120e-01 1.51107252e+00 -7.46959329e-01 -1.37564135e+00 1.33219790e+00 -3.27879697e-01 -3.58959347e-01 2.92613775e-01 -5.94739653e-02 -5.94731420e-02 -1.48133889e-01 2.44779184e-01 6.63829684e-01 1.10505462e-01 -1.21999156e+00 -4.66944396e-01 -3.83384049e-01 1.48232907e-01 4.34818938e-02 -1.93404973e-01 1.68388695e-01 -7.78163597e-02 -5.35561264e-01 2.64608175e-01 -4.77460951e-01 1.57430798e-01 -4.51969951e-01 -2.69919753e-01 -5.06784737e-01 2.55331695e-01 -9.34739590e-01 1.18596208e+00 -1.70018303e+00 5.78472316e-02 -2.49412924e-01 -1.29926160e-01 2.15515614e-01 -1.06213495e-01 7.89564371e-01 -2.43579388e-01 5.59728265e-01 -1.03289977e-01 -3.33608419e-01 1.55287310e-01 2.89455980e-01 -6.50160193e-01 1.02082163e-01 1.80353045e-01 1.05647826e+00 -8.60989273e-01 -3.28025043e-01 2.51694918e-01 1.51981935e-01 -4.76146758e-01 1.26001224e-01 -4.98223931e-01 6.24155104e-01 -2.36286640e-01 4.76409853e-01 1.66716367e-01 -7.21804202e-02 3.70130152e-01 -7.03647286e-02 -4.63143766e-01 1.28771913e+00 -5.55695772e-01 1.58526146e+00 -9.10045147e-01 8.11277628e-01 -1.54372126e-01 -1.19677091e+00 8.34535658e-01 6.06238008e-01 -1.34324238e-01 -8.93296957e-01 3.96404684e-01 4.66992617e-01 1.08788133e-01 -3.93148184e-01 3.21419090e-01 -4.27203149e-01 -1.65538982e-01 6.65807962e-01 3.94621223e-01 -1.69302374e-01 4.54216301e-01 7.99207836e-02 7.19747305e-01 3.78540665e-01 4.29104805e-01 -5.66075385e-01 5.48476994e-01 2.47022316e-01 3.64559978e-01 4.81806636e-01 2.32022032e-02 5.87421879e-02 4.91890490e-01 -3.48306447e-01 -1.02376986e+00 -8.17248046e-01 -4.03151810e-01 1.31707048e+00 -2.13298097e-01 -5.98834217e-01 -8.23498845e-01 -1.82400212e-01 -5.35988688e-01 1.25964069e+00 -6.27849460e-01 -1.50501639e-01 -6.91078246e-01 -4.44632232e-01 6.57693207e-01 6.41725957e-01 4.12526459e-01 -1.50898027e+00 -5.45738637e-01 4.03429419e-01 -3.10993999e-01 -1.15381598e+00 2.12992966e-01 2.09429443e-01 -1.10569119e+00 -8.56026232e-01 -1.51318058e-01 -1.03530467e+00 3.06603462e-01 -3.77206773e-01 1.40336621e+00 2.08165288e-01 4.59091999e-02 9.62941572e-02 -2.43442744e-01 -4.30951118e-01 -9.55705702e-01 2.11427823e-01 -1.04461253e-01 -4.41028535e-01 4.49551105e-01 -6.33532822e-01 3.35077965e-03 -2.03742459e-02 -6.71061993e-01 4.57010984e-01 3.18643808e-01 7.08122194e-01 1.32905571e-02 -1.21073820e-01 3.93628359e-01 -1.07126367e+00 7.66733944e-01 -3.78128946e-01 -4.99299526e-01 3.49451810e-01 -5.42893410e-01 1.49778962e-01 5.29808819e-01 -3.07339191e-01 -1.24666405e+00 -4.02422965e-01 -4.67898488e-01 2.30224848e-01 -2.47616664e-01 8.93639982e-01 -1.00765608e-01 4.12337363e-01 7.69851208e-01 3.04788530e-01 -2.40055546e-01 -6.19261920e-01 4.38735366e-01 3.74434888e-01 3.75919461e-01 -1.00946271e+00 6.02620423e-01 5.74208237e-02 -5.26351035e-01 -7.41348267e-01 -1.06678581e+00 -9.86992270e-02 -8.58058333e-01 -7.15635568e-02 6.23858988e-01 -7.26226389e-01 -3.83443534e-01 5.39050579e-01 -1.49235225e+00 -7.58891523e-01 -1.64078817e-01 1.96703762e-01 -4.36136276e-01 4.62732315e-02 -9.94672179e-01 -6.66963696e-01 -2.49888018e-01 -9.20312047e-01 7.94823170e-01 1.05798297e-01 -5.19493043e-01 -1.62530220e+00 1.35224396e-02 3.74505341e-01 3.96645278e-01 -1.66085631e-01 1.63026202e+00 -7.38328218e-01 -4.94624734e-01 2.31709898e-01 8.80656317e-02 1.64713413e-01 -3.53502035e-02 -4.78608496e-02 -1.23653340e+00 1.01943299e-01 1.07399903e-01 -3.53256553e-01 5.16169965e-01 7.53782094e-02 6.00256503e-01 -4.93292451e-01 -3.00014526e-01 2.29471996e-01 1.25802433e+00 5.46935648e-02 4.74846423e-01 4.96781856e-01 3.59860659e-01 1.05462563e+00 3.14064950e-01 -5.18243723e-02 6.70043766e-01 4.66642827e-01 9.17498693e-02 2.33537257e-01 -9.00565311e-02 -6.86116040e-01 6.28650606e-01 1.17566645e+00 2.38790929e-01 -9.00506377e-02 -1.44362235e+00 9.18110907e-01 -1.70378232e+00 -6.58685684e-01 -1.36803091e-01 1.86324298e+00 1.25231779e+00 3.87144268e-01 -2.69761771e-01 1.62136570e-01 5.18391669e-01 4.77618352e-02 -8.57216492e-03 -7.92062223e-01 -1.83362961e-01 4.03332114e-01 -1.40439764e-01 9.18774903e-01 -5.54384649e-01 1.50317466e+00 6.44034624e+00 4.78559971e-01 -1.06002355e+00 2.37871677e-01 5.55613458e-01 5.64688504e-01 -6.39697015e-01 2.74735987e-01 -8.25797915e-01 4.85814363e-02 1.23249805e+00 -3.42689276e-01 4.08354461e-01 5.31140804e-01 2.87731051e-01 -2.08227992e-01 -1.47559094e+00 4.01974678e-01 -2.21123487e-01 -1.42749298e+00 3.38737935e-01 -2.45457560e-01 2.78405368e-01 1.19229943e-01 -1.66144058e-01 5.09713948e-01 3.31069618e-01 -9.78851438e-01 1.24245358e+00 2.92643756e-01 6.92157865e-01 -1.00769594e-01 6.14087224e-01 8.76395881e-01 -1.04995227e+00 -1.17473848e-01 -2.64195055e-01 -2.95675427e-01 1.35617331e-01 2.89245546e-01 -1.02273571e+00 2.91193902e-01 4.74545300e-01 5.38516343e-01 -8.68660867e-01 3.83845150e-01 -8.45759511e-01 9.79968846e-01 -2.44817764e-01 -8.09186772e-02 2.64800549e-01 -1.40980229e-01 4.83829379e-01 1.14284623e+00 3.32067221e-01 -1.14614300e-01 -7.36376271e-02 1.25493670e+00 9.02159214e-02 3.92289191e-01 -7.76589096e-01 -2.22507551e-01 5.80067575e-01 9.07884598e-01 -7.09815383e-01 -4.34358716e-01 -2.04013988e-01 3.76296610e-01 4.94204730e-01 1.74651649e-02 -4.80763763e-01 2.39626214e-01 1.70859694e-01 3.37774038e-01 -8.15147981e-02 -2.62347072e-01 -5.40146649e-01 -1.13930333e+00 -6.84802011e-02 -7.76829362e-01 2.20490649e-01 -1.16055870e+00 -1.00607491e+00 6.50682569e-01 2.25942492e-01 -5.37222207e-01 -7.09722400e-01 -9.58914340e-01 -6.65458500e-01 1.06493843e+00 -1.37289035e+00 -1.32335758e+00 1.28836617e-01 3.52094442e-01 6.65634394e-01 -1.67291492e-01 1.08477676e+00 8.46220851e-02 -4.69658434e-01 1.21011622e-01 -3.82621318e-01 2.93447465e-01 3.53911638e-01 -1.29010892e+00 5.13989627e-01 8.32906485e-01 4.04963315e-01 9.98614490e-01 7.57058740e-01 -3.94028693e-01 -9.25106108e-01 -6.29995763e-01 1.81314349e+00 -6.50262892e-01 9.89184260e-01 -5.40106297e-01 -9.58958447e-01 9.75805163e-01 5.19069433e-01 -4.45248067e-01 4.09443706e-01 3.57961327e-01 -3.78298283e-01 -1.60057098e-02 -8.24694574e-01 7.01657832e-01 1.07210290e+00 -8.41274917e-01 -1.20330870e+00 2.55630851e-01 5.99127948e-01 -3.20127100e-01 -6.67065918e-01 2.19857201e-01 2.27771759e-01 -1.07269025e+00 5.53885877e-01 -5.45166850e-01 5.72776079e-01 2.58677397e-02 -1.03446737e-01 -1.38370049e+00 -2.54822552e-01 -4.46749926e-01 3.34098935e-01 1.48472118e+00 8.86365235e-01 -8.57997954e-01 3.55951816e-01 4.29519325e-01 -2.54321307e-01 -5.01713932e-01 -8.11630845e-01 -3.70695651e-01 5.59855640e-01 -8.13650548e-01 3.46090436e-01 8.90081823e-01 3.24505776e-01 8.89022112e-01 3.47179711e-01 -7.44157806e-02 8.05584341e-02 8.30008835e-02 3.79746675e-01 -1.40298057e+00 -1.76610649e-01 -7.18204856e-01 1.21769898e-01 -8.59260738e-01 5.24881005e-01 -1.26612127e+00 5.90223521e-02 -1.76381767e+00 -1.04957491e-01 -4.86287743e-01 8.64157826e-02 8.91879380e-01 2.41782695e-01 5.04882038e-02 1.15116671e-01 2.40171880e-01 2.13570282e-01 2.60535896e-01 1.04698861e+00 1.09298237e-01 -1.10535011e-01 -2.22482651e-01 -9.00446594e-01 9.27246928e-01 9.28066015e-01 -3.03850412e-01 -5.37940979e-01 -7.68362999e-01 5.90076149e-01 5.02460748e-02 3.21639568e-01 -6.53674901e-01 1.63646981e-01 2.77910680e-02 -2.54226010e-02 -3.23090315e-01 7.86293074e-02 -5.02382636e-01 -7.62825981e-02 3.91386837e-01 -3.43561471e-01 2.25962803e-01 6.05750561e-01 -1.86326802e-01 -2.98269540e-01 -5.34946144e-01 7.44356096e-01 -5.97440064e-01 -7.80914307e-01 -1.31162956e-01 -7.76034474e-01 1.74286932e-01 5.16306043e-01 -1.49999246e-01 -4.65935200e-01 -1.22637346e-01 -9.13426995e-01 -2.55479966e-03 3.13691109e-01 5.81683159e-01 2.48512715e-01 -1.02217174e+00 -5.69419622e-01 1.76392883e-01 4.38717566e-02 -5.44264838e-02 -2.44486660e-01 8.17268550e-01 -5.17554641e-01 8.23481917e-01 -1.65327907e-01 -5.32482088e-01 -8.25477064e-01 3.90568674e-01 4.84849423e-01 -4.92179334e-01 -5.10699451e-01 9.40802038e-01 1.98058337e-01 -8.68566632e-01 -1.07448742e-01 -5.65014124e-01 -4.80746865e-01 4.35497761e-01 2.08828717e-01 -1.12551257e-01 2.38512561e-01 -7.57378280e-01 -2.41902605e-01 4.40672189e-01 8.59282352e-03 -3.97180378e-01 1.42254949e+00 -9.78409350e-02 -4.40123260e-01 9.97616768e-01 8.29430819e-01 7.56772980e-02 -5.55931211e-01 -3.92139286e-01 5.42982519e-01 1.53494388e-01 -7.79877380e-02 -9.52259541e-01 -5.86635709e-01 1.28268898e+00 7.43205696e-02 4.32091445e-01 8.03753972e-01 3.59158188e-01 5.56116045e-01 2.76196539e-01 4.65748787e-01 -9.42859888e-01 -1.65329680e-01 1.00256562e+00 1.10372698e+00 -6.89122081e-01 -5.35082817e-01 -3.57646704e-01 -3.55835944e-01 1.08794045e+00 5.55782437e-01 3.44726481e-02 5.33097267e-01 3.09792340e-01 2.69606441e-01 -2.21876934e-01 -1.21808648e+00 -1.85242444e-01 1.64372072e-01 3.35885108e-01 8.99960220e-01 -8.44492093e-02 -4.08383757e-01 6.97475731e-01 -7.69807756e-01 -2.27446333e-01 4.77165103e-01 7.42546260e-01 -4.66023266e-01 -1.33565509e+00 -2.39698917e-01 1.71274878e-02 -3.75566870e-01 -6.11835957e-01 -5.91599822e-01 9.79600906e-01 2.08110780e-01 7.68624485e-01 -2.70613320e-02 -8.63572136e-02 1.00691587e-01 6.82815015e-01 5.83845615e-01 -1.01405549e+00 -7.08111286e-01 -1.02519907e-01 4.95635659e-01 -2.65973747e-01 -3.63177896e-01 -7.38461494e-01 -1.21114278e+00 -2.51164496e-01 -2.10539192e-01 2.44424373e-01 6.06668651e-01 1.32521653e+00 -1.12747364e-01 3.55008930e-01 1.59902751e-01 -8.36321890e-01 -2.27358311e-01 -1.29152501e+00 -1.58860177e-01 8.99772346e-02 4.50087339e-02 -5.65108955e-01 -3.71868104e-01 1.33461118e-01]
[10.524857521057129, 9.345829963684082]
5656f1f5-5257-4231-b2aa-76cbfcef9a76
example-based-image-synthesis-via-randomized
1609.0737
null
http://arxiv.org/abs/1609.07370v1
http://arxiv.org/pdf/1609.07370v1.pdf
Example-Based Image Synthesis via Randomized Patch-Matching
Image and texture synthesis is a challenging task that has long been drawing attention in the fields of image processing, graphics, and machine learning. This problem consists of modelling the desired type of images, either through training examples or via a parametric modeling, and then generating images that belong to the same statistical origin. This work addresses the image synthesis task, focusing on two specific families of images -- handwritten digits and face images. This paper offers two main contributions. First, we suggest a simple and intuitive algorithm capable of generating such images in a unified way. The proposed approach taken is pyramidal, consisting of upscaling and refining the estimated image several times. For each upscaling stage, the algorithm randomly draws small patches from a patch database, and merges these to form a coherent and novel image with high visual quality. The second contribution is a general framework for the evaluation of the generation performance, which combines three aspects: the likelihood, the originality and the spread of the synthesized images. We assess the proposed synthesis scheme and show that the results are similar in nature, and yet different from the ones found in the training set, suggesting that true synthesis effect has been obtained.
['Yi Ren', 'Michael Elad', 'Yaniv Romano']
2016-09-23
null
null
null
null
['patch-matching']
['computer-vision']
[ 7.55428553e-01 2.08462819e-01 2.72845715e-01 -2.64314860e-01 -5.42989612e-01 -2.94783324e-01 9.53435183e-01 -1.23551726e-01 -1.09680198e-01 7.42346525e-01 -2.06352919e-01 2.56400406e-01 -2.93549806e-01 -8.86608005e-01 -7.65357792e-01 -1.04910541e+00 1.67251751e-01 5.79106867e-01 1.37758300e-01 -8.92150849e-02 4.77500647e-01 8.54450881e-01 -2.05964446e+00 1.71336472e-01 7.30035186e-01 9.02777791e-01 3.53343755e-01 7.08680332e-01 -3.02548837e-02 4.39396352e-01 -7.01763272e-01 -5.51226258e-01 2.15272367e-01 -7.95466423e-01 -5.13861120e-01 9.20910120e-01 4.22974706e-01 -9.39685553e-02 3.03840756e-01 1.01489711e+00 4.17947173e-01 -9.88576468e-03 1.02858269e+00 -9.96063590e-01 -3.57806832e-01 2.47125596e-01 -5.89563131e-01 -3.87561947e-01 1.42031193e-01 2.20973585e-02 6.33451223e-01 -8.50459874e-01 7.92427003e-01 1.36596012e+00 3.73450398e-01 4.51319277e-01 -1.58022690e+00 -3.88920337e-01 -3.72887731e-01 -1.56272724e-01 -1.45382440e+00 -5.41341364e-01 7.82934844e-01 -6.73259616e-01 2.16293007e-01 4.08445776e-01 5.15181005e-01 7.96659946e-01 1.75044134e-01 5.46476603e-01 1.56223023e+00 -9.44461942e-01 3.89068335e-01 6.06281340e-01 -2.63610989e-01 3.79330128e-01 9.55448151e-02 2.23916486e-01 -3.51239949e-01 -4.54331608e-03 9.81059968e-01 -4.65610832e-01 -1.65618628e-01 -5.33656836e-01 -9.90058362e-01 6.90205872e-01 1.12903953e-01 6.80653870e-01 -5.25834858e-01 -6.44003376e-02 -6.07424937e-02 -1.27189653e-02 6.04644179e-01 4.00383085e-01 -6.10395595e-02 2.46082544e-01 -1.20026386e+00 4.86331195e-01 6.25059903e-01 7.87987888e-01 9.70764041e-01 8.34446326e-02 -1.16214417e-01 1.07006872e+00 5.96960038e-02 5.67397892e-01 2.80919075e-01 -7.56121099e-01 7.29310885e-02 3.14702749e-01 6.20247647e-02 -1.27269530e+00 -2.93678977e-03 -4.33652520e-01 -8.53297174e-01 8.08983147e-01 4.13347065e-01 7.23260939e-02 -7.73983061e-01 1.60534072e+00 3.99733633e-01 -6.60385415e-02 2.91654025e-03 4.22281235e-01 5.18483222e-01 8.19927990e-01 -2.72905439e-01 -3.46929848e-01 1.29640591e+00 -8.57220590e-01 -7.17103839e-01 5.57452552e-02 -1.38383344e-01 -1.14158666e+00 7.78880537e-01 7.27458715e-01 -1.22064161e+00 -9.54392314e-01 -9.73391533e-01 3.07603598e-01 -4.15146425e-02 6.45418108e-01 1.63981795e-01 8.27388287e-01 -1.06178188e+00 8.08466554e-01 -2.97253698e-01 -3.15197200e-01 2.58541226e-01 1.84197307e-01 -3.26039374e-01 1.17172953e-02 -4.78570372e-01 7.24201858e-01 5.45148492e-01 9.91297606e-03 -5.75080037e-01 -3.74011576e-01 -6.44388258e-01 -2.63405666e-02 2.14677360e-02 -5.11082828e-01 9.94287431e-01 -1.50703013e+00 -1.65528452e+00 8.74278665e-01 -2.76191533e-01 -1.53281286e-01 7.10811913e-01 2.30503455e-01 -2.11555615e-01 3.46404538e-02 1.24786347e-01 8.12671900e-01 1.47519815e+00 -1.77957261e+00 -5.27973056e-01 -3.07123363e-01 -4.75300580e-01 1.31717026e-01 -1.22323092e-02 -1.53015450e-01 -4.99320000e-01 -9.35228825e-01 -1.02415495e-01 -7.89800704e-01 -2.38658831e-01 -1.50499389e-01 -3.77735764e-01 1.06545694e-01 4.76925850e-01 -5.61325550e-01 9.08144593e-01 -2.31256938e+00 2.68164337e-01 5.65975010e-01 -1.55514225e-01 8.91744420e-02 -6.13795407e-02 5.36785066e-01 -2.80778706e-01 -1.42583266e-01 -4.85335469e-01 -5.32601416e-01 -1.99934170e-01 1.31138727e-01 -3.01556081e-01 2.64990211e-01 4.77983922e-01 5.97978294e-01 -4.23817366e-01 -5.33560991e-01 4.00844455e-01 5.13230801e-01 -3.85146737e-01 1.66741237e-01 -3.15944225e-01 6.50039017e-01 -2.48543337e-01 3.85676235e-01 9.46957827e-01 4.41108160e-02 2.86033481e-01 -3.45404506e-01 -2.61059791e-01 -5.20626366e-01 -1.56549764e+00 1.23297322e+00 -4.30436760e-01 6.57596350e-01 -1.51121184e-01 -9.95585203e-01 1.42208183e+00 1.30349934e-01 3.37749243e-01 -6.38635874e-01 1.62389636e-01 3.92284423e-01 -1.53095454e-01 -4.69447523e-01 5.65898240e-01 -3.27087402e-01 2.91363120e-01 3.92566204e-01 2.77251452e-01 -6.27969742e-01 4.62623656e-01 -3.09118837e-01 3.22724104e-01 3.08173597e-01 4.69818026e-01 -3.33607614e-01 6.88154280e-01 -7.43028671e-02 -1.11134350e-02 6.17587388e-01 5.71429312e-01 9.71878588e-01 6.16677999e-01 -3.18637490e-01 -1.47042966e+00 -7.63486028e-01 -3.33419740e-01 5.10222912e-01 7.98262805e-02 -6.93777576e-02 -1.18357432e+00 -1.28775135e-01 -2.58753836e-01 6.31650925e-01 -7.97292173e-01 2.30037421e-01 -4.99412179e-01 -8.61249626e-01 1.61988556e-01 -6.54315129e-02 4.84377116e-01 -1.24971390e+00 -6.06336534e-01 1.00367762e-01 -5.22231460e-02 -8.13497126e-01 -3.88837643e-02 -2.23408535e-01 -9.11551833e-01 -7.88022757e-01 -1.07095683e+00 -7.31432676e-01 9.18466568e-01 1.79559793e-02 1.15445054e+00 1.17954656e-01 -5.11390984e-01 1.45983085e-01 -2.72125125e-01 -3.13890040e-01 -8.26585114e-01 -2.28488907e-01 -3.82671773e-01 7.65455663e-01 -3.58905137e-01 -4.40030068e-01 -2.99779266e-01 3.83653760e-01 -1.24428189e+00 2.01555371e-01 9.52423990e-01 9.08414662e-01 7.23341286e-01 5.84245145e-01 3.74286234e-01 -9.83168960e-01 5.79362869e-01 -8.42508599e-02 -7.56709635e-01 2.92548448e-01 -4.98809129e-01 2.66760856e-01 5.75394988e-01 -2.88196295e-01 -1.30715001e+00 3.87969971e-01 -3.40391427e-01 -1.69748202e-01 -5.31653702e-01 1.31319180e-01 -2.30770171e-01 -2.77709454e-01 8.41061056e-01 5.32774091e-01 3.69584024e-01 -5.42566597e-01 3.51674199e-01 4.50080335e-01 5.62920570e-01 -5.99814475e-01 8.22081745e-01 3.64138901e-01 1.72580063e-01 -1.33433020e+00 -1.64415419e-01 -7.36700669e-02 -6.02923930e-01 -5.40975153e-01 7.25251496e-01 -3.80209893e-01 -2.98827142e-01 8.64859998e-01 -1.26786089e+00 -2.67002940e-01 -5.78221917e-01 2.94344485e-01 -7.66094327e-01 3.67702365e-01 -1.04581207e-01 -8.67644370e-01 7.86297172e-02 -1.42793787e+00 1.27200639e+00 1.77418485e-01 -4.31987531e-02 -9.27231967e-01 2.20263556e-01 1.49033234e-01 2.11188048e-01 4.67572570e-01 9.27948058e-01 -1.93186134e-01 -6.78804398e-01 -1.45731315e-01 -1.47499725e-01 6.75712883e-01 2.25037873e-01 4.07027066e-01 -1.08524239e+00 -1.12559088e-01 8.10634568e-02 -1.24939866e-01 6.75595939e-01 5.69097936e-01 9.59444523e-01 -2.13550746e-01 -1.91483349e-01 2.19772756e-01 1.66390169e+00 3.24741423e-01 1.02037358e+00 1.64011165e-01 1.47024527e-01 9.80725110e-01 5.99037886e-01 4.07555848e-01 -2.48209521e-01 9.72219646e-01 3.06043893e-01 -3.01085889e-01 -3.71865511e-01 -4.63972315e-02 1.56895618e-03 3.38809311e-01 -2.80444235e-01 -3.47497493e-01 -4.76806760e-01 3.83247077e-01 -1.67103815e+00 -1.08882976e+00 -1.95561692e-01 2.32886481e+00 5.91813624e-01 -1.06882006e-01 1.54941633e-01 4.79055464e-01 9.08162951e-01 -4.09889147e-02 -8.36091861e-02 -4.89425153e-01 -2.41964281e-01 5.41636586e-01 2.71311551e-01 7.57236123e-01 -8.89182329e-01 5.86917043e-01 6.54456329e+00 1.26455939e+00 -1.25068462e+00 -3.36751878e-01 1.04082632e+00 5.18188000e-01 -3.62671405e-01 -7.90632516e-02 -6.79935038e-01 4.59455013e-01 4.56386268e-01 -3.22905667e-02 2.89663315e-01 6.78867280e-01 1.38741478e-01 -4.89871502e-01 -8.18615377e-01 9.37819898e-01 3.88732046e-01 -1.40480053e+00 3.70879233e-01 1.68571383e-01 9.71281111e-01 -8.87096643e-01 3.04479539e-01 -3.12183678e-01 -2.01646283e-01 -1.16962624e+00 1.03344226e+00 8.64548385e-01 7.60068655e-01 -7.11330056e-01 6.69983566e-01 4.44787860e-01 -8.89491796e-01 1.57051221e-01 -1.94961756e-01 1.72181234e-01 8.30761194e-02 8.30895543e-01 -8.30175757e-01 7.06380963e-01 4.42539245e-01 3.87208909e-01 -7.58394122e-01 1.13460207e+00 -9.86070782e-02 2.20245779e-01 -1.38550326e-01 -4.30471860e-02 -7.27108791e-02 -5.86184084e-01 3.75353307e-01 1.13087022e+00 6.59545243e-01 -1.09616205e-01 -2.51221627e-01 1.12101734e+00 3.61947477e-01 4.22961265e-01 -5.76039135e-01 1.84268892e-01 5.87838367e-02 1.33879399e+00 -1.05876768e+00 -4.38073337e-01 -1.14758320e-01 9.60254669e-01 -8.75235423e-02 1.66268274e-01 -5.34360886e-01 -3.91137630e-01 -5.97178191e-03 1.95895836e-01 4.73342717e-01 1.01611763e-01 -3.60669553e-01 -8.68273616e-01 9.98645648e-02 -1.02862823e+00 -1.57425597e-01 -8.22649717e-01 -9.46418226e-01 9.46313798e-01 2.79317528e-01 -1.30537355e+00 -7.06641257e-01 -5.59061050e-01 -4.01394486e-01 1.07784271e+00 -1.19309735e+00 -1.08548152e+00 -4.39573526e-01 3.83392662e-01 6.50592148e-01 -2.65450746e-01 7.03909576e-01 3.02247524e-01 -3.32967132e-01 3.40935051e-01 3.13753784e-01 -4.25944209e-01 4.78259385e-01 -1.07204533e+00 2.03249291e-01 8.41424346e-01 1.13234758e-01 2.61991799e-01 1.01149380e+00 -4.00682062e-01 -7.72408664e-01 -8.39763582e-01 8.96511614e-01 -1.83523521e-02 1.07002757e-01 -1.78937867e-01 -6.72648013e-01 5.03742695e-02 2.88499922e-01 -2.68227220e-01 1.36545062e-01 -5.62448621e-01 1.37387887e-01 -1.97392151e-01 -1.24048972e+00 5.16067326e-01 4.03187484e-01 -5.70333228e-02 -2.50974566e-01 1.12178721e-01 -1.80958491e-03 -1.52455747e-01 -6.50893211e-01 3.31631422e-01 5.43651879e-01 -1.51738846e+00 9.10585582e-01 1.78634319e-02 7.64347315e-01 -3.18893254e-01 -3.84190343e-02 -1.27097666e+00 -2.61198401e-01 -4.77665454e-01 5.49155176e-01 1.44052756e+00 3.63850981e-01 -4.22916472e-01 7.77268052e-01 7.62157142e-02 2.41053462e-01 -6.32189393e-01 -6.28873289e-01 -7.31160522e-01 -1.73731536e-01 -1.95292205e-01 5.74549317e-01 5.05575359e-01 -6.09496832e-01 3.45902920e-01 -5.43871939e-01 6.75835758e-02 8.52210939e-01 2.53299117e-01 1.14439058e+00 -1.11679447e+00 -5.19506574e-01 -5.30761063e-01 -5.70383966e-01 -8.32355618e-01 -1.61440074e-01 -4.16824460e-01 2.32872933e-01 -1.24175227e+00 2.00220626e-02 -5.32117248e-01 4.35322315e-01 -7.74138644e-02 9.14464220e-02 5.43009162e-01 5.78570738e-02 1.80204287e-01 1.50312170e-01 3.72053385e-01 1.38466311e+00 9.83190089e-02 -8.29503983e-02 2.71118253e-01 -4.13055450e-01 6.12342119e-01 6.58472240e-01 -1.75065681e-01 -4.28002298e-01 -4.53623906e-02 9.51049291e-03 2.24822193e-01 5.43147802e-01 -1.10164702e+00 -1.52440891e-01 -5.86270988e-02 5.03941655e-01 -4.29024220e-01 4.23526496e-01 -8.99432659e-01 8.05732012e-01 4.98324573e-01 -1.81461871e-01 -1.56551942e-01 5.61921000e-02 2.96255976e-01 -5.06688237e-01 -7.72568583e-01 1.27888346e+00 -1.86282650e-01 -6.58544898e-01 2.95023881e-02 -2.87247360e-01 -3.74653310e-01 1.17970526e+00 -5.24197340e-01 1.18539914e-01 -4.08665031e-01 -8.55444551e-01 -7.06160665e-01 5.10798931e-01 1.22936413e-01 5.33627510e-01 -1.32953584e+00 -8.39681268e-01 5.76718688e-01 8.99771005e-02 -2.26636410e-01 3.81038547e-01 4.89020526e-01 -8.53916228e-01 2.33241580e-02 -4.81450886e-01 -7.61822701e-01 -1.34058440e+00 5.80250561e-01 2.40449011e-01 -1.52725086e-01 -3.12099338e-01 4.98919368e-01 4.92911905e-01 -2.00184941e-01 1.74076650e-02 -8.72603878e-02 -3.81633401e-01 7.87510276e-02 5.14329076e-01 2.26750165e-01 1.75781041e-01 -8.44448805e-01 1.13591649e-01 9.71223891e-01 2.98014283e-01 -3.73326302e-01 1.27721238e+00 1.53299747e-02 -3.56612056e-01 2.09181771e-01 9.93481755e-01 1.84397668e-01 -1.10818195e+00 -6.82079569e-02 -2.23741591e-01 -8.14940929e-01 -3.41510504e-01 -5.18073142e-01 -1.03139174e+00 6.84734344e-01 6.24537885e-01 4.55811441e-01 1.38160908e+00 -1.13946311e-01 1.31025493e-01 -8.84175673e-02 1.88098356e-01 -8.16157460e-01 3.22150618e-01 -8.38587433e-03 1.21719396e+00 -8.11969519e-01 6.16322504e-03 -7.43584216e-01 -4.81529951e-01 1.36063933e+00 1.91290110e-01 -3.35098028e-01 4.42496121e-01 1.62741646e-01 -8.20766538e-02 -9.96363834e-02 -3.73543948e-01 -7.96807036e-02 4.62943405e-01 7.26330101e-01 2.94352144e-01 -8.77643079e-02 -5.36767840e-01 6.51988462e-02 -2.16866881e-01 -1.36876320e-02 5.23652256e-01 5.76959670e-01 -3.71662259e-01 -1.35679793e+00 -7.44257569e-01 1.67443186e-01 -2.43884191e-01 8.94206539e-02 -3.22352648e-01 8.27198505e-01 3.79664421e-01 6.91093624e-01 2.65401928e-03 -1.70930088e-01 3.59600008e-01 -1.26698494e-01 7.37819254e-01 -4.04687881e-01 -1.96150601e-01 3.32378328e-01 -3.91760394e-02 -1.34979114e-01 -6.61781430e-01 -6.53557241e-01 -4.30906743e-01 -1.23113871e-01 -4.63274568e-01 2.10289001e-01 8.31082761e-01 7.50418723e-01 6.44041300e-02 4.78693634e-01 8.91640604e-01 -1.26619124e+00 -3.24578971e-01 -7.21436679e-01 -8.19687366e-01 4.52717304e-01 1.19062580e-01 -6.56575441e-01 -2.32209593e-01 5.28208435e-01]
[11.671029090881348, -0.7433927059173584]
4dda640d-f906-4ed1-b9cc-37b0b83b3ad1
modeling-of-spatio-temporal-hawkes-processes
2003.03671
null
https://arxiv.org/abs/2003.03671v2
https://arxiv.org/pdf/2003.03671v2.pdf
Modeling of Spatio-Temporal Hawkes Processes with Randomized Kernels
We investigate spatio-temporal event analysis using point processes. Inferring the dynamics of event sequences spatiotemporally has many practical applications including crime prediction, social media analysis, and traffic forecasting. In particular, we focus on spatio-temporal Hawkes processes that are commonly used due to their capability to capture excitations between event occurrences. We introduce a novel inference framework based on randomized transformations and gradient descent to learn the process. We replace the spatial kernel calculations by randomized Fourier feature-based transformations. The introduced randomization by this representation provides flexibility while modeling the spatial excitation between events. Moreover, the system described by the process is expressed within closed-form in terms of scalable matrix operations. During the optimization, we use maximum likelihood estimation approach and gradient descent while properly handling positivity and orthonormality constraints. The experiment results show the improvements achieved by the introduced method in terms of fitting capability in synthetic and real datasets with respect to the conventional inference methods in the spatio-temporal Hawkes process literature. We also analyze the triggering interactions between event types and how their dynamics change in space and time through the interpretation of learned parameters.
['Suleyman Serdar Kozat', 'Fatih Ilhan']
2020-03-07
null
null
null
null
['crime-prediction']
['miscellaneous']
[ 3.65018994e-02 -3.76338720e-01 1.91014066e-01 -1.55832976e-01 -3.47020566e-01 -4.62358952e-01 1.03107691e+00 5.21099627e-01 -5.74599147e-01 6.54222429e-01 3.96703184e-01 -1.42042443e-01 -5.86035252e-01 -1.00498152e+00 -7.13676214e-01 -9.36316967e-01 -3.61537129e-01 4.16581750e-01 1.75365135e-01 -9.14680064e-02 1.55701697e-01 6.17339194e-01 -1.33591235e+00 -1.46094576e-01 7.21360505e-01 4.50904995e-01 -1.29939720e-01 6.00656927e-01 7.00822920e-02 8.09889972e-01 -1.53803483e-01 -2.26750925e-01 2.45015562e-01 -9.18641537e-02 -2.65060186e-01 -4.15953994e-02 -3.68824035e-01 -5.91308959e-02 -3.57517362e-01 5.52472174e-01 3.59425604e-01 6.79984570e-01 8.96662414e-01 -1.27684617e+00 -3.72126549e-01 1.71790197e-01 -8.80581915e-01 6.19334340e-01 2.57519096e-01 3.25223356e-02 6.76568568e-01 -8.70495319e-01 3.90842706e-01 1.16199911e+00 7.29260981e-01 -4.52714637e-02 -1.28830731e+00 -3.20890307e-01 1.25567779e-01 4.64767277e-01 -1.52161503e+00 -4.55492362e-02 8.05590332e-01 -6.99395835e-01 8.48718882e-01 4.15052682e-01 5.22615194e-01 1.04868925e+00 2.93524235e-01 6.04389787e-01 8.66908312e-01 -3.35605949e-01 2.47989371e-01 -1.95833400e-01 2.66205698e-01 3.50194007e-01 5.11750840e-02 -1.43420801e-01 -5.89390159e-01 -5.03474057e-01 7.62643456e-01 3.72410625e-01 6.18068129e-02 -1.64866090e-01 -1.22832835e+00 9.62391853e-01 1.38008535e-01 2.75263280e-01 -7.15895295e-01 1.93671376e-01 3.86937022e-01 -2.01160491e-01 7.46058285e-01 -1.46445483e-01 1.28590360e-01 -2.36743361e-01 -1.04842401e+00 5.13630211e-01 5.75364769e-01 4.83022213e-01 5.92513740e-01 -7.34390616e-02 -4.85191524e-01 4.87292320e-01 8.91445577e-02 7.81037927e-01 -8.86853188e-02 -3.98271471e-01 7.49318719e-01 3.19481730e-01 4.41425920e-01 -1.48072517e+00 -5.07093489e-01 -2.74725527e-01 -1.18002570e+00 -3.74881744e-01 4.28112864e-01 -3.36071491e-01 -1.84583083e-01 1.67285740e+00 7.14587450e-01 9.79776084e-01 -2.80144572e-01 5.97698331e-01 1.61431625e-01 1.11865664e+00 2.58424520e-01 -5.57358682e-01 1.35613823e+00 -3.50632936e-01 -1.06196260e+00 4.05245245e-01 2.18989328e-01 -5.59182942e-01 7.23516762e-01 8.00169930e-02 -1.14289320e+00 -2.72710621e-01 -3.53197217e-01 2.02803493e-01 -3.22387755e-01 1.84344649e-01 4.95181680e-01 4.21441972e-01 -7.33699262e-01 3.94378781e-01 -1.18261886e+00 -3.70600849e-01 1.35467485e-01 9.98055190e-02 -1.42283604e-01 5.65202713e-01 -1.25304544e+00 6.90032423e-01 2.75728077e-01 3.52779597e-01 -6.51196241e-01 -7.63639688e-01 -6.72173858e-01 2.81776071e-01 2.78370351e-01 -4.68763322e-01 6.45302236e-01 -2.05442935e-01 -1.20873213e+00 4.16568816e-01 -5.12165248e-01 -5.95146716e-01 7.11302042e-01 -3.05408657e-01 -4.68750119e-01 1.47050172e-01 1.71133101e-01 -9.18581411e-02 8.28478813e-01 -6.45833433e-01 -5.49645782e-01 -1.69239834e-01 -2.65079170e-01 2.07090393e-01 -3.17482352e-01 2.55864531e-01 -1.53405160e-01 -8.21296632e-01 -1.91827733e-02 -9.47114408e-01 -1.38649717e-01 -3.13406914e-01 -3.03156435e-01 -2.24335104e-01 7.88287103e-01 -7.35175133e-01 1.47217071e+00 -1.99548876e+00 1.13932915e-01 5.18601775e-01 -3.20913717e-02 -4.39198874e-02 3.37582082e-01 1.17244780e+00 9.65580065e-03 -2.25167617e-01 -3.72454256e-01 -5.09446144e-01 -3.82814370e-02 8.32133740e-02 -7.45959997e-01 8.17142308e-01 4.36595947e-01 6.38476968e-01 -8.88542473e-01 -3.73026937e-01 5.39164662e-01 7.11288035e-01 -3.86249363e-01 1.72385927e-02 2.02173740e-01 9.11764324e-01 -5.35779119e-01 -7.65544921e-02 5.64473093e-01 -2.14869156e-01 -1.69019759e-01 8.38830397e-02 -5.15657008e-01 -5.39568178e-02 -1.55967820e+00 1.05480182e+00 -4.96128500e-01 4.77799296e-01 -2.49248773e-01 -1.14448857e+00 7.58382678e-01 4.88016188e-01 7.30263114e-01 -2.39139467e-01 -1.34856865e-01 -2.75376201e-01 -3.41690123e-01 -5.97430587e-01 3.85015368e-01 -4.71226759e-02 1.63100809e-02 6.17342770e-01 -2.93972880e-01 3.65642101e-01 3.27545613e-01 -7.07150949e-03 9.74777341e-01 1.04325823e-02 3.45314711e-01 -2.68774956e-01 7.47329414e-01 -1.60108864e-01 4.77373749e-01 8.07104945e-01 1.85828328e-01 2.08317801e-01 5.19104540e-01 -5.42115331e-01 -1.15190017e+00 -1.23982334e+00 -3.17772955e-01 9.04661953e-01 -5.13140112e-02 -1.32112175e-01 -6.26724184e-01 7.96311945e-02 -1.98194772e-01 8.60069036e-01 -8.55096936e-01 -1.49172172e-02 -8.07129800e-01 -1.32495391e+00 5.52261710e-01 3.92282009e-01 5.37903070e-01 -9.39050019e-01 -7.76054621e-01 3.47403735e-01 -2.48773083e-01 -1.00613761e+00 -2.39493385e-01 -2.80326158e-01 -6.74447179e-01 -7.89222658e-01 -5.50871670e-01 -3.03548485e-01 3.05940598e-01 -9.61556360e-02 5.05014837e-01 -5.18267512e-01 -3.99789721e-01 5.46909213e-01 -1.70931831e-01 -2.16331556e-01 -4.62650172e-02 -3.53397392e-02 5.45810387e-02 8.04570317e-01 3.55757058e-01 -8.22966278e-01 -6.93438351e-01 1.78844914e-01 -1.17779374e+00 2.50687823e-02 1.53424799e-01 6.26287460e-01 3.35813195e-01 3.58842820e-01 5.55318654e-01 -5.05730033e-01 7.64908135e-01 -9.26498532e-01 -6.87526584e-01 2.79049456e-01 7.06887618e-02 2.03182593e-01 4.05600429e-01 -4.46631044e-01 -1.54043412e+00 -1.07524157e-01 4.20739770e-01 -1.41817629e-01 -2.31024086e-01 3.83280218e-01 1.80820465e-01 3.84446383e-01 4.35504943e-01 4.35896784e-01 -3.84253770e-01 -2.23524600e-01 3.37477595e-01 2.17583865e-01 4.14956719e-01 -7.03006268e-01 8.00345957e-01 1.20411277e+00 3.70167524e-01 -1.08074450e+00 -4.53190595e-01 -8.09269786e-01 -5.78789830e-01 -3.38339120e-01 1.21721041e+00 -6.81010842e-01 -1.23816073e+00 4.50499147e-01 -1.35683227e+00 2.75325514e-02 -1.08005576e-01 7.46432722e-01 -5.84704220e-01 2.91878164e-01 -7.52562225e-01 -1.29619622e+00 -1.55562058e-01 -7.34942555e-01 1.26354229e+00 -2.42841728e-02 -2.22242445e-01 -1.51523817e+00 5.62615335e-01 -3.75389010e-02 2.24032536e-01 6.95314765e-01 8.56395185e-01 -3.19711804e-01 -5.20647883e-01 -1.79836869e-01 -1.40140235e-01 -2.72059977e-01 1.22994937e-01 6.13742471e-02 -7.58306801e-01 -2.64619123e-02 2.08667234e-01 6.13620281e-01 5.78268409e-01 6.32469237e-01 9.29398954e-01 -2.97843993e-01 -3.63271177e-01 4.33974743e-01 1.39315319e+00 4.25975174e-02 4.02307063e-01 2.16229573e-01 6.94717765e-01 8.10241401e-01 4.32475865e-01 9.20461476e-01 2.33303800e-01 7.53469229e-01 4.92538549e-02 5.07420301e-02 5.92875183e-01 -4.30415422e-01 1.41105250e-01 6.43052638e-01 -4.49364990e-01 -1.37102515e-01 -1.21324790e+00 7.15530813e-01 -2.43411350e+00 -1.47595215e+00 -5.50674975e-01 2.38695383e+00 4.92439300e-01 -7.64099285e-02 2.36131638e-01 1.36322886e-01 9.42011535e-01 1.78691164e-01 -2.29519829e-01 -2.69610345e-01 -1.57216206e-01 2.36996010e-01 6.58027291e-01 6.49558544e-01 -1.23593724e+00 6.55765176e-01 5.75931549e+00 9.37903404e-01 -1.00654066e+00 4.35064495e-01 3.68328750e-01 -2.68086165e-01 -7.32399374e-02 -8.63069296e-02 -8.63210678e-01 5.86599767e-01 8.49272132e-01 -1.31221160e-01 2.95509905e-01 1.52149066e-01 1.15104616e+00 -1.13139242e-01 -7.12712884e-01 1.02762103e+00 -2.34636784e-01 -1.24762332e+00 -1.46410828e-02 5.30587696e-02 7.30635107e-01 -5.33947527e-01 7.17759877e-02 -1.02963567e-01 2.02382773e-01 -9.14828241e-01 6.06935680e-01 1.04426050e+00 6.24209382e-02 -8.11286688e-01 3.56650889e-01 6.01791501e-01 -1.30009353e+00 -2.63277367e-02 -8.60016495e-02 -4.69198555e-01 8.35038483e-01 9.20336246e-01 -8.79574120e-01 3.80657166e-01 4.41774487e-01 6.93450093e-01 -2.39753738e-01 8.35673273e-01 -1.67794541e-01 9.60271358e-01 -7.03741312e-01 -8.24958906e-02 3.51634830e-01 -7.06335068e-01 8.43034148e-01 1.18457305e+00 3.74994546e-01 3.91702950e-02 4.86798249e-02 9.27258015e-01 6.13319099e-01 1.67141587e-01 -6.77130520e-01 3.15015823e-01 4.37954277e-01 1.00561798e+00 -9.08877492e-01 -1.51863620e-01 -2.17963502e-01 7.44985998e-01 2.28487879e-01 6.01385832e-01 -1.21962905e+00 8.12628772e-03 5.03188550e-01 4.12228256e-01 3.13268512e-01 -5.49790919e-01 -2.21704707e-01 -1.11663485e+00 2.59129763e-01 -2.01877907e-01 3.03635687e-01 -5.39542615e-01 -1.22973728e+00 2.20123574e-01 6.87478006e-01 -9.96228337e-01 -3.37275922e-01 -2.11253345e-01 -8.69272470e-01 8.54741514e-01 -1.02343524e+00 -1.11035621e+00 -6.48737699e-03 8.73378813e-01 3.15747589e-01 2.41391316e-01 4.84574199e-01 3.31818551e-01 -7.36795425e-01 7.66602391e-03 2.96081722e-01 -7.21214563e-02 2.70424604e-01 -1.03177142e+00 3.90942484e-01 1.02618027e+00 1.55472711e-01 6.89328492e-01 9.14912939e-01 -8.46245110e-01 -1.01744580e+00 -1.02892280e+00 1.02370286e+00 -4.19469506e-01 1.02993274e+00 -5.70733190e-01 -9.49361801e-01 5.97426951e-01 5.36726229e-02 -1.23301558e-01 6.21463478e-01 5.56374788e-02 1.27031639e-01 -2.42035557e-02 -8.96859825e-01 7.64844298e-01 7.92903006e-01 -4.70354736e-01 -5.14540374e-01 5.71203113e-01 2.55766511e-01 -5.52838221e-02 -7.04461753e-01 1.99364334e-01 1.10596523e-01 -6.60201192e-01 1.02729928e+00 -7.48804510e-01 2.13031068e-01 -4.53152925e-01 2.20769063e-01 -9.79394674e-01 -4.06513244e-01 -7.85963058e-01 -5.45491017e-02 1.24715126e+00 1.17960639e-01 -7.69957006e-01 4.14183527e-01 6.89329982e-01 5.25469363e-01 -5.77214956e-01 -1.00911653e+00 -4.95010912e-01 -2.70936370e-01 -6.18923903e-01 6.23578846e-01 8.36176991e-01 -1.56057015e-01 1.53038934e-01 -8.16376626e-01 6.24101341e-01 8.43887806e-01 -1.40621707e-01 6.68115258e-01 -9.86039400e-01 -5.31095326e-01 -7.87824541e-02 -4.87847358e-01 -6.00187838e-01 1.50392503e-01 -4.45998698e-01 -1.69960424e-01 -1.21077430e+00 1.50190458e-01 -3.50512832e-01 -1.31824042e-03 7.04990551e-02 -2.90111482e-01 -1.06879577e-01 -5.58121912e-02 3.01216990e-01 -2.47033283e-01 6.96280539e-01 6.73632979e-01 3.37278217e-01 -3.90082181e-01 2.09156066e-01 2.16349766e-01 7.36725688e-01 8.40111256e-01 -4.55199987e-01 -5.01211524e-01 -2.90867507e-01 6.47786081e-01 2.16592044e-01 8.31190169e-01 -8.92090559e-01 4.00695324e-01 -2.39797279e-01 1.30674755e-02 -6.17792964e-01 5.56397855e-01 -7.78895020e-01 5.21748066e-01 1.44811645e-01 -4.63768065e-01 3.35432410e-01 1.83009401e-01 9.58061576e-01 -2.78381228e-01 -1.44569933e-01 2.97520161e-01 2.13156879e-01 -4.05417591e-01 2.43740305e-01 -7.73134232e-01 -1.01295905e-02 1.35402501e+00 -8.31303224e-02 7.49722049e-02 -6.59723938e-01 -9.98245776e-01 1.00619718e-01 -1.01485148e-01 1.70574307e-01 2.65228152e-01 -1.27181363e+00 -7.43693948e-01 -3.41880508e-02 -2.71611840e-01 -2.59558648e-01 6.35737479e-01 1.37803268e+00 -5.11050940e-01 4.31877673e-01 5.46406470e-02 -6.96279347e-01 -1.08006799e+00 5.03992319e-01 9.54736620e-02 -6.15994513e-01 -5.61224580e-01 2.96125233e-01 3.10503185e-01 -2.05976948e-01 -1.26631320e-01 -3.20538342e-01 -3.52230549e-01 1.03389725e-01 4.47917879e-01 8.60291302e-01 -9.34192091e-02 -6.70167744e-01 -2.77953833e-01 5.41066289e-01 3.50639373e-01 -4.92814869e-01 1.47031450e+00 -2.50686407e-01 -1.33404091e-01 9.05708075e-01 9.48234141e-01 6.32565767e-02 -1.29488575e+00 -1.81627408e-01 8.94391984e-02 -2.27471724e-01 -1.99726135e-01 -8.32377840e-03 -4.78441328e-01 8.96154284e-01 4.10847723e-01 4.78292137e-01 8.42543840e-01 -2.33160108e-01 5.49052119e-01 2.30844691e-01 2.09039927e-01 -1.00354898e+00 -3.78488451e-01 4.19282824e-01 7.01866150e-01 -8.11594605e-01 -2.70589918e-01 -6.33515537e-01 -4.34774965e-01 8.93772185e-01 -1.03563093e-01 -6.01648092e-01 1.04340100e+00 3.86409694e-03 -5.46887994e-01 -1.31744340e-01 -5.52452087e-01 -1.06436595e-01 1.88497573e-01 2.78252423e-01 1.89120650e-01 2.11971149e-01 -6.00009680e-01 1.54263452e-01 -1.74847752e-01 -9.15987492e-02 2.39619389e-01 6.48670435e-01 -1.39234483e-01 -6.94852233e-01 -7.27250278e-01 1.89465880e-01 -5.05682528e-01 -1.22856930e-01 2.20611796e-01 7.20269561e-01 1.32270008e-01 1.01479614e+00 2.91549563e-01 3.88323545e-01 2.36495972e-01 1.94661260e-01 2.79399157e-01 -3.63073766e-01 -5.19922853e-01 -2.90304851e-02 -1.42093033e-01 -3.08499813e-01 -5.05072176e-01 -1.16700184e+00 -1.04315782e+00 -3.02893907e-01 -9.15549770e-02 3.58405709e-01 5.45174658e-01 1.18789995e+00 2.08003998e-01 4.66816872e-01 5.74246466e-01 -5.85955143e-01 -3.03457826e-01 -8.18659663e-01 -4.99877721e-01 5.72148621e-01 2.53138453e-01 -8.24900329e-01 -2.20027640e-01 2.02574208e-01]
[6.820353031158447, 3.593109607696533]
6f09ea46-d38d-42b5-a162-33b522953fd1
suspicious-vehicle-detection-using-licence
2304.14507
null
https://arxiv.org/abs/2304.14507v1
https://arxiv.org/pdf/2304.14507v1.pdf
Suspicious Vehicle Detection Using Licence Plate Detection And Facial Feature Recognition
With the increasing need to strengthen vehicle safety and detection, the availability of pre-existing methods of catching criminals and identifying vehicles manually through the various traffic surveillance cameras is not only time-consuming but also inefficient. With the advancement of technology in every field the use of real-time traffic surveillance models will help facilitate an easy approach. Keeping this in mind, the main focus of our paper is to develop a combined face recognition and number plate recognition model to ensure vehicle safety and real-time tracking of running-away criminals and stolen vehicles.
['Manoj Kumar Rajagopal', 'Bala Murugan MS', 'Manideep Ramisetty', 'Aaron George Pichappa', 'Vrinda Agarwal']
2023-04-18
null
null
null
null
['face-recognition']
['computer-vision']
[-3.18878181e-02 -4.49941605e-01 -1.23708867e-01 -1.89240575e-01 -1.91200338e-02 -7.10900605e-01 6.90015256e-01 -1.67157993e-01 -6.67044759e-01 6.25029266e-01 -5.42076170e-01 -5.96591055e-01 2.59364881e-02 -7.71081030e-01 -2.26298526e-01 -4.74214047e-01 1.90371871e-01 2.21308634e-01 5.56361973e-01 -5.35002649e-02 6.23484731e-01 1.15892828e+00 -1.66388762e+00 -2.68041581e-01 4.56595421e-01 7.36100078e-01 1.32170469e-02 6.23608232e-01 -1.40124306e-01 4.93457973e-01 -3.54219288e-01 -7.96889842e-01 4.38947141e-01 -6.01837449e-02 -4.37838614e-01 3.09655424e-02 2.82693148e-01 -9.41978037e-01 -5.37047088e-01 1.09034133e+00 1.47083431e-01 6.90996423e-02 6.05913222e-01 -1.37594402e+00 -3.76247227e-01 -2.96783924e-01 -5.42911589e-01 3.50251824e-01 2.90496558e-01 3.19806904e-01 -1.87012434e-01 -5.55983603e-01 3.99172097e-01 1.12986648e+00 6.72588348e-01 8.02069545e-01 -7.26710021e-01 -1.12894475e+00 -2.79580802e-01 3.14668238e-01 -1.64235520e+00 -8.16653788e-01 6.65485501e-01 -6.30938232e-01 7.89881051e-01 7.27496818e-02 5.31365156e-01 8.50031734e-01 -1.80169549e-02 1.53395385e-01 7.40075171e-01 -5.79379499e-01 -8.53493214e-02 5.48109472e-01 4.39682215e-01 6.92287564e-01 1.12496769e+00 3.20909888e-01 8.06760117e-02 -5.48008606e-02 7.29513824e-01 5.63308060e-01 2.82257557e-01 7.73854479e-02 -3.22847664e-01 8.08051646e-01 -4.23437655e-01 5.99248528e-01 -3.22319597e-01 -2.91454792e-02 4.82812703e-01 -1.71346292e-01 2.11804360e-01 -2.38639876e-01 -8.82249326e-03 -3.64802092e-01 -1.00652039e+00 -7.52139762e-02 4.67294514e-01 7.05295146e-01 7.53558993e-01 1.11062102e-01 2.90174633e-01 3.33536148e-01 6.54742777e-01 1.01196265e+00 -1.77545756e-01 -8.63609314e-01 5.15577734e-01 9.34402764e-01 1.93438321e-01 -1.19817376e+00 -1.23015745e-02 3.68329287e-01 -4.73411292e-01 5.81485093e-01 5.54455400e-01 -3.74626160e-01 -7.52353013e-01 9.49649513e-01 3.41101378e-01 1.40857205e-01 -2.83077538e-01 3.17478895e-01 3.85825962e-01 6.44390583e-01 4.38078970e-01 -3.67709011e-01 1.32100117e+00 -3.37500274e-01 -1.08301079e+00 -1.24729639e-02 4.33393270e-01 -6.01155639e-01 1.46015525e-01 1.01549461e-01 -5.69754779e-01 -4.73131329e-01 -7.74129808e-01 4.15065795e-01 -6.78376853e-01 1.73416361e-01 4.55389440e-01 1.62989593e+00 -8.84646595e-01 -1.52448164e-02 -6.97920918e-01 -6.04744792e-01 5.77206433e-01 6.73077941e-01 -7.14179158e-01 -1.98032632e-01 -8.02942693e-01 1.13283312e+00 2.62075923e-02 3.98718685e-01 -6.17288709e-01 -3.05900544e-01 -6.90081716e-01 -2.82361448e-01 3.32764119e-01 8.17685127e-02 7.99180150e-01 -7.97641158e-01 -1.16083682e+00 9.39559817e-01 -3.60278338e-01 -1.14106856e-01 6.09886050e-01 -8.28418359e-02 -8.46760869e-01 2.89762348e-01 4.37102057e-02 2.80002743e-01 7.14754105e-01 -1.16159451e+00 -9.15835738e-01 -6.53678477e-01 -4.63284642e-01 -2.51535714e-01 -3.58552516e-01 7.24518657e-01 -2.22908840e-01 9.56248194e-02 -6.22905731e-01 -7.26355612e-01 9.99171063e-02 -1.28564999e-01 4.88770045e-02 -5.30179799e-01 1.67816412e+00 -8.58264387e-01 1.27353311e+00 -2.23951530e+00 -8.73795033e-01 2.45836750e-01 6.40711114e-02 1.31889892e+00 1.30253106e-01 2.94777691e-01 1.24572232e-01 5.90994358e-02 1.91405848e-01 -1.44859523e-01 -3.03988546e-01 1.73516303e-01 -9.02237650e-03 5.83790481e-01 2.50495225e-01 4.83990133e-01 -7.24833846e-01 -6.77242398e-01 8.96046162e-01 7.56592453e-01 1.24901816e-01 1.13754272e-01 6.98196709e-01 9.87209082e-02 -5.86309075e-01 8.89827430e-01 1.00697708e+00 4.29026663e-01 -3.91336046e-02 4.15243417e-01 -5.32794476e-01 -2.61242092e-01 -1.23602962e+00 2.98617363e-01 1.89874515e-01 7.64670968e-01 1.82109639e-01 -7.45445371e-01 9.09397066e-01 6.20224416e-01 4.31600183e-01 -8.44279587e-01 3.43825191e-01 2.68843919e-01 -4.99788016e-01 -8.70075166e-01 6.14030838e-01 3.58173586e-02 3.45221668e-01 1.28972515e-01 -2.06329688e-01 6.67685151e-01 1.87806234e-01 -1.25058144e-01 8.15128863e-01 -2.04019472e-01 1.99435517e-01 4.55320291e-02 9.78073716e-01 1.17967486e-01 3.25117379e-01 1.92878962e-01 -8.90349209e-01 -5.69091029e-02 7.88107291e-02 -5.25227845e-01 -8.05186749e-01 -7.23253250e-01 -2.31790673e-02 6.76999390e-01 -1.48991540e-01 3.88955176e-01 -9.92035210e-01 -6.89411938e-01 -2.31370796e-02 4.48231220e-01 -2.76950300e-01 5.38622327e-02 -7.38806129e-01 -4.54247504e-01 8.33810270e-01 2.93758392e-01 6.30784452e-01 -8.07303846e-01 -4.93126422e-01 1.86259493e-01 2.23356575e-01 -1.10745513e+00 -1.63640127e-01 -3.90236288e-01 -7.99050450e-01 -1.52517092e+00 -8.45667183e-01 -7.40528286e-01 1.07964289e+00 8.73147130e-01 1.85968027e-01 5.78976035e-01 -1.32661596e-01 4.68156606e-01 -3.08544159e-01 -6.41319156e-01 -5.61167777e-01 -2.38854051e-01 1.01554312e-01 2.68007308e-01 1.02074718e+00 7.66273215e-02 -3.59786302e-01 3.20846975e-01 -7.54494369e-01 -6.59063816e-01 2.23969415e-01 -1.07211098e-01 -3.12189579e-01 3.11210454e-01 6.82059944e-01 -4.90473598e-01 4.71063614e-01 -4.09799486e-01 -1.19726789e+00 1.92348287e-01 -6.08879209e-01 -5.91521800e-01 3.47700745e-01 -2.80549318e-01 -1.02619648e+00 3.26279789e-01 -6.21976964e-02 -1.81452483e-01 -8.30207169e-01 -3.99927586e-01 -2.38188535e-01 -4.82253492e-01 1.84516713e-01 3.27662945e-01 4.25695211e-01 -3.46803695e-01 -1.64236858e-01 9.64236140e-01 4.34609354e-01 2.32595474e-01 9.03301716e-01 4.93584454e-01 1.63210049e-01 -1.32148278e+00 6.94955960e-02 -8.36278617e-01 -6.80258751e-01 -9.50209618e-01 1.04463947e+00 -5.87434590e-01 -1.29233265e+00 1.02057433e+00 -1.52979195e+00 2.38319173e-01 4.38996643e-01 5.49585938e-01 2.76959658e-01 9.00363266e-01 -2.32975170e-01 -1.58811057e+00 -1.19779401e-01 -9.75798249e-01 7.16356754e-01 4.03480530e-01 1.27131075e-01 -8.72373879e-01 2.02832058e-01 6.75335407e-01 4.87795442e-01 6.81231245e-02 2.81141341e-01 -3.77554655e-01 -7.36910880e-01 -9.35883403e-01 -5.07633507e-01 5.45271277e-01 2.76582628e-01 4.97764826e-01 -8.49024773e-01 1.24208875e-01 6.87572593e-03 2.82247275e-01 4.43926662e-01 3.55277747e-01 3.27735037e-01 -1.03209518e-01 -5.55309713e-01 7.36633018e-02 1.52697885e+00 9.04993057e-01 1.06525862e+00 2.98723966e-01 4.71872658e-01 9.39982116e-01 4.53951508e-01 1.30864844e-01 4.36923325e-01 4.82002646e-01 4.10928398e-01 7.70103261e-02 8.65610689e-02 3.73128057e-02 4.10427898e-01 1.84455380e-01 -6.13716304e-01 -5.28063718e-03 -9.58620489e-01 5.74861705e-01 -1.51170540e+00 -1.61353767e+00 -5.62531769e-01 2.17588234e+00 1.62216395e-01 -1.30666509e-01 5.62763333e-01 4.04486984e-01 1.34771252e+00 -3.80087912e-01 1.63936585e-01 -4.76001114e-01 4.34770256e-01 -2.14576200e-01 8.72371316e-01 4.92612928e-01 -1.13025105e+00 7.03666389e-01 7.48027372e+00 6.62303507e-01 -1.19186759e+00 -1.38376588e-02 6.62251413e-01 2.93196052e-01 3.05434138e-01 -3.70450556e-01 -1.24731290e+00 7.42085457e-01 1.16477096e+00 2.06018180e-01 5.06630063e-01 6.64724350e-01 3.76942098e-01 -4.30747241e-01 -4.97777253e-01 9.09628391e-01 1.80897102e-01 -1.06116676e+00 -2.33843908e-01 3.34021837e-01 1.49063572e-01 -3.36639434e-01 -1.94841266e-01 1.57702968e-01 -1.66447744e-01 -8.26425910e-01 4.66715664e-01 3.72314930e-01 6.10405922e-01 -9.02702808e-01 1.01570594e+00 3.97259533e-01 -1.27562022e+00 -8.93972069e-02 -1.76770613e-01 -2.36788437e-01 3.44565421e-01 1.17911220e-01 -7.21066177e-01 1.14492975e-01 3.48498046e-01 6.96157292e-02 -4.07049328e-01 1.01554894e+00 9.67046618e-02 6.81229532e-01 -3.28446060e-01 -3.08157027e-01 2.43906423e-01 -3.37821305e-01 1.78290829e-01 1.29446113e+00 2.84530103e-01 1.05147973e-01 -3.93296629e-01 4.45852935e-01 3.61726135e-01 -2.23056555e-01 -1.21958947e+00 -1.22987546e-01 5.25418222e-01 1.22786081e+00 -8.76293838e-01 -1.55900642e-01 -8.74137700e-01 4.57381666e-01 -1.83861151e-01 8.96525830e-02 -9.45882738e-01 -3.96971166e-01 6.71200097e-01 6.01283133e-01 2.76240915e-01 -5.76247752e-01 -2.14368224e-01 -5.11268258e-01 -6.48575723e-02 -2.43918747e-01 1.18393235e-01 -1.67126417e-01 -8.27332914e-01 2.57538885e-01 3.70627880e-01 -1.03116786e+00 -6.43685758e-02 -9.23325181e-01 -7.89950371e-01 7.60081112e-01 -1.46767342e+00 -1.18449044e+00 4.74837013e-02 6.78588390e-01 3.07774603e-01 -5.03048599e-01 4.95451242e-01 7.16099441e-01 -7.26821899e-01 4.36485350e-01 -5.82928583e-02 5.63794434e-01 1.89842120e-01 -2.66510993e-01 1.77257791e-01 1.06295466e+00 -5.68331659e-01 5.56253791e-01 3.76754284e-01 -9.77279067e-01 -1.47851431e+00 -9.50339437e-01 1.18718457e+00 -7.28240132e-01 5.33274829e-01 -2.60368973e-01 -6.11465752e-01 4.77875382e-01 1.49361134e-01 -3.03709239e-01 5.13852477e-01 -3.04896504e-01 -6.82481155e-02 -3.05150509e-01 -1.48372972e+00 2.93519706e-01 4.68383551e-01 -5.17629206e-01 -6.11868024e-01 9.20952018e-03 1.44073078e-02 3.64862978e-01 -9.48594064e-02 -4.93670516e-02 6.17023170e-01 -7.58446515e-01 9.08922374e-01 -2.05397204e-01 -3.71219218e-01 -2.07710117e-01 1.95672169e-01 -6.08153865e-02 -1.53638974e-01 -5.11284053e-01 3.72895509e-01 1.61441314e+00 5.16390987e-02 -8.72473300e-01 9.43708837e-01 1.18020236e+00 3.46797764e-01 8.00397322e-02 -1.20545638e+00 -9.43235815e-01 -4.39391524e-01 -5.87986350e-01 3.54426414e-01 7.18300521e-01 7.51999673e-03 1.55593781e-02 -5.91275096e-01 3.90405655e-01 8.53939354e-01 -8.28738868e-01 6.34007931e-01 -1.37991226e+00 6.48844063e-01 -3.59215170e-01 -8.65336478e-01 -1.63360551e-01 -7.51733333e-02 -3.44881192e-02 -2.00403795e-01 -1.16206479e+00 3.97722453e-01 -5.81401819e-03 8.23601112e-02 3.43667716e-01 1.10744290e-01 3.05451959e-01 1.70989782e-01 -2.63253637e-02 -5.88035464e-01 -4.85989116e-02 7.58687139e-01 1.70667037e-01 8.56466293e-02 2.90516108e-01 -4.67013210e-01 7.44102538e-01 8.01986575e-01 -7.61440575e-01 -1.29740566e-01 -1.92732096e-01 -1.00294180e-01 -5.15508167e-02 6.33278847e-01 -1.02059817e+00 4.70457882e-01 -3.27499241e-01 1.54511929e-01 -8.58842432e-01 1.59695029e-01 -1.43591499e+00 1.52231619e-01 7.02150881e-01 5.59961021e-01 1.40877485e-01 3.87585908e-01 5.63377440e-01 3.86087969e-02 -5.92770159e-01 7.56123245e-01 -2.93067209e-02 -7.60273099e-01 1.83987781e-01 -1.23615348e+00 -5.68871677e-01 1.79884791e+00 -9.39361572e-01 -5.65159738e-01 -6.10278249e-02 5.50829209e-02 1.03370789e-02 5.04102886e-01 3.39770347e-01 4.85147536e-01 -1.03264797e+00 -3.30362380e-01 4.58887160e-01 -3.24971467e-01 -5.48763394e-01 3.81468713e-01 4.05461788e-01 -9.31801736e-01 9.34249401e-01 -7.13883758e-01 -1.65136546e-01 -1.76380336e+00 8.10948789e-01 9.21521634e-02 1.05367914e-01 -2.36464128e-01 3.13473761e-01 -2.97149986e-01 4.65704463e-02 3.10917228e-01 3.29054207e-01 -6.72054708e-01 -1.59181729e-01 9.86635387e-01 1.25401330e+00 -1.11010000e-01 -1.32143295e+00 -6.60384476e-01 6.22303307e-01 1.11397929e-01 1.35539815e-01 9.85094368e-01 -2.69257069e-01 -1.66595906e-01 -7.48507306e-02 8.88253093e-01 -2.21646596e-02 -1.02676237e+00 4.97170717e-01 1.13443710e-01 -7.66285717e-01 1.82213001e-02 -4.04815674e-01 -9.96190369e-01 7.58119583e-01 7.88488984e-01 5.17327666e-01 8.96531045e-01 -3.74862880e-01 7.93735445e-01 3.72369438e-01 6.45157754e-01 -1.15935707e+00 -4.27298129e-01 3.84342194e-01 2.05309510e-01 -1.51754487e+00 -1.41701862e-01 -5.45098007e-01 -2.72559851e-01 1.23981082e+00 5.50848424e-01 -3.72183360e-02 5.37484050e-01 3.09894770e-01 1.00788638e-01 -7.91682228e-02 -7.79091939e-02 -1.71115413e-01 6.97881207e-02 1.15183318e+00 1.48703605e-01 2.74794213e-02 -4.44411218e-01 2.42000371e-01 7.09921956e-01 4.51066464e-01 4.31367189e-01 1.01332176e+00 -8.46774757e-01 -1.29865944e+00 -9.56459939e-01 2.14635208e-01 -7.61263967e-01 5.17473400e-01 -5.07282615e-01 1.07292533e+00 3.88679773e-01 1.44952321e+00 -1.60942394e-02 -2.24175841e-01 2.17074722e-01 5.61901331e-02 1.79090023e-01 -1.36839524e-01 -4.82692689e-01 -4.10456926e-01 1.44331738e-01 -1.66731432e-01 -7.08900452e-01 -8.82507443e-01 -9.27131534e-01 -1.03119123e+00 -3.23698372e-01 -3.66979279e-02 1.07618117e+00 1.10018122e+00 8.12428668e-02 -2.28593811e-01 5.23994565e-01 -7.75560319e-01 -2.97560275e-01 -5.57354808e-01 -6.70719802e-01 1.57718316e-01 3.28545839e-01 -6.92026138e-01 -1.90113887e-01 9.56732854e-02]
[13.053977966308594, 0.9787275791168213]
5374385b-06d7-47ad-8c76-3483a87a2f96
unsupervised-cross-spectral-stereo-matching
1903.01078
null
http://arxiv.org/abs/1903.01078v1
http://arxiv.org/pdf/1903.01078v1.pdf
Unsupervised Cross-spectral Stereo Matching by Learning to Synthesize
Unsupervised cross-spectral stereo matching aims at recovering disparity given cross-spectral image pairs without any supervision in the form of ground truth disparity or depth. The estimated depth provides additional information complementary to individual semantic features, which can be helpful for other vision tasks such as tracking, recognition and detection. However, there are large appearance variations between images from different spectral bands, which is a challenge for cross-spectral stereo matching. Existing deep unsupervised stereo matching methods are sensitive to the appearance variations and do not perform well on cross-spectral data. We propose a novel unsupervised cross-spectral stereo matching framework based on image-to-image translation. First, a style adaptation network transforms images across different spectral bands by cycle consistency and adversarial learning, during which appearance variations are minimized. Then, a stereo matching network is trained with image pairs from the same spectra using view reconstruction loss. At last, the estimated disparity is utilized to supervise the spectral-translation network in an end-to-end way. Moreover, a novel style adaptation network F-cycleGAN is proposed to improve the robustness of spectral translation. Our method can tackle appearance variations and enhance the robustness of unsupervised cross-spectral stereo matching. Experimental results show that our method achieves good performance without using depth supervision or explicit semantic information.
['You Song', 'Mingyang Liang', 'Hongsheng Li', 'Xiaoyang Guo', 'Xiaogang Wang']
2019-03-04
null
null
null
null
['stereo-matching']
['computer-vision']
[ 8.30464363e-01 -3.47813874e-01 -4.72382531e-02 -4.23303187e-01 -6.80932045e-01 -4.87744182e-01 3.29532087e-01 -3.01740944e-01 -3.80389035e-01 4.48525816e-01 -4.23229672e-02 1.37909621e-01 2.15280697e-01 -9.79152501e-01 -8.77977431e-01 -8.57432961e-01 8.49450350e-01 9.74530652e-02 2.97649682e-01 -1.61574945e-01 1.98537067e-01 2.26421565e-01 -1.53053260e+00 2.73505058e-02 1.18733788e+00 9.72908854e-01 4.25776958e-01 1.35104448e-01 -8.91232118e-02 4.61478233e-01 -1.32627159e-01 -3.15456718e-01 6.48192108e-01 -4.97105956e-01 -4.07272846e-01 7.18070030e-01 7.68550634e-01 -4.61803079e-01 -4.36910391e-01 1.55548882e+00 5.31474233e-01 7.44189397e-02 3.48736823e-01 -9.82268572e-01 -2.93991357e-01 -8.89630802e-03 -7.27269292e-01 -3.30386758e-01 3.35720181e-01 5.07554710e-01 6.64157510e-01 -7.83729732e-01 4.27440077e-01 1.18179476e+00 6.09787345e-01 4.36080843e-01 -1.40286803e+00 -9.49866772e-01 -8.61428678e-02 2.97468528e-02 -1.35021532e+00 -4.98431116e-01 1.38874078e+00 -4.11756516e-01 2.58164883e-01 -6.62632510e-02 7.25824833e-01 8.24012399e-01 -2.14461550e-01 4.90500122e-01 1.32618797e+00 -3.65082264e-01 -8.92465934e-02 -2.13511549e-02 -6.46002412e-01 7.20314324e-01 -1.00293964e-01 5.76088071e-01 -6.33367419e-01 3.36390972e-01 9.01716590e-01 2.44926482e-01 -5.26091456e-01 -7.04959869e-01 -1.30948365e+00 5.94363570e-01 8.09554219e-01 8.82127602e-03 -8.86021033e-02 -1.03878662e-01 2.72289157e-01 3.49213243e-01 4.64750141e-01 7.56923854e-02 -3.58150601e-01 4.78319466e-01 -6.43697262e-01 -1.40655845e-01 1.81590691e-02 8.86433303e-01 1.31276536e+00 2.52235472e-01 3.20375651e-01 9.66861427e-01 3.19868743e-01 9.30215895e-01 5.11294782e-01 -1.05041063e+00 7.47451901e-01 6.54321194e-01 4.60820310e-02 -1.10866749e+00 -1.47187173e-01 -4.08972323e-01 -1.04216611e+00 3.62352312e-01 2.14468420e-01 1.83948860e-01 -7.50429451e-01 1.75114489e+00 5.26726484e-01 8.78915936e-02 4.78836522e-02 1.14998341e+00 5.22720814e-01 4.22697842e-01 -2.08936870e-01 -3.01453859e-01 8.74802589e-01 -8.03245306e-01 -4.81075525e-01 -4.33017403e-01 2.18335032e-01 -1.08072829e+00 9.02713120e-01 5.37718311e-02 -1.05916691e+00 -7.92769909e-01 -1.08384216e+00 -1.40882730e-01 -2.11533353e-01 -1.17415607e-01 3.26381683e-01 6.88429594e-01 -6.99827790e-01 3.24405938e-01 -5.27970850e-01 -1.96626931e-01 2.36932516e-01 2.45509997e-01 -3.28216136e-01 -3.36015344e-01 -1.09329343e+00 5.51766276e-01 5.82646191e-01 9.40418318e-02 -4.82462466e-01 -5.12121379e-01 -1.22968829e+00 -4.20882434e-01 3.69534850e-01 -9.58711386e-01 7.24507570e-01 -1.63607562e+00 -1.66047406e+00 1.20149326e+00 -1.62989005e-01 -2.58194566e-01 7.30383992e-01 8.87071639e-02 -3.70816827e-01 2.96579242e-01 3.53482574e-01 6.34148717e-01 1.19019496e+00 -1.19056749e+00 -5.64277768e-01 -5.75150251e-01 -6.29849210e-02 5.09993255e-01 -1.58579811e-01 -2.69177884e-01 -6.76357985e-01 -7.41835535e-01 4.70280945e-01 -7.58339047e-01 2.22303029e-02 3.03096771e-01 -3.57487857e-01 4.64774132e-01 7.33064950e-01 -5.34335256e-01 5.72200775e-01 -2.10787964e+00 2.43013620e-01 1.17454574e-01 -1.83974624e-01 1.01804957e-01 -2.43660569e-01 -1.21004451e-02 -1.69745877e-01 -6.43436193e-01 -7.25054145e-01 -2.12381318e-01 -4.34863925e-01 7.77301639e-02 -2.48059928e-02 7.83524573e-01 1.16946287e-01 6.71437144e-01 -1.03084266e+00 -5.65996408e-01 7.61819601e-01 4.17941362e-01 -4.85537767e-01 3.81233096e-01 -2.29691595e-01 9.66874242e-01 -2.53060758e-01 6.14178836e-01 1.07422674e+00 7.52185956e-02 4.34410796e-02 -6.19793117e-01 -9.89891514e-02 -1.76614225e-02 -1.01310229e+00 2.35166359e+00 -6.83887780e-01 4.89597261e-01 4.33689877e-02 -1.10507214e+00 1.00631917e+00 -7.80449659e-02 5.21238685e-01 -1.06755900e+00 -3.38104963e-02 5.35350323e-01 -2.78589875e-01 -3.19709420e-01 7.23620877e-02 -3.32295120e-01 1.93942368e-01 5.05171865e-02 -2.45352224e-01 -7.07432866e-01 -1.40924901e-01 -2.85046637e-01 3.99666548e-01 1.99501008e-01 -1.03912741e-01 3.17476690e-04 1.05893898e+00 -2.71810591e-01 7.77914762e-01 1.07228279e-01 9.44044590e-02 8.78579974e-01 -3.02198410e-01 -3.35310161e-01 -1.30364621e+00 -1.20650434e+00 -2.58448236e-02 4.44936752e-01 5.82262039e-01 1.22111998e-01 -7.66014516e-01 -4.49174911e-01 -6.61777630e-02 1.62985861e-01 -3.10128152e-01 -2.34831899e-01 -2.96877533e-01 -2.11358771e-01 1.95915639e-01 3.53970289e-01 1.09355807e+00 -8.21179032e-01 -3.52439195e-01 1.32097840e-01 -5.15908241e-01 -1.35096097e+00 -1.00433922e+00 -2.17940271e-01 -9.88208830e-01 -1.27683568e+00 -7.54290223e-01 -8.92154217e-01 7.64701068e-01 6.59053445e-01 8.06111336e-01 -1.48934990e-01 -4.00031328e-01 2.80948788e-01 -2.79478915e-02 -2.06285380e-02 -3.30749035e-01 -8.07133541e-02 -3.71763036e-02 5.55547237e-01 1.55187190e-01 -7.94696510e-01 -1.01511335e+00 7.20147908e-01 -1.13721228e+00 4.89844143e-01 4.63952005e-01 1.08250141e+00 7.19308019e-01 8.66712779e-02 1.36055619e-01 -8.47794533e-01 -1.92234367e-01 1.54087633e-01 -8.98458481e-01 -5.80826811e-02 -5.60057342e-01 -3.57956067e-02 7.44242966e-01 -2.84406334e-01 -1.26539016e+00 5.36708415e-01 3.05712130e-02 -6.91183686e-01 -4.01791893e-02 9.75760594e-02 -6.03718162e-01 -3.58431607e-01 5.40436327e-01 7.62043595e-01 3.98434043e-01 -1.94976926e-01 1.77202493e-01 5.55097044e-01 8.90364230e-01 -3.11822653e-01 1.26279449e+00 9.48187292e-01 1.22782156e-01 -7.65218139e-01 -9.88806069e-01 -6.26240730e-01 -7.28588402e-01 -2.93261826e-01 9.25577760e-01 -1.35429513e+00 -4.19224828e-01 9.36399281e-01 -1.01911449e+00 -3.45352829e-01 1.04914177e-02 4.76407111e-01 -8.56459677e-01 8.21995199e-01 -4.78818595e-01 -4.72449332e-01 -2.53017306e-01 -1.17272091e+00 1.21485090e+00 2.74492741e-01 3.50135386e-01 -1.06199884e+00 -1.72533289e-01 8.71850610e-01 1.96694404e-01 1.90728039e-01 6.90813303e-01 4.56343234e-01 -8.46269786e-01 9.79335606e-02 -4.47035730e-01 6.41144872e-01 4.73447591e-01 -4.10191655e-01 -9.50888097e-01 -4.44496274e-01 4.65635881e-02 -2.86306888e-01 8.27778339e-01 2.32211173e-01 1.20023179e+00 5.43805510e-02 1.75054874e-02 1.28146601e+00 1.68618751e+00 6.60539567e-02 7.59797931e-01 4.94196534e-01 1.06824744e+00 8.96374583e-01 8.78441811e-01 7.97494054e-02 2.66672194e-01 8.31752777e-01 5.88446736e-01 -4.14291143e-01 -2.11932272e-01 -4.16952491e-01 4.54924971e-01 5.95642805e-01 7.51646161e-02 3.36045265e-01 -6.17864847e-01 4.97153252e-01 -1.65560544e+00 -9.32259023e-01 -1.07026204e-01 2.55861783e+00 9.41400588e-01 -1.07328452e-01 -1.13993898e-01 1.92504525e-01 1.07003629e+00 2.31715590e-01 -8.94483089e-01 3.20344955e-01 -5.17510831e-01 1.61421955e-01 7.43661463e-01 5.61169386e-01 -9.91939306e-01 9.62862611e-01 4.46783257e+00 8.53407860e-01 -1.26916504e+00 1.14121214e-01 5.10362923e-01 1.77678078e-01 -4.14203018e-01 -2.08961964e-03 -7.69711956e-02 6.90955758e-01 7.23704100e-02 1.26657188e-01 6.94119334e-01 4.60805595e-01 2.22328439e-01 -1.38160095e-01 -8.34587038e-01 1.45829320e+00 2.01066792e-01 -1.04676986e+00 -5.64549305e-02 -1.06986284e-01 1.03325963e+00 3.57261561e-02 2.23726898e-01 -3.47340584e-01 -2.26369128e-02 -6.47426844e-01 6.40407264e-01 3.47426802e-01 1.18413413e+00 -9.51340199e-01 4.27340031e-01 3.14122677e-01 -1.33089960e+00 1.58848956e-01 -5.20077586e-01 3.21550846e-01 2.17088014e-02 5.24246097e-01 -1.53965503e-01 7.72623301e-01 5.52034616e-01 1.07930958e+00 -3.02157223e-01 7.71284521e-01 -3.71423423e-01 -9.11659822e-02 -2.08844051e-01 6.74981356e-01 6.07571285e-03 -6.71214759e-01 5.90849161e-01 7.32774019e-01 2.62565434e-01 -2.23095357e-01 2.42810279e-01 8.41484010e-01 3.58846635e-02 5.96355647e-02 -6.79087996e-01 2.83907145e-01 3.15789610e-01 9.46501315e-01 -4.04683858e-01 -1.45799071e-01 -6.25010014e-01 1.53448999e+00 -7.11134821e-02 3.58797938e-01 -6.22126818e-01 -1.87325567e-01 6.49367332e-01 3.08065861e-01 1.40097007e-01 -4.62893248e-02 -1.69886693e-01 -1.51822376e+00 2.18334675e-01 -8.50580752e-01 2.09256262e-01 -1.14495683e+00 -1.35383558e+00 2.21929327e-01 -5.12313128e-01 -1.85425353e+00 -7.60667175e-02 -4.79493827e-01 -4.71913695e-01 9.53147829e-01 -1.93605423e+00 -1.34736967e+00 -6.96012735e-01 1.13528264e+00 5.84604979e-01 -2.25879848e-01 4.90333080e-01 3.72626513e-01 -3.73186499e-01 5.91797113e-01 3.08448255e-01 2.05709398e-01 1.13614178e+00 -9.94031072e-01 1.07149646e-01 9.12161827e-01 -3.41229290e-01 2.84402460e-01 4.83568966e-01 -4.88639355e-01 -1.36713386e+00 -1.42755032e+00 3.06424379e-01 2.16217399e-01 3.68958741e-01 -4.95099798e-02 -7.41351068e-01 3.05419356e-01 -9.97055322e-05 1.38405263e-01 3.73373628e-01 -4.53824878e-01 -5.57482600e-01 -4.89778906e-01 -1.04950452e+00 4.44510251e-01 1.21875405e+00 -1.11542845e+00 -1.94830149e-01 3.14881384e-01 6.65155172e-01 -5.95045090e-01 -6.87998533e-01 3.71035427e-01 5.32514155e-01 -1.36414754e+00 1.22768021e+00 7.54533708e-02 6.23947263e-01 -6.46783233e-01 -1.46723732e-01 -1.32517338e+00 1.06226966e-01 -4.71935868e-01 5.73489964e-01 1.28572309e+00 8.47375616e-02 -8.28694344e-01 7.74637282e-01 2.19552338e-01 -1.28184289e-01 -9.41462740e-02 -6.63271487e-01 -8.27776790e-01 -7.65864551e-02 -3.15589845e-01 4.87931699e-01 1.23665357e+00 -5.66242635e-01 2.90321410e-01 -4.30796981e-01 3.66036803e-01 1.30467618e+00 5.80260336e-01 1.04407096e+00 -1.01533997e+00 -3.43939662e-01 -3.32502723e-01 -5.88768303e-01 -1.09520102e+00 3.57696891e-01 -8.31513524e-01 5.65854609e-02 -1.01037478e+00 2.23799154e-01 -2.27453902e-01 -4.64681089e-02 -3.44789475e-02 -1.97698623e-01 5.82428217e-01 5.78586385e-02 4.09754992e-01 -1.65680230e-01 7.95110047e-01 1.52148283e+00 -5.33111691e-01 -3.76967080e-02 -1.15602314e-01 -2.60875076e-01 7.66669989e-01 6.63129508e-01 -2.47640014e-01 -5.40466309e-01 -5.80930829e-01 6.45492375e-02 3.43983829e-01 5.11681139e-01 -1.04943407e+00 3.08152258e-01 -2.20787317e-01 4.74998534e-01 -5.44486344e-01 2.76996732e-01 -1.13602602e+00 2.01042667e-01 4.17090356e-01 -4.01895829e-02 -3.76266003e-01 -5.03700562e-02 7.40888953e-01 -5.09279430e-01 3.44083011e-02 1.23598337e+00 -1.29899696e-01 -6.75058842e-01 6.48306370e-01 2.17732385e-01 -4.34692018e-02 8.54807854e-01 -5.14976382e-01 6.02548569e-02 -3.07595819e-01 -3.60230386e-01 1.59728080e-01 1.15341151e+00 2.95134485e-01 7.15392232e-01 -1.52616167e+00 -5.23940146e-01 4.47840393e-01 6.23768210e-01 4.66891885e-01 6.17832124e-01 7.86238432e-01 -6.82067394e-01 4.79290560e-02 -4.76581514e-01 -1.00444341e+00 -1.19622612e+00 6.63564801e-01 7.22237468e-01 1.43931717e-01 -4.80851114e-01 5.55056274e-01 6.65905237e-01 -5.99855900e-01 3.84271368e-02 1.03532113e-01 2.75423527e-01 -9.72144008e-02 2.75601983e-01 -3.23834270e-02 -5.30210435e-02 -8.81826222e-01 -1.38699263e-01 1.25984323e+00 1.53771251e-01 -7.29237646e-02 1.03578496e+00 -5.85582614e-01 -1.16760790e-01 2.69478619e-01 1.53992796e+00 5.21255359e-02 -1.52289963e+00 -8.61435354e-01 -7.00221300e-01 -9.41212595e-01 1.84327468e-01 -3.37519199e-01 -1.60163975e+00 1.05850911e+00 8.68931651e-01 -3.48650992e-01 1.59046125e+00 -5.14481902e-01 8.95489991e-01 -8.95080268e-02 4.55467939e-01 -1.12038016e+00 1.86622530e-01 1.99966341e-01 5.44352412e-01 -1.64857697e+00 -1.63125023e-01 -6.36043906e-01 -3.68659079e-01 1.21743011e+00 7.13171601e-01 1.01599339e-02 2.78936386e-01 -1.80842012e-01 2.08973765e-01 5.60245104e-02 -1.18856234e-02 -5.04907846e-01 2.46390179e-01 7.99907982e-01 1.91630527e-01 -1.29831225e-01 5.38581423e-02 -3.28391314e-01 3.71124744e-02 -1.92521960e-01 1.44131064e-01 4.96997863e-01 -1.83810771e-01 -1.10116899e+00 -4.86236334e-01 -1.86206952e-01 -3.12496610e-02 -1.15563802e-01 -1.89524442e-01 3.98687333e-01 2.86351204e-01 7.93601632e-01 6.58593327e-02 -4.78453517e-01 3.13262105e-01 -1.96136415e-01 6.69194102e-01 -4.88113374e-01 -2.44681105e-01 2.76410282e-01 -3.63860220e-01 -6.93814874e-01 -8.77026916e-01 -6.25984251e-01 -9.30029929e-01 -3.24445605e-01 -2.64490664e-01 -3.24969113e-01 5.38668871e-01 7.05216169e-01 -7.53615499e-02 2.37234473e-01 1.18999493e+00 -8.65175366e-01 -1.18141115e-01 -5.83566070e-01 -5.91876209e-01 8.82499158e-01 5.40905178e-01 -5.74315846e-01 -4.32774067e-01 3.29894036e-01]
[9.074911117553711, -2.337958335876465]
b766ef30-03b4-4614-85c5-a92580a9a46f
finding-the-law-enhancing-statutory-article
2301.12847
null
https://arxiv.org/abs/2301.12847v1
https://arxiv.org/pdf/2301.12847v1.pdf
Finding the Law: Enhancing Statutory Article Retrieval via Graph Neural Networks
Statutory article retrieval (SAR), the task of retrieving statute law articles relevant to a legal question, is a promising application of legal text processing. In particular, high-quality SAR systems can improve the work efficiency of legal professionals and provide basic legal assistance to citizens in need at no cost. Unlike traditional ad-hoc information retrieval, where each document is considered a complete source of information, SAR deals with texts whose full sense depends on complementary information from the topological organization of statute law. While existing works ignore these domain-specific dependencies, we propose a novel graph-augmented dense statute retriever (G-DSR) model that incorporates the structure of legislation via a graph neural network to improve dense retrieval performance. Experimental results show that our approach outperforms strong retrieval baselines on a real-world expert-annotated SAR dataset.
['Gerasimos Spanakis', 'Gijs Van Dijck', 'Antoine Louis']
2023-01-30
null
null
null
null
['ad-hoc-information-retrieval']
['natural-language-processing']
[ 1.97251037e-01 3.56574118e-01 -7.96714664e-01 -7.92191476e-02 -1.31892848e+00 -7.84984708e-01 5.02277374e-01 6.97681546e-01 -4.14174587e-01 5.41491807e-01 9.71550941e-01 -8.43957365e-01 -7.79654264e-01 -1.06773508e+00 -4.70045686e-01 -1.54373229e-01 3.41209114e-01 9.68275428e-01 1.54130086e-01 -7.23595321e-01 5.34641802e-01 5.85955441e-01 -6.92241669e-01 5.27331591e-01 1.47634196e+00 4.15845692e-01 -2.03566015e-01 7.24703670e-02 -4.12211567e-01 1.35408640e+00 -7.32216656e-01 -7.61782765e-01 3.25852871e-01 7.66948014e-02 -1.12784958e+00 -5.42582810e-01 8.95188749e-01 -4.62740928e-01 -1.20628154e+00 1.30163968e+00 4.61994052e-01 1.29869983e-01 9.41145301e-01 -2.44352937e-01 -1.14395201e+00 9.45920348e-01 -8.06850255e-01 5.47300875e-01 8.80684733e-01 -3.70618671e-01 1.65085649e+00 -3.33456397e-01 1.36895001e+00 1.23007870e+00 4.35978621e-01 -5.12448512e-03 -8.13150704e-01 -1.41253054e-01 6.80169389e-02 4.92198050e-01 -1.20481849e+00 -3.01846057e-01 1.11106646e+00 -3.71851981e-01 1.22549975e+00 2.10201927e-02 4.56748873e-01 8.47850263e-01 4.57583725e-01 9.86220300e-01 4.88780379e-01 -3.18641514e-01 1.93442684e-02 -6.80029988e-01 7.79484212e-01 5.17808020e-01 9.15359795e-01 -5.47262132e-01 1.60967082e-01 -5.60655832e-01 3.43212724e-01 1.64543450e-01 -1.09438427e-01 4.03987095e-02 -6.20771587e-01 8.90089929e-01 4.25193727e-01 7.52457857e-01 -6.15152240e-01 -2.17044279e-02 5.37653029e-01 5.11203408e-01 3.87557447e-01 6.22770965e-01 -1.47005618e-01 7.61552751e-02 -1.00083959e+00 6.21053874e-01 7.75698066e-01 1.11312437e+00 5.29466093e-01 -4.62020785e-01 -6.80066705e-01 9.16402876e-01 1.93828970e-01 7.19980180e-01 9.70193297e-02 -7.90154278e-01 1.36956692e+00 1.36171842e+00 -2.25082129e-01 -1.33678269e+00 -3.52601171e-01 -5.11880636e-01 -6.13502264e-01 -3.98850560e-01 2.64239848e-01 1.35358617e-01 -1.01309597e+00 8.81660879e-01 1.79222986e-01 -5.52285075e-01 7.91356564e-02 7.39665985e-01 1.23571384e+00 5.87247133e-01 5.56587540e-02 7.26703927e-02 1.40832520e+00 -4.33894783e-01 -9.19076383e-01 -2.62448460e-01 8.39326024e-01 -6.83385432e-01 7.16934562e-01 -1.84814945e-01 -1.04644513e+00 8.69797368e-04 -6.54409885e-01 -9.26448941e-01 -6.70983016e-01 4.43680175e-02 7.24910378e-01 2.22774297e-01 -8.25036764e-01 2.93293566e-01 -2.44685590e-01 -2.40306646e-01 9.62970793e-01 -2.05509022e-01 -2.94223279e-01 -6.52504504e-01 -1.52393949e+00 9.48215485e-01 2.71458566e-01 7.73228821e-04 -4.18281034e-02 -8.12083662e-01 -1.08586633e+00 3.72698486e-01 8.81690025e-01 -6.68197930e-01 9.43915844e-01 2.60263622e-01 -4.62681323e-01 9.58953738e-01 -2.52828240e-01 -5.51392019e-01 2.96799779e-01 -1.83600560e-02 -5.01259446e-01 6.66394949e-01 5.65902352e-01 1.42010245e-02 3.40721399e-01 -7.59607434e-01 -1.65106416e-01 -6.36100113e-01 5.87126672e-01 2.08881646e-01 -1.87243998e-01 3.40905249e-01 -6.29794776e-01 -5.88907123e-01 8.92764181e-02 -4.54935551e-01 -4.08268571e-01 -5.55965781e-01 -4.66535658e-01 -5.73908985e-01 6.17332935e-01 -1.03525472e+00 1.61907887e+00 -1.73732221e+00 1.45340676e-03 6.83152497e-01 6.09538138e-01 3.23074102e-01 -2.86225855e-01 9.99424338e-01 3.33116889e-01 1.76585749e-01 -3.78726423e-02 3.57567072e-01 1.66001588e-01 2.22741351e-01 -6.82415187e-01 3.86902392e-01 -3.05957109e-01 1.58121562e+00 -9.41086292e-01 -9.06409025e-01 -1.77226424e-01 1.79033607e-01 -4.18217599e-01 -5.80657363e-01 -2.21732557e-01 -2.50008047e-01 -1.21426749e+00 1.00531244e+00 4.22730386e-01 -5.08433759e-01 4.03559297e-01 -1.42510563e-01 4.04497415e-01 7.13578105e-01 -3.57177317e-01 1.81567633e+00 -1.23381294e-01 6.21569574e-01 -6.99551171e-03 -1.06441832e+00 6.73307121e-01 1.92456126e-01 9.08953071e-01 -1.27731109e+00 4.11942452e-02 -3.28624360e-02 -1.11081928e-01 -6.12706900e-01 8.60220075e-01 3.83889407e-01 -3.67951065e-01 5.22404194e-01 -4.24049556e-01 2.74459943e-02 7.87260413e-01 1.14168954e+00 1.85782254e+00 -3.51466954e-01 4.89198953e-01 -7.16588646e-02 2.87025571e-01 5.01108825e-01 4.07507539e-01 1.00809145e+00 5.68808652e-02 8.33146572e-02 6.74296737e-01 -4.08193916e-01 -8.33734810e-01 -7.86640227e-01 -2.29491785e-01 7.04978406e-01 6.08607568e-02 -5.01515210e-01 -3.31287622e-01 -9.88800228e-01 4.70129788e-01 5.75390816e-01 -4.63376909e-01 1.75274521e-01 -8.35684121e-01 -1.23568371e-01 5.18113613e-01 3.91894728e-01 2.72476315e-01 -1.09473717e+00 -8.23839828e-02 1.74588069e-01 -3.75816703e-01 -1.38245547e+00 -6.81609035e-01 -5.62239230e-01 -7.70439088e-01 -1.61469245e+00 -5.56563258e-01 -6.47754967e-01 8.68594110e-01 5.28968930e-01 1.13205063e+00 4.57822800e-01 -5.08605063e-01 5.48402429e-01 -4.81026113e-01 -2.05803886e-01 -3.12390924e-01 5.09271801e-01 -5.16786277e-01 -6.69449329e-01 8.27487946e-01 -5.08482099e-01 -4.47107345e-01 -2.78252810e-01 -1.24494636e+00 -4.83427614e-01 6.53708816e-01 2.69040257e-01 4.86302674e-01 3.14147621e-02 6.78720355e-01 -1.40008223e+00 1.48367178e+00 -4.04452622e-01 -8.84999037e-01 7.15798676e-01 -8.47051680e-01 1.19464152e-01 4.26726878e-01 2.42740840e-01 -1.02602828e+00 -4.49293703e-01 -1.08001031e-01 -2.08030082e-02 4.70761470e-02 1.27149057e+00 4.66923006e-02 2.02997923e-01 8.42538714e-01 -2.23749373e-02 -2.71870345e-01 -5.87530255e-01 6.72372103e-01 7.61119485e-01 5.64538360e-01 -6.58379972e-01 9.98931050e-01 4.75180894e-01 1.54635027e-01 -5.88049948e-01 -1.25424576e+00 -1.18701696e+00 -8.60799015e-01 -2.80402780e-01 6.26541853e-01 -9.30452049e-01 -4.36153650e-01 -4.07772720e-01 -1.46177697e+00 2.94794649e-01 -3.77114505e-01 2.13568136e-01 -3.57320197e-02 8.96960676e-01 -6.64162815e-01 -4.50898290e-01 -6.32899106e-01 -6.09017730e-01 1.30856907e+00 -2.67942786e-01 -8.54483917e-02 -1.10354125e+00 3.57772678e-01 1.32483506e+00 5.44076003e-02 3.36124480e-01 1.31233191e+00 -8.28717768e-01 -6.41218901e-01 -8.39226723e-01 -5.30986786e-01 -1.98326826e-01 1.24274008e-01 -4.55809474e-01 -2.92456269e-01 -1.15139596e-01 -4.26806122e-01 -1.59322709e-01 1.30965650e+00 4.80174333e-01 7.49245465e-01 -8.11733365e-01 -6.16378784e-01 1.38654083e-01 1.37340462e+00 1.81533694e-01 1.05310845e+00 3.73789400e-01 9.24329042e-01 4.57771540e-01 4.70139802e-01 1.75893888e-01 6.57314062e-01 4.40537572e-01 3.52096893e-02 -1.77686125e-01 -2.59063840e-01 -4.93171632e-01 1.00411095e-01 6.47940576e-01 -3.82255197e-01 -3.56079191e-01 -1.25962782e+00 6.46677077e-01 -2.10073876e+00 -1.62420094e+00 -3.07635933e-01 1.56757641e+00 7.68991172e-01 2.11515259e-02 2.71148793e-03 -8.05620328e-02 5.16975582e-01 4.91260052e-01 -5.45535922e-01 -1.50624931e-01 -4.44122314e-01 2.14124694e-01 7.71685243e-01 5.10823309e-01 -9.87017393e-01 1.27513838e+00 6.51413107e+00 1.06088221e+00 -2.83809304e-01 -3.77952904e-02 6.38261810e-02 -8.64150897e-02 -8.16135406e-01 8.17413554e-02 -7.39765167e-01 1.74504891e-01 4.87346798e-01 -5.73085368e-01 3.36133957e-01 9.25964355e-01 2.04280287e-01 1.85758993e-01 -6.57856584e-01 6.00637019e-01 1.27499044e-01 -1.82302582e+00 4.45666343e-01 5.99019229e-01 8.54525566e-01 3.39476556e-01 -1.75739631e-01 4.28714812e-01 7.51683772e-01 -8.03714633e-01 2.99777925e-01 6.87110424e-01 7.40772188e-01 -4.66509312e-01 8.85440171e-01 2.50296295e-01 -1.06085372e+00 -1.89936668e-01 -6.91185653e-01 1.86581045e-01 3.36745381e-01 7.51365244e-01 -7.57115424e-01 1.06989777e+00 2.33368874e-01 1.20664465e+00 -6.98416889e-01 9.20908391e-01 -5.68923295e-01 5.02383649e-01 5.62917851e-02 -1.44607201e-01 4.43506539e-01 -2.84948856e-01 8.12780797e-01 1.16337156e+00 -8.41234401e-02 4.19933379e-01 1.87262401e-01 6.40973330e-01 -5.83573580e-01 5.12278497e-01 -1.30541611e+00 -2.13728905e-01 3.95550340e-01 1.08433068e+00 -4.58116710e-01 -5.41122735e-01 -5.49462616e-01 3.56399685e-01 5.60838938e-01 6.78995967e-01 -3.73394907e-01 -6.43063545e-01 9.31370258e-02 5.21023631e-01 4.08679068e-01 -1.90307781e-01 -1.41810402e-01 -1.51134574e+00 3.91873151e-01 -8.54537129e-01 1.11657357e+00 -8.24191570e-01 -1.44405878e+00 1.65269315e-01 1.09488063e-01 -1.18401372e+00 -3.71648520e-01 -4.87542689e-01 -3.35864872e-01 5.14529705e-01 -1.79009616e+00 -1.22821701e+00 3.65673810e-01 6.04284048e-01 2.01430097e-01 -4.82553422e-01 3.22030813e-01 5.52390039e-01 -2.53631532e-01 2.51163185e-01 2.74337173e-01 7.15122640e-01 4.87420559e-01 -1.09635746e+00 2.30069444e-01 8.51927519e-01 3.47734928e-01 1.15078592e+00 1.89564198e-01 -1.29693115e+00 -1.55479372e+00 -1.03112340e+00 1.15964794e+00 -5.27949512e-01 9.84936953e-01 -6.82571307e-02 -9.13311660e-01 6.39238775e-01 3.60700339e-01 -5.29893219e-01 6.77625239e-01 6.34114325e-01 -8.31100821e-01 -9.25719365e-02 -9.69050944e-01 6.57857537e-01 1.43973100e+00 -9.34536278e-01 -1.36344576e+00 9.47473764e-01 4.82188880e-01 -2.82854378e-01 -1.05802965e+00 8.73372033e-02 4.12660599e-01 -2.07791552e-01 1.14564455e+00 -1.05890214e+00 5.94685674e-01 -4.67720628e-02 2.29354590e-01 -7.00741172e-01 -6.12975776e-01 -4.79003638e-01 -1.58549219e-01 7.22162545e-01 6.30427718e-01 -4.53542501e-01 5.12998760e-01 7.91583180e-01 -1.74355134e-01 -4.64112669e-01 -1.11108077e+00 -9.89648223e-01 1.94102019e-01 -3.78162235e-01 7.83502698e-01 1.21785867e+00 6.75712168e-01 7.24859834e-01 -3.50057073e-02 2.18717903e-02 6.50563180e-01 6.43004715e-01 5.49699843e-01 -1.56858504e+00 9.51331556e-02 -5.84205389e-01 -5.62670290e-01 -1.14439571e+00 5.89479327e-01 -1.54827118e+00 -4.85745221e-01 -2.68386650e+00 4.39225584e-01 1.15322553e-01 -1.33469671e-01 7.82130480e-01 -2.83723082e-02 -3.13548744e-01 -2.04837769e-01 6.13274455e-01 -1.13866997e+00 2.50435561e-01 1.47188568e+00 -8.54259789e-01 -1.21980004e-01 -2.87334830e-01 -1.16618645e+00 4.21082407e-01 4.37241197e-01 -5.20119905e-01 -5.09043753e-01 -6.91737413e-01 7.99675107e-01 2.03432381e-01 2.15617828e-02 -4.25116241e-01 8.91118526e-01 -2.17258573e-01 -2.65273619e-02 -9.33905959e-01 -1.74297348e-01 -8.05905402e-01 -5.36539495e-01 1.19657762e-01 -5.49272954e-01 -1.05808496e-01 -6.96208104e-02 9.45116460e-01 -4.07700121e-01 -4.11376238e-01 -2.59880256e-02 -1.90076202e-01 -2.87908763e-01 4.51308668e-01 -4.44282204e-01 3.04070711e-01 3.86712968e-01 -1.71357661e-01 -1.15695286e+00 -5.81234097e-01 -2.45821163e-01 5.00571489e-01 1.49510518e-01 3.47764790e-01 7.30728507e-01 -1.13906395e+00 -9.82128143e-01 -5.33365250e-01 4.17072594e-01 -3.61591652e-02 2.08959654e-01 4.94559824e-01 -3.80602777e-01 1.08120227e+00 3.28290612e-01 6.26903996e-02 -1.27023005e+00 6.62977993e-01 -2.13471770e-01 -9.92147207e-01 -1.04157853e+00 2.45324546e-03 -3.28210652e-01 -1.74791485e-01 -1.32965922e-01 -3.63799095e-01 -8.28072906e-01 1.02968633e-01 5.19732952e-01 2.27500558e-01 1.36169523e-01 -7.56014764e-01 -1.49063036e-01 7.10580111e-01 -6.67474806e-01 8.11973959e-02 1.67826021e+00 3.45912814e-01 -5.95599771e-01 -4.21451956e-01 9.04317915e-01 5.35489082e-01 -1.11869812e-01 -5.56624830e-01 3.72569978e-01 -4.33153600e-01 2.60165572e-01 -7.76858985e-01 -1.07218170e+00 2.43743077e-01 -4.17571753e-01 3.14715326e-01 8.45306754e-01 3.49522382e-01 8.81604791e-01 1.26026368e+00 3.59205693e-01 -1.41381168e+00 -2.55854726e-01 7.85132587e-01 1.16907275e+00 -9.75545406e-01 6.62089527e-01 -4.76724535e-01 -4.77095395e-01 1.06776643e+00 -5.70555478e-02 1.21411547e-01 4.62615073e-01 -5.29567786e-02 6.11083172e-02 -1.10411763e+00 -5.27253628e-01 -4.16699022e-01 9.86137986e-01 3.85072231e-01 2.15605080e-01 -4.72828239e-01 -8.36380541e-01 2.57661730e-01 -7.40565360e-02 -4.91328426e-02 3.42760772e-01 1.02098632e+00 -5.16283035e-01 -1.16063082e+00 -1.00516945e-01 1.01323915e+00 -5.52068532e-01 -4.96106476e-01 -9.70668435e-01 7.46989250e-01 -5.45683801e-01 1.00287867e+00 -3.56673688e-01 1.32511005e-01 4.30229276e-01 -1.14689097e-01 3.15898925e-01 -7.94822037e-01 -6.25840783e-01 -4.80292290e-02 4.50258374e-01 -4.76146281e-01 -4.11937743e-01 -7.58519351e-01 -1.43021369e+00 -2.28233561e-01 -2.33288724e-02 4.15684909e-01 3.66718322e-01 1.17372370e+00 5.83887339e-01 4.63629812e-01 1.45178204e-02 3.66429538e-01 -4.30887967e-01 -9.06940699e-01 -6.43963039e-01 6.20080054e-01 1.27971962e-01 -2.34247655e-01 -2.40927581e-02 -2.06649035e-01]
[9.82319164276123, 9.050939559936523]
3c9cf053-b25a-48aa-aead-768c734e17d3
representation-learning-for-person-or-entity
2305.0564
null
https://arxiv.org/abs/2305.05640v2
https://arxiv.org/pdf/2305.05640v2.pdf
Representation Learning for Person or Entity-centric Knowledge Graphs: An Application in Healthcare
Knowledge graphs (KGs) are a popular way to organise information based on ontologies or schemas and have been used across a variety of scenarios from search to recommendation. Despite advances in KGs, representing knowledge remains a non-trivial task across industries and it is especially challenging in the biomedical and healthcare domains due to complex interdependent relations between entities, heterogeneity, lack of standardization, and sparseness of data. KGs are used to discover diagnoses or prioritize genes relevant to disease, but they often rely on schemas that are not centred around a node or entity of interest, such as a person. Entity-centric KGs are relatively unexplored but hold promise in representing important facets connected to a central node and unlocking downstream tasks beyond graph traversal and reasoning, such as generating graph embeddings and training graph neural networks for a wide range of predictive tasks. This paper presents an end-to-end representation learning framework to extract entity-centric KGs from structured and unstructured data. We introduce a star-shaped ontology to represent the multiple facets of a person and use it to guide KG creation. Compact representations of the graphs are created leveraging graph neural networks and experiments are conducted using different levels of heterogeneity or explicitness. A readmission prediction task is used to evaluate the results of the proposed framework, showing a stable system, robust to missing data, that outperforms a range of baseline machine learning classifiers. We highlight that this approach has several potential applications across domains and is open-sourced. Lastly, we discuss lessons learned, challenges, and next steps for the adoption of the framework in practice.
['Joao Bettencourt-Silva', 'Thaddeus Stappenbeck', 'Natasha Mulligan', 'Christos Theodoropoulos']
2023-05-09
null
null
null
null
['person-centric-knowledge-graphs', 'readmission-prediction']
['graphs', 'medical']
[ 2.66914666e-01 8.84395659e-01 -4.28163320e-01 -3.08722526e-01 -3.16420764e-01 -1.72532752e-01 5.57586551e-01 1.03680634e+00 -1.09881930e-01 7.22715974e-01 6.88487828e-01 -1.98315904e-01 -7.38452375e-01 -1.14431822e+00 -4.91202623e-01 -4.49793279e-01 -4.44563746e-01 7.67795980e-01 -1.51712531e-02 -2.56196678e-01 -1.81708515e-01 4.28460240e-01 -1.36932373e+00 3.35541189e-01 7.62321174e-01 1.05968225e+00 -9.39054191e-02 2.57121831e-01 -2.86541373e-01 8.34104300e-01 -5.52430451e-01 -6.20939851e-01 1.42171821e-02 -3.13256383e-02 -8.58490646e-01 -9.83175710e-02 2.31171742e-01 5.20454109e-01 -4.44089800e-01 8.26855063e-01 6.40510857e-01 -2.39768680e-02 6.93222821e-01 -1.23919845e+00 -9.11666811e-01 6.28176689e-01 2.91497558e-02 9.30339023e-02 4.84186649e-01 -3.38620842e-01 1.27337039e+00 -3.48851681e-01 1.00642252e+00 1.19382799e+00 7.96249270e-01 4.29219157e-01 -9.84650671e-01 -3.20457399e-01 2.15557113e-01 2.86229521e-01 -1.29470444e+00 -1.19737916e-01 4.74227577e-01 -4.32049721e-01 1.21168375e+00 3.12928200e-01 7.81475425e-01 1.14751542e+00 2.55720049e-01 4.09151345e-01 8.05365682e-01 -3.19420785e-01 1.90072075e-01 2.08902314e-01 3.00228000e-01 9.98585999e-01 7.77921438e-01 -3.62496585e-01 -5.50070286e-01 -4.97514457e-01 4.07599002e-01 2.62350410e-01 -4.40271795e-01 -6.47098005e-01 -1.24201620e+00 9.07260776e-01 7.61295676e-01 2.41725937e-01 -6.18379176e-01 -1.02761865e-01 5.58270276e-01 8.98007825e-02 4.34307128e-01 7.52921402e-01 -3.99101138e-01 2.10718721e-01 -5.78665257e-01 1.59423307e-01 9.42861438e-01 1.00143838e+00 4.46755052e-01 -1.61614746e-01 -3.60723644e-01 8.56528282e-01 2.60760188e-02 -1.13552898e-01 4.79725212e-01 -1.79281577e-01 5.26952624e-01 1.44549441e+00 -2.82388628e-01 -1.38050139e+00 -8.67295027e-01 -6.19680703e-01 -1.11511934e+00 -3.79189402e-01 -2.64041889e-02 1.24337047e-01 -1.17875707e+00 1.54585028e+00 4.71627384e-01 3.54924053e-01 2.78766632e-01 6.22549057e-01 1.38420367e+00 1.85799450e-01 3.01991224e-01 2.14770764e-01 1.64408982e+00 -7.29861975e-01 -6.97265923e-01 -7.47995526e-02 8.29678655e-01 4.72759595e-03 5.34292519e-01 2.48115540e-01 -6.59905136e-01 -1.79392621e-01 -8.13874543e-01 -2.13662073e-01 -1.08167338e+00 -1.55523747e-01 9.61657524e-01 4.18646365e-01 -1.11729658e+00 5.34347892e-01 -5.47309458e-01 -7.50615180e-01 7.59517133e-01 4.29736197e-01 -7.47954428e-01 -4.09992546e-01 -1.55820918e+00 8.83614480e-01 7.25371897e-01 -7.97061995e-02 -4.29251313e-01 -9.07436490e-01 -1.09792268e+00 3.06124747e-01 5.11615217e-01 -1.23926413e+00 3.11989099e-01 -3.87658745e-01 -7.53486216e-01 8.22663009e-01 2.47827768e-01 -7.57334471e-01 1.18689768e-01 1.54620446e-02 -9.51329112e-01 9.44232270e-02 1.31594524e-01 4.91241693e-01 4.70492661e-01 -6.64361238e-01 -5.96822560e-01 -4.79015291e-01 2.24318281e-01 2.50209808e-01 -5.82500339e-01 -4.15124238e-01 -5.57594776e-01 -5.65040827e-01 -8.50067362e-02 -8.28955770e-01 -2.58364797e-01 -2.93460190e-01 -7.67002940e-01 -4.11540210e-01 5.27592063e-01 -5.72509646e-01 1.50221384e+00 -1.70154715e+00 4.03038621e-01 4.11834657e-01 7.12141037e-01 2.80263513e-01 7.50813559e-02 7.89995909e-01 -1.92066222e-01 2.92398989e-01 -1.35612592e-01 1.61692739e-01 -1.11089110e-01 2.66580284e-01 1.05118923e-01 1.83053344e-01 3.90000701e-01 1.22039676e+00 -1.03347468e+00 -2.61653274e-01 -2.61864327e-02 6.18064404e-01 -4.20032710e-01 4.55885082e-02 -2.73791462e-01 9.21140760e-02 -5.72423697e-01 8.74004662e-01 1.44842371e-01 -7.64469981e-01 5.14458060e-01 -4.97456074e-01 4.81015712e-01 3.34847957e-01 -1.09129858e+00 1.46511590e+00 -2.05883488e-01 1.81291610e-01 -3.14733744e-01 -1.31970835e+00 9.05221939e-01 3.37410778e-01 5.09393632e-01 -4.42683131e-01 -9.41786170e-02 7.23225251e-03 -1.79815348e-02 -6.10545456e-01 3.06471497e-01 2.89814826e-02 -1.70207977e-01 -2.57565593e-03 2.02219889e-01 3.53577137e-01 2.84500867e-01 4.02247816e-01 1.65589762e+00 -2.57709831e-01 7.85749853e-01 -1.85435265e-01 4.13481534e-01 2.35816509e-01 4.98151362e-01 4.11849916e-01 1.13598235e-01 2.48552725e-01 7.23841131e-01 -6.30021513e-01 -7.07733333e-01 -8.19160759e-01 -2.51434982e-01 7.08644032e-01 -1.24994472e-01 -7.22994804e-01 -1.55129597e-01 -8.36419940e-01 4.69725430e-01 4.26413238e-01 -1.00017393e+00 -3.81579876e-01 -1.16667613e-01 -8.22638631e-01 6.11956000e-01 3.94608676e-01 8.11538398e-02 -1.07785881e+00 -3.50225717e-01 3.44596475e-01 2.80497726e-02 -1.35915995e+00 4.90439050e-02 1.42583102e-01 -8.08661282e-01 -1.63498628e+00 -5.88140011e-01 -7.56299675e-01 8.24893355e-01 6.04833243e-04 1.48694170e+00 7.85701349e-02 -5.33249259e-01 6.05170071e-01 -4.00064230e-01 -5.09841144e-01 -2.82131732e-01 3.84505510e-01 -4.29833606e-02 -5.41000590e-02 4.51116025e-01 -3.45154375e-01 -7.10908294e-01 1.45246815e-02 -9.74951327e-01 8.02112520e-02 7.03162968e-01 9.42547977e-01 6.97305918e-01 1.65298715e-01 8.32535028e-01 -1.61407268e+00 9.44681287e-01 -1.04126167e+00 -1.62854195e-01 4.30922747e-01 -8.75624895e-01 1.82296723e-01 5.67649662e-01 -9.05121118e-02 -5.69130063e-01 -3.17234248e-01 8.95697102e-02 -3.07798654e-01 -1.03015177e-01 1.04222643e+00 -6.03734776e-02 1.35991007e-01 7.18541563e-01 -8.49687010e-02 -1.95761919e-02 -4.15658951e-01 4.80148077e-01 5.13780057e-01 3.14471483e-01 -3.65544617e-01 5.08536696e-01 3.66314739e-01 3.78036320e-01 -7.88084924e-01 -7.80101001e-01 -6.22152865e-01 -3.72955173e-01 1.72170773e-01 8.61144364e-01 -9.86913264e-01 -6.30409658e-01 -1.06543079e-01 -7.12631822e-01 1.12621225e-01 -2.94739813e-01 2.21490324e-01 -1.89266726e-01 1.58698961e-01 -3.06191772e-01 -1.44115701e-01 -6.65380716e-01 -8.13403070e-01 1.12324798e+00 1.60082936e-01 -2.99709350e-01 -1.39437664e+00 3.63387261e-03 5.56865096e-01 2.98536897e-01 7.12285578e-01 1.59125733e+00 -1.11085367e+00 -4.42490339e-01 -3.77027422e-01 -2.61552304e-01 -1.74814146e-02 6.01123631e-01 -5.36866426e-01 -6.51290894e-01 -2.91588455e-01 -8.81096542e-01 -2.54147977e-01 7.34867752e-01 1.42183434e-02 1.06608987e+00 -4.61281359e-01 -8.47519815e-01 5.92337191e-01 1.32138205e+00 -1.46857694e-01 5.05209506e-01 3.64421844e-01 1.05343843e+00 7.15118408e-01 2.80446887e-01 3.36788684e-01 7.82798529e-01 5.72351873e-01 5.48578382e-01 -1.26189739e-01 -1.78091988e-01 -2.19932348e-01 -3.80916111e-02 6.41752183e-01 -7.74341896e-02 -3.54973495e-01 -1.14162827e+00 7.48642981e-01 -1.73516452e+00 -7.95635402e-01 -1.51517734e-01 2.12578726e+00 7.31269836e-01 6.05710149e-02 1.16440222e-01 4.98586241e-03 6.12182617e-01 -1.90420414e-03 -6.00449204e-01 -4.71210808e-01 -9.76296738e-02 3.03838700e-01 3.16308171e-01 -3.93309817e-02 -9.35919344e-01 8.94072533e-01 5.44165516e+00 3.89286846e-01 -1.12335622e+00 -1.05729766e-01 4.05425161e-01 1.66212425e-01 -2.62264878e-01 -3.01045299e-01 -7.80275643e-01 2.72322297e-01 1.08372676e+00 -3.11716855e-01 2.20575079e-01 8.19655120e-01 -1.79367155e-01 3.08337361e-01 -1.17055619e+00 8.48668337e-01 1.37972504e-01 -1.75765836e+00 3.33664536e-01 1.64228037e-01 6.63920045e-01 2.16666982e-01 -2.21749932e-01 4.11736846e-01 4.45946097e-01 -1.39279103e+00 5.16051352e-02 6.46845877e-01 8.21067393e-01 -5.65549493e-01 8.26455116e-01 7.12505588e-03 -1.28953433e+00 -1.26385733e-01 -3.78216445e-01 2.43310079e-01 -8.41303542e-02 7.15508163e-01 -1.50821948e+00 1.13206434e+00 6.60326779e-01 8.22434962e-01 -7.78889477e-01 1.06098771e+00 -9.05095115e-02 3.46443892e-01 -9.80285332e-02 4.92207259e-02 1.96445301e-01 6.51319772e-02 3.17374885e-01 1.37796950e+00 4.68544215e-01 -8.29413682e-02 1.43438846e-01 5.84848702e-01 -3.70972753e-01 3.14086974e-01 -1.05907393e+00 -4.72324133e-01 3.78373295e-01 1.45332003e+00 -5.83925486e-01 -3.24625909e-01 -5.06424844e-01 6.64944649e-01 7.04038858e-01 3.08970898e-01 -4.86936957e-01 -4.59107101e-01 8.04246843e-01 3.99840087e-01 3.63428503e-01 2.43022531e-01 1.40571579e-01 -1.05005717e+00 -1.54721126e-01 -9.63939726e-01 1.05220926e+00 -5.49911559e-01 -1.51687062e+00 7.36502707e-01 3.44703905e-02 -1.07489812e+00 -2.91190952e-01 -7.97713459e-01 -2.69615501e-01 8.07607710e-01 -1.73569024e+00 -1.33202958e+00 -5.51826358e-01 5.90297997e-01 -4.36726920e-02 -3.26853573e-01 1.14090765e+00 3.17572057e-01 -4.74889606e-01 5.15142620e-01 -1.27334878e-01 1.82850540e-01 5.40582657e-01 -1.35552573e+00 2.81125754e-01 2.38368049e-01 2.68313259e-01 8.29837680e-01 3.10648769e-01 -8.84849131e-01 -1.57838500e+00 -1.56310248e+00 9.69197631e-01 -4.73748028e-01 6.44285858e-01 -4.17100608e-01 -9.63146687e-01 6.86062455e-01 -6.54614493e-02 4.55693156e-01 1.05588210e+00 5.74813485e-01 -4.08370167e-01 -1.08466975e-01 -1.19559956e+00 5.07659197e-01 1.33769631e+00 -4.16536510e-01 -4.13193941e-01 5.28166711e-01 6.37245655e-01 -3.79716277e-01 -1.39219737e+00 5.07284641e-01 3.61836523e-01 -7.16391206e-01 1.02140808e+00 -1.17438710e+00 1.81725353e-01 -1.39642790e-01 2.31902245e-02 -1.67348909e+00 -5.35275757e-01 -4.31815982e-01 -4.81476307e-01 1.06749141e+00 4.69797999e-01 -8.07145834e-01 8.05539072e-01 4.55172330e-01 -2.16168284e-01 -1.27242863e+00 -8.02203000e-01 -5.11151075e-01 -3.30290735e-01 -1.85660541e-01 7.77363598e-01 1.30905855e+00 2.75743753e-01 5.98842382e-01 -1.44728854e-01 4.48488176e-01 4.17326838e-01 1.18307367e-01 5.85695267e-01 -1.73382175e+00 -1.29919022e-01 -2.08892778e-01 -1.11339712e+00 -2.03808367e-01 1.39337122e-01 -1.58261132e+00 -6.44251943e-01 -2.25151896e+00 2.01098789e-02 -6.71616852e-01 -6.36249959e-01 7.37918317e-01 -1.82631522e-01 -8.39865506e-02 -9.07803103e-02 -9.09260958e-02 -5.51432729e-01 2.36684933e-01 1.03996766e+00 -4.49542731e-01 -1.31010875e-01 -2.87038386e-01 -1.10404480e+00 5.17738163e-01 7.12566674e-01 -4.52123821e-01 -7.06396520e-01 -1.66641325e-01 5.58605611e-01 -2.42182370e-02 2.80522734e-01 -8.55261564e-01 3.13819110e-01 1.13186672e-01 2.90779531e-01 -1.14180692e-01 2.62893617e-01 -8.44280899e-01 5.14435887e-01 2.82572359e-01 -2.74681449e-01 2.11764544e-01 1.22065775e-01 9.69861746e-01 -1.66030064e-01 1.23275071e-01 2.80355155e-01 -1.69666231e-01 -8.05542946e-01 4.72688943e-01 2.48046190e-01 4.05318886e-01 1.08076143e+00 -1.56536087e-01 -5.67673981e-01 -2.08304241e-01 -9.21009183e-01 3.01949590e-01 1.39184952e-01 7.02071667e-01 6.65559530e-01 -1.30469155e+00 -6.26900971e-01 8.19026157e-02 6.72513068e-01 5.25345877e-02 1.57579929e-01 7.31614172e-01 -3.96460950e-01 7.48809934e-01 -1.65886357e-01 -4.17522997e-01 -1.20916677e+00 6.89444423e-01 1.54980317e-01 -7.25455582e-01 -9.62773085e-01 6.18457079e-01 1.92224517e-01 -3.72156233e-01 2.42535412e-01 -4.83747303e-01 -5.97742140e-01 3.25208157e-01 3.04742187e-01 2.97318041e-01 4.55594420e-01 -6.69656456e-01 -4.97613192e-01 1.64025635e-01 -2.62383282e-01 7.38661706e-01 1.62202322e+00 3.07493180e-01 -1.66110113e-01 2.58688062e-01 9.54365373e-01 -2.21215457e-01 -3.54008853e-01 -3.30549717e-01 3.78868401e-01 -4.94466722e-02 -1.04651675e-01 -9.59213972e-01 -1.13845992e+00 5.97300172e-01 4.00024563e-01 3.83710206e-01 8.41267467e-01 2.33918130e-01 6.05722785e-01 5.02580464e-01 3.96495700e-01 -8.01191688e-01 -1.07735321e-01 1.48093566e-01 8.97338390e-01 -1.09217966e+00 2.37903848e-01 -5.82269430e-01 -6.90595627e-01 1.03506219e+00 3.66561264e-01 1.42866731e-01 6.92227781e-01 -1.14380352e-01 -1.23676002e-01 -7.18221366e-01 -8.30644250e-01 -4.00364727e-01 7.35738397e-01 8.57772112e-01 4.37792569e-01 2.52698392e-01 -1.69975847e-01 6.67767823e-01 -3.87108922e-02 1.14398286e-01 2.87110895e-01 7.06301212e-01 -1.85162481e-02 -1.20745099e+00 9.72544122e-03 1.16105402e+00 -5.72805822e-01 -2.07690030e-01 -3.79433841e-01 8.53000104e-01 2.83393979e-01 7.03599632e-01 -3.18502903e-01 -3.18705022e-01 4.83491510e-01 3.09304357e-01 2.15159982e-01 -9.79506195e-01 -5.78379452e-01 -5.01643717e-01 5.56164801e-01 -4.95783389e-01 -2.64511883e-01 -3.23346913e-01 -1.30695152e+00 1.55239664e-02 -3.79680730e-02 1.07891485e-01 2.96864033e-01 6.86293542e-01 1.09981608e+00 8.51388633e-01 1.33554998e-03 -2.43010655e-01 -5.98965883e-02 -6.67673230e-01 -6.08814061e-01 7.60196745e-01 9.80281234e-02 -9.40194130e-01 9.72769856e-02 -2.14254797e-01]
[8.451953887939453, 7.760760307312012]
d546b951-576d-4784-9f79-738995485296
semipfl-personalized-semi-supervised
2203.08176
null
https://arxiv.org/abs/2203.08176v2
https://arxiv.org/pdf/2203.08176v2.pdf
SemiPFL: Personalized Semi-Supervised Federated Learning Framework for Edge Intelligence
Recent advances in wearable devices and Internet-of-Things (IoT) have led to massive growth in sensor data generated in edge devices. Labeling such massive data for classification tasks has proven to be challenging. In addition, data generated by different users bear various personal attributes and edge heterogeneity, rendering it impractical to develop a global model that adapts well to all users. Concerns over data privacy and communication costs also prohibit centralized data accumulation and training. We propose SemiPFL that supports edge users having no label or limited labeled datasets and a sizable amount of unlabeled data that is insufficient to train a well-performing model. In this work, edge users collaborate to train a Hyper-network in the server, generating personalized autoencoders for each user. After receiving updates from edge users, the server produces a set of base models for each user, which the users locally aggregate them using their own labeled dataset. We comprehensively evaluate our proposed framework on various public datasets from a wide range of application scenarios, from wearable health to IoT, and demonstrate that SemiPFL outperforms state-of-art federated learning frameworks under the same assumptions regarding user performance, network footprint, and computational consumption. We also show that the solution performs well for users without label or having limited labeled datasets and increasing performance for increased labeled data and number of users, signifying the effectiveness of SemiPFL for handling data heterogeneity and limited annotation. We also demonstrate the stability of SemiPFL for handling user hardware resource heterogeneity in three real-time scenarios.
['Peyman Servati', 'Z. Jane Wang', 'Wenwen Zhang', 'Arvin Tashakori']
2022-03-15
null
null
null
null
['semi-supervised-time-series-classification']
['time-series']
[-2.81320084e-02 -3.01929209e-02 -3.71632248e-01 -6.51263297e-01 -3.94752949e-01 -4.40274417e-01 -2.28546396e-01 3.23013067e-01 -3.54170233e-01 7.92719126e-01 1.63867608e-01 5.37631214e-02 -1.52822003e-01 -9.15945411e-01 -6.79614842e-01 -6.29902601e-01 -1.86169177e-01 5.11758029e-01 -5.04816957e-02 5.03212631e-01 -8.32358122e-01 2.33297631e-01 -1.55084968e+00 3.13216269e-01 6.69484854e-01 1.54648209e+00 -2.07760483e-02 4.22570914e-01 7.80867264e-02 4.81133848e-01 -6.25732601e-01 -3.14813197e-01 4.78036523e-01 1.55656725e-01 -4.79810536e-01 9.39781740e-02 2.85915911e-01 -2.92498261e-01 -1.57193035e-01 8.66009057e-01 1.05204809e+00 -7.52463117e-02 1.35409692e-02 -1.52226865e+00 -5.04791081e-01 8.89191091e-01 -8.05064887e-02 -1.78502023e-01 6.19458221e-02 -2.05998912e-01 6.96497679e-01 -3.91055971e-01 2.77586997e-01 6.10166192e-01 1.08696926e+00 7.84106910e-01 -9.33674157e-01 -7.65250504e-01 2.15128332e-01 -8.81669670e-02 -1.36407232e+00 -3.68681967e-01 5.98555028e-01 -8.60044081e-03 6.82357192e-01 4.22655702e-01 6.12107396e-01 1.22452080e+00 3.14331683e-03 6.25769556e-01 7.90646195e-01 -1.48713931e-01 7.13886321e-01 3.76459301e-01 3.50926369e-01 3.69340926e-01 5.90659738e-01 -2.89928317e-01 -7.04556465e-01 -5.42794704e-01 3.39879453e-01 4.37013507e-01 -1.18668310e-01 -2.02968150e-01 -1.05516350e+00 2.52405286e-01 2.93744385e-01 1.44790307e-01 -8.57811034e-01 7.08114579e-02 5.62954187e-01 2.22023815e-01 4.09681648e-01 -2.98299432e-01 -1.03525376e+00 3.84816639e-02 -7.26486981e-01 -3.64204079e-01 1.05887198e+00 1.39406109e+00 6.96378350e-01 -5.60731590e-02 -6.33825734e-02 5.62362850e-01 1.17668390e-01 4.20839190e-01 6.75105393e-01 -8.50242376e-01 2.56984293e-01 7.56336331e-01 1.79009527e-01 -6.83147371e-01 -7.49948859e-01 -6.58115149e-01 -1.20456696e+00 -4.59524602e-01 1.41308950e-02 -7.00977504e-01 -5.51854372e-01 1.93966651e+00 6.81530237e-01 3.78978193e-01 4.46050130e-02 5.64535677e-01 8.15980494e-01 3.42036217e-01 4.21573669e-01 -3.23825866e-01 1.35903931e+00 -7.26346254e-01 -6.56389117e-01 -1.52928919e-01 5.91635108e-01 -8.81022215e-02 1.04037929e+00 4.80110705e-01 -8.95706952e-01 -5.03192723e-01 -7.82509267e-01 2.30620846e-01 -3.80683810e-01 1.52312577e-01 8.52111936e-01 1.19730747e+00 -1.09385908e+00 4.91208404e-01 -9.23192143e-01 -6.06567383e-01 7.06912696e-01 8.53609085e-01 -2.47999057e-01 1.27655074e-01 -9.46536601e-01 1.31361678e-01 5.44711053e-01 -1.98429033e-01 -5.42978287e-01 -8.45749378e-01 -3.71917069e-01 2.23134637e-01 1.88575640e-01 -1.07156098e+00 1.03429306e+00 -8.16039443e-01 -1.17367327e+00 3.96174163e-01 1.82918638e-01 -4.40127701e-01 4.56689835e-01 -7.46707022e-02 -9.22714829e-01 -2.63099998e-01 -1.20721579e-01 4.47941691e-01 6.27676249e-01 -1.02250051e+00 -7.61300266e-01 -6.63847744e-01 -3.52569968e-02 -2.89516873e-03 -1.15613639e+00 -3.14039856e-01 -4.60265219e-01 -2.17088416e-01 -1.99474141e-01 -9.46586668e-01 -1.76007524e-01 3.13424021e-02 -3.50095659e-01 -1.20394915e-01 1.04828572e+00 -3.56007159e-01 1.22175837e+00 -2.12595654e+00 -5.61400473e-01 3.64195436e-01 3.00072908e-01 8.68475810e-02 5.78315631e-02 2.23603714e-02 2.35175341e-01 6.70872629e-02 1.76524147e-01 -5.57552278e-01 3.08853444e-02 5.78584731e-01 3.22508276e-01 2.23029122e-01 -6.25387192e-01 7.59449005e-01 -8.27694833e-01 -4.68980134e-01 5.23719527e-02 7.59566486e-01 -5.36633551e-01 2.60197550e-01 -2.35280856e-01 6.08838260e-01 -6.37605071e-01 8.06800127e-01 4.69560713e-01 -8.59823763e-01 4.11313593e-01 -6.66841328e-01 3.34966511e-01 -1.57866105e-01 -1.47043014e+00 1.59494805e+00 -7.86493838e-01 -9.72178057e-02 2.50020653e-01 -8.12181175e-01 6.50781691e-01 6.52677655e-01 1.08897650e+00 -4.36249375e-01 4.74182814e-01 8.01121742e-02 -5.48891425e-01 -5.64203024e-01 3.77994627e-02 2.19283313e-01 -2.63099670e-01 7.01711595e-01 2.74104983e-01 9.02998984e-01 -5.15202545e-02 1.95338484e-02 1.31257856e+00 -2.78841257e-01 1.86952278e-01 -2.81296879e-01 1.96163371e-01 -6.17385328e-01 8.35637808e-01 9.25347030e-01 -1.77791119e-01 1.52697131e-01 -3.95183027e-01 -7.95830905e-01 -7.65691876e-01 -9.93057549e-01 -1.29519194e-01 1.42354274e+00 1.17681988e-01 -5.36319315e-01 -7.77881980e-01 -8.88730288e-01 1.82059072e-02 4.16482329e-01 -4.48290318e-01 -4.36530337e-02 -6.66019246e-02 -1.16990924e+00 4.14062440e-01 6.45624697e-01 8.33827972e-01 -8.62325013e-01 -1.07322335e+00 2.61685729e-01 -2.84614146e-01 -1.27631068e+00 -4.45848167e-01 1.48014814e-01 -8.39079022e-01 -9.75768924e-01 -2.05596805e-01 -5.39120615e-01 7.84341037e-01 6.10493235e-02 1.16372216e+00 -1.36171609e-01 -3.00099313e-01 8.23556900e-01 -2.67654359e-01 -6.84155047e-01 -8.05637613e-02 3.51292372e-01 4.57041919e-01 4.42220151e-01 3.50407660e-01 -7.74574459e-01 -8.81467760e-01 4.36971962e-01 -9.29913104e-01 -2.18338504e-01 3.16384614e-01 4.62277293e-01 6.16250873e-01 3.53909701e-01 9.57401335e-01 -1.15650463e+00 5.09314775e-01 -7.93666184e-01 -1.95302010e-01 5.28741300e-01 -9.63287354e-01 -1.67582095e-01 8.53942454e-01 -6.90015435e-01 -8.93639266e-01 3.47879827e-01 -7.02174306e-02 -1.89867362e-01 -2.71211773e-01 2.28746757e-01 -7.26410210e-01 -1.19628578e-01 8.24486613e-01 -1.57583505e-01 -4.16048586e-01 -5.63379347e-01 3.06657344e-01 1.17008388e+00 5.51865935e-01 -6.12429857e-01 2.54515618e-01 5.68858385e-01 -2.45038152e-01 -7.38212347e-01 -8.49477232e-01 -3.49618614e-01 -2.51490027e-01 -1.28431290e-01 6.40838087e-01 -1.10336292e+00 -1.12607908e+00 3.90886009e-01 -6.38163447e-01 -2.96510994e-01 -6.34238958e-01 3.62339586e-01 -2.18519375e-01 1.06860630e-01 -4.74597603e-01 -7.98789799e-01 -1.08485603e+00 -7.58633554e-01 9.14735854e-01 4.73714083e-01 -1.14601061e-01 -1.03265774e+00 -3.47282022e-01 2.44436651e-01 7.76535153e-01 3.03895384e-01 5.13545215e-01 -8.98398519e-01 -3.38628411e-01 -5.95754385e-01 4.46195602e-02 1.99347228e-01 5.64921260e-01 -6.43709004e-01 -1.16605532e+00 -6.30007923e-01 1.61862168e-02 -2.92148113e-01 1.24729604e-01 1.56637207e-01 1.69322610e+00 -6.04606926e-01 -5.55687785e-01 7.25656509e-01 1.34730494e+00 -7.84019157e-02 1.03647754e-01 -1.11998804e-01 7.82709777e-01 2.02802539e-01 7.04860836e-02 8.84152472e-01 5.73719740e-01 2.07540631e-01 4.30803686e-01 -1.77993000e-01 1.68577999e-01 5.29310433e-03 7.89828226e-02 6.51368797e-01 4.28009033e-02 -5.21605730e-01 -4.81478035e-01 5.04892051e-01 -2.02926517e+00 -5.20751476e-01 1.73888430e-01 2.51165223e+00 5.75572014e-01 -1.43842921e-01 4.11614865e-01 1.30831301e-01 7.22685695e-01 -2.73267567e-01 -1.04859197e+00 5.52448928e-02 6.14505038e-02 1.65607721e-01 8.87314320e-01 -1.58043832e-01 -1.03683257e+00 3.33753645e-01 5.96187639e+00 2.35370263e-01 -1.22283697e+00 7.05026746e-01 7.92324662e-01 -2.06881985e-01 7.22836377e-03 -4.44555730e-01 -6.61107004e-01 8.05359721e-01 1.37875748e+00 -1.94107085e-01 3.57249975e-01 1.17627394e+00 1.63049966e-01 3.78922433e-01 -1.07071221e+00 1.33508873e+00 -1.36633828e-01 -1.15000713e+00 -2.67753422e-01 -2.76226681e-02 8.62167835e-01 5.22877634e-01 -2.86923409e-01 6.37643179e-03 5.74557126e-01 -4.75170970e-01 3.46863478e-01 2.97271252e-01 8.66736531e-01 -4.90856707e-01 6.69953763e-01 5.74438632e-01 -1.26986277e+00 -5.35681963e-01 -2.73972392e-01 1.72113508e-01 1.42027110e-01 9.30245519e-01 -6.17929280e-01 5.62267423e-01 1.08474636e+00 3.06776017e-01 -4.05115426e-01 1.02736127e+00 2.14291736e-01 5.79556346e-01 -7.59518445e-01 1.52027652e-01 -4.78405684e-01 2.39322737e-01 -9.09472257e-02 1.01848686e+00 5.96818745e-01 9.67170745e-02 5.60841858e-01 1.61224544e-01 -5.39385855e-01 2.81637043e-01 -2.56762981e-01 2.63862997e-01 9.31334674e-01 1.50731063e+00 -4.82894808e-01 -4.44434673e-01 -6.41157269e-01 1.03537190e+00 1.34100154e-01 2.71205664e-01 -7.36017168e-01 1.67130515e-01 4.79652852e-01 2.73310304e-01 -1.60116911e-01 1.07520483e-01 -2.19862849e-01 -1.22445703e+00 1.32090002e-01 -5.13338387e-01 1.00351620e+00 -3.78835022e-01 -1.59051776e+00 7.40495086e-01 -2.23972648e-01 -9.95570660e-01 2.14685500e-03 -2.34609738e-01 -4.58330244e-01 3.71970981e-01 -9.67425764e-01 -1.38249791e+00 -9.43752885e-01 1.02416062e+00 1.45996213e-01 6.91038147e-02 1.17925680e+00 8.99697721e-01 -7.28135049e-01 9.60431039e-01 -2.62336731e-02 5.69304675e-02 5.67773879e-01 -9.11550879e-01 1.95604041e-01 5.04354835e-01 1.96703468e-02 4.77710307e-01 3.46047759e-01 -7.58229494e-01 -1.65513730e+00 -1.72603488e+00 6.43604934e-01 -1.47449717e-01 3.16012472e-01 -4.53716904e-01 -5.68449557e-01 9.90033627e-01 -6.50231838e-02 7.59177446e-01 1.21957958e+00 1.30681217e-01 -1.26336709e-01 -7.27439582e-01 -1.66758132e+00 2.08897561e-01 1.26987326e+00 -2.80626535e-01 3.57203007e-01 6.25317454e-01 5.46654701e-01 -1.42281055e-01 -1.31015623e+00 3.86139154e-01 5.78056514e-01 -6.70923054e-01 8.09473097e-01 -4.28115934e-01 -7.27969944e-01 -1.98408410e-01 -2.56175369e-01 -1.08386922e+00 -2.45537460e-01 -7.91125238e-01 -4.81293112e-01 1.38788915e+00 2.86492378e-01 -8.88492167e-01 1.19327545e+00 1.43432105e+00 4.55527641e-02 -4.29172605e-01 -7.79770434e-01 -7.12302089e-01 -6.49648428e-01 -5.23345053e-01 1.02052689e+00 9.92549717e-01 2.97517404e-02 1.91244692e-01 -6.25762284e-01 3.20907384e-01 8.75699699e-01 -6.71112835e-02 6.84635103e-01 -1.66440821e+00 -5.09381354e-01 4.18072611e-01 -2.92782992e-01 -7.58335412e-01 -1.37643933e-01 -8.54883134e-01 -3.69709015e-01 -1.48933136e+00 -4.68297079e-02 -1.04283357e+00 -6.71166956e-01 1.00593245e+00 1.34753734e-01 4.75943476e-01 -5.34095690e-02 1.75070822e-01 -9.80529010e-01 1.16864748e-01 4.93751138e-01 -1.31089211e-01 -4.83034521e-01 3.73082697e-01 -8.22765887e-01 5.82979798e-01 1.06485093e+00 -3.77728939e-01 -7.50123501e-01 -6.51493073e-01 3.30393672e-01 -1.63976163e-01 9.61024165e-02 -1.44217849e+00 4.99296695e-01 1.63503081e-01 6.82076275e-01 -1.10989727e-01 1.08877882e-01 -1.57212198e+00 8.11879694e-01 2.63803750e-01 2.45821709e-03 -2.15579495e-02 -8.44396800e-02 5.09748578e-01 5.14007390e-01 2.21673280e-01 4.45024729e-01 -1.49761796e-01 -4.25415725e-01 8.91387343e-01 2.55699396e-01 -2.51264274e-01 1.17785311e+00 -1.35746285e-01 -2.20182106e-01 -3.84694308e-01 -1.13912237e+00 3.26429278e-01 5.70274174e-01 2.66340554e-01 1.43325076e-01 -1.33007896e+00 -3.34643930e-01 4.10540283e-01 1.77348569e-01 8.01930949e-02 5.11008441e-01 5.08969843e-01 7.09373727e-02 -2.37214610e-01 -1.33072168e-01 -5.31308711e-01 -1.15738368e+00 8.40756297e-01 1.95051759e-01 3.65208052e-02 -6.55794263e-01 4.41246182e-01 -2.52442122e-01 -4.41146314e-01 6.13398910e-01 -1.76275834e-01 2.43470117e-01 -5.30150309e-02 6.70984030e-01 5.37283838e-01 5.25623620e-01 -3.93199027e-01 -3.61067504e-01 2.01114193e-01 2.23350003e-01 5.77404380e-01 1.36305285e+00 -3.65871698e-01 2.12136492e-01 3.05520415e-01 1.09532106e+00 -2.11843178e-01 -1.08736503e+00 -3.30968380e-01 -2.54799873e-01 -9.37178135e-02 -1.47096096e-02 -9.18825626e-01 -1.52822661e+00 2.58441091e-01 1.13345575e+00 3.18968117e-01 1.54936218e+00 3.73838395e-02 1.06678367e+00 3.51928204e-01 9.88461375e-01 -1.15659845e+00 -3.20756793e-01 -1.94268242e-01 -4.73745316e-02 -1.11305678e+00 -1.75737590e-01 -3.80160093e-01 -4.03820336e-01 6.16145194e-01 4.87050027e-01 3.04272085e-01 1.02540541e+00 6.28153741e-01 1.99395776e-01 3.14779170e-02 -5.79344988e-01 1.06383130e-01 -1.76712394e-01 8.39345872e-01 1.77416086e-01 3.86099100e-01 -1.33769497e-01 1.10924554e+00 -1.27415121e-01 4.96235281e-01 1.92444503e-01 9.98981178e-01 -1.82727143e-01 -1.34987104e+00 -2.15719089e-01 8.79268050e-01 -6.48691595e-01 2.26769745e-01 1.26190647e-01 1.69356525e-01 5.26490152e-01 1.12941730e+00 1.03213839e-01 -4.91625279e-01 2.26682499e-01 2.58277088e-01 1.64193019e-01 -2.87811190e-01 -8.88660908e-01 -5.81777804e-02 1.50866881e-02 -6.42318308e-01 -3.68841588e-01 -3.61228645e-01 -1.32471156e+00 -1.42237350e-01 -1.18209824e-01 1.46308050e-01 9.18128133e-01 6.63423836e-01 9.86923516e-01 4.97384876e-01 6.21023059e-01 -5.90291023e-01 -5.00819206e-01 -7.14114130e-01 -6.03253484e-01 6.06248438e-01 3.02237633e-04 -2.32366368e-01 4.90422845e-02 1.89127907e-01]
[6.021755695343018, 6.179160118103027]
11857e00-8a53-4ede-83dd-e122b5dbf2bf
efficient-reward-poisoning-attacks-on-online
2205.14842
null
https://arxiv.org/abs/2205.14842v2
https://arxiv.org/pdf/2205.14842v2.pdf
Efficient Reward Poisoning Attacks on Online Deep Reinforcement Learning
We study reward poisoning attacks on online deep reinforcement learning (DRL), where the attacker is oblivious to the learning algorithm used by the agent and the dynamics of the environment. We demonstrate the intrinsic vulnerability of state-of-the-art DRL algorithms by designing a general, black-box reward poisoning framework called adversarial MDP attacks. We instantiate our framework to construct two new attacks which only corrupt the rewards for a small fraction of the total training timesteps and make the agent learn a low-performing policy. We provide a theoretical analysis of the efficiency of our attack and perform an extensive empirical evaluation. Our results show that our attacks efficiently poison agents learning in several popular classical control and MuJoCo environments with a variety of state-of-the-art DRL algorithms, such as DQN, PPO, SAC, etc.
['Gagandeep Singh', 'Qi Zeng', 'Yinglun Xu']
2022-05-30
null
null
null
null
['data-poisoning']
['adversarial']
[-7.15554118e-01 -6.89166784e-02 -2.20541179e-01 2.31264293e-01 -6.56395614e-01 -1.09176981e+00 7.11640894e-01 4.19164784e-02 -1.01273727e+00 1.06549358e+00 -2.78732508e-01 -5.25582075e-01 4.38115522e-02 -9.34159636e-01 -1.05525517e+00 -9.14011955e-01 -8.18658471e-01 5.15167832e-01 3.34421009e-01 -3.81994158e-01 7.86314756e-02 6.89968109e-01 -8.91812205e-01 -2.24090397e-01 5.59214294e-01 7.27151155e-01 -5.16182840e-01 1.04823470e+00 4.45539713e-01 1.49730515e+00 -9.98030663e-01 -3.32092941e-01 6.22023106e-01 -4.08673018e-01 -7.38027513e-01 -3.81838500e-01 -2.64930725e-02 -1.20018601e+00 -9.58395123e-01 1.38250661e+00 6.34718716e-01 -8.22181925e-02 1.82689890e-01 -1.64557338e+00 -3.40177417e-01 1.20532906e+00 -4.98883218e-01 2.33860761e-01 -1.07945755e-01 7.82924533e-01 7.09761620e-01 1.11620635e-01 4.92131084e-01 1.45030582e+00 2.83245176e-01 1.06297505e+00 -1.21344256e+00 -9.11639035e-01 1.36143535e-01 -5.11972718e-02 -5.72312593e-01 -4.71082814e-02 3.97016734e-01 -3.63950059e-02 8.31290126e-01 -1.62681118e-01 4.49478328e-01 1.43339360e+00 4.08394426e-01 1.00423551e+00 1.47569966e+00 -7.98657164e-02 7.66954005e-01 -3.14728737e-01 -9.56405103e-02 7.18655944e-01 4.14095372e-01 9.72294211e-01 -2.54001021e-01 -8.73208284e-01 1.00475144e+00 -8.92889202e-02 1.48662403e-01 -6.52828097e-01 -1.01568782e+00 1.12431943e+00 3.05009931e-01 -2.37656862e-01 -4.23891515e-01 1.15513432e+00 8.40999603e-01 9.06164229e-01 -1.19191200e-01 4.96544719e-01 -4.86762971e-01 -3.59983325e-01 -1.27045020e-01 8.31594765e-01 1.03266811e+00 5.82310557e-01 4.67952192e-01 5.85247040e-01 -1.22785144e-01 -4.32262495e-02 5.61552085e-02 7.67745435e-01 2.94337988e-01 -1.66363597e+00 3.93635005e-01 -3.01587820e-01 7.12411463e-01 -2.22096771e-01 -3.38167578e-01 -2.36942142e-01 -1.35203645e-01 9.85748291e-01 7.06163824e-01 -9.43035781e-01 -3.87127370e-01 1.98679531e+00 4.14973229e-01 3.27923410e-02 4.68273759e-01 6.78441226e-01 -1.58980638e-01 5.08955419e-01 1.98918670e-01 -1.55672148e-01 9.04734254e-01 -8.95745695e-01 -3.64427328e-01 1.42147541e-01 9.13992047e-01 -6.47977218e-02 8.13214123e-01 4.59237933e-01 -1.26274025e+00 1.01004682e-01 -9.72312629e-01 5.13853610e-01 -1.00022770e-01 -4.96179193e-01 6.49855137e-01 7.42846310e-01 -6.99119329e-01 1.02735174e+00 -1.26746595e+00 3.36623132e-01 8.10545683e-01 3.86765063e-01 8.50689337e-02 2.60918766e-01 -1.20373201e+00 7.86781371e-01 2.64580220e-01 -6.39548779e-01 -2.32356906e+00 -6.44628584e-01 -4.35685247e-01 -8.99656117e-03 8.40539753e-01 -3.96434635e-01 1.87349296e+00 -8.68361473e-01 -1.79296494e+00 4.97140408e-01 9.05890226e-01 -1.19839764e+00 9.77825582e-01 -4.42016691e-01 1.37545746e-02 5.65823019e-01 -6.81567192e-02 2.12886006e-01 9.67081368e-01 -1.07636118e+00 -6.83085084e-01 -1.64718479e-01 6.80637538e-01 1.33091435e-01 -8.48999247e-02 1.24687955e-01 8.54667902e-01 -5.30996442e-01 -1.12806106e+00 -9.00421500e-01 -4.83038694e-01 -4.68575507e-02 -6.70514926e-02 -3.37864369e-01 9.24610317e-01 -1.18492387e-01 4.86992657e-01 -2.20679402e+00 -1.43639311e-01 4.44786251e-02 3.92918169e-01 3.53821814e-01 -4.00060177e-01 7.41344750e-01 1.81963846e-01 2.82162670e-02 7.16207102e-02 -2.63234111e-03 6.53034747e-01 5.84558189e-01 -9.25911367e-01 1.02899313e+00 -2.61399031e-01 9.19945419e-01 -1.34714699e+00 -4.88368236e-02 -9.26792920e-02 -5.72147407e-02 -6.26955986e-01 5.29666126e-01 -7.74059534e-01 3.86642277e-01 -8.40772033e-01 2.03854501e-01 4.01187092e-01 2.06052661e-01 1.13809630e-01 8.22592258e-01 -6.05994798e-02 5.06795406e-01 -9.23435032e-01 1.28841209e+00 -3.80700082e-02 1.00108869e-01 2.71144360e-01 -7.32793868e-01 2.98593611e-01 4.42010731e-01 5.54488897e-01 -6.87163711e-01 2.57897854e-01 3.06786984e-01 2.56764114e-01 -1.52846441e-01 1.96812198e-01 -1.11472875e-01 -2.31866494e-01 1.09038568e+00 4.15165126e-02 4.40094359e-02 7.48931915e-02 3.85161102e-01 1.58751845e+00 2.57351041e-01 1.86837658e-01 -2.35219643e-01 9.37296003e-02 1.12787932e-01 4.28581119e-01 1.62522423e+00 -7.97688663e-01 -6.85057521e-01 1.19609380e+00 -4.04277295e-01 -1.32723641e+00 -1.11362743e+00 4.36106294e-01 1.26314449e+00 -1.07585676e-01 -1.71601504e-01 -9.92474198e-01 -1.38887203e+00 4.39803898e-01 6.52999103e-01 -6.21533990e-01 -3.64160210e-01 -7.08031714e-01 -4.31621403e-01 1.17249787e+00 3.67523015e-01 8.72705162e-01 -1.37204576e+00 -1.07372093e+00 3.36473018e-01 6.02838814e-01 -7.87376702e-01 -4.42008406e-01 3.32689255e-01 -7.31410623e-01 -1.26143324e+00 -2.16291934e-01 -3.03442329e-01 1.58416659e-01 7.31566846e-02 8.88474286e-01 -3.17363180e-02 6.72263652e-02 6.96954846e-01 -2.00209573e-01 -3.56110007e-01 -9.92857218e-01 -6.54185936e-02 3.67691994e-01 -4.78640735e-01 3.26393582e-02 -5.81930339e-01 -5.77979445e-01 -8.29767361e-02 -9.12818849e-01 -7.50798881e-01 1.30821794e-01 7.97948182e-01 1.82371885e-01 2.39372447e-01 7.59093523e-01 -9.63095188e-01 9.21036005e-01 -4.06071782e-01 -1.55423832e+00 -2.94693802e-02 -3.29892218e-01 3.71274948e-01 1.30172515e+00 -7.95137584e-01 -4.68093753e-01 -1.43722877e-01 1.08756199e-02 -6.72734141e-01 4.53321263e-02 -1.24398880e-01 6.26535341e-02 -4.99998689e-01 9.20859694e-01 2.78542697e-01 1.03784800e-01 -2.28344813e-01 5.54630220e-01 4.87605818e-02 4.56959128e-01 -1.27677870e+00 1.11737108e+00 4.73610669e-01 2.87671149e-01 -2.42897213e-01 -5.07712424e-01 4.42828983e-01 1.48602113e-01 -4.38536443e-02 4.83495384e-01 -7.32566714e-01 -1.75113523e+00 8.16861033e-01 -9.28736925e-01 -1.10805118e+00 -8.35479200e-01 2.73239106e-01 -1.01761305e+00 3.47456634e-01 -1.11091113e+00 -9.68286633e-01 -2.41814181e-01 -1.21207976e+00 3.25291961e-01 2.02661484e-01 6.21585429e-01 -7.86879897e-01 6.27106667e-01 -3.36896807e-01 4.93646622e-01 2.78526992e-01 9.42029536e-01 -9.94583309e-01 -6.41838372e-01 2.45746419e-01 2.46519029e-01 4.83033180e-01 -4.16508645e-01 -2.46656850e-01 -7.64911771e-01 -6.81959987e-01 2.52670407e-01 -1.03342807e+00 5.82429051e-01 9.14136097e-02 1.22043800e+00 -9.79926765e-01 1.93026870e-01 3.92054349e-01 1.46407437e+00 4.14343655e-01 4.80698109e-01 6.38463855e-01 4.70044196e-01 7.87009969e-02 6.03577733e-01 7.81726956e-01 -1.79943088e-02 1.35524869e-01 9.42655742e-01 4.42416370e-01 4.90954489e-01 -3.69592577e-01 9.14567351e-01 -8.64395648e-02 1.02028057e-01 1.65437847e-01 -3.67756188e-01 2.48687789e-01 -1.94994998e+00 -1.19503736e+00 3.73024762e-01 2.22557187e+00 1.28354394e+00 1.76450506e-01 9.36236024e-01 -1.61015689e-01 2.10806563e-01 2.06529081e-01 -1.07170117e+00 -7.44384944e-01 2.39238590e-01 6.11103058e-01 1.27448618e+00 5.49560905e-01 -1.12228489e+00 1.24453068e+00 6.84818792e+00 8.75428617e-01 -8.93019557e-01 3.59725118e-01 3.49005222e-01 -3.69732320e-01 5.78219071e-02 -1.01749285e-03 -5.97775161e-01 3.01431656e-01 1.18884039e+00 -4.37400937e-01 1.31684756e+00 1.39685357e+00 1.66500807e-01 1.61733776e-01 -1.13862395e+00 5.27557611e-01 -7.19776690e-01 -1.31287074e+00 -4.63692009e-01 3.48144799e-01 7.17243195e-01 4.44353104e-01 1.02028772e-01 7.50117183e-01 1.65065658e+00 -9.63398337e-01 7.28552997e-01 -8.86091962e-02 3.22143197e-01 -1.35970819e+00 2.57755160e-01 3.84022743e-01 -5.13246715e-01 -4.73913997e-01 -4.10780072e-01 -2.79532790e-01 -3.60926688e-01 -1.09198943e-01 -3.15910608e-01 5.91432080e-02 5.27279735e-01 2.24378407e-01 -2.59704173e-01 6.69739962e-01 -7.13133395e-01 8.34977746e-01 -4.04762745e-01 -1.36595160e-01 8.57471466e-01 -1.64359976e-02 5.85268915e-01 6.63478971e-01 -4.23789591e-01 4.32404913e-02 5.36234915e-01 9.13707554e-01 -5.09816647e-01 -5.35195708e-01 -7.19393730e-01 -2.95968682e-01 5.33105016e-01 1.03698993e+00 -3.91108915e-02 -3.67259949e-01 -1.27050221e-01 7.46066749e-01 6.02495372e-01 3.46144468e-01 -1.21626675e+00 -2.86552042e-01 1.09001601e+00 -3.69089097e-01 4.11360115e-01 -4.92666095e-01 2.19262868e-01 -1.04155660e+00 -1.90456495e-01 -1.43560708e+00 4.18833882e-01 -1.43131733e-01 -1.50715125e+00 -2.28229836e-02 -1.85249150e-01 -6.41415417e-01 -4.23135281e-01 -5.45001745e-01 -5.58510482e-01 4.63566273e-01 -1.76952183e+00 -3.76874387e-01 5.31096935e-01 9.68795180e-01 -6.44105300e-02 -4.09534752e-01 7.66328454e-01 -1.22907452e-01 -4.91031855e-01 5.70658743e-01 5.76514482e-01 4.67566192e-01 3.02106619e-01 -1.53996515e+00 5.20899832e-01 9.08367395e-01 -1.25144318e-01 2.57006913e-01 7.76753604e-01 -5.53993642e-01 -1.74510396e+00 -9.79037106e-01 -1.60314947e-01 -8.23069215e-02 1.31351566e+00 -2.88109213e-01 -5.51295102e-01 8.38225961e-01 2.59984195e-01 3.21771950e-01 2.69686915e-02 -4.94305760e-01 -5.30850589e-01 -1.60088912e-01 -1.42703557e+00 1.01750088e+00 7.17931688e-01 -6.16293132e-01 -4.14598644e-01 4.38464969e-01 9.53248024e-01 -4.65618253e-01 -6.58131778e-01 -3.18605930e-01 3.57232451e-01 -7.15132713e-01 7.91044474e-01 -1.32592261e+00 3.34841192e-01 7.17019429e-03 2.35394202e-02 -1.47747266e+00 8.41240063e-02 -1.27464998e+00 -6.13132834e-01 5.16881347e-01 -2.14667395e-01 -1.00748444e+00 7.33188629e-01 3.75767261e-01 1.51124045e-01 -3.30812335e-01 -1.23839176e+00 -1.17473423e+00 9.54824746e-01 1.63605884e-02 4.64163721e-01 8.30058217e-01 1.87718272e-01 -1.37202427e-01 -6.03202403e-01 3.18431199e-01 1.17974448e+00 -1.97263286e-01 8.56041431e-01 -5.02959549e-01 -9.02598679e-01 -3.65642995e-01 9.06885564e-02 -6.80598676e-01 4.51447010e-01 -5.76613963e-01 -1.88533872e-01 -7.01469481e-01 1.07226454e-01 -4.56790000e-01 -3.50000620e-01 8.30345988e-01 2.89182574e-01 -3.24720204e-01 5.09681821e-01 2.02675194e-01 -8.43049288e-01 6.99090600e-01 1.33920336e+00 -2.44741552e-02 -1.21792279e-01 9.64989588e-02 -3.70911598e-01 5.19895554e-01 1.12252629e+00 -1.15456748e+00 -3.81164014e-01 -2.16239661e-01 1.46109208e-01 3.89901221e-01 6.00353599e-01 -8.12256753e-01 1.64679345e-02 -6.39770985e-01 -1.51989564e-01 8.00629240e-03 -2.26036742e-01 -7.15622008e-01 -5.06993592e-01 1.39970708e+00 -7.85072744e-01 2.26017565e-01 -1.45824682e-02 5.55303335e-01 5.81574023e-01 -2.34012544e-01 1.26726031e+00 -3.19478750e-01 5.56411745e-04 4.72783387e-01 -7.09632576e-01 5.91295183e-01 1.20969510e+00 8.26771438e-01 -1.02406979e+00 -4.59009588e-01 -4.45739955e-01 5.10614097e-01 5.20339131e-01 -1.90238371e-01 2.58682430e-01 -1.05346274e+00 -6.63505971e-01 -1.35989189e-01 -4.60761487e-01 -3.98325622e-01 -1.77989051e-01 4.77454752e-01 -7.34049797e-01 1.24588206e-01 -6.21628165e-01 2.09674418e-01 -7.34042823e-01 1.13459361e+00 9.21607316e-01 -7.05323517e-01 -8.97883475e-01 1.41974553e-01 1.38230184e-02 -1.85341582e-01 6.03754044e-01 -1.48334742e-01 2.87780106e-01 -5.50666094e-01 6.05889559e-01 5.42541265e-01 -4.65325266e-01 2.82390833e-01 -1.51864752e-01 -3.84284943e-01 -3.74703616e-01 -5.80420077e-01 1.23580670e+00 4.86590713e-01 6.01078793e-02 4.29953672e-02 7.91725457e-01 -6.56818375e-02 -1.88423431e+00 -2.51133084e-01 -3.36419463e-01 -2.43211970e-01 -4.05335836e-02 -7.67586648e-01 -1.04765987e+00 6.79158747e-01 5.07423818e-01 4.75776017e-01 6.84821606e-01 -4.20918763e-01 8.01699817e-01 9.19868231e-01 8.97929311e-01 -1.23025036e+00 3.27890366e-01 6.96195960e-01 3.89019936e-01 -7.24901795e-01 -1.58781394e-01 5.56280732e-01 -7.05402195e-01 9.32991505e-01 4.49184567e-01 -1.12708473e+00 3.38774413e-01 5.56283891e-01 -1.77749172e-01 6.92996904e-02 -9.08918083e-01 -1.00223005e-01 -1.04188228e+00 6.61281586e-01 -5.50478876e-01 1.16950028e-01 -2.57873029e-01 3.86401355e-01 1.24345616e-01 -5.36435358e-02 9.76743877e-01 1.28138196e+00 -6.60844624e-01 -1.35772324e+00 -3.57448548e-01 -1.31065950e-01 -1.00377643e+00 1.52322367e-01 -3.50790024e-01 9.53024983e-01 -4.69271690e-01 7.22307324e-01 -3.56383801e-01 5.43386787e-02 -2.95123830e-02 -1.51695356e-01 6.26072407e-01 -1.05808169e-01 -1.19308746e+00 -8.79017338e-02 2.50349753e-02 -1.00928175e+00 -1.31171346e-01 -4.87998843e-01 -1.50027227e+00 -8.86299849e-01 3.48682016e-01 1.20160438e-01 4.40539628e-01 9.17798042e-01 8.39198828e-02 2.73759753e-01 1.25885868e+00 -4.47934479e-01 -1.89973176e+00 -4.84725386e-01 -9.64579046e-01 8.51587132e-02 7.30281472e-01 -4.63965446e-01 -5.34076393e-01 -6.55860484e-01]
[3.9621288776397705, 2.3787074089050293]
49688cd6-4168-4572-be67-3740bfc1f781
learning-graphs-for-knowledge-transfer-with
null
null
http://openaccess.thecvf.com//content/CVPR2021/html/Ghosh_Learning_Graphs_for_Knowledge_Transfer_With_Limited_Labels_CVPR_2021_paper.html
http://openaccess.thecvf.com//content/CVPR2021/papers/Ghosh_Learning_Graphs_for_Knowledge_Transfer_With_Limited_Labels_CVPR_2021_paper.pdf
Learning Graphs for Knowledge Transfer With Limited Labels
Fixed input graphs are a mainstay in approaches that utilize Graph Convolution Networks (GCNs) for knowledge transfer. The standard paradigm is to utilize relationships in the input graph to transfer information using GCNs from training to testing nodes in the graph; for example, the semi-supervised, zero-shot, and few-shot learning setups. We propose a generalized framework for learning and improving the input graph as part of the standard GCN-based learning setup. Moreover, we use additional constraints between similar and dissimilar neighbors for each node in the graph by applying triplet loss on the intermediate layer output. We present results of semi-supervised learning on Citeseer, Cora, and Pubmed benchmarking datasets, and zero/few-shot action recognition on UCF101 and HMDB51 datasets, significantly outperforming current approaches. We also present qualitative results visualizing the graph connections that our approach learns to update.
['Abhinav Shrivastava', 'Larry S. Davis', 'Nirat Saini', 'Pallabi Ghosh']
2021-06-19
null
null
null
cvpr-2021-1
['few-shot-action-recognition']
['computer-vision']
[ 3.57212514e-01 5.33489645e-01 -4.87472713e-01 -4.62978333e-01 -2.64109850e-01 -5.78536570e-01 6.47971511e-01 6.01855993e-01 -4.29329365e-01 9.14774299e-01 3.20211411e-01 -2.96292245e-01 -4.19358373e-01 -1.17224061e+00 -1.03235006e+00 -5.24546325e-01 -1.93341345e-01 6.48824930e-01 2.16384560e-01 -1.29742652e-01 5.38419709e-02 4.94396359e-01 -1.14050555e+00 4.44530934e-01 5.99307060e-01 6.78384006e-01 -1.45462051e-01 7.17827976e-01 -3.09886158e-01 7.99146652e-01 -3.38024735e-01 -6.59478605e-01 -2.35324930e-02 -4.07611370e-01 -1.13557661e+00 -1.31486014e-01 4.78388220e-01 1.14289686e-01 -6.69992387e-01 9.68422651e-01 4.61995572e-01 4.40405875e-01 7.48933077e-01 -1.30855358e+00 -1.04001784e+00 8.35218430e-01 -1.90472677e-01 4.11146432e-01 2.24319503e-01 1.78539589e-01 1.30649126e+00 -4.30538684e-01 1.14709759e+00 1.16028166e+00 5.50762653e-01 6.25010192e-01 -1.38061440e+00 -4.65019614e-01 1.85777634e-01 3.89260560e-01 -1.08347034e+00 -2.10791454e-01 5.58246195e-01 -4.95365947e-01 1.37308192e+00 -3.17565538e-02 5.83161771e-01 1.21668184e+00 8.17704350e-02 5.65219760e-01 6.06552184e-01 -4.37895030e-01 2.40675047e-01 -7.60840327e-02 5.78177035e-01 1.13958001e+00 2.83580154e-01 -3.53078283e-02 -7.18263626e-01 -1.71417564e-01 5.30403197e-01 2.46339843e-01 -3.04332435e-01 -7.52199769e-01 -8.79616320e-01 8.77218604e-01 1.00131047e+00 2.68915802e-01 -1.55265778e-01 3.52469385e-01 4.91140306e-01 5.24275839e-01 5.12227535e-01 5.34964204e-01 -3.00110161e-01 1.54792160e-01 -5.14551878e-01 -2.63021618e-01 9.32562172e-01 9.28119540e-01 1.10578036e+00 -3.40610772e-01 -6.24302447e-01 5.05146801e-01 8.18418562e-02 -1.64937109e-01 3.29907089e-01 -5.77720284e-01 3.67401928e-01 8.99169266e-01 -3.86911362e-01 -6.38295770e-01 -4.93359476e-01 -4.43200231e-01 -7.31681108e-01 7.69515634e-02 2.25187376e-01 -1.04091205e-02 -1.47453332e+00 1.75110996e+00 6.51849657e-02 5.27393043e-01 -2.62291450e-02 4.66802388e-01 1.27881992e+00 2.01898336e-01 2.43027061e-01 -6.46091849e-02 6.37875617e-01 -1.11214411e+00 -5.53078175e-01 1.63879655e-02 1.01207507e+00 4.12761234e-02 1.06337380e+00 4.55024466e-03 -8.12609255e-01 -3.21810782e-01 -9.14500356e-01 -2.47077018e-01 -1.05384350e+00 -3.72653157e-01 8.43185663e-01 2.78379709e-01 -1.28695285e+00 1.21219802e+00 -8.41338634e-01 -6.86709881e-01 8.95289183e-01 4.76651371e-01 -5.35064876e-01 -4.86484796e-01 -1.32223463e+00 9.59158957e-01 6.16936505e-01 -2.63968974e-01 -1.14126670e+00 -1.04489863e+00 -1.11227322e+00 4.17124361e-01 5.19795060e-01 -8.50721657e-01 7.93453693e-01 -8.71859372e-01 -1.07011592e+00 1.07159257e+00 2.13447824e-01 -7.69510925e-01 1.68036595e-01 2.41355807e-01 -2.36438587e-01 1.44726545e-01 -2.32284591e-01 6.86628044e-01 5.12476325e-01 -8.44706893e-01 1.88075099e-02 -3.88483942e-01 2.75931299e-01 2.04728127e-01 -2.80597597e-01 -2.35771239e-01 -3.91816616e-01 -2.93753743e-01 -3.59616935e-01 -6.54365599e-01 -2.29677737e-01 6.13187253e-02 -6.82752132e-01 -3.30518037e-01 6.85200632e-01 -5.51345646e-01 1.02621162e+00 -1.91762352e+00 4.32024091e-01 3.67963642e-01 5.87079227e-01 3.89351606e-01 -4.57546800e-01 6.39997125e-01 -4.32352692e-01 1.14002042e-01 -3.20168912e-01 -1.56381547e-01 -1.03539348e-01 4.66860920e-01 7.23420978e-02 3.59572142e-01 2.21769676e-01 1.43345666e+00 -1.13935542e+00 -1.79461330e-01 1.32291541e-01 5.33560336e-01 -5.59764028e-01 -8.10556337e-02 -5.21566272e-01 2.24699929e-01 -2.29784653e-01 4.87397462e-01 8.43378603e-02 -8.57773006e-01 1.80730626e-01 -2.99032122e-01 3.38213265e-01 2.42344335e-01 -7.31753230e-01 1.95968807e+00 -1.58357784e-01 4.78034556e-01 -3.22929472e-01 -1.31498718e+00 6.18610322e-01 1.26129329e-01 2.40500197e-01 -4.96164948e-01 3.37772481e-02 -2.68809259e-01 7.84637034e-02 -2.17755079e-01 -2.60373857e-02 -1.22510344e-02 1.74179718e-01 4.53964472e-01 9.16290283e-01 3.30120176e-01 5.42697072e-01 7.02104151e-01 1.67556715e+00 1.79370314e-01 3.66162062e-01 5.02748117e-02 1.57615885e-01 -1.09673806e-01 1.64552227e-01 9.17182624e-01 9.85922851e-03 4.26235557e-01 9.32220817e-01 -3.99744332e-01 -6.72746658e-01 -1.08991516e+00 2.54174203e-01 1.24899685e+00 -3.25830340e-01 -5.38961947e-01 -4.71544236e-01 -9.93384719e-01 2.36176044e-01 8.00818920e-01 -1.11542630e+00 -6.92067921e-01 -9.93008241e-02 -4.61325258e-01 4.29455370e-01 5.40388942e-01 1.25919223e-01 -1.26018465e+00 -3.56546678e-02 7.81536326e-02 3.87667716e-01 -7.97089040e-01 -4.29484129e-01 5.75555325e-01 -8.39545071e-01 -1.41196096e+00 -5.58277071e-01 -7.56347239e-01 8.45342278e-01 -3.37134413e-02 1.21931398e+00 2.13642657e-01 -7.60367334e-01 6.15827560e-01 -3.28817904e-01 -1.21727742e-01 -2.05464706e-01 1.56670317e-01 -2.95593590e-01 -2.34179720e-01 3.22821349e-01 -7.01316535e-01 -3.08094054e-01 -2.05366284e-01 -7.20163405e-01 1.68412812e-02 5.08717835e-01 9.36762571e-01 6.45225644e-01 -3.46669167e-01 4.90402550e-01 -1.52021015e+00 6.51798725e-01 -6.11518383e-01 -4.48894680e-01 7.22401083e-01 -8.57353806e-01 4.07980829e-01 5.77252746e-01 -2.35113442e-01 -7.76794910e-01 -2.01737555e-03 1.44250214e-01 -7.64160812e-01 -5.02239093e-02 9.17917609e-01 -5.58098741e-02 -3.46410930e-01 8.47493172e-01 -5.31881873e-04 1.38990968e-01 -3.98854077e-01 7.90556252e-01 2.10283950e-01 4.60366428e-01 -2.04108760e-01 6.62395775e-01 3.84231329e-01 1.72498956e-01 -5.41078925e-01 -1.08362889e+00 -5.55301428e-01 -9.64053988e-01 -8.83087069e-02 1.08556068e+00 -6.20801032e-01 -7.35531688e-01 1.49847746e-01 -8.96740913e-01 -7.26988435e-01 -5.65936387e-01 3.08956087e-01 -4.47542459e-01 1.76815301e-01 -6.37973547e-01 -2.54990131e-01 -3.34719121e-01 -6.95404053e-01 8.35015357e-01 2.00227752e-01 3.14921290e-02 -1.46724975e+00 3.42428714e-01 1.40057534e-01 3.72454911e-01 4.07395780e-01 1.40063858e+00 -1.14310420e+00 -4.97030795e-01 -6.88493028e-02 -3.81076992e-01 1.80510283e-01 2.58970171e-01 -2.92167775e-02 -7.80801892e-01 -3.84581178e-01 -9.95090842e-01 -7.24368751e-01 1.34858263e+00 3.24928761e-01 1.21031892e+00 -1.04319401e-01 -6.31142676e-01 8.36598694e-01 1.37018895e+00 -1.68435737e-01 6.14507198e-01 -8.78191069e-02 9.87176955e-01 3.33050489e-01 1.64200619e-01 1.06883109e-01 2.10476071e-01 2.49599606e-01 3.28973055e-01 -1.85539126e-01 -4.62561190e-01 -4.12316829e-01 1.36632249e-01 4.40014064e-01 3.33043411e-02 -4.62346613e-01 -9.17855263e-01 4.10420477e-01 -1.90810621e+00 -1.01462972e+00 2.41844505e-01 2.24803615e+00 8.75041664e-01 1.74911499e-01 -5.72579168e-02 -4.39177543e-01 7.59781301e-01 2.09755465e-01 -8.99612188e-01 -2.88115442e-01 -1.13280397e-02 6.59182549e-01 5.95149875e-01 4.50560361e-01 -9.56468761e-01 1.10782480e+00 6.44985390e+00 4.53631967e-01 -8.41251314e-01 -7.55715743e-03 5.76012015e-01 -2.11167797e-01 -3.11135620e-01 3.84552330e-02 -5.59405148e-01 7.13394433e-02 1.19185829e+00 -3.96394998e-01 7.51019418e-01 5.41670918e-01 -2.86743820e-01 1.90154612e-01 -1.46120918e+00 7.90490568e-01 2.67039865e-01 -1.90132582e+00 2.54301935e-01 -4.73766439e-02 7.39578903e-01 4.76171017e-01 -2.45104805e-01 5.68976223e-01 8.85705352e-01 -1.29414523e+00 -2.40622640e-01 7.83788562e-01 8.91984701e-01 -5.58827162e-01 4.57852811e-01 -1.53316148e-02 -8.66916656e-01 2.45883524e-01 -4.11364079e-01 7.71528557e-02 -6.42350689e-02 2.96332210e-01 -9.99413252e-01 6.78156435e-01 5.82137227e-01 1.12569785e+00 -8.03585768e-01 1.11858821e+00 -4.87290144e-01 6.03951454e-01 2.86473501e-02 -9.77370050e-03 2.83887714e-01 -1.73416704e-01 3.27146590e-01 1.05920374e+00 -8.52915272e-02 -5.09409271e-02 2.83580631e-01 8.87299061e-01 -7.83480108e-01 -3.51106748e-02 -8.90481889e-01 -6.00923777e-01 1.80523455e-01 1.23419368e+00 -6.48281395e-01 -4.69759673e-01 -5.06969690e-01 1.07084250e+00 1.01941144e+00 5.82186639e-01 -4.51305687e-01 -6.59308374e-01 4.45177764e-01 -7.92647749e-02 4.29881692e-01 1.57917798e-01 2.54210770e-01 -1.00324190e+00 -4.21780020e-01 -5.37769198e-01 1.10410917e+00 -8.06648314e-01 -1.46329558e+00 1.67318270e-01 -6.56682551e-02 -6.17336035e-01 -5.99211343e-02 -8.52428019e-01 -9.61121082e-01 8.20516706e-01 -1.49820602e+00 -1.22157359e+00 -3.38224858e-01 7.62584507e-01 2.84759160e-02 -7.94188529e-02 9.55765009e-01 6.89909905e-02 -5.49706399e-01 5.27664542e-01 5.18022850e-02 3.63260478e-01 6.80607677e-01 -1.36598849e+00 5.56432605e-01 4.95649576e-01 4.25073415e-01 7.08708942e-01 3.63917530e-01 -9.30611253e-01 -1.31683505e+00 -1.38445973e+00 6.95667386e-01 -2.11393312e-01 8.91844690e-01 -5.58250546e-01 -1.16527164e+00 1.17795289e+00 2.56515682e-01 5.03479421e-01 8.80355239e-01 5.05810916e-01 -6.47450686e-01 2.43073255e-01 -1.12706757e+00 4.51151282e-01 1.46598160e+00 -7.10714638e-01 -6.12348497e-01 6.94763958e-01 1.04408145e+00 -1.77971929e-01 -9.43712592e-01 2.17563733e-01 2.70167202e-01 -5.14393985e-01 8.60009670e-01 -1.56348252e+00 2.75693774e-01 3.75301801e-02 1.99885339e-01 -1.80967033e+00 -5.04091799e-01 -7.17469096e-01 -2.56395519e-01 8.95767987e-01 6.68897390e-01 -5.70741057e-01 9.64380383e-01 3.55394036e-01 -3.13535482e-01 -6.43880188e-01 -6.74689114e-01 -6.23767376e-01 4.46817614e-02 1.30207673e-01 2.85112232e-01 1.29579270e+00 1.67868748e-01 7.11296856e-01 -1.45648018e-01 -4.46834527e-02 6.10351980e-01 2.10298728e-02 5.38171530e-01 -1.39763331e+00 -6.06121182e-01 -2.76324481e-01 -7.38455594e-01 -2.51045823e-01 4.30966079e-01 -1.83054411e+00 -2.35144734e-01 -2.02827835e+00 4.21039402e-01 -2.12661121e-02 -7.83988595e-01 9.83116090e-01 -1.62071079e-01 2.69880760e-02 -2.33013153e-01 -2.85075217e-01 -9.43803191e-01 4.80291814e-01 1.06351817e+00 -3.11718017e-01 -9.89341736e-02 -2.55996615e-01 -6.11401737e-01 3.57499242e-01 6.18542492e-01 -4.17561054e-01 -6.55417085e-01 -2.88043261e-01 2.41942778e-01 -1.76521778e-01 3.92635912e-01 -7.75593162e-01 4.77801621e-01 7.23382309e-02 3.97102118e-01 -3.72137815e-01 2.77145207e-02 -3.14419776e-01 7.95826390e-02 6.18572295e-01 -7.82694817e-01 -2.24581689e-01 1.18132308e-01 9.01617110e-01 7.13105649e-02 -1.20526277e-01 7.73695886e-01 -3.53734195e-01 -8.17445874e-01 7.43249416e-01 3.08131099e-01 3.56540293e-01 9.07189548e-01 2.49211602e-02 -7.64965296e-01 -3.14486057e-01 -1.25792634e+00 4.53728378e-01 3.49273980e-01 3.69117022e-01 6.97961092e-01 -1.19638574e+00 -3.98769140e-01 4.93063964e-02 5.06510973e-01 -2.84588158e-01 1.06702127e-01 8.46066833e-01 -2.03875557e-01 2.30446026e-01 -4.05210942e-01 -1.63867712e-01 -1.25284219e+00 9.35255229e-01 3.70534927e-01 -4.30779785e-01 -6.50105357e-01 1.03927898e+00 2.50611063e-02 -6.94556832e-01 5.62824190e-01 -9.03073475e-02 -1.46789595e-01 4.81191650e-02 3.28245252e-01 2.51895219e-01 1.56430528e-01 -1.68476462e-01 -2.38310874e-01 2.21456178e-02 -3.62826645e-01 2.69811213e-01 1.56364155e+00 3.66896987e-01 -2.29122162e-01 6.15957141e-01 1.44127965e+00 -5.99204540e-01 -9.25692022e-01 -5.61240256e-01 1.77873358e-01 -1.64952260e-02 -4.23562014e-03 -9.65459287e-01 -9.54249978e-01 8.42364132e-01 3.79296988e-01 6.41735643e-03 5.62685668e-01 3.98138940e-01 2.41177425e-01 9.35189486e-01 2.33681440e-01 -1.01808345e+00 1.49208322e-01 3.67412299e-01 6.55895054e-01 -1.25791609e+00 2.37637553e-02 -3.38669360e-01 -4.81552720e-01 1.05701149e+00 6.02924109e-01 -1.17983893e-01 6.55786574e-01 4.03565494e-03 -4.10640001e-01 -7.71135569e-01 -1.03175008e+00 -5.30085385e-01 5.01800358e-01 7.03679979e-01 5.25863826e-01 -1.15354404e-01 -9.63931810e-03 2.53171057e-01 3.83849055e-01 9.29353610e-02 3.97390634e-01 9.90235031e-01 -3.72498572e-01 -1.07140350e+00 5.09925544e-01 1.14164770e+00 -1.54354289e-01 -4.30186957e-01 -8.35885525e-01 8.02691936e-01 -3.35185647e-01 5.76101065e-01 6.62462860e-02 -3.69010121e-01 2.45118976e-01 5.09938598e-01 6.33942068e-01 -1.16992390e+00 -6.59323096e-01 -4.23458874e-01 2.36578688e-01 -8.11358690e-01 -2.34095052e-01 -4.00292099e-01 -1.51010168e+00 9.81242768e-03 -2.39631608e-01 5.06180823e-02 2.67807603e-01 8.89749467e-01 5.75448632e-01 8.20867002e-01 1.41239852e-01 -4.31474030e-01 -4.05896127e-01 -9.08619225e-01 -6.36527181e-01 6.60035908e-01 1.28032267e-01 -6.62334859e-01 -3.99545252e-01 -1.40227601e-01]
[6.947887897491455, 6.2983551025390625]
379fe1ca-a6b3-4e83-b14c-4e4097030b3e
counting-from-sky-a-large-scale-dataset-for
2008.1247
null
https://arxiv.org/abs/2008.12470v1
https://arxiv.org/pdf/2008.12470v1.pdf
Counting from Sky: A Large-scale Dataset for Remote Sensing Object Counting and A Benchmark Method
Object counting, whose aim is to estimate the number of objects from a given image, is an important and challenging computation task. Significant efforts have been devoted to addressing this problem and achieved great progress, yet counting the number of ground objects from remote sensing images is barely studied. In this paper, we are interested in counting dense objects from remote sensing images. Compared with object counting in a natural scene, this task is challenging in the following factors: large scale variation, complex cluttered background, and orientation arbitrariness. More importantly, the scarcity of data severely limits the development of research in this field. To address these issues, we first construct a large-scale object counting dataset with remote sensing images, which contains four important geographic objects: buildings, crowded ships in harbors, large-vehicles and small-vehicles in parking lots. We then benchmark the dataset by designing a novel neural network that can generate a density map of an input image. The proposed network consists of three parts namely attention module, scale pyramid module and deformable convolution module to attack the aforementioned challenging factors. Extensive experiments are performed on the proposed dataset and one crowd counting datset, which demonstrate the challenges of the proposed dataset and the superiority and effectiveness of our method compared with state-of-the-art methods.
['Qingjie Liu', 'Yunhong Wang', 'Guangshuai Gao']
2020-08-28
null
null
null
null
['object-counting']
['computer-vision']
[ 7.44631663e-02 -5.78807831e-01 3.80525559e-01 -2.39495143e-01 -3.07325035e-01 -2.37171784e-01 5.35870135e-01 4.26447429e-02 -8.72795641e-01 8.59783828e-01 8.05996135e-02 -1.66279316e-01 8.65007862e-02 -1.15251768e+00 -4.40630078e-01 -6.29911125e-01 5.87472878e-02 3.53349537e-01 4.75225538e-01 1.05062433e-01 3.70918185e-01 4.77043241e-01 -1.56924629e+00 -3.16426218e-01 1.03881788e+00 8.41153324e-01 5.57568669e-01 5.63710153e-01 -3.22947264e-01 1.02487803e+00 -7.23399818e-01 -2.74725407e-01 1.13142848e-01 -1.97207913e-01 -3.60187024e-01 1.60081953e-01 4.11740631e-01 -7.15418041e-01 -3.75028580e-01 1.49556124e+00 6.60802066e-01 2.38067150e-01 7.29138911e-01 -9.19213235e-01 -9.95788574e-01 2.74164945e-01 -1.09378755e+00 1.00810564e+00 -1.53119192e-01 1.23715974e-01 4.70948845e-01 -9.05657351e-01 -2.09730327e-01 1.32824755e+00 5.43444395e-01 3.38902175e-01 -4.37768906e-01 -8.95935535e-01 2.03228205e-01 1.69120952e-01 -1.84245956e+00 -9.28031206e-02 4.30361748e-01 -4.96906698e-01 6.24521971e-01 5.93815073e-02 5.52569747e-01 4.15374219e-01 -2.02661708e-01 6.57453716e-01 1.04679263e+00 -1.97726935e-01 3.67632687e-01 1.09284535e-01 1.28272429e-01 6.54718280e-01 8.41465652e-01 -3.54768544e-01 1.80662647e-01 5.37568145e-02 9.14824963e-01 4.46305960e-01 -1.12948686e-01 3.61545384e-01 -1.11917818e+00 9.03499007e-01 6.99888587e-01 3.91587585e-01 -3.26020628e-01 1.54507890e-01 9.86556038e-02 -4.89034802e-01 6.44737482e-01 -2.36297935e-01 -2.41046827e-02 2.03287795e-01 -9.28145349e-01 3.34146410e-01 6.02973580e-01 9.77481902e-01 6.51301324e-01 7.92237893e-02 -2.95896322e-01 6.71546102e-01 1.05653055e-01 9.28114593e-01 2.86504537e-01 -4.74589735e-01 1.01870060e+00 6.63727522e-01 4.45842206e-01 -1.54996228e+00 -4.79667604e-01 -2.06453890e-01 -1.33129573e+00 1.47787491e-02 2.84964204e-01 -2.31440127e-01 -7.44889677e-01 1.27609921e+00 4.69649047e-01 2.66579360e-01 -2.11260110e-01 1.12063718e+00 9.52915490e-01 8.34517002e-01 4.40524399e-01 -6.12790398e-02 1.52455962e+00 -7.42918372e-01 -6.30999267e-01 -4.38391447e-01 3.77260298e-02 -3.96330416e-01 9.28662062e-01 -1.41069576e-01 -8.29770148e-01 -7.07132101e-01 -9.38020229e-01 -6.66844100e-02 -5.03636956e-01 5.93438864e-01 7.17164636e-01 6.05497062e-01 -5.09345710e-01 1.74853876e-01 -7.83046186e-01 -3.62086415e-01 8.76728296e-01 1.86367154e-01 2.24109576e-03 -1.20234475e-01 -9.81766820e-01 7.86218524e-01 5.44328749e-01 5.25128663e-01 -5.54454803e-01 -3.32731873e-01 -8.16667020e-01 2.01878995e-01 2.34800160e-01 -4.25188273e-01 8.81771386e-01 -5.16739011e-01 -6.66984558e-01 7.66368210e-01 6.17940770e-03 -9.34411064e-02 7.17695296e-01 -1.03416562e-01 -4.24746692e-01 -1.11747503e-01 4.93824422e-01 4.69676912e-01 5.72453618e-01 -1.00967920e+00 -1.04197931e+00 -6.70182884e-01 7.89305791e-02 1.30524412e-01 -3.74049217e-01 2.18772337e-01 -2.82512784e-01 -5.60954869e-01 -6.60500899e-02 -5.55604577e-01 -1.98416576e-01 -2.19498370e-02 -3.43679756e-01 -2.54338473e-01 8.18730772e-01 -6.86523139e-01 1.24798512e+00 -1.91285014e+00 -4.30133522e-01 -2.04821214e-01 3.18138659e-01 4.53913748e-01 5.17368205e-02 -2.95946486e-02 4.29096967e-01 9.47419554e-02 -4.28579122e-01 -1.52508304e-01 -1.74139634e-01 1.76764697e-01 -2.26591930e-01 6.46910787e-01 5.45712888e-01 7.82359660e-01 -1.04905748e+00 -9.67968941e-01 4.77964699e-01 5.81588507e-01 -2.22273886e-01 1.49852395e-01 2.51841396e-02 6.67945385e-01 -7.78624296e-01 1.00641620e+00 1.30558562e+00 -3.36444736e-01 -2.62849808e-01 -1.04452539e-02 -3.19572657e-01 -4.10495520e-01 -1.39662802e+00 9.33161557e-01 -3.17808777e-01 2.83175409e-01 -1.28702551e-01 -7.90342391e-01 1.04552174e+00 -2.56313741e-01 3.82218845e-02 -7.24325120e-01 3.65112185e-01 1.89294606e-01 -2.89304703e-01 -7.35580683e-01 7.21259713e-01 -1.89315394e-01 -1.03003472e-01 1.24486104e-01 -4.18379664e-01 -8.31004381e-02 5.13135433e-01 -8.22331160e-02 9.89712536e-01 -3.53527963e-01 5.51342905e-01 -1.63250357e-01 6.90812886e-01 -2.70968322e-02 4.96343523e-01 7.08826840e-01 -4.65645730e-01 4.75672573e-01 1.46294996e-01 -7.38633454e-01 -9.95502770e-01 -8.65207851e-01 -1.96102828e-01 9.34903800e-01 4.29814249e-01 3.46283376e-01 -7.49053240e-01 -4.66016442e-01 5.01299500e-02 3.48379731e-01 -7.67819703e-01 4.42086846e-01 -8.06807280e-01 -1.10429823e+00 4.08945590e-01 8.11298430e-01 1.38426065e+00 -1.21256912e+00 -7.07252324e-01 1.92504004e-01 -4.92884070e-01 -1.40683687e+00 -5.82810521e-01 -3.20970327e-01 -6.92306280e-01 -1.20361662e+00 -9.71313715e-01 -8.44855428e-01 7.58294463e-01 8.81983340e-01 1.18640912e+00 3.51284921e-01 -5.20451307e-01 -1.42686531e-01 -1.32357702e-01 -9.57083702e-01 8.38411748e-02 3.72121334e-02 -7.93737471e-02 6.83982149e-02 8.88809443e-01 -4.50197488e-01 -8.58033538e-01 4.20988649e-01 -1.18880165e+00 -3.39436740e-01 7.52809882e-01 5.20607710e-01 5.09088159e-01 5.22357881e-01 5.14481902e-01 -6.21536791e-01 6.49288774e-01 -7.50815868e-01 -1.01936412e+00 1.20773256e-01 1.65175170e-01 -2.88308710e-01 5.80145836e-01 -4.19517785e-01 -9.32226896e-01 5.19467220e-02 9.00018662e-02 -1.49509251e-01 -2.51416147e-01 2.17251033e-01 -1.91329017e-01 4.72329259e-02 4.55408037e-01 3.58246624e-01 -3.77062947e-01 -2.30844110e-01 -4.83408980e-02 9.02739882e-01 5.34137309e-01 -4.01182860e-01 8.14470351e-01 8.23989272e-01 -9.75319222e-02 -9.64747250e-01 -1.04843175e+00 -7.12545991e-01 -5.79460025e-01 -1.08554207e-01 1.11269331e+00 -1.18227327e+00 -8.63260508e-01 5.93999445e-01 -1.36695325e+00 1.71295628e-01 -9.23650414e-02 4.94422913e-01 -5.03770486e-02 4.26873237e-01 -3.37869942e-01 -1.26006007e+00 -5.16994119e-01 -7.87728071e-01 1.21856201e+00 6.23912454e-01 5.03002346e-01 -7.10079610e-01 6.17531277e-02 3.38436544e-01 5.66400826e-01 3.54853779e-01 5.37284255e-01 -2.85853863e-01 -8.03571701e-01 -3.03984046e-01 -9.59455132e-01 4.95897621e-01 9.97416377e-02 -1.92522585e-01 -8.63414705e-01 -8.27206746e-02 -1.68583430e-02 -2.93363363e-01 9.95319784e-01 4.99951392e-01 1.30451405e+00 -2.80189991e-01 -2.79152036e-01 3.35570753e-01 1.60367453e+00 2.90563200e-02 7.97827780e-01 1.87571883e-01 8.43661368e-01 4.19123888e-01 5.39936543e-01 7.75023818e-01 6.82810605e-01 2.08075762e-01 6.54695928e-01 -1.21517472e-01 2.21435875e-01 -1.22434273e-01 -2.82420307e-01 7.58911490e-01 -5.95141590e-01 -1.40935168e-01 -9.53099251e-01 7.55487740e-01 -1.68269181e+00 -1.31357038e+00 -2.24629492e-01 1.83173811e+00 3.53157848e-01 -2.03899249e-01 1.84515521e-01 1.87797710e-01 1.17673767e+00 2.30193317e-01 -4.89605337e-01 3.20976079e-01 -1.70167118e-01 -1.75213646e-02 6.61089778e-01 5.85572049e-02 -1.47384715e+00 7.27421343e-01 5.18437052e+00 8.66982222e-01 -7.80846119e-01 1.15472309e-01 8.57782364e-01 2.18428746e-01 1.75691277e-01 -4.98659819e-01 -1.20411849e+00 9.01985705e-01 2.24667147e-01 2.52833664e-01 3.73578876e-01 8.91900837e-01 1.77927643e-01 -3.71874541e-01 -4.87527758e-01 1.20193684e+00 1.59034789e-01 -1.02949524e+00 9.68349874e-02 -6.73541725e-02 7.34340608e-01 8.52715150e-02 -4.54950705e-02 5.24698198e-01 3.14284503e-01 -1.15010357e+00 7.37733245e-01 4.63375479e-01 7.63736427e-01 -7.76709318e-01 1.07976317e+00 7.05212653e-01 -1.62012982e+00 -2.17572287e-01 -9.60612714e-01 -4.23929185e-01 7.87210092e-02 8.40892851e-01 -3.17056060e-01 6.50452003e-02 8.56979132e-01 3.79827917e-01 -5.32175064e-01 1.14330065e+00 -1.25413656e-01 2.54538417e-01 -4.04729724e-01 -5.17159581e-01 3.61922413e-01 -4.33450550e-01 2.26265099e-02 1.33560371e+00 4.01503265e-01 6.29266143e-01 3.15228313e-01 1.11277676e+00 -1.79834515e-01 3.99244875e-02 -6.73406720e-01 1.57868713e-01 5.32318950e-01 1.58009970e+00 -9.15057778e-01 -3.81085336e-01 -4.70075518e-01 5.64711332e-01 4.29822415e-01 1.14458986e-01 -1.22784257e+00 -4.66958523e-01 2.78052986e-01 2.20792741e-01 4.92055565e-01 -2.96450287e-01 -1.06489524e-01 -1.19878209e+00 3.00732642e-01 -4.64772224e-01 3.43597770e-01 -6.05619609e-01 -1.36147046e+00 3.78247112e-01 2.81764101e-02 -1.13510931e+00 4.65577751e-01 -4.62509751e-01 -7.93555021e-01 9.39216137e-01 -1.95629466e+00 -1.10590744e+00 -1.12804461e+00 6.12505555e-01 5.50944269e-01 2.02864502e-02 3.79684925e-01 6.72208786e-01 -4.81572837e-01 2.01743171e-01 6.52712733e-02 6.30588233e-01 2.26429943e-02 -9.13831830e-01 5.31668901e-01 7.91385651e-01 -2.84151822e-01 3.44414860e-01 1.98595226e-01 -6.41078413e-01 -1.11110771e+00 -1.59982061e+00 7.68872678e-01 -4.54435617e-01 5.26465774e-01 -2.59864330e-01 -7.88624108e-01 5.26880801e-01 -3.41916889e-01 6.29221380e-01 2.85603464e-01 -4.16044176e-01 -5.43520227e-03 1.37000354e-02 -1.42072678e+00 2.48783350e-01 9.78169203e-01 -9.31035280e-02 -3.77817303e-01 4.21010524e-01 3.91490877e-01 -3.11541647e-01 -3.48444402e-01 2.82042921e-01 2.51739323e-01 -9.31064427e-01 1.08590138e+00 -1.37992054e-01 6.06224716e-01 -4.47951496e-01 -2.71732867e-01 -9.47089314e-01 -4.37056690e-01 1.59310743e-01 1.26239762e-01 1.25173008e+00 -2.59303093e-01 -5.03139079e-01 8.51235390e-01 3.44422042e-01 1.19844079e-01 -3.47011626e-01 -8.92124951e-01 -6.64472282e-01 1.45241871e-01 1.12345926e-02 9.66088414e-01 8.21880996e-01 -5.54916859e-01 3.48280579e-01 -3.41386348e-01 4.95429993e-01 8.00044477e-01 -1.48255542e-01 8.75147104e-01 -1.22023666e+00 1.21703312e-01 -1.77763566e-01 -6.74008965e-01 -1.07278156e+00 -1.76749706e-01 -4.41812247e-01 1.18147299e-01 -1.67782366e+00 6.47808552e-01 -6.86402678e-01 1.28630057e-01 8.77009109e-02 -6.98880792e-01 4.05477256e-01 1.60544857e-01 3.17954779e-01 -8.46680343e-01 5.23135066e-01 1.33529365e+00 -6.00407898e-01 5.46100885e-02 -6.38880185e-04 -5.99042356e-01 9.91518438e-01 9.40805495e-01 -4.22271371e-01 -9.58921984e-02 -8.81056786e-01 6.35786802e-02 -6.33182973e-02 6.29125297e-01 -1.42167163e+00 2.62296379e-01 -2.21738666e-01 6.56190038e-01 -9.73102272e-01 1.82792231e-01 -8.61872017e-01 -2.36746013e-01 5.48813581e-01 3.27281356e-01 1.28844425e-01 1.40303358e-01 7.60369778e-01 -2.15341419e-01 -1.91128924e-01 8.74907613e-01 -5.81979334e-01 -7.92952061e-01 5.62662423e-01 1.26666538e-02 3.67112994e-01 1.07165277e+00 -3.03407550e-01 -5.42598188e-01 5.68283908e-02 -2.55717129e-01 2.17488155e-01 1.18323490e-02 1.80587843e-01 6.15467608e-01 -1.32907224e+00 -1.19824219e+00 1.14698865e-01 1.50802702e-01 4.88833219e-01 4.95750606e-01 6.17243111e-01 -9.30002689e-01 3.44193697e-01 -1.64599970e-01 -5.58360040e-01 -8.91538620e-01 6.08826339e-01 1.76695079e-01 -3.42439532e-01 -4.29007441e-01 7.52087295e-01 2.82259196e-01 -5.10961711e-01 2.31681168e-02 -6.11443818e-01 -5.18115819e-01 -5.98945059e-02 9.96846259e-01 7.53270984e-01 -5.23661934e-02 -7.87672758e-01 -4.51168418e-01 8.22791815e-01 8.93954635e-02 4.31871861e-01 1.24106538e+00 -1.80403903e-01 -8.20851550e-02 2.38293812e-01 9.22592402e-01 -1.01971254e-01 -1.09292769e+00 -3.43736500e-01 -3.77188057e-01 -7.71276593e-01 -1.16891861e-01 -2.76417762e-01 -1.15305793e+00 1.15794528e+00 6.51850045e-01 3.20527226e-01 8.55700791e-01 -1.04048491e-01 7.45174646e-01 4.60019171e-01 6.09734654e-01 -1.03012252e+00 1.01464666e-01 6.24782681e-01 5.81983566e-01 -1.81048489e+00 1.67618886e-01 -2.73347467e-01 -4.43485081e-01 7.59954929e-01 6.93039358e-01 -2.86392272e-01 6.04711115e-01 2.45465785e-01 -1.20951742e-01 -3.16944748e-01 7.80708119e-02 -5.33897877e-01 -7.25879595e-02 6.68082416e-01 1.56084850e-01 2.26225093e-01 -1.82126537e-01 4.41338658e-01 -2.66314708e-02 2.62074113e-01 4.23863411e-01 9.63913202e-01 -8.69096339e-01 -5.83841242e-02 -8.40872943e-01 6.83659136e-01 -6.15107954e-01 -2.81660646e-01 1.16368562e-01 9.16804790e-01 4.44108844e-01 1.03208041e+00 4.61354017e-01 -1.10192457e-02 3.12907130e-01 -7.13061750e-01 2.82297730e-01 -4.02464420e-01 -2.87591755e-01 -4.67769206e-01 -3.23641628e-01 3.94980349e-02 -6.74326241e-01 -4.33066398e-01 -1.13483107e+00 -8.12461078e-01 -6.44256234e-01 -5.18613867e-02 6.55159473e-01 8.00310791e-01 -1.60518706e-01 4.91569847e-01 5.14432728e-01 -1.11472821e+00 -6.96981907e-01 -1.16746008e+00 -9.66055572e-01 3.71189386e-01 9.23339948e-02 -7.91750133e-01 -4.00408626e-01 -3.04893758e-02]
[8.498454093933105, -0.26599931716918945]
b87e7bec-97e9-49f2-af19-b6fdb9be2aa5
separation-free-spectral-super-resolution-via
2211.15361
null
https://arxiv.org/abs/2211.15361v1
https://arxiv.org/pdf/2211.15361v1.pdf
Separation-Free Spectral Super-Resolution via Convex Optimization
Atomic norm methods have recently been proposed for spectral super-resolution with flexibility in dealing with missing data and miscellaneous noises. A notorious drawback of these convex optimization methods however is their lower resolution in the high signal-to-noise (SNR) regime as compared to conventional methods such as ESPRIT. In this paper, we devise a simple weighting scheme in existing atomic norm methods and show that the resolution of the resulting convex optimization method can be made arbitrarily high in the absence of noise, achieving the so-called separation-free super-resolution. This is proved by a novel, kernel-free construction of the dual certificate whose existence guarantees exact super-resolution using the proposed method. Numerical results corroborating our analysis are provided.
['Zongben Xu', 'Gongguo Tang', 'Yi-Lin Mo', 'Zai Yang']
2022-11-28
null
null
null
null
['spectral-super-resolution', 'miscellaneous']
['computer-vision', 'miscellaneous']
[ 5.54488957e-01 1.99433062e-02 1.90701693e-01 1.06296837e-01 -1.13926661e+00 -3.05632263e-01 1.30769596e-01 -3.38009179e-01 -3.20415825e-01 1.24213505e+00 2.66377538e-01 1.64482713e-01 -7.41936207e-01 -5.25550246e-01 -3.95629674e-01 -1.19365239e+00 8.27338025e-02 9.56433043e-02 -1.44602224e-01 -3.51621240e-01 -7.57592591e-03 4.31607187e-01 -1.46343529e+00 -6.55812696e-02 1.23375857e+00 8.82364690e-01 1.66519418e-01 5.61072826e-01 3.19277138e-01 3.63407135e-01 -3.63976955e-01 -2.28708565e-01 4.61423814e-01 -5.25477946e-01 -4.48982388e-01 3.15347284e-01 3.82173747e-01 2.55691446e-03 -1.87690303e-01 1.47824502e+00 7.30533302e-01 2.92961478e-01 2.79730588e-01 -7.66326845e-01 -5.10051012e-01 5.10061741e-01 -8.87881041e-01 4.65249717e-01 3.87726784e-01 -4.30276066e-01 8.90981853e-01 -1.24635923e+00 5.65232754e-01 9.75732505e-01 8.49994779e-01 2.77236134e-01 -1.63525069e+00 -4.26954120e-01 -3.33115190e-01 1.42720804e-01 -1.61554241e+00 -6.41236126e-01 8.76509607e-01 -1.35154530e-01 2.73387372e-01 5.18941522e-01 2.02674627e-01 6.87541783e-01 -1.82469964e-01 3.16083670e-01 1.64362741e+00 -5.41556716e-01 7.36592785e-02 1.21677473e-01 -1.72244802e-01 3.96271616e-01 5.55287659e-01 4.57957499e-02 -5.48938215e-01 -4.06124562e-01 9.78298485e-01 -3.16015899e-01 -8.73323143e-01 -3.68325025e-01 -1.20715737e+00 7.10339487e-01 2.50133663e-01 6.01415753e-01 -5.84815204e-01 -2.31177032e-01 2.24571273e-01 2.32782841e-01 8.00014615e-01 1.93238139e-01 5.65852178e-03 1.52390733e-01 -1.14239645e+00 3.22364420e-01 6.04891121e-01 8.86175156e-01 2.68646896e-01 6.48737073e-01 9.38061029e-02 8.57066095e-01 -1.65841132e-01 5.96172333e-01 5.60458153e-02 -1.17039251e+00 5.67267954e-01 -2.24442706e-01 6.01545811e-01 -9.40671444e-01 -3.81897897e-01 -1.01434290e+00 -1.29788494e+00 2.40769163e-01 5.54303467e-01 -2.67078936e-01 -2.56305993e-01 1.73733759e+00 4.34529752e-01 4.89118218e-01 3.91109079e-01 1.22472107e+00 5.04555166e-01 6.16632879e-01 -3.71130496e-01 -1.08008921e+00 1.11287403e+00 -2.27217928e-01 -1.08716929e+00 2.69380957e-01 1.25873592e-02 -7.67759919e-01 5.74538410e-01 6.49594367e-01 -1.42095792e+00 -4.14273649e-01 -9.06399786e-01 2.97328353e-01 3.47293258e-01 -5.46849146e-02 2.36332193e-01 7.29172647e-01 -8.58520925e-01 6.54581368e-01 -3.30727011e-01 -8.20361916e-03 2.44711250e-01 1.54333070e-01 -4.00701553e-01 1.83467567e-01 -1.14432108e+00 7.91921258e-01 3.45145226e-01 3.26223612e-01 -3.05061579e-01 -9.63007569e-01 -5.04318714e-01 2.41323799e-01 7.51515210e-01 -5.35930812e-01 8.35381389e-01 -9.60985899e-01 -1.27644777e+00 4.74493027e-01 -1.09929889e-01 -4.00428832e-01 6.67483449e-01 -3.13116729e-01 -5.70433378e-01 3.44361633e-01 6.93323463e-02 -3.69517893e-01 1.12888372e+00 -1.50405657e+00 -5.92751265e-01 -3.62837344e-01 -7.81791583e-02 1.83656722e-01 -5.52495271e-02 -9.34051126e-02 1.69832721e-01 -7.90091932e-01 4.20500755e-01 -6.58689618e-01 -5.29793680e-01 -2.23427668e-01 -2.56344587e-01 3.46449971e-01 6.14157856e-01 -8.18279147e-01 1.05185556e+00 -2.12635446e+00 5.08118451e-01 2.43721917e-01 2.48833358e-01 3.62314969e-01 6.55591488e-02 5.14781833e-01 -3.48124832e-01 -1.33669257e-01 -4.81256872e-01 -8.97856876e-02 -3.11312735e-01 -6.89579174e-02 -2.81211615e-01 9.91185427e-01 4.80798027e-03 1.87649548e-01 -9.44351912e-01 -4.67179596e-01 2.83520311e-01 7.93459058e-01 -3.72780561e-01 -1.24970511e-01 4.85778511e-01 7.00925887e-01 -3.28422308e-01 5.28538227e-01 9.49630737e-01 -6.36195913e-02 2.50079781e-01 -4.71332759e-01 -4.37545687e-01 -2.44792238e-01 -1.85668707e+00 1.33156145e+00 -4.89822179e-01 3.92629743e-01 9.57298160e-01 -1.44191110e+00 7.06412733e-01 6.63346112e-01 7.79822588e-01 -5.99678040e-01 -4.90084365e-02 4.12117451e-01 -1.96062490e-01 -2.99086124e-01 4.54307169e-01 -7.47998476e-01 7.55574629e-02 -1.02550760e-02 -1.39692053e-01 1.70063198e-01 1.68909393e-02 1.09919183e-01 6.41025841e-01 -7.16763139e-02 5.54645061e-01 -8.77803087e-01 8.42474759e-01 -2.99171686e-01 8.53606880e-01 7.61205375e-01 -1.13603830e-01 5.79930842e-01 2.03067243e-01 -3.75375780e-03 -1.48980296e+00 -9.48278546e-01 -6.39689207e-01 4.40818787e-01 1.59045413e-01 -6.16718344e-02 -7.80933857e-01 7.34131336e-02 -1.06719382e-01 5.21073997e-01 -3.21919531e-01 2.63564229e-01 -6.51902854e-01 -1.11441743e+00 3.20203066e-01 2.92592704e-01 6.69751823e-01 -3.03549826e-01 -4.66954678e-01 4.65370536e-01 -5.26647747e-01 -1.55567169e+00 -2.00225070e-01 1.18588597e-01 -9.34735775e-01 -9.27133441e-01 -8.88186693e-01 -4.14143562e-01 4.74389017e-01 5.75597227e-01 5.93076587e-01 -2.72889733e-01 -1.71975389e-01 5.37596047e-01 -1.72539562e-01 7.24950209e-02 -3.58113497e-01 -5.21344244e-01 4.66724187e-01 4.46004957e-01 -2.72030771e-01 -8.67867947e-01 -4.79062438e-01 2.17283100e-01 -7.66010463e-01 -1.27606347e-01 4.00314629e-01 1.11085224e+00 7.46340096e-01 5.41561723e-01 9.65735257e-01 -7.72411644e-01 7.28673160e-01 -4.74242657e-01 -8.50480735e-01 -1.63003162e-01 -6.40920639e-01 -3.98587249e-02 9.69935775e-01 -4.04480934e-01 -1.35119772e+00 1.16062768e-01 5.52072898e-02 -3.03902298e-01 1.79389521e-01 4.83740568e-01 -1.78742647e-01 -4.53991950e-01 6.16422951e-01 4.43919331e-01 -1.39127374e-01 -6.57696545e-01 3.55428010e-01 3.90600711e-01 1.00956440e+00 -5.48700750e-01 1.29436839e+00 7.51672685e-01 5.52428007e-01 -1.34913611e+00 -9.34937537e-01 -6.11817122e-01 -6.38673604e-01 -1.74027100e-01 5.16523242e-01 -8.71795535e-01 -7.57417202e-01 1.24092326e-01 -8.79335284e-01 3.64912063e-01 -3.85918409e-01 6.33022130e-01 -7.89161861e-01 7.75520802e-01 -4.08497274e-01 -1.36763692e+00 -2.81969011e-01 -7.23686576e-01 5.65190971e-01 2.55012572e-01 1.62637189e-01 -1.00717163e+00 -1.37239411e-01 5.09814262e-01 6.08117640e-01 7.33500302e-01 3.26397598e-01 -1.12310708e-01 -3.82089049e-01 -6.62274426e-03 -4.24222142e-01 3.66352379e-01 2.74720951e-03 -5.77672958e-01 -8.15772891e-01 -5.19790173e-01 7.08089590e-01 8.35141614e-02 6.62922740e-01 5.61508000e-01 7.46429026e-01 -5.67526817e-01 -8.06557760e-02 7.41302609e-01 2.06968164e+00 -2.54615724e-01 6.14472687e-01 1.50076836e-01 4.59060669e-01 3.65945727e-01 6.73165143e-01 7.28559673e-01 -3.00634623e-01 8.93486202e-01 3.94390345e-01 -1.87708214e-01 5.88220358e-03 4.36623573e-01 5.03196009e-02 6.81311429e-01 -6.58594251e-01 1.00531988e-01 -2.14074761e-01 6.90078974e-01 -1.91665149e+00 -1.31208527e+00 -6.26952708e-01 2.51508498e+00 8.85709465e-01 -7.54436627e-02 7.60443732e-02 6.56990230e-01 7.99466789e-01 1.41762108e-01 -1.97241902e-01 -2.41376281e-01 -5.34409761e-01 2.75328010e-01 7.88518131e-01 7.71057606e-01 -1.09037244e+00 2.42920622e-01 6.41617727e+00 8.85635972e-01 -7.70317972e-01 5.08669019e-01 -4.67892811e-02 -1.58091068e-01 -1.89461291e-01 -1.22911327e-01 -4.30315077e-01 3.69651616e-01 8.65067840e-01 -5.80290258e-01 4.86235529e-01 5.23666084e-01 5.99052608e-01 -2.19980434e-01 -3.20594043e-01 1.20062923e+00 2.26901714e-02 -1.17872012e+00 -3.11538786e-01 -5.74533176e-03 7.78385162e-01 -6.00209892e-01 -5.87546714e-02 -3.10699075e-01 -2.07840160e-01 -8.62923443e-01 4.06805843e-01 6.06192708e-01 9.60710049e-01 -1.05689371e+00 6.63930118e-01 3.47268313e-01 -1.16116381e+00 -1.72554612e-01 -4.15238231e-01 1.80341806e-02 4.91245866e-01 1.09030032e+00 -2.28646576e-01 1.05052650e+00 4.00753111e-01 3.23840886e-01 2.63229966e-01 1.09995306e+00 1.42135695e-01 7.06852078e-01 -3.55873108e-01 6.08498693e-01 1.27272084e-01 -6.89438820e-01 1.33856058e+00 1.29427373e+00 6.08572185e-01 6.68753326e-01 7.31511861e-02 9.37667072e-01 7.52514675e-02 1.89895749e-01 -4.96625990e-01 3.85470748e-01 4.27944630e-01 1.16713595e+00 -4.93580580e-01 -1.59136146e-01 -5.57227969e-01 9.22954738e-01 -1.28077641e-01 6.93028271e-01 -7.17228293e-01 -3.04847658e-01 4.20864224e-01 3.58108073e-01 4.38405901e-01 -2.42220834e-01 -2.79895186e-01 -1.23236823e+00 2.42860496e-01 -9.01465595e-01 3.29520583e-01 -4.08733070e-01 -1.03408754e+00 3.33554327e-01 1.16765209e-01 -1.34674692e+00 5.31728156e-02 -2.70658314e-01 -1.65071964e-01 1.04876995e+00 -1.44535267e+00 -9.18354452e-01 -1.42054841e-01 6.58320606e-01 4.31995779e-01 -4.10108492e-02 6.31286025e-01 5.24296165e-01 -2.88959205e-01 3.75909269e-01 6.02696300e-01 -2.40359500e-01 4.18764353e-01 -1.26428568e+00 -4.21403885e-01 1.13271821e+00 -1.51697487e-01 4.35218424e-01 1.45530951e+00 -3.78139496e-01 -1.56572020e+00 -7.80532837e-01 6.45780265e-01 1.11333482e-01 7.27742136e-01 -1.62893653e-01 -1.20727074e+00 2.88829088e-01 1.19385654e-02 1.88109741e-01 4.34330881e-01 -1.77031700e-02 -2.16614619e-01 -2.61836618e-01 -1.31472301e+00 2.36706033e-01 9.11256373e-01 -4.47798878e-01 -4.39454228e-01 3.62125963e-01 3.66449267e-01 -3.94609690e-01 -1.14099109e+00 4.91170943e-01 3.50261837e-01 -1.05154002e+00 1.33056498e+00 -3.93551618e-01 3.81965525e-02 -4.41341996e-01 -4.01736915e-01 -1.23659289e+00 -6.62741423e-01 -1.02506363e+00 -1.27960160e-01 1.08461344e+00 8.37490559e-02 -7.55812168e-01 3.97782236e-01 1.66650251e-01 4.09658020e-03 -4.64182138e-01 -1.45676672e+00 -1.15003860e+00 -1.45169616e-01 -1.34921640e-01 2.37091817e-02 1.12615716e+00 1.50245920e-01 2.18255162e-01 -1.03558004e+00 6.05751514e-01 1.45260775e+00 2.50531808e-02 3.61541837e-01 -1.37739718e+00 -3.82721543e-01 -1.56990349e-01 -2.81654924e-01 -5.81975341e-01 -7.20748678e-02 -6.30367398e-01 -7.19924197e-02 -1.21024740e+00 2.68247098e-01 -3.31835628e-01 -3.32917541e-01 -2.23460093e-01 -1.96151957e-01 4.49298769e-01 7.45662674e-02 3.12272370e-01 -3.27546060e-01 4.52333957e-01 1.26947248e+00 1.41051799e-01 -2.29686514e-01 9.45013762e-02 -7.80936718e-01 6.83906376e-01 5.54851890e-01 -4.09512192e-01 -3.40030432e-01 9.53798145e-02 1.82154536e-01 6.80100501e-01 4.98429149e-01 -9.94546592e-01 8.80586579e-02 -1.24054871e-01 9.84476209e-02 -3.95801276e-01 5.81882238e-01 -9.75809693e-01 4.40496355e-01 2.26874828e-01 -2.12872908e-01 -4.43905264e-01 9.31631550e-02 7.84967721e-01 -1.72133878e-01 -2.73909867e-01 1.20297754e+00 1.29086692e-02 -3.83274794e-01 -1.20462514e-01 -1.86416030e-01 8.09676051e-02 8.54841352e-01 -2.95498192e-01 -7.28256181e-02 -6.54600620e-01 -9.97020245e-01 -1.05816238e-01 2.24125609e-01 -2.57624388e-01 6.15132332e-01 -1.43833780e+00 -1.39240718e+00 -1.96645185e-01 -3.73426050e-01 -3.75172317e-01 4.65699494e-01 1.43936312e+00 4.27897274e-02 2.38125205e-01 -1.44910917e-01 -3.73052955e-01 -1.30497909e+00 6.38864636e-01 3.84706140e-01 -1.09729514e-01 -1.08146191e+00 3.94174516e-01 -6.94623729e-03 1.04262881e-01 -1.54337615e-01 1.96473598e-01 -1.03994519e-01 5.67546822e-02 7.24522889e-01 7.81858444e-01 -1.97638124e-02 -9.56077993e-01 -1.78528354e-01 6.76196635e-01 3.64764631e-01 -2.01492161e-01 1.41892302e+00 -4.74442393e-01 -2.23999217e-01 2.35435486e-01 1.00606728e+00 5.74338317e-01 -1.13511539e+00 -5.36295652e-01 -1.17995210e-01 -6.46581709e-01 2.80135274e-01 -3.17194194e-01 -1.07918167e+00 4.20555294e-01 5.72301507e-01 4.34832007e-01 1.41146111e+00 -3.86634022e-01 7.72480905e-01 7.72923008e-02 4.49005514e-01 -1.15089011e+00 -4.37842727e-01 4.96332012e-02 1.02999210e+00 -9.88996804e-01 3.99495840e-01 -8.14067781e-01 -1.40152782e-01 9.72272754e-01 2.19615605e-02 -3.08920622e-01 3.79378885e-01 3.75938326e-01 -3.29868317e-01 1.44831151e-01 -4.11402255e-01 -3.78052056e-01 2.62530327e-01 7.72103488e-01 3.07808906e-01 1.16333924e-01 -8.41109037e-01 6.69537187e-01 1.43408582e-01 -4.18078639e-02 9.19400871e-01 6.13363922e-01 -6.43557310e-01 -6.96444213e-01 -1.04263890e+00 1.40616223e-01 -8.64895999e-01 -3.06405518e-02 1.55310705e-01 8.20616782e-01 -9.70749483e-02 1.11076915e+00 -5.23966551e-01 1.12566344e-01 4.24448103e-01 -3.21787149e-02 5.84215760e-01 -2.04491735e-01 -2.11838976e-01 4.70833331e-01 3.11907113e-01 -6.09786510e-01 -7.39458799e-01 -7.87202835e-01 -1.03507614e+00 -2.55628437e-01 -5.59219003e-01 3.59962851e-01 3.08724672e-01 8.70608687e-01 5.96126132e-02 3.93737555e-01 6.69054151e-01 -8.15788805e-01 -7.35809565e-01 -6.74031436e-01 -9.60666239e-01 3.59190047e-01 6.40677631e-01 -6.00815475e-01 -7.01421559e-01 8.93697217e-02]
[6.560561656951904, 1.5051029920578003]
8e9680db-d287-4d64-8f5c-13b280c67eec
conscious-inference-for-object-detection
null
null
https://openreview.net/forum?id=HygYqs0qKX
https://openreview.net/pdf?id=HygYqs0qKX
Conscious Inference for Object Detection
Current Convolutional Neural Network (CNN)-based object detection models adopt strictly feedforward inference to predict the final detection results. However, the widely used one-way inference is agnostic to the global image context and the interplay between input image and task semantics. In this work, we present a general technique to improve off-the-shelf CNN-based object detection models in the inference stage without re-training, architecture modification or ground-truth requirements. We propose an iterative, bottom-up and top-down inference mechanism, which is named conscious inference, as it is inspired by prevalent models for human consciousness with top-down guidance and temporal persistence. While the downstream pass accumulates category-specific evidence over time, it subsequently affects the proposal calculation and the final detection. Feature activations are updated in line with no additional memory cost. Our approach advances the state of the art using popular detection models (Faster-RCNN, YOLOv2, YOLOv3) on 2D object detection and 6D object pose estimation.
['Gang Hua', 'Ying Wu', 'Nikolaos Karianakis', 'Jiahuan Zhou']
2018-09-27
null
null
null
null
['6d-pose-estimation']
['computer-vision']
[ 6.27825558e-02 -1.01198256e-01 3.91696334e-01 -2.62076408e-01 2.10841939e-01 -4.17598397e-01 8.42814744e-01 4.07046646e-01 -8.50291908e-01 2.98526436e-01 -2.87642568e-01 -1.32691905e-01 2.01094389e-01 -8.60912323e-01 -7.21349597e-01 -6.02054954e-01 1.22856110e-01 4.00221884e-01 1.01112139e+00 -8.45478922e-02 7.50232339e-01 6.86895311e-01 -2.01655912e+00 3.51025969e-01 3.63953263e-01 1.25543630e+00 6.98076308e-01 9.22454298e-01 -4.05864418e-02 7.61424363e-01 -3.88680935e-01 -4.05490428e-01 1.54113710e-01 -2.13447660e-01 -5.03523290e-01 -1.55254183e-02 6.32381678e-01 -4.80879813e-01 -2.09736675e-01 1.28832340e+00 4.93353069e-01 -1.02732189e-01 7.69638419e-01 -9.68718886e-01 -7.70265162e-01 3.57683361e-01 -3.71281564e-01 6.05658531e-01 8.81726146e-02 4.84794140e-01 4.81958508e-01 -1.37364256e+00 5.13181150e-01 1.25190067e+00 7.38045514e-01 5.87102592e-01 -1.01676977e+00 -2.82442003e-01 5.86250246e-01 5.24363995e-01 -1.21548462e+00 -3.55936527e-01 5.66195607e-01 -6.55756950e-01 1.44457173e+00 -1.21917427e-01 1.08485627e+00 9.98822987e-01 2.40225568e-01 6.51293278e-01 1.23070860e+00 -4.96718675e-01 3.66519481e-01 5.68074882e-02 2.43898228e-01 1.01676333e+00 6.27109826e-01 4.49860334e-01 -7.49471307e-01 3.75940204e-01 1.06817615e+00 -5.39527498e-02 4.38256934e-02 -5.97351432e-01 -1.20208693e+00 6.21675730e-01 8.30803871e-01 1.47644803e-01 -6.42138839e-01 3.10723931e-01 3.31138641e-01 7.79979676e-02 1.38663784e-01 2.28270665e-01 -3.35972071e-01 3.08014631e-01 -1.10474825e+00 3.68084699e-01 7.93817997e-01 8.26424837e-01 7.25618243e-01 8.42602737e-03 -5.68728447e-01 2.45680749e-01 6.02498949e-01 2.35922813e-01 4.10567999e-01 -7.53877699e-01 4.80332337e-02 8.01385164e-01 1.64650634e-01 -9.49903846e-01 -6.64453328e-01 -7.62849092e-01 -5.36878824e-01 6.71818912e-01 4.58432078e-01 2.98003525e-01 -1.15645027e+00 1.53543782e+00 2.83554554e-01 4.78142202e-02 -1.33127749e-01 1.27873361e+00 1.07323575e+00 1.79251537e-01 2.75141865e-01 -5.37935793e-02 1.72902501e+00 -1.07143474e+00 -5.43127298e-01 -5.20428598e-01 9.57683846e-02 -6.90927088e-01 6.19929314e-01 4.51693058e-01 -1.40456879e+00 -9.15245891e-01 -1.20512772e+00 -4.22239602e-01 -7.45780349e-01 5.21054208e-01 7.82507658e-01 4.70554113e-01 -1.18809557e+00 4.67681795e-01 -7.81566083e-01 -6.56198144e-01 6.73601031e-01 4.92063344e-01 -8.19327235e-02 1.56287253e-01 -8.25173497e-01 1.43379486e+00 6.79304659e-01 4.56159860e-01 -1.48062432e+00 -3.21805060e-01 -5.32356560e-01 -1.11092255e-02 4.55463439e-01 -1.25287569e+00 1.44706666e+00 -6.73795819e-01 -1.65223968e+00 1.29756975e+00 -6.76003844e-02 -7.39252329e-01 7.37018049e-01 -3.16529214e-01 8.31452385e-02 1.24512486e-01 -1.39609784e-01 1.23869693e+00 1.33306170e+00 -1.21371841e+00 -6.16755962e-01 -4.81288016e-01 2.06396639e-01 2.44712010e-01 1.84584275e-01 -5.11244386e-02 -5.01756430e-01 -3.44419509e-01 4.68124151e-01 -6.61253631e-01 -1.12971559e-01 4.95141953e-01 -1.58937529e-01 -5.74907064e-01 6.83153033e-01 -2.00789362e-01 8.05915833e-01 -1.82379603e+00 2.76879132e-01 -3.25910836e-01 2.78476685e-01 3.49612325e-01 6.18370846e-02 9.07362998e-03 2.44565248e-01 -4.31839883e-01 -1.61482126e-01 -5.42173207e-01 8.28029681e-03 4.59292531e-02 -2.35939115e-01 6.56597376e-01 4.89248008e-01 1.08990824e+00 -9.31058407e-01 -5.60634971e-01 6.55479252e-01 3.21879953e-01 -8.23108315e-01 1.92944452e-01 -4.56519783e-01 2.97209501e-01 -1.02620134e-02 6.78896010e-01 7.87565351e-01 -2.55453408e-01 -3.40370536e-02 -6.62767887e-01 -6.87755644e-01 3.24760884e-01 -1.06093204e+00 1.99668360e+00 -3.01464975e-01 6.86240494e-01 1.89903304e-02 -1.09475338e+00 8.25829446e-01 -1.24253510e-02 -3.08292180e-01 -6.89501643e-01 5.82616627e-01 1.79742366e-01 1.86937392e-01 -4.82062280e-01 4.72035646e-01 -1.24158682e-02 2.39270717e-01 1.50662914e-01 4.91812766e-01 -1.19118311e-01 1.90600917e-01 1.01747094e-02 5.19567072e-01 5.29990315e-01 6.06923044e-01 -2.34343737e-01 4.62745219e-01 1.20932184e-01 1.25659898e-01 1.20318878e+00 -5.98625064e-01 6.44444406e-01 2.80483395e-01 -5.74210167e-01 -8.39007139e-01 -1.14679992e+00 -1.85487181e-01 1.21155417e+00 3.92070621e-01 -3.12348865e-02 -7.85918951e-01 -4.77320343e-01 -5.32568358e-02 5.80264390e-01 -9.40243721e-01 -2.27858156e-01 -4.44152921e-01 -7.21749723e-01 3.10581833e-01 5.97155154e-01 7.81460643e-01 -1.49632418e+00 -1.42805326e+00 4.37162220e-01 1.56519622e-01 -1.17861462e+00 2.21881866e-01 4.42015469e-01 -9.73376691e-01 -9.24353778e-01 -5.37803292e-01 -6.65025771e-01 7.61307657e-01 1.72302231e-01 1.14712048e+00 1.49133112e-02 -6.35768294e-01 2.17129543e-01 -5.33676008e-03 -7.36172378e-01 -6.16478920e-02 -4.29929085e-02 1.01282552e-01 -1.81254372e-01 5.48778713e-01 -4.07769382e-01 -9.57050622e-01 -2.24857591e-02 -5.07804811e-01 2.48620987e-01 8.95795941e-01 5.27211487e-01 4.52310413e-01 -6.20916009e-01 3.57138693e-01 -3.85465026e-01 3.81498218e-01 -1.26320636e-02 -7.83022106e-01 1.28226817e-01 -3.97091329e-01 1.11703031e-01 1.84477299e-01 -4.82057005e-01 -1.05614924e+00 2.85235524e-01 5.31893261e-02 -6.71314478e-01 -4.03165102e-01 7.28591010e-02 3.62243772e-01 -3.06504548e-01 7.79924810e-01 4.12101746e-01 -3.24812174e-01 -5.44445992e-01 4.89690930e-01 1.50021315e-01 7.87348747e-01 -1.36093780e-01 5.05770743e-01 7.16899335e-01 3.56471986e-02 -4.78722662e-01 -1.12610543e+00 -3.42687786e-01 -1.18596101e+00 -4.28562611e-01 1.29392290e+00 -1.02783680e+00 -1.12495899e+00 7.14446366e-01 -1.62704360e+00 -2.01366201e-01 -1.92783475e-01 4.35177267e-01 -5.62340975e-01 -2.30691098e-02 -4.88811642e-01 -7.66460598e-01 -4.13346916e-01 -9.79778528e-01 1.08254778e+00 2.97370732e-01 3.73819540e-03 -5.95764220e-01 -1.50804326e-01 -1.98826194e-01 5.47892690e-01 -1.45274311e-01 6.35438979e-01 -2.95872867e-01 -9.49156463e-01 -6.36106879e-02 -8.54032159e-01 1.01203777e-01 -4.62484628e-01 -9.74841341e-02 -1.20739651e+00 -1.38285100e-01 7.42468759e-02 -2.81395674e-01 1.28325772e+00 5.54797351e-01 1.16026151e+00 -2.45381054e-02 -2.64643908e-01 5.47133267e-01 1.49984586e+00 -3.42849046e-01 4.40486044e-01 3.58259916e-01 4.06221420e-01 5.18788755e-01 3.91325772e-01 3.93930703e-01 3.97372812e-01 6.76263273e-01 1.00245452e+00 2.45322511e-01 -6.29724860e-01 -2.00208351e-01 1.74375981e-01 6.79039299e-01 -4.99029726e-01 1.08987585e-01 -7.79216528e-01 4.95205522e-01 -1.89055383e+00 -1.13838792e+00 -9.06854719e-02 2.04479003e+00 6.26660645e-01 6.19955182e-01 -3.84691320e-02 -1.50936469e-01 5.70927620e-01 -1.36400521e-01 -6.58483624e-01 -4.15638566e-01 -2.79515497e-02 1.49401441e-01 2.98611403e-01 2.85665274e-01 -1.11825621e+00 1.24602008e+00 6.44263840e+00 4.01455253e-01 -1.27823925e+00 3.65275353e-01 8.28393083e-03 -7.51210228e-02 4.12885487e-01 -3.08858114e-03 -1.17423320e+00 1.03741668e-01 4.72446591e-01 2.30107516e-01 2.50262916e-01 9.48731244e-01 -7.04994367e-04 -3.75242651e-01 -1.27900243e+00 1.04935241e+00 2.25450471e-01 -1.59317946e+00 1.94536552e-01 -2.68691331e-01 4.84906763e-01 3.23200405e-01 3.59828249e-02 3.98425817e-01 1.12648057e-02 -7.91741669e-01 1.39395666e+00 7.53598094e-01 4.00016099e-01 -3.72874767e-01 6.36242807e-01 4.68663007e-01 -1.14829659e+00 -3.35392475e-01 -6.40499413e-01 -3.24045062e-01 6.39761984e-02 3.28437656e-01 -8.70064139e-01 6.47445992e-02 8.50132287e-01 5.94667256e-01 -9.71913636e-01 1.30569410e+00 -5.58354497e-01 1.59130767e-01 -1.87798664e-01 -3.96965772e-01 3.85494053e-01 2.08634183e-01 7.00127840e-01 1.51940465e+00 8.35131481e-02 1.20933503e-01 3.58724743e-02 1.18569374e+00 8.64065140e-02 -4.99702781e-01 -3.03968847e-01 5.62157810e-01 3.63348007e-01 1.26686263e+00 -1.07612562e+00 -5.42617977e-01 -2.11823747e-01 1.13875020e+00 6.31841481e-01 1.95210233e-01 -8.39376092e-01 -9.26318243e-02 2.76792437e-01 1.32790906e-02 8.20083916e-01 -4.03975755e-01 -4.51693475e-01 -1.09026802e+00 -2.29376137e-01 -3.29384744e-01 1.44803792e-01 -1.07411957e+00 -1.18570542e+00 4.53014761e-01 -3.10786013e-02 -1.01350057e+00 7.95540884e-02 -1.18159616e+00 -5.52165329e-01 5.64713299e-01 -1.74919236e+00 -1.16436839e+00 -6.21192455e-01 5.48504889e-01 6.27872169e-01 7.43246153e-02 7.23673284e-01 9.14723277e-02 -3.79586995e-01 2.19206423e-01 -7.05425262e-01 -6.99986098e-03 3.86613578e-01 -1.12989259e+00 2.92101204e-01 9.47588027e-01 9.20494795e-02 7.41993189e-01 6.70579195e-01 -4.09138113e-01 -1.23665309e+00 -9.89475489e-01 8.65831435e-01 -5.86967587e-01 5.15384555e-01 -5.98591626e-01 -7.31307805e-01 5.32414794e-01 2.37177894e-01 2.46987954e-01 4.91562933e-02 -1.29133435e-02 -4.85069990e-01 -3.97807881e-02 -1.07276249e+00 6.77049518e-01 1.41636407e+00 -4.01914477e-01 -1.03686178e+00 3.38493288e-02 6.39851987e-01 -3.57536465e-01 -2.80301481e-01 4.23176199e-01 6.68678522e-01 -1.23996377e+00 1.12101972e+00 -4.60754633e-01 1.26260787e-01 -6.30375743e-01 -1.21161237e-01 -6.90842032e-01 -6.65588379e-01 -1.29290909e-01 -6.03022635e-01 5.35770237e-01 1.68085769e-01 -3.10770243e-01 4.73931521e-01 3.03700683e-03 -4.03734773e-01 -5.34036458e-01 -1.08726525e+00 -4.55907941e-01 -2.70545572e-01 -4.82228905e-01 1.72306925e-01 3.97154570e-01 -3.43755305e-01 4.09385532e-01 3.46167125e-02 3.79205942e-01 9.52237904e-01 3.75988305e-01 4.88616556e-01 -1.39581311e+00 -1.04675099e-01 -8.38026404e-01 -6.68722093e-01 -1.19834626e+00 -2.94615746e-01 -7.60452628e-01 2.20908031e-01 -1.58789432e+00 2.56340027e-01 -8.17145705e-02 -3.46566319e-01 4.58914548e-01 -2.53773592e-02 5.32165706e-01 4.05516148e-01 2.85355031e-01 -9.61393893e-01 3.65527213e-01 1.29788721e+00 2.49271095e-01 -1.51854038e-01 -6.04221039e-02 -2.74275094e-01 9.63082314e-01 6.79973543e-01 -5.07878780e-01 -5.66892736e-02 -6.15067124e-01 4.10851657e-01 -4.83437121e-01 1.21355486e+00 -1.27471924e+00 7.03284144e-01 1.84650689e-01 6.98453546e-01 -1.12008476e+00 4.29319859e-01 -6.11891925e-01 -4.18379277e-01 9.50626194e-01 -1.89653486e-01 -3.30251940e-02 2.45551258e-01 6.98783875e-01 1.46579802e-01 -4.49240923e-01 1.01584148e+00 -6.49392664e-01 -1.31460059e+00 2.95808852e-01 -5.39330482e-01 -3.48483801e-01 9.36758280e-01 -4.14307535e-01 -3.38193744e-01 2.51457632e-01 -1.14870930e+00 -2.16968954e-01 2.36576155e-01 4.24018025e-01 7.74328053e-01 -1.00565648e+00 -6.39263809e-01 2.13080868e-01 2.17309088e-01 -2.30062995e-02 1.08191870e-01 1.14130914e+00 -4.03508127e-01 5.71064472e-01 -4.61672276e-01 -9.32604373e-01 -8.86001289e-01 7.22055495e-01 5.05430400e-01 1.18743837e-01 -5.74613154e-01 1.22442746e+00 2.32174933e-01 -2.83594102e-01 3.54501754e-01 -5.86907685e-01 -2.74071962e-01 2.26442292e-01 4.40709144e-01 1.81121230e-01 1.66026652e-01 -2.68610597e-01 -5.44474125e-01 7.08404839e-01 2.84931455e-02 -1.10344045e-01 1.13032115e+00 -1.56559914e-01 -1.95233196e-01 5.21100760e-01 7.37482488e-01 -6.42511725e-01 -1.61646521e+00 -1.51896358e-01 -1.69217601e-01 -2.50586241e-01 2.87334919e-01 -9.64928627e-01 -7.39558041e-01 1.35799050e+00 8.79745185e-01 1.06915846e-01 8.95362020e-01 1.85115382e-01 2.44919490e-02 7.05820739e-01 4.36092675e-01 -1.02675164e+00 5.97590029e-01 7.01106369e-01 1.05287457e+00 -1.37368643e+00 1.63093939e-01 -1.91673815e-01 -2.29784310e-01 1.20019686e+00 1.00064671e+00 -4.86689687e-01 5.62146008e-01 1.00091457e-01 -3.84437263e-01 -4.13220048e-01 -8.95803332e-01 -6.75906897e-01 2.83675224e-01 5.99178374e-01 1.62548527e-01 -1.17080957e-01 -1.76204415e-03 3.81788433e-01 -5.99299781e-02 1.15360469e-01 9.57024917e-02 8.09880674e-01 -9.98863697e-01 -4.04910654e-01 -2.10257068e-01 1.51315153e-01 -6.50376678e-02 -3.27187777e-01 -3.47234756e-01 7.61509001e-01 7.76978433e-01 4.97331738e-01 4.35981214e-01 1.39655620e-01 2.19486311e-01 4.84180339e-02 1.13394201e+00 -7.51883328e-01 -6.66384220e-01 -1.79655492e-01 -2.50943214e-01 -3.91137689e-01 -7.44206131e-01 -7.62898862e-01 -1.20459533e+00 1.32279247e-02 -5.53740203e-01 -5.03185749e-01 9.33248341e-01 9.65245426e-01 4.12612647e-01 7.30563581e-01 2.94101294e-02 -1.45708716e+00 -4.00250226e-01 -1.05683804e+00 9.91529748e-02 5.32936305e-02 3.65486294e-01 -1.07160079e+00 -1.15027219e-01 1.42406464e-01]
[9.306233406066895, 0.29851096868515015]
7ab4b1af-d639-4215-a82e-1c99ca1a2227
improving-image-captioning-descriptiveness-by
2306.11593
null
https://arxiv.org/abs/2306.11593v1
https://arxiv.org/pdf/2306.11593v1.pdf
Improving Image Captioning Descriptiveness by Ranking and LLM-based Fusion
State-of-The-Art (SoTA) image captioning models often rely on the Microsoft COCO (MS-COCO) dataset for training. This dataset contains annotations provided by human annotators, who typically produce captions averaging around ten tokens. However, this constraint presents a challenge in effectively capturing complex scenes and conveying detailed information. Furthermore, captioning models tend to exhibit bias towards the ``average'' caption, which captures only the more general aspects. What would happen if we were able to automatically generate longer captions, thereby making them more detailed? Would these captions, evaluated by humans, be more or less representative of the image content compared to the original MS-COCO captions? In this paper, we present a novel approach to address previous challenges by showcasing how captions generated from different SoTA models can be effectively fused, resulting in richer captions. Our proposed method leverages existing models from the literature, eliminating the need for additional training. Instead, it utilizes an image-text based metric to rank the captions generated by SoTA models for a given image. Subsequently, the top two captions are fused using a Large Language Model (LLM). Experimental results demonstrate the effectiveness of our approach, as the captions generated by our model exhibit higher consistency with human judgment when evaluated on the MS-COCO test set. By combining the strengths of various SoTA models, our method enhances the quality and appeal of image captions, bridging the gap between automated systems and the rich, informative nature of human-generated descriptions. This advance opens up new possibilities for generating captions that are more suitable for the training of both vision-language and captioning models.
['Paolo Napoletano', 'Marco Donzella', 'Luigi Celona', 'Simone Bianco']
2023-06-20
null
null
null
null
['image-captioning']
['computer-vision']
[ 5.06444216e-01 3.73023748e-01 -8.52984935e-02 -3.41596454e-01 -1.14996803e+00 -6.62612021e-01 7.78376222e-01 1.22900940e-01 -2.73972780e-01 7.44160771e-01 4.73436266e-01 -1.48156077e-01 4.30843592e-01 -5.12762070e-01 -1.00578117e+00 -4.54691648e-01 4.64709431e-01 4.98639435e-01 9.57700908e-02 -2.21308112e-01 3.32769640e-02 -1.38658434e-02 -1.65072048e+00 6.62840426e-01 1.01613069e+00 1.00558650e+00 5.82471490e-01 4.71120387e-01 -2.45509401e-01 7.55454659e-01 -6.42750323e-01 -8.06120098e-01 5.09563014e-02 -5.18897593e-01 -6.39691949e-01 4.03071553e-01 8.51886570e-01 -3.04117590e-01 -5.34731988e-03 9.12975252e-01 1.58620283e-01 -4.55113024e-01 6.42343521e-01 -1.33552241e+00 -1.05858791e+00 6.45073175e-01 -4.81741667e-01 -2.86553204e-01 5.37132442e-01 3.76970023e-01 1.15998971e+00 -8.76432478e-01 6.95618570e-01 1.16165078e+00 4.24533427e-01 7.02169657e-01 -1.18302131e+00 -5.17884672e-01 1.21193573e-01 -3.49260233e-02 -1.06672692e+00 -3.76537442e-01 6.20984137e-01 -5.38124263e-01 4.61800784e-01 2.75402486e-01 5.04195511e-01 1.37068617e+00 -2.22511917e-01 1.11990881e+00 1.23959494e+00 -5.52441359e-01 5.20809516e-02 4.74405885e-01 -1.57886505e-01 3.92168880e-01 4.08096999e-01 -1.92693233e-01 -3.73027951e-01 -1.46481385e-02 5.30549943e-01 -2.37239733e-01 -4.85210031e-01 -3.96475255e-01 -1.49546623e+00 7.10301697e-01 5.21977425e-01 2.72839427e-01 -6.30291998e-01 1.40451416e-01 3.30386400e-01 -2.87815094e-01 3.87710094e-01 7.40213871e-01 -8.49022642e-02 -6.88465461e-02 -9.69366193e-01 2.43931398e-01 5.47311664e-01 1.05663264e+00 7.00768888e-01 -2.25143149e-01 -6.82491899e-01 8.61543417e-01 3.75366002e-01 9.01641786e-01 3.27364713e-01 -1.02142251e+00 5.79893351e-01 4.58341420e-01 4.43556666e-01 -8.96581948e-01 2.58578300e-01 -4.91380423e-01 -5.09286761e-01 -4.84949984e-02 2.71326482e-01 1.13741616e-02 -1.21842027e+00 1.86829305e+00 -5.29556386e-02 -7.17459470e-02 2.82382458e-01 1.03110385e+00 8.02159607e-01 8.50715458e-01 4.82126176e-01 1.15821650e-02 1.50092721e+00 -1.27898872e+00 -8.15867543e-01 -5.95684350e-01 5.10762453e-01 -1.00194263e+00 1.25746238e+00 -3.08253965e-03 -1.04063833e+00 -6.07670546e-01 -9.35873151e-01 1.12416975e-01 -2.22784147e-01 2.89200753e-01 4.79296982e-01 4.32047218e-01 -1.26771390e+00 8.98378789e-02 -3.19051981e-01 -3.02715302e-01 6.11123145e-01 -1.08563103e-01 -4.05862778e-01 -4.29544896e-01 -1.09136045e+00 1.06770754e+00 4.19631988e-01 1.09955825e-01 -9.50797856e-01 -4.96187598e-01 -1.12113822e+00 2.06134822e-02 3.09979916e-01 -8.14202130e-01 1.42444789e+00 -1.39368773e+00 -1.09109628e+00 9.14700210e-01 -3.67180556e-01 -5.62359691e-01 5.22278965e-01 -2.47367844e-01 -1.52348444e-01 4.57577556e-01 4.70498800e-01 1.45691371e+00 6.91881418e-01 -1.92240679e+00 -4.92193311e-01 1.86983898e-01 2.94062227e-01 3.11508358e-01 -2.67957270e-01 -6.12739064e-02 -6.97291195e-01 -5.11754394e-01 -4.35506016e-01 -1.03838956e+00 -2.78567493e-01 4.30361144e-02 -3.75982434e-01 -7.00407550e-02 6.17284298e-01 -5.37519693e-01 9.38762248e-01 -2.10163164e+00 -1.14475243e-01 -2.26322755e-01 1.12494335e-01 5.58289647e-01 -4.70100850e-01 5.17329037e-01 1.37145653e-01 3.15412313e-01 -4.41908628e-01 -5.63665509e-01 1.82642713e-02 3.17341894e-01 -5.30115724e-01 -1.37337521e-01 5.36713541e-01 1.12132692e+00 -1.25666583e+00 -7.90341020e-01 2.92971283e-01 5.26462913e-01 -2.27368653e-01 3.91015202e-01 -5.60356975e-01 4.39843863e-01 -3.81718338e-01 4.51714575e-01 5.80645025e-01 -5.55127859e-01 -8.81428346e-02 -3.03309560e-01 1.41247272e-01 -5.25678173e-02 -5.24788797e-01 1.47733283e+00 -6.68210745e-01 8.29819560e-01 -2.48329014e-01 -5.49621344e-01 8.64116430e-01 4.67839956e-01 2.83854634e-01 -7.11993337e-01 1.40439257e-01 3.84004265e-01 -2.08723038e-01 -6.57089770e-01 5.82660556e-01 -9.56496969e-02 -1.76767752e-01 4.00542915e-01 8.74171872e-03 -3.84830147e-01 4.70323712e-01 4.45783883e-01 6.61565721e-01 3.23021561e-01 6.08284734e-02 2.98796535e-01 4.25457984e-01 3.65660667e-01 2.22804636e-01 9.48004127e-01 -3.05991322e-01 1.16124475e+00 3.18915963e-01 -1.18759409e-01 -1.44604123e+00 -9.51352537e-01 -7.82508496e-03 6.88664377e-01 1.59138352e-01 -4.16054964e-01 -1.10240269e+00 -7.09459543e-01 -2.73309767e-01 8.56847286e-01 -7.51459002e-01 1.02537259e-01 -3.50028187e-01 -2.92273641e-01 4.78306085e-01 4.36820745e-01 4.88171786e-01 -1.37026966e+00 -6.11793101e-01 7.67231882e-02 -7.70225942e-01 -1.63248944e+00 -6.22042239e-01 -4.07557547e-01 -5.16749561e-01 -8.03023219e-01 -1.19468820e+00 -8.08266997e-01 8.59988213e-01 5.26445627e-01 1.32213974e+00 8.77460390e-02 1.27823194e-02 5.89805603e-01 -6.76558256e-01 -6.31885707e-01 -7.84983158e-01 -1.65323064e-01 -3.06360692e-01 3.06364685e-01 2.99468488e-01 -1.79305315e-01 -5.42420924e-01 1.74692124e-01 -1.25021183e+00 7.37588644e-01 1.08737135e+00 7.66517043e-01 4.60755318e-01 -6.21849716e-01 6.80729806e-01 -8.78945291e-01 5.78807890e-01 -4.90141690e-01 -2.27650389e-01 4.58483279e-01 -4.59284365e-01 7.59642348e-02 6.36326432e-01 -5.47920406e-01 -1.00614250e+00 2.30054110e-01 6.90108389e-02 -6.19408846e-01 -2.11098820e-01 5.40682971e-01 2.44142804e-02 1.86575517e-01 5.57353854e-01 4.45587188e-01 3.28945108e-02 -1.44164562e-01 7.08469689e-01 8.85459125e-01 7.61917830e-01 -6.59729600e-01 9.17240143e-01 4.32338864e-01 -3.71691138e-01 -5.59007406e-01 -1.34272408e+00 -4.34019178e-01 -2.81740427e-01 -4.61801976e-01 1.00127757e+00 -1.06411183e+00 -2.27834016e-01 2.98915416e-01 -1.58284402e+00 1.08138494e-01 -1.29021600e-01 3.07538778e-01 -5.59017241e-01 3.87724012e-01 -2.89893955e-01 -8.49492371e-01 -5.07039726e-01 -1.35121179e+00 1.51251590e+00 3.83353025e-01 -2.46445313e-01 -7.82599270e-01 -8.06391537e-02 9.23650742e-01 4.21242297e-01 4.10249710e-01 5.84184527e-01 -4.60716039e-01 -5.35898149e-01 -3.27743620e-01 -6.11610770e-01 5.49360931e-01 7.77298063e-02 -1.66081339e-01 -1.15953481e+00 -1.15248762e-01 -3.45879793e-01 -6.27504289e-01 6.40544951e-01 1.32305786e-01 8.99109602e-01 -3.96090209e-01 -1.53030753e-01 1.72299030e-03 1.47487974e+00 -1.23962402e-01 7.46989727e-01 4.44144100e-01 7.48000801e-01 8.03158462e-01 7.92378485e-01 -3.11977081e-02 5.84242702e-01 7.01852322e-01 6.03638887e-01 -4.58500713e-01 -4.52263385e-01 -5.66638410e-01 4.44098830e-01 6.56852603e-01 1.59383371e-01 -4.85564977e-01 -9.20766175e-01 9.76861954e-01 -1.94178414e+00 -8.12090516e-01 -1.54219002e-01 1.90840089e+00 1.04034758e+00 -1.62304081e-02 -1.83165416e-01 -2.59842306e-01 9.55290258e-01 9.72654223e-02 -3.17933738e-01 -3.00081551e-01 -2.02184692e-01 -5.21894284e-02 4.30908173e-01 2.93661773e-01 -9.30999935e-01 9.58263159e-01 6.30542040e+00 7.03697085e-01 -1.05298114e+00 3.72578725e-02 7.33220220e-01 1.70059443e-01 -4.98226941e-01 1.61661431e-01 -5.86947739e-01 5.64536870e-01 9.97114301e-01 -2.02542216e-01 2.27693040e-02 8.27275574e-01 3.73618722e-01 -1.64724007e-01 -1.14219248e+00 1.08518803e+00 5.28355896e-01 -1.44491267e+00 6.69549167e-01 8.10348317e-02 9.13395107e-01 -4.93974388e-02 1.88665718e-01 2.13070169e-01 1.57139614e-01 -1.11023128e+00 1.12041545e+00 4.17609841e-01 8.18545759e-01 -2.18411401e-01 1.08723056e+00 1.04623646e-01 -7.96781957e-01 1.35406002e-01 -1.43116772e-01 9.58470851e-02 4.84086215e-01 5.01214862e-01 -1.16785586e+00 6.97622776e-01 4.37902868e-01 5.69594443e-01 -9.05382037e-01 1.19913566e+00 -2.99196541e-01 5.86201429e-01 -2.36170199e-02 -1.56869531e-01 6.27869844e-01 2.21796148e-02 4.49783206e-01 1.33018136e+00 2.94441491e-01 -1.61861688e-01 2.15239495e-01 9.41330671e-01 -2.12540939e-01 1.74763501e-01 -7.04275846e-01 -3.17776978e-01 3.64519596e-01 1.32396936e+00 -5.66610277e-01 -5.73900223e-01 -4.17768091e-01 8.51085842e-01 2.05465421e-01 3.72945935e-01 -8.69094014e-01 -1.29088998e-01 1.40047640e-01 1.92315862e-01 2.68076509e-01 5.40534109e-02 -2.72535235e-01 -1.13108754e+00 3.06605905e-01 -1.06516647e+00 1.33500323e-01 -1.46123970e+00 -1.37734997e+00 1.03491485e+00 7.28379413e-02 -1.50158727e+00 -3.13966334e-01 -5.92231750e-01 -4.03839082e-01 7.98748374e-01 -1.72026789e+00 -1.65029180e+00 -5.44346571e-01 1.26543507e-01 6.89630866e-01 2.42620349e-01 6.47009492e-01 2.38487259e-01 -1.93442598e-01 3.83587539e-01 -2.62424171e-01 1.55885294e-01 9.04808700e-01 -1.11195254e+00 2.32599452e-01 9.72043395e-01 3.14087093e-01 5.13958752e-01 1.03985786e+00 -5.77385902e-01 -8.34251285e-01 -1.20689869e+00 9.48384762e-01 -7.05967367e-01 4.89980668e-01 -3.11408699e-01 -9.25525963e-01 4.16633844e-01 4.93315339e-01 -1.89263299e-01 5.31838715e-01 -2.72679090e-01 -5.48848808e-01 7.28977472e-02 -9.09072399e-01 7.24797368e-01 6.80397689e-01 -4.69615102e-01 -7.18040884e-01 3.00581396e-01 9.54029560e-01 -2.52422631e-01 -4.60595280e-01 3.34561527e-01 4.44335282e-01 -7.91115522e-01 7.94570982e-01 -3.21333051e-01 1.04391778e+00 -5.27321398e-01 -2.24295128e-02 -1.20513570e+00 3.45907770e-02 -4.41692203e-01 1.95087925e-01 1.39754558e+00 6.37986362e-01 -1.97078839e-01 4.41934526e-01 5.77634096e-01 -2.11725026e-01 -6.37191176e-01 -5.86258054e-01 -7.20988393e-01 -9.54295173e-02 -4.16130573e-01 5.25828183e-01 7.70588934e-01 -2.56923229e-01 4.73524213e-01 -5.33765554e-01 3.35175805e-02 5.91183364e-01 9.05728247e-03 8.02408516e-01 -9.57665622e-01 -8.50507393e-02 -3.36175442e-01 -3.06610793e-01 -9.07502115e-01 1.25657186e-01 -8.33896637e-01 4.67946202e-01 -1.83183062e+00 5.94087064e-01 -3.35463166e-01 1.74333602e-02 6.44044042e-01 -4.45065111e-01 7.51110137e-01 5.64869046e-01 4.05288428e-01 -8.73080969e-01 5.49444675e-01 1.65263236e+00 -2.07400069e-01 1.42463982e-01 -4.59150821e-01 -9.86738563e-01 4.67234761e-01 5.31727910e-01 -3.98007393e-01 -3.91221881e-01 -7.22068965e-01 8.51608738e-02 -9.67259631e-02 5.48422813e-01 -1.02290452e+00 5.37048131e-02 -1.77779645e-01 2.08188042e-01 -3.28095496e-01 3.88829947e-01 -6.68400109e-01 1.39877275e-01 1.54090062e-01 -4.15546477e-01 -1.20548747e-01 2.35551760e-01 5.58692694e-01 -4.74049062e-01 -4.07722026e-01 6.96159899e-01 -1.63679838e-01 -5.11530578e-01 1.27714232e-01 -2.23914832e-01 -7.70260617e-02 1.04373801e+00 -2.18939796e-01 -4.62070376e-01 -7.98474848e-01 -4.05974656e-01 3.60897243e-01 6.89166427e-01 6.59326017e-01 5.70561171e-01 -1.37941718e+00 -1.04867578e+00 -3.63311499e-01 6.88608050e-01 6.35176301e-02 1.07944660e-01 5.14126718e-01 -5.09795606e-01 6.37411654e-01 -1.80561438e-01 -8.07176352e-01 -1.02956033e+00 5.05283296e-01 3.21695432e-02 -2.45431051e-01 -4.35853392e-01 3.68297905e-01 4.37843591e-01 -6.03977703e-02 -8.60573798e-02 -1.08086064e-01 -2.61783749e-01 -7.55489394e-02 6.10773325e-01 -4.03686970e-01 -3.35678846e-01 -9.45807576e-01 -9.14218798e-02 4.93265271e-01 -2.37785399e-01 -3.77899528e-01 1.25299108e+00 -2.76690543e-01 1.32825412e-02 2.01964617e-01 9.45589840e-01 -2.07621548e-02 -1.34058535e+00 -1.85350254e-01 -8.61425921e-02 -4.47134376e-01 -9.87511650e-02 -1.05905688e+00 -7.18011081e-01 8.69257331e-01 3.18645924e-01 6.56646192e-02 1.02521813e+00 3.02789330e-01 1.00643909e+00 7.49765486e-02 3.92658025e-01 -7.41573513e-01 3.81663829e-01 2.60830969e-01 1.01328492e+00 -1.49446058e+00 -3.88673455e-01 -5.23028493e-01 -1.10016990e+00 9.65494812e-01 7.05914974e-01 1.72777981e-01 -2.22749114e-01 -2.33545020e-01 4.00446355e-01 -5.28699066e-03 -6.07500672e-01 -4.60416526e-01 4.15545344e-01 7.57052779e-01 3.03803355e-01 -2.79663987e-02 -1.10782579e-01 4.44209337e-01 -1.43532187e-01 1.45443706e-02 9.22375083e-01 7.53944218e-01 -3.64222109e-01 -1.10815847e+00 -5.16747057e-01 1.49347872e-01 -3.09342682e-01 -3.08433473e-01 -4.71934438e-01 5.22389531e-01 2.54422110e-02 1.11190593e+00 -9.04144570e-02 -1.03750050e-01 6.66651577e-02 2.18213648e-02 3.35009247e-01 -8.05038989e-01 -2.41738543e-01 -1.08643211e-01 1.08788252e-01 -2.67400473e-01 -6.08611047e-01 -4.44719374e-01 -9.07550871e-01 2.54122764e-01 -2.36231208e-01 4.07851577e-01 7.75641978e-01 1.07150328e+00 4.74529117e-01 3.49357724e-01 4.96142864e-01 -8.80063415e-01 -1.98620155e-01 -1.00884926e+00 1.49315625e-01 9.23992693e-01 3.16186488e-01 -5.05843461e-01 -3.64693761e-01 5.11100650e-01]
[10.987144470214844, 1.0575937032699585]
cda6d9a8-1e6a-4fcf-8328-d1085e205956
intent-recognition-in-conversational
2212.03721
null
https://arxiv.org/abs/2212.03721v1
https://arxiv.org/pdf/2212.03721v1.pdf
Intent Recognition in Conversational Recommender Systems
Any organization needs to improve their products, services, and processes. In this context, engaging with customers and understanding their journey is essential. Organizations have leveraged various techniques and technologies to support customer engagement, from call centres to chatbots and virtual agents. Recently, these systems have used Machine Learning (ML) and Natural Language Processing (NLP) to analyze large volumes of customer feedback and engagement data. The goal is to understand customers in context and provide meaningful answers across various channels. Despite multiple advances in Conversational Artificial Intelligence (AI) and Recommender Systems (RS), it is still challenging to understand the intent behind customer questions during the customer journey. To address this challenge, in this paper, we study and analyze the recent work in Conversational Recommender Systems (CRS) in general and, more specifically, in chatbot-based CRS. We introduce a pipeline to contextualize the input utterances in conversations. We then take the next step towards leveraging reverse feature engineering to link the contextualized input and learning model to support intent recognition. Since performance evaluation is achieved based on different ML models, we use transformer base models to evaluate the proposed approach using a labelled dialogue dataset (MSDialogue) of question-answering interactions between information seekers and answer providers.
['Sahar Moradizeyveh']
2022-12-06
null
null
null
null
['feature-engineering', 'intent-recognition']
['methodology', 'natural-language-processing']
[ 2.02053905e-01 4.71348494e-01 3.00774761e-02 -8.34852099e-01 -8.99930596e-01 -8.00932646e-01 8.71003985e-01 4.63769644e-01 -1.08957648e-01 3.10915828e-01 8.21770370e-01 -4.28843886e-01 -1.30272180e-01 -4.60916728e-01 2.00637147e-01 -4.26685117e-04 3.30913067e-01 7.89258718e-01 -1.18434012e-01 -8.42272222e-01 8.79505515e-01 3.05887252e-01 -1.25723839e+00 1.20156729e+00 6.35388911e-01 9.00628388e-01 2.39468634e-01 1.15530169e+00 -9.31957781e-01 1.44988048e+00 -5.19509137e-01 -6.26415968e-01 -9.85949337e-02 -4.68878865e-01 -1.49657845e+00 5.72259119e-03 -1.36153281e-01 -1.49693832e-01 1.51120201e-01 1.16679072e-01 2.79208958e-01 4.91073936e-01 3.43478441e-01 -1.26663864e+00 -6.78357899e-01 7.98601449e-01 1.73481867e-01 1.12939395e-01 9.86553133e-01 5.50167635e-02 1.33196640e+00 -9.17181015e-01 5.51687598e-01 1.41170442e+00 4.69041288e-01 5.47643185e-01 -9.31097388e-01 -1.19084239e-01 2.70787686e-01 4.36033994e-01 -5.81099808e-01 -5.56385040e-01 7.41915584e-01 -4.89505112e-01 1.39858115e+00 4.60404396e-01 2.67595828e-01 1.01140702e+00 -2.48408526e-01 1.00440788e+00 6.96952641e-01 -6.30127847e-01 2.08725017e-02 8.17094743e-01 8.08383942e-01 2.34685197e-01 -7.55788922e-01 -5.61912835e-01 -5.80706060e-01 -2.11854860e-01 1.11103311e-01 2.71530151e-01 8.34617913e-02 3.20844889e-01 -8.00215662e-01 1.25567424e+00 1.82090238e-01 5.65614104e-01 -5.58335245e-01 -3.91103953e-01 5.11516988e-01 6.10280037e-01 2.74172544e-01 9.16346729e-01 -4.85754430e-01 -8.30675662e-01 -1.30268574e-01 1.66266710e-01 1.84830165e+00 8.88442576e-01 7.06670046e-01 -6.21197045e-01 -3.43059778e-01 1.30758345e+00 4.96248782e-01 -1.65173158e-01 4.04102683e-01 -1.09408760e+00 6.66129172e-01 1.01926112e+00 2.74749726e-01 -1.14806461e+00 -5.44626713e-01 -5.09312823e-02 -2.70318270e-01 -5.45314908e-01 3.44236434e-01 -1.30467430e-01 7.33902853e-04 9.15355921e-01 2.41808698e-01 -1.63731217e-01 3.87205184e-01 7.24996150e-01 1.29266310e+00 6.34521723e-01 -1.15952387e-01 -1.44502193e-01 1.50625193e+00 -1.31920123e+00 -8.59488308e-01 -3.08228970e-01 8.66351604e-01 -1.06720066e+00 1.18161309e+00 3.95326972e-01 -7.75511622e-01 -4.45781827e-01 -4.63892490e-01 -2.80760407e-01 -3.59764695e-01 -1.62188128e-01 7.27949858e-01 5.17707884e-01 -6.49071455e-01 3.05685908e-01 -2.90889174e-01 -7.04917669e-01 4.90043052e-02 2.89225161e-01 -1.24593459e-01 -1.30422249e-01 -9.74417031e-01 9.61428404e-01 -2.88158208e-01 2.02694505e-01 -2.19952792e-01 -5.85636020e-01 -6.26714587e-01 1.77006889e-02 3.51909250e-01 -3.05708647e-01 1.91214907e+00 -6.95610642e-01 -2.00208688e+00 5.19429266e-01 -2.86611766e-01 -4.87374812e-01 3.24295573e-02 -4.52142388e-01 -3.56323093e-01 -1.03138342e-01 -2.10474670e-01 2.55801529e-01 1.83009073e-01 -1.09982908e+00 -1.00173020e+00 -3.26825917e-01 4.11006898e-01 2.35043168e-01 -1.98089838e-01 5.02903938e-01 -1.89661443e-01 3.57594222e-01 -2.42241565e-02 -8.50310445e-01 -1.83170304e-01 -1.06232643e+00 -3.52718890e-01 -6.28556669e-01 6.36807263e-01 -7.26018310e-01 1.10864568e+00 -1.92966950e+00 -1.12950236e-01 -6.32079616e-02 1.47016287e-01 2.42533684e-01 -1.62093461e-01 1.29981017e+00 4.10890996e-01 1.27927922e-02 3.21791917e-01 -6.51231050e-01 3.01476389e-01 3.16663682e-01 -4.90531951e-01 -2.32061416e-01 2.73143649e-01 9.81288791e-01 -8.07281375e-01 -3.26043487e-01 2.39787176e-01 3.99103463e-01 -8.67090583e-01 7.96580911e-01 -3.19863141e-01 7.15180516e-01 -5.68535566e-01 4.28142101e-01 1.30157292e-01 -2.15381995e-01 3.35082769e-01 -5.28971851e-02 -2.22179249e-01 7.40195453e-01 -7.98021853e-01 1.36278963e+00 -1.17658687e+00 7.38834620e-01 2.52600700e-01 -9.84934807e-01 1.25996184e+00 4.05132264e-01 3.24296296e-01 -8.42217863e-01 2.70596780e-02 1.59035012e-01 -4.98864949e-02 -1.05187821e+00 1.00505114e+00 1.51697099e-02 -2.38564566e-01 8.15523565e-01 -3.99099961e-02 -2.55588256e-02 -1.15703046e-01 3.32401901e-01 1.24948537e+00 -2.70873725e-01 2.86929369e-01 4.10385460e-01 8.53542507e-01 2.78499454e-01 1.24267302e-01 6.83755815e-01 -1.48393095e-01 4.38134402e-01 5.45153797e-01 -3.83588374e-01 -4.59746838e-01 -4.50058758e-01 2.97389358e-01 1.70524108e+00 -2.30890185e-01 -5.09480178e-01 -4.64898825e-01 -6.89072847e-01 -1.81324139e-01 1.09436011e+00 -2.90010452e-01 2.52784848e-01 -7.11820602e-01 -1.71442822e-01 2.41975129e-01 2.97885120e-01 2.13240728e-01 -1.34693658e+00 -3.85138839e-01 5.80392540e-01 -5.75608909e-01 -1.34323585e+00 -2.74453044e-01 -5.36800250e-02 -4.70341057e-01 -1.16864145e+00 3.75313475e-03 -6.55077338e-01 9.85832810e-02 4.47106719e-01 1.31267619e+00 2.86319554e-02 -1.69848099e-01 8.56346846e-01 -9.04356599e-01 -2.73623407e-01 -6.95944667e-01 3.71684402e-01 -3.41774672e-01 3.03214878e-01 6.92356288e-01 -5.27387619e-01 -6.17158711e-01 7.68884778e-01 -4.86587048e-01 -1.19162865e-01 3.67731899e-01 7.55836844e-01 -1.85939506e-01 -5.99383891e-01 1.08417559e+00 -1.22405815e+00 1.34700286e+00 -8.04846466e-01 4.04351391e-02 3.72908980e-01 -3.96991640e-01 -1.94443122e-01 5.95261097e-01 -8.35329518e-02 -1.34682453e+00 -2.06384599e-01 -5.83203614e-01 3.42407435e-01 -4.57549214e-01 6.35037839e-01 5.43937348e-02 3.46968651e-01 7.34420598e-01 -7.96107426e-02 -6.85688201e-03 -6.49983704e-01 6.45356417e-01 1.46338391e+00 3.19962084e-01 -3.47454935e-01 3.35089713e-02 1.32043391e-01 -7.01354206e-01 -8.44929993e-01 -1.01039886e+00 -1.23112154e+00 -4.89717543e-01 -4.55385804e-01 6.78087890e-01 -2.93663234e-01 -1.64633429e+00 -2.50979334e-01 -1.38558137e+00 -2.28395671e-01 -2.49313757e-01 2.96931088e-01 -5.55379391e-01 2.07330376e-01 -6.94453359e-01 -1.33876562e+00 -5.26962817e-01 -9.07886088e-01 7.24279523e-01 3.61884177e-01 -7.36923754e-01 -1.07980752e+00 2.10766584e-01 1.23743415e+00 7.56151855e-01 -1.33028179e-01 7.88243294e-01 -1.48463809e+00 -3.28553349e-01 -4.25669700e-01 -2.50746369e-01 2.83135325e-01 1.84436426e-01 -3.03985894e-01 -1.02487230e+00 2.58144200e-01 2.34628588e-01 -4.59149957e-01 2.77577102e-01 -1.60499215e-01 4.37799722e-01 -6.40672982e-01 -2.92243939e-02 -2.82734543e-01 8.34851801e-01 3.78177434e-01 4.86908019e-01 2.71960646e-01 4.27139580e-01 1.53769743e+00 9.88462865e-01 6.00417674e-01 9.87910867e-01 7.39744961e-01 2.84023583e-01 4.65864331e-01 3.03777963e-01 -1.59518585e-01 3.69746715e-01 1.07971871e+00 2.22624093e-01 -8.86433125e-02 -9.83252823e-01 4.12670553e-01 -2.22501540e+00 -9.70657766e-01 -3.60224128e-01 1.64904821e+00 6.16114140e-01 -1.37790546e-01 1.40818104e-01 -1.95630446e-01 3.32599431e-01 -1.68072477e-01 -2.93359309e-01 -1.17010427e+00 5.41328669e-01 -6.37006238e-02 -2.12868676e-01 8.17035139e-01 -4.84303385e-01 8.67361069e-01 5.36983013e+00 2.39285603e-01 -6.79864287e-01 2.39375025e-01 4.56794858e-01 1.83422267e-01 -3.59890938e-01 6.92529976e-02 -1.02064323e+00 2.56929640e-03 1.30034781e+00 4.80236346e-03 7.48265862e-01 8.89609993e-01 3.92910540e-01 -1.01295225e-01 -1.49464738e+00 9.84118223e-01 2.27237120e-01 -1.49606049e+00 -4.96661365e-01 -2.11244434e-01 2.25033313e-01 -3.18069803e-03 -1.43970236e-01 8.88083994e-01 4.59900588e-01 -1.12571454e+00 -4.70028892e-02 7.26883829e-01 -2.42294341e-01 -6.37385726e-01 1.02312303e+00 8.04954708e-01 -8.80450249e-01 -5.58478236e-01 2.84478012e-02 -3.81462216e-01 3.98830831e-01 7.06113055e-02 -1.81075704e+00 4.01348233e-01 4.10769701e-01 5.55590212e-01 -2.18285829e-01 5.24197698e-01 1.38000607e-01 4.15190905e-01 -5.40604219e-02 -2.94562697e-01 2.45164067e-01 -4.34449077e-01 1.68418556e-01 1.40288532e+00 -4.69453726e-03 4.03466582e-01 2.44974732e-01 5.80007434e-01 -4.22437303e-02 4.94084805e-01 -5.29489398e-01 -2.75704950e-01 4.58773971e-01 1.62085760e+00 -3.88811976e-01 2.19756290e-02 -6.06983602e-01 7.71951020e-01 2.10539430e-01 1.53102323e-01 -2.85362393e-01 -2.52069890e-01 6.59044623e-01 1.57011971e-02 2.88402662e-02 -1.10972486e-01 -1.02872260e-01 -7.20821023e-01 -2.10856616e-01 -1.23533499e+00 3.94370526e-01 -4.42490041e-01 -1.40057755e+00 7.17672229e-01 -4.65551406e-01 -7.52864242e-01 -7.51189947e-01 -2.71014333e-01 -7.88087130e-01 8.96844149e-01 -1.42680931e+00 -1.07363284e+00 -4.50798064e-01 3.19287002e-01 1.06826293e+00 -2.05446064e-01 9.96712387e-01 3.93462092e-01 -4.38990116e-01 3.27334642e-01 -2.30445832e-01 -3.80649380e-02 7.89279938e-01 -1.06806493e+00 2.59154409e-01 -1.79782555e-01 1.25547886e-01 8.24207962e-01 6.60260677e-01 -2.23693460e-01 -1.80477321e+00 -7.58234382e-01 1.26722348e+00 -9.05744135e-01 7.10053563e-01 -4.18922096e-01 -8.44233751e-01 7.34933376e-01 5.77950716e-01 -7.07458079e-01 1.32599473e+00 7.63882041e-01 -1.12798117e-01 2.39252113e-04 -1.15695608e+00 3.45125645e-01 5.98964870e-01 -7.99266636e-01 -8.31485093e-01 4.76010829e-01 9.02153611e-01 -9.97622013e-02 -1.12890911e+00 -9.01645571e-02 5.70984721e-01 -1.13620269e+00 7.41874933e-01 -9.58112180e-01 4.04556453e-01 2.26747334e-01 -4.24099803e-01 -1.15777779e+00 2.17646524e-01 -9.27196681e-01 3.07855196e-03 1.67857647e+00 5.56721389e-01 -6.00123644e-01 7.28261888e-01 1.12110054e+00 -3.38549495e-01 -6.83123887e-01 -3.47577453e-01 1.53799802e-02 -4.51664209e-01 -7.78461993e-01 5.53317070e-01 8.72120976e-01 7.35702455e-01 1.10030913e+00 -2.99505413e-01 -1.64191723e-01 -1.34248376e-01 3.94358844e-01 1.18851388e+00 -1.16406119e+00 -4.18696940e-01 -2.70923316e-01 5.55393435e-02 -1.32249951e+00 1.09074293e-02 -7.51217782e-01 1.61268115e-01 -1.85527706e+00 -1.19032875e-01 -4.89475340e-01 1.74899504e-01 1.07061930e-01 2.07152382e-01 -1.01426557e-01 2.68257439e-01 2.46496186e-01 -9.25126731e-01 2.31902257e-01 8.83602798e-01 7.45179877e-02 -7.80617297e-01 4.84189570e-01 -1.10457933e+00 6.80421650e-01 9.28639829e-01 -1.76680565e-01 -3.13472778e-01 -4.13001925e-02 5.69086254e-01 4.56338257e-01 -3.47760431e-02 -3.02251667e-01 5.69797635e-01 -1.08608350e-01 -4.89830762e-01 -7.63587058e-01 5.78713357e-01 -6.55528843e-01 -3.32042247e-01 -3.59479301e-02 -8.91728640e-01 -8.88420418e-02 -1.65045351e-01 4.16247010e-01 -4.36790824e-01 -4.83949661e-01 1.52836397e-01 -8.93005505e-02 -5.34090698e-01 -2.53346056e-01 -6.83510900e-01 7.52704665e-02 7.04885006e-01 -3.90741527e-02 -5.22806287e-01 -1.01811147e+00 -8.62350762e-01 5.93634725e-01 -6.56187460e-02 8.01203668e-01 6.33131027e-01 -7.81576753e-01 -6.92235410e-01 -1.50792256e-01 3.67474765e-01 -3.17130893e-01 3.43748033e-01 1.00138414e+00 -2.36586615e-01 9.47263062e-01 9.44910571e-02 -4.18822289e-01 -1.31400406e+00 1.33958369e-01 1.14304245e-01 -4.71792668e-01 -2.74141163e-01 8.63607705e-01 -1.77854598e-01 -1.00166178e+00 4.20863956e-01 -2.15744510e-01 -1.02586627e+00 5.60979843e-01 7.21529484e-01 4.94921088e-01 1.50683209e-01 -4.32858348e-01 -1.91493541e-01 -3.54270115e-02 -3.83535713e-01 -1.65607616e-01 1.45957923e+00 -4.55053002e-01 -7.00279027e-02 5.97677886e-01 1.30290127e+00 -5.67797795e-02 -6.26598716e-01 -4.01120216e-01 7.33992040e-01 -4.59448963e-01 -3.29033136e-01 -8.17433894e-01 -4.68978763e-01 9.93409097e-01 2.24628702e-01 9.76826608e-01 5.68112552e-01 2.97949046e-01 9.15714681e-01 8.90513957e-01 1.44993216e-01 -1.28672278e+00 4.78731990e-01 8.87067139e-01 1.07052433e+00 -1.54276109e+00 -4.99572337e-01 -4.79466051e-01 -1.23535466e+00 1.18774223e+00 4.00788784e-01 3.04328531e-01 5.31294048e-01 -6.24510050e-02 4.33922857e-01 -4.00379956e-01 -1.27238238e+00 -2.31493220e-01 -1.68951489e-02 4.42079961e-01 8.68997574e-01 -1.06988631e-01 -7.50401244e-02 1.00457847e+00 -2.41301179e-01 -7.06163868e-02 5.98381698e-01 9.49902534e-01 -5.14730513e-01 -1.47214437e+00 -1.53695896e-01 5.03955901e-01 -4.12696898e-01 -1.16811760e-01 -8.51697683e-01 3.00144017e-01 -4.84740496e-01 1.93255258e+00 -2.72802021e-02 -5.93847394e-01 7.46513724e-01 3.40127558e-01 -1.32175371e-01 -1.16223502e+00 -1.40408957e+00 -3.48834872e-01 9.51831818e-01 -6.14509404e-01 -4.92688835e-01 -7.04321742e-01 -1.31361496e+00 -2.57107973e-01 -4.39002663e-01 6.64065897e-01 1.07003963e+00 1.19638693e+00 5.98696232e-01 4.51639086e-01 1.04543316e+00 -5.43800235e-01 -4.25786406e-01 -1.13566184e+00 -1.15042403e-01 2.18297467e-01 6.06541894e-02 -1.41704530e-01 -2.56101668e-01 -2.22011834e-01]
[12.465458869934082, 7.894413471221924]
d8020847-529f-4fa4-a648-56e2c4ca0d5a
your-face-mirrors-your-deepest-beliefs
2112.12455
null
https://arxiv.org/abs/2112.12455v1
https://arxiv.org/pdf/2112.12455v1.pdf
Your Face Mirrors Your Deepest Beliefs-Predicting Personality and Morals through Facial Emotion Recognition
Can we really "read the mind in the eyes"? Moreover, can AI assist us in this task? This paper answers these two questions by introducing a machine learning system that predicts personality characteristics of individuals on the basis of their face. It does so by tracking the emotional response of the individual's face through facial emotion recognition (FER) while watching a series of 15 short videos of different genres. To calibrate the system, we invited 85 people to watch the videos, while their emotional responses were analyzed through their facial expression. At the same time, these individuals also took four well-validated surveys of personality characteristics and moral values: the revised NEO FFI personality inventory, the Haidt moral foundations test, the Schwartz personal value system, and the domain-specific risk-taking scale (DOSPERT). We found that personality characteristics and moral values of an individual can be predicted through their emotional response to the videos as shown in their face, with an accuracy of up to 86% using gradient-boosted trees. We also found that different personality characteristics are better predicted by different videos, in other words, there is no single video that will provide accurate predictions for all personality characteristics, but it is the response to the mix of different videos that allows for accurate prediction.
['T. Schaefer', 'L. Ripperger', 'M. F. Kaiser', 'C. Cetinkaya', 'E. Altuntas', 'A. Fronzetti Colladon', 'P. A. Gloor']
2021-12-23
null
null
null
null
['facial-emotion-recognition']
['computer-vision']
[-4.76679236e-01 2.18730614e-01 -1.65743575e-01 -8.49064529e-01 1.06590129e-02 -2.43997321e-01 1.92069858e-02 -3.64248276e-01 -3.88395309e-01 5.85614979e-01 1.76286206e-01 5.98488033e-01 -4.03080881e-02 -5.58589756e-01 -4.79283556e-02 -4.76369649e-01 -6.84317499e-02 3.27246517e-01 -4.23404187e-01 -4.76237804e-01 3.41955602e-01 3.21002543e-01 -1.59419310e+00 4.35488909e-01 5.74069202e-01 1.32633913e+00 -5.72958648e-01 6.22818232e-01 3.95446122e-01 1.13670146e+00 -5.22600889e-01 -1.20296025e+00 7.61599541e-02 -4.69237983e-01 -6.12591624e-01 1.88616276e-01 5.62644660e-01 -7.71661937e-01 -1.35989085e-01 1.01873815e+00 1.08177043e-01 1.98311999e-01 5.72656929e-01 -1.39068067e+00 -8.58594418e-01 -5.14577813e-02 -4.30616796e-01 -1.47464395e-01 9.07964051e-01 2.59984016e-01 8.04787219e-01 -5.66614211e-01 7.72064209e-01 1.58015954e+00 7.78627634e-01 1.01639771e+00 -1.13457727e+00 -7.33387649e-01 -2.23173395e-01 3.98397744e-01 -1.08871150e+00 -4.19931650e-01 9.53315496e-01 -7.74556100e-01 6.33939266e-01 1.43165886e-01 1.15797520e+00 1.38437808e+00 4.51402247e-01 2.73044497e-01 1.48790145e+00 -1.22814123e-02 5.95788890e-03 7.62611449e-01 1.33077934e-01 7.95748353e-01 -6.15004376e-02 1.58887357e-01 -5.44101357e-01 -3.19779634e-01 6.81292892e-01 -3.43715459e-01 2.02979639e-01 1.43263102e-01 -7.35462964e-01 8.04347456e-01 -2.43026819e-02 2.68835753e-01 -4.82871592e-01 -1.63868621e-01 5.43544233e-01 6.97293043e-01 4.58266586e-01 3.95109028e-01 -1.52520359e-01 -4.10963178e-01 -7.07820714e-01 2.50301003e-01 9.59144056e-01 3.37429196e-01 6.03338301e-01 1.31765857e-01 3.48180458e-02 7.88460672e-01 2.75959998e-01 4.12374496e-01 5.52397490e-01 -1.63645923e+00 -3.59063774e-01 5.79075873e-01 2.16536462e-01 -1.60985410e+00 -4.92561698e-01 5.55845648e-02 -4.19800729e-01 5.52164972e-01 6.25127792e-01 -5.49814105e-01 1.57397017e-02 1.76696396e+00 2.22315714e-01 -3.82775366e-01 -1.86702937e-01 9.91536915e-01 7.55718589e-01 2.84433395e-01 1.71739250e-01 -4.75623310e-01 1.43833590e+00 -3.94706547e-01 -7.95427203e-01 -1.40548289e-01 4.10391748e-01 -1.80177256e-01 8.71858597e-01 7.59866476e-01 -1.29877794e+00 -7.78004289e-01 -6.16286933e-01 2.96921849e-01 4.62899730e-03 2.04276770e-01 6.19103611e-01 1.06246543e+00 -1.16894031e+00 9.10239398e-01 -1.09352276e-01 -7.05884099e-01 3.14653069e-01 4.28342611e-01 -7.22145140e-01 3.14007819e-01 -1.22640073e+00 1.07861328e+00 -2.62465328e-01 -1.54439524e-01 -4.99582112e-01 -2.87029773e-01 -5.39921582e-01 9.95306745e-02 -3.07486746e-02 -5.26811242e-01 9.76876378e-01 -2.22017908e+00 -1.75889874e+00 1.34639812e+00 -2.64207602e-01 -1.17294841e-01 4.00716335e-01 4.17416357e-02 -6.60103559e-01 5.15090108e-01 -1.23704351e-01 8.19735229e-01 1.15524304e+00 -7.66780317e-01 -3.50077778e-01 -7.29123831e-01 5.26961051e-02 1.27772138e-01 -6.98483586e-01 6.97368920e-01 3.59353840e-01 -2.59958297e-01 -6.11844301e-01 -8.70444894e-01 2.85645664e-01 4.95920330e-03 1.64834052e-01 -4.15114403e-01 4.68443274e-01 -1.03822017e+00 9.34467316e-01 -2.36227441e+00 6.22840561e-02 2.42991835e-01 2.90401876e-01 -3.15800011e-02 2.85719354e-02 1.02885567e-01 -3.58155407e-02 2.87645519e-01 4.72154945e-01 -4.97965924e-02 1.62318200e-01 -8.79749805e-02 1.86576024e-01 4.44950372e-01 1.09701820e-01 5.70034027e-01 -7.11705744e-01 -7.64553070e-01 -1.52404204e-01 2.99269080e-01 -5.55228174e-01 2.59165823e-01 2.80953258e-01 3.75086635e-01 -6.19951822e-02 6.25746429e-01 3.47669572e-01 1.45666525e-01 3.81497711e-01 2.54812658e-01 2.61296660e-01 -4.09028530e-01 -4.76245850e-01 6.96047485e-01 1.16958559e-01 7.89615273e-01 2.97084063e-01 -7.59498954e-01 1.45934212e+00 4.50677186e-01 6.10558212e-01 -5.89946926e-01 3.28228503e-01 -2.23202690e-01 1.46629557e-01 -9.19764340e-01 2.78205395e-01 -6.76334202e-01 -1.42033100e-01 4.07746434e-01 -1.45102710e-01 2.69598603e-01 8.53722468e-02 -7.23374858e-02 8.18560600e-01 1.71492979e-01 2.02205256e-01 -5.97526133e-03 6.18170559e-01 -2.06339702e-01 8.66624951e-01 2.57595181e-01 -8.74297559e-01 2.36828835e-03 1.06022072e+00 -7.43440986e-01 -8.20558310e-01 -7.12405741e-01 -4.52733673e-02 1.40317047e+00 -2.23111063e-01 -2.04325810e-01 -1.06335258e+00 -5.16310513e-01 2.14016080e-01 7.53397822e-01 -8.47819805e-01 -2.67105043e-01 -3.53340432e-02 -4.07565564e-01 3.86078566e-01 3.39172065e-01 4.22191173e-01 -1.31156933e+00 -5.11557102e-01 -2.50949651e-01 -3.61498028e-01 -9.73610222e-01 -8.69328827e-02 -7.31792033e-01 -7.86878645e-01 -8.56162488e-01 -6.35212243e-01 -4.86535132e-01 5.50560772e-01 -2.79925078e-01 1.05621457e+00 3.44249271e-02 -4.45108349e-03 7.87124872e-01 -3.44964385e-01 -2.33815879e-01 -4.13815349e-01 -6.27285480e-01 6.18162453e-01 4.15961206e-01 8.19567680e-01 -5.21959543e-01 -3.55610847e-01 4.99109954e-01 -9.22641233e-02 -2.61516958e-01 3.49858671e-01 5.62611759e-01 -1.42632812e-01 1.38895974e-01 6.94256306e-01 -4.86194700e-01 6.85644329e-01 -5.86519659e-01 -9.60360747e-03 4.44236584e-02 -5.85636735e-01 -7.66312361e-01 6.61574841e-01 -4.35997784e-01 -1.36829674e+00 3.66922356e-02 -2.64907293e-02 -4.41706568e-01 -3.74195933e-01 1.60531364e-02 2.48419032e-01 -2.96452373e-01 5.67745388e-01 -1.17349267e-01 4.76268917e-01 -1.81661129e-01 -3.09516549e-01 8.10128152e-01 5.53289056e-01 -4.72101361e-01 4.62778836e-01 3.83105904e-01 7.59137375e-03 -1.06155050e+00 -6.41773522e-01 5.11425920e-02 -7.11293280e-01 -1.16598892e+00 9.74375844e-01 -7.88311839e-01 -1.70852971e+00 4.65578318e-01 -7.17048407e-01 -1.08055726e-01 4.70086858e-02 6.21157229e-01 -8.12355280e-01 5.05861759e-01 -7.70587921e-01 -1.22509217e+00 -1.98889837e-01 -6.64392769e-01 5.32509148e-01 4.28996354e-01 -9.46214139e-01 -8.93630981e-01 8.96694809e-02 7.87501872e-01 2.82250643e-01 3.10805678e-01 6.60807908e-01 -4.06699896e-01 2.66720384e-01 -4.67614979e-01 -1.29661888e-01 6.17057920e-01 -2.13050097e-01 3.06469172e-01 -9.01952922e-01 -1.40934020e-01 1.83223829e-01 -8.13676596e-01 3.47395301e-01 4.37897742e-01 8.80601406e-01 -2.89041281e-01 1.89161643e-01 4.30590868e-01 8.63047600e-01 4.04161066e-01 9.46481764e-01 1.77937105e-01 3.32002670e-01 1.36462665e+00 5.72026312e-01 6.59759998e-01 3.88756335e-01 6.24651432e-01 1.86414316e-01 4.62506443e-01 4.41647440e-01 -5.22821471e-02 1.18083704e+00 3.15415263e-01 -5.28935611e-01 3.07654679e-01 -4.91891176e-01 3.33075486e-02 -1.53192484e+00 -1.41286576e+00 -9.23226550e-02 2.07572865e+00 3.41577232e-01 -1.12667277e-01 9.49961066e-01 -1.15437485e-01 8.48806798e-01 -1.34844318e-01 -6.94132090e-01 -1.12407231e+00 -1.84783177e-03 -3.13300520e-01 -1.44936383e-01 2.31735229e-01 -8.04367483e-01 7.51055956e-01 7.41992092e+00 1.57568634e-01 -9.92427945e-01 -2.01819703e-01 1.11238706e+00 -4.23440576e-01 7.21240044e-02 -1.65922582e-01 -6.51253939e-01 5.44767022e-01 1.35058761e+00 -3.33150595e-01 6.39059722e-01 1.08389139e+00 4.21809047e-01 -2.41179124e-01 -8.10111642e-01 1.06208789e+00 4.53541309e-01 -4.88855541e-01 -5.83762348e-01 2.90884674e-01 1.62050977e-01 -8.00306380e-01 2.46263310e-01 4.38204736e-01 7.39922971e-02 -1.08705342e+00 5.87410986e-01 1.21154952e+00 7.40195334e-01 -6.23015106e-01 6.14084542e-01 9.49357972e-02 -4.65557873e-01 -5.19212961e-01 -4.81830746e-01 -7.12133765e-01 -5.05676903e-02 2.33247772e-01 -6.77911282e-01 -2.27321416e-01 9.30469990e-01 7.07159221e-01 -4.64848131e-01 3.91295791e-01 5.03139868e-02 4.13581371e-01 9.35841948e-02 -3.04249197e-01 -2.56702155e-01 -3.66746902e-01 4.49672818e-01 9.04074907e-01 3.97180706e-01 3.39138597e-01 -3.23237300e-01 8.19857597e-01 1.05858974e-01 3.10451835e-01 -5.42927980e-01 -3.06388497e-01 3.13215218e-02 1.72573411e+00 -2.10670888e-01 -3.00822884e-01 -4.65846449e-01 9.03247893e-01 3.05290103e-01 6.74270540e-02 -5.79039931e-01 -2.47039869e-02 7.88443506e-01 2.51358867e-01 -1.71993434e-01 2.88834512e-01 -2.03226045e-01 -1.17553687e+00 -2.24582195e-01 -8.57032359e-01 4.71048057e-01 -1.23986626e+00 -1.68377590e+00 3.15898538e-01 -2.25893065e-01 -8.76007557e-01 -4.44605380e-01 -6.66359246e-01 -6.34992599e-01 6.02244377e-01 -7.02315927e-01 -8.43069136e-01 -5.10019004e-01 5.47429919e-01 -3.57599370e-02 -4.02285010e-01 6.75375938e-01 -4.19178754e-02 -9.31543410e-01 6.34314716e-01 -3.17900240e-01 3.20735723e-01 1.15567398e+00 -1.13244796e+00 -3.02289039e-01 -5.38489176e-03 -6.50616229e-01 2.40719989e-01 6.62158549e-01 -7.70862579e-01 -1.23780394e+00 -4.72595274e-01 9.18526173e-01 -4.72133517e-01 5.54670691e-01 1.23095356e-01 -6.96050882e-01 7.51541495e-01 1.51983291e-01 -4.32759523e-01 1.12081838e+00 2.27152824e-01 -1.24087758e-01 -2.96350420e-01 -1.64408267e+00 4.08510208e-01 7.96334445e-01 -3.36300552e-01 -6.07055187e-01 1.96816579e-01 -1.17297538e-01 2.58886248e-01 -1.34133244e+00 -1.11251771e-01 1.33274257e+00 -1.69046283e+00 7.56669879e-01 -8.82591009e-01 7.94518411e-01 4.67565238e-01 -5.73535226e-02 -1.27978516e+00 -8.95991623e-01 -5.14866769e-01 1.91586152e-01 1.23532438e+00 -2.05887288e-01 -6.46870494e-01 9.66102660e-01 1.35503983e+00 5.00218272e-01 -6.78747714e-01 -6.57920301e-01 -5.32461762e-01 3.91797833e-02 -3.20901684e-02 3.21605116e-01 9.46935892e-01 8.05946171e-01 1.38163596e-01 -8.06098163e-01 -4.26129878e-01 6.92532301e-01 -2.80835003e-01 8.17761838e-01 -1.61841667e+00 -1.81578353e-01 -4.87797648e-01 -7.63230741e-01 3.37335467e-02 6.49074733e-01 -5.77131867e-01 -6.55659497e-01 -8.94771159e-01 4.77241337e-01 2.06418440e-01 1.47708550e-01 3.66731644e-01 5.59169240e-02 2.11854070e-01 2.85628349e-01 4.67669703e-02 -4.21312779e-01 2.95577288e-01 1.05837786e+00 3.71380419e-01 2.90942919e-02 -2.08329350e-01 -1.00985312e+00 1.16106737e+00 8.30717683e-01 -5.69073223e-02 4.08806615e-02 3.88699621e-01 4.59932506e-01 6.35035694e-01 5.80480099e-01 -9.41249847e-01 -2.03980669e-01 -4.11224127e-01 1.04682386e+00 1.18374638e-01 8.22816133e-01 -6.07399642e-01 1.72765926e-01 4.60116118e-01 -3.62147421e-01 5.29868603e-02 -6.70889243e-02 1.78659827e-01 7.77469128e-02 -4.04940456e-01 1.04710948e+00 -3.95823151e-01 -7.50285685e-01 -1.29287094e-01 -8.44291389e-01 -2.20960096e-01 1.30601454e+00 -3.91662121e-01 -3.91533762e-01 -1.14253640e+00 -9.72146094e-01 5.10117561e-02 8.08038294e-01 3.81241888e-02 5.75700879e-01 -1.30949152e+00 -7.03934014e-01 1.79941863e-01 -2.65984029e-01 -1.32619643e+00 6.35303319e-01 1.23645806e+00 -5.28990105e-02 5.31020761e-02 -1.08112049e+00 -7.91492909e-02 -1.69512880e+00 6.19092166e-01 5.39264500e-01 2.28876948e-01 -1.50329471e-01 7.22851396e-01 2.42081717e-01 1.38424188e-01 -3.95599715e-02 7.03151047e-01 -7.28468895e-01 5.30663669e-01 8.40483546e-01 8.25325131e-01 -5.79451203e-01 -1.22253156e+00 -1.50210679e-01 4.95942533e-01 -6.10641530e-03 -4.81560454e-02 1.15159035e+00 -2.65591532e-01 -2.63646871e-01 4.77483362e-01 9.75516975e-01 -5.43812290e-02 -9.88064289e-01 3.04695576e-01 -3.71853530e-01 -9.51821983e-01 -1.61023021e-01 -8.42726469e-01 -9.53144610e-01 7.76778579e-01 4.66113269e-01 4.47988212e-01 1.20872343e+00 -3.69895875e-01 8.21616471e-01 2.71403164e-01 2.51510680e-01 -1.53394175e+00 3.52779061e-01 9.51737538e-02 8.31782460e-01 -1.18106651e+00 -4.06283215e-02 -1.94955632e-01 -1.49144018e+00 1.30938780e+00 8.60518694e-01 -3.26293409e-01 3.40936154e-01 -2.76224256e-01 4.55495939e-02 -1.86551094e-01 -1.05964959e+00 2.08101258e-01 2.09710404e-01 7.86873937e-01 4.15406048e-01 3.41043800e-01 -2.84652889e-01 8.76648962e-01 -4.48872685e-01 4.73963320e-01 7.35501766e-01 2.95850366e-01 -6.53311312e-01 -8.23886752e-01 -7.14178264e-01 8.13487291e-01 -5.39921403e-01 6.68619215e-01 -1.11501706e+00 5.73892474e-01 6.43362328e-02 9.35165167e-01 1.15596212e-01 -9.13839042e-01 1.80770218e-01 5.19672871e-01 5.33789694e-01 -1.42516688e-01 -6.35775208e-01 -2.36202508e-01 4.49911416e-01 -6.89247191e-01 -4.34284478e-01 -1.26826084e+00 -7.46419787e-01 -9.23625946e-01 4.91709173e-01 -9.17564780e-02 1.50339797e-01 7.20998704e-01 2.00961098e-01 -1.69517934e-01 9.47089136e-01 -9.97031748e-01 -6.08075500e-01 -8.78237724e-01 -1.11338460e+00 6.83599770e-01 -5.69049641e-02 -7.71448791e-01 -6.61572576e-01 2.58023236e-02]
[13.364453315734863, 2.0258219242095947]
b32d1b11-9445-499c-89b6-e2d6f469728a
core-a-knowledge-graph-entity-type-prediction
2112.10067
null
https://arxiv.org/abs/2112.10067v1
https://arxiv.org/pdf/2112.10067v1.pdf
CORE: A Knowledge Graph Entity Type Prediction Method via Complex Space Regression and Embedding
Entity type prediction is an important problem in knowledge graph (KG) research. A new KG entity type prediction method, named CORE (COmplex space Regression and Embedding), is proposed in this work. The proposed CORE method leverages the expressive power of two complex space embedding models; namely, RotatE and ComplEx models. It embeds entities and types in two different complex spaces using either RotatE or ComplEx. Then, we derive a complex regression model to link these two spaces. Finally, a mechanism to optimize embedding and regression parameters jointly is introduced. Experiments show that CORE outperforms benchmarking methods on representative KG entity type inference datasets. Strengths and weaknesses of various entity type prediction methods are analyzed.
['C. -C. Jay Kuo', 'Bin Wang', 'Yun-Cheng Wang', 'Xiou Ge']
2021-12-19
null
null
null
null
['type-prediction']
['computer-code']
[-3.82163912e-01 3.57007653e-01 -7.88811862e-01 -3.17800760e-01 1.07086837e-01 -5.56601584e-01 5.20939231e-01 4.75452423e-01 -3.76084089e-01 6.88783526e-01 3.32438082e-01 -2.34375596e-01 -4.88733917e-01 -1.40669572e+00 -5.92592239e-01 -3.67328227e-01 -1.01406731e-01 2.65292674e-01 9.58434716e-02 -2.39059448e-01 5.16461767e-02 1.22027457e-01 -1.37113750e+00 7.89789110e-03 1.11606979e+00 1.08181000e+00 -2.18321711e-01 4.30593759e-01 -3.67891461e-01 1.02715254e+00 -1.98825106e-01 -8.03242505e-01 2.32584670e-01 1.82428136e-01 -9.62851942e-01 -8.77499104e-01 4.36043978e-01 5.63756377e-02 -5.08600235e-01 7.59906709e-01 5.51186144e-01 1.11842543e-01 8.82646680e-01 -1.88266826e+00 -1.01006544e+00 8.70746434e-01 -1.51334003e-01 4.53862036e-03 5.31098425e-01 -3.60996008e-01 1.57640302e+00 -1.08444393e+00 8.91707897e-01 1.10401320e+00 1.25108135e+00 3.30987304e-01 -1.26802158e+00 -4.00219202e-01 9.36919749e-02 6.36848330e-01 -1.62798572e+00 -6.60349503e-02 9.03909326e-01 -6.14004493e-01 1.13010502e+00 2.70948678e-01 9.48011041e-01 7.21961498e-01 -3.51896621e-02 1.02731264e+00 9.11426723e-01 -2.76433736e-01 1.80518091e-01 4.77650583e-01 4.61770564e-01 9.32561040e-01 9.62047219e-01 -9.85855609e-02 -3.10014486e-01 -4.17483389e-01 4.43001777e-01 -5.98967932e-02 -2.18025774e-01 -8.53691638e-01 -1.37359905e+00 8.54930937e-01 7.70849049e-01 8.32214952e-02 -2.52331674e-01 1.97352245e-01 3.32241893e-01 2.46289983e-01 4.81763273e-01 7.69412994e-01 -7.98178852e-01 2.34961491e-02 -4.68103319e-01 4.09225285e-01 1.21732843e+00 1.39647651e+00 8.28787386e-01 -8.17177910e-03 -2.10434228e-01 8.81553292e-01 7.77917504e-01 5.07230520e-01 2.34311700e-01 -2.72721916e-01 5.34727335e-01 1.37390447e+00 -1.95308968e-01 -1.36233830e+00 -7.08697200e-01 -1.90342903e-01 -6.65512741e-01 -4.80448008e-01 -1.56882823e-01 -6.08631447e-02 -4.24741536e-01 1.54781449e+00 8.52437854e-01 6.34049237e-01 2.52834469e-01 4.69119459e-01 1.39322412e+00 3.53849560e-01 3.40416670e-01 2.61866450e-01 1.37223327e+00 -9.95952070e-01 -8.10567677e-01 1.39493302e-01 1.22463787e+00 -1.59467056e-01 6.62910938e-01 -3.58936012e-01 -9.71372485e-01 -2.01320648e-01 -1.09802210e+00 -4.06360775e-01 -1.20159817e+00 2.47528076e-01 9.60686266e-01 8.62403691e-01 -7.96816349e-01 6.85883999e-01 -5.69547236e-01 -2.04916090e-01 4.10787076e-01 4.59745169e-01 -4.32646692e-01 9.52338502e-02 -1.70508969e+00 9.04154360e-01 6.73184991e-01 7.65845627e-02 2.45262265e-01 -1.33367980e+00 -1.26323450e+00 1.23550892e-01 2.97664136e-01 -1.43829572e+00 6.64030790e-01 -6.12269975e-02 -1.13706279e+00 5.36846280e-01 5.31195104e-02 -3.43589753e-01 1.54292107e-01 -3.43718201e-01 -6.80321753e-01 -9.96059254e-02 5.87749854e-03 3.51203024e-01 5.26425958e-01 -1.18599558e+00 -7.11064041e-01 -4.66682434e-01 8.21334198e-02 2.06686363e-01 -7.04578400e-01 -5.39352238e-01 -3.81376773e-01 -8.61670196e-01 2.01129243e-02 -9.02101815e-01 1.25443498e-02 -3.45518053e-01 -6.44647717e-01 -6.92653775e-01 7.36239851e-01 -6.16626382e-01 1.94025147e+00 -1.75972271e+00 5.09099603e-01 4.73310918e-01 7.76522160e-01 -4.58426550e-02 7.31183961e-02 4.66820270e-01 -1.21074565e-01 4.99821335e-01 1.14811450e-01 -1.08651333e-01 6.35397971e-01 2.80239612e-01 -7.69332051e-02 3.31767857e-01 9.29989144e-02 1.40320742e+00 -1.02480412e+00 -7.14060962e-01 1.07939072e-01 3.57406944e-01 -8.93495440e-01 4.74620312e-02 -1.95838213e-02 -5.84578693e-01 -5.99190593e-01 9.80006278e-01 7.82587171e-01 -3.38033646e-01 4.71671730e-01 -8.53561223e-01 -4.48802933e-02 2.86300421e-01 -1.39798963e+00 1.21300018e+00 -2.59814888e-01 1.67665422e-01 -5.95332623e-01 -1.00111187e+00 7.67523050e-01 1.68661475e-01 7.54070759e-01 -2.48269305e-01 -1.70522079e-01 2.27630839e-01 -3.47010672e-01 -5.25900245e-01 1.02641046e+00 2.26933613e-01 -3.73073369e-01 1.16254747e-01 3.09955478e-01 1.95885703e-01 3.09381366e-01 3.03907573e-01 1.09075940e+00 1.32293701e-01 3.84985417e-01 -2.64929712e-01 4.58489954e-01 -1.95985913e-01 9.73225415e-01 5.69150090e-01 -1.64227724e-01 -1.34671345e-01 6.81598783e-01 -5.14949739e-01 -9.09436285e-01 -1.12306368e+00 -1.98188514e-01 9.14318979e-01 4.45453376e-01 -1.12893772e+00 -1.77860573e-01 -1.23347116e+00 7.16963470e-01 5.47594070e-01 -8.39226782e-01 -3.44193459e-01 -4.61703211e-01 -7.58457243e-01 7.99850941e-01 8.54961872e-01 4.77681339e-01 -7.96709299e-01 3.01794618e-01 3.34716178e-02 -1.84034437e-01 -1.12222135e+00 -9.91552696e-02 -9.17776302e-02 -7.21552670e-01 -1.32849205e+00 -1.94939181e-01 -8.56234312e-01 5.12423515e-01 4.33678627e-02 1.18325198e+00 2.61940122e-01 -6.63271248e-02 6.13250256e-01 -6.33859873e-01 -2.60762900e-01 8.23587999e-02 6.14104569e-01 2.81142145e-01 -1.17800429e-01 4.68952477e-01 -4.60150540e-01 -5.66386402e-01 1.59869567e-01 -6.02635503e-01 2.87564807e-02 8.54507685e-01 9.66661334e-01 5.30949473e-01 6.81277290e-02 5.32359421e-01 -1.34407163e+00 4.87570822e-01 -9.95052993e-01 -3.48029405e-01 5.86561441e-01 -1.27419233e+00 3.74326557e-01 5.33360720e-01 -1.19163796e-01 -7.71391153e-01 -3.76000345e-01 1.09465063e-01 -4.19794828e-01 2.48224452e-01 1.00849104e+00 -3.92701507e-01 -1.51160806e-01 2.27532506e-01 2.50478033e-02 -4.72348541e-01 -4.37866449e-01 7.31099844e-01 4.59895134e-01 2.19926015e-01 -6.12270236e-01 1.24864852e+00 1.17175914e-01 9.09752920e-02 -8.00948203e-01 -5.57067633e-01 -6.64572656e-01 -8.87938857e-01 1.28536448e-01 7.09486663e-01 -8.93863559e-01 -7.39108622e-01 2.57854968e-01 -7.32101560e-01 -1.36275783e-01 -3.14564824e-01 4.78892267e-01 -5.07503986e-01 3.67614031e-01 -4.41909701e-01 -3.26963812e-01 -2.97666311e-01 -6.73254132e-01 1.02412891e+00 1.95342436e-01 5.15202759e-03 -1.50288749e+00 2.52348095e-01 3.39344829e-01 1.89803958e-01 3.17993879e-01 1.11447859e+00 -7.42779255e-01 -7.14604318e-01 -1.30849034e-01 -3.66014481e-01 -2.14844257e-01 5.98445013e-02 6.61390200e-02 -4.53505009e-01 -8.17037225e-02 -9.58946466e-01 3.20055634e-02 8.86515796e-01 8.01441912e-03 1.02365589e+00 -7.53142834e-01 -7.73666143e-01 1.12620604e+00 1.56778848e+00 -2.84948736e-01 7.53761530e-01 6.60310149e-01 1.34201276e+00 3.11754167e-01 7.38520324e-01 4.52030867e-01 1.25317287e+00 9.67962384e-01 -1.40368054e-02 4.63676378e-02 1.25423878e-01 -6.86013162e-01 1.84289649e-01 1.21030331e+00 -3.63028347e-01 -7.41915852e-02 -9.49199796e-01 4.58042502e-01 -1.98227000e+00 -1.27232182e+00 -3.19028974e-01 1.71672761e+00 9.03605998e-01 -3.16685438e-01 2.32781380e-01 1.47300303e-01 4.52640414e-01 1.39183044e-01 -3.94887149e-01 -1.91897437e-01 -2.22039476e-01 1.39392257e-01 8.10494781e-01 1.38381168e-01 -1.38345027e+00 1.14674997e+00 6.15904284e+00 7.14708567e-01 -5.74949861e-01 -1.48443401e-01 -1.91497639e-01 4.24743414e-01 -7.22628295e-01 2.76855558e-01 -1.00612664e+00 6.23990715e-01 5.81242025e-01 -5.50884545e-01 2.45199040e-01 1.13721514e+00 -4.25467730e-01 4.72457319e-01 -1.04116309e+00 9.54200089e-01 -1.00074954e-01 -1.61036491e+00 1.67786792e-01 1.32676244e-01 5.30780673e-01 -2.89153486e-01 1.07104601e-02 1.03849721e+00 4.77919847e-01 -8.87759924e-01 3.73771518e-01 5.94618440e-01 6.67606235e-01 -6.87506378e-01 7.22036481e-01 -1.94725201e-01 -1.79239500e+00 -1.72256261e-01 -3.71852487e-01 4.11545962e-01 -1.12356044e-01 4.86908972e-01 -8.84240508e-01 1.12338281e+00 6.99195266e-01 1.28049648e+00 -9.02440250e-01 8.72287333e-01 -3.90217006e-01 4.28490460e-01 -1.21425599e-01 -1.39029995e-01 -4.35521454e-01 -2.29974657e-01 5.59299588e-01 1.33979332e+00 1.31556571e-01 1.65572226e-01 -1.95649564e-02 9.25284982e-01 -3.26421320e-01 1.55127019e-01 -7.19916165e-01 -1.66731939e-01 1.13944829e+00 1.57266879e+00 -8.65980461e-02 -1.96111381e-01 -6.89618587e-01 7.13527739e-01 6.76638126e-01 2.99264073e-01 -9.34679508e-01 -6.94454253e-01 8.40426922e-01 -2.46500269e-01 5.60778201e-01 -3.05944085e-02 -2.13695407e-01 -1.68386388e+00 1.44450823e-02 -4.69457358e-01 8.72280240e-01 -4.31752205e-01 -1.46421051e+00 -7.22006783e-02 1.69294074e-01 -1.11193752e+00 -4.99641672e-02 -8.04622352e-01 -3.73383403e-01 3.72327179e-01 -1.87578118e+00 -1.58114934e+00 -4.41641808e-01 5.23172379e-01 -2.83056319e-01 -3.33687186e-01 7.81733274e-01 4.87516642e-01 -1.06937444e+00 1.18543780e+00 2.04213664e-01 4.45137054e-01 3.41919720e-01 -1.93892694e+00 3.52493405e-01 3.22133332e-01 1.25615180e-01 9.30336833e-01 1.84630841e-01 -7.31876254e-01 -2.07201195e+00 -1.33018780e+00 1.09317160e+00 -6.37115717e-01 7.96446979e-01 -2.84062624e-01 -7.28481650e-01 9.93054092e-01 -2.94357836e-01 2.79635310e-01 1.15733588e+00 7.58815765e-01 -7.19625592e-01 -1.06668413e-01 -1.07422972e+00 5.71470380e-01 1.18552220e+00 -4.73911792e-01 -7.37447202e-01 -6.35090694e-02 8.53692532e-01 -2.96897739e-01 -1.94799829e+00 8.08598638e-01 8.57317030e-01 -4.24806178e-01 1.44251990e+00 -8.59000027e-01 2.79634327e-01 -2.59180784e-01 -2.04474613e-01 -1.56217849e+00 -7.03994036e-01 -3.42121661e-01 -1.13511896e+00 1.28651690e+00 6.05568707e-01 -8.61900568e-01 9.28151548e-01 5.42925656e-01 -3.24006453e-02 -1.13866246e+00 -5.17068565e-01 -9.16605532e-01 5.64450920e-02 -3.58645059e-02 9.87065732e-01 1.70762527e+00 3.28593463e-01 4.96969163e-01 3.80353630e-02 2.93162942e-01 5.39774060e-01 2.87328571e-01 1.03529024e+00 -1.50652194e+00 -6.06982224e-02 -5.90106130e-01 -1.04302692e+00 -6.45822883e-01 3.04134190e-01 -1.40659869e+00 -8.60589683e-01 -1.71033740e+00 3.45536880e-02 -1.11041343e+00 -6.38229370e-01 6.67163193e-01 -6.23938203e-01 -1.96310699e-01 1.88976765e-01 2.29695484e-01 -5.82034409e-01 9.55414355e-01 6.85073376e-01 -2.80648232e-01 -2.31492430e-01 -2.16439411e-01 -7.69447803e-01 5.82103729e-01 6.83335066e-01 -1.26171798e-01 -3.32783908e-01 -1.15777642e-01 8.99666071e-01 -3.94151390e-01 2.62477487e-01 -6.98787093e-01 1.81288943e-01 -1.46527231e-01 3.86533499e-01 -7.85484493e-01 1.70449957e-01 -7.42756784e-01 1.55480221e-01 1.60055712e-01 -1.68315947e-01 1.81543514e-01 -6.14822321e-02 7.37983048e-01 -6.44341186e-02 -5.51773943e-02 1.82316780e-01 4.78915334e-01 -1.25747991e+00 7.28197575e-01 5.24399936e-01 1.77149400e-01 1.20633507e+00 -3.30438673e-01 -7.23030090e-01 2.18503118e-01 -7.19105005e-01 6.12334967e-01 3.94080698e-01 7.21999586e-01 7.41000831e-01 -2.03860378e+00 -5.25032282e-01 5.04370918e-03 5.04452765e-01 -9.64413583e-02 2.07224250e-01 1.16301799e+00 -4.26635683e-01 3.36095095e-01 1.63529366e-01 -2.52121031e-01 -1.17382526e+00 7.12645710e-01 1.59631684e-01 -5.40543675e-01 -5.30857205e-01 5.96804321e-01 -1.06563851e-01 -1.06081522e+00 -1.16012029e-01 -2.21235499e-01 -7.12358117e-01 3.41420919e-01 6.17814362e-02 7.92221546e-01 -1.67398117e-02 -5.60869694e-01 -4.10759538e-01 4.48720723e-01 -9.01546031e-02 4.12438631e-01 1.62367630e+00 5.02444990e-02 -2.73411125e-01 4.42392677e-01 1.36047530e+00 2.40290731e-01 -4.70792055e-01 -3.17083687e-01 3.66334647e-01 -4.18391913e-01 4.65723947e-02 -4.88036871e-01 -9.36080456e-01 1.71384215e-01 2.48569876e-01 3.36735845e-01 8.45361769e-01 -3.74108888e-02 8.69679570e-01 6.17442191e-01 2.04806149e-01 -1.40942025e+00 -1.78931922e-01 5.03609538e-01 4.97365773e-01 -1.01745355e+00 3.85724843e-01 -8.15216303e-01 -4.12991017e-01 8.46073389e-01 7.48834133e-01 8.34805612e-03 8.82673919e-01 9.33518540e-03 -5.45731068e-01 -5.53390503e-01 -9.43708658e-01 -2.89327383e-01 7.49802411e-01 6.84355974e-01 4.30491477e-01 3.27116579e-01 -3.95656526e-01 9.57111359e-01 -4.51160014e-01 -1.49636671e-01 2.50644982e-01 1.03291929e+00 -3.88008654e-02 -1.18156421e+00 7.36249015e-02 8.44553649e-01 -3.19501273e-02 -1.72427207e-01 -4.84058827e-01 1.06761861e+00 2.64171094e-01 4.66146499e-01 -2.61027008e-01 -8.39497089e-01 3.81893426e-01 7.46292546e-02 5.15679181e-01 -4.79221284e-01 -4.65016633e-01 -7.74059236e-01 2.22261280e-01 -5.83373010e-01 -2.80714035e-01 -5.17650187e-01 -1.01590598e+00 -5.56200564e-01 -7.49653220e-01 5.09656556e-02 5.17242134e-01 6.00085139e-01 7.68919528e-01 3.95073265e-01 8.28468800e-01 -1.33650050e-01 -4.28812027e-01 -7.55904257e-01 -7.54770517e-01 5.55799007e-01 -1.79993547e-02 -1.24604774e+00 -2.30920315e-01 -8.76803100e-02]
[8.741496086120605, 7.9024529457092285]
3b295faa-5221-4210-a8ec-daeb6fce980c
fusion-for-visual-infrared-person-reid-in
2305.0032
null
https://arxiv.org/abs/2305.00320v1
https://arxiv.org/pdf/2305.00320v1.pdf
Fusion for Visual-Infrared Person ReID in Real-World Surveillance Using Corrupted Multimodal Data
Visible-infrared person re-identification (V-I ReID) seeks to match images of individuals captured over a distributed network of RGB and IR cameras. The task is challenging due to the significant differences between V and I modalities, especially under real-world conditions, where images are corrupted by, e.g, blur, noise, and weather. Indeed, state-of-art V-I ReID models cannot leverage corrupted modality information to sustain a high level of accuracy. In this paper, we propose an efficient model for multimodal V-I ReID -- named Multimodal Middle Stream Fusion (MMSF) -- that preserves modality-specific knowledge for improved robustness to corrupted multimodal images. In addition, three state-of-art attention-based multimodal fusion models are adapted to address corrupted multimodal data in V-I ReID, allowing to dynamically balance each modality importance. Recently, evaluation protocols have been proposed to assess the robustness of ReID models under challenging real-world scenarios. However, these protocols are limited to unimodal V settings. For realistic evaluation of multimodal (and cross-modal) V-I person ReID models, we propose new challenging corrupted datasets for scenarios where V and I cameras are co-located (CL) and not co-located (NCL). Finally, the benefits of our Masking and Local Multimodal Data Augmentation (ML-MDA) strategy are explored to improve the robustness of ReID models to multimodal corruption. Our experiments on clean and corrupted versions of the SYSU-MM01, RegDB, and ThermalWORLD datasets indicate the multimodal V-I ReID models that are more likely to perform well in real-world operational conditions. In particular, our ML-MDA is an important strategy for a V-I person ReID system to sustain high accuracy and robustness when processing corrupted multimodal images. Also, our multimodal ReID model MMSF outperforms every method under CL and NCL camera scenarios.
['Eric Granger', 'Rafael M. O. Cruz', 'Mahdi Alehdaghi', 'Arthur Josi']
2023-04-29
null
null
null
null
['person-re-identification']
['computer-vision']
[ 1.81480907e-02 -6.21053934e-01 3.20962995e-01 -2.88408369e-01 -7.54138052e-01 -7.07472622e-01 6.41482115e-01 -1.42751843e-01 -5.76004565e-01 5.00582576e-01 2.13083863e-01 1.20366417e-01 -8.06598663e-02 -2.94643909e-01 -6.92682743e-01 -8.12502265e-01 1.10313781e-01 1.30519137e-01 -3.20360005e-01 -3.50960255e-01 -1.86628819e-01 5.98133326e-01 -1.95265722e+00 2.58389533e-01 6.60474896e-01 6.96481824e-01 -8.51785541e-02 8.37903798e-01 1.03928402e-01 5.08137941e-01 -8.08355153e-01 -9.07160521e-01 5.02043605e-01 -2.35551596e-01 -4.88557607e-01 -1.82881683e-01 8.16959083e-01 -3.50220114e-01 -5.12893021e-01 8.75883639e-01 9.28502440e-01 1.13824345e-01 6.38420165e-01 -1.77244937e+00 -7.14935780e-01 2.93620944e-01 -5.97638905e-01 5.11281230e-02 9.51708198e-01 6.12176538e-01 1.78217873e-01 -6.82989120e-01 3.63782048e-01 1.57659364e+00 9.15015638e-01 7.90027916e-01 -1.23476398e+00 -8.16284955e-01 1.46060780e-01 2.87951589e-01 -1.70922959e+00 -6.57224655e-01 5.35430849e-01 -2.96767682e-01 8.11619222e-01 5.32127261e-01 3.69523138e-01 1.57771385e+00 1.24972565e-02 5.46478450e-01 1.06385601e+00 -1.54796481e-01 2.89920606e-02 3.08384508e-01 2.40348279e-01 8.86634812e-02 3.80805582e-01 1.90569267e-01 -8.52145731e-01 -2.52124310e-01 3.31512690e-01 -1.05753948e-03 -4.01214242e-01 5.68277650e-02 -1.30691302e+00 2.07982212e-01 3.19062442e-01 2.80260965e-02 -4.07057583e-01 1.28793856e-02 2.65313208e-01 2.84315377e-01 2.42172573e-02 6.15246408e-02 -3.34055163e-02 1.07413894e-02 -7.04360723e-01 2.71931350e-01 4.60654974e-01 1.08239162e+00 5.55979848e-01 1.74172595e-02 -3.77861500e-01 7.21205294e-01 3.77814919e-01 1.14077044e+00 6.47730380e-02 -7.12302744e-01 6.30129039e-01 4.93151158e-01 3.48169833e-01 -8.02297831e-01 -4.06971425e-01 -2.42963836e-01 -1.07095850e+00 1.08948365e-01 3.50852579e-01 -6.96259588e-02 -8.91616344e-01 1.86213994e+00 3.43422264e-01 2.78677583e-01 4.37514484e-01 1.04221225e+00 1.30286014e+00 3.65216136e-01 3.99480313e-01 8.05563480e-02 1.29096115e+00 -1.99044779e-01 -7.16930032e-01 -1.63881376e-01 1.06066391e-01 -7.14705467e-01 7.67257988e-01 3.87136899e-02 -1.08090568e+00 -8.41616750e-01 -7.24056542e-01 2.20247999e-01 -6.00342095e-01 4.00192700e-02 9.17779431e-02 1.05198884e+00 -1.35535681e+00 -1.48188889e-01 -2.48068556e-01 -8.14091742e-01 8.89561698e-02 5.71414053e-01 -9.65605736e-01 -3.65353078e-01 -1.16756785e+00 8.40481579e-01 6.11605793e-02 6.03989422e-01 -9.02188241e-01 -5.73065341e-01 -1.05239236e+00 -2.33683169e-01 -1.05019823e-01 -1.03353190e+00 7.36162543e-01 -8.29847157e-01 -1.08943713e+00 8.83581400e-01 -3.71061295e-01 -3.78304012e-02 6.08303130e-01 -7.02248886e-03 -6.76601052e-01 1.77522495e-01 -1.11473568e-01 8.76914203e-01 8.94128203e-01 -2.09987092e+00 -3.71509463e-01 -5.78401446e-01 6.86810538e-02 4.03758973e-01 -3.06484759e-01 3.30762655e-01 -6.21442616e-01 -3.36654216e-01 -2.49218509e-01 -9.17687654e-01 2.73995519e-01 -3.43143314e-01 -4.20195043e-01 8.07265267e-02 8.06563020e-01 -7.17384100e-01 6.65575683e-01 -2.23785830e+00 2.02249922e-02 2.79804200e-01 -7.32332021e-02 4.26170021e-01 -5.32093346e-01 3.88152272e-01 -1.28448769e-01 2.51579046e-01 2.09146403e-02 -9.76431370e-01 1.14772514e-01 2.87231445e-01 6.44386485e-02 7.94793785e-01 2.30691186e-03 9.32905316e-01 -4.90140587e-01 -4.15093631e-01 4.27244335e-01 8.95856023e-01 -4.91626635e-02 5.08061767e-01 4.67755884e-01 8.62730503e-01 1.68217629e-01 1.04762244e+00 1.15207469e+00 3.77235413e-01 -2.05095351e-01 -5.84740222e-01 -4.27279025e-02 -6.77997231e-01 -1.29988766e+00 1.34113288e+00 -2.65677840e-01 3.72294128e-01 5.32918751e-01 -3.96887273e-01 7.28427589e-01 3.29477817e-01 3.31937850e-01 -9.11813378e-01 3.34416240e-01 -2.35982314e-01 -4.00552005e-01 -6.51230574e-01 8.57983410e-01 2.63617754e-01 -1.80719152e-01 3.55760381e-02 -5.92000559e-02 4.03118163e-01 -6.39358461e-02 3.12767804e-01 7.71551490e-01 -5.53010926e-02 -3.66001815e-01 2.46493012e-01 7.23794639e-01 -2.29709610e-01 5.23132324e-01 1.19177556e+00 -5.19352317e-01 9.31042254e-01 -7.02635124e-02 -8.47707242e-02 -8.31244349e-01 -1.27763951e+00 -6.21356480e-02 8.33101034e-01 7.85083830e-01 8.06277022e-02 -7.02041328e-01 -3.05834234e-01 1.39403954e-01 4.92740661e-01 -5.97339749e-01 -1.61348805e-01 -4.32421744e-01 -9.06938553e-01 1.14087784e+00 4.19090509e-01 7.65072167e-01 -6.92872524e-01 -2.66327560e-01 -3.34641039e-01 -5.75356305e-01 -1.37533116e+00 -4.62482095e-01 -5.03809869e-01 -1.76691756e-01 -1.24902058e+00 -9.87305641e-01 -4.54018444e-01 6.58699512e-01 7.26623058e-01 9.08076167e-01 1.40743658e-01 -2.44091943e-01 1.31800580e+00 -4.66205478e-01 -2.54044712e-01 -2.40029067e-01 -3.75865161e-01 4.11512822e-01 4.71439213e-01 3.08621228e-01 2.39343084e-02 -6.24406159e-01 7.77692139e-01 -9.88572359e-01 -3.65238219e-01 4.13448364e-01 7.32428551e-01 2.77088642e-01 -1.42672509e-01 3.30126852e-01 -6.69383407e-02 4.81039375e-01 -4.90021944e-01 -3.24081659e-01 6.57734513e-01 -1.68675199e-01 -4.05708700e-01 2.00887457e-01 -6.64174438e-01 -1.35842478e+00 -7.79667543e-03 1.75017506e-01 -5.87796688e-01 -6.44996405e-01 2.47433633e-01 -6.32170856e-01 -2.87379622e-01 4.72193897e-01 2.36815229e-01 -2.99616922e-02 -3.29688162e-01 3.18662196e-01 9.95972991e-01 1.21603537e+00 -6.10440731e-01 1.02746689e+00 5.71977735e-01 -1.05109476e-01 -9.45180893e-01 1.51762709e-01 -5.99760115e-01 -5.58016062e-01 -6.12121642e-01 1.03284514e+00 -1.30678797e+00 -1.23134065e+00 1.14868867e+00 -1.21142435e+00 -8.14280659e-02 2.78301209e-01 3.90732944e-01 -1.43834669e-02 7.34018505e-01 -4.60095137e-01 -1.28001177e+00 -3.82310420e-01 -1.24527538e+00 1.26189494e+00 6.05293572e-01 7.69070610e-02 -6.96571112e-01 -1.27917424e-01 9.18023229e-01 5.35996020e-01 2.45773211e-01 3.00681502e-01 -3.81889790e-01 -3.50046188e-01 -2.96779633e-01 -4.12185341e-01 1.16896428e-01 5.25025502e-02 -4.16625403e-02 -1.45507336e+00 -6.90877318e-01 -5.21486998e-01 -1.54101431e-01 6.43427432e-01 1.68835253e-01 6.53164029e-01 -5.69773465e-02 -2.14352876e-01 8.03791225e-01 1.20244896e+00 1.66698322e-02 8.06519866e-01 2.95808703e-01 8.27939808e-01 7.25341022e-01 4.79296863e-01 3.63543093e-01 8.48268628e-01 7.84774303e-01 6.59427822e-01 -4.39279288e-01 -2.64853030e-01 -3.91465835e-02 6.12139583e-01 1.80693939e-01 -2.08548352e-01 -6.86854780e-01 -8.27402294e-01 5.85881174e-01 -1.82582843e+00 -1.09345043e+00 -3.64972293e-01 2.52344847e+00 2.34321877e-01 -5.64025640e-01 4.19351369e-01 2.36750357e-02 1.09900177e+00 -1.56720877e-01 -3.87328625e-01 -5.09429388e-02 -7.50438511e-01 -3.78206551e-01 6.77633524e-01 2.25585073e-01 -1.08615434e+00 4.86247092e-01 5.93897390e+00 2.53152341e-01 -8.30166161e-01 1.79891184e-01 3.11997980e-01 -1.01680465e-01 -1.99865401e-01 -3.13141972e-01 -8.71395290e-01 6.57718956e-01 9.53977108e-01 3.43920171e-01 5.41498661e-01 3.22204113e-01 2.47334152e-01 -2.82326370e-01 -9.53391552e-01 1.69782877e+00 5.85426509e-01 -7.65960395e-01 -8.35460499e-02 2.90502552e-02 5.70834339e-01 -1.67046294e-01 3.66954565e-01 1.30241096e-01 2.09560961e-01 -1.07436514e+00 9.27258730e-01 7.70235777e-01 8.56727540e-01 -8.12566638e-01 1.29369235e+00 1.89416427e-02 -1.39920402e+00 -2.06873730e-01 -1.91405088e-01 4.53773886e-01 2.84464031e-01 -4.84399237e-02 -2.82619864e-01 1.04502213e+00 1.24511147e+00 3.48078966e-01 -9.77930307e-01 1.05260122e+00 1.36324972e-01 1.20161422e-01 -3.02470714e-01 3.94714743e-01 -3.26140195e-01 6.63242862e-02 6.58318996e-01 1.40418899e+00 2.52185374e-01 4.77308892e-02 3.33988741e-02 7.06470370e-01 6.79823086e-02 -3.80177438e-01 -6.80207372e-01 5.30208647e-01 5.56075454e-01 1.03573203e+00 -9.33091193e-02 -9.98886526e-02 -4.02950883e-01 1.21648073e+00 -1.43662661e-01 8.11710238e-01 -1.00315213e+00 -7.65021797e-03 1.10802519e+00 -1.51229322e-01 -6.03294671e-02 -1.58678725e-01 -1.36326641e-01 -1.22390378e+00 -1.47385569e-02 -9.06897306e-01 7.12056041e-01 -1.08193731e+00 -1.58889389e+00 6.98410809e-01 4.71553147e-01 -1.27001774e+00 -3.92554421e-03 -4.06495959e-01 -4.19129938e-01 1.10039270e+00 -1.65591049e+00 -1.90483701e+00 -9.49695945e-01 1.22794044e+00 2.50666495e-02 -4.06108499e-01 5.59117675e-01 5.10913253e-01 -8.88323009e-01 1.18953764e+00 -1.21593503e-02 1.14588566e-01 1.03883553e+00 -9.37229991e-01 1.34738207e-01 1.32671440e+00 -4.25276518e-01 7.64606893e-01 6.25350654e-01 -7.66078413e-01 -2.14597273e+00 -1.12198746e+00 4.05501664e-01 -5.82987547e-01 6.05817959e-02 -3.67409706e-01 -7.95728922e-01 4.67685252e-01 3.03777248e-01 -8.35585967e-02 6.89380229e-01 -2.16596663e-01 -5.33841372e-01 -4.15616333e-01 -1.62962997e+00 4.37533259e-01 9.96924877e-01 -6.83165371e-01 -3.65604371e-01 -3.77635658e-02 2.29236543e-01 -2.67363369e-01 -1.02088273e+00 3.74975592e-01 6.43983603e-01 -9.78646934e-01 1.41185880e+00 -1.28439978e-01 -3.83645147e-01 -7.01270998e-01 -3.50171655e-01 -1.11765993e+00 2.82719731e-02 -5.78824520e-01 2.60277003e-01 1.83763254e+00 -2.06924230e-02 -8.78910124e-01 2.87781924e-01 1.28969967e+00 2.12119892e-01 4.71845239e-01 -1.13358295e+00 -8.73588085e-01 -3.33443254e-01 -4.76603866e-01 9.25834894e-01 7.94546485e-01 -3.58200788e-01 -5.04605830e-01 -7.41861820e-01 8.32481682e-01 1.04908776e+00 -4.29640353e-01 1.39580417e+00 -9.17917132e-01 8.29963759e-02 -2.92980254e-01 -7.02565551e-01 -4.25331801e-01 1.98846698e-01 -4.64718670e-01 -2.77103670e-02 -1.40156865e+00 4.10316318e-01 -3.49896401e-01 -3.56336266e-01 5.86164117e-01 -4.04138058e-01 5.99144280e-01 4.90257859e-01 3.58728707e-01 -5.52863359e-01 4.48470503e-01 4.76609558e-01 -5.15832186e-01 -2.20787033e-01 -2.49890700e-01 -5.42787075e-01 1.56158835e-01 5.53448737e-01 -1.19871549e-01 -1.70707181e-01 -5.65403283e-01 -4.75858152e-02 -7.65762180e-02 9.57150817e-01 -9.29313183e-01 6.15013063e-01 -6.69135824e-02 6.34947777e-01 -7.01440752e-01 7.52268970e-01 -1.00658131e+00 7.64496922e-01 1.24606863e-01 9.45677534e-02 4.11673218e-01 4.24687743e-01 4.10722584e-01 -1.16848879e-01 2.59883195e-01 7.49920011e-01 4.96221036e-02 -8.15982878e-01 9.95217562e-02 -2.53690690e-01 -4.43463624e-01 9.23893809e-01 -5.68459153e-01 -9.58604753e-01 -6.50461137e-01 -2.82893330e-01 4.28073913e-01 8.94428134e-01 6.25777304e-01 8.74185205e-01 -1.14903808e+00 -1.02943516e+00 3.01347286e-01 4.95181918e-01 -3.62249434e-01 8.33279908e-01 9.00137484e-01 -3.05597454e-01 8.67717117e-02 -1.95762336e-01 -8.28124523e-01 -1.82124746e+00 6.95979536e-01 5.39266348e-01 3.27878296e-01 -2.33447507e-01 5.41348398e-01 4.89401966e-02 -5.57665467e-01 4.95622188e-01 1.31931156e-01 -1.42773136e-01 -4.72713932e-02 7.85680056e-01 5.90995133e-01 9.64361802e-02 -1.52252233e+00 -7.11318195e-01 7.70610154e-01 2.95536548e-01 -8.75307992e-02 9.24265504e-01 -5.11103690e-01 -1.47142217e-01 -1.37592226e-01 8.92138362e-01 -2.03374416e-01 -9.44456697e-01 2.66475827e-02 -5.86819112e-01 -6.14675105e-01 -1.91911876e-01 -9.28198218e-01 -9.76794362e-01 4.40426946e-01 1.22219586e+00 -2.20114931e-01 1.27969432e+00 -1.23838767e-01 4.61466342e-01 1.81394160e-01 4.69885379e-01 -8.17948699e-01 -2.27390423e-01 1.23382052e-02 8.04065824e-01 -1.35247934e+00 -4.73305993e-02 -2.52425015e-01 -8.97402287e-01 6.90280139e-01 6.57480240e-01 6.78509593e-01 1.31469861e-01 3.05510521e-01 4.84331042e-01 1.00325093e-01 -1.88387662e-01 -4.28956360e-01 2.33505592e-01 1.18795276e+00 -1.54653579e-01 -6.73168451e-02 4.24672186e-01 4.59171742e-01 7.47219697e-02 -3.39452893e-01 4.54241753e-01 8.31724763e-01 3.42970610e-01 -9.79856849e-01 -1.41958094e+00 -1.22537911e-01 -1.46725297e-01 2.91813850e-01 -6.20119035e-01 7.89425492e-01 2.47122601e-01 1.83918631e+00 -8.38312805e-02 -7.32765675e-01 4.65943754e-01 -1.00028932e-01 4.18543458e-01 2.97607630e-01 -1.09679174e+00 -9.99362469e-02 -1.51317855e-02 -4.80376929e-01 -8.94948244e-01 -6.91268265e-01 -6.32804096e-01 -9.60534334e-01 -1.20297723e-01 -3.17287654e-01 9.30204034e-01 7.55248368e-01 4.36827093e-01 1.54460177e-01 5.96564353e-01 -1.22192383e+00 -1.17391087e-01 -7.51052558e-01 -3.67931813e-01 9.73013937e-01 3.87955368e-01 -6.32097304e-01 -4.48233843e-01 4.13931981e-02]
[14.575874328613281, 0.9999595880508423]
26fe264a-854d-4c4f-8b22-32cf07856cf2
machine-learning-for-real-time-anomaly
2306.10741
null
https://arxiv.org/abs/2306.10741v1
https://arxiv.org/pdf/2306.10741v1.pdf
Machine Learning for Real-Time Anomaly Detection in Optical Networks
This work proposes a real-time anomaly detection scheme that leverages the multi-step ahead prediction capabilities of encoder-decoder (ED) deep learning models with recurrent units. Specifically, an encoder-decoder is used to model soft-failure evolution over a long future horizon (i.e., for several days ahead) by analyzing past quality-of-transmission (QoT) observations. This information is subsequently used for real-time anomaly detection (e.g., of attack incidents), as the knowledge of how the QoT is expected to evolve allows capturing unexpected network behavior. Specifically, for anomaly detection, a statistical hypothesis testing scheme is used, alleviating the limitations of supervised (SL) and unsupervised learning (UL) schemes, usually applied for this purpose. Indicatively, the proposed scheme eliminates the need for labeled anomalies, required when SL is applied, and the need for on-line analyzing entire datasets to identify abnormal instances (i.e., UL). Overall, it is shown that by utilizing QoT evolution information, the proposed approach can effectively detect abnormal deviations in real-time. Importantly, it is shown that the information concerning soft-failure evolution (i.e., QoT predictions) is essential to accurately detect anomalies.
['Georgios Ellinas', 'Tania Panayiotou', 'Sadananda Behera']
2023-06-19
null
null
null
null
['anomaly-detection']
['methodology']
[ 4.22553197e-02 -5.86067922e-02 1.07720010e-02 -1.04881175e-01 -5.54765761e-01 -4.44765508e-01 3.08447152e-01 7.09115565e-01 7.91375618e-03 5.39825678e-01 -2.40467414e-01 -7.16882706e-01 -1.72850400e-01 -8.90701354e-01 -6.48849666e-01 -7.02015102e-01 -5.74098170e-01 1.21113975e-02 6.34291843e-02 -1.82882130e-01 1.98127583e-01 6.53005660e-01 -1.34823525e+00 -1.07556775e-01 5.92086792e-01 1.38925135e+00 -3.29884470e-01 7.73595810e-01 2.15143323e-01 1.00534713e+00 -8.86469543e-01 -2.70267576e-01 1.05254829e-01 -3.21931958e-01 -4.50212657e-01 2.22499430e-01 -4.30781275e-01 -8.66458297e-01 -4.40661222e-01 7.19912648e-01 1.00922465e-01 -1.19392946e-01 3.98064584e-01 -1.47384953e+00 5.63584454e-02 3.55633855e-01 -1.94666252e-01 7.21055627e-01 2.03940690e-01 3.78096879e-01 1.19323039e+00 -6.45910263e-01 6.89540356e-02 7.22966850e-01 3.97694856e-01 1.54019259e-02 -9.33494449e-01 -4.46649760e-01 3.79388660e-01 4.33033407e-01 -1.05083060e+00 -3.10571462e-01 9.91559029e-01 -4.50246364e-01 9.10654247e-01 2.50591692e-02 6.32454216e-01 1.05574429e+00 7.19643354e-01 8.74861419e-01 3.13674390e-01 -3.01099896e-01 6.01837277e-01 -3.06443036e-01 7.46442005e-02 4.21848685e-01 1.06710829e-01 2.92573094e-01 -3.18361282e-01 -2.07071826e-01 3.78595531e-01 4.46993083e-01 1.37555346e-01 -2.32431423e-02 -6.69148684e-01 5.09683669e-01 -5.69896176e-02 2.42522389e-01 -9.56003487e-01 6.81628436e-02 9.41488445e-01 7.25395322e-01 6.46053195e-01 3.46145958e-01 -5.30858397e-01 -4.91796821e-01 -1.04551101e+00 -1.77735299e-01 6.87822998e-01 6.89982772e-01 7.01032460e-01 9.11847651e-01 -2.10489586e-01 2.90578038e-01 1.09344020e-01 2.73380041e-01 4.13602769e-01 -5.67772090e-01 3.21909100e-01 5.41182935e-01 -1.77900381e-02 -7.83115566e-01 -4.91739899e-01 -9.07318473e-01 -8.30196619e-01 -1.06305502e-01 9.87309515e-02 -5.08570850e-01 -6.66396379e-01 1.63469100e+00 5.00067621e-02 5.25357425e-01 1.59874409e-01 2.84497499e-01 -1.69442654e-01 6.24659359e-01 -2.39760563e-01 -4.74542141e-01 9.14841712e-01 -3.25538009e-01 -6.82303131e-01 -2.85854489e-02 1.09550548e+00 -2.23759696e-01 4.54924017e-01 5.58317184e-01 -8.41387749e-01 -2.48639166e-01 -1.00586200e+00 9.75626111e-01 -2.37344593e-01 -9.91427004e-02 3.42221260e-01 5.24304867e-01 -6.71238422e-01 5.61936080e-01 -1.12628484e+00 -2.92164385e-01 2.85940677e-01 1.50833622e-01 1.62463754e-01 1.27213985e-01 -1.45880651e+00 5.37938476e-01 5.16071260e-01 1.86740771e-01 -1.38146055e+00 -5.34717381e-01 -6.85608804e-01 3.91378015e-01 5.81944585e-01 -2.46912651e-02 1.24355876e+00 -7.26249158e-01 -1.31019926e+00 5.24041206e-02 -6.59234002e-02 -1.03695643e+00 2.98782408e-01 -2.37979159e-01 -1.10030389e+00 3.04300338e-01 -2.26381257e-01 -4.77468580e-01 1.07931638e+00 -6.89077914e-01 -8.05844963e-01 -2.34423146e-01 1.30555987e-01 -1.52366057e-01 -7.02612460e-01 8.38999748e-02 -1.39524698e-01 -7.37781405e-01 7.40649849e-02 -6.84827030e-01 -1.13893613e-01 -5.33948123e-01 -5.34371614e-01 -1.50951669e-01 1.04144442e+00 -9.31543350e-01 1.77847707e+00 -2.27588320e+00 -3.52067590e-01 7.68665791e-01 9.10014734e-02 1.98564395e-01 1.93300471e-01 1.03622735e+00 -2.17680484e-01 -1.71661347e-01 -2.39014238e-01 -1.95310056e-01 -2.55067438e-01 4.23658073e-01 -6.39454782e-01 3.38204175e-01 5.42581379e-01 6.50791883e-01 -7.63616920e-01 4.53189760e-02 2.86579281e-01 -2.18883544e-01 -3.67795944e-01 4.78688091e-01 -3.34439635e-01 5.72633266e-01 -4.95359391e-01 8.58742595e-01 1.10746019e-01 -3.02554518e-01 1.52143911e-02 2.49081254e-01 -7.51854703e-02 2.78560549e-01 -8.67515504e-01 1.02603912e+00 -7.91153908e-01 5.23765087e-01 -3.15841347e-01 -1.52326584e+00 1.03760433e+00 7.76799917e-01 8.30049932e-01 -8.88613224e-01 -5.69166504e-02 3.47393811e-01 7.51152784e-02 -2.62883633e-01 2.67557465e-02 2.68093824e-01 -1.39636502e-01 9.11465406e-01 -6.06722385e-02 6.44991279e-01 2.52618015e-01 1.50418729e-01 1.65602672e+00 -4.29123282e-01 4.47340488e-01 3.13301593e-01 7.00120211e-01 -2.53662944e-01 8.93058836e-01 5.93050957e-01 -3.40413675e-02 -8.38949308e-02 8.48170280e-01 -5.44862866e-01 -1.17297828e+00 -1.14552200e+00 1.40289307e-01 8.87707710e-01 -2.30384707e-01 -3.81495237e-01 -2.86875606e-01 -6.57727301e-01 -4.58155051e-02 9.28302586e-01 -4.35564250e-01 -6.66637480e-01 -5.85150599e-01 -7.05591559e-01 4.46944833e-01 5.93591094e-01 2.15798318e-01 -1.12678564e+00 -7.80936301e-01 8.03107858e-01 2.98050076e-01 -1.04693079e+00 1.19542256e-01 4.32962745e-01 -9.08414245e-01 -9.33514655e-01 -2.63936192e-01 1.24643452e-01 6.33113563e-01 -1.14903904e-01 8.24323595e-01 3.34665664e-02 -1.77234247e-01 5.52870274e-01 -5.31867385e-01 -2.87737310e-01 -5.94969094e-01 -2.47898474e-01 3.84245396e-01 5.05804479e-01 3.18569660e-01 -8.81053507e-01 -5.85774839e-01 1.55020103e-01 -1.03012633e+00 -5.20828128e-01 8.01055789e-01 9.55749154e-01 3.86392891e-01 5.27674317e-01 1.15471423e+00 -6.29956663e-01 6.54653370e-01 -8.37583542e-01 -7.61576533e-01 3.24739844e-01 -9.04757142e-01 1.18540354e-01 1.21586156e+00 -2.21738875e-01 -7.22795665e-01 -5.85746586e-01 -9.45677832e-02 -4.78089631e-01 -1.65838405e-01 8.69570076e-01 7.33661577e-02 4.28955048e-01 2.15571418e-01 6.34710908e-01 -1.96479201e-01 -3.65914792e-01 -2.07477301e-01 8.30247223e-01 3.99372727e-01 -2.69550085e-01 8.41620445e-01 2.65371114e-01 8.53342265e-02 -7.15519369e-01 -8.08024824e-01 -6.10834837e-01 -5.17829061e-01 -5.28547704e-01 2.59374589e-01 -8.13565731e-01 -6.23432934e-01 5.88750362e-01 -8.37389648e-01 6.56890273e-02 -3.38992417e-01 4.74955112e-01 -4.60815728e-01 2.73048282e-01 -7.88024545e-01 -1.20540345e+00 -4.90982801e-01 -6.71418309e-01 6.75076604e-01 -6.01166748e-02 -2.40140334e-01 -1.11536074e+00 -1.18329160e-01 -1.36867449e-01 3.20787907e-01 3.21283758e-01 9.59198058e-01 -1.21101236e+00 -5.31398296e-01 -8.17331493e-01 5.53308539e-02 6.55066848e-01 3.13286036e-01 1.09202519e-01 -8.05253088e-01 -7.40042210e-01 -1.08600315e-03 7.63622532e-03 1.85104996e-01 5.78140095e-02 1.45578706e+00 -5.64071357e-01 -2.71066781e-02 2.44758695e-01 1.07851696e+00 6.61602676e-01 3.05798113e-01 3.25510204e-01 3.45992327e-01 3.78555000e-01 7.79211044e-01 1.19257903e+00 2.52235472e-01 5.53003132e-01 6.88647568e-01 1.18104974e-02 5.81624210e-01 -2.06593558e-01 6.76694453e-01 8.14648092e-01 3.95740718e-01 -2.70141989e-01 -1.14260995e+00 6.70438409e-01 -1.71301806e+00 -9.27480400e-01 2.66088158e-01 2.40317059e+00 2.69561291e-01 7.45939374e-01 1.65323451e-01 8.68319273e-01 4.19424206e-01 5.45291118e-02 -1.00382161e+00 -6.22476220e-01 1.75773263e-01 9.04686749e-02 2.70799100e-01 3.55430469e-02 -8.97803843e-01 3.46418262e-01 5.16643381e+00 5.12311161e-01 -1.21133149e+00 -1.59371361e-01 4.42858458e-01 1.83136370e-02 1.99172571e-02 6.13229200e-02 -4.04935658e-01 7.08094716e-01 1.51730978e+00 -2.88399965e-01 1.18301608e-01 7.55220950e-01 5.40461063e-01 1.78659931e-01 -1.14830256e+00 5.33491552e-01 -1.39540389e-01 -9.15742934e-01 -1.50265917e-01 1.08742520e-01 2.67973155e-01 -1.02267349e-02 3.52419913e-02 3.47285479e-01 5.83477579e-02 -5.88854790e-01 3.64966005e-01 6.93989336e-01 4.11470026e-01 -1.15108943e+00 1.11259401e+00 5.80851018e-01 -1.00263739e+00 -7.87158668e-01 1.53050005e-01 5.91616705e-03 4.32713479e-01 1.19746888e+00 -9.93543684e-01 8.56654823e-01 6.28116965e-01 9.17939544e-01 -4.19703245e-01 8.50633681e-01 -1.61364108e-01 1.06870735e+00 -2.67549545e-01 3.84941310e-01 3.10430229e-01 1.62882973e-02 7.61423230e-01 8.58808100e-01 4.40905094e-01 -3.94274473e-01 2.91333735e-01 4.42088991e-01 2.57853866e-01 -2.37058446e-01 -5.57480752e-01 -2.83843249e-01 6.14992738e-01 8.78926218e-01 -3.80118459e-01 -3.00918579e-01 -5.01923084e-01 7.06033468e-01 -1.72984153e-01 4.85087425e-01 -6.13487363e-01 -3.75904530e-01 5.82049310e-01 9.11458060e-02 3.24297607e-01 -2.93473840e-01 -1.17095046e-01 -8.69617403e-01 2.80665398e-01 -6.99485600e-01 5.26964128e-01 -3.68490368e-01 -1.20275760e+00 4.88072544e-01 -2.19245821e-01 -1.86995625e+00 -9.74886775e-01 -1.57740518e-01 -1.14346266e+00 5.79099476e-01 -1.46949029e+00 -7.47833550e-01 1.98739499e-01 5.49614966e-01 4.78771150e-01 -3.41429055e-01 5.30917525e-01 2.67971873e-01 -1.14182413e+00 7.65331805e-01 2.37744302e-01 2.77532011e-01 1.81921422e-01 -9.97528911e-01 4.24408913e-01 1.35321224e+00 9.90338624e-02 1.77655593e-01 7.20694840e-01 -5.45799017e-01 -1.15941548e+00 -1.26358438e+00 5.03731668e-01 8.28988776e-02 1.11722624e+00 -2.07936332e-01 -1.17929316e+00 6.02449417e-01 -4.29735035e-01 1.31036460e-01 7.30278492e-01 -6.08284958e-02 3.95784667e-03 -3.84111136e-01 -8.46343160e-01 4.85534906e-01 3.88360023e-01 -6.80844426e-01 -2.79349655e-01 2.19850048e-01 5.67370117e-01 -4.63429689e-02 -1.01557589e+00 5.44469655e-01 2.43847042e-01 -8.48292232e-01 5.03671944e-01 -5.98840714e-01 9.75788236e-02 -1.05604671e-01 -5.49637824e-02 -1.32000279e+00 -3.53801996e-02 -6.30264223e-01 -1.02817714e+00 1.04687452e+00 4.36356008e-01 -8.94736826e-01 5.67013860e-01 2.36872718e-01 -3.83409858e-01 -9.56265211e-01 -1.24301088e+00 -1.01658332e+00 -5.05063176e-01 -6.63237691e-01 6.70950294e-01 7.28595138e-01 -2.21930034e-02 -9.59058255e-02 -5.24953961e-01 7.23782480e-01 4.18858498e-01 1.61926717e-01 4.66333658e-01 -1.23181319e+00 -3.61211419e-01 -1.44927949e-01 -6.72682762e-01 -7.34198213e-01 5.29277772e-02 -3.61252606e-01 -2.84652293e-01 -8.69432867e-01 -4.61763084e-01 -2.03993380e-01 -9.37116563e-01 4.97424424e-01 9.12090112e-03 -1.85900673e-01 -1.14769064e-01 3.06756407e-01 -5.88417590e-01 7.67475009e-01 4.15931255e-01 1.49497181e-01 -2.31215909e-01 6.22608840e-01 -8.62274915e-02 5.36396921e-01 1.19733667e+00 -4.31454360e-01 -4.19161022e-01 1.08213745e-01 2.95577675e-01 4.46178168e-01 4.53717172e-01 -1.21514618e+00 2.36031905e-01 5.50535843e-02 1.02224179e-01 -8.44035745e-01 1.16175346e-01 -8.06456149e-01 -2.43283793e-01 7.37046480e-01 -1.55749068e-01 4.38123196e-01 5.48185632e-02 9.46557820e-01 -5.59329927e-01 4.53153327e-02 3.43278527e-01 4.33707565e-01 -7.74636805e-01 5.53943872e-01 -6.47491097e-01 -1.16866603e-01 1.31816936e+00 -1.72335565e-01 1.56734213e-01 -6.93665385e-01 -6.84988976e-01 5.96140325e-01 3.75629868e-04 3.53596389e-01 7.18002737e-01 -1.09005499e+00 -5.27010381e-01 3.95613313e-01 4.77432132e-01 -2.41246179e-01 3.90956551e-01 1.04623592e+00 -8.17362443e-02 4.35517699e-01 1.74517441e-03 -6.50725007e-01 -8.55619431e-01 5.35168231e-01 2.92835385e-01 -5.99404275e-01 -9.01180983e-01 2.36410126e-01 -3.65881175e-01 1.06718309e-01 1.55033976e-01 -2.37171320e-04 -1.20998189e-01 9.94774476e-02 3.47698867e-01 3.94904256e-01 4.60039377e-01 -1.13481514e-01 -2.29862198e-01 -2.32706126e-02 -4.33548570e-01 1.08207867e-01 1.27527881e+00 -1.49676427e-01 -3.00940666e-02 8.25708985e-01 8.68632436e-01 -3.33681583e-01 -1.41891325e+00 -6.11722767e-01 4.89049464e-01 6.06792346e-02 1.50426507e-01 -5.95461488e-01 -1.12381196e+00 9.83926356e-01 4.92529094e-01 6.33953035e-01 1.39762795e+00 -3.41442227e-01 1.04963171e+00 4.73077089e-01 3.01226079e-01 -1.21886110e+00 3.79457995e-02 5.39418399e-01 3.98661345e-01 -9.03816640e-01 -3.08037072e-01 2.61304885e-01 -4.51547712e-01 1.13781953e+00 3.98350477e-01 1.95731044e-01 8.87469828e-01 1.53035685e-01 -2.10112587e-01 -2.25993857e-01 -1.22727668e+00 5.11259027e-02 -9.73486435e-03 2.06140906e-01 4.86510284e-02 1.26251727e-01 1.72031567e-01 3.44144702e-01 -7.62033910e-02 -3.82799715e-01 7.19440162e-01 1.12818456e+00 -3.66802305e-01 -8.88021171e-01 -1.32600218e-01 8.33823383e-01 -4.95133102e-01 5.83562851e-02 -5.83636872e-02 4.48937505e-01 -3.09096217e-01 1.15001965e+00 3.29109520e-01 -4.84114766e-01 3.42591941e-01 3.13293546e-01 -4.21928614e-01 -5.53204358e-01 -3.14435840e-01 -2.42977917e-01 -1.27993599e-01 -7.54189551e-01 1.15348674e-01 -4.98067021e-01 -1.05267560e+00 -1.82324395e-01 -1.63626313e-01 1.03573456e-01 4.42072421e-01 1.33321631e+00 4.06401753e-01 9.50316310e-01 1.55013096e+00 -1.15672365e-01 -7.39872038e-01 -9.45545018e-01 -7.09140599e-01 1.93519637e-01 6.85499310e-01 -5.79536557e-01 -6.22581899e-01 -4.22399491e-01]
[7.308893203735352, 2.6909446716308594]
1a5b7c2c-e90d-4800-8d4e-f11eadb473dc
markbert-marking-word-boundaries-improves
null
null
https://openreview.net/forum?id=7uE-SSLTgxw
https://openreview.net/pdf?id=7uE-SSLTgxw
MarkBERT: Marking Word Boundaries Improves Chinese BERT
We present a Chinese BERT model dubbed MarkBERT that uses word information in this work. Existing word-based BERT models regard words as basic units, however, due to the vocabulary limit of BERT, they only cover high-frequency words and fall back to character level when encountering out-of-vocabulary (OOV) words. Different from existing works, MarkBERT keeps the vocabulary being Chinese characters and inserts boundary markers between contiguous words. Such design enables the model to handle any words in the same way, no matter they are OOV words or not. Besides, our model has two additional benefits: first, it is convenient to add word-level learning objectives over markers, which is complementary to traditional character and sentence-level pretraining tasks; second, it can easily incorporate richer semantics such as POS tags of words by replacing generic markers with POS tag-specific markers. MarkBERT pushes the state-of-the-art of Chinese named entity recognition from 95.4\% to 96.5\% on the MSRA dataset and from 82.8\% to 84.2\% on the OntoNotes dataset, respectively. Compared to previous word-based BERT models, MarkBERT achieves better accuracy on text classification, keyword recognition, and semantic similarity tasks.\footnote{All the codes and models will be made publicly available at \url{https://github.com/}}
['Anonymous']
2021-11-16
null
null
null
acl-arr-november-2021-11
['chinese-named-entity-recognition']
['natural-language-processing']
[-2.86226809e-01 -2.94113308e-01 -6.02996171e-01 -1.93676963e-01 -8.63611877e-01 -6.50458336e-01 2.23916814e-01 3.70935142e-01 -9.37187433e-01 6.75669909e-01 3.51106703e-01 -5.63525379e-01 2.45066479e-01 -8.39055657e-01 -3.93002421e-01 -3.28356206e-01 1.90246925e-01 4.15341914e-01 4.67378289e-01 -3.97167563e-01 3.44811946e-01 1.09425060e-01 -9.03804898e-01 3.16472560e-01 1.16892505e+00 8.40974033e-01 5.05160093e-01 2.66934037e-01 -7.64130712e-01 3.63327295e-01 -6.31012678e-01 -5.57435572e-01 -5.77984862e-02 1.74185187e-02 -7.11720109e-01 -3.52773160e-01 4.21046764e-02 1.43624060e-02 -5.81598759e-01 1.13504136e+00 5.56379855e-01 1.31115735e-01 5.09881914e-01 -8.17957163e-01 -1.26813722e+00 1.30673635e+00 -3.49797010e-01 2.92144299e-01 2.40894064e-01 -3.20748657e-01 1.46843874e+00 -1.18908715e+00 4.40396428e-01 8.61130118e-01 6.66062593e-01 6.79258704e-01 -6.99656010e-01 -8.83817255e-01 5.32389164e-01 2.77533382e-01 -1.72225153e+00 -2.13148475e-01 3.68770093e-01 -2.19798610e-01 1.22559774e+00 2.87170231e-01 4.55157489e-01 8.93141508e-01 -1.00736879e-01 1.14879680e+00 7.86715627e-01 -7.11080492e-01 2.28456641e-03 -8.61726627e-02 7.73657501e-01 4.44209427e-01 2.75342315e-01 -2.62128264e-01 -4.04206902e-01 5.15265614e-02 4.50730801e-01 9.15852711e-02 -3.50300074e-01 2.29249969e-01 -1.31619549e+00 7.91763365e-01 2.12615743e-01 6.72056973e-01 -3.97687331e-02 1.85533389e-01 4.36481446e-01 -8.47513303e-02 3.84903044e-01 4.53100622e-01 -7.89352477e-01 -3.24448258e-01 -9.07347143e-01 -1.43527761e-01 7.01277912e-01 1.53454053e+00 6.39476120e-01 2.20701143e-01 -2.74841130e-01 1.12922621e+00 2.13145912e-01 6.54221117e-01 1.07901549e+00 -2.78862584e-02 7.21426547e-01 6.37707353e-01 -8.63940194e-02 -5.34553409e-01 -3.83704811e-01 -5.18610895e-01 -6.83619022e-01 -7.53080666e-01 9.84305516e-02 -3.19858074e-01 -1.20717454e+00 1.44743407e+00 -1.85577378e-01 1.97740436e-01 1.37924120e-01 5.84239364e-01 1.31602108e+00 8.61725330e-01 2.72111773e-01 -5.10220379e-02 1.69462681e+00 -1.06660628e+00 -9.49311793e-01 -4.98374343e-01 9.28861082e-01 -8.73693883e-01 1.45097613e+00 8.54536369e-02 -8.87205124e-01 -4.39077526e-01 -9.89845753e-01 -9.01129842e-02 -8.59022617e-01 3.42207015e-01 4.73819405e-01 9.60961759e-01 -8.25756133e-01 1.95988417e-01 -6.12160325e-01 -2.24029094e-01 6.68939650e-02 2.43606403e-01 -6.02233000e-02 9.69000743e-04 -1.69181931e+00 7.02745140e-01 7.94212043e-01 -1.54840454e-01 -3.54824543e-01 -5.14963865e-01 -1.00146246e+00 1.80990994e-01 3.21817458e-01 -9.25488025e-02 1.23973238e+00 -6.66611433e-01 -1.14271581e+00 7.68954396e-01 -4.56754744e-01 -2.60039419e-01 7.76871145e-02 -4.51447487e-01 -9.03931797e-01 -1.07655235e-01 4.44469117e-02 5.16476691e-01 1.85926363e-01 -9.02374387e-01 -6.90332532e-01 -3.59547883e-02 -1.95482135e-01 1.94353417e-01 -8.68424118e-01 3.61547709e-01 -1.04171741e+00 -1.19289923e+00 -5.85991796e-03 -8.53179514e-01 2.99773067e-02 -6.04370832e-01 -5.36795974e-01 -4.33674276e-01 4.52077478e-01 -6.68287337e-01 1.90151381e+00 -2.28011513e+00 -3.29355180e-01 6.39024377e-02 -1.95828993e-02 6.82235003e-01 -2.23214835e-01 8.01754594e-01 7.29631260e-02 6.55791640e-01 -4.10831608e-02 -2.98006803e-01 1.73657000e-01 1.33923739e-01 -3.69490147e-01 1.60548493e-01 -2.87495553e-02 1.32759643e+00 -7.97352195e-01 -3.75362724e-01 -4.65826094e-02 2.22470880e-01 -1.22241430e-01 -1.90814748e-01 -1.13622330e-01 -2.44533345e-01 -5.57673693e-01 7.43776500e-01 7.39110172e-01 -1.80190518e-01 1.87632650e-01 2.09740669e-01 -2.10162863e-01 6.33108497e-01 -1.15205979e+00 1.36191905e+00 -6.10845387e-01 2.93053865e-01 -3.94477457e-01 -9.40642655e-01 1.19769514e+00 4.82923567e-01 1.43986031e-01 -5.36369860e-01 2.15622321e-01 4.38449889e-01 -1.07093111e-01 -2.19602272e-01 8.11234951e-01 -7.56593570e-02 -4.28658456e-01 8.49029571e-02 5.45159122e-03 -1.76828932e-02 1.75549418e-01 7.95317888e-02 8.92812312e-01 -2.57074147e-01 4.80555922e-01 -3.18771362e-01 5.42102695e-01 1.10043064e-01 8.31812799e-01 5.98567545e-01 -1.48573950e-01 7.15944946e-01 8.90080258e-02 3.72865424e-02 -8.27810645e-01 -1.02420688e+00 -3.37702751e-01 1.23314285e+00 4.61464643e-01 -6.99580610e-01 -4.91415113e-01 -7.05459833e-01 1.64033040e-01 9.77329791e-01 -4.07064587e-01 -2.23523468e-01 -6.92338943e-01 -5.74981391e-01 8.92121494e-01 8.43517482e-01 5.43597281e-01 -1.15790129e+00 2.35771239e-01 3.44568580e-01 -2.33203843e-01 -1.20053422e+00 -1.03545558e+00 2.39120901e-01 -6.55881226e-01 -7.38714993e-01 -1.00570476e+00 -1.42920578e+00 4.29841518e-01 3.80715966e-01 1.06236279e+00 2.94497043e-01 -3.91540080e-02 4.07654308e-02 -1.02883077e+00 -6.78540528e-01 3.36890966e-02 4.17704284e-01 -7.40554035e-02 -3.81790489e-01 8.95896614e-01 3.12241055e-02 -4.02782708e-01 4.44856107e-01 -8.73787940e-01 -1.34468734e-01 5.36824465e-01 9.61038470e-01 5.76208472e-01 -1.97590217e-01 7.32335150e-01 -8.84604394e-01 5.31036496e-01 -4.45330769e-01 -3.47556919e-01 5.22286415e-01 -6.86692417e-01 -1.97726920e-01 7.86939919e-01 -6.91350043e-01 -7.38351285e-01 -1.51801884e-01 -4.79483068e-01 2.84303706e-02 1.53952911e-02 8.69637668e-01 -3.63588125e-01 2.48383388e-01 2.65809476e-01 5.52449107e-01 -5.58074534e-01 -7.36448586e-01 3.67058724e-01 1.09231603e+00 2.16211557e-01 -5.49562156e-01 6.64538085e-01 1.08640529e-01 -8.83795500e-01 -8.06475818e-01 -8.35873842e-01 -1.04950488e+00 -7.34984815e-01 2.87086070e-01 8.69681597e-01 -1.07036924e+00 -4.68619704e-01 5.29480815e-01 -9.45315063e-01 -1.84095070e-01 -1.20207950e-01 6.35636151e-01 3.76767591e-02 6.88214540e-01 -8.93840730e-01 -7.11045504e-01 -6.18631899e-01 -8.55925620e-01 9.90742743e-01 2.68666863e-01 -2.00396940e-01 -1.09987354e+00 -2.94716120e-01 8.42806548e-02 2.33606458e-01 -5.50933480e-01 1.04804945e+00 -1.06023276e+00 2.11433563e-02 -4.05532271e-01 -1.40402243e-01 3.89892548e-01 9.31639597e-02 -2.56103784e-01 -5.93819320e-01 -1.80077866e-01 -4.41286683e-01 -2.08114341e-01 1.02957809e+00 1.51601747e-01 1.17469823e+00 -2.20894590e-01 -4.74200040e-01 6.14519715e-01 1.34592021e+00 5.46943426e-01 6.21254921e-01 6.66972041e-01 6.59608305e-01 2.47520030e-01 6.87992215e-01 4.38532472e-01 4.66915250e-01 6.30919993e-01 8.36448818e-02 -7.14475960e-02 -1.26892805e-01 -4.51353610e-01 3.47440064e-01 1.23977721e+00 4.66238856e-01 -5.79337120e-01 -1.33063483e+00 7.25023210e-01 -1.75925040e+00 -5.91242373e-01 -3.50345850e-01 2.17441154e+00 1.15386534e+00 1.55865178e-01 -1.06816985e-01 -8.00622627e-02 1.10406518e+00 1.62237942e-01 -2.64529586e-01 -2.82614112e-01 -5.56921601e-01 3.66422176e-01 7.79078484e-01 3.50997925e-01 -1.22844684e+00 1.60959530e+00 5.61658764e+00 1.37762117e+00 -1.11749601e+00 1.64493963e-01 4.34566498e-01 2.51496196e-01 -5.10304987e-01 5.76646067e-02 -1.50207198e+00 6.58018172e-01 6.67065144e-01 -4.13744807e-01 1.12860557e-02 7.60809958e-01 -7.66663477e-02 2.40266681e-01 -7.01687217e-01 1.05989325e+00 2.40409821e-01 -1.20680189e+00 1.47653922e-01 -2.74294406e-01 5.64916253e-01 1.46816999e-01 -3.59943733e-02 6.23227775e-01 4.28230286e-01 -1.15341604e+00 8.80332053e-01 1.09732449e-01 1.07386267e+00 -7.59823561e-01 1.12851512e+00 4.44406897e-01 -1.49463654e+00 1.07440814e-01 -6.32565200e-01 1.03887841e-01 3.58402550e-01 5.03742814e-01 -5.32470584e-01 5.54985285e-01 4.31050509e-01 6.61500216e-01 -5.91694653e-01 1.18329191e+00 -5.42748988e-01 9.65927839e-01 -1.56359956e-01 -7.16561615e-01 4.43471193e-01 -2.42174473e-02 2.01137245e-01 1.65661728e+00 3.51556003e-01 1.78835034e-01 4.32520628e-01 2.07935736e-01 -2.55438507e-01 6.90370679e-01 -1.48989037e-01 -1.90523401e-01 8.74032438e-01 1.05563581e+00 -7.06325352e-01 -5.14032722e-01 -6.21801317e-01 1.08316135e+00 4.43112582e-01 4.02581066e-01 -8.04197550e-01 -9.87493575e-01 5.21110952e-01 -2.09189504e-01 5.84407806e-01 -5.34846365e-01 -5.19534647e-01 -1.36600423e+00 3.64178680e-02 -7.01328278e-01 5.42935610e-01 -5.20931184e-01 -1.44560015e+00 6.29607022e-01 -2.53917247e-01 -1.27425301e+00 3.18493456e-01 -7.95607746e-01 -5.39380848e-01 9.61385131e-01 -1.68222928e+00 -1.13220334e+00 -7.56517500e-02 3.54514092e-01 7.10436702e-01 -2.55855620e-01 9.57344472e-01 3.21546495e-01 -6.98515236e-01 1.13338268e+00 6.32811487e-01 8.24638724e-01 8.84459138e-01 -1.16758013e+00 7.08508849e-01 8.53064716e-01 4.45729136e-01 9.20912981e-01 3.76313888e-02 -9.08145487e-01 -1.16548228e+00 -1.03784013e+00 1.55668819e+00 -2.18797669e-01 9.77157414e-01 -6.59232616e-01 -9.86668587e-01 6.35356128e-01 1.63815871e-01 -8.63986090e-02 9.48903620e-01 2.89256930e-01 -4.65996772e-01 1.00274779e-01 -5.91052532e-01 8.88572276e-01 1.09702849e+00 -4.67949092e-01 -9.43087578e-01 3.02667946e-01 1.08188331e+00 -3.52245003e-01 -5.96133828e-01 3.37487608e-01 3.23928893e-01 -2.92396367e-01 7.74689138e-01 -6.14245892e-01 -6.09647743e-02 -1.77923769e-01 -1.88015506e-01 -1.24332201e+00 -4.92341727e-01 -5.06411850e-01 1.86611023e-02 1.56297898e+00 7.76321411e-01 -7.17988372e-01 6.81061924e-01 3.86471987e-01 -5.28178155e-01 -6.53580427e-01 -9.80441272e-01 -1.21138966e+00 5.00645339e-01 -7.47166574e-01 7.48974800e-01 1.11058545e+00 4.26635355e-01 2.61196673e-01 -1.39139503e-01 9.64977294e-02 -1.24558367e-01 -3.12965252e-02 1.39693141e-01 -8.57497573e-01 5.71254874e-03 -6.21239245e-01 -3.19764137e-01 -1.58181167e+00 4.06747729e-01 -1.22166324e+00 1.20275162e-01 -1.61093748e+00 1.45910710e-01 -8.78899515e-01 -4.88859713e-01 7.80770600e-01 -5.93238175e-01 2.74726987e-01 3.46937031e-01 3.08951765e-01 -4.17242736e-01 7.42297113e-01 9.01653588e-01 -1.22820795e-01 -1.43714666e-01 -1.04849145e-01 -7.66182303e-01 4.96468395e-01 1.02851403e+00 -3.34525704e-01 1.65121872e-02 -7.35135376e-01 9.47329402e-02 -4.17329878e-01 -2.70605475e-01 -6.19055331e-01 2.81443715e-01 -1.62659377e-01 1.82086065e-01 -6.76585257e-01 8.53298232e-02 -5.26537120e-01 -5.08507729e-01 4.63450670e-01 -2.85364389e-01 4.00645435e-01 3.19849908e-01 3.38685066e-01 -3.83294851e-01 -7.51835585e-01 5.38861752e-01 -1.10910848e-01 -1.25677323e+00 3.49107832e-01 -5.21845281e-01 5.26443243e-01 9.25331175e-01 -3.18123370e-01 -3.75605077e-01 -3.14875662e-01 -4.05731320e-01 4.29390758e-01 2.30706692e-01 6.84074521e-01 6.83000922e-01 -1.26929867e+00 -6.73503041e-01 -6.04223087e-02 4.97336000e-01 -1.29811078e-01 6.99794739e-02 5.24717033e-01 -5.61898947e-01 6.94208086e-01 3.90348643e-01 -1.19870014e-01 -1.20461988e+00 6.62989378e-01 -5.96738458e-02 -9.64306891e-02 -6.72627509e-01 1.09976351e+00 1.46870255e-01 -6.22225642e-01 4.19848233e-01 -2.94190913e-01 -4.78275120e-01 1.49962962e-01 6.31876886e-01 8.81896541e-02 1.72770381e-01 -6.69008017e-01 -6.11395001e-01 6.70522273e-01 -2.46346548e-01 -1.05519496e-01 1.04926085e+00 -1.38589814e-01 4.44373079e-02 5.61341465e-01 1.11101365e+00 5.50087273e-01 -5.04026949e-01 -5.20207345e-01 4.45714653e-01 -2.91829318e-01 -1.06534101e-01 -8.29891562e-01 -8.11217546e-01 8.90619814e-01 1.20330438e-01 -1.74324244e-01 8.14158380e-01 9.71884429e-02 1.24019611e+00 4.66035932e-01 2.65816540e-01 -1.34097385e+00 -1.03131369e-01 1.25993204e+00 4.18225676e-01 -1.05322206e+00 -2.60061055e-01 -6.69936955e-01 -9.03057158e-01 1.05980408e+00 5.70231736e-01 1.94770452e-02 7.91772306e-01 2.97925591e-01 3.98354858e-01 1.91715986e-01 -4.38268602e-01 -5.01107454e-01 4.61482435e-01 5.21783292e-01 7.80508161e-01 2.27108061e-01 -9.54556465e-01 1.14999473e+00 -3.75598431e-01 -3.09365660e-01 4.53300565e-01 9.41115618e-01 -5.63694954e-01 -1.36035180e+00 -1.22854583e-01 4.36759084e-01 -7.00032353e-01 -9.86137927e-01 -2.60622114e-01 8.54266703e-01 -2.07307339e-02 1.02533579e+00 9.31099355e-02 -3.72744888e-01 3.30711454e-01 2.99670249e-01 -3.18078578e-01 -1.11777484e+00 -6.43366218e-01 1.28592744e-01 2.19918400e-01 2.78894622e-02 1.19619668e-01 -5.21928966e-01 -1.70133996e+00 -1.62850022e-01 -7.30857849e-01 5.86040676e-01 4.45956498e-01 8.75876009e-01 2.16177881e-01 2.99993426e-01 4.48852658e-01 -1.70395836e-01 -6.10362291e-01 -1.00729930e+00 -8.34485590e-01 3.30398411e-01 -1.23685181e-01 -3.80120724e-01 -9.97716635e-02 -2.05086350e-01]
[10.112523078918457, 10.033121109008789]
223327ef-c20a-4d85-b467-78b2f030dccf
improving-audio-language-learning-with-mixgen
2210.17143
null
https://arxiv.org/abs/2210.17143v2
https://arxiv.org/pdf/2210.17143v2.pdf
Exploring Train and Test-Time Augmentations for Audio-Language Learning
In this paper, we aim to unveil the impact of data augmentation in audio-language multi-modal learning, which has not been explored despite its importance. We explore various augmentation methods at not only train-time but also test-time and find out that proper data augmentation can lead to substantial improvements. Specifically, applying our proposed audio-language paired augmentation PairMix, which is the first multi-modal audio-language augmentation method, outperforms the baselines for both automated audio captioning and audio-text retrieval tasks. To fully take advantage of data augmentation, we also present multi-level test-time augmentation (Multi-TTA) for the test-time. We successfully incorporate the two proposed methods and uni-modal augmentations and achieve 47.5 SPIDEr on audio captioning, which is an 18.2% relative increase over the baseline. In audio-text retrieval, the proposed methods also show an improvement in performance as well.
['Kyogu Lee', 'Jinwoo Lee', 'Jaeheon Sim', 'Minju Park', 'KyungSu Kim', 'Yoori Oh', 'Jinhee Kim', 'Eungbeom Kim']
2022-10-31
null
null
null
null
['audio-captioning']
['audio']
[ 4.22624648e-01 -1.77204445e-01 -5.21056466e-02 -2.43629470e-01 -2.05800533e+00 -7.40599632e-01 6.44136727e-01 2.83056408e-01 -4.62724894e-01 4.60633785e-01 5.39003909e-01 -2.13534027e-01 1.28817618e-01 -2.83554912e-01 -7.35152006e-01 -5.14062047e-01 -8.13688189e-02 5.99794745e-01 3.99810150e-02 -2.38499552e-01 -1.03912450e-01 1.11723103e-01 -1.60236859e+00 6.70840442e-01 6.62597716e-01 9.56338048e-01 -2.26166278e-01 9.60655749e-01 -1.03737332e-01 4.12665844e-01 -6.63192570e-01 -1.10176086e-01 8.21630135e-02 -1.31534651e-01 -8.48897815e-01 -9.26601738e-02 8.46524060e-01 -2.53329009e-01 -2.24909201e-01 4.75257754e-01 1.03319514e+00 1.53412372e-01 4.24274206e-01 -1.48314655e+00 -2.70394385e-01 7.79503584e-01 -9.95097339e-01 2.17010021e-01 4.99285936e-01 -2.17794001e-01 1.27185583e+00 -1.06515396e+00 -9.04421136e-02 1.46951938e+00 5.25412738e-01 2.54691839e-01 -9.84723568e-01 -1.03141201e+00 8.94039199e-02 1.63339898e-01 -1.40564215e+00 -6.56014979e-01 7.91076481e-01 -2.94456095e-01 7.17190921e-01 3.15222502e-01 2.03014180e-01 1.02665627e+00 -2.37785459e-01 1.15619373e+00 1.06460953e+00 -7.94345319e-01 -3.21920931e-01 1.33751974e-01 1.12357708e-02 2.90827692e-01 -3.52293491e-01 -2.01926365e-01 -7.25128829e-01 -2.37291902e-01 3.91517371e-01 -3.55786502e-01 -7.35851079e-02 2.78395295e-01 -1.34595895e+00 5.49711108e-01 6.09301552e-02 3.26406658e-01 -3.46822202e-01 3.42124939e-01 7.11100459e-01 3.30065668e-01 5.37735164e-01 3.86255950e-01 -4.46727276e-01 -3.67515206e-01 -1.24088955e+00 1.52300015e-01 1.67519093e-01 7.24355817e-01 4.94628251e-01 3.87616545e-01 -5.42406023e-01 1.28086603e+00 4.52908933e-01 6.92661583e-01 6.80438697e-01 -6.09720111e-01 7.62532830e-01 2.27492809e-01 -1.47181287e-01 -4.67686146e-01 -1.95017159e-01 -4.99058366e-01 -7.72125304e-01 -3.89033198e-01 5.62001616e-02 -3.29755604e-01 -1.08686650e+00 1.85593629e+00 2.42269069e-01 5.45149326e-01 1.99261516e-01 5.56343377e-01 9.09807026e-01 1.02001834e+00 2.07946718e-01 -3.22328508e-01 1.41954494e+00 -1.29488456e+00 -1.08092022e+00 -2.35683680e-01 7.59872913e-01 -1.35998619e+00 1.27898097e+00 3.75099659e-01 -1.22424757e+00 -7.59119570e-01 -9.88386512e-01 4.94513847e-02 -1.57480493e-01 2.04463542e-01 4.73975152e-01 6.06928170e-01 -1.08352172e+00 -6.24942891e-02 -6.28909945e-01 -2.04454169e-01 -1.89016629e-02 4.34434414e-01 -3.44361126e-01 -6.49876744e-02 -1.41073227e+00 2.65183955e-01 2.87380248e-01 -1.56587675e-01 -1.08292305e+00 -8.52135420e-01 -7.41139412e-01 -3.50668728e-02 3.67902726e-01 -3.37098420e-01 1.61091125e+00 -5.42621672e-01 -1.22409439e+00 5.00985742e-01 -3.65848035e-01 -4.30245876e-01 8.49423558e-02 -5.91048300e-01 -7.59795070e-01 1.69262782e-01 4.07015532e-02 1.09278929e+00 7.38852561e-01 -1.09892726e+00 -6.50402248e-01 -2.16972888e-01 -6.94294199e-02 4.38281566e-01 -9.06086326e-01 1.28053978e-01 -7.80732751e-01 -1.03046954e+00 -1.08485594e-01 -1.15961909e+00 1.18474275e-01 -4.04704213e-01 -3.33407283e-01 -1.88507706e-01 1.02507055e+00 -7.30660677e-01 1.42993641e+00 -2.25075626e+00 -2.88364828e-01 1.06495922e-03 -3.23772848e-01 3.60905290e-01 -5.89477003e-01 5.92524767e-01 -3.38215321e-01 2.90156692e-01 -7.00071035e-03 -7.67430902e-01 -5.57912663e-02 1.40160974e-02 -5.48272789e-01 -5.48518915e-03 4.20745879e-01 8.52268159e-01 -4.45607603e-01 -6.23693168e-01 1.09253392e-01 5.92481375e-01 -6.93139315e-01 2.05965728e-01 -1.20791957e-01 4.48076844e-01 -3.39552343e-01 8.98898959e-01 4.58277702e-01 4.96743284e-02 -3.30177069e-01 -2.22336292e-01 4.07850632e-04 3.10914844e-01 -1.16947579e+00 1.81267238e+00 -6.50292516e-01 8.01924229e-01 -8.19927230e-02 -6.90276921e-01 7.39449263e-01 9.27272141e-01 5.44193447e-01 -6.78714693e-01 -3.25033888e-02 2.84906864e-01 6.77457079e-02 -3.04054886e-01 6.70738459e-01 -2.01847941e-01 -2.70159423e-01 5.95685780e-01 2.58924425e-01 6.03962131e-02 1.72505289e-01 2.70781815e-01 9.10806417e-01 -1.51655406e-01 -1.21211573e-01 4.88239266e-02 7.99886823e-01 -3.27994794e-01 6.54355288e-02 8.62584233e-01 -2.20999029e-02 8.10479105e-01 -6.98404387e-02 2.90723205e-01 -9.73464668e-01 -7.86104321e-01 2.60893684e-02 1.78643298e+00 -4.39170331e-01 -4.66174990e-01 -4.08549786e-01 -4.10346985e-01 -1.89468428e-01 4.01603729e-01 -2.82420009e-01 -6.35907203e-02 -4.50734198e-01 -7.94700384e-01 1.12428200e+00 6.03722751e-01 4.30768281e-01 -8.58965218e-01 3.17952216e-01 2.23507464e-01 -7.03545153e-01 -1.54242444e+00 -7.56156981e-01 4.90882620e-02 -8.24263752e-01 -4.25883800e-01 -9.80532765e-01 -8.11903834e-01 1.69711930e-04 3.03029299e-01 7.44062901e-01 -2.03815728e-01 1.86323631e-03 7.02786148e-01 -6.28010094e-01 -7.42995679e-01 -4.80115980e-01 4.09476578e-01 2.40218520e-01 2.30766773e-01 1.02949820e-01 -7.54589677e-01 -3.19201797e-01 2.96692729e-01 -9.57903206e-01 -1.31627277e-01 1.06706965e+00 6.07663155e-01 6.66566908e-01 -1.33608714e-01 1.06823206e+00 -5.29095650e-01 7.33666956e-01 -3.34327847e-01 -7.50967488e-02 1.75500140e-01 -4.24134791e-01 -1.27560776e-02 2.39530027e-01 -6.75655544e-01 -8.10553312e-01 -1.77634470e-02 -5.01777411e-01 -5.04975557e-01 -2.99897909e-01 8.21543217e-01 -9.56078842e-02 -2.20922884e-02 4.36097652e-01 2.27014765e-01 -1.33558452e-01 -7.03172028e-01 3.23127270e-01 1.08034706e+00 5.79627097e-01 -6.47521675e-01 9.63764608e-01 1.85852289e-01 -2.24849924e-01 -1.05453241e+00 -8.64343226e-01 -9.64599431e-01 -4.59589392e-01 -1.61764309e-01 6.87781394e-01 -1.46092832e+00 -3.72116417e-01 2.83913434e-01 -1.17411518e+00 -2.14387160e-02 -5.82643189e-02 6.56637073e-01 -4.15249854e-01 2.19878525e-01 -5.10849297e-01 -1.19839799e+00 -5.54147840e-01 -1.11199224e+00 1.46611702e+00 -1.88713029e-01 -2.21132338e-02 -7.08716869e-01 2.05637366e-01 7.07477868e-01 3.82319748e-01 1.12071428e-02 7.50031769e-01 -1.24311399e+00 -3.50273609e-01 -4.42287236e-01 -8.51566344e-02 2.53060669e-01 4.17822897e-02 -1.22391045e-01 -1.56384325e+00 -5.30088902e-01 -3.98783803e-01 -6.28419876e-01 6.85529470e-01 2.63228536e-01 1.22703135e+00 -2.08316758e-01 -4.83651794e-02 8.73520225e-03 9.86068189e-01 3.95393044e-01 5.93537688e-01 2.35109746e-01 6.12503231e-01 4.11086082e-01 9.93462145e-01 3.89199376e-01 2.43581712e-01 9.00477290e-01 3.48735571e-01 -4.04490352e-01 -2.45300680e-01 -2.58346260e-01 5.58714092e-01 1.55144691e+00 2.72646844e-01 -4.59050447e-01 -9.95575905e-01 9.09885585e-01 -1.57576382e+00 -6.12129748e-01 -2.12749794e-01 2.13473296e+00 1.00844240e+00 2.33942211e-01 4.55685169e-01 6.79653525e-01 6.11617565e-01 1.21462271e-01 -1.86327517e-01 -2.64290512e-01 -6.31203363e-03 3.17044944e-01 2.20569730e-01 6.25216305e-01 -1.28313434e+00 7.34196842e-01 6.59314108e+00 1.13584375e+00 -9.76418734e-01 3.05385828e-01 4.35186923e-01 -1.49938956e-01 -1.86050922e-01 -2.89854437e-01 -8.41720521e-01 6.07609563e-02 1.45286560e+00 -1.58970535e-01 1.63156956e-01 4.76432890e-01 2.11176679e-01 2.17000633e-01 -1.17878342e+00 9.46184933e-01 1.25351071e-01 -9.08948600e-01 4.62345570e-01 7.97200873e-02 6.25711441e-01 7.85063207e-02 2.28986964e-01 7.59327888e-01 -3.25672328e-01 -8.59451473e-01 5.83629072e-01 5.07383458e-02 8.02205205e-01 -8.90343606e-01 9.50095952e-01 7.45874792e-02 -1.53251600e+00 6.99073076e-02 3.03701788e-01 1.59873024e-01 3.72720629e-01 2.80911773e-01 -1.11880946e+00 7.90367186e-01 5.26413977e-01 3.77187788e-01 -7.07145929e-01 1.30917883e+00 1.41292587e-01 1.10322452e+00 -3.42707932e-01 2.61545837e-01 4.19033676e-01 3.83376360e-01 6.47163451e-01 1.54528236e+00 4.19926941e-01 -2.31734291e-01 4.01491761e-01 1.01324730e-01 -2.34214425e-01 4.63202685e-01 -4.40586925e-01 -5.67220926e-01 5.57035089e-01 1.22155857e+00 -2.37936750e-01 -3.44654918e-01 -3.34267139e-01 7.04786777e-01 -2.86225259e-01 3.43887329e-01 -9.21111047e-01 -5.03813803e-01 5.23713052e-01 9.72932428e-02 -3.95660326e-02 -1.92787960e-01 9.48897973e-02 -6.09748423e-01 9.67613086e-02 -9.95579541e-01 5.39026260e-01 -8.98763180e-01 -1.05909574e+00 6.66992724e-01 1.90011874e-01 -1.40058851e+00 -4.87081379e-01 -1.60487533e-01 -3.08698565e-01 7.50858486e-01 -1.70195901e+00 -1.48896241e+00 -9.41143632e-02 6.22609079e-01 8.39551330e-01 -3.37206483e-01 9.46205318e-01 1.15349543e+00 -3.18477333e-01 1.18757248e+00 2.67346352e-02 -2.86261272e-02 1.25112832e+00 -9.70162451e-01 1.43082082e-01 6.36687279e-01 6.04756713e-01 3.81179273e-01 6.77858651e-01 -2.34223902e-01 -1.20577848e+00 -1.28593087e+00 7.78622746e-01 -4.70126987e-01 9.69011307e-01 -4.66875017e-01 -1.12619174e+00 7.45250762e-01 5.13042569e-01 -3.37590486e-01 1.00458014e+00 2.25281641e-01 -3.11701268e-01 -2.74459988e-01 -6.88470602e-01 3.75712216e-01 3.27264339e-01 -8.68898392e-01 -5.47718287e-01 4.88763243e-01 1.35685432e+00 -3.00540566e-01 -9.31103587e-01 6.05331242e-01 5.10786951e-01 -2.84810483e-01 1.13001168e+00 -5.29418945e-01 2.80274749e-01 -2.02049688e-01 -3.30423832e-01 -1.00858128e+00 9.21923667e-02 -7.28397965e-01 5.80307618e-02 1.93242812e+00 6.29657209e-01 -3.90365273e-01 4.90772396e-01 -1.60913616e-01 -4.74812001e-01 -4.48409945e-01 -9.79838729e-01 -7.88429439e-01 3.67428958e-02 -9.38816845e-01 3.36407572e-01 9.44530666e-01 -1.85431004e-01 6.17646217e-01 -7.52988040e-01 3.94739926e-01 3.28549653e-01 -3.11587185e-01 8.01754594e-01 -1.17015219e+00 -2.31419235e-01 -1.64261281e-01 -2.09540814e-01 -9.66259062e-01 1.09462924e-01 -8.21828127e-01 1.03587672e-01 -1.37826157e+00 2.03417987e-01 -2.90849477e-01 -6.62086427e-01 7.56541014e-01 -1.01132914e-01 8.43948781e-01 1.16245463e-01 8.61023217e-02 -7.48303056e-01 6.88651383e-01 1.06398308e+00 -2.40021020e-01 -4.05185044e-01 -4.33407314e-02 -7.11361766e-01 4.44629073e-01 7.73140490e-01 -4.70376551e-01 -4.85049963e-01 -4.73742425e-01 -1.30427882e-01 2.11519778e-01 6.49371520e-02 -1.18503964e+00 1.80184722e-01 2.01676220e-01 -3.36359888e-02 -9.95942235e-01 8.81556392e-01 -7.29552507e-01 -1.90053329e-01 4.29713055e-02 -6.56484783e-01 3.03808153e-01 1.01806092e+00 7.54979372e-01 -7.15535760e-01 -5.92201948e-02 4.84919667e-01 3.31346422e-01 -2.27472633e-01 2.46766865e-01 -5.22241950e-01 8.68730620e-02 5.42008102e-01 1.68359026e-01 -1.79708555e-01 -7.83577919e-01 -7.95943916e-01 4.34557736e-01 -4.04172599e-01 7.60981679e-01 4.16405261e-01 -1.76114774e+00 -9.33056772e-01 1.12072207e-01 3.66018236e-01 -2.52452344e-01 3.17208350e-01 9.23698366e-01 1.85985997e-01 8.96544337e-01 9.55689698e-02 -8.44716668e-01 -1.64966619e+00 1.51836783e-01 -2.24853277e-01 -4.06764328e-01 2.11810730e-02 8.10302079e-01 9.70292911e-02 -4.06400770e-01 6.67671442e-01 -1.78999618e-01 -3.32794577e-01 2.64964938e-01 5.92578351e-01 1.48430362e-01 3.80697072e-01 -6.98388338e-01 -2.37084046e-01 6.33965373e-01 -3.56065571e-01 -5.97334743e-01 1.11104798e+00 -1.23538569e-01 -1.19800176e-02 9.55557227e-01 1.22497749e+00 1.84696749e-01 -4.23955381e-01 -3.31009150e-01 -1.60386339e-02 -2.85715312e-01 2.86653966e-01 -9.91426766e-01 -7.27465034e-01 1.29132783e+00 8.47623706e-01 2.07916096e-01 1.26455009e+00 -2.42657270e-02 9.35236216e-01 3.76470000e-01 3.28249112e-02 -9.35509920e-01 5.62988698e-01 4.87876683e-01 1.03213954e+00 -1.36969244e+00 -1.84406400e-01 -2.79057443e-01 -5.50711453e-01 7.49660194e-01 7.08500981e-01 3.45603049e-01 3.74096185e-01 2.02355295e-01 1.04327939e-01 1.13984689e-01 -9.11234140e-01 -3.55770141e-01 8.40128779e-01 3.57961923e-01 9.32797909e-01 -2.39726201e-01 7.20827132e-02 4.13940519e-01 -1.35774136e-01 -2.78891146e-01 3.49565953e-01 8.35850894e-01 -3.61030251e-01 -1.35517967e+00 -6.70239329e-01 3.29254478e-01 -7.45432436e-01 -2.83361495e-01 -4.46561545e-01 8.07472169e-01 -3.69571805e-01 1.11882818e+00 3.03394478e-02 -6.20602489e-01 4.58541751e-01 6.00005567e-01 1.41843483e-01 -5.96762717e-01 -6.85958624e-01 7.59353459e-01 2.22692505e-01 -1.75113916e-01 -6.96662128e-01 -4.38905418e-01 -1.09433293e+00 1.27659842e-01 -6.97510064e-01 5.05897760e-01 7.04181612e-01 9.05309796e-01 3.65206689e-01 1.04276168e+00 5.41431367e-01 -7.89776742e-01 -2.80327737e-01 -1.40357399e+00 -2.52531111e-01 7.43392855e-02 6.31950438e-01 -5.55636644e-01 -4.73102242e-01 -3.70527692e-02]
[15.19153881072998, 5.086999893188477]
6b5bb9c7-89f7-4f98-934d-14023039d135
augmenting-message-passing-by-retrieving
2206.00362
null
https://arxiv.org/abs/2206.00362v3
https://arxiv.org/pdf/2206.00362v3.pdf
Retrieval-enhanced Graph Neural Networks for Graph Property Prediction
Graph Neural Networks~(GNNs) are effective tools for graph representation learning. Most GNNs rely on a recursive neighborhood aggregation scheme, named message passing, thereby their theoretical expressive power is limited to the first order Weisfeiler-Lehman test (1-WL). Motivated by the success of retrieval-based models and off-the-shelf high-performance retrieval systems, we propose a non-parametric and model-agnostic scheme called GraphRetrieval to boost existing GNN models. In GraphRetrieval, similar training graphs associated with their ground-truth labels are retrieved as an enhancement to be jointly utilized with the input graph representation to complete various graph property predictive tasks. In particular, to effectively "absorb" useful information from retrieved graphs and "ignore" possible noise, we introduce an adapter based on self-attention to explicitly learn the interaction between an input graph and its retrieved similar graphs. By experimenting with three classic GNN models on 12 different datasets, we have demonstrated GraphRetrieval is able to bring substantial improvements to existing GNN models without comprising the model size and the prediction efficiency. Our work also firstly validates the feasibility and effectiveness of retrieved-enhanced graph neural networks.
['Bernardo Cuenca Grau', 'Qi Liu', 'Song Le', 'Jian Tang', 'Linfeng Song', 'Hanchen Wang', 'Shengchao Liu', 'Dingmin Wang']
2022-06-01
null
null
null
null
['graph-property-prediction']
['graphs']
[ 1.35876238e-01 3.12156200e-01 -4.20611322e-01 -2.30393037e-01 -5.35535216e-01 -3.01708430e-01 6.50398374e-01 6.54132426e-01 -2.75869429e-01 6.02043211e-01 1.07550502e-01 -3.48218352e-01 -5.73052108e-01 -1.26844656e+00 -8.73175561e-01 -3.93938780e-01 -5.26541591e-01 7.57023275e-01 2.01027408e-01 -4.44443434e-01 1.67604368e-02 7.16515124e-01 -1.48104608e+00 1.10519893e-01 8.53291988e-01 8.85847151e-01 1.92049086e-01 8.18284512e-01 -3.39413702e-01 1.06179476e+00 -4.49593157e-01 -4.25955951e-01 1.00075804e-01 -1.54567182e-01 -7.59678304e-01 -2.71725357e-01 5.13565421e-01 -1.81571677e-01 -1.06249630e+00 1.04371428e+00 3.68505627e-01 4.61442530e-01 7.02846825e-01 -1.30799544e+00 -1.03492761e+00 8.63772631e-01 -1.52483866e-01 1.82165593e-01 3.09353173e-01 -3.28417942e-02 1.48539364e+00 -6.38484657e-01 8.34921181e-01 1.23690534e+00 6.46909833e-01 3.22991997e-01 -1.12430763e+00 -5.53593099e-01 3.64758730e-01 2.84437418e-01 -1.46254885e+00 -1.87388554e-01 1.02846551e+00 1.19967200e-01 1.23338997e+00 4.19068366e-01 8.00791502e-01 8.42415690e-01 1.72284484e-01 8.98487508e-01 4.02018160e-01 -4.65107143e-01 -8.21061581e-02 -3.31396237e-02 4.25072253e-01 1.08443832e+00 5.09372771e-01 5.53612113e-02 -3.50344509e-01 -2.30551869e-01 7.25952744e-01 2.99060106e-01 -4.47279245e-01 -5.57061076e-01 -7.13739753e-01 9.15065825e-01 1.23805499e+00 3.91408116e-01 -3.83030474e-01 5.57934225e-01 4.13792133e-01 5.93395233e-01 7.63413370e-01 5.45195282e-01 -1.22748502e-01 5.25306106e-01 -6.13296151e-01 -8.30906034e-02 9.07903671e-01 1.01496947e+00 1.06342959e+00 -1.77218355e-02 -1.77414373e-01 8.13024282e-01 3.82430404e-01 4.10010815e-01 4.60382313e-01 1.24734954e-03 4.46203232e-01 1.22454357e+00 -4.93671477e-01 -1.45340466e+00 -5.49917519e-01 -8.43706548e-01 -1.03719485e+00 -4.62792039e-01 -1.96118951e-01 2.89525419e-01 -1.16637385e+00 1.64500058e+00 3.95117216e-02 3.51884991e-01 -5.93079552e-02 6.42093718e-01 1.29998159e+00 5.31530023e-01 2.58059353e-01 -1.15883406e-02 7.17083156e-01 -1.06474257e+00 -4.23549235e-01 -2.06766620e-01 1.06715679e+00 -1.84477389e-01 9.95753348e-01 5.86996600e-02 -8.40007961e-01 -3.31696093e-01 -1.07545257e+00 -6.41106367e-02 -8.12090993e-01 -1.50590450e-01 1.06034863e+00 3.85797232e-01 -1.51702976e+00 9.65517282e-01 -4.93940830e-01 -5.85803330e-01 4.45974290e-01 5.52165866e-01 -5.32628775e-01 -4.53751266e-01 -1.33256042e+00 7.42364585e-01 7.19393253e-01 1.52522966e-01 -9.82612252e-01 -4.49343354e-01 -8.80154133e-01 3.29699636e-01 5.17299056e-01 -8.02035153e-01 6.91783547e-01 -7.58485794e-01 -1.08001339e+00 5.40222943e-01 2.46625274e-01 -6.64883673e-01 2.63814256e-02 -1.14941746e-01 -6.00572526e-01 3.20113629e-01 -2.84165442e-01 3.65088463e-01 6.85262620e-01 -1.22452092e+00 -1.41709790e-01 -1.02990508e-01 2.64519751e-01 3.96694303e-01 -6.45032346e-01 -3.66928220e-01 -7.41638303e-01 -5.36500990e-01 2.67572343e-01 -7.78261900e-01 -3.72813582e-01 -3.05783510e-01 -4.81823951e-01 -4.16509062e-01 7.17544615e-01 -2.78269231e-01 1.43016934e+00 -1.88330543e+00 1.12862922e-01 7.38445044e-01 8.75770271e-01 5.16190052e-01 -7.95385182e-01 8.06438208e-01 -1.02169983e-01 1.30504519e-01 9.48919058e-02 -9.60331187e-02 -1.15825489e-01 3.77407342e-01 -1.53346881e-01 4.14942950e-01 8.56008232e-02 1.33192062e+00 -1.15223575e+00 -4.10864800e-01 2.30696902e-01 4.71333444e-01 -3.37469220e-01 3.15728247e-01 -4.95986640e-01 -3.23089182e-01 -5.46221673e-01 5.67987382e-01 4.55715656e-01 -9.03628528e-01 3.02009016e-01 -2.02751413e-01 7.70044208e-01 2.13321671e-01 -8.81357789e-01 1.47909915e+00 -3.17113131e-01 4.83474612e-01 -2.31847346e-01 -1.07490349e+00 9.60536838e-01 -9.33444127e-02 2.79186755e-01 -8.15172255e-01 7.49354437e-02 -3.57125513e-02 -1.79069325e-01 -7.79763088e-02 6.96480513e-01 2.22701386e-01 2.48334289e-01 4.17771935e-01 3.62946153e-01 8.84067118e-02 2.55498648e-01 8.61485302e-01 1.48482490e+00 -1.76704437e-01 3.12095821e-01 -1.81917876e-01 4.91347194e-01 -9.75973755e-02 -1.43855169e-01 1.15168905e+00 1.08613797e-01 3.53439510e-01 3.94782275e-01 -4.31484491e-01 -5.33930421e-01 -9.67526734e-01 4.64646757e-01 1.46499848e+00 2.78321922e-01 -8.71933877e-01 -3.42974663e-01 -9.49782491e-01 1.25645325e-01 5.19322991e-01 -6.64059222e-01 -6.72110677e-01 -4.86253053e-01 -6.14820540e-01 5.65694690e-01 4.75375265e-01 2.81994730e-01 -1.19771695e+00 2.56491184e-01 3.04441661e-01 1.80135369e-01 -7.52108812e-01 -3.63326758e-01 2.16963544e-01 -1.11322963e+00 -1.35008168e+00 -3.40676814e-01 -7.67267108e-01 8.53915036e-01 5.35535753e-01 1.56500709e+00 8.57710123e-01 -1.45589799e-01 6.34091139e-01 -4.86664563e-01 -2.28003129e-01 -3.41225833e-01 4.49214667e-01 -1.44561082e-01 -2.47991294e-01 4.47897166e-02 -7.38082349e-01 -5.74580550e-01 -3.55724096e-02 -1.00494492e+00 -8.96671191e-02 9.17231381e-01 8.28600705e-01 5.80363393e-01 1.11733362e-01 6.19839668e-01 -1.32067275e+00 1.08378685e+00 -5.84255278e-01 -4.70013291e-01 5.84165871e-01 -1.02994478e+00 3.39375079e-01 7.46586263e-01 -4.72983152e-01 -5.00384808e-01 -3.55985194e-01 4.85841706e-02 -6.23435855e-01 3.48806262e-01 1.01342702e+00 6.21067360e-02 -5.28350711e-01 7.72431433e-01 2.83156961e-01 -4.30230312e-02 -2.89933741e-01 7.33094513e-01 1.29048660e-01 3.07818979e-01 -3.92581344e-01 8.91868770e-01 1.39340937e-01 2.50761598e-01 -7.32574105e-01 -7.85943508e-01 -6.60847366e-01 -2.85052627e-01 -2.50013769e-01 2.08924800e-01 -7.79494464e-01 -7.01317847e-01 1.46467522e-01 -8.36495042e-01 -3.29783827e-01 -2.14426205e-01 3.03812891e-01 -2.05610901e-01 4.79939431e-01 -8.24612141e-01 -6.45447016e-01 -7.87710547e-01 -6.53812110e-01 8.71553123e-01 6.59153238e-02 2.81796515e-01 -1.34021556e+00 6.62872642e-02 -1.31938994e-01 7.01947153e-01 6.13737628e-02 1.28880596e+00 -1.20935059e+00 -8.68659556e-01 -6.04959846e-01 -6.45954311e-01 1.03908502e-01 -8.77439752e-02 -3.36689621e-01 -7.79296756e-01 -6.19629264e-01 -5.34691393e-01 -3.31507385e-01 1.18793595e+00 2.11148754e-01 1.02228022e+00 -4.17165726e-01 -6.23347759e-01 5.18187642e-01 1.69383788e+00 -2.68296152e-01 6.16954446e-01 7.86857028e-03 1.20050502e+00 2.05117792e-01 2.18659118e-01 -8.24416801e-02 3.47355008e-01 1.58832714e-01 7.22543716e-01 -1.37264311e-01 -3.87627244e-01 -6.06571972e-01 1.34226695e-01 1.17118502e+00 -2.60526575e-02 -8.54774356e-01 -7.52412379e-01 5.11248767e-01 -1.97156537e+00 -5.68919659e-01 -8.65862966e-02 2.06465697e+00 3.59625846e-01 1.65663734e-01 -2.13301852e-01 -2.30510324e-01 7.59335637e-01 5.13910532e-01 -4.38504040e-01 -8.06416571e-02 -1.22011766e-01 3.34455937e-01 5.81425250e-01 3.57524395e-01 -9.36984777e-01 1.20273757e+00 6.26337481e+00 1.05895793e+00 -8.73030186e-01 -1.70626223e-01 3.12548965e-01 1.70250192e-01 -5.18909991e-01 -1.69090703e-02 -4.01236534e-01 -7.58956447e-02 1.24176121e+00 -3.67329866e-01 6.84115529e-01 9.04200613e-01 -5.78990281e-01 2.79271632e-01 -1.18124783e+00 8.95802379e-01 2.65041173e-01 -1.50437021e+00 7.54659951e-01 -5.12275845e-02 4.17665124e-01 3.90212446e-01 -1.39177740e-01 8.13368857e-01 6.92269266e-01 -1.22667181e+00 -4.12754528e-02 7.42515743e-01 6.42992556e-01 -6.63697720e-01 8.83081138e-01 1.43840656e-01 -1.55341947e+00 -6.94371620e-03 -7.71844625e-01 5.10492586e-02 -2.45115072e-01 5.06425321e-01 -1.20510399e+00 1.14541519e+00 3.60679060e-01 8.13239694e-01 -9.20409143e-01 1.03472984e+00 -2.16036350e-01 4.32324171e-01 -3.45538706e-01 -4.84705836e-01 4.16025907e-01 -1.76568508e-01 5.34426808e-01 1.04693234e+00 1.18742086e-01 1.08459532e-01 3.01802665e-01 6.01673007e-01 -7.29116499e-01 4.18689102e-01 -9.23668444e-01 -4.81006444e-01 3.79197359e-01 1.23057222e+00 -7.41523027e-01 -4.49842155e-01 -2.43490949e-01 6.75986409e-01 9.98010218e-01 5.23408651e-01 -5.07874489e-01 -5.45814753e-01 1.66859463e-01 9.66152102e-02 2.27016613e-01 4.49567363e-02 2.82653809e-01 -1.00330377e+00 -1.84022769e-01 -5.91318429e-01 8.36690724e-01 -7.68024683e-01 -1.61254406e+00 9.86263812e-01 -8.22492391e-02 -8.88843536e-01 -8.78819674e-02 -5.60071826e-01 -4.67309207e-01 6.39663160e-01 -1.56747472e+00 -1.47249770e+00 -3.82469416e-01 7.94534266e-01 -2.73244195e-02 -1.79748505e-01 9.13251281e-01 2.94451296e-01 -3.09348047e-01 6.42131269e-01 6.15351535e-02 1.95755810e-01 4.20829892e-01 -1.30767930e+00 4.10253406e-01 6.43244147e-01 5.46037078e-01 8.00100267e-01 3.94937277e-01 -8.68129611e-01 -1.74875867e+00 -1.41531408e+00 5.86580634e-01 -3.28488588e-01 8.11985075e-01 -3.26360226e-01 -1.13776493e+00 8.56454551e-01 -1.08898170e-01 3.93012971e-01 3.37474942e-01 5.51339746e-01 -4.76239800e-01 -2.89224893e-01 -8.51023376e-01 7.04773962e-01 1.39629591e+00 -7.86112130e-01 -2.84891009e-01 6.27810895e-01 1.14925611e+00 -1.82428151e-01 -8.69804680e-01 5.29204190e-01 1.94213673e-01 -6.43160701e-01 1.02331591e+00 -9.84395027e-01 -1.92470010e-02 6.46076500e-02 -7.96666965e-02 -1.25317860e+00 -5.10905743e-01 -4.70818937e-01 -4.96008605e-01 7.78408527e-01 4.52071071e-01 -7.58286297e-01 9.68979537e-01 3.99413556e-01 -7.31783658e-02 -8.20674896e-01 -5.55358231e-01 -6.68065786e-01 -2.86007196e-01 -4.01941538e-01 7.18491375e-01 8.84828269e-01 -1.25331491e-01 5.95596731e-01 -3.23119968e-01 3.88509333e-01 5.11249363e-01 6.11513406e-02 8.41039777e-01 -1.58999896e+00 -2.10001245e-01 -4.93736416e-01 -7.08412647e-01 -1.06205237e+00 3.39032292e-01 -1.46682715e+00 -2.29012787e-01 -1.97184277e+00 3.15908164e-01 -4.32037532e-01 -9.19743896e-01 4.76066500e-01 -2.11851597e-01 1.49058715e-01 7.38840997e-02 2.05060095e-01 -1.24816275e+00 6.19799674e-01 1.17654288e+00 -3.73167396e-01 -2.23072216e-01 -1.35473192e-01 -6.78992689e-01 2.94660985e-01 4.72185135e-01 -5.33876181e-01 -8.62213016e-01 -2.22667456e-01 4.77858216e-01 1.47200167e-01 3.31295639e-01 -8.64542305e-01 5.64320862e-01 2.11854279e-01 1.71504676e-01 -6.12019897e-01 1.73397809e-01 -7.85917580e-01 2.07815722e-01 3.95826042e-01 -4.38769102e-01 1.05761841e-01 1.66790351e-01 1.17922533e+00 -1.44435272e-01 -1.14635654e-01 2.57185549e-01 -1.09494053e-01 -7.03508317e-01 7.63861358e-01 1.18624367e-01 -1.21773295e-01 6.22810483e-01 -2.75245234e-02 -6.71587765e-01 -6.38817728e-01 -6.03718638e-01 2.28798211e-01 2.06524938e-01 2.42581084e-01 9.28912699e-01 -1.27836716e+00 -5.46037138e-01 2.10952178e-01 4.10827905e-01 -1.50327578e-01 1.82932317e-01 3.99073899e-01 -4.67656910e-01 4.94560897e-01 1.46087930e-01 -3.96239489e-01 -1.05665183e+00 9.42500889e-01 3.02008629e-01 -9.07793999e-01 -8.05367649e-01 9.30347025e-01 1.71398029e-01 -6.32219791e-01 3.11800718e-01 -2.47582167e-01 -2.96437234e-01 -2.50667661e-01 2.03858674e-01 2.54517794e-01 3.21906656e-01 -2.41151750e-01 -2.14607343e-01 6.90066442e-02 -5.30154109e-01 5.14058590e-01 1.46157539e+00 8.83523822e-02 -4.15079296e-01 1.25250041e-01 1.36622286e+00 -2.04761237e-01 -6.60019100e-01 -5.30093074e-01 1.67421773e-01 -1.92003429e-01 1.55910164e-01 -6.58103526e-01 -1.21180713e+00 4.53888088e-01 2.86454886e-01 5.18483818e-01 1.02695262e+00 9.99038145e-02 7.05631793e-01 9.57051814e-01 3.84999454e-01 -6.61741436e-01 1.70421153e-01 4.49649036e-01 9.24141943e-01 -1.20492840e+00 3.57276380e-01 -4.70053077e-01 -1.73975885e-01 9.54950333e-01 5.02575159e-01 -4.50055838e-01 7.63051689e-01 -2.38082409e-01 -3.33249152e-01 -7.56859303e-01 -9.66636419e-01 -4.06116962e-01 8.58833790e-01 6.62754476e-01 2.27747291e-01 -8.08026816e-04 -6.44361600e-02 3.08671564e-01 1.15693420e-01 -2.24429786e-01 1.49293065e-01 6.46549881e-01 -4.22986388e-01 -9.31665659e-01 2.83078730e-01 1.02631617e+00 -2.83375476e-03 -4.65492547e-01 -5.80491483e-01 1.20333338e+00 -7.30838597e-01 8.29559267e-01 -1.92212895e-01 -6.45667851e-01 3.12238663e-01 -1.53053656e-01 3.16683710e-01 -7.26069033e-01 -7.02277482e-01 -2.15255052e-01 3.29401135e-01 -7.65591085e-01 -2.12487847e-01 1.22412950e-01 -1.06047881e+00 -5.11507928e-01 -7.54949868e-01 3.03274274e-01 3.36870521e-01 6.98536217e-01 5.69661140e-01 8.00554693e-01 3.34901452e-01 -5.85173607e-01 -4.53057081e-01 -9.92241383e-01 -7.02154815e-01 4.94214237e-01 2.51900733e-01 -5.14467657e-01 -4.63049501e-01 -6.89084768e-01]
[7.081363677978516, 6.2723164558410645]
5da809b2-d9c7-47af-b10b-017509e6d523
ambient-search-a-document-retrieval-system
null
null
https://aclanthology.org/C16-1196
https://aclanthology.org/C16-1196.pdf
Ambient Search: A Document Retrieval System for Speech Streams
We present Ambient Search, an open source system for displaying and retrieving relevant documents in real time for speech input. The system works ambiently, that is, it unobstructively listens to speech streams in the background, identifies keywords and keyphrases for query construction and continuously serves relevant documents from its index. Query terms are ranked with Word2Vec and TF-IDF and are continuously updated to allow for ongoing querying of a document collection. The retrieved documents, in our case Wikipedia articles, are visualized in real time in a browser interface. Our evaluation shows that Ambient Search compares favorably to another implicit information retrieval system on speech streams. Furthermore, we extrinsically evaluate multiword keyphrase generation, showing positive impact for manual transcriptions.
['Max M{\\"u}hlh{\\"a}user', 'Chris Biemann', 'Benjamin Milde', 'Stefan Radomski', 'Jonas Wacker']
2016-12-01
ambient-search-a-document-retrieval-system-1
https://aclanthology.org/C16-1196
https://aclanthology.org/C16-1196.pdf
coling-2016-12
['keyphrase-generation']
['natural-language-processing']
[ 2.94381171e-01 4.94712405e-02 -4.00808193e-02 1.74242586e-01 -1.44094586e+00 -9.78311896e-01 8.31530988e-01 5.32683909e-01 -7.81254828e-01 3.88793111e-01 7.73605168e-01 -4.35046911e-01 -2.02781603e-01 -6.44434869e-01 -3.28932762e-01 -4.58543658e-01 -7.59631619e-02 4.01948273e-01 8.22108865e-01 -5.48434794e-01 3.15125376e-01 2.16102794e-01 -1.95440161e+00 4.15662646e-01 4.10934955e-01 8.47649932e-01 5.76386034e-01 1.31554341e+00 -6.97300553e-01 6.19398236e-01 -1.09759116e+00 -1.76144898e-01 -1.77364424e-01 3.09410989e-01 -9.04125273e-01 -7.02516913e-01 5.10772824e-01 -2.41029173e-01 -6.83575094e-01 8.07726622e-01 8.90536249e-01 3.52724016e-01 2.05990165e-01 -1.11767900e+00 -3.23556840e-01 6.26551986e-01 4.76695180e-01 7.04813361e-01 1.17949927e+00 2.84370547e-03 1.09795487e+00 -1.07359457e+00 6.69758439e-01 1.12360430e+00 6.66307136e-02 2.92431355e-01 -9.97160852e-01 -5.40246427e-01 1.25780508e-01 1.78462043e-01 -1.51862001e+00 -8.89834106e-01 4.47413117e-01 -4.57478203e-02 1.43420768e+00 1.34850717e+00 6.31208241e-01 1.35061145e+00 -1.72751218e-01 7.58315563e-01 3.48871529e-01 -6.10539794e-01 2.27777213e-01 4.43978608e-01 4.54626292e-01 4.51328427e-01 -3.06858152e-01 -3.86702381e-02 -1.36492062e+00 -8.67282033e-01 2.73636490e-01 -4.41474617e-01 -6.27270639e-01 -3.68901674e-04 -1.45046389e+00 2.96543926e-01 -3.19667488e-01 2.52887487e-01 -4.15474206e-01 -3.85354571e-02 4.89226401e-01 5.73935747e-01 3.70965242e-01 4.67712492e-01 -3.46637189e-01 -6.77482724e-01 -7.15796411e-01 4.10071820e-01 1.31174278e+00 1.22208965e+00 4.49581295e-01 -2.04218626e-01 -6.03826225e-01 1.10130394e+00 2.83737987e-01 1.16624010e+00 6.86407983e-01 -1.02867913e+00 5.88873804e-01 1.37587517e-01 2.62186766e-01 -1.08628583e+00 -3.36263962e-02 2.67287660e-02 -6.35834336e-02 -1.34952247e-01 -2.32368216e-01 -1.57031715e-02 -4.97565031e-01 1.30124593e+00 1.78018600e-01 -2.03351483e-01 1.58703506e-01 5.72785139e-01 1.11725068e+00 9.06530201e-01 1.16299503e-01 -3.72866154e-01 1.54909098e+00 -7.23604798e-01 -1.26305485e+00 -1.10472150e-01 1.77808583e-01 -9.34242725e-01 1.38199091e+00 4.40579891e-01 -1.01370442e+00 -3.53585869e-01 -9.10414338e-01 -1.03817739e-01 -8.77109766e-01 -3.04207414e-01 2.85133213e-01 5.42317688e-01 -1.35943067e+00 3.92723717e-02 -5.93665779e-01 -5.69505811e-01 -4.15310413e-01 1.61351278e-01 -5.64544350e-02 6.69222414e-01 -1.60334802e+00 4.39659476e-01 -5.52753657e-02 -6.24186933e-01 -8.03692043e-01 -7.58606553e-01 -5.28543234e-01 2.91289598e-01 6.49818540e-01 -5.22746503e-01 1.78719461e+00 -9.98966470e-02 -1.41174114e+00 4.65658307e-01 -5.66327453e-01 -3.94677311e-01 -2.83080172e-02 -3.27220559e-01 -1.16084564e+00 6.01100087e-01 1.73033029e-01 2.90781081e-01 9.42990005e-01 -1.01112163e+00 -8.80412638e-01 -2.36871988e-01 1.27111018e-01 5.30952811e-01 -8.51992309e-01 3.91064286e-01 -1.24713576e+00 -7.49002278e-01 -9.10019130e-02 -6.56596124e-01 2.92679239e-02 -1.72307879e-01 -3.70325238e-01 -3.53755742e-01 9.59427476e-01 -5.76806903e-01 2.20278573e+00 -2.00679111e+00 -3.14019352e-01 5.91538072e-01 1.41717985e-01 2.37400353e-01 5.38572781e-02 9.88059819e-01 2.41062164e-01 3.60955805e-01 5.13030410e-01 2.63641216e-02 9.94125828e-02 -1.14617951e-01 -5.82147062e-01 -2.43412569e-01 -6.31087899e-01 6.71723783e-01 -9.75997448e-01 -7.60907829e-01 1.59697220e-01 5.64305305e-01 -1.17517792e-01 2.58626074e-01 -9.74380448e-02 -3.08543742e-01 -5.37459612e-01 3.81568313e-01 -2.30352976e-03 -1.39484592e-02 -1.76486045e-01 -5.54831475e-02 -3.63392770e-01 8.77260864e-01 -1.07609296e+00 1.84570634e+00 -7.76260555e-01 8.73467386e-01 2.75986493e-01 -5.68945259e-02 5.11972606e-01 8.36759865e-01 4.67333913e-01 -1.03236687e+00 -4.17741209e-01 1.11952648e-01 -8.82394373e-01 -7.44782805e-01 8.58416498e-01 9.68156099e-01 -3.12568918e-02 7.24688768e-01 -2.12619901e-01 -1.07936449e-01 1.89822897e-01 8.98026705e-01 1.34900832e+00 -5.41546106e-01 -2.49727920e-01 -7.46125355e-02 6.46290362e-01 -7.26205558e-02 -4.32333469e-01 1.07075357e+00 1.17024109e-01 3.39820832e-01 -1.32860377e-01 -1.67298689e-02 -6.28535509e-01 -1.29352891e+00 1.34627476e-01 1.75409412e+00 9.28641856e-02 -1.14272702e+00 -7.07543135e-01 -4.88896847e-01 -1.57576993e-01 9.02417481e-01 -5.94078712e-02 -2.80983120e-01 -3.31359714e-01 3.71933401e-01 6.60439789e-01 1.72336400e-01 -2.23324612e-01 -1.28923154e+00 -9.22411025e-01 1.21237516e-01 -4.75345820e-01 -7.39854693e-01 -9.73405123e-01 3.10154766e-01 -4.86665308e-01 -8.21481884e-01 -8.82763922e-01 -6.55213833e-01 1.58277467e-01 7.02996969e-01 1.47524524e+00 2.56803539e-02 -2.84319878e-01 1.33603513e+00 -5.58072090e-01 -3.86169791e-01 -3.78721803e-01 1.48250356e-01 4.37306315e-02 -3.19720566e-01 3.69730651e-01 -3.43949407e-01 -7.66891479e-01 2.48568624e-01 -9.06175017e-01 -4.11870390e-01 -8.80743712e-02 3.16883832e-01 4.99930978e-01 -1.67503804e-01 3.57477993e-01 -5.38075447e-01 1.42871785e+00 -2.76998878e-01 -4.35468316e-01 4.30162787e-01 -8.46267402e-01 5.57482317e-02 -2.31783167e-02 -2.91173369e-01 -9.06082332e-01 -2.60058314e-01 -3.05323303e-01 -2.94534892e-01 -2.73353934e-01 3.80088180e-01 2.43835710e-02 2.14314148e-01 8.32132518e-01 4.71696645e-01 -3.39868277e-01 -5.79933941e-01 5.25309384e-01 1.20236158e+00 6.03989422e-01 -2.87520289e-01 6.25264585e-01 2.63534904e-01 -8.25902283e-01 -1.42016995e+00 6.93207383e-02 -1.08922684e+00 -1.26867652e-01 -4.59065825e-01 7.39010692e-01 -7.85018921e-01 -7.80949354e-01 -4.80234623e-02 -1.10465240e+00 -4.03551422e-02 -4.29693252e-01 2.81165868e-01 -2.41184682e-01 3.08850586e-01 -2.36111462e-01 -1.30026245e+00 -7.35417068e-01 -8.56051564e-01 1.48923028e+00 -3.49063217e-03 -6.02351248e-01 -6.75235450e-01 2.13807315e-01 2.26385564e-01 4.54023302e-01 -6.81436062e-01 7.18209982e-01 -1.20643282e+00 -4.53026623e-01 -4.00587648e-01 2.31250063e-01 -1.94389611e-01 1.86370924e-01 -1.07200630e-02 -1.43087220e+00 8.34407192e-03 -4.94259268e-01 -1.09206386e-01 5.34090757e-01 -1.53104827e-01 1.23474002e+00 -6.56320572e-01 -6.31920397e-01 6.74526542e-02 6.73095286e-01 7.61319816e-01 2.05664232e-01 3.39532375e-01 1.68497533e-01 6.84417009e-01 5.09884417e-01 5.70137739e-01 2.55701482e-01 9.46892500e-01 1.10235721e-01 1.53486222e-01 -8.15416276e-02 -5.33303857e-01 2.58487761e-01 1.13977575e+00 4.70967680e-01 -6.25441134e-01 -8.03274274e-01 5.32314241e-01 -1.53481889e+00 -9.67875898e-01 8.97769704e-02 2.42851996e+00 9.16774213e-01 1.17315196e-01 1.73650548e-01 1.93156645e-01 4.78676081e-01 2.47641623e-01 -2.58342624e-01 -3.13070953e-01 -2.68296916e-02 3.47341239e-01 3.91796917e-01 7.64043033e-01 -8.26422513e-01 8.39688301e-01 6.74554348e+00 8.39652419e-01 -9.62291360e-01 1.07362598e-01 5.75014353e-02 -4.66721416e-01 -7.18601882e-01 -3.07734489e-01 -5.11674583e-01 3.94077897e-01 1.28130662e+00 -6.28396630e-01 6.16109967e-01 7.61929214e-01 3.20206434e-01 -1.00655012e-01 -8.74197304e-01 1.21848583e+00 6.91298619e-02 -1.38585830e+00 1.34053782e-01 -6.99870242e-03 -8.03395659e-02 1.99577361e-01 1.40996501e-01 3.65406930e-01 1.72988251e-01 -4.43696856e-01 8.49462330e-01 7.10089326e-01 7.46967256e-01 -7.34016657e-01 -7.60002360e-02 2.16555208e-01 -1.21270359e+00 1.45242333e-01 1.04302898e-01 5.01313508e-01 1.87429830e-01 9.57852677e-02 -1.04740059e+00 9.13997218e-02 1.13606572e+00 -2.06167042e-01 -6.53688848e-01 9.23113346e-01 2.11669251e-01 5.42272389e-01 -5.09056211e-01 -4.72402364e-01 4.10878807e-02 1.81197539e-01 1.04774439e+00 1.57003951e+00 3.64528954e-01 -5.55242747e-02 7.48946071e-02 2.48404592e-01 1.81815624e-01 5.86295187e-01 -1.02297068e+00 -2.01224506e-01 1.11656773e+00 1.09302235e+00 -5.18209934e-01 -6.97789371e-01 -2.97227532e-01 1.05736446e+00 -4.17555124e-01 7.70101249e-01 -4.85651940e-01 -1.06597579e+00 6.71028435e-01 5.41054122e-02 2.13249996e-01 -4.93159741e-02 5.83814204e-01 -8.34335506e-01 1.39620006e-01 -1.22625566e+00 4.98332143e-01 -1.17006230e+00 -6.01354837e-01 9.00524855e-01 6.60014823e-02 -1.06078672e+00 -7.31366754e-01 4.88331728e-03 -3.90900612e-01 1.04680717e+00 -9.03725445e-01 -3.51151705e-01 -1.95076615e-01 1.05266857e+00 1.03981853e+00 -2.64084637e-01 1.40040267e+00 4.81817186e-01 -4.56928574e-02 6.04450107e-01 6.70680329e-02 -1.24394327e-01 7.46121645e-01 -1.34938157e+00 8.54924798e-01 3.85850847e-01 9.94066536e-01 1.09517825e+00 7.12504387e-01 -7.53463566e-01 -1.94415820e+00 -4.27510411e-01 1.12848616e+00 -5.45972943e-01 8.90231133e-01 -6.42868757e-01 -8.95685494e-01 4.43124846e-02 6.82677805e-01 -3.86913061e-01 8.14947665e-01 2.03104347e-01 -5.03100097e-01 -2.90402472e-01 -5.94307959e-01 9.78289723e-01 9.79931235e-01 -1.25554979e+00 -5.48308134e-01 5.32998383e-01 1.47593927e+00 -3.43520939e-01 -5.47171891e-01 1.97323645e-03 7.42646635e-01 -6.37273192e-01 1.31201208e+00 -4.67577398e-01 -6.18749082e-01 -1.16024949e-01 -2.76971698e-01 -1.13559926e+00 1.52405754e-01 -1.18163109e+00 -4.82158303e-01 1.12266707e+00 9.20942247e-01 -4.07697260e-01 3.89945984e-01 8.19765389e-01 -1.60668641e-02 -2.78136581e-01 -9.63906646e-01 -4.94334131e-01 -8.44620764e-01 -1.08662438e+00 7.07325459e-01 2.90519804e-01 5.42708933e-01 5.97020328e-01 8.45668241e-02 2.52427787e-01 1.52832597e-01 -3.10093999e-01 6.40744567e-01 -1.14020288e+00 -4.67313305e-02 -4.31091607e-01 9.74525735e-02 -1.23548055e+00 -1.77005798e-01 -7.69845605e-01 6.04330674e-02 -1.34951663e+00 -5.31467319e-01 -2.38618925e-01 -3.70877326e-01 8.05286318e-02 1.43172711e-01 -1.55230686e-01 8.14866871e-02 2.90034294e-01 -8.53123426e-01 1.55328795e-01 6.63611293e-01 -5.03546596e-01 -4.41720039e-01 2.44579837e-01 -5.69209337e-01 4.26149398e-01 4.78306949e-01 -2.84786642e-01 -7.89201736e-01 -3.62668514e-01 5.82200885e-01 3.35953414e-01 2.11977780e-01 -1.05575764e+00 8.13602507e-01 1.70344025e-01 -2.42285021e-02 -8.67483854e-01 6.06231749e-01 -9.85573113e-01 -1.62696198e-01 -1.76705986e-01 -9.59432602e-01 3.97202671e-01 2.04629272e-01 7.63639927e-01 -1.39400735e-01 -3.12242061e-01 -3.89970571e-01 -7.38726812e-04 -5.37860513e-01 2.09934756e-01 -1.16944969e+00 1.86091304e-01 1.18819140e-01 -8.03958625e-02 -1.71578959e-01 -1.05424654e+00 -8.56019855e-01 1.31172374e-01 -1.39691100e-01 8.05381477e-01 8.90672565e-01 -1.21260786e+00 -8.11076462e-02 2.66090214e-01 4.74393189e-01 -5.81583798e-01 -2.14600891e-01 2.19787702e-01 -9.17375758e-02 9.49737132e-01 7.76794553e-01 -4.33897823e-01 -1.53306949e+00 5.38189948e-01 -1.15447119e-01 2.04662323e-01 -6.62896454e-01 8.39044750e-01 -1.79532602e-01 -3.39358114e-02 1.24548483e+00 -3.61070573e-01 -5.25437236e-01 3.49310994e-01 1.04774415e+00 4.55836505e-01 7.09764421e-01 -3.23785275e-01 -2.46328831e-01 9.06109884e-02 1.55994788e-01 -1.10169077e+00 7.60394931e-01 -6.41419530e-01 9.73275527e-02 8.05196643e-01 1.38587487e+00 6.16964638e-01 -4.18009758e-01 -3.56537670e-01 3.69232178e-01 -3.06552231e-01 2.21911862e-01 -9.50369298e-01 -6.58580363e-01 4.75802273e-01 9.30721521e-01 9.37399149e-01 8.85149300e-01 1.41584069e-01 8.74218345e-01 1.14696038e+00 4.43300217e-01 -1.33968782e+00 -2.75821816e-02 5.43294907e-01 9.93824363e-01 -7.27381110e-01 -3.89374226e-01 4.98335510e-02 -5.80381989e-01 1.02042019e+00 2.06363052e-01 7.77126551e-01 8.04928839e-01 5.44312537e-01 6.40757084e-01 -4.98504847e-01 -1.21380591e+00 -2.99991876e-01 3.95775169e-01 7.57416070e-01 3.81544530e-01 -2.76877165e-01 1.64576888e-01 3.47317100e-01 -4.56582397e-01 -5.89156270e-01 -1.80726022e-01 9.81687069e-01 -6.90982103e-01 -9.47986662e-01 -4.98697460e-01 3.41048002e-01 -5.68910778e-01 -6.04398549e-01 -6.44856274e-01 1.54557586e-01 -4.73101139e-01 1.45449626e+00 9.58453715e-02 -4.91616786e-01 6.43476844e-01 5.86358130e-01 -2.98541903e-01 -4.90065128e-01 -9.98642385e-01 5.00011861e-01 5.63441217e-01 -9.58378255e-01 1.08271316e-01 -5.68523467e-01 -1.03016901e+00 4.97280478e-01 -4.74642515e-01 1.01639056e+00 1.07932603e+00 2.11115748e-01 7.10088611e-01 6.32491171e-01 4.94810611e-01 -3.67189616e-01 -3.94548178e-01 -9.30538058e-01 -3.73619981e-02 8.27062130e-02 4.59254771e-01 -1.41532823e-01 -5.29821873e-01 1.73487991e-01]
[14.151311874389648, 6.420164585113525]
2a88eec0-86f4-4bed-a2b6-700948fdc5d3
developing-quantum-annealer-driven-data
1603.0798
null
http://arxiv.org/abs/1603.07980v1
http://arxiv.org/pdf/1603.07980v1.pdf
Developing Quantum Annealer Driven Data Discovery
Machine learning applications are limited by computational power. In this paper, we gain novel insights into the application of quantum annealing (QA) to machine learning (ML) through experiments in natural language processing (NLP), seizure prediction, and linear separability testing. These experiments are performed on QA simulators and early-stage commercial QA hardware and compared to an unprecedented number of traditional ML techniques. We extend QBoost, an early implementation of a binary classifier that utilizes a quantum annealer, via resampling and ensembling of predicted probabilities to produce a more robust class estimator. To determine the strengths and weaknesses of this approach, resampled QBoost (RQBoost) is tested across several datasets and compared to QBoost and traditional ML. We show and explain how QBoost in combination with a commercial QA device are unable to perfectly separate binary class data which is linearly separable via logistic regression with shrinkage. We further explore the performance of RQBoost in the space of NLP and seizure prediction and find QA-enabled ML using QBoost and RQBoost is outperformed by traditional techniques. Additionally, we provide a detailed discussion of algorithmic constraints and trade-offs imposed by the use of this QA hardware. Through these experiments, we provide unique insights into the state of quantum ML via boosting and the use of quantum annealing hardware that are valuable to institutions interested in applying QA to problems in ML and beyond.
['Michael Kim', 'Joseph Dulny III']
2016-03-25
null
null
null
null
['seizure-prediction']
['medical']
[ 2.87977904e-01 -7.79755320e-03 -1.64776593e-01 -6.20600581e-01 -1.62534654e+00 -4.28640246e-01 2.89736032e-01 4.54946935e-01 -6.19819880e-01 8.99802089e-01 -8.59571099e-02 -7.08442032e-01 -1.75298601e-01 -6.89110875e-01 -5.64224899e-01 -6.47310495e-01 -3.22095513e-01 4.92636561e-01 -7.52206966e-02 -4.66615260e-01 3.23331207e-01 3.45418781e-01 -1.51632857e+00 3.93404305e-01 8.15057993e-01 1.07586229e+00 -4.28673953e-01 8.59751761e-01 5.97996891e-01 5.40280759e-01 -7.80625761e-01 -1.87022731e-01 2.86820889e-01 -6.60436094e-01 -8.25407147e-01 -8.08521450e-01 3.48292172e-01 -3.23558748e-01 -1.43407211e-01 8.83194804e-01 9.59342539e-01 4.27961024e-03 5.96544087e-01 -1.25928235e+00 -1.15643926e-01 4.57820892e-01 -1.60703231e-02 5.20789325e-01 7.27772713e-01 3.68529081e-01 8.98092508e-01 -1.56170711e-01 1.04750954e-01 1.10019088e+00 8.19343626e-01 7.24559069e-01 -1.51635695e+00 -9.63228524e-01 -1.03267777e+00 3.06111217e-01 -1.23249841e+00 -5.92329264e-01 3.72497827e-01 2.56429370e-02 1.61851418e+00 3.38991582e-01 8.61728251e-01 8.35572004e-01 1.07071126e+00 6.06772721e-01 1.75895727e+00 -8.28691363e-01 9.92463112e-01 1.65626153e-01 7.12784231e-01 7.88326025e-01 -3.28640521e-01 8.18131864e-01 -1.13547468e+00 -7.46743083e-01 -1.38787165e-01 -6.37533069e-01 1.79863796e-01 -2.22360669e-03 -6.19888604e-01 9.63061452e-01 4.51295257e-01 2.26460472e-02 -6.40645064e-03 4.82873112e-01 2.69238204e-01 3.88822079e-01 2.93950707e-01 5.70035100e-01 -5.51291347e-01 -3.55152518e-01 -9.94503736e-01 2.02418506e-01 6.60587430e-01 5.47693551e-01 7.15537310e-01 -1.29613936e-01 -3.62170935e-02 3.29231530e-01 2.71174312e-01 5.94942868e-01 8.63541365e-01 -5.80430865e-01 -7.24886358e-03 3.64367291e-02 -2.13071167e-01 3.43949497e-02 -9.00501788e-01 -2.52970666e-01 -3.22667807e-01 2.49535739e-01 1.76473394e-01 -1.98412299e-01 -9.73299682e-01 1.55396056e+00 1.88932478e-01 -1.12594351e-01 4.44979787e-01 5.03011465e-01 6.03919446e-01 6.75676107e-01 3.12383890e-01 -2.50030756e-01 1.64615583e+00 -4.36350226e-01 -3.97265017e-01 -3.68413657e-01 1.15154600e+00 -3.92644405e-01 8.62539232e-01 8.16611469e-01 -8.18373621e-01 -2.68622756e-01 -1.63479924e+00 -7.32348338e-02 -2.07150251e-01 -2.85530001e-01 1.08727074e+00 1.71604538e+00 -1.09714127e+00 8.04042757e-01 -1.10567415e+00 -2.82814473e-01 4.27491963e-01 9.73384857e-01 1.99210215e-02 1.61143571e-01 -1.48802841e+00 1.18570101e+00 3.24546546e-01 -3.80728722e-01 -6.17603064e-01 -2.64753610e-01 -7.73997962e-01 -1.02428406e-01 -5.37781894e-01 -5.67931414e-01 1.44105530e+00 -6.32175267e-01 -1.82294631e+00 5.34562290e-01 -6.54548258e-02 -1.01409328e+00 7.99641237e-02 1.34721354e-01 -5.43218315e-01 2.37232998e-01 -1.65317222e-01 6.62783325e-01 6.77561045e-01 -1.38862059e-01 -5.55167615e-01 -6.08099401e-01 -3.38941008e-01 1.21964812e-01 2.45488770e-02 1.36753291e-01 4.80683386e-01 7.74837062e-02 3.34108561e-01 -9.38454926e-01 -4.63907234e-02 -7.18625128e-01 -9.01938155e-02 -2.88305491e-01 1.71259239e-01 -4.95321691e-01 9.13323700e-01 -2.14052486e+00 -5.39027631e-01 3.84236932e-01 -2.21585855e-01 -2.07852453e-01 2.70954996e-01 3.69029909e-01 -5.16642295e-02 -1.74009651e-01 -3.44231457e-01 7.97335953e-02 1.76644906e-01 -6.94254264e-02 -4.17615056e-01 9.12454724e-01 4.71529812e-02 1.10448647e+00 -7.37803221e-01 -4.83565331e-01 3.87478545e-02 2.25777626e-01 -5.81388831e-01 -2.55759656e-01 9.89994854e-02 3.73769403e-01 -2.64439225e-01 8.18269074e-01 5.05662918e-01 4.08359654e-02 1.17454253e-01 -1.18141277e-02 3.19506437e-01 8.29890966e-01 -6.71108842e-01 1.58849573e+00 -3.36323380e-01 5.64531088e-01 -4.43076909e-01 -9.06461060e-01 8.51668477e-01 8.55173245e-02 1.38829783e-01 -1.28577530e+00 3.81772399e-01 4.23610032e-01 1.02647096e-01 -8.78480151e-02 5.02836764e-01 -6.71617925e-01 -4.88515973e-01 7.65678108e-01 2.74463862e-01 -7.21122861e-01 -2.49927998e-01 4.35993001e-02 1.47138298e+00 -2.75348332e-02 6.04003012e-01 -5.00094593e-01 8.49476457e-02 3.22627604e-01 3.68974686e-01 1.12714088e+00 -8.09824228e-01 1.84594452e-01 7.48419538e-02 -3.55015814e-01 -7.83581078e-01 -1.28120756e+00 -1.06251466e+00 1.02913415e+00 1.67881757e-01 -5.98125517e-01 -7.88983583e-01 -5.62595785e-01 -2.07949698e-01 9.89942849e-01 -2.68280745e-01 -5.59928119e-01 -2.67642111e-01 -1.83152390e+00 8.46874475e-01 2.32776031e-01 3.98335040e-01 -9.08310473e-01 -1.17566884e+00 3.50596637e-01 2.29045942e-01 -7.96093762e-01 4.28311557e-01 1.06335878e+00 -1.13661540e+00 -6.02410138e-01 1.14961520e-01 -4.01605755e-01 6.74628541e-02 -6.24211729e-01 9.30007815e-01 -2.39289656e-01 -4.52591956e-01 2.88402170e-01 -4.68578607e-01 -4.40393865e-01 -8.27325702e-01 -5.79630062e-02 3.81160915e-01 -7.47577369e-01 8.72755468e-01 -4.40500140e-01 -5.67951381e-01 -1.25107229e-01 -6.53217912e-01 -2.48225853e-01 5.93204618e-01 1.28632653e+00 3.61236855e-02 1.39025658e-01 6.07667804e-01 -5.98737657e-01 7.72421122e-01 -2.74151713e-01 -5.69385231e-01 2.26793692e-01 -9.16234136e-01 5.04106641e-01 4.70009267e-01 -1.59501001e-01 -6.31384134e-01 2.32660130e-01 -5.69109678e-01 6.12561703e-01 1.15099110e-01 1.46927834e-01 1.82990134e-01 -5.63508272e-01 1.36142874e+00 1.97214097e-01 -2.24913970e-01 8.49702209e-02 2.97128201e-01 1.16089499e+00 2.18678504e-01 -4.03315574e-01 1.43688992e-01 3.95227402e-01 1.47938222e-01 -8.50244939e-01 -3.84290904e-01 -1.36809304e-01 -2.15251312e-01 9.92098898e-02 7.48562634e-01 -7.96872973e-01 -8.65687072e-01 2.33060464e-01 -6.54025733e-01 -3.77816558e-01 -3.08191091e-01 7.76161730e-01 -9.04752731e-01 2.72267699e-01 -9.46861267e-01 -1.09474814e+00 -7.68113852e-01 -1.07540321e+00 1.16567552e+00 2.98573971e-01 -3.51276040e-01 -5.10183692e-01 2.54605502e-01 7.22447276e-01 6.27533253e-03 -3.25710148e-01 9.46361780e-01 -8.65209460e-01 -2.09889516e-01 -3.09163570e-01 2.74426132e-01 6.42503276e-02 -6.17097378e-01 -4.57588822e-01 -1.50077295e+00 -5.86344719e-01 3.03350091e-01 -1.05282378e+00 7.49576032e-01 3.42159152e-01 5.57059705e-01 2.16118008e-01 -2.48863623e-01 5.73955953e-01 1.13074791e+00 3.07938695e-01 5.57653010e-01 5.50882339e-01 -2.46473014e-01 2.14976519e-01 7.09739327e-01 3.68203908e-01 1.28989816e-01 5.40248454e-01 -6.92094713e-02 2.65419960e-01 3.14019710e-01 -1.05089448e-01 6.99428141e-01 7.17288256e-01 6.49255991e-01 1.67949155e-01 -1.19968855e+00 -4.92911227e-03 -1.33585906e+00 -6.48717463e-01 1.40943006e-01 2.22592354e+00 9.17937815e-01 6.04042888e-01 -3.65282595e-03 5.25430739e-01 3.15027833e-01 -4.07591581e-01 -5.53450823e-01 -1.09329903e+00 -5.74283265e-02 1.19031382e+00 8.27642441e-01 4.31031615e-01 -1.09554994e+00 9.40117538e-01 7.09389687e+00 1.30830371e+00 -9.39506948e-01 5.20138323e-01 8.39423597e-01 1.14698015e-01 1.21709652e-01 3.61561894e-01 -7.13511825e-01 3.57625961e-01 1.85223937e+00 7.26612508e-02 6.28332019e-01 6.15130186e-01 -3.24607521e-01 -7.94654310e-01 -1.26505768e+00 1.06641579e+00 -7.69955516e-02 -9.89797890e-01 -6.80661321e-01 -1.39741689e-01 4.29516435e-01 5.10492623e-01 7.96301067e-02 6.41757250e-01 4.59338248e-01 -1.22396922e+00 7.26707041e-01 -1.67739228e-03 7.74775147e-01 -6.97287261e-01 6.82360590e-01 6.39640093e-01 -4.37240243e-01 -4.12944794e-01 -4.44368243e-01 -3.99184346e-01 -1.74400881e-01 5.12848973e-01 -1.11572456e+00 3.47704738e-01 9.29087639e-01 -6.84989020e-02 -7.04805136e-01 8.61253619e-01 -2.99959540e-01 8.43657970e-01 -7.87253320e-01 -5.64978302e-01 6.32741898e-02 5.56410402e-02 4.21352237e-02 1.08329058e+00 -4.82947752e-02 4.36151445e-01 -2.86455482e-01 6.03995979e-01 3.68930191e-01 -8.06507841e-03 -1.04542188e-01 1.93960488e-01 3.79789352e-01 9.75897372e-01 -7.21245527e-01 -4.05396372e-01 1.33200303e-01 7.73079336e-01 1.64677307e-01 -2.61239856e-01 -7.23245919e-01 -3.11893344e-01 -1.35885244e-02 -5.13757825e-01 -2.95429170e-01 2.61864029e-02 -5.78516126e-01 -1.09235847e+00 -3.73841971e-01 -9.71136332e-01 5.05404174e-01 -8.10682595e-01 -1.09663904e+00 5.97802281e-01 1.56884417e-02 -9.53941584e-01 -4.48140979e-01 -4.72661853e-01 -2.12655485e-01 8.11988890e-01 -9.87920761e-01 -7.05916643e-01 2.65179962e-01 2.11340532e-01 6.34575039e-02 2.87439357e-02 1.13253772e+00 -7.38658234e-02 -2.26637706e-01 8.09824288e-01 5.81953764e-01 -3.42537701e-01 5.82130849e-01 -1.02537107e+00 6.52943492e-01 4.94819641e-01 3.95979226e-01 3.84938717e-01 8.79050016e-01 -5.56878209e-01 -1.58222246e+00 -3.70069802e-01 5.79463303e-01 -6.59540951e-01 5.40544689e-01 -6.85946345e-01 -3.85285020e-01 3.73948306e-01 -5.00305742e-02 -1.75116539e-01 8.98249745e-01 1.62112713e-01 -3.45938131e-02 -2.80160069e-01 -1.50680482e+00 1.04690827e-01 4.57933366e-01 -1.06445038e+00 -9.20940816e-01 5.98043680e-01 1.76860631e-01 -4.06511396e-01 -6.99083507e-01 5.96024215e-01 8.68441641e-01 -1.15961123e+00 7.84116507e-01 -3.71423185e-01 -2.16603652e-01 1.05899006e-01 -1.37689769e-01 -1.19692278e+00 2.10388571e-01 -7.27966726e-01 2.13064060e-01 6.31217957e-01 3.95941854e-01 -8.10712755e-01 1.13507450e+00 7.94048071e-01 1.24193348e-01 -7.06793785e-01 -1.65624106e+00 -7.16597676e-01 5.16583025e-01 -9.83286738e-01 4.10788447e-01 4.67125833e-01 9.22409058e-01 4.23427969e-01 -3.15415375e-02 7.21735433e-02 7.29739845e-01 -4.38753180e-02 3.44156176e-01 -5.45423985e-01 -7.39580154e-01 1.28545426e-02 -1.07744503e+00 -7.94680774e-01 6.62782341e-02 -1.03269470e+00 2.62804091e-01 -6.95425034e-01 2.95559525e-01 -6.04652643e-01 -3.59485894e-01 3.37958127e-01 1.18781619e-01 6.19025350e-01 -3.87058109e-02 2.66665965e-01 -2.87177026e-01 4.89808440e-01 1.93213284e-01 -3.34503889e-01 -3.48955035e-01 -5.95600680e-02 -2.12572232e-01 4.01073813e-01 9.64348912e-01 -8.62229228e-01 -7.08937943e-02 1.29592419e-01 4.41233695e-01 4.61571753e-01 2.62473822e-01 -1.50600672e+00 5.36235273e-01 4.08638537e-01 6.05296254e-01 -5.00049591e-01 6.43837452e-01 -4.00459856e-01 -1.80501819e-01 9.04190719e-01 -2.80584961e-01 -8.08254778e-02 4.29137200e-01 3.36439997e-01 1.15004510e-01 -3.87559533e-01 9.35917139e-01 1.92122355e-01 -5.78273058e-01 -4.24545288e-01 -7.11728334e-01 -9.92110297e-02 1.10878575e+00 -1.59534112e-01 -6.16945505e-01 -1.70825332e-01 -6.22726917e-01 6.33061081e-02 3.54557455e-01 -2.30379000e-01 4.52266276e-01 -6.89012110e-01 -3.66889060e-01 5.59182823e-01 5.15857600e-02 -7.55172908e-01 8.01554844e-02 8.90373230e-01 -8.61192703e-01 5.24598241e-01 -3.33127290e-01 -7.73151398e-01 -1.13092589e+00 3.10932308e-01 5.07035613e-01 -9.77942571e-02 -3.21723551e-01 9.16221380e-01 -4.14290637e-01 -5.72800159e-01 -7.52468407e-02 -2.17276603e-01 2.83027172e-01 -2.82248229e-01 2.50507027e-01 1.62392065e-01 7.01707840e-01 -3.01615387e-01 -7.16068149e-01 2.41347045e-01 1.45671755e-01 -7.40868330e-01 1.02991319e+00 3.20169389e-01 -3.91060859e-02 2.93167293e-01 1.08971334e+00 -1.39511481e-01 -7.48043060e-01 2.78101325e-01 1.32931858e-01 1.28099173e-01 3.80051225e-01 -1.07044613e+00 -2.79694293e-02 1.09539747e+00 1.39463091e+00 7.44319707e-02 1.31218481e+00 2.16969308e-02 7.06591606e-01 6.83216810e-01 7.88172960e-01 -1.22085404e+00 -4.01547313e-01 1.92577705e-01 -1.19800806e-01 -1.09924078e+00 5.09524822e-01 8.00334588e-02 -2.39462167e-01 1.18356979e+00 5.20382002e-02 -1.26787096e-01 5.84712625e-01 3.11760992e-01 -3.23936641e-02 -2.43418157e-01 -7.94217885e-01 5.28685823e-02 -2.49830484e-02 3.15050602e-01 1.87902510e-01 4.00674582e-01 -5.27561069e-01 5.28207242e-01 -9.09760594e-01 5.88649251e-02 4.50171798e-01 1.36541450e+00 -5.02977133e-01 -1.13181388e+00 -7.07542956e-01 7.08642006e-01 -4.54450876e-01 -5.27657509e-01 -2.50865281e-01 5.02349317e-01 1.22877747e-01 1.29552138e+00 3.93623263e-02 -7.70337105e-01 -3.27158332e-01 3.04805011e-01 8.14936876e-01 -3.66043448e-01 -9.87642169e-01 -2.34249562e-01 2.12072164e-01 -5.62756240e-01 -3.91893424e-02 -7.77715564e-01 -1.45407915e+00 -2.33888254e-01 -9.39594805e-01 4.65046734e-01 1.18763411e+00 1.13333797e+00 1.15719214e-01 1.49064273e-01 4.20507848e-01 -5.06777525e-01 -1.10424232e+00 -7.47004747e-01 -7.99618721e-01 -2.00412720e-01 2.79706180e-01 -2.87802339e-01 -5.20021260e-01 -5.11237800e-01]
[5.546063423156738, 4.942598342895508]
5aba3db7-854d-4071-aeb1-941dfd43ffcf
towards-generalizable-person-re
2206.09362
null
https://arxiv.org/abs/2206.09362v2
https://arxiv.org/pdf/2206.09362v2.pdf
Towards Generalizable Person Re-identification with a Bi-stream Generative Model
Generalizable person re-identification (re-ID) has attracted growing attention due to its powerful adaptation capability in the unseen data domain. However, existing solutions often neglect either crossing cameras (e.g., illumination and resolution differences) or pedestrian misalignments (e.g., viewpoint and pose discrepancies), which easily leads to poor generalization capability when adapted to the new domain. In this paper, we formulate these difficulties as: 1) Camera-Camera (CC) problem, which denotes the various human appearance changes caused by different cameras; 2) Camera-Person (CP) problem, which indicates the pedestrian misalignments caused by the same identity person under different camera viewpoints or changing pose. To solve the above issues, we propose a Bi-stream Generative Model (BGM) to learn the fine-grained representations fused with camera-invariant global feature and pedestrian-aligned local feature, which contains an encoding network and two stream decoding sub-networks. Guided by original pedestrian images, one stream is employed to learn a camera-invariant global feature for the CC problem via filtering cross-camera interference factors. For the CP problem, another stream learns a pedestrian-aligned local feature for pedestrian alignment using information-complete densely semantically aligned part maps. Moreover, a part-weighted loss function is presented to reduce the influence of missing parts on pedestrian alignment. Extensive experiments demonstrate that our method outperforms the state-of-the-art methods on the large-scale generalizable re-ID benchmarks, involving domain generalization setting and cross-domain setting.
['Qi Tian', 'Ruiming Hu', 'Zheng Wang', 'Wei Liu', 'Xin Xu']
2022-06-19
null
null
null
null
['generalizable-person-re-identification']
['computer-vision']
[ 1.54537857e-01 -5.75108945e-01 1.09871253e-01 -3.97364408e-01 -8.17700624e-01 -4.43202466e-01 5.74813724e-01 -3.24474752e-01 -3.37304711e-01 7.07320452e-01 4.21771675e-01 3.78029436e-01 2.17907965e-01 -5.64106464e-01 -8.53005171e-01 -8.88567686e-01 4.42468196e-01 3.99442762e-01 3.10818627e-02 -1.36250824e-01 -1.33711964e-01 2.54215658e-01 -1.54118657e+00 1.07082762e-01 7.07004249e-01 7.44292617e-01 1.05223432e-01 3.36708814e-01 2.51210272e-01 1.34831443e-01 -6.33006275e-01 -7.16703653e-01 4.89229947e-01 -4.01895881e-01 -3.78601313e-01 6.64268076e-01 7.32769072e-01 -5.49869776e-01 -5.63955426e-01 1.27785397e+00 7.34699786e-01 3.51894021e-01 4.20525849e-01 -1.51003969e+00 -1.00309575e+00 -4.71002655e-03 -8.05195808e-01 1.58067435e-01 5.81160963e-01 3.60190690e-01 5.65651834e-01 -1.00114119e+00 2.97081321e-01 1.55443478e+00 6.13309622e-01 7.45076478e-01 -1.23974192e+00 -8.63845766e-01 6.08967125e-01 5.44202149e-01 -1.57612538e+00 -3.45436901e-01 8.16843927e-01 -3.82662863e-01 7.16555476e-01 1.48833588e-01 5.94549656e-01 1.65851891e+00 -2.41002515e-01 7.07818925e-01 8.86128902e-01 -1.39695108e-01 -3.59206013e-02 4.90496904e-02 1.47984680e-02 1.20734297e-01 6.67225540e-01 2.41982564e-01 -2.20633358e-01 1.18183516e-01 8.17832828e-01 2.87246108e-01 -3.85639459e-01 -5.18947542e-01 -1.34084332e+00 4.41128165e-01 5.17814338e-01 -6.50582388e-02 -8.32689852e-02 -2.35868007e-01 5.65933168e-01 1.48120701e-01 1.25323653e-01 -1.15599543e-01 -3.27321231e-01 1.68020114e-01 -4.04980540e-01 5.07923484e-01 3.25119227e-01 1.46115947e+00 6.46888196e-01 -6.69987649e-02 -1.91537693e-01 9.85793233e-01 1.70322508e-01 7.43599474e-01 5.18220484e-01 -4.70299542e-01 9.96459007e-01 5.01554251e-01 3.86125386e-01 -1.13546002e+00 -3.02027553e-01 -5.24508774e-01 -1.29365849e+00 -2.25914985e-01 6.13507152e-01 -8.15033466e-02 -6.86357021e-01 1.98605573e+00 2.80130208e-01 5.03396988e-01 -7.83191808e-03 1.38603711e+00 9.37228799e-01 3.18489403e-01 1.95255309e-01 -7.19146132e-02 1.62739897e+00 -1.12935901e+00 -3.91794294e-01 -4.43797231e-01 7.87104890e-02 -6.43655837e-01 6.89798653e-01 1.48183763e-01 -7.35293269e-01 -1.08342957e+00 -9.82535720e-01 -1.16077969e-02 -3.65300804e-01 2.84409374e-01 7.39119276e-02 7.75855541e-01 -7.21058547e-01 -7.71651939e-02 -3.45813483e-01 -7.01920092e-01 2.34701991e-01 3.09320122e-01 -6.11360431e-01 -6.23900414e-01 -1.09766614e+00 6.46370053e-01 2.85090357e-01 2.64048696e-01 -7.78046250e-01 -5.13844669e-01 -9.67735589e-01 2.22117733e-02 3.85627866e-01 -1.14073491e+00 8.23218465e-01 -1.15404022e+00 -1.31302631e+00 8.92536342e-01 -1.71053275e-01 -4.55613770e-02 7.75490224e-01 -3.37227911e-01 -8.19013298e-01 -1.73696026e-01 3.48865092e-01 5.11145234e-01 8.95303488e-01 -1.40873551e+00 -7.64861345e-01 -7.01035798e-01 -1.48801729e-02 5.96098900e-01 -1.71296731e-01 1.49288267e-01 -8.28688443e-01 -9.13932979e-01 -7.44498670e-02 -9.84398842e-01 -7.43391514e-02 -1.88172117e-01 -5.43667197e-01 -1.05160780e-01 6.42104864e-01 -9.61424589e-01 8.31331015e-01 -2.12272072e+00 2.92066187e-01 -9.98242721e-02 1.85518116e-02 2.25183815e-01 -3.46512318e-01 1.32703751e-01 -4.14191306e-01 -1.93636015e-01 3.48226987e-02 -5.18518150e-01 -1.24877272e-02 1.06800064e-01 -1.28805250e-01 6.00138664e-01 6.13192767e-02 7.88915217e-01 -9.13478553e-01 -3.42718720e-01 3.37391645e-01 5.62385738e-01 -4.35289204e-01 3.58992130e-01 4.00879532e-01 8.77059817e-01 -3.68777901e-01 6.50057554e-01 1.21316195e+00 -4.16678600e-02 -1.23883694e-01 -6.72141016e-01 2.31040902e-02 -2.51183867e-01 -1.41579175e+00 1.69277930e+00 -1.69844732e-01 7.32436310e-03 4.30179015e-03 -1.04790866e+00 8.37063789e-01 2.62037992e-01 3.33486676e-01 -7.56193161e-01 5.05249463e-02 4.83945571e-03 -2.96489209e-01 -3.65642071e-01 2.74031669e-01 7.72767290e-02 -2.71032065e-01 7.85758495e-02 7.62750208e-02 6.64860904e-01 -7.05642104e-02 -8.99767280e-02 6.75661266e-01 1.73646286e-01 3.51379603e-01 -7.87253305e-02 1.00121367e+00 -3.66098732e-01 9.56579268e-01 6.42732739e-01 -5.48704386e-01 9.32919979e-01 -1.31672785e-01 -6.99144185e-01 -1.18757975e+00 -1.20731878e+00 1.01299010e-01 9.44977105e-01 7.93615282e-01 -2.35008508e-01 -6.43732369e-01 -7.42179692e-01 -1.06663220e-01 2.78860658e-01 -4.24711615e-01 -2.36321643e-01 -7.76061714e-01 -9.73088205e-01 3.47886652e-01 8.15870225e-01 1.18762648e+00 -4.93726820e-01 -3.07844412e-02 1.20826945e-01 -7.19772220e-01 -1.43862462e+00 -1.00058484e+00 -5.30247808e-01 -3.93163294e-01 -1.15211380e+00 -1.24976861e+00 -8.59909773e-01 8.95061135e-01 8.90057623e-01 8.59626234e-01 -2.33616427e-01 -2.49862820e-01 6.15142822e-01 -3.04359645e-01 1.42675787e-01 1.37727320e-01 -2.89071739e-01 4.55322176e-01 3.93822014e-01 7.09927380e-01 -4.03847128e-01 -9.24657226e-01 8.85753989e-01 -4.29198325e-01 1.06096789e-01 4.46891367e-01 1.07296336e+00 6.24803305e-01 1.76459208e-01 3.96935940e-01 -2.75969625e-01 3.51551741e-01 -4.68297333e-01 -3.61981422e-01 3.82840216e-01 -1.80874303e-01 -3.95687938e-01 7.13200510e-01 -6.20874345e-01 -1.26099622e+00 1.95936441e-01 6.64524287e-02 -4.44884270e-01 -6.10581100e-01 -1.97892681e-01 -1.09085512e+00 3.80686224e-02 3.72052521e-01 5.13124526e-01 -4.19315755e-01 -5.87926030e-01 2.59395450e-01 5.88727713e-01 1.03123081e+00 -8.64443660e-01 1.16847253e+00 4.45748031e-01 -2.24964157e-01 -5.61261296e-01 -6.41660035e-01 -7.01541901e-01 -7.98675120e-01 -1.77919790e-01 1.10430205e+00 -1.55427778e+00 -4.18273777e-01 9.09404218e-01 -1.19823945e+00 3.58034343e-01 -4.66736294e-02 4.34505165e-01 -3.99175823e-01 7.12975323e-01 -3.18061739e-01 -4.54450220e-01 -1.61112174e-01 -1.29090440e+00 1.28414941e+00 5.51355779e-01 2.17474416e-01 -6.51844919e-01 -1.42041221e-01 6.49861991e-01 1.34356365e-01 2.15153337e-01 3.81238639e-01 -5.00546575e-01 -5.42015493e-01 -2.64626443e-01 -5.85321009e-01 4.68638867e-01 3.33039075e-01 -5.53479075e-01 -8.71367693e-01 -7.25998342e-01 -2.79382139e-01 -1.23298327e-02 5.39777875e-01 1.93413004e-01 1.04588103e+00 -4.53529924e-01 -5.43951869e-01 8.87346983e-01 1.25995052e+00 -6.46218210e-02 5.54400206e-01 4.67469305e-01 1.08863819e+00 3.72416198e-01 4.47150856e-01 5.10322332e-01 8.42448831e-01 1.17205715e+00 2.10926339e-01 -1.01418905e-01 -2.41595820e-01 -4.78390455e-01 4.16569442e-01 3.77540797e-01 -3.72292519e-01 -2.51163542e-01 -4.79528606e-01 4.32818115e-01 -2.01377583e+00 -1.11523342e+00 3.38488445e-02 2.39219856e+00 2.29762226e-01 -2.80189067e-01 5.48004925e-01 -1.23745434e-01 1.34929299e+00 -6.50581792e-02 -7.57456958e-01 3.17766041e-01 -4.31349099e-01 -4.52825576e-01 4.41075772e-01 8.73550102e-02 -1.35832131e+00 7.22049832e-01 4.48586321e+00 7.39829004e-01 -6.79565847e-01 2.43920475e-01 5.43253422e-01 1.31419331e-01 9.81802940e-02 -2.11270198e-01 -1.08181834e+00 9.28750694e-01 2.26896480e-01 -7.10555315e-02 5.14863849e-01 8.57505679e-01 -4.80831191e-02 3.68191093e-01 -1.14306629e+00 1.65427554e+00 5.39022326e-01 -7.59501576e-01 2.08982125e-01 1.83169767e-02 6.85113132e-01 -3.28689635e-01 6.91401213e-02 4.30596352e-01 2.16641441e-01 -7.96595931e-01 7.37574160e-01 3.84477884e-01 9.79121387e-01 -7.79029012e-01 8.01367819e-01 2.34881490e-01 -1.72366190e+00 -2.54303545e-01 -6.09811544e-01 1.36114836e-01 2.02790871e-01 1.26117215e-01 -2.43196830e-01 9.35929894e-01 1.06989825e+00 8.49118352e-01 -7.47636735e-01 1.09741461e+00 -6.72539026e-02 4.69613411e-02 -2.54896045e-01 5.46091676e-01 -2.23972946e-01 -2.32909858e-01 7.87502110e-01 1.29196787e+00 4.73034918e-01 6.45976290e-02 3.53107959e-01 7.37534523e-01 1.33306563e-01 -1.90147713e-01 -3.40843678e-01 6.96238637e-01 5.28969944e-01 1.12450612e+00 -1.92995489e-01 -1.97654441e-01 -7.64847457e-01 1.51734674e+00 1.26568779e-01 7.93792844e-01 -9.60233510e-01 1.09095991e-01 9.72685814e-01 1.04157083e-01 4.84262049e-01 -7.57032037e-02 3.95938270e-02 -1.80452847e+00 2.82120645e-01 -9.66486037e-01 6.82880163e-01 -5.65664470e-01 -1.89409316e+00 4.68356550e-01 8.26063305e-02 -1.57791674e+00 -1.82800349e-02 -5.43192685e-01 -6.29413486e-01 1.01035893e+00 -1.38655651e+00 -1.75796437e+00 -8.43674362e-01 1.11030686e+00 6.73794389e-01 -4.79576886e-01 5.73553205e-01 7.29415476e-01 -1.03459847e+00 1.19156778e+00 -5.47276512e-02 4.68981326e-01 9.81292307e-01 -9.53134000e-01 4.41361904e-01 1.17884660e+00 -3.97440672e-01 6.04361296e-01 4.86636043e-01 -5.39573371e-01 -1.34825718e+00 -1.51897895e+00 7.68897712e-01 -4.81146127e-01 2.13509202e-01 -4.38536376e-01 -6.37886047e-01 8.66013885e-01 -2.27654025e-01 3.39381844e-01 6.13563657e-01 -3.82218249e-02 -7.62020469e-01 -3.73072177e-01 -1.13694990e+00 5.06834567e-01 1.46002448e+00 -3.55633289e-01 -4.97862995e-01 2.21462429e-01 4.09481615e-01 -4.78984088e-01 -7.00458109e-01 3.26129347e-01 4.57576543e-01 -8.85452330e-01 1.54259670e+00 -6.22809887e-01 -2.43151113e-02 -6.48160756e-01 -3.79639655e-01 -1.28326023e+00 -7.56018937e-01 -3.98544729e-01 1.19662143e-01 1.64505947e+00 -3.74666274e-01 -7.43300557e-01 3.97681504e-01 6.82831407e-01 2.29275841e-02 -9.08254012e-02 -1.16551495e+00 -1.01016915e+00 5.58154145e-03 7.07924888e-02 1.07083881e+00 8.01259220e-01 -3.26693177e-01 3.97791773e-01 -7.46818304e-01 5.73976696e-01 9.38660562e-01 4.51867469e-02 1.18603873e+00 -1.00295889e+00 -2.98955888e-01 -3.28697324e-01 -7.56195366e-01 -1.45029271e+00 1.00646041e-01 -5.74799240e-01 -6.56579956e-02 -1.15984833e+00 6.80695355e-01 -3.08182329e-01 -2.38705263e-01 1.10825531e-01 -5.35491765e-01 1.26918957e-01 4.00436372e-01 3.42359036e-01 -7.66148627e-01 6.84645951e-01 1.17691696e+00 -3.48398089e-01 1.59227133e-01 1.21721201e-01 -9.08104420e-01 6.39195144e-01 5.12223721e-01 -2.04993952e-02 -2.44088277e-01 -6.81442261e-01 -8.95095989e-02 -1.51808217e-01 1.01951861e+00 -1.40513778e+00 3.36226940e-01 -3.71601135e-02 9.40914035e-01 -3.89599413e-01 3.93406421e-01 -8.81643236e-01 2.49620721e-01 7.69650564e-02 1.24996640e-01 2.24758878e-01 1.47899939e-02 9.82602835e-01 -2.02821463e-01 2.34623402e-01 8.37406158e-01 -1.60468444e-01 -8.83184314e-01 6.71185553e-01 1.58206582e-01 6.94790930e-02 1.03696585e+00 -4.26495999e-01 -5.07733405e-01 -3.25012743e-01 -5.41648746e-01 3.16455066e-01 6.05904639e-01 7.86172092e-01 5.73519588e-01 -1.68596959e+00 -1.04426599e+00 4.23169881e-01 4.86140996e-01 1.61780149e-01 8.15293372e-01 6.36059165e-01 5.03486991e-02 2.93289393e-01 -3.52013111e-01 -6.51142716e-01 -1.22591233e+00 7.69101441e-01 5.11800468e-01 -1.34951279e-01 -7.58371890e-01 8.44368219e-01 8.26609910e-01 -4.41362083e-01 1.63205534e-01 1.03203356e-01 -1.45238355e-01 -2.17616633e-01 7.84093142e-01 5.11784792e-01 -2.08242089e-01 -1.28025854e+00 -5.79607725e-01 9.37509596e-01 -1.58385515e-01 2.79556960e-01 8.80351543e-01 -5.94880819e-01 3.56613100e-01 -1.52221784e-01 1.09793329e+00 -2.60122895e-01 -1.60799932e+00 -5.66653371e-01 -6.02766216e-01 -6.83026791e-01 -5.26645660e-01 -7.72680938e-01 -9.57366526e-01 7.21242428e-01 8.65570366e-01 -4.26093727e-01 1.24545693e+00 -9.73157361e-02 1.04961705e+00 7.74374008e-02 5.17581642e-01 -1.09942102e+00 7.54414126e-02 2.98557699e-01 9.00100052e-01 -1.46769106e+00 -5.15061021e-02 -4.50141847e-01 -8.12108159e-01 8.98329556e-01 9.62134063e-01 -9.08372328e-02 3.89422297e-01 -2.03857914e-01 -1.09863184e-01 4.03046966e-01 -9.57747549e-02 -2.17957959e-01 3.38587791e-01 1.20927370e+00 -1.26146525e-01 2.70547956e-01 2.00024351e-01 1.20968401e+00 -1.61770806e-01 -2.11844802e-01 2.42246799e-02 2.94567943e-01 -1.59005541e-02 -9.81146216e-01 -7.04003692e-01 -8.79639164e-02 1.30227739e-02 1.88388258e-01 -2.22888425e-01 8.08360219e-01 7.49249578e-01 1.06407726e+00 8.39236081e-02 -5.78984380e-01 5.73448658e-01 -3.49934578e-01 6.21515214e-01 -2.92078644e-01 -3.26815456e-01 -9.15551372e-03 -1.57249346e-01 -5.74708939e-01 -3.99493098e-01 -9.33434367e-01 -7.03881681e-01 -3.84775341e-01 -1.39792025e-01 -2.69337684e-01 1.29283383e-01 9.08407629e-01 3.91632974e-01 4.24689144e-01 5.17071962e-01 -1.06661844e+00 -5.93267798e-01 -8.18167627e-01 -4.56843346e-01 1.03256834e+00 3.59080970e-01 -9.30062711e-01 -2.16602117e-01 3.03951412e-01]
[14.668105125427246, 0.9311332106590271]
1f6528de-77a8-46d9-b6f3-0399c54fe253
tace-time-aware-convolutional-embedding
null
null
https://openreview.net/forum?id=hopfHdHZGYe
https://openreview.net/pdf?id=hopfHdHZGYe
TaCE: Time-aware Convolutional Embedding Learning for Temporal Knowledge Graph Completion
Temporal knowledge graph completion (TKGC) is a challenging task to infer the missing component for quadruples. The key challenge lies at how to integrate time information into the embeddings of entities and relations. Recent TKGC methods tend to capture temporal patterns via linear or multilinear models, which are fast but not expressive enough. In this study, we propose a novel time-aware convolutional embedding model (TaCE) to represent the time-dependent facts in the task of TKGC. It highlights its novelty to feasibly convert timestamps as temporal convolutional filters to fully interact with entities and relations and learn temporal patterns in knowledge graphs (KGs). An extensive comparison proves that our model outperforms the state-of-the-art models on three public benchmark datasets of ICEWS14, ICEWS05-15 and GDELT. Results also demonstrate good temporal expressiveness and computation efficiency performed by our TaCE.
['Chen Peng', 'YanFeng Hu', 'Hong Shen', 'Jin Luo']
2021-09-29
null
null
null
null
['temporal-knowledge-graph-completion']
['knowledge-base']
[-6.71680748e-01 -6.49200976e-02 -5.96442342e-01 -2.34385863e-01 -4.17363159e-02 -7.94662714e-01 7.30953693e-01 2.50036687e-01 -3.91549885e-01 6.28745556e-01 4.14965689e-01 -2.97237426e-01 -6.25679493e-01 -1.02061033e+00 -8.01791847e-01 -3.27116013e-01 -9.24929798e-01 5.17787158e-01 2.85298884e-01 -4.47272241e-01 -4.37941819e-01 3.51276666e-01 -1.16328561e+00 1.55177370e-01 7.52023876e-01 9.68318284e-01 -5.78932881e-01 6.08214140e-01 -1.82193816e-01 1.19076753e+00 -9.66235176e-02 -8.25603247e-01 2.07814872e-01 3.11661929e-01 -1.02911651e+00 -6.33067489e-01 1.02252722e-01 -3.19076955e-01 -1.39142704e+00 2.99440354e-01 1.10349990e-01 2.90380955e-01 4.26924199e-01 -1.89663064e+00 -1.00588036e+00 8.20434213e-01 -2.56104320e-01 4.22191024e-01 3.45861375e-01 -5.17505035e-02 1.22697973e+00 -5.99519551e-01 1.09667230e+00 9.69693184e-01 1.07458341e+00 4.49074619e-03 -1.19546878e+00 -6.09903097e-01 4.78339136e-01 9.49984848e-01 -1.66184556e+00 -1.12863779e-01 6.58950031e-01 -4.38245416e-01 1.39190066e+00 2.21920893e-01 8.69968176e-01 1.24365163e+00 1.86500192e-01 6.87069178e-01 5.14764667e-01 7.84099475e-02 -4.33972850e-02 -3.99835110e-01 3.77641559e-01 8.66633892e-01 2.34089822e-01 2.60340929e-01 -1.07785642e+00 -6.85119629e-02 4.72844481e-01 1.15876555e-01 -1.71431482e-01 -3.88960630e-01 -1.54505670e+00 5.76746881e-01 6.71259940e-01 2.68578678e-01 -4.40539688e-01 7.03760505e-01 7.52865553e-01 5.30516505e-01 7.04413474e-01 1.69604570e-01 -8.54814470e-01 -3.11826050e-01 -5.97222865e-01 3.79770130e-01 1.03189540e+00 1.19700253e+00 4.97327119e-01 -1.21733971e-01 -3.84611011e-01 2.18967140e-01 -7.07942769e-02 1.72491834e-01 2.92683076e-02 -5.17954230e-01 6.61705911e-01 8.42245579e-01 3.23840678e-01 -1.17691803e+00 -6.77282333e-01 -2.73816198e-01 -6.25461936e-01 -5.71419597e-01 3.17481071e-01 3.56549211e-02 -9.27147150e-01 1.79941607e+00 5.24426520e-01 9.43496227e-01 1.41773731e-01 7.01724470e-01 9.63145137e-01 7.89014637e-01 1.57693595e-01 -5.93406036e-02 1.26808703e+00 -9.00070131e-01 -1.01562583e+00 5.29314540e-02 8.11262190e-01 -1.65041611e-01 5.37438631e-01 -3.29405852e-02 -6.72356546e-01 -2.38108501e-01 -1.00563502e+00 -4.42341149e-01 -1.06672943e+00 1.79854125e-01 1.52116776e+00 1.01887502e-01 -9.65796828e-01 8.34170938e-01 -1.25972354e+00 -3.23527277e-01 1.20883666e-01 3.82147491e-01 -6.06737912e-01 -1.08360015e-01 -1.93983614e+00 9.71996427e-01 8.53448868e-01 4.15989965e-01 -7.26940453e-01 -1.29962242e+00 -1.13258660e+00 8.58682394e-02 5.87999284e-01 -7.40178347e-01 9.33454633e-01 -1.21116741e-02 -1.10606647e+00 5.71071386e-01 -1.12047136e-01 -9.33279693e-01 5.93075275e-01 -1.03577256e-01 -1.15321589e+00 4.74641286e-02 1.05891414e-01 2.71041185e-01 4.74202216e-01 -7.31597424e-01 -5.07407367e-01 -2.58894473e-01 3.01642835e-01 -2.83414096e-01 -4.77374494e-01 -3.18345696e-01 -1.10159183e+00 -5.60500920e-01 -1.03424534e-01 -9.38039064e-01 -2.25685723e-02 -1.29186705e-01 -3.53623539e-01 -5.86787641e-01 8.63248527e-01 -9.60038722e-01 1.87776089e+00 -1.93020558e+00 3.41234595e-01 -3.97936739e-02 4.11014259e-01 1.36760816e-01 -4.01065350e-02 1.04785538e+00 -1.57558724e-01 -4.96552698e-02 -7.92541057e-02 -4.51912731e-01 4.76468176e-01 7.55990684e-01 -5.79535186e-01 4.43380892e-01 3.54969740e-01 1.53790617e+00 -1.36834168e+00 -4.57587063e-01 1.45878673e-01 5.85204840e-01 -1.95752233e-01 -7.86633417e-02 -5.02250731e-01 1.40189389e-02 -3.79692405e-01 6.32361591e-01 5.76970339e-01 -3.74528140e-01 6.10769928e-01 -5.94332874e-01 -1.86121657e-01 3.91886294e-01 -1.03165829e+00 2.04775095e+00 -3.68520409e-01 5.40368021e-01 -6.80553734e-01 -6.60802841e-01 4.47997600e-01 4.67084348e-01 8.89008939e-01 -8.25919211e-01 -1.36150345e-01 -4.58674058e-02 -3.79512578e-01 -6.40923321e-01 8.76424074e-01 2.81975746e-01 -3.04604918e-01 3.58971864e-01 4.64448035e-01 4.07584727e-01 4.54313338e-01 6.35016799e-01 1.32410336e+00 4.96806145e-01 6.02724478e-02 3.67416777e-02 3.07821948e-02 9.47841704e-02 9.04644966e-01 1.17285825e-01 -7.33912513e-02 8.73570913e-04 7.77454734e-01 -1.03145421e+00 -7.17195928e-01 -1.20731843e+00 2.76321083e-01 9.29416001e-01 -5.65675180e-03 -1.02256036e+00 1.59656078e-01 -9.58492517e-01 5.53414464e-01 5.37968218e-01 -1.18923092e+00 -2.60728359e-01 -5.85612476e-01 -5.41349292e-01 8.57919574e-01 9.03285742e-01 2.83954382e-01 -4.51825827e-01 -1.92811191e-01 4.77856189e-01 -4.43900615e-01 -1.59999955e+00 -5.02827883e-01 8.53692219e-02 -5.41796744e-01 -1.42166495e+00 -1.53218508e-01 -4.31119770e-01 2.99336910e-01 -1.24346418e-03 1.14940393e+00 -2.95610070e-01 -3.48998010e-01 4.93641078e-01 -5.55447340e-01 -3.37814361e-01 3.77511501e-01 2.19919667e-01 1.98214725e-01 3.57934162e-02 4.50816095e-01 -9.12492692e-01 -6.13163292e-01 1.18059263e-01 -8.90085101e-01 1.35959713e-02 8.81864876e-02 5.86101592e-01 4.85935658e-01 3.79335433e-01 3.99324089e-01 -8.55437338e-01 3.75396818e-01 -7.64272809e-01 -7.05996811e-01 6.10744357e-01 -8.73135388e-01 3.57804984e-01 6.36586130e-01 -4.91333425e-01 -9.35591161e-01 -9.68694761e-02 4.55396295e-01 -8.83078277e-01 6.99978232e-01 9.49748695e-01 3.36005360e-01 6.46678135e-02 4.62054491e-01 2.50505626e-01 -5.28765321e-01 -3.13912123e-01 9.31518078e-01 -1.50443316e-01 8.23031247e-01 -8.53799999e-01 9.99179602e-01 8.23658228e-01 1.84692994e-01 -3.07860374e-01 -7.63660967e-01 -6.62961543e-01 -8.31429720e-01 -2.05888022e-02 5.82432508e-01 -1.12263584e+00 -9.75005448e-01 2.19615102e-01 -1.21925116e+00 -4.38706875e-01 -3.28869522e-01 5.08215606e-01 -2.83609629e-01 3.88683647e-01 -7.50114501e-01 -5.24583340e-01 -3.59017849e-01 -3.79126757e-01 9.41348374e-01 -8.71661603e-02 -8.32836553e-02 -1.35296071e+00 2.62931108e-01 4.99271564e-02 2.74054825e-01 8.48106325e-01 9.22739744e-01 -3.32696199e-01 -8.25481653e-01 -2.04022169e-01 -3.26600581e-01 -2.45356634e-01 6.16086796e-02 3.28324556e-01 -6.78882837e-01 -1.57660499e-01 -8.86819720e-01 -1.56825393e-01 1.11950076e+00 5.40882051e-02 8.65107000e-01 -4.18891311e-01 -5.41750491e-01 9.44562495e-01 1.51212180e+00 -2.20644493e-02 6.70163333e-01 2.20696837e-01 9.13896680e-01 3.22344393e-01 7.34413564e-01 5.40446281e-01 1.23674452e+00 6.39817059e-01 3.96438420e-01 3.96919996e-01 -1.19218342e-01 -6.30640626e-01 2.41418108e-01 1.06769514e+00 -4.03882623e-01 -2.88612336e-01 -1.16026175e+00 1.18378603e+00 -2.45221925e+00 -1.07862318e+00 -4.22020644e-01 1.74780607e+00 8.50409865e-01 -1.32814899e-01 -9.64839235e-02 1.37723815e-02 1.66685596e-01 5.53574204e-01 -3.98367375e-01 -1.89982355e-01 -1.23531617e-01 2.76504695e-01 8.47080588e-01 2.36562565e-01 -1.22630405e+00 1.11504328e+00 5.78192091e+00 5.42402506e-01 -7.85084248e-01 2.36815423e-01 -1.68726802e-01 -1.50727808e-01 -3.86764854e-01 2.42008865e-01 -5.86096525e-01 3.98695976e-01 1.19235933e+00 -5.53155184e-01 5.22210419e-01 3.57868224e-01 -1.34079665e-01 2.99341440e-01 -1.33889496e+00 9.29059803e-01 -2.65996397e-01 -1.64645553e+00 -1.72706231e-01 -2.10013896e-01 9.59951699e-01 1.57940388e-01 -2.47240271e-02 8.00124764e-01 7.52961695e-01 -9.75861192e-01 6.31314218e-01 8.53771985e-01 9.85858619e-01 -6.73005521e-01 4.43159282e-01 -2.31536135e-01 -1.92442536e+00 -4.37292829e-02 -1.06146038e-01 1.69885941e-02 2.86406308e-01 5.17633021e-01 -1.01571774e+00 1.57184649e+00 8.37712705e-01 1.21371949e+00 -4.56838250e-01 7.73979366e-01 -4.53433603e-01 5.59710801e-01 -3.66636753e-01 3.50982606e-01 4.47748840e-01 -3.85437794e-02 3.14473867e-01 1.21693754e+00 1.51114032e-01 8.26108232e-02 -1.42346919e-01 7.45173097e-01 -3.50985140e-01 -4.17970002e-01 -6.72092199e-01 -6.69640481e-01 5.78068495e-01 1.16188681e+00 -3.62857938e-01 -1.30766943e-01 -5.27238131e-01 1.08758819e+00 7.02441931e-01 6.15443110e-01 -1.33591747e+00 -3.11190605e-01 9.28025544e-01 -2.02929690e-01 4.41792488e-01 -7.49180317e-01 1.58256605e-01 -1.46805978e+00 2.13794708e-01 -3.58733803e-01 1.12176645e+00 -6.65481746e-01 -1.23059320e+00 3.20115149e-01 1.62489325e-01 -1.10570991e+00 -1.23923376e-01 -4.70651746e-01 -4.74687010e-01 5.75071573e-01 -1.72608674e+00 -1.82339776e+00 -2.74173290e-01 9.37035799e-01 -8.11427012e-02 3.09778661e-01 7.85452425e-01 6.55222416e-01 -5.76107919e-01 5.33104718e-01 1.04145043e-01 3.42918932e-01 5.89198351e-01 -1.31569350e+00 6.83174908e-01 1.05993593e+00 2.71519303e-01 7.03223348e-01 5.01002371e-01 -7.25465953e-01 -2.06703043e+00 -1.59875906e+00 1.28675008e+00 -6.92832053e-01 1.28423524e+00 -3.90048951e-01 -7.10963666e-01 1.44286668e+00 7.28746876e-02 5.97371221e-01 5.57688951e-01 6.67770207e-01 -9.76566851e-01 -3.58856708e-01 -5.87044537e-01 6.17047608e-01 1.49676228e+00 -8.88347089e-01 -5.90714931e-01 3.84245068e-01 1.33686817e+00 -5.62527776e-01 -1.57368565e+00 5.55096149e-01 5.83164692e-01 -3.47329855e-01 1.14620590e+00 -1.12234592e+00 2.93626755e-01 -5.38351536e-01 4.47503403e-02 -1.31311798e+00 -3.06673795e-01 -8.36297870e-01 -1.08738256e+00 1.22582960e+00 2.60404527e-01 -6.30780220e-01 6.61776245e-01 5.64799190e-01 -2.91378312e-02 -7.16369271e-01 -1.04357326e+00 -1.22198820e+00 -2.77860850e-01 -6.79025888e-01 9.77225244e-01 1.64367580e+00 2.26246744e-01 -5.51448725e-02 -5.90884805e-01 5.50475478e-01 4.45305884e-01 5.08271813e-01 5.61522245e-01 -1.25127947e+00 1.74123161e-02 -1.23479730e-02 -6.21834338e-01 -4.53985482e-01 4.07112807e-01 -1.00065005e+00 -6.88645184e-01 -1.90718484e+00 -2.12082207e-01 -4.03979242e-01 -5.23110390e-01 8.86206210e-01 4.40723822e-02 -2.99806118e-01 -1.55976281e-01 -1.37407511e-01 -8.75140011e-01 8.78108621e-01 9.43027794e-01 -5.16230464e-01 -9.45261642e-02 -4.19268727e-01 -2.43326962e-01 1.07364967e-01 4.68271822e-01 -3.77559453e-01 -7.80461252e-01 -7.35076904e-01 8.12407255e-01 2.19760746e-01 5.54637611e-01 -4.69163746e-01 8.04890871e-01 -1.40253425e-01 4.71819527e-02 -8.14333141e-01 4.81214076e-01 -8.67284894e-01 7.21159637e-01 2.54105300e-01 1.80737421e-01 2.08783582e-01 4.56500709e-01 1.20702875e+00 -4.47576255e-01 6.56712711e-01 -1.76376179e-01 2.44776666e-01 -1.43711901e+00 9.33471501e-01 2.29607746e-01 2.88785417e-02 1.13916099e+00 1.96645379e-01 -7.42232919e-01 -1.59928203e-01 -8.79237235e-01 9.50110734e-01 -5.71506843e-02 7.37227917e-01 5.74475288e-01 -1.77260888e+00 -4.72448409e-01 -3.81824464e-01 6.62459135e-01 -3.89244407e-02 5.67817032e-01 1.23970985e+00 -4.19281244e-01 7.04115450e-01 1.59521297e-01 -1.10135496e-01 -8.48570526e-01 9.42346275e-01 1.49930477e-01 -7.56576061e-01 -7.32531369e-01 7.68151104e-01 -4.12924975e-01 -5.30643284e-01 2.49026939e-01 -8.01676571e-01 -7.36365747e-03 3.57055455e-01 1.01135544e-01 5.71973205e-01 2.34401450e-01 -2.39745274e-01 -7.45996833e-01 2.54188657e-01 -3.27351764e-02 1.85103238e-01 1.71802425e+00 1.36199072e-01 -2.39439592e-01 4.56642658e-01 1.32607448e+00 -2.50861496e-01 -1.08794355e+00 -6.50226295e-01 2.69083291e-01 -2.95868397e-01 -9.03822854e-02 -6.63347840e-01 -1.08391094e+00 4.51440573e-01 1.56345498e-02 8.74156505e-02 1.05912590e+00 -1.78661749e-01 1.01730251e+00 5.30482709e-01 6.74672425e-01 -1.14205396e+00 -2.59807587e-01 7.05343723e-01 8.09187472e-01 -9.68084633e-01 7.24324360e-02 -4.95909184e-01 -6.57410741e-01 1.13013887e+00 4.87497926e-01 1.03949681e-01 9.19946551e-01 7.05585778e-02 -4.93954450e-01 -6.01574600e-01 -1.25706935e+00 -5.61960995e-01 4.73217398e-01 5.09719014e-01 -4.45951000e-02 3.10146034e-01 -3.56914848e-01 8.16209376e-01 -1.12627283e-01 2.76710302e-01 1.10717393e-01 9.10188556e-01 5.79617620e-01 -1.05592108e+00 2.93067038e-01 2.42177904e-01 -2.04272464e-01 -5.20880371e-02 -2.40055650e-01 1.20612741e+00 2.41174534e-01 8.21387231e-01 -1.13407083e-01 -8.80709112e-01 4.82605398e-01 1.36230811e-01 4.46494520e-01 -1.56257048e-01 -4.69906449e-01 -7.40406275e-01 5.02086639e-01 -9.86997306e-01 -2.99513012e-01 -4.09363717e-01 -1.38082325e+00 -7.12095320e-01 6.83755521e-03 1.82849795e-01 3.11348855e-01 5.01220882e-01 7.08317518e-01 9.00639713e-01 3.32526922e-01 -2.64262736e-01 -7.69637078e-02 -4.70771790e-01 -7.18794584e-01 4.25860435e-01 2.82393396e-01 -9.60527778e-01 -2.98317652e-02 -5.67574911e-02]
[8.550195693969727, 7.9062724113464355]
b3720b7a-b238-4097-9324-388304636ec5
playing-20-question-game-with-policy-based
1808.07645
null
https://arxiv.org/abs/1808.07645v3
https://arxiv.org/pdf/1808.07645v3.pdf
Playing 20 Question Game with Policy-Based Reinforcement Learning
The 20 Questions (Q20) game is a well known game which encourages deductive reasoning and creativity. In the game, the answerer first thinks of an object such as a famous person or a kind of animal. Then the questioner tries to guess the object by asking 20 questions. In a Q20 game system, the user is considered as the answerer while the system itself acts as the questioner which requires a good strategy of question selection to figure out the correct object and win the game. However, the optimal policy of question selection is hard to be derived due to the complexity and volatility of the game environment. In this paper, we propose a novel policy-based Reinforcement Learning (RL) method, which enables the questioner agent to learn the optimal policy of question selection through continuous interactions with users. To facilitate training, we also propose to use a reward network to estimate the more informative reward. Compared to previous methods, our RL method is robust to noisy answers and does not rely on the Knowledge Base of objects. Experimental results show that our RL method clearly outperforms an entropy-based engineering system and has competitive performance in a noisy-free simulation environment.
['Xianchao Wu', 'Huang Hu', 'Chongyang Tao', 'Wei Wu', 'Zhan Chen', 'Can Xu', 'Bingfeng Luo']
2018-08-23
playing-20-question-game-with-policy-based-1
https://aclanthology.org/D18-1361
https://aclanthology.org/D18-1361.pdf
emnlp-2018-10
['question-selection']
['natural-language-processing']
[-2.47449771e-01 7.47581571e-02 1.15363784e-01 1.51619986e-01 -4.35398877e-01 -5.67689419e-01 1.48284033e-01 2.14266125e-02 -6.65332079e-01 7.49469161e-01 -3.98399979e-01 -4.40353870e-01 -2.54659444e-01 -1.19414103e+00 -3.20271254e-01 -3.62544030e-01 3.48716885e-01 5.74950397e-01 6.30105495e-01 -6.03945374e-01 6.31137371e-01 -1.29226133e-01 -1.31248391e+00 -3.15518916e-01 1.17560208e+00 1.22878897e+00 9.01933908e-01 7.58644998e-01 -2.87074059e-01 1.28292882e+00 -8.97853553e-01 -3.95917088e-01 3.78552616e-01 -9.64849949e-01 -1.09219575e+00 -1.09338693e-01 -5.61821640e-01 -5.80987751e-01 -1.31222069e-01 1.20711017e+00 5.35236061e-01 5.03408432e-01 2.01688141e-01 -1.19770968e+00 -4.49049056e-01 4.85817403e-01 2.81366650e-02 2.55423814e-01 8.29768956e-01 4.23805684e-01 1.17221487e+00 -5.56185246e-01 4.28530484e-01 8.31408739e-01 7.02919886e-02 5.69220066e-01 -6.30225360e-01 -6.09050691e-01 1.49483949e-01 7.61532247e-01 -1.26337266e+00 1.08277544e-01 9.52616394e-01 -7.24807987e-03 5.11992693e-01 2.75826275e-01 1.14387202e+00 4.55962002e-01 2.80863672e-01 7.62194633e-01 1.16898823e+00 -3.43291700e-01 1.04157054e+00 3.11869353e-01 -3.13598722e-01 6.12596571e-01 -1.67787552e-01 2.12255120e-01 -4.36781585e-01 -3.73173237e-01 9.23147798e-01 -2.38954023e-01 -2.58101732e-01 -3.62799287e-01 -8.75477314e-01 9.96266484e-01 4.23720837e-01 2.74528772e-01 -8.10716808e-01 1.89346686e-01 -1.35280252e-01 6.82414293e-01 -2.43471965e-01 1.07106638e+00 -3.66587728e-01 -5.84859133e-01 -5.94423413e-01 6.41651154e-01 1.33542132e+00 6.07453644e-01 5.69264233e-01 -4.68479991e-02 -1.91542283e-01 6.45516872e-01 4.43327814e-01 3.10118765e-01 6.08524144e-01 -1.46218431e+00 1.83801293e-01 7.72208631e-01 6.34363115e-01 -9.12858725e-01 -6.22633956e-02 -2.45774046e-01 -3.12362164e-01 4.65858191e-01 6.04907811e-01 -4.36475009e-01 -3.17537725e-01 1.44725096e+00 4.88187551e-01 9.58182439e-02 -3.30962949e-02 1.13969338e+00 7.48787701e-01 6.22566938e-01 -3.36172640e-01 -3.09317887e-01 1.56723356e+00 -7.42465854e-01 -8.38296175e-01 -2.14236811e-01 -7.64124617e-02 -3.86041373e-01 1.18936491e+00 6.69663906e-01 -1.12897217e+00 -3.44027370e-01 -8.32273245e-01 3.66378635e-01 -2.17807278e-01 -1.44480094e-01 3.07908684e-01 5.71826637e-01 -7.31670797e-01 5.65709531e-01 -2.72519827e-01 -3.40518504e-01 1.91258609e-01 3.48113507e-01 2.58685827e-01 2.88597763e-01 -1.65423095e+00 9.59696233e-01 5.12606740e-01 -7.28975609e-02 -9.52858806e-01 -1.83543891e-01 -3.62378389e-01 2.29089037e-01 1.19833124e+00 -5.70041180e-01 1.70179963e+00 -8.37628663e-01 -2.32336688e+00 2.11743310e-01 2.26526439e-01 -4.00626361e-01 4.55783725e-01 2.51326002e-02 5.50693162e-02 5.72316766e-01 -6.10843338e-02 3.93811762e-01 8.58227968e-01 -1.12255991e+00 -6.98170304e-01 -8.93353224e-02 9.15148437e-01 5.93202770e-01 9.77648199e-02 -5.43985851e-02 -2.41195828e-01 -2.93511271e-01 1.36102028e-02 -7.37628460e-01 -4.97775674e-01 -6.88097775e-02 8.19353852e-03 -4.45236295e-01 4.10816699e-01 -5.54199457e-01 1.18047917e+00 -1.61871600e+00 -2.78263092e-01 2.82085478e-01 2.84366101e-01 4.24738556e-01 1.04831144e-01 5.43413341e-01 3.47619206e-01 -5.39247878e-02 9.36201215e-02 5.45979500e-01 1.13284655e-01 2.32621208e-01 -9.31884199e-02 -5.82267269e-02 -9.06885117e-02 9.46726441e-01 -1.24207783e+00 -5.28483391e-01 -8.92064907e-03 -2.34400526e-01 -8.31811428e-01 8.74138772e-01 -6.28854394e-01 1.91355556e-01 -9.58645940e-01 3.49883676e-01 3.63385677e-01 -5.24151921e-01 2.28536755e-01 4.60348040e-01 2.35012382e-01 4.31558520e-01 -1.51452351e+00 1.20841825e+00 -3.45317423e-01 5.05915508e-02 1.58852756e-01 -9.50407147e-01 1.16206205e+00 3.65510315e-01 2.13214263e-01 -8.58107984e-01 3.06703329e-01 7.52510875e-03 2.90484488e-01 -8.51285040e-01 6.68168426e-01 -2.69916445e-01 -1.73531458e-01 8.57132733e-01 -2.04892531e-01 -7.01906860e-01 6.96135387e-02 1.87651709e-01 1.19441342e+00 1.85812771e-01 6.89316034e-01 2.29359660e-02 6.11674368e-01 -1.35964528e-01 5.56989491e-01 1.21300650e+00 -4.87680554e-01 1.43040344e-01 5.91361105e-01 -2.00240955e-01 -4.85047787e-01 -9.81256664e-01 5.34476340e-01 1.04384446e+00 5.65203190e-01 -3.30846876e-01 -8.42225552e-01 -6.55602813e-01 -2.51594901e-01 9.25073683e-01 -4.05452937e-01 -2.16975927e-01 -3.45536411e-01 -1.36681214e-01 -1.62521582e-02 1.19200312e-01 9.42508936e-01 -1.54603553e+00 -1.22338152e+00 5.21227598e-01 -5.23371339e-01 -6.06155932e-01 -6.23952568e-01 1.23545446e-01 -6.09465539e-01 -1.15941811e+00 -4.71094042e-01 -6.64016068e-01 3.78660589e-01 1.37474716e-01 9.52696741e-01 4.05963510e-01 4.57912832e-01 5.18630624e-01 -4.98741269e-01 -4.26269978e-01 -4.81177658e-01 -1.68834981e-02 -2.19775796e-01 -2.57929236e-01 3.68407875e-01 -4.63814229e-01 -8.71376276e-01 4.55318481e-01 -7.37965584e-01 -2.62239009e-01 4.47519392e-01 8.47457945e-01 3.01992804e-01 3.86556596e-01 1.01546383e+00 -5.77559412e-01 1.47425568e+00 -4.69436109e-01 -8.02523971e-01 3.55631024e-01 -7.19534755e-01 3.69708478e-01 8.98408651e-01 -6.17605090e-01 -8.71735632e-01 -1.80046067e-01 -1.05738573e-01 4.63124067e-02 6.74962252e-02 4.68772680e-01 -2.61135995e-01 -7.15520009e-02 6.47318184e-01 5.54475427e-01 1.49190098e-01 -1.65971711e-01 1.95375621e-01 7.47034132e-01 2.15319082e-01 -5.67800760e-01 6.95496082e-01 -1.22178383e-01 -3.54083866e-01 -5.01312613e-01 -3.76623064e-01 -2.53856987e-01 3.38586830e-02 -5.62880695e-01 6.24191463e-01 -4.00906205e-01 -1.63963401e+00 2.88518280e-01 -1.11216140e+00 -2.87104905e-01 -4.33391571e-01 2.90106565e-01 -8.45184207e-01 3.12641203e-01 -4.01117116e-01 -1.17028010e+00 -3.23195130e-01 -9.98674572e-01 3.53497058e-01 7.47394264e-01 -1.51586324e-01 -5.67635119e-01 -1.39562637e-01 5.18372476e-01 5.66541970e-01 -2.66052932e-01 6.86171055e-01 -6.96109235e-01 -9.01839912e-01 -1.04786500e-01 1.67455599e-01 1.19218379e-01 7.69071057e-02 -5.33873439e-01 -3.72698694e-01 4.70394343e-02 5.67161441e-01 -4.40258026e-01 3.08319032e-01 2.82473266e-02 1.00690293e+00 -5.30442119e-01 2.75345534e-01 -7.56712481e-02 1.23586488e+00 7.44038582e-01 7.42610633e-01 4.38698798e-01 9.51428525e-03 5.27896285e-01 8.55316579e-01 7.88269937e-01 7.67430246e-01 4.29974198e-01 6.47622049e-01 5.46262920e-01 5.08574963e-01 -3.85134012e-01 2.85931855e-01 6.00374460e-01 -1.83392525e-01 -1.10478140e-01 -5.05571842e-01 2.72726208e-01 -1.85271084e+00 -1.05498719e+00 3.92753184e-01 2.17486525e+00 1.22132206e+00 2.33639151e-01 2.87133664e-01 1.84757590e-01 5.20923078e-01 -3.81883048e-02 -9.91172552e-01 -2.84120500e-01 3.90315682e-01 3.46321553e-01 1.11958534e-01 4.45399106e-01 -4.71392155e-01 8.98866296e-01 6.14559364e+00 9.23422575e-01 -7.07455397e-01 -1.23405203e-01 3.94288540e-01 1.38756102e-02 -3.16538930e-01 2.16796979e-01 -3.69634092e-01 6.20443940e-01 7.69973040e-01 -5.87615967e-01 9.56028819e-01 8.51049185e-01 3.50872427e-01 -9.18309510e-01 -6.81714356e-01 8.75135541e-01 -2.09358543e-01 -9.86449420e-01 -2.97921330e-01 -1.39341563e-01 3.37035745e-01 -5.85964918e-01 -2.18521059e-01 4.36635286e-01 6.54526949e-01 -8.03122461e-01 7.80186296e-01 8.65489900e-01 -2.94890646e-02 -8.43763292e-01 6.78353250e-01 1.03784811e+00 -9.74197984e-01 -3.14836353e-01 -4.24960345e-01 -5.12398481e-01 -1.71843112e-01 1.97712798e-02 -1.00832343e+00 3.26343536e-01 4.91014510e-01 -2.72783577e-01 -3.65514904e-01 1.19054317e+00 -7.56393611e-01 6.92017555e-01 -3.27540189e-01 -9.87284660e-01 1.04290865e-01 -3.46626550e-01 3.86389494e-01 2.31380701e-01 3.20047885e-01 8.50065947e-01 3.05691779e-01 1.16899776e+00 -8.28055106e-03 2.64501154e-01 -2.37645715e-01 -2.59678420e-02 5.93417645e-01 1.06573296e+00 -7.70154119e-01 -1.49709761e-01 5.53714000e-02 8.63534510e-01 2.27505311e-01 3.14764649e-01 -7.42515624e-01 -6.05325103e-01 1.24432221e-01 2.37539351e-01 4.79774088e-01 -8.88680294e-02 6.33417815e-02 -8.48523676e-01 -1.05407257e-02 -1.25136888e+00 2.91673422e-01 -1.04416025e+00 -8.75053942e-01 4.51335430e-01 -3.50531638e-01 -1.11169195e+00 -5.80843270e-01 -2.07769096e-01 -7.09445357e-01 7.90236115e-01 -1.45737982e+00 -1.62740529e-01 -1.77774698e-01 6.04157388e-01 1.76009253e-01 -8.61286223e-02 6.25894964e-01 -2.40657389e-01 -1.23951994e-01 3.20326865e-01 -1.66679800e-01 9.78858024e-02 2.76054263e-01 -1.42093611e+00 -1.03012912e-01 3.70567530e-01 -1.16949111e-01 4.61508811e-01 7.63884485e-01 -5.46076357e-01 -1.57763231e+00 -5.13652503e-01 6.87720060e-01 -9.19252187e-02 5.83263874e-01 -1.70320392e-01 -8.49825382e-01 -9.61665139e-02 3.02057952e-01 -3.00166756e-01 5.90137303e-01 -4.32109028e-01 3.67085993e-01 -1.59361422e-01 -1.43909907e+00 8.38841319e-01 4.77939904e-01 -4.49833304e-01 -9.74243402e-01 1.01091057e-01 6.88417673e-01 -4.22902375e-01 -6.10054553e-01 -1.59546077e-01 4.43567693e-01 -1.08466983e+00 5.53905249e-01 -3.19662780e-01 2.19222993e-01 -4.76580590e-01 1.50394469e-01 -1.37732220e+00 -1.68830872e-01 -9.44115162e-01 -1.72611564e-01 7.21125066e-01 1.69648245e-01 -7.39206135e-01 1.01660156e+00 8.61402690e-01 6.46466672e-01 -9.76540208e-01 -9.33916152e-01 -7.26941168e-01 -1.55899497e-02 -2.44535312e-01 8.37469578e-01 4.92813051e-01 4.15988863e-01 3.94143373e-01 -4.68806535e-01 8.10608035e-04 2.79540807e-01 3.95091176e-01 5.79023063e-01 -1.10062087e+00 -7.14674175e-01 -3.57202649e-01 5.35789467e-02 -1.62031424e+00 -2.95742720e-01 -4.84661430e-01 4.71445382e-01 -1.70345628e+00 -2.04383343e-01 -2.59272248e-01 -2.31019482e-01 1.11154854e-01 -5.61650336e-01 -3.58776689e-01 3.69305938e-01 -3.12014483e-02 -1.10822725e+00 7.27846980e-01 1.67958117e+00 8.29502493e-02 -3.80210161e-01 4.69999731e-01 -1.03246140e+00 5.68668962e-01 1.08732760e+00 -4.47161525e-01 -5.48320532e-01 2.86642611e-01 7.38178313e-01 8.49686086e-01 2.62295097e-01 -8.47146928e-01 6.57983601e-01 -5.07809520e-01 5.93997650e-02 -4.76803273e-01 3.38329047e-01 -9.16109025e-01 -1.46135509e-01 8.29296529e-01 -3.63105714e-01 -1.09146632e-01 -3.15723628e-01 5.19749761e-01 -6.90563163e-03 -7.97389090e-01 5.47283709e-01 -5.70883095e-01 -4.80865240e-01 1.39470831e-01 -7.95285046e-01 2.89854527e-01 1.11396694e+00 -3.20355684e-01 2.68912781e-02 -1.18897855e+00 -4.22298223e-01 7.58050323e-01 1.42429369e-02 1.54194951e-01 8.96991789e-01 -1.26818669e+00 -4.65686768e-01 -1.37386903e-01 -1.22836478e-01 -1.54205814e-01 -6.97666183e-02 2.44112119e-01 -3.91455233e-01 8.87317508e-02 6.75119683e-02 -1.89562474e-04 -8.30388665e-01 5.61139524e-01 6.52560592e-01 -5.17764151e-01 -1.76501915e-01 6.54885232e-01 -2.12775066e-01 -4.65228796e-01 2.24402830e-01 -3.27670217e-01 -6.20893955e-01 -4.53479700e-02 5.65718055e-01 2.98375428e-01 -3.27616483e-01 -2.38397606e-02 -1.10257506e-01 1.05707824e-01 9.06224549e-02 -2.60377765e-01 1.08293426e+00 -1.35100305e-01 1.00848172e-02 3.32517326e-01 4.27844346e-01 -2.70965040e-01 -1.10063088e+00 -4.96351212e-01 -7.66083822e-02 -6.19696975e-01 -1.21195257e-01 -9.47288930e-01 -8.06663632e-01 4.59600687e-01 3.31204295e-01 8.34079921e-01 1.21900237e+00 -2.00378522e-01 8.65352571e-01 8.10390770e-01 7.32174575e-01 -1.61361873e+00 7.69227624e-01 6.24302328e-01 9.11373317e-01 -1.11480606e+00 -1.73796728e-01 1.14162350e-02 -8.46294284e-01 1.04850674e+00 8.82954359e-01 -2.61296183e-01 7.69543588e-01 -1.51646035e-02 8.28972459e-02 -2.55349517e-01 -9.03074324e-01 -3.75101775e-01 -1.62270620e-01 4.79640186e-01 -1.59410402e-01 -2.36673877e-02 -5.92269838e-01 8.89798880e-01 -5.51806331e-01 3.57871920e-01 7.55258799e-01 9.88147318e-01 -8.93992007e-01 -1.23935390e+00 -5.41818202e-01 5.51311672e-01 -3.52108419e-01 -7.13466927e-02 -4.40467387e-01 3.11279505e-01 1.17353491e-01 1.54597604e+00 -3.42596680e-01 -3.62079173e-01 4.50696975e-01 -4.77085635e-02 4.20458257e-01 -5.32394767e-01 -9.47584748e-01 -2.98942119e-01 -1.96962669e-01 -5.67518592e-01 -2.48978540e-01 -4.07925248e-01 -1.48527169e+00 -2.92508751e-01 -6.69009745e-01 7.16381490e-01 3.84280950e-01 1.02675390e+00 2.77156472e-01 3.80947113e-01 1.01419294e+00 -7.89210200e-02 -9.95327294e-01 -8.30048442e-01 -7.05601633e-01 2.91906651e-02 -2.21134014e-02 -5.78144550e-01 -2.47240320e-01 -4.41411644e-01]
[3.894944667816162, 1.5167460441589355]
80a7698a-34b6-4051-801e-e2c6963a9533
movingfashion-a-benchmark-for-the-video-to
2110.02627
null
https://arxiv.org/abs/2110.02627v4
https://arxiv.org/pdf/2110.02627v4.pdf
MovingFashion: a Benchmark for the Video-to-Shop Challenge
Retrieving clothes which are worn in social media videos (Instagram, TikTok) is the latest frontier of e-fashion, referred to as "video-to-shop" in the computer vision literature. In this paper we present MovingFashion, the first publicly available dataset to cope with this challenge. MovingFashion is composed of 14855 social videos, each one of them associated to e-commerce "shop" images where the corresponding clothing items are clearly portrayed. In addition, we present a network for retrieving the shop images in this scenario, dubbed SEAM Match-RCNN. The model is trained by image-to-video domain adaptation, allowing to use video sequences where only their association with a shop image is given, eliminating the need of millions of annotated bounding boxes. SEAM Match-RCNN builds an embedding, where an attention-based weighted sum of few frames (10) of a social video is enough to individuate the correct product within the first 5 retrieved items in a 14K+ shop element gallery with an accuracy of 80%. This provides the best performance on MovingFashion, comparing exhaustively against the related state-of-the-art approaches and alternative baselines.
['Marco Cristani', 'Geri Skenderi', 'Christian Joppi', 'Marco Godi']
2021-10-06
null
null
null
null
['video-to-shop']
['computer-vision']
[ 1.02650754e-01 -3.13899010e-01 -2.17203572e-01 -2.72855937e-01 -7.93265343e-01 -7.32512832e-01 4.17478532e-01 -7.96544328e-02 -4.72524285e-01 4.12147373e-01 3.54176849e-01 5.05262613e-01 -1.23359829e-01 -6.04844570e-01 -1.26590204e+00 -5.88732898e-01 -8.59889314e-02 1.91159457e-01 1.55356526e-01 -2.57600456e-01 6.70756027e-03 7.64497891e-02 -1.64104736e+00 7.70255089e-01 1.14662670e-01 1.61028850e+00 4.80481714e-01 7.55331933e-01 1.96944773e-01 7.29335129e-01 -5.48132479e-01 -9.67014968e-01 4.15593237e-01 -1.41489267e-01 -5.91913044e-01 4.91980165e-01 1.17682409e+00 -7.53829658e-01 -6.08918548e-01 1.04939377e+00 1.06890097e-01 2.55335599e-01 4.96897757e-01 -1.05933785e+00 -1.34569466e+00 4.12342250e-01 -4.09038126e-01 1.26729414e-01 7.05293417e-01 -9.23194587e-02 1.18712640e+00 -1.44682825e+00 1.36427915e+00 1.27172136e+00 7.26699829e-01 3.26193124e-01 -1.05327153e+00 -3.29795361e-01 3.47015224e-02 5.57014823e-01 -1.50350952e+00 -4.44547683e-01 8.17692757e-01 -2.73087472e-01 6.98533654e-01 3.19012970e-01 1.10198247e+00 1.53482103e+00 2.17364594e-01 1.14880955e+00 6.04731917e-01 -1.48415729e-01 7.50691965e-02 1.34552181e-01 -9.13673714e-02 7.25362062e-01 3.65056992e-02 -2.47422263e-01 -8.20456505e-01 1.03886865e-01 9.55158889e-01 4.02915627e-01 -1.90729469e-01 -5.28268635e-01 -1.30680001e+00 8.47568512e-01 6.57696128e-01 2.07030892e-01 -4.91475016e-01 3.60402822e-01 6.03711069e-01 4.43955570e-01 5.88251472e-01 3.14019233e-01 -2.90910989e-01 1.78463340e-01 -1.04005897e+00 4.48446482e-01 5.97470224e-01 1.25801623e+00 4.50727612e-01 -1.38067469e-01 -3.58212739e-02 9.07441258e-01 1.07440345e-01 4.93483394e-01 5.82454205e-02 -1.03959548e+00 4.70085502e-01 5.20910084e-01 4.09768999e-01 -1.49225879e+00 4.07875590e-02 -2.47080494e-02 -5.22646129e-01 -3.91412705e-01 4.46948826e-01 7.53876194e-02 -7.70343959e-01 1.13692176e+00 3.13977838e-01 1.58879265e-01 -2.86337703e-01 1.39342010e+00 1.04792929e+00 8.76424670e-01 -1.12586409e-01 7.98788816e-02 1.63591528e+00 -1.28956759e+00 -7.50492990e-01 -4.08759862e-02 5.02467416e-02 -1.04817247e+00 9.39059794e-01 5.28789461e-01 -1.11359811e+00 -7.58306205e-01 -9.85686362e-01 -3.95681679e-01 -7.73624480e-01 4.80487108e-01 5.23331285e-01 2.11121023e-01 -9.59248662e-01 7.59200573e-01 -2.39160016e-01 -5.76987386e-01 5.52377462e-01 2.20667005e-01 -7.79134274e-01 -4.80399251e-01 -1.13986599e+00 6.07831180e-01 3.25295031e-01 2.76619494e-01 -1.21287096e+00 -6.35533273e-01 -1.05297518e+00 -1.80782065e-01 8.58059466e-01 -3.75528544e-01 8.92703235e-01 -1.43626249e+00 -1.07086003e+00 1.00198364e+00 2.28360325e-01 -5.14871955e-01 5.58522403e-01 -8.23316395e-01 -8.52682233e-01 5.83118498e-01 1.31569445e-01 8.51460040e-01 1.44853485e+00 -1.31646323e+00 -6.00245416e-01 -2.83171624e-01 4.98938233e-01 3.55173051e-02 -4.08267617e-01 1.90123960e-01 -1.11399508e+00 -9.34606850e-01 -2.42154390e-01 -1.17995799e+00 1.00930005e-01 3.13081771e-01 -2.84054607e-01 -1.59972504e-01 8.79272103e-01 -1.07622898e+00 1.15999138e+00 -2.33468866e+00 5.04309416e-01 -1.08003512e-01 1.24735847e-01 9.43684801e-02 -3.32294524e-01 6.37938738e-01 2.54542649e-01 -3.30269784e-01 1.37168676e-01 -3.21244866e-01 1.37157381e-01 1.31755024e-01 -3.88228558e-02 6.62768543e-01 1.45000458e-01 1.07204199e+00 -9.64250863e-01 -4.07170385e-01 4.87567186e-01 7.09258497e-01 -4.06233966e-01 5.59405833e-02 -2.72169977e-01 6.79908320e-02 -6.56742454e-02 9.99997795e-01 7.56428361e-01 -2.37888306e-01 3.85010868e-01 -7.83000827e-01 1.78196833e-01 -4.68606740e-01 -1.07980609e+00 2.09193850e+00 -2.62170345e-01 6.30949140e-01 3.48964036e-01 -5.41267514e-01 5.81046581e-01 7.62727037e-02 5.09217978e-01 -6.02063239e-01 2.56554127e-01 1.15556702e-01 -7.05999970e-01 -7.25649297e-01 9.72974896e-01 5.00316381e-01 -3.80122960e-01 -1.15122370e-01 5.54231882e-01 2.78173625e-01 5.01383722e-01 2.59007931e-01 7.71180987e-01 3.69605869e-01 7.92561658e-03 -2.11961627e-01 3.63265276e-01 7.54266158e-02 2.04105914e-01 5.83820105e-01 -1.73927248e-01 7.61716545e-01 6.83406442e-02 -9.34989512e-01 -1.28417826e+00 -9.55726266e-01 1.59138933e-01 1.47799718e+00 5.06830275e-01 -6.09050095e-01 -8.69169176e-01 -9.00254130e-01 3.26521933e-01 1.89886481e-01 -8.66573632e-01 1.60298571e-01 -6.79117620e-01 -1.55943692e-01 1.98394395e-02 5.12372553e-01 5.92459857e-01 -1.26501560e+00 -3.66672486e-01 2.57695287e-01 -1.66416615e-01 -1.32481027e+00 -9.81835425e-01 -2.55547315e-01 -3.30596805e-01 -1.35074139e+00 -1.18649900e+00 -1.15678525e+00 6.76200867e-01 5.44341028e-01 1.35568750e+00 -3.35363224e-02 -4.00428295e-01 5.65522254e-01 -5.93954325e-01 -3.00703541e-04 -7.06862658e-02 -1.10166430e-01 6.18187748e-02 6.19461656e-01 6.06679499e-01 -7.56381378e-02 -9.86196995e-01 4.07121330e-01 -9.78051186e-01 -1.83810309e-01 4.13594395e-01 6.72522187e-01 6.26899779e-01 -1.59172028e-01 3.23293120e-01 -6.83556616e-01 3.37450147e-01 -6.76934838e-01 -2.72504359e-01 2.36691833e-01 1.25175461e-01 -7.26525605e-01 7.70191371e-01 -6.23577416e-01 -4.64780807e-01 3.22152168e-01 -3.07434648e-02 -1.15639043e+00 -1.23582162e-01 9.22694728e-02 1.03234813e-01 -2.69131940e-02 3.40401918e-01 2.44446352e-01 -1.12225354e-01 -6.88014328e-01 6.09382629e-01 4.15623754e-01 5.90143085e-01 -1.46068603e-01 5.15698373e-01 4.74469721e-01 -4.24198300e-01 -8.81529808e-01 -8.46412003e-01 -6.71945989e-01 -6.04016840e-01 -7.50822783e-01 1.16619635e+00 -1.02132094e+00 -7.37905681e-01 3.21827441e-01 -9.21751559e-01 7.36127496e-02 -3.21170032e-01 3.79327863e-01 -5.23605704e-01 1.64489761e-01 -9.13364947e-01 -5.51954865e-01 -1.62833050e-01 -9.56331789e-01 1.33123100e+00 -1.39557302e-01 -4.18312363e-02 -6.42373562e-01 -4.47276652e-01 6.53029680e-01 2.77782440e-01 3.82267296e-01 4.08925563e-01 -4.68030185e-01 -5.51525772e-01 -3.85293394e-01 -2.75914550e-01 6.02945030e-01 -5.40563650e-02 -2.09339380e-01 -7.44047701e-01 -4.99614477e-01 -3.98766160e-01 -3.15660030e-01 1.11273634e+00 3.98937941e-01 1.22359812e+00 -4.66586947e-01 -1.67702094e-01 1.98783562e-01 1.60257196e+00 1.93886869e-02 6.20919883e-01 2.95842975e-01 8.97883117e-01 4.77242887e-01 8.97422075e-01 3.39969635e-01 3.84708643e-01 9.50048387e-01 7.91517019e-01 -4.43314128e-02 -2.76252151e-01 -5.37218392e-01 5.78930438e-01 8.68243635e-01 -3.41617435e-01 -8.91892835e-02 -1.43426329e-01 9.32563245e-01 -1.96537364e+00 -9.72019970e-01 2.57540885e-02 1.95140922e+00 2.43362188e-01 -1.34561792e-01 4.97845232e-01 -2.18591332e-01 8.22492421e-01 5.20908177e-01 -4.18716699e-01 -2.03904554e-01 -1.61991343e-01 -1.28048986e-01 5.98574579e-01 8.90103579e-02 -1.53976393e+00 8.05369854e-01 6.02916527e+00 1.10135615e+00 -6.28189206e-01 3.07367593e-01 4.57903206e-01 -3.21008116e-01 5.98681159e-02 -5.17065287e-01 -6.80845201e-01 7.05077529e-01 8.52203965e-01 5.48311889e-01 6.92775846e-01 1.14528883e+00 -7.43191317e-02 -5.31324036e-02 -1.11146402e+00 1.10753322e+00 7.50679672e-01 -1.64667404e+00 1.40145615e-01 -8.85152072e-02 8.20722818e-01 -3.98527801e-01 1.66278541e-01 3.65204513e-01 -1.26190990e-01 -8.71191740e-01 1.19579375e+00 5.07026017e-01 9.66507077e-01 -7.49416769e-01 7.73522973e-01 -2.94448197e-01 -1.39574587e+00 -2.95137078e-01 -5.83523750e-01 4.80962664e-01 2.19440043e-01 2.42665380e-01 -3.76967549e-01 6.27678871e-01 1.21072900e+00 1.06907916e+00 -5.52527070e-01 6.90024614e-01 3.31411511e-01 1.58795610e-01 -6.81377724e-02 -1.11168168e-01 4.58577275e-01 -2.33322561e-01 4.97280121e-01 1.29215014e+00 4.49045807e-01 -2.55181581e-01 2.11739615e-01 5.42829335e-01 -5.79822719e-01 2.16493636e-01 -7.40083992e-01 -8.10925215e-02 1.24153242e-01 1.25214803e+00 -7.24806130e-01 -4.46532786e-01 -7.01725423e-01 1.55553496e+00 1.35869741e-01 1.55160084e-01 -9.58202124e-01 -4.40385826e-02 5.52939177e-01 4.17464107e-01 1.07549632e+00 -1.58008248e-01 8.10288310e-01 -1.14688897e+00 2.54563481e-01 -9.69716251e-01 3.26615959e-01 -9.07256901e-01 -1.61312687e+00 5.84043324e-01 -3.93751189e-02 -1.33673871e+00 2.69029409e-01 -8.37212801e-01 1.27837881e-01 1.19614024e-02 -1.18485153e+00 -1.58965480e+00 -3.32049340e-01 7.69193053e-01 1.26833022e+00 -2.27305606e-01 7.93338478e-01 7.25059211e-01 -2.90314019e-01 3.92037719e-01 3.84497434e-01 2.69726217e-01 8.39522243e-01 -1.11624324e+00 3.20707351e-01 3.71742487e-01 4.56801713e-01 4.48532045e-01 5.44794917e-01 -6.24350369e-01 -1.92703784e+00 -1.30575609e+00 6.96987271e-01 -6.47255778e-01 6.96662188e-01 -6.49618089e-01 -4.49180663e-01 7.23978877e-01 4.36165154e-01 3.55237126e-01 5.22067368e-01 -2.28222713e-01 -3.92027110e-01 -3.24896842e-01 -1.13155329e+00 4.29140657e-01 1.25901020e+00 -5.69493532e-01 -3.77976745e-01 6.44897223e-01 6.33355319e-01 -5.01208425e-01 -1.30026221e+00 1.28634961e-03 8.84824634e-01 -9.08007205e-01 1.25052476e+00 -5.59045851e-01 7.71247208e-01 -2.79721051e-01 -6.54707432e-01 -1.04250431e+00 -2.95573771e-01 -4.63867307e-01 -4.94657367e-01 1.01042068e+00 5.07407151e-02 1.38454348e-01 7.02546120e-01 1.80873945e-01 2.69660223e-02 -7.78385580e-01 -8.70864153e-01 -6.48923099e-01 -5.44711590e-01 -2.09511712e-01 5.89259505e-01 9.23588455e-01 -6.78922951e-01 1.63619854e-02 -1.10941017e+00 -6.80449158e-02 6.46260142e-01 1.95857346e-01 7.65561640e-01 -9.74052608e-01 -1.75900370e-01 1.60979286e-01 -6.07808888e-01 -1.23493707e+00 -1.06947497e-01 -7.07822502e-01 -1.54266849e-01 -1.24685884e+00 2.60299057e-01 2.16075867e-01 -3.81120920e-01 -2.38433816e-02 2.81128913e-01 1.03114033e+00 7.43963957e-01 9.40173641e-02 -1.29599094e+00 1.79999247e-01 1.38280261e+00 -4.44668323e-01 1.24677859e-01 -3.84341598e-01 -3.74009550e-01 6.57699823e-01 2.18690455e-01 -2.53642976e-01 6.48651225e-03 -4.74524051e-01 2.45105073e-01 1.07641794e-01 6.21636152e-01 -9.30649757e-01 1.69321112e-02 2.13941365e-01 7.99938142e-01 -6.70199573e-01 8.40421021e-01 -1.32303631e+00 5.11744857e-01 6.28795251e-02 -3.42746794e-01 2.77438581e-01 -4.65734750e-02 1.03651381e+00 -2.02279225e-01 -2.32678369e-01 4.21211720e-01 -5.27340353e-01 -1.35597658e+00 2.32086465e-01 -2.28717417e-01 -3.79048735e-01 1.23452199e+00 -2.13699684e-01 -1.39054224e-01 -4.34709579e-01 -1.10848308e+00 -9.05299261e-02 4.55010533e-01 9.46529865e-01 8.39465499e-01 -1.73379552e+00 -6.22525394e-01 1.23855814e-01 2.87420303e-01 -5.28636932e-01 7.04750896e-01 5.26953042e-01 -6.27801418e-01 2.94232339e-01 -3.89365554e-01 -5.37169456e-01 -1.33973241e+00 8.91349554e-01 -1.59302816e-01 -5.86269535e-02 -9.17020380e-01 8.92803729e-01 -2.73756757e-02 -6.75687492e-02 2.08172262e-01 -9.01721045e-02 -2.10492149e-01 4.09807205e-01 6.80267215e-01 3.38960558e-01 8.19949880e-02 -1.17660785e+00 -3.23057622e-01 7.01965392e-01 -2.74699748e-01 4.10389870e-01 1.43340242e+00 -2.43474096e-01 1.02730528e-01 4.04771060e-01 1.42490435e+00 -8.16770941e-02 -1.42477655e+00 -1.90379977e-01 -3.44803631e-01 -9.70193744e-01 -5.22424877e-02 -6.26090646e-01 -1.36902487e+00 3.07461590e-01 8.27276826e-01 3.93838376e-01 9.80769575e-01 2.45694327e-03 1.19031847e+00 2.13709235e-01 6.38554394e-01 -1.40747225e+00 3.35310191e-01 4.86830957e-02 1.20895207e+00 -1.38652289e+00 6.93564638e-02 -3.43083888e-01 -9.66062367e-01 1.01732600e+00 3.69679272e-01 -7.55056739e-01 5.41086674e-01 -2.65299112e-01 -2.26954430e-01 -4.01843190e-01 -3.50624233e-01 -4.87846583e-02 4.82625932e-01 4.70293105e-01 -9.66418907e-02 1.19526558e-01 1.93612538e-02 6.89363718e-01 1.82560459e-01 -6.96414635e-02 9.82581452e-02 8.30262184e-01 -1.09477088e-01 -5.88303983e-01 -3.82327050e-01 4.89172816e-01 -6.24496460e-01 6.34569582e-03 -3.28175396e-01 1.03107607e+00 3.33253771e-01 9.00522053e-01 2.20739186e-01 -4.91836965e-01 4.61483866e-01 -1.29329160e-01 5.57121158e-01 -3.02493900e-01 -9.40088809e-01 2.63997585e-01 3.56743097e-01 -8.73792052e-01 -7.61812508e-01 -3.26902419e-01 -4.03966397e-01 -5.51970661e-01 -2.65876621e-01 -2.06662416e-01 6.16655648e-01 5.08285820e-01 4.10909474e-01 2.80052036e-01 4.50668395e-01 -1.47370553e+00 -7.78892636e-02 -7.00666010e-01 -9.83415008e-01 8.93315136e-01 4.05062109e-01 -7.20071435e-01 -6.03488237e-02 3.80364925e-01]
[10.317032814025879, 0.837133526802063]
cd4ad536-cedd-4464-b0a9-faed5401fb33
positional-spectral-temporal-attention-in-3d
2110.09955
null
https://arxiv.org/abs/2110.09955v2
https://arxiv.org/pdf/2110.09955v2.pdf
Positional-Spectral-Temporal Attention in 3D Convolutional Neural Networks for EEG Emotion Recognition
Recognizing the feelings of human beings plays a critical role in our daily communication. Neuroscience has demonstrated that different emotion states present different degrees of activation in different brain regions, EEG frequency bands and temporal stamps. In this paper, we propose a novel structure to explore the informative EEG features for emotion recognition. The proposed module, denoted by PST-Attention, consists of Positional, Spectral and Temporal Attention modules to explore more discriminative EEG features. Specifically, the Positional Attention module is to capture the activate regions stimulated by different emotions in the spatial dimension. The Spectral and Temporal Attention modules assign the weights of different frequency bands and temporal slices respectively. Our method is adaptive as well as efficient which can be fit into 3D Convolutional Neural Networks (3D-CNN) as a plug-in module. We conduct experiments on two real-world datasets. 3D-CNN combined with our module achieves promising results and demonstrate that the PST-Attention is able to capture stable patterns for emotion recognition from EEG.
['Dongmei Jiang', 'Hao Wu', 'Yanxi Zhao', 'Jiyao Liu']
2021-10-13
null
null
null
null
['eeg-emotion-recognition']
['miscellaneous']
[-3.00287545e-01 -4.62303966e-01 3.26418996e-01 -5.09820700e-01 -1.50454134e-01 -1.55199587e-01 2.40282729e-01 -4.37676087e-02 -3.04017037e-01 6.11684263e-01 3.63099068e-01 3.18666607e-01 -2.05091134e-01 -5.58930635e-01 -1.84257627e-01 -6.70374990e-01 -4.97315019e-01 -2.91486472e-01 -3.99949811e-02 -4.97724935e-02 4.48501617e-01 6.44771934e-01 -1.37460220e+00 4.80583578e-01 6.95859134e-01 1.40354419e+00 1.49178222e-01 2.86887378e-01 -4.57987823e-02 2.54400820e-01 -6.08886123e-01 2.38324776e-01 -1.70644775e-01 -4.74847227e-01 -6.23436153e-01 -7.21229985e-02 -4.13015246e-01 2.47623354e-01 -1.34562269e-01 1.11753738e+00 8.26346099e-01 2.24603295e-01 6.40756547e-01 -1.25812888e+00 -7.17602670e-01 4.12672758e-01 -8.51252615e-01 8.03597212e-01 5.09670317e-01 -4.57054079e-02 5.59166074e-01 -9.47126567e-01 7.30354562e-02 9.65997219e-01 6.20370865e-01 3.50307226e-01 -9.12127137e-01 -8.91232967e-01 2.68156946e-01 7.54101574e-01 -1.39517164e+00 -3.32751721e-02 1.21421099e+00 -3.53452355e-01 1.25685906e+00 2.17930689e-01 1.20718789e+00 1.25799322e+00 8.35901856e-01 7.65394986e-01 1.33913362e+00 -4.59599234e-02 2.25120917e-01 1.52063295e-01 4.89470363e-01 1.82119533e-01 -3.53127062e-01 -3.04220766e-01 -7.87529051e-01 -1.75921228e-02 6.52309775e-01 1.54672936e-01 -5.95424533e-01 1.31093308e-01 -1.11901402e+00 5.16710520e-01 6.60816193e-01 7.74274051e-01 -9.14903045e-01 1.63788691e-01 4.36700344e-01 1.38501495e-01 6.37279093e-01 4.49176639e-01 -6.57299995e-01 -2.69832164e-01 -7.28069842e-01 -1.89084664e-01 3.10158670e-01 6.48662806e-01 6.87397838e-01 -1.79768521e-02 -3.94828618e-01 7.84423470e-01 1.48770571e-01 1.46154389e-01 1.01289213e+00 -3.58549446e-01 -2.22125679e-01 6.04857445e-01 -2.78928012e-01 -1.14048922e+00 -9.49302495e-01 -6.20714366e-01 -1.04805374e+00 -1.61984906e-01 -4.67563182e-01 -3.80396456e-01 -6.88354671e-01 1.75859821e+00 1.49965554e-01 5.78248560e-01 -2.21961245e-01 1.10538340e+00 8.30367744e-01 7.35226393e-01 1.29091471e-01 -2.88791865e-01 1.72796845e+00 -6.15236402e-01 -9.14638400e-01 -1.42746940e-01 2.24695858e-02 -5.03817856e-01 9.09069717e-01 4.43142503e-01 -1.00432169e+00 -7.16643512e-01 -1.00231814e+00 1.18653484e-01 -5.22384822e-01 5.17701395e-02 7.16254234e-01 3.18230242e-01 -9.90949154e-01 4.83039439e-01 -7.71799982e-01 -4.57734704e-01 5.40926576e-01 4.68704641e-01 -3.59917492e-01 6.38598144e-01 -1.40348887e+00 8.57769907e-01 2.12188005e-01 3.63032699e-01 -5.81774473e-01 -5.19963741e-01 -6.18460715e-01 5.09668052e-01 -3.39430839e-01 -4.19537961e-01 6.88714802e-01 -1.24565101e+00 -1.58061957e+00 6.33681297e-01 -1.13636926e-01 -1.63084149e-01 -3.87557834e-01 -1.71816960e-01 -7.49819160e-01 3.14334899e-01 -9.84486118e-02 5.45353174e-01 8.02691758e-01 -6.46831870e-01 -2.99675703e-01 -6.06574535e-01 -3.92582297e-01 1.73604235e-01 -6.34050310e-01 2.29591817e-01 -3.84006262e-01 -6.76243126e-01 1.52661771e-01 -5.79845190e-01 -2.96861120e-02 -4.79452223e-01 -1.11607395e-01 -3.90009820e-01 7.05603600e-01 -3.88691783e-01 1.11957526e+00 -2.26313639e+00 2.14293033e-01 3.46548855e-01 1.45011529e-01 -2.10179806e-01 -2.83362959e-02 8.76357779e-02 -5.60750365e-01 -7.15438351e-02 -8.16309676e-02 2.03908876e-01 1.50023445e-01 -1.88716710e-01 -1.50950640e-01 5.64154148e-01 6.15347326e-01 8.85973454e-01 -6.87988341e-01 -2.70443827e-01 1.69595405e-01 7.30325460e-01 -4.57978457e-01 2.32226267e-01 3.08013052e-01 5.99217296e-01 -6.18055999e-01 4.70590115e-01 8.09446931e-01 -7.55039304e-02 -4.09455746e-02 -4.64464724e-01 -3.42324942e-01 1.95684701e-01 -8.75923574e-01 1.87667370e+00 -2.01475635e-01 8.18323910e-01 -1.06819063e-01 -1.18104267e+00 1.06256938e+00 5.02253652e-01 6.55781567e-01 -1.03006256e+00 6.28072023e-01 -4.04115431e-02 3.74777876e-02 -8.95215631e-01 1.81673944e-01 -1.65349990e-01 -1.24910578e-01 4.70337212e-01 5.19529104e-01 2.74778426e-01 -3.25366259e-01 -2.80191928e-01 9.55810845e-01 -1.79307342e-01 2.89808303e-01 -7.55049586e-01 5.36191106e-01 -6.11516893e-01 5.47922313e-01 9.19987634e-02 -3.04362923e-01 2.03585938e-01 6.93156183e-01 -5.05912185e-01 -4.03374076e-01 -6.81274414e-01 -3.70316237e-01 8.75645280e-01 3.18907946e-01 -1.42740905e-01 -6.11564279e-01 -3.88282537e-01 -2.41741493e-01 4.02457535e-01 -1.04877090e+00 -5.09452641e-01 -5.41069731e-02 -9.33180392e-01 1.39815181e-01 7.26784348e-01 6.11057043e-01 -1.55142689e+00 -1.16931057e+00 8.71439725e-02 4.18152027e-02 -7.04810143e-01 -3.20329964e-01 7.60031581e-01 -6.28020227e-01 -7.10048854e-01 -7.85847008e-01 -8.42634976e-01 4.94342834e-01 -2.24080868e-02 8.04996371e-01 -3.66767973e-01 -3.95410746e-01 4.39426601e-01 -5.05598366e-01 -6.50308907e-01 8.36894393e-01 9.44339335e-02 -9.68434103e-03 5.05648673e-01 8.50637197e-01 -1.12857890e+00 -9.03901875e-01 1.26808107e-01 -6.70844138e-01 -2.49292135e-01 5.26027322e-01 7.15492964e-01 5.94346642e-01 -3.25458199e-02 8.61940384e-01 -2.67249465e-01 1.12890816e+00 -7.87462831e-01 -5.42627275e-02 -7.59661989e-03 -1.12589918e-01 -1.66399826e-04 3.65105957e-01 -5.29587686e-01 -8.45209420e-01 -8.29506963e-02 -2.56820410e-01 -5.34284413e-01 -3.39965791e-01 6.70463085e-01 -1.85109273e-01 -1.04037307e-01 3.77909780e-01 3.21020097e-01 -4.73152310e-01 -2.90651917e-01 -1.19710237e-01 7.78852284e-01 4.05862242e-01 -5.03802478e-01 1.25911757e-02 1.26668140e-01 -3.39319885e-01 -7.43249238e-01 -4.33904588e-01 -5.02000034e-01 -5.15628397e-01 -4.72026646e-01 1.14202249e+00 -8.47454429e-01 -9.08089519e-01 3.32705110e-01 -1.15503681e+00 -8.24802443e-02 3.41901556e-03 8.92944038e-01 -3.90939832e-01 3.43810231e-03 -5.09825945e-01 -7.50366390e-01 -5.98184526e-01 -1.00242722e+00 1.08407521e+00 6.64963841e-01 -4.46101695e-01 -7.54591167e-01 1.83877856e-01 -6.15613282e-01 3.98189217e-01 1.99685887e-01 8.70925725e-01 -5.00297606e-01 -7.37535283e-02 2.73756813e-02 -1.83603436e-01 1.13960560e-02 7.68216476e-02 -2.56271273e-01 -1.32236993e+00 -1.82401594e-02 3.82039100e-01 -2.31427416e-01 8.02367330e-01 6.51629925e-01 1.65901244e+00 8.71789604e-02 -2.86631137e-01 7.69470513e-01 1.28923285e+00 6.43636405e-01 8.81623685e-01 1.68611910e-02 1.70969695e-01 4.42720592e-01 1.27501041e-01 6.86623216e-01 -1.26427878e-02 4.30021167e-01 3.07092458e-01 -2.66671300e-01 4.33049470e-01 4.15427744e-01 3.00705254e-01 1.05073690e+00 -2.58985966e-01 -5.15276715e-02 -7.89396167e-01 6.47725165e-01 -1.67310381e+00 -1.03034472e+00 3.08269989e-02 1.75491500e+00 6.39830410e-01 -3.44948657e-02 -1.01268120e-01 1.77713633e-01 5.44542730e-01 -5.29031223e-03 -5.43965995e-01 -8.62664104e-01 7.79577484e-03 7.41415441e-01 -2.43927628e-01 -1.30081058e-01 -9.80025887e-01 4.13261145e-01 6.12742758e+00 5.46132267e-01 -1.53452694e+00 3.09900612e-01 5.90285003e-01 -2.18987405e-01 -1.01077087e-01 -6.07150912e-01 -2.70494461e-01 6.09025896e-01 9.83462811e-01 -3.55156720e-01 2.96963722e-01 6.03581607e-01 1.09128542e-01 4.77558933e-03 -9.63288426e-01 1.27488017e+00 8.71791393e-02 -9.23202753e-01 -2.72941351e-01 -2.02981934e-01 3.72412056e-01 -1.13796562e-01 1.48251668e-01 3.20784003e-01 -4.23487008e-01 -1.08083427e+00 6.30421221e-01 9.78246987e-01 6.92525625e-01 -1.01906562e+00 9.29084897e-01 -3.16985101e-02 -1.35212493e+00 -9.82042775e-02 -3.99245352e-01 -1.36109114e-01 -3.37679721e-02 6.41402423e-01 -2.30021328e-01 4.37098622e-01 1.18681550e+00 1.17845690e+00 -2.63746172e-01 1.04935682e+00 -4.21425253e-02 3.80627632e-01 -6.48660436e-02 -4.23377991e-01 2.66410083e-01 -8.35617185e-02 1.99580744e-01 1.45407033e+00 6.55857921e-01 6.34781837e-01 -2.67628223e-01 1.07422340e+00 6.65837899e-02 1.35132506e-01 -3.68047833e-01 -7.76547566e-02 3.37235034e-02 1.61900401e+00 -9.62855458e-01 -1.33247554e-01 -3.05691779e-01 1.14408338e+00 1.95294097e-01 4.12129462e-01 -1.03424942e+00 -7.91807950e-01 7.75857508e-01 -5.45878053e-01 1.32286221e-01 -1.25031143e-01 -2.45239452e-01 -1.10847533e+00 -1.26647905e-01 -5.28435588e-01 2.44316474e-01 -1.24913275e+00 -1.33865869e+00 1.10135031e+00 -3.58969688e-01 -1.10480571e+00 4.42504793e-01 -6.32618308e-01 -8.91669512e-01 1.18683779e+00 -1.45382023e+00 -5.22229254e-01 -6.12625360e-01 1.17821550e+00 5.13414323e-01 1.26513779e-01 1.09319985e+00 3.49180430e-01 -7.97340035e-01 3.65034819e-01 -2.79899627e-01 -8.04262161e-02 6.40434682e-01 -1.05092800e+00 -2.19922990e-01 5.12670696e-01 5.99958263e-02 6.27149403e-01 3.21335405e-01 -2.47954562e-01 -1.14799976e+00 -8.37030411e-01 6.09177530e-01 1.39384508e-01 5.44458508e-01 -4.70852524e-01 -8.88518691e-01 4.83400226e-01 8.14952970e-01 -7.19092563e-02 9.58061099e-01 3.41716707e-01 1.84474271e-02 -3.49995881e-01 -8.98678064e-01 2.27690488e-01 6.76869988e-01 -7.03932405e-01 -7.18211353e-01 1.76320925e-01 5.00168025e-01 -1.26248524e-01 -9.33984697e-01 2.54663408e-01 6.02079451e-01 -9.54654694e-01 6.85565174e-01 -4.15441096e-01 4.34425235e-01 -3.92057300e-02 1.13049531e-02 -1.63289523e+00 -7.75347650e-01 -3.64008158e-01 6.89705759e-02 9.15467620e-01 2.67470509e-01 -6.94560468e-01 2.33511761e-01 3.15782815e-01 -3.94063890e-01 -9.68432724e-01 -8.95848930e-01 -1.94835067e-01 -3.77152622e-01 -5.80599070e-01 8.13308239e-01 9.67159808e-01 7.72814393e-01 5.33039868e-01 -9.89113152e-02 2.26383075e-01 2.04411615e-02 3.32736224e-01 -4.31447625e-02 -1.22215760e+00 6.17686324e-02 -6.86570823e-01 -6.91553891e-01 -7.34285295e-01 1.67392522e-01 -8.56512606e-01 2.18927506e-02 -1.30755031e+00 4.12212461e-01 6.43790737e-02 -1.13057172e+00 6.20269001e-01 -2.56387256e-02 2.28447720e-01 -2.19537526e-01 -2.92260766e-01 -6.20522380e-01 9.52734292e-01 1.01303566e+00 -1.52184680e-01 -3.50988328e-01 -3.71156424e-01 -6.33565784e-01 6.63644373e-01 9.36357677e-01 -2.19874367e-01 -3.73923510e-01 -3.33447963e-01 2.54050016e-01 -1.01730518e-01 3.63793939e-01 -1.38508773e+00 2.64458358e-01 1.76184103e-01 1.01765847e+00 -6.05733275e-01 3.30151826e-01 -9.49967444e-01 7.94698969e-02 3.36430311e-01 -2.33508393e-01 3.08460861e-01 6.31801903e-01 2.35790640e-01 -3.95427495e-01 3.00817460e-01 7.35234737e-01 -8.73454884e-02 -7.89338768e-01 4.34339970e-01 -6.77813351e-01 -3.03165883e-01 1.08529997e+00 -2.10132688e-01 -1.55038089e-01 -1.40605032e-01 -1.04166257e+00 2.47842357e-01 -3.69845480e-01 3.06650937e-01 8.82066607e-01 -1.68152237e+00 -3.29909325e-01 7.24940717e-01 -1.36686014e-02 -5.93517721e-01 6.31492257e-01 1.18398106e+00 -1.12076011e-02 5.10508478e-01 -9.61483419e-01 -5.69481373e-01 -1.02460039e+00 5.80833673e-01 4.73357916e-01 1.46183167e-02 -4.31167841e-01 1.07789695e+00 2.58929551e-01 6.81162104e-02 1.45939082e-01 -6.24956131e-01 -8.13075006e-01 4.52080458e-01 5.19741833e-01 2.91118268e-02 9.56812054e-02 -4.77419674e-01 -7.05928385e-01 7.74746835e-01 2.67584890e-01 1.40811587e-02 1.84491551e+00 6.86755497e-03 -4.26899105e-01 6.57587171e-01 1.34395826e+00 -2.96422660e-01 -1.06532848e+00 1.11071371e-01 -1.30291760e-01 -9.22030732e-02 1.50271028e-01 -8.38313520e-01 -1.43766689e+00 1.26984441e+00 8.80277634e-01 4.30025131e-01 1.77229583e+00 -8.87275189e-02 6.49172068e-01 -5.26400469e-02 4.12275702e-01 -1.07470870e+00 1.24132343e-01 4.05831128e-01 1.08709502e+00 -7.02929318e-01 -2.59368718e-01 1.95750013e-01 -5.95596910e-01 1.21692801e+00 6.71233833e-01 -3.37106317e-01 1.12302291e+00 2.10855961e-01 -2.78582901e-01 -6.82051361e-01 -8.03358674e-01 -7.06416592e-02 6.37118816e-01 3.85826647e-01 5.14928997e-01 6.30997047e-02 -4.37711447e-01 1.48209286e+00 -5.82245849e-02 5.88310212e-02 4.15867455e-02 6.57835186e-01 -4.25746977e-01 -5.21773577e-01 -2.45652005e-01 4.12754655e-01 -4.87127244e-01 -4.04883251e-02 -2.86531866e-01 4.13877070e-01 4.66947615e-01 6.98833048e-01 4.64158416e-01 -8.83337200e-01 3.54409844e-01 4.07968193e-01 4.96981591e-01 -4.09022480e-01 -8.35395098e-01 3.45318556e-01 -3.92748058e-01 -9.16113555e-01 -7.21434712e-01 -4.42133635e-01 -1.38820529e+00 2.25320056e-01 -7.36041227e-03 3.46432537e-01 4.55946863e-01 8.03720236e-01 7.01590896e-01 1.10424268e+00 7.01822817e-01 -1.02080333e+00 3.69487941e-01 -1.23881435e+00 -9.96132374e-01 2.38160536e-01 2.42552489e-01 -7.66364396e-01 -1.86294436e-01 1.59642883e-02]
[13.096226692199707, 3.4180030822753906]
b1cc3d97-8efd-432c-b61c-7537f090410c
nonnegative-tucker-decomposition-with-beta
2110.14434
null
https://arxiv.org/abs/2110.14434v4
https://arxiv.org/pdf/2110.14434v4.pdf
Nonnegative Tucker Decomposition with Beta-divergence for Music Structure Analysis of Audio Signals
Nonnegative Tucker decomposition (NTD), a tensor decomposition model, has received increased interest in the recent years because of its ability to blindly extract meaningful patterns, in particular in Music Information Retrieval. Nevertheless, existing algorithms to compute NTD are mostly designed for the Euclidean loss. This work proposes a multiplicative updates algorithm to compute NTD with the beta-divergence loss, often considered a better loss for audio processing. We notably show how to implement efficiently the multiplicative rules using tensor algebra. Finally, we show on a music structure analysis task that unsupervised NTD fitted with beta-divergence loss outperforms earlier results obtained with the Euclidean loss.
['Frédéric Bimbot', 'Jérémy E. Cohen', 'Valentin Leplat', 'Florian Voorwinden', 'Axel Marmoret']
2021-10-27
null
null
null
null
['music-information-retrieval']
['music']
[ 1.90622613e-01 -4.42500204e-01 3.89845110e-02 -1.30662143e-01 -6.88472092e-01 -7.01295853e-01 3.09584498e-01 3.05979755e-02 -4.07551497e-01 1.76306844e-01 6.77285016e-01 -1.08716935e-01 -6.28102124e-01 -2.71257550e-01 -2.54108906e-01 -7.08746910e-01 -6.15282178e-01 2.56569624e-01 -1.92687571e-01 -1.08063288e-01 2.76792467e-01 5.06472349e-01 -1.47931600e+00 3.49809051e-01 4.74738121e-01 1.02462935e+00 -2.57994860e-01 6.94659591e-01 3.32013160e-01 5.69125891e-01 -3.25047880e-01 -7.41220057e-01 6.96514487e-01 -3.31672937e-01 -7.30718732e-01 9.94894886e-04 6.16817892e-01 -2.72245020e-01 -2.23237514e-01 1.11634147e+00 6.58995509e-01 2.93588638e-01 5.69497228e-01 -1.14434040e+00 -2.15514615e-01 7.23845184e-01 -4.59967077e-01 2.06567243e-01 5.39542139e-01 -4.82207596e-01 1.68063104e+00 -1.25519443e+00 4.43447381e-01 1.27458668e+00 9.48570073e-01 8.95454362e-02 -1.69512570e+00 -4.16265607e-01 -2.20848829e-01 3.74692649e-01 -1.33615780e+00 -3.57057422e-01 9.09015059e-01 -5.20597816e-01 7.57717311e-01 6.80345297e-01 4.44757730e-01 8.58769834e-01 -8.00330266e-02 1.05749118e+00 9.32168305e-01 -6.33869052e-01 6.54310128e-03 -4.16681319e-01 1.20430306e-01 4.53692108e-01 2.66676277e-01 2.32330449e-02 -8.99222434e-01 -5.17709851e-01 6.01645410e-01 -1.47744209e-01 -2.50461131e-01 -6.24763727e-01 -1.51821911e+00 5.81785560e-01 -4.39576469e-02 4.01264101e-01 -2.67301381e-01 2.90692359e-01 7.25140989e-01 6.64022982e-01 6.25530481e-01 5.20218492e-01 -2.49650046e-01 -4.92057532e-01 -1.08363569e+00 4.47666615e-01 9.02907610e-01 5.80524623e-01 3.26377273e-01 7.90023953e-02 -2.80647039e-01 1.09475398e+00 2.60274917e-01 2.19853312e-01 3.50783825e-01 -1.28795207e+00 5.44197738e-01 2.07646653e-01 7.30774775e-02 -1.29722202e+00 -2.73263693e-01 -7.39620686e-01 -1.12241220e+00 1.97628558e-01 6.05900466e-01 2.26981580e-01 -2.19163358e-01 1.77226412e+00 2.71501511e-01 3.17460895e-01 -8.78722742e-02 1.11618423e+00 1.81394428e-01 2.63921231e-01 -5.46277702e-01 -2.92979568e-01 1.21550810e+00 -5.45517147e-01 -7.86911726e-01 4.18356925e-01 4.51514125e-01 -1.41254437e+00 7.95493305e-01 1.12003076e+00 -1.21871567e+00 -2.96570510e-01 -1.08026862e+00 -1.84881493e-01 1.82109758e-01 2.60262340e-01 8.41522694e-01 9.00941849e-01 -9.99237537e-01 1.06617391e+00 -7.20776558e-01 -2.14682072e-01 -4.40218933e-02 2.27827474e-01 -2.58136034e-01 8.21559597e-03 -8.62609148e-01 4.16876137e-01 6.06667511e-02 2.23321334e-01 -7.60804534e-01 -6.63908541e-01 -3.11864585e-01 -6.98387474e-02 2.37921983e-01 -9.57909286e-01 1.20023203e+00 -5.37830472e-01 -1.53424168e+00 7.19293773e-01 -8.24575424e-02 -7.11817801e-01 6.18548751e-01 -3.85561675e-01 -2.54549652e-01 2.84633607e-01 3.19633521e-02 -2.49250066e-02 1.36794734e+00 -7.84768999e-01 -2.05147624e-01 -4.81190115e-01 -2.12334022e-02 2.23179549e-01 -4.61399227e-01 2.54521277e-02 -1.40126973e-01 -1.49115658e+00 8.39749634e-01 -8.85856688e-01 -1.82026342e-01 -6.51800931e-02 -3.19041222e-01 -2.41848290e-01 4.42519873e-01 -7.40317225e-01 1.43773675e+00 -2.36678338e+00 5.95303297e-01 4.22051966e-01 2.71697134e-01 -2.35012591e-01 -5.64968772e-02 7.00117111e-01 -3.17777693e-01 -1.79841116e-01 -4.31194216e-01 -6.48647785e-01 3.69813204e-01 1.90788954e-01 -6.81946099e-01 5.29358983e-01 -2.04415292e-01 2.24555254e-01 -8.77219081e-01 -1.52405441e-01 -2.03157067e-01 2.63709724e-01 -1.06099820e+00 -1.06516264e-01 2.44448841e-01 3.34468149e-02 1.69845652e-02 4.54171181e-01 5.59308648e-01 1.39321730e-01 1.05968215e-01 -5.43610215e-01 -1.80326670e-01 4.22986299e-01 -1.59929729e+00 1.96726489e+00 -9.47549418e-02 6.63677216e-01 5.47208905e-01 -9.89774346e-01 5.95704794e-01 5.15221417e-01 8.42858553e-01 -1.21155061e-01 -9.32808891e-02 4.96567905e-01 -5.11072129e-02 -9.38375741e-02 8.19471538e-01 -5.37395060e-01 4.87627089e-03 7.77892351e-01 8.34978595e-02 -1.89634740e-01 3.96963120e-01 4.01011169e-01 1.15440929e+00 -6.26584589e-02 -7.89777637e-02 -1.42799884e-01 4.80129927e-01 -3.01189989e-01 5.07076383e-01 6.56878293e-01 -2.95479149e-02 8.92568171e-01 5.03317177e-01 -2.20242158e-01 -9.39341366e-01 -1.27277529e+00 -2.46570170e-01 1.11628759e+00 -3.53354156e-01 -7.96754837e-01 -3.65337580e-01 -2.56455749e-01 1.06679171e-01 3.80957365e-01 -3.14667284e-01 3.12982593e-03 -4.12769228e-01 -6.45228922e-01 7.86931992e-01 2.42026687e-01 2.30822697e-01 -3.64205897e-01 -1.60929784e-02 4.05952334e-01 -3.55579793e-01 -7.56386220e-01 -7.88799882e-01 1.82323068e-01 -1.39101338e+00 -7.50001669e-01 -9.36119616e-01 -5.08392930e-01 2.26052627e-01 4.93416071e-01 9.22999918e-01 -4.55104083e-01 -3.46819967e-01 9.42095220e-01 -5.55778325e-01 -1.76499873e-01 -1.84317976e-01 -1.84767768e-01 6.33782446e-01 4.80817109e-01 2.63786554e-01 -1.17792273e+00 -3.97823960e-01 3.00677389e-01 -9.77887392e-01 -1.98836952e-01 3.72477740e-01 7.24522889e-01 6.17330492e-01 1.35024786e-01 1.52632728e-01 -2.81626910e-01 1.03506851e+00 -4.73691896e-02 -2.55614012e-01 -1.80289254e-01 -5.95901132e-01 2.39219770e-01 3.43928158e-01 -3.40624124e-01 -5.30768335e-01 -2.23785684e-01 2.55866759e-02 -7.46833503e-01 3.72014910e-01 6.10249639e-01 1.88593313e-01 -1.04304619e-01 5.71554363e-01 2.26717949e-01 -5.85463457e-02 -1.07634807e+00 4.56400335e-01 4.07987297e-01 5.31825781e-01 -9.27711904e-01 9.40956652e-01 6.81853652e-01 3.52397054e-01 -9.73918259e-01 -8.65240633e-01 -9.44797933e-01 -5.67142129e-01 -1.63238063e-01 5.09723306e-01 -8.46369207e-01 -8.77598226e-01 2.09196523e-01 -9.92810845e-01 3.57639939e-01 -5.09154916e-01 1.11535609e+00 -6.50659025e-01 8.66365075e-01 -7.20728338e-01 -9.95005548e-01 -4.02204126e-01 -6.06491983e-01 9.02260184e-01 -6.42176270e-01 -3.27169150e-01 -6.62935734e-01 4.60127652e-01 3.06385130e-01 7.02746063e-02 2.07376815e-02 8.83947432e-01 -3.73353004e-01 -4.14535195e-01 -2.05590457e-01 -9.79401357e-03 8.25115144e-01 -1.25118554e-01 -2.89232016e-01 -9.86278653e-01 -3.88109416e-01 2.93600798e-01 7.45764449e-02 1.01697135e+00 3.14420432e-01 9.44580734e-01 -3.76798332e-01 3.70634556e-01 5.65907061e-01 1.13759577e+00 -3.24114889e-01 3.57565343e-01 1.31417274e-01 5.09174228e-01 4.79588777e-01 3.51943195e-01 9.48293626e-01 -7.36160576e-02 8.42156112e-01 3.41432005e-01 5.03486931e-01 -2.17963904e-01 -2.30132103e-01 6.35320961e-01 1.67324638e+00 -5.52529514e-01 4.16708112e-01 -6.38347089e-01 4.40998584e-01 -1.84222519e+00 -1.24659073e+00 -2.75000870e-01 2.47549176e+00 6.78430021e-01 1.02706157e-01 4.73568112e-01 8.68032396e-01 3.48284751e-01 1.86033562e-01 -2.00821429e-01 -5.61989605e-01 -4.09840912e-01 3.15270245e-01 4.43206489e-01 4.93128926e-01 -1.14438868e+00 2.59611130e-01 6.68720388e+00 9.77543235e-01 -5.63642323e-01 1.66188002e-01 -2.02319682e-01 -1.47221550e-01 -3.96660984e-01 1.08652540e-01 -2.57803947e-01 2.98231188e-02 5.49979270e-01 -1.19564123e-01 6.76913381e-01 6.54320657e-01 3.27072084e-01 2.82733738e-01 -1.20736289e+00 1.42652166e+00 8.07222202e-02 -8.68923843e-01 2.45160848e-01 2.53619999e-01 6.63042784e-01 -9.20046121e-02 2.79333711e-01 -3.01809069e-02 -7.27808401e-02 -5.05387485e-01 9.18988287e-01 5.26666641e-01 2.38213271e-01 -7.44499862e-01 3.15434575e-01 1.16796300e-01 -1.25309491e+00 -4.76432927e-02 -5.50538421e-01 -3.49359602e-01 9.00908783e-02 1.10429668e+00 -7.73142576e-01 7.26489902e-01 5.55267453e-01 9.46071863e-01 -3.08971673e-01 1.53603220e+00 2.96963956e-02 8.40956211e-01 -6.39047444e-01 1.36948228e-01 2.62520581e-01 -9.03325260e-01 1.41725874e+00 1.12195373e+00 6.68103933e-01 -2.12724105e-01 -8.58159065e-02 5.82395256e-01 -2.57234648e-02 2.11465478e-01 -4.07453418e-01 -1.26811624e-01 -8.88968930e-02 8.78295183e-01 -4.43567872e-01 7.31686875e-02 -1.03274360e-01 1.10605454e+00 -1.87773615e-01 4.35383171e-01 -3.10833305e-01 -5.45615613e-01 9.93528724e-01 6.42089397e-02 4.42154825e-01 -7.33952582e-01 -3.01537037e-01 -1.40912485e+00 2.93247670e-01 -9.82231677e-01 5.35834014e-01 -6.58182919e-01 -1.55934227e+00 1.21000588e-01 -9.14348438e-02 -1.90698302e+00 -8.67790356e-02 -7.05924571e-01 -1.78480834e-01 7.08864868e-01 -8.28773558e-01 -5.90148628e-01 4.73733485e-01 6.91219985e-01 1.86934739e-01 -1.19734548e-01 9.59760070e-01 7.33745933e-01 -3.42749476e-01 5.62080681e-01 5.95002532e-01 9.50126871e-02 8.37785542e-01 -1.45659900e+00 -5.65351471e-02 8.02182198e-01 8.67203116e-01 8.62426102e-01 1.13543248e+00 -2.03050181e-01 -1.57958150e+00 -5.15277147e-01 9.69930112e-01 -3.02849025e-01 9.18182969e-01 -1.87483534e-01 -6.03623569e-01 4.42038059e-01 5.27785085e-02 -4.59082186e-01 1.03071129e+00 5.73638976e-01 -8.37467968e-01 -4.99495089e-01 -8.73189390e-01 5.78752637e-01 1.05471134e+00 -8.90170097e-01 -7.82184720e-01 3.71340483e-01 5.42188585e-01 -7.43174478e-02 -9.42799926e-01 4.00519997e-01 7.98523188e-01 -1.16844714e+00 1.18761969e+00 -6.13051891e-01 8.65924433e-02 -6.16304874e-01 -5.92564762e-01 -1.00304174e+00 -4.03859437e-01 -1.28241992e+00 -1.74627230e-01 9.62854028e-01 2.95580685e-01 -3.90301675e-01 6.22595251e-01 2.00651184e-01 -1.58400953e-01 -3.38595003e-01 -1.52038813e+00 -1.06525874e+00 -3.30144793e-01 -1.22753906e+00 2.62847811e-01 8.70949447e-01 2.13650376e-01 3.96287560e-01 -6.12079620e-01 -7.90912285e-03 1.05147231e+00 3.06528330e-01 5.85412025e-01 -1.27555311e+00 -8.23026299e-01 -6.99065924e-01 -7.42348373e-01 -1.37089264e+00 -2.84957528e-01 -1.25663030e+00 -4.25082594e-01 -1.17855060e+00 1.01453967e-01 -1.70410097e-01 -5.09588838e-01 1.19096518e-01 3.11913520e-01 5.43038011e-01 3.30321282e-01 4.09608871e-01 -6.20921075e-01 6.34602726e-01 1.13767004e+00 -2.52043307e-01 -9.99095067e-02 2.36683220e-01 -3.69703531e-01 7.87966013e-01 4.68416780e-01 -5.98190963e-01 -2.11832806e-01 -3.50652546e-01 1.00994754e+00 -1.17557049e-01 2.47090563e-01 -1.18214214e+00 1.70664117e-01 3.85152876e-01 1.65206157e-02 -8.50512862e-01 6.95621192e-01 -8.58914375e-01 3.62267457e-02 2.94436365e-01 -2.98499376e-01 -2.33617499e-02 -3.90028916e-02 8.66215885e-01 -5.74838221e-01 -4.23462391e-01 2.23024562e-01 3.19849133e-01 -1.99247897e-01 1.39854997e-01 -4.74184543e-01 -7.66166067e-03 2.83484161e-01 -2.36240596e-01 4.49577361e-01 -5.15541136e-01 -1.18623388e+00 -2.63378918e-01 -1.23889647e-01 2.46678129e-01 6.93866730e-01 -1.76440036e+00 -9.62177455e-01 2.32272536e-01 7.22180959e-03 -5.55436254e-01 1.62533671e-01 1.38745070e+00 -4.06283110e-01 2.87065387e-01 2.13580325e-01 -4.71912265e-01 -1.43588316e+00 3.26664537e-01 -1.16852954e-01 -4.35201168e-01 -5.03333569e-01 9.55874562e-01 -2.30390474e-01 -2.68966556e-01 4.63480800e-01 -4.13854390e-01 1.03493713e-01 6.06045008e-01 6.58526063e-01 7.85320759e-01 3.53153199e-01 -5.40635228e-01 -2.74674535e-01 5.99887609e-01 2.67088801e-01 -6.53900266e-01 1.36958873e+00 -1.22788258e-01 -6.18870497e-01 6.58442974e-01 1.32299662e+00 3.21904927e-01 -6.44964993e-01 -3.75435114e-01 1.96157590e-01 -6.41914010e-01 1.11177929e-01 -3.94521296e-01 -6.11737967e-01 9.36515331e-01 4.50900167e-01 5.68341136e-01 1.28517938e+00 -4.04402107e-01 7.50776172e-01 5.64645588e-01 5.46308041e-01 -9.19425070e-01 1.03822708e-01 6.93420351e-01 1.18588734e+00 -5.30517340e-01 2.34421670e-01 -1.47260994e-01 -2.75957733e-01 1.35468662e+00 -4.59887564e-01 -2.96020955e-01 9.91895854e-01 -6.77545890e-02 -2.94758886e-01 8.62539858e-02 -5.33522129e-01 -2.98401892e-01 6.28676116e-01 3.83140534e-01 5.69868684e-01 1.54175699e-01 -5.20618856e-01 4.88065720e-01 -4.66182381e-01 -2.23163843e-01 3.20949137e-01 5.76586723e-01 -2.36511737e-01 -1.42671645e+00 -7.69323528e-01 5.49937308e-01 -7.93371558e-01 -2.85788834e-01 -5.34471691e-01 2.12253034e-01 -7.45857134e-02 9.73372817e-01 -2.89319724e-01 -5.82651377e-01 4.44769293e-01 1.09254584e-01 7.76176512e-01 -4.46036935e-01 -6.71921849e-01 5.44172168e-01 2.05140501e-01 -6.01022959e-01 -7.55342245e-01 -9.01276171e-01 -6.97384894e-01 -4.29481655e-01 -1.26243740e-01 3.83079529e-01 5.66813588e-01 5.21532416e-01 2.03991622e-01 1.75320998e-01 6.69650435e-01 -8.37210655e-01 -9.19140697e-01 -8.91770720e-01 -1.16920805e+00 4.77059662e-01 3.20164263e-01 -5.29377699e-01 -8.15278232e-01 -7.96148553e-02]
[15.529335021972656, 5.482382774353027]
ec088b3f-c47e-459b-a399-ba7bc112f6f3
improving-fairness-and-robustness-in-end-to
2306.06083
null
https://arxiv.org/abs/2306.06083v1
https://arxiv.org/pdf/2306.06083v1.pdf
Improving Fairness and Robustness in End-to-End Speech Recognition through unsupervised clustering
The challenge of fairness arises when Automatic Speech Recognition (ASR) systems do not perform equally well for all sub-groups of the population. In the past few years there have been many improvements in overall speech recognition quality, but without any particular focus on advancing Equality and Equity for all user groups for whom systems do not perform well. ASR fairness is therefore also a robustness issue. Meanwhile, data privacy also takes priority in production systems. In this paper, we present a privacy preserving approach to improve fairness and robustness of end-to-end ASR without using metadata, zip codes, or even speaker or utterance embeddings directly in training. We extract utterance level embeddings using a speaker ID model trained on a public dataset, which we then use in an unsupervised fashion to create acoustic clusters. We use cluster IDs instead of speaker utterance embeddings as extra features during model training, which shows improvements for all demographic groups and in particular for different accents.
['Pascale Fung', 'Irina-Elena Veliche']
2023-06-06
null
null
null
null
['automatic-speech-recognition']
['speech']
[-3.16524170e-02 2.82824934e-01 -8.94857571e-02 -7.99414039e-01 -1.00786185e+00 -4.81549054e-01 6.99980497e-01 1.64429575e-01 -8.38722646e-01 5.62171221e-01 6.74720585e-01 -5.65860808e-01 1.27220199e-01 -3.70140672e-01 -2.28590101e-01 -4.00081426e-01 2.31186926e-01 2.96522409e-01 -5.48340440e-01 -2.33565599e-01 -2.17847135e-02 5.77879906e-01 -1.14296865e+00 -8.18499085e-03 8.23469877e-01 7.21277654e-01 -5.26445031e-01 6.91462278e-01 6.02926165e-02 5.90193152e-01 -8.09633136e-01 -1.09706318e+00 5.27274668e-01 -5.08409850e-02 -7.26790011e-01 -8.72992352e-02 5.32520235e-01 -5.70441127e-01 -7.00879216e-01 1.03902233e+00 9.72386360e-01 2.48781964e-01 4.01228160e-01 -1.15425813e+00 -8.94431472e-01 7.69276917e-01 -2.83827007e-01 -2.24763781e-01 9.19415727e-02 4.38583270e-03 1.00404847e+00 -5.71238101e-01 9.72673818e-02 1.37121618e+00 4.44771051e-01 9.62605476e-01 -1.19271863e+00 -8.44551384e-01 -4.99830581e-02 4.54790751e-03 -1.62881458e+00 -1.26370502e+00 4.04029995e-01 -1.32063523e-01 7.91798532e-01 8.37607920e-01 1.19615376e-01 1.07769823e+00 -5.07001460e-01 6.02700114e-01 8.48857284e-01 -3.03309977e-01 3.64144742e-01 5.52968025e-01 1.04016341e-01 1.02508202e-01 -5.33681437e-02 2.28419509e-02 -4.90462273e-01 -4.85650688e-01 4.12572213e-02 -1.76899147e-03 -9.30500701e-02 9.96119455e-02 -9.14931178e-01 1.01694059e+00 -8.03041682e-02 1.74930468e-01 -1.67217359e-01 -4.37202863e-02 4.47398752e-01 4.00463253e-01 6.80508971e-01 2.94141769e-01 -4.26139623e-01 -2.15846792e-01 -1.21729255e+00 2.81191021e-01 7.94310689e-01 5.19347310e-01 4.37809974e-01 4.80445214e-02 -2.82387465e-01 1.33115792e+00 2.27242693e-01 5.48927903e-01 5.50458729e-01 -1.05203879e+00 6.04831696e-01 1.00709140e-01 -2.20837872e-02 -1.08855140e+00 -2.01063186e-01 -4.07478780e-01 -9.04052258e-01 7.93526694e-02 5.11831462e-01 -3.04134518e-01 -7.21692741e-01 1.98670244e+00 1.29203260e-01 -4.71326485e-02 2.79765964e-01 7.23551273e-01 4.78673786e-01 5.69920421e-01 2.07631707e-01 1.10267485e-02 1.31964314e+00 -6.40763104e-01 -7.36634433e-01 -3.43659855e-02 5.57850480e-01 -7.25093782e-01 7.61322439e-01 2.89231479e-01 -1.07908833e+00 -2.03793406e-01 -7.92702496e-01 -1.67324781e-01 -5.05034268e-01 3.77197303e-02 3.71696502e-01 1.62138927e+00 -1.27648604e+00 3.37950438e-01 -4.79304850e-01 -2.22066641e-01 6.35418892e-01 5.93731642e-01 -7.80352175e-01 -2.68678367e-02 -1.33348131e+00 9.39880848e-01 -1.03182815e-01 -1.46283433e-01 -3.82041335e-01 -8.39565516e-01 -1.02118707e+00 3.32989812e-01 3.82015705e-02 -2.83352703e-01 1.14612436e+00 -8.40821028e-01 -1.71651947e+00 9.36183929e-01 -2.15007514e-01 -7.55233228e-01 6.18556619e-01 3.91403586e-02 -6.26390576e-01 -3.48437726e-01 -1.50981009e-01 5.40652335e-01 7.42916048e-01 -9.42164183e-01 -6.09618843e-01 -6.94153666e-01 -1.42845094e-01 3.19754630e-01 -6.59331203e-01 5.85934281e-01 -1.55358374e-01 -5.95090270e-01 -3.40992123e-01 -8.48714411e-01 -2.60079145e-01 -3.54455858e-01 -4.91421580e-01 -1.13769382e-01 5.79362929e-01 -1.30265462e+00 1.39524126e+00 -2.42171574e+00 -3.75145316e-01 4.68480259e-01 8.99026766e-02 5.33530653e-01 3.52733885e-03 1.36171475e-01 -2.00934738e-01 5.34612417e-01 -2.14860037e-01 -8.38126719e-01 4.01400536e-01 5.67726493e-02 -3.23824495e-01 5.12708604e-01 1.33700252e-01 4.30383354e-01 -4.92690712e-01 -1.45245671e-01 2.97661901e-01 4.88333881e-01 -8.10295641e-01 2.42009640e-01 3.89514655e-01 2.02273771e-01 -1.12139704e-02 2.94148475e-01 1.00329602e+00 4.70511675e-01 1.01160526e-01 4.26179111e-01 -6.95824921e-02 4.80642825e-01 -1.03745687e+00 1.34520078e+00 -4.66731995e-01 4.68564868e-01 4.04801518e-01 -1.12426353e+00 9.21214938e-01 5.27814090e-01 2.53009200e-01 -7.70259559e-01 -8.87462869e-03 2.67983437e-01 2.86831498e-01 -1.78649768e-01 7.02152610e-01 -3.36698234e-01 -3.25968236e-01 3.31728101e-01 -7.84007758e-02 2.09557310e-01 -4.43981230e-01 2.32208177e-01 8.95904541e-01 -8.07431817e-01 1.48107827e-01 -1.07662804e-01 6.58526719e-01 -5.48200488e-01 5.52722692e-01 6.29436016e-01 -5.69052994e-01 9.49952543e-01 3.08599204e-01 1.07098773e-01 -1.10757232e+00 -1.00573421e+00 -2.86544651e-01 1.20496488e+00 -6.05087936e-01 -2.30110243e-01 -8.57457697e-01 -6.18721604e-01 2.04978883e-01 1.06131136e+00 -4.40768600e-01 -1.61803737e-01 -2.48539358e-01 -7.81416535e-01 1.03534722e+00 8.72123390e-02 2.05908880e-01 -4.88354146e-01 5.30696437e-02 -1.54648786e-02 5.10821268e-02 -1.00416303e+00 -7.37728059e-01 7.40160514e-03 -1.71274394e-01 -2.55401254e-01 -1.16007674e+00 -3.03122818e-01 3.10546935e-01 -1.70546491e-02 6.27879620e-01 -1.42448872e-01 -1.83015525e-01 4.60589468e-01 -1.88439384e-01 -6.21945441e-01 -7.52695382e-01 1.75761715e-01 3.67193907e-01 4.56289709e-01 4.22335833e-01 -3.69835496e-01 -4.43910867e-01 2.35308141e-01 -5.46128333e-01 -4.60134208e-01 1.03819266e-01 7.17141688e-01 -1.15112513e-01 -1.08033441e-01 9.52882648e-01 -8.72087240e-01 6.72580540e-01 -5.12132525e-01 -2.01763213e-01 1.91627577e-01 -6.20796561e-01 -1.19717918e-01 6.56390667e-01 5.68398684e-02 -1.06515670e+00 -1.05603516e-01 -7.06002712e-01 -2.09194809e-01 -3.92416984e-01 1.28124312e-01 -6.42396092e-01 2.07323104e-01 3.52136970e-01 2.91779805e-02 4.84086096e-01 -4.66727048e-01 3.41322869e-01 1.63312864e+00 4.00948137e-01 -2.75401890e-01 5.66405714e-01 1.88809276e-01 -5.10941863e-01 -1.29056680e+00 -2.25275457e-01 -5.34586072e-01 -1.31111637e-01 2.52813339e-01 7.60480940e-01 -9.84064758e-01 -7.21512258e-01 4.95020211e-01 -8.54178607e-01 -4.15999293e-02 -1.84823409e-01 5.60028493e-01 -2.23964870e-01 4.19526875e-01 -3.78682226e-01 -1.31607723e+00 -4.24869388e-01 -1.09720790e+00 8.29555035e-01 6.06719963e-02 -3.62286240e-01 -7.71041632e-01 -1.64956525e-01 7.71922290e-01 6.73554361e-01 -1.97299853e-01 5.70830643e-01 -1.36151528e+00 -5.33763133e-02 -4.40977067e-01 -1.40780658e-01 8.17643166e-01 3.09910834e-01 -3.74381518e-04 -1.40297425e+00 -2.81874001e-01 -2.10931510e-01 -3.22675258e-02 6.20357037e-01 4.13241357e-01 1.35197186e+00 -7.44661212e-01 2.99585741e-02 4.24885869e-01 8.42161655e-01 1.96957961e-01 8.37796330e-01 -2.07431510e-01 7.40907311e-01 1.03603792e+00 1.70183212e-01 5.38090885e-01 5.81828475e-01 7.93587804e-01 1.08695531e-03 -7.15930611e-02 2.50788659e-01 -1.86254635e-01 4.86277193e-01 6.90989554e-01 2.60980099e-01 -1.81094259e-01 -8.73464525e-01 7.11970687e-01 -1.54784095e+00 -1.33433640e+00 2.37452671e-01 2.78180742e+00 9.10627246e-01 -3.73346895e-01 5.29998839e-01 9.11950022e-02 8.41846764e-01 1.62590295e-01 -2.81407714e-01 -8.16871464e-01 -2.01168627e-01 2.44526953e-01 8.41102064e-01 7.11762786e-01 -1.04676366e+00 8.07186365e-01 6.15351868e+00 9.31880236e-01 -1.21680927e+00 2.68311083e-01 1.30441022e+00 -2.99737841e-01 -4.61707294e-01 -1.36957899e-01 -5.31329513e-01 5.98402262e-01 1.46612644e+00 -4.11167294e-01 6.43743038e-01 8.14914703e-01 3.14088643e-01 3.99208963e-01 -9.07619238e-01 1.19509780e+00 7.76086375e-02 -1.13972056e+00 -3.29183429e-01 5.14701784e-01 3.65019947e-01 3.00552766e-03 5.48239768e-01 3.58971804e-01 3.67755622e-01 -1.39333749e+00 7.47087896e-01 1.24239385e-01 8.56335163e-01 -1.13201463e+00 5.96311152e-01 2.51633018e-01 -3.60713422e-01 -7.40269572e-02 -4.93928939e-01 1.46521643e-01 -1.21019132e-01 4.64333147e-01 -9.34909940e-01 5.19597709e-01 6.27405941e-01 8.85474607e-02 -4.62895155e-01 7.08216429e-01 4.21925992e-01 8.03062797e-01 -4.38265115e-01 -6.43333187e-03 -1.01208147e-02 -1.97509639e-02 3.11091036e-01 1.34364831e+00 3.34312290e-01 -5.29006720e-02 -4.12411839e-01 4.89163071e-01 -4.35913742e-01 4.96424854e-01 -6.28363609e-01 -2.51879096e-01 7.16205180e-01 1.16313946e+00 2.88167316e-03 -1.45462871e-01 -4.80712324e-01 1.03928876e+00 1.63089931e-01 2.12038934e-01 -5.12206912e-01 -4.28218693e-01 1.30230999e+00 1.75613329e-01 -8.78994837e-02 -3.36714089e-02 -3.91791046e-01 -1.17962205e+00 -1.52410492e-01 -1.28984952e+00 4.17843580e-01 -1.27391830e-01 -1.33540392e+00 7.67009556e-02 -4.15664673e-01 -7.20876038e-01 -3.67762625e-01 -4.36311841e-01 -5.44228673e-01 1.27335191e+00 -1.25893331e+00 -8.52565229e-01 4.72149760e-01 4.95138586e-01 5.57461791e-02 -4.69259948e-01 1.05041611e+00 7.28853822e-01 -6.25783265e-01 1.58234596e+00 3.80756229e-01 4.71820444e-01 9.18417573e-01 -1.16204572e+00 5.38638234e-01 7.72286892e-01 1.31166756e-01 9.52438533e-01 4.97570306e-01 -1.67475954e-01 -1.05186033e+00 -1.04535103e+00 1.27263927e+00 -5.95049918e-01 3.22636634e-01 -7.83821166e-01 -7.67014265e-01 5.77994466e-01 2.84346074e-01 -1.89701915e-01 1.19949877e+00 4.72834945e-01 -5.03450334e-01 -3.40675145e-01 -1.47198975e+00 5.75705767e-01 8.01341414e-01 -9.59032655e-01 -2.40283951e-01 1.99248567e-02 8.76890421e-01 -8.16072449e-02 -8.89174998e-01 -7.94828217e-03 5.83456278e-01 -8.65739465e-01 9.11529779e-01 -7.13932037e-01 -1.40360249e-02 6.76377094e-05 -5.86343944e-01 -1.26047218e+00 -1.08499192e-01 -8.96968365e-01 3.80938321e-01 1.91539490e+00 7.23228037e-01 -1.02819490e+00 9.15964782e-01 1.48509371e+00 8.82990435e-02 -2.55294889e-01 -1.30643117e+00 -5.66399872e-01 4.31515157e-01 -5.90592384e-01 1.00212014e+00 1.09339559e+00 6.00119531e-02 8.05771202e-02 -6.18512034e-01 3.23703557e-01 5.80771983e-01 -5.31484127e-01 9.88653421e-01 -8.20859849e-01 -2.29403466e-01 -5.67658544e-01 -5.40779293e-01 -4.91850436e-01 4.05360013e-01 -1.06895626e+00 -2.01673716e-01 -9.61069107e-01 5.91333583e-02 -6.68405712e-01 -3.30774486e-01 4.04937267e-01 -2.65504837e-01 -9.84239057e-02 3.37625891e-01 -1.87648982e-01 -1.09438762e-01 5.41463435e-01 4.10150826e-01 -1.92249566e-01 -3.19944397e-02 3.41270000e-01 -1.32298470e+00 2.33109996e-01 1.00088418e+00 -3.64912093e-01 -2.56531924e-01 -3.65929097e-01 -4.52212304e-01 -6.97094575e-02 2.99074173e-01 -9.31125998e-01 8.26275907e-03 -8.39663893e-02 2.43312314e-01 -9.12143290e-02 2.88969934e-01 -9.20087039e-01 1.56105578e-01 2.33953729e-01 -6.59127533e-01 -2.36612052e-01 1.36361225e-02 3.22253525e-01 -6.85872659e-02 -2.50460923e-01 8.71995151e-01 1.73472896e-01 -1.14429995e-01 4.58842367e-01 -3.16156328e-01 5.03728203e-02 6.63849711e-01 2.49055438e-02 -4.63006347e-02 -8.67874622e-01 -6.68032944e-01 1.72282264e-01 5.50647974e-01 5.91288686e-01 3.42389882e-01 -1.22435224e+00 -1.07804811e+00 3.23725373e-01 1.71366096e-01 -4.89474505e-01 5.67845583e-01 4.47682023e-01 -1.27889961e-01 4.49334294e-01 1.08572215e-01 -1.52611986e-01 -1.44559765e+00 4.38318968e-01 3.71708632e-01 5.72259426e-02 4.05898923e-03 9.54371810e-01 1.12540536e-01 -9.54536259e-01 4.04327363e-01 4.64521646e-02 -2.52250079e-02 1.05596170e-01 7.96073794e-01 3.99300396e-01 2.63594538e-01 -1.02032161e+00 -4.72288907e-01 -1.53171957e-01 -1.40336692e-01 -5.60426176e-01 1.13661945e+00 -3.54024053e-01 1.91748649e-01 1.23550110e-01 1.63439131e+00 4.43513811e-01 -7.99212933e-01 -2.83605736e-02 -1.57166377e-01 -7.64050841e-01 -2.03528740e-02 -6.93314254e-01 -1.07541907e+00 1.13846517e+00 6.52720153e-01 3.25402588e-01 7.48735070e-01 -1.75821543e-01 7.66264796e-01 -7.54875988e-02 2.56127566e-01 -1.31066740e+00 -5.99670470e-01 2.52378851e-01 6.25300288e-01 -1.23276401e+00 -3.61990213e-01 9.06311721e-02 -9.89530504e-01 6.16377771e-01 7.11668879e-02 3.67447346e-01 7.10248530e-01 1.88056510e-02 2.87847370e-01 5.07430077e-01 -3.30342114e-01 -9.27770063e-02 -2.92863902e-02 8.81020606e-01 4.95894700e-01 5.05844057e-01 -1.92830220e-01 9.14528251e-01 -6.37353897e-01 -4.35797125e-01 3.87789935e-01 3.15118551e-01 -2.26172715e-01 -1.44835234e+00 -3.68712217e-01 6.43773377e-01 -1.05912471e+00 -2.29197949e-01 -3.46416563e-01 2.50482172e-01 -1.88085772e-02 1.30768645e+00 1.83959559e-01 -6.26496255e-01 3.59862477e-01 3.48283470e-01 -1.84919760e-02 -6.16804659e-01 -8.32456052e-01 -4.04218405e-01 5.08567512e-01 -2.45991990e-01 1.76458672e-01 -1.08790529e+00 -1.02423728e+00 -9.46653962e-01 -1.57135636e-01 3.31118315e-01 1.06105554e+00 5.79972804e-01 6.90619469e-01 2.04558112e-02 1.07471657e+00 -3.29180241e-01 -1.07919264e+00 -7.10848212e-01 -6.96042955e-01 4.59954470e-01 4.67542619e-01 3.09779886e-02 -4.17693198e-01 -3.17169547e-01]
[14.00651741027832, 5.917810916900635]
bee909fc-7a66-41c2-9123-9a590a6511ff
cuda-optimized-real-time-rendering-of-a
2012.08655
null
https://arxiv.org/abs/2012.08655v1
https://arxiv.org/pdf/2012.08655v1.pdf
CUDA-Optimized real-time rendering of a Foveated Visual System
The spatially-varying field of the human visual system has recently received a resurgence of interest with the development of virtual reality (VR) and neural networks. The computational demands of high resolution rendering desired for VR can be offset by savings in the periphery, while neural networks trained with foveated input have shown perceptual gains in i.i.d and o.o.d generalization. In this paper, we present a technique that exploits the CUDA GPU architecture to efficiently generate Gaussian-based foveated images at high definition (1920x1080 px) in real-time (165 Hz), with a larger number of pooling regions than previous Gaussian-based foveation algorithms by several orders of magnitude, producing a smoothly foveated image that requires no further blending or stitching, and that can be well fit for any contrast sensitivity function. The approach described can be adapted from Gaussian blurring to any eccentricity-dependent image processing and our algorithm can meet demand for experimentation to evaluate the role of spatially-varying processing across biological and artificial agents, so that foveation can be added easily on top of existing systems rather than forcing their redesign (emulated foveated renderer). Altogether, this paper demonstrates how a GPU, with a CUDA block-wise architecture, can be employed for radially-variant rendering, with opportunities for more complex post-processing to ensure a metameric foveation scheme. Code is provided.
['Tomaso Poggio', 'Arturo Deza', 'Elian Malkin']
2020-12-15
null
https://openreview.net/forum?id=ZMsqkUadtZ7
https://openreview.net/pdf?id=ZMsqkUadtZ7
neurips-workshop-svrhm-2020-12
['foveation']
['computer-vision']
[ 3.83584291e-01 -1.04806952e-01 8.59854877e-01 -8.73059705e-02 -9.24102515e-02 -6.27597332e-01 6.30194902e-01 -1.93292841e-01 -6.97120667e-01 6.36032343e-01 -2.22658291e-01 -4.66473877e-01 -1.03020459e-01 -7.08004653e-01 -4.95546162e-01 -6.57703400e-01 -2.57450104e-01 -2.78835952e-01 6.61121726e-01 -3.87429565e-01 3.71430606e-01 9.19292986e-01 -2.26346922e+00 3.76517594e-01 6.20563507e-01 9.43863809e-01 7.28034377e-01 1.26221251e+00 4.49029922e-01 5.08239865e-01 -9.01532650e-01 9.05307531e-02 4.10992503e-01 -5.36730513e-02 -3.02524894e-01 -7.81868473e-02 8.53032112e-01 -4.38050747e-01 -3.48981291e-01 5.97623348e-01 1.02449226e+00 2.45961428e-01 4.13414299e-01 -8.15967977e-01 -8.79571617e-01 -2.56315500e-01 -6.29124045e-01 6.43992126e-01 3.95275414e-01 5.82124650e-01 1.08072795e-01 -7.23056912e-01 6.28858685e-01 1.18320787e+00 4.30385083e-01 4.08972204e-01 -1.40699041e+00 -1.91243798e-01 -3.31104785e-01 -4.06630114e-02 -1.45669520e+00 -3.47270846e-01 2.75625646e-01 -3.99353653e-01 1.56717288e+00 5.54787457e-01 9.08962369e-01 6.07745886e-01 5.18033028e-01 6.65462688e-02 1.35222566e+00 -3.64276111e-01 2.93680042e-01 1.21043123e-01 -3.73447120e-01 5.50207853e-01 2.20017791e-01 4.85859931e-01 -2.05726430e-01 -8.47083256e-02 1.71235812e+00 -4.88838911e-01 -7.80426085e-01 -2.79098690e-01 -1.20418763e+00 4.81926888e-01 5.73819220e-01 3.00911278e-01 -4.89967525e-01 3.13830405e-01 4.53085572e-01 1.70839921e-01 2.89731473e-01 8.56921196e-01 -2.87642658e-01 -1.94283664e-01 -8.57187808e-01 4.80134636e-01 3.21020335e-01 8.47404957e-01 3.36892635e-01 4.74043161e-01 -4.43080030e-02 6.88749135e-01 -5.83746731e-02 3.89815032e-01 6.33034289e-01 -1.42276585e+00 -2.29785979e-01 1.12934755e-02 3.75064671e-01 -9.71715748e-01 -5.22535443e-01 -4.50576097e-01 -5.21191299e-01 1.20373523e+00 5.20697594e-01 -3.62436354e-01 -9.39031541e-01 1.34473157e+00 4.10154253e-01 3.99583727e-02 -3.43605946e-03 1.13893175e+00 6.62949920e-01 5.75173438e-01 -1.49275690e-01 -5.76992966e-02 1.58783352e+00 -3.69068712e-01 -5.23650706e-01 9.26407650e-02 2.32212022e-01 -9.32043254e-01 1.19408381e+00 6.43829644e-01 -1.29819989e+00 -7.10297465e-01 -1.30658948e+00 -2.59927273e-01 -3.45350504e-01 -1.10234760e-01 9.16654110e-01 9.89319801e-01 -1.72816110e+00 5.60688674e-01 -6.92152917e-01 -3.16202015e-01 2.15051085e-01 4.52076435e-01 -2.21254542e-01 2.15945140e-01 -9.16670442e-01 1.05126226e+00 3.84471625e-01 1.21451467e-01 -3.30471963e-01 -8.63056421e-01 -6.86422050e-01 8.05869233e-03 -2.13475985e-04 -1.10722208e+00 1.08642364e+00 -9.52872455e-01 -1.66737163e+00 7.08516121e-01 -2.98109408e-02 -5.55794179e-01 4.71604139e-01 1.41284123e-01 -2.56597787e-01 3.69482249e-01 -4.35354680e-01 1.04204750e+00 8.70553553e-01 -1.27442193e+00 -7.32447445e-01 -2.44809017e-01 3.20030332e-01 4.22380447e-01 2.30247676e-01 3.30475271e-01 -1.02406479e-01 -4.25361097e-01 -2.29741886e-01 -6.72306418e-01 -2.53766924e-01 2.92166859e-01 3.81524444e-01 3.38050537e-02 7.39362657e-01 -6.38352096e-01 9.81929719e-01 -2.14319944e+00 -3.11613351e-01 -1.00820102e-01 4.59065169e-01 5.15028179e-01 -4.42598760e-02 6.34788126e-02 -2.13093519e-01 -3.05472791e-01 2.53140423e-02 -5.33348508e-02 -2.96733797e-01 -5.70067316e-02 -1.64789468e-01 5.70177555e-01 1.50383204e-01 6.35422170e-01 -8.56718957e-01 -1.56061754e-01 4.69047517e-01 1.11696935e+00 -5.92347324e-01 -8.46914798e-02 1.60759330e-01 2.08251476e-01 2.26191670e-01 2.67684668e-01 9.38756645e-01 -5.59320301e-02 -1.97598353e-01 -1.28198057e-01 -6.05667353e-01 -1.72592491e-01 -1.15232170e+00 1.58108795e+00 -5.65631986e-01 1.30041921e+00 3.04030150e-01 -1.82479516e-01 9.52271760e-01 2.60936409e-01 1.42893359e-01 -8.63142788e-01 5.02598107e-01 2.49618739e-01 2.74347901e-01 -4.02125984e-01 1.03754425e+00 -2.76382156e-02 5.36580086e-01 1.27392441e-01 -1.67607054e-01 -3.38789493e-01 -2.07668319e-02 -1.04731873e-01 7.64735520e-01 4.65450794e-01 6.84703439e-02 -4.22196597e-01 1.36139274e-01 1.09107375e-01 -1.82583809e-01 5.75582564e-01 -2.98289657e-01 1.01831901e+00 1.40065834e-01 -4.21579212e-01 -1.42091608e+00 -9.38620448e-01 -4.87685591e-01 8.51822078e-01 7.50774518e-03 5.09163737e-02 -1.02139723e+00 4.07698184e-01 -3.03722054e-01 6.36124671e-01 -6.20054603e-01 2.03208044e-01 -6.52351320e-01 -5.58335006e-01 4.61935133e-01 4.11451936e-01 3.97519439e-01 -1.11265647e+00 -1.74337649e+00 2.05420896e-01 4.34708297e-01 -7.48154759e-01 -6.95248619e-02 1.37025923e-01 -6.84667945e-01 -7.74290025e-01 -1.18905985e+00 -7.44322896e-01 4.84045923e-01 6.26615167e-01 8.52278292e-01 -2.21099928e-01 -6.45810902e-01 4.01057541e-01 -3.04918159e-02 -3.31126064e-01 3.43915187e-02 -4.76995528e-01 5.27381971e-02 -4.22232598e-01 3.14164124e-02 -6.78411305e-01 -1.01226652e+00 2.12808535e-01 -9.70566034e-01 2.43729159e-01 2.25745603e-01 5.64154685e-01 3.84338468e-01 -7.22962022e-02 3.35881323e-01 -3.75754327e-01 9.79443133e-01 4.23307456e-02 -8.57212543e-01 -2.30948538e-01 -3.11908007e-01 -2.25815162e-01 5.69859684e-01 -6.69383049e-01 -1.33787882e+00 -1.51845485e-01 7.29185566e-02 -6.04702890e-01 -2.79228747e-01 1.38333946e-01 3.30242932e-01 -4.54938114e-01 1.39710331e+00 -3.43462713e-02 2.48856246e-01 7.51539841e-02 5.39301693e-01 7.63563514e-01 8.72325301e-01 -1.64273053e-01 3.42386901e-01 6.59793556e-01 -6.01909161e-02 -1.07171464e+00 3.52415800e-01 -6.46849275e-02 -3.77285451e-01 -3.57885748e-01 9.15250778e-01 -8.95634890e-01 -9.03988242e-01 5.46803296e-01 -1.24535394e+00 -5.79674363e-01 -3.82121176e-01 6.42316341e-01 -7.69725740e-01 1.24843635e-01 -4.55745429e-01 -8.90672386e-01 -1.96598753e-01 -1.08999228e+00 7.87585616e-01 7.45106518e-01 -3.28133345e-01 -7.32961953e-01 -5.03466371e-03 7.74068618e-03 8.34645271e-01 4.00830626e-01 5.52184165e-01 2.14781195e-01 -6.87396288e-01 -1.79372858e-02 -6.17024124e-01 4.92971614e-02 2.51385868e-02 2.75257289e-01 -1.43685782e+00 -3.50628108e-01 1.88636303e-01 -4.68400493e-03 5.25368869e-01 8.67873371e-01 6.09375536e-01 3.08223013e-02 -9.71258506e-02 8.09477866e-01 1.55449152e+00 5.26447296e-01 1.12034810e+00 5.49391806e-01 4.32842821e-01 6.11958981e-01 4.61916775e-01 3.63885820e-01 2.48947274e-02 6.62692249e-01 4.10976619e-01 -5.14799356e-01 -4.38889623e-01 3.89778018e-01 -7.75688142e-02 -9.93669927e-02 -5.54647386e-01 -1.56666845e-01 -6.46468759e-01 4.28170741e-01 -1.32972133e+00 -1.04405594e+00 -2.28846654e-01 2.50978088e+00 4.99977559e-01 -9.49674286e-03 1.91541120e-01 -7.23022223e-02 7.20810175e-01 -4.33464162e-02 -4.84965324e-01 -1.11908901e+00 -1.76587492e-01 4.70829904e-01 5.41659772e-01 5.70703804e-01 -6.91482604e-01 6.37821376e-01 6.93179274e+00 6.65427804e-01 -1.61673319e+00 -3.33274752e-02 5.31472802e-01 -4.10027236e-01 -1.06211998e-01 -3.62310588e-01 -3.80578279e-01 3.13234746e-01 1.08959663e+00 -2.25356221e-01 6.66901708e-01 7.34179854e-01 4.86950785e-01 -5.22669435e-01 -7.04077065e-01 1.22444999e+00 7.91635215e-02 -1.31116629e+00 -4.25294399e-01 2.01330468e-01 5.10803580e-01 1.51259691e-01 4.54619318e-01 -1.41525537e-01 5.49375173e-03 -1.15419996e+00 6.22132897e-01 4.66957211e-01 1.13387799e+00 -8.09880495e-01 3.59536707e-01 1.01829618e-01 -1.00524485e+00 1.21912599e-01 -8.01293671e-01 -2.73128062e-01 1.05264569e-02 3.55193168e-02 -8.83084774e-01 2.99560040e-01 7.77260780e-01 -1.45667240e-01 -5.89633048e-01 1.41862798e+00 3.06328058e-01 -2.17928037e-01 -5.77374041e-01 -2.32469440e-01 1.24158159e-01 -1.33855343e-01 4.54909384e-01 1.10787535e+00 4.89658892e-01 2.02569708e-01 -6.83444917e-01 9.37463701e-01 6.93499386e-01 -6.38305862e-03 -5.86211920e-01 4.75950956e-01 3.59470248e-01 1.24513018e+00 -7.83367872e-01 -2.49249384e-01 -2.89808184e-01 1.14081085e+00 8.09084550e-02 6.37335300e-01 -9.23226893e-01 -8.22004080e-01 6.10134602e-01 4.26320344e-01 5.03266394e-01 -4.00912315e-01 -2.52330244e-01 -7.50609696e-01 -1.53955832e-01 -7.11807847e-01 -2.89622992e-01 -1.51272666e+00 -5.87476730e-01 1.23007882e+00 -3.60646322e-02 -1.02678287e+00 -4.20701593e-01 -8.28409314e-01 -4.29012746e-01 1.49086440e+00 -1.20014834e+00 -8.46742630e-01 -3.46810728e-01 5.36159337e-01 1.76107869e-01 1.26189530e-01 6.80989683e-01 1.98684350e-01 1.13158248e-01 4.07346874e-01 -6.41017780e-02 -4.67704952e-01 5.44288516e-01 -1.20253348e+00 4.35316622e-01 8.38882506e-01 -1.83603942e-01 7.47434020e-01 9.77689862e-01 -3.52947801e-01 -1.15195751e+00 -7.72840798e-01 3.79682600e-01 -3.87361467e-01 3.18811357e-01 -2.46808127e-01 -9.65108454e-01 2.58593321e-01 6.21073484e-01 4.37086038e-02 1.82084143e-01 -2.98462898e-01 -7.59886950e-02 1.70143142e-01 -1.34885263e+00 9.95193005e-01 8.28884244e-01 -4.65833426e-01 -2.36499280e-01 6.30900543e-03 7.11416006e-01 -8.33420634e-01 -7.37402141e-01 -6.83078915e-03 7.21607327e-01 -1.41050065e+00 9.83810663e-01 4.83610407e-02 1.22083157e-01 -7.23190904e-01 1.48704395e-01 -1.32565272e+00 -4.59152788e-01 -8.65581453e-01 2.60766447e-01 6.54085636e-01 1.90520659e-01 -9.30442989e-01 5.32782674e-01 7.00100422e-01 -2.80019224e-01 -6.18838310e-01 -1.07088006e+00 -5.05656660e-01 -2.03598678e-01 -2.67932475e-01 2.52985746e-01 6.10107958e-01 -7.21274465e-02 5.50971553e-02 -5.38090132e-02 2.69794703e-01 3.44196528e-01 -2.60963768e-01 6.76369607e-01 -8.32390308e-01 -5.15708923e-01 -5.73365808e-01 -7.97395587e-01 -1.01436758e+00 -7.13654935e-01 -3.87670338e-01 -1.43568590e-01 -1.43514609e+00 -3.96281093e-01 -3.29500377e-01 3.15473735e-01 4.89907227e-02 -4.32142802e-02 4.71968174e-01 1.84584424e-01 1.03036119e-02 2.86816150e-01 1.16868861e-01 1.56249869e+00 5.13319850e-01 -5.22261977e-01 -4.15373087e-01 -5.17545640e-01 5.48993170e-01 6.69510543e-01 1.51020721e-01 -7.64493048e-01 -4.66343194e-01 1.05016038e-01 4.83045764e-02 5.82326174e-01 -1.33653069e+00 2.54209638e-01 1.92560539e-01 6.76657438e-01 -2.40321994e-01 6.50578558e-01 -6.48344934e-01 4.49606448e-01 3.43846142e-01 4.84084599e-02 3.03879738e-01 5.82891762e-01 2.79243559e-01 1.72182828e-01 -7.80938640e-02 9.67267931e-01 -1.43780023e-01 -6.58394754e-01 -9.28880349e-02 -6.92494452e-01 -3.89921993e-01 1.12771451e+00 -8.22707772e-01 -6.55908406e-01 -2.74271488e-01 -6.66578054e-01 -4.27117258e-01 9.74932075e-01 1.30859062e-01 7.60190070e-01 -9.07522321e-01 -5.80338538e-01 4.24296677e-01 -3.24801475e-01 -2.65720896e-02 6.16144061e-01 6.33942008e-01 -1.22197890e+00 5.08053064e-01 -6.49183452e-01 -6.82213247e-01 -1.27963138e+00 8.38190854e-01 7.01268733e-01 4.54468340e-01 -7.89073646e-01 9.54458535e-01 4.19304758e-01 2.27948904e-01 -5.11254817e-02 -5.32041132e-01 -2.92822599e-01 -1.60650775e-01 9.02482927e-01 3.14228088e-01 6.30897209e-02 -4.27198976e-01 -4.61141467e-02 4.83922362e-01 1.12401456e-01 -4.35377032e-01 1.05865121e+00 -1.99933082e-01 1.82174936e-01 1.89355776e-01 8.28552485e-01 2.65443064e-02 -1.78768575e+00 3.26225460e-01 -7.29882419e-01 -8.18360925e-01 3.55550796e-01 -8.17668378e-01 -9.30722356e-01 7.97680855e-01 1.04266965e+00 3.33913118e-01 1.28137660e+00 -3.16420197e-01 3.29879761e-01 -1.18933581e-01 4.90123957e-01 -8.44442666e-01 -2.85495877e-01 1.34432390e-01 1.00023580e+00 -5.21229565e-01 -4.35857214e-02 -5.66019058e-01 -6.34447038e-01 1.04247141e+00 5.49173534e-01 -5.24596870e-01 2.05366790e-01 6.11053586e-01 1.62402183e-01 -1.77775592e-01 -8.08076262e-01 -2.01615587e-01 2.41592098e-02 1.22243345e+00 4.45403844e-01 -3.55301648e-02 -1.97705477e-01 -1.73627347e-01 -4.20109093e-01 8.27022716e-02 9.36758578e-01 1.02091897e+00 -6.34672642e-01 -4.94218647e-01 -5.71190417e-01 1.70858860e-01 -2.38150001e-01 -3.19366932e-01 2.15346143e-01 7.94823468e-01 2.10150629e-01 7.53092945e-01 6.06607437e-01 -2.61229407e-02 4.06340480e-01 -3.36901218e-01 1.01161635e+00 -3.08144540e-01 -7.59053469e-01 3.27815354e-01 7.03856423e-02 -5.58409095e-01 -1.97287783e-01 -4.89301175e-01 -1.16740787e+00 -5.63646793e-01 -2.36263916e-01 -3.63294214e-01 9.11346316e-01 2.56981701e-01 6.29011035e-01 8.10832143e-01 7.78684467e-02 -1.37896836e+00 2.47188453e-02 -7.98052073e-01 -6.45182312e-01 -1.22596174e-01 3.39462399e-01 -6.17341518e-01 -4.22615975e-01 1.19623885e-01]
[11.019867897033691, -2.1465799808502197]
e34e83d4-d151-4bb3-9283-36cf9bf1f968
structure-clip-enhance-multi-modal-language
2305.06152
null
https://arxiv.org/abs/2305.06152v1
https://arxiv.org/pdf/2305.06152v1.pdf
Structure-CLIP: Enhance Multi-modal Language Representations with Structure Knowledge
Large-scale vision-language pre-training has shown promising advances on various downstream tasks and achieved significant performance in multi-modal understanding and generation tasks. However, existing methods often perform poorly on image-text matching tasks that require a detailed semantics understanding of the text. Although there have been some works on this problem, they do not sufficiently exploit the structural knowledge present in sentences to enhance multi-modal language representations, which leads to poor performance. In this paper, we present an end-to-end framework Structure-CLIP, which integrates latent detailed semantics from the text to enhance fine-grained semantic representations. Specifically, (1) we use scene graphs in order to pay more attention to the detailed semantic learning in the text and fully explore structured knowledge between fine-grained semantics, and (2) we utilize the knowledge-enhanced framework with the help of the scene graph to make full use of representations of structured knowledge. To verify the effectiveness of our proposed method, we pre-trained our models with the aforementioned approach and conduct experiments on different downstream tasks. Numerical results show that Structure-CLIP can often achieve state-of-the-art performance on both VG-Attribution and VG-Relation datasets. Extensive experiments show its components are effective and its predictions are interpretable, which proves that our proposed method can enhance detailed semantic representation well.
['Wen Zhang', 'Zhipeng Hu', 'Tangjie Lv', 'Zeng Zhao', 'WeiJie Chen', 'Xinfeng Zhang', 'Rongsheng Zhang', 'Zhuo Chen', 'Jiji Tang', 'Yufeng Huang']
2023-05-06
null
null
null
null
['text-matching']
['natural-language-processing']
[ 3.11876297e-01 2.96669826e-02 -3.55254799e-01 -5.06813228e-01 -6.88995600e-01 -2.53426552e-01 8.90351593e-01 -2.14725249e-02 -2.28342697e-01 4.86283749e-01 6.54017568e-01 -8.44363570e-02 -2.54016761e-02 -8.21963847e-01 -8.48178089e-01 -4.69122708e-01 5.79407513e-01 3.14779490e-01 2.99317598e-01 -3.14793348e-01 1.06841505e-01 -2.30190203e-01 -1.52906108e+00 6.11274540e-01 9.69019473e-01 9.68691587e-01 6.16790652e-01 1.54965669e-01 -5.76884270e-01 7.95586705e-01 -2.09736004e-01 -4.90905881e-01 -2.15513855e-02 -3.84717554e-01 -8.45534742e-01 4.16066736e-01 3.31287175e-01 -2.39985764e-01 -5.72910845e-01 1.10317123e+00 2.87118971e-01 1.08882017e-01 6.29365385e-01 -1.26835787e+00 -9.64657903e-01 6.58634305e-01 -7.62806237e-01 -7.16088340e-02 2.11826950e-01 6.28359243e-02 1.03244805e+00 -9.18460548e-01 5.46026111e-01 1.58141565e+00 4.93521124e-01 4.38269943e-01 -8.37871730e-01 -6.22166097e-01 5.01977742e-01 5.12604296e-01 -1.30937088e+00 -2.75413722e-01 9.42837596e-01 -2.37428337e-01 8.28791916e-01 -2.27932200e-01 4.27834958e-01 1.11440074e+00 -1.82112038e-01 1.17186737e+00 1.12786996e+00 -4.65823621e-01 -2.64464051e-01 2.02091023e-01 2.53434926e-01 9.97784197e-01 1.13585547e-01 -9.58771557e-02 -4.92079943e-01 1.19200714e-01 5.50927818e-01 2.98582047e-01 -3.08445901e-01 -2.68396080e-01 -1.31910729e+00 9.40627933e-01 7.41510093e-01 3.61418098e-01 -3.12089294e-01 3.67622554e-01 5.53671718e-01 -8.42986815e-03 4.73181814e-01 -6.57551065e-02 -1.85317010e-01 1.86477572e-01 -8.33786666e-01 3.37267295e-02 3.29572618e-01 1.06873930e+00 9.76738393e-01 -6.74206540e-02 -4.15549964e-01 1.07339180e+00 4.85513985e-01 5.70743144e-01 6.56439662e-01 -7.73138106e-01 8.26250672e-01 8.21446538e-01 -4.35612023e-01 -1.07610106e+00 -9.04597417e-02 -3.87576610e-01 -1.05819464e+00 -3.44611377e-01 1.50714099e-01 1.24548450e-01 -1.13570786e+00 1.81893897e+00 1.91514671e-01 3.14713329e-01 3.43694329e-01 8.95740628e-01 1.07159352e+00 6.52936816e-01 4.93351460e-01 1.04629889e-01 1.49038303e+00 -1.42936647e+00 -7.10827887e-01 -5.41018128e-01 7.24597991e-01 -5.86550891e-01 1.22127748e+00 -1.95757523e-01 -5.88681161e-01 -7.60714829e-01 -9.12909865e-01 -2.52380311e-01 -4.20791894e-01 2.12107986e-01 7.23094344e-01 2.28878006e-01 -8.63434792e-01 2.87365496e-01 -5.91125607e-01 -4.99245524e-01 7.40358353e-01 -1.26291513e-01 -3.76603693e-01 -5.34017742e-01 -1.42925024e+00 5.60274661e-01 7.31068552e-01 -7.59310871e-02 -9.22050178e-01 -5.28906584e-01 -1.13168859e+00 1.77007914e-02 6.35874689e-01 -1.13031328e+00 9.57489669e-01 -1.00422919e+00 -9.19169307e-01 9.10666704e-01 -4.53512013e-01 -3.38027567e-01 3.65153462e-01 -7.34496266e-02 -1.40195951e-01 4.64807510e-01 3.68330210e-01 9.40069318e-01 8.20810735e-01 -1.35546231e+00 -5.49409211e-01 -5.80625355e-01 3.72360170e-01 4.79517549e-01 -6.41487479e-01 -2.79347628e-01 -8.81440759e-01 -8.39874148e-01 1.30622193e-01 -7.58444011e-01 -1.56055704e-01 5.09044901e-02 -4.86973971e-01 -3.12034339e-01 8.60494971e-01 -6.92781210e-01 1.02889812e+00 -1.91499245e+00 -6.15335740e-02 -1.59467056e-01 1.03729866e-01 2.77886808e-01 -3.06718171e-01 4.59888190e-01 1.74829498e-01 5.53943031e-02 -2.42915347e-01 -5.36035895e-01 1.31684959e-01 3.31807554e-01 -5.17882347e-01 1.10012017e-01 5.99075668e-02 1.33430088e+00 -7.70066082e-01 -8.11073363e-01 4.13158387e-01 4.79169041e-01 -3.80099982e-01 1.00421600e-01 -4.26040769e-01 2.24628851e-01 -1.04665613e+00 5.04125357e-01 4.75444078e-01 -6.10520005e-01 -1.29399985e-01 -5.32772362e-01 4.66342539e-01 -2.26789471e-02 -9.01747286e-01 2.08449554e+00 -5.68275928e-01 4.02617037e-01 -2.84241080e-01 -1.34464204e+00 7.67450750e-01 6.38739988e-02 1.64952040e-01 -8.91752899e-01 7.19919917e-04 1.28261941e-02 -4.89102513e-01 -5.93728185e-01 2.57525295e-01 -4.60811913e-01 -2.43081246e-02 3.52744967e-01 7.18339533e-02 -2.19131447e-02 -8.38297885e-03 6.12841904e-01 7.34205842e-01 3.91934097e-01 2.35621020e-01 5.18787652e-02 8.04292321e-01 1.85236380e-01 2.25197345e-01 6.58384860e-01 -1.33151248e-01 4.47273701e-01 2.01289639e-01 3.36974189e-02 -7.38132238e-01 -7.90941060e-01 1.88580915e-01 1.16195428e+00 5.28509259e-01 -5.10722697e-01 -6.53874516e-01 -7.67101467e-01 -4.35002372e-02 8.34192455e-01 -5.70243239e-01 -3.10873479e-01 -1.43803880e-01 -5.73754370e-01 6.55458868e-01 8.89528692e-01 1.01428902e+00 -1.02572680e+00 -1.39576152e-01 1.95199307e-02 -5.66170633e-01 -1.52130115e+00 -4.60307151e-01 -4.22258258e-01 -8.26284111e-01 -1.00195920e+00 -7.67196000e-01 -9.10033107e-01 6.60770833e-01 7.35163212e-01 9.57419634e-01 1.23883523e-01 -2.53649861e-01 6.63766503e-01 -4.90147859e-01 -2.54207700e-01 -2.43452907e-01 -9.66098234e-02 -4.25315201e-01 6.15172721e-02 4.00155306e-01 -3.76733512e-01 -4.34208840e-01 2.10258141e-01 -9.59367096e-01 5.47682822e-01 6.40346706e-01 8.95302415e-01 7.52584100e-01 3.15001346e-02 6.44067466e-01 -8.58906090e-01 4.30734187e-01 -4.71543789e-01 -1.79636151e-01 5.30937552e-01 -6.53868377e-01 3.42935622e-01 5.95917761e-01 -1.45926610e-01 -1.42503667e+00 -9.12134126e-02 -1.29398316e-01 -8.11475396e-01 -2.23117679e-01 7.31919587e-01 -4.03069496e-01 2.03514189e-01 1.93950921e-01 6.81854725e-01 2.23227683e-02 -4.78592485e-01 7.51644552e-01 7.73999751e-01 5.36857665e-01 -7.92399764e-01 7.95129180e-01 7.45919883e-01 2.30707061e-02 -6.65632129e-01 -1.39729548e+00 -6.17317498e-01 -4.60901916e-01 3.91096957e-02 9.53924477e-01 -1.38255191e+00 -3.37548107e-01 4.16743547e-01 -1.00944459e+00 -9.77066308e-02 1.35823250e-01 2.62082309e-01 -5.96960247e-01 7.22276270e-01 -4.10030395e-01 -5.91633201e-01 -5.10975420e-01 -9.90637898e-01 1.42572439e+00 2.79558659e-01 3.55768204e-01 -1.41903031e+00 -3.73675406e-01 8.99591088e-01 2.96625704e-01 1.46719965e-03 9.61256683e-01 -5.70493340e-01 -6.35179400e-01 9.92413759e-02 -7.63014495e-01 2.29383826e-01 1.19048029e-01 -4.88339961e-01 -1.02990925e+00 6.43165040e-05 -1.95341781e-01 -6.97844207e-01 1.34520316e+00 1.52278587e-01 1.51398110e+00 -1.66382656e-01 -5.55469871e-01 6.20691955e-01 1.57910860e+00 -3.00923020e-01 5.27001262e-01 4.78424519e-01 1.12661612e+00 8.78041446e-01 8.60390007e-01 2.95383304e-01 8.88636649e-01 7.09405601e-01 4.51597273e-01 -2.46409550e-01 -4.64300722e-01 -7.60989666e-01 3.26018691e-01 6.73213720e-01 1.12060897e-01 -1.92110747e-01 -7.78646052e-01 7.32129395e-01 -2.14452004e+00 -9.60710287e-01 -1.70566961e-01 1.55615497e+00 7.85614252e-01 -4.09241058e-02 -2.71822333e-01 -2.39910662e-01 8.47029686e-01 4.60825175e-01 -4.94440377e-01 1.38262600e-01 -2.82498896e-01 -4.12335023e-02 3.83325487e-01 1.75716594e-01 -1.04923999e+00 1.42118526e+00 4.80690002e+00 1.29819298e+00 -8.30743611e-01 1.96129307e-01 4.92303699e-01 3.07793707e-01 -5.95343709e-01 1.03956997e-01 -8.68140280e-01 2.54383504e-01 4.32117194e-01 -4.13663507e-01 1.51030645e-01 8.67913902e-01 2.48673528e-01 1.23020910e-01 -9.16857779e-01 1.13486612e+00 3.33139032e-01 -1.38203895e+00 6.55501783e-01 -1.47434592e-01 7.21854806e-01 -2.57663634e-02 -1.16366111e-01 5.11048377e-01 1.69045240e-01 -9.63255942e-01 5.74609518e-01 5.08035064e-01 8.36325347e-01 -6.03752077e-01 7.62321115e-01 6.04690075e-01 -1.53914320e+00 5.74473701e-02 -6.24242485e-01 1.33122772e-01 3.35992157e-01 5.21794081e-01 -6.27875209e-01 9.95660603e-01 4.19088125e-01 1.20106995e+00 -7.93561876e-01 6.10892296e-01 -5.08647442e-01 5.47208428e-01 -1.00558139e-01 3.76392305e-02 6.35409236e-01 -3.93544734e-02 1.75261155e-01 1.15678120e+00 1.40185997e-01 1.40912443e-01 4.60773885e-01 1.01598597e+00 -2.40356565e-01 2.25110710e-01 -6.50765479e-01 -2.85338908e-01 2.70493358e-01 1.12814260e+00 -4.69484657e-01 -6.89284027e-01 -7.18469024e-01 1.02680731e+00 3.62326622e-01 4.55816776e-01 -9.41838443e-01 -2.35623673e-01 4.94678348e-01 -1.02466084e-02 3.78369600e-01 -2.91235615e-02 -2.40977824e-01 -1.34679472e+00 3.59188169e-02 -5.78716874e-01 6.07871652e-01 -1.16576385e+00 -1.49466622e+00 3.17993581e-01 6.19033314e-02 -9.06125665e-01 -1.85647294e-01 -4.98187035e-01 -4.03893471e-01 8.47234249e-01 -1.93457460e+00 -1.81105316e+00 -6.29262507e-01 1.13223290e+00 9.57175493e-01 -1.69495150e-01 5.43634117e-01 1.57044977e-01 -3.26961428e-01 5.89211524e-01 -9.86706540e-02 3.30217928e-01 6.95842028e-01 -9.07252312e-01 6.76395297e-02 8.09428394e-01 3.18636000e-01 4.98389453e-01 2.80832112e-01 -8.03255975e-01 -1.32532489e+00 -1.52942753e+00 6.97446406e-01 -2.40734920e-01 7.79705226e-01 -1.33124426e-01 -1.02823329e+00 7.29003370e-01 1.06829025e-01 -1.61215290e-02 4.61672753e-01 -4.28525507e-02 -5.27200758e-01 9.34804454e-02 -7.26320803e-01 5.32697916e-01 1.44627440e+00 -7.78535485e-01 -1.13649583e+00 4.52247977e-01 9.92665052e-01 -1.64008394e-01 -5.23583055e-01 4.70882148e-01 2.34042346e-01 -7.73856819e-01 1.12978089e+00 -6.52618945e-01 8.54304016e-01 -2.48694018e-01 -3.78057122e-01 -1.15094852e+00 -2.27715552e-01 6.97593614e-02 6.60974160e-02 1.46663141e+00 1.70337662e-01 -5.46172678e-01 6.44280195e-01 3.50668848e-01 -3.50195169e-01 -7.27525592e-01 -5.45627236e-01 -6.94061756e-01 -1.37681291e-02 -6.52790785e-01 5.29855788e-01 1.02040040e+00 -2.62666672e-01 7.40515471e-01 -3.64206374e-01 4.67707589e-02 7.71367490e-01 5.11280000e-01 7.27343619e-01 -8.93755198e-01 -2.80566394e-01 -3.36776406e-01 -3.29646498e-01 -1.34101987e+00 6.92316175e-01 -1.15955722e+00 -1.25601918e-01 -2.02370381e+00 7.37921596e-01 -4.17518198e-01 -3.25105399e-01 7.52820373e-01 -6.54679418e-01 2.08502263e-01 2.73782432e-01 3.52881879e-01 -8.31920147e-01 9.90801692e-01 1.47675443e+00 -4.68566597e-01 2.55777687e-01 -3.54907960e-01 -1.04196739e+00 8.24135602e-01 4.51317847e-01 -2.13468418e-01 -8.30077589e-01 -5.94516575e-01 -3.42728421e-02 -1.37358652e-02 7.35326529e-01 -6.87971294e-01 1.08163364e-01 -1.68261752e-01 2.34617531e-01 -5.51200867e-01 2.79144108e-01 -7.48599708e-01 -1.75209895e-01 2.31920853e-01 -4.23492104e-01 -4.49563146e-01 1.43040448e-01 9.97488081e-01 -5.60670972e-01 -1.43133879e-01 5.64019799e-01 -2.53564149e-01 -1.34222353e+00 5.06165683e-01 2.60763496e-01 2.65829474e-01 9.05010104e-01 -2.62570709e-01 -4.35996920e-01 -4.49709237e-01 -5.88394701e-01 5.28349698e-01 5.51941633e-01 7.24942446e-01 6.53822064e-01 -1.41928685e+00 -7.01721966e-01 3.85041423e-02 6.74212694e-01 -3.23485360e-02 5.57589531e-01 7.22764432e-01 -6.48036897e-02 6.35337293e-01 4.56895530e-02 -8.39049041e-01 -1.23908055e+00 7.24882841e-01 1.84652880e-01 -3.96181077e-01 -7.97355592e-01 8.17711055e-01 9.54424381e-01 -2.51907945e-01 -3.85710411e-02 -2.01251321e-02 -3.62176090e-01 -2.52961810e-03 5.31878233e-01 -1.40689164e-01 -2.79607415e-01 -8.09992671e-01 -3.46685112e-01 8.71300817e-01 2.47830823e-02 4.86826198e-03 1.12358356e+00 -5.39347768e-01 5.34290187e-02 2.67815828e-01 1.28525567e+00 -3.28286350e-01 -1.13747668e+00 -7.32253432e-01 -2.13770092e-01 -5.41856945e-01 1.92383230e-01 -6.94961250e-01 -1.23090267e+00 1.35094631e+00 2.95443505e-01 -3.16607714e-01 1.23212910e+00 3.43128115e-01 1.10747445e+00 4.00058657e-01 4.93021548e-01 -9.34714198e-01 4.08338815e-01 4.94131356e-01 7.52340794e-01 -1.45971036e+00 -9.14094895e-02 -7.34304607e-01 -1.07573485e+00 9.09655988e-01 6.84647441e-01 1.61356777e-01 4.24642771e-01 -2.84208298e-01 -8.94954428e-02 -2.61663318e-01 -7.45708466e-01 -6.41481757e-01 5.13018787e-01 6.49143338e-01 1.83495477e-01 2.11519306e-03 -1.17355004e-01 8.07314157e-01 8.48730877e-02 1.03972636e-01 1.12413891e-01 6.12874448e-01 -5.65814674e-01 -9.51738358e-01 -1.85534939e-01 3.99598360e-01 -1.95071995e-01 -3.49169105e-01 -3.14623415e-01 7.37764001e-01 4.53129504e-03 9.38771367e-01 -2.62737244e-01 -2.60247767e-01 2.90347308e-01 1.20223545e-01 4.60723817e-01 -7.14585245e-01 -5.05297035e-02 9.83595774e-02 1.69677109e-01 -6.67328238e-01 -6.53846860e-01 -3.01458389e-01 -1.59740281e+00 -2.69400068e-02 -2.21891850e-01 -5.47132492e-02 5.60453713e-01 1.43163168e+00 4.48639929e-01 6.81105912e-01 3.18394005e-01 -5.17731249e-01 -4.90625381e-01 -9.28393543e-01 -5.33488750e-01 8.70351434e-01 -7.82190415e-04 -8.93423915e-01 -2.90632904e-01 3.43622535e-01]
[10.665877342224121, 1.4490554332733154]
26a06edc-b01e-42c5-a774-22c62bfe6975
building-korean-sign-language-augmentation
2207.05261
null
https://arxiv.org/abs/2207.05261v1
https://arxiv.org/pdf/2207.05261v1.pdf
Building Korean Sign Language Augmentation (KoSLA) Corpus with Data Augmentation Technique
We present an efficient framework of corpus for sign language translation. Aided with a simple but dramatic data augmentation technique, our method converts text into annotated forms with minimum information loss. Sign languages are composed of manual signals, non-manual signals, and iconic features. According to professional sign language interpreters, non-manual signals such as facial expressions and gestures play an important role in conveying exact meaning. By considering the linguistic features of sign language, our proposed framework is a first and unique attempt to build a multimodal sign language augmentation corpus (hereinafter referred to as the KoSLA corpus) containing both manual and non-manual modalities. The corpus we built demonstrates confident results in the hospital context, showing improved performance with augmented datasets. To overcome data scarcity, we resorted to data augmentation techniques such as synonym replacement to boost the efficiency of our translation model and available data, while maintaining grammatical and semantic structures of sign language. For the experimental support, we verify the effectiveness of data augmentation technique and usefulness of our corpus by performing a translation task between normal sentences and sign language annotations on two tokenizers. The result was convincing, proving that the BLEU scores with the KoSLA corpus were significant.
['Hyunshim Han', 'Sumi Lee', 'Ohkyoon Kwon', 'Dongmyeong Noh', 'Eunkyung Han', 'Changnam An']
2022-07-12
null
null
null
null
['sign-language-translation']
['computer-vision']
[ 5.08621097e-01 3.15367669e-01 -2.91443974e-01 -5.10353625e-01 -6.84308112e-01 -5.29323280e-01 6.93501115e-01 -5.15774608e-01 -7.87134290e-01 8.34161758e-01 7.65275717e-01 -1.19416714e-01 1.40536129e-01 -2.96285361e-01 -3.97258312e-01 -4.47287381e-01 1.21312946e-01 3.81967366e-01 -2.42887780e-01 -5.16616046e-01 3.86875379e-03 3.45307022e-01 -1.53853869e+00 4.24585730e-01 8.83860588e-01 6.60042048e-01 -1.21989422e-01 5.48813105e-01 -3.30100238e-01 7.06196249e-01 -5.75629056e-01 -5.83382905e-01 3.36178839e-01 -9.41758931e-01 -7.32803583e-01 3.41136120e-02 5.67119241e-01 -5.34652770e-01 -2.53358841e-01 7.96204805e-01 7.16734946e-01 -1.86554775e-01 5.24592340e-01 -1.21945298e+00 -6.24692500e-01 7.36018062e-01 -1.94616094e-01 -3.74902844e-01 6.09195054e-01 1.86288059e-01 1.00105798e+00 -7.61374056e-01 1.14581096e+00 1.04375315e+00 4.94487911e-01 9.42586660e-01 -8.61341178e-01 -6.79567277e-01 1.42444726e-02 1.83513492e-01 -1.12751722e+00 -3.92621398e-01 8.85181785e-01 -2.70145476e-01 6.43590987e-01 5.05098641e-01 8.35574210e-01 1.26092279e+00 -4.04057771e-01 1.02778161e+00 1.29476321e+00 -9.04271424e-01 -6.61944821e-02 2.57911384e-01 9.96561348e-02 7.17636168e-01 2.28524394e-02 -2.87491940e-02 -7.31395364e-01 -3.43678817e-02 6.04172885e-01 -4.36315089e-01 -3.96839947e-01 -1.81825101e-01 -1.51971900e+00 3.73479784e-01 4.91753779e-02 8.06525767e-01 -3.57791126e-01 8.28154907e-02 5.91227591e-01 4.73836064e-01 -8.87977183e-02 4.60105926e-01 -3.41970265e-01 -4.89807546e-01 -6.90578759e-01 -1.24643415e-01 6.69356525e-01 1.19571996e+00 9.47746933e-02 -9.34897512e-02 -2.65180409e-01 8.41208398e-01 3.39261055e-01 9.42256689e-01 9.79609787e-01 -6.56615853e-01 6.58379495e-01 8.47362459e-01 -1.70874208e-01 -5.59256315e-01 -2.97089636e-01 2.87200771e-02 -6.43797159e-01 1.41476467e-01 5.27594090e-01 -1.25365481e-01 -1.16617000e+00 1.96217108e+00 -4.54016924e-02 -4.20253396e-01 2.25915015e-01 1.02987778e+00 9.32525218e-01 1.75854601e-02 3.60891283e-01 -1.89218715e-01 1.53476751e+00 -5.71910262e-01 -1.19191027e+00 1.14776291e-01 8.75236154e-01 -9.15454209e-01 1.50808370e+00 2.39794984e-01 -8.11096966e-01 -3.03232700e-01 -8.56958151e-01 -1.63067207e-01 -2.76890635e-01 5.60204566e-01 7.65770078e-01 8.36619079e-01 -8.71879995e-01 8.80136341e-02 -6.23901963e-01 -7.42650092e-01 3.82222176e-01 4.22370732e-01 -8.84145498e-01 1.19758956e-01 -1.07458138e+00 9.84873772e-01 3.55574548e-01 1.51593879e-01 -8.74172598e-02 -1.10195845e-01 -7.43617475e-01 -3.62296611e-01 7.92456716e-02 -5.20930827e-01 1.02127123e+00 -1.21573842e+00 -1.63687825e+00 1.12248683e+00 -2.36414239e-01 -1.75414622e-01 7.94831455e-01 -1.66287705e-01 -5.59659421e-01 2.12590083e-01 -2.53358305e-01 6.63432717e-01 7.34081924e-01 -1.17301059e+00 -5.71512043e-01 -3.11032802e-01 -1.09603949e-01 2.27958933e-01 -4.22603905e-01 2.11659327e-01 -3.29717755e-01 -7.70384848e-01 4.26351368e-01 -8.82263720e-01 8.49606991e-02 8.47795382e-02 -2.46162832e-01 1.00700006e-01 4.96454090e-01 -1.06947172e+00 1.21970057e+00 -2.23255205e+00 1.36453256e-01 6.23989880e-01 7.38402829e-02 2.79614419e-01 -3.59945118e-01 3.94547075e-01 -2.77452674e-02 6.29649162e-02 -5.95832586e-01 -3.29162091e-01 1.23849750e-01 5.41363001e-01 -1.73365444e-01 1.96781829e-01 1.87294483e-01 1.20007634e+00 -8.09708655e-01 -8.75470698e-01 3.00950885e-01 4.16045636e-01 -5.15189528e-01 -1.67668425e-02 7.18452260e-02 6.69418097e-01 -3.82849127e-01 1.01015890e+00 1.89165041e-01 3.07531625e-01 2.09424838e-01 -3.64219189e-01 6.48954883e-02 1.64966099e-02 -9.43736255e-01 1.76314366e+00 -5.16723394e-01 6.19437754e-01 -1.49180278e-01 -6.01262987e-01 9.32570517e-01 4.99732971e-01 6.16293311e-01 -9.02058661e-01 5.52093089e-01 5.56694448e-01 1.61224842e-01 -1.08839822e+00 5.75508356e-01 -2.86532521e-01 6.87146187e-03 4.42144424e-01 1.35318309e-01 -2.40989536e-01 4.44281191e-01 1.04573645e-01 8.08525085e-01 3.92174631e-01 3.55641067e-01 3.13298821e-01 5.59980929e-01 1.56968728e-01 6.07218519e-02 4.28639740e-01 -2.73639828e-01 4.82931703e-01 2.09299967e-01 -2.90434789e-02 -1.12619090e+00 -7.81849325e-01 1.67730376e-02 8.77384126e-01 -2.79921979e-01 -2.58208185e-01 -6.02246165e-01 -6.12948179e-01 -3.25861663e-01 5.20401359e-01 -4.58629221e-01 1.39846042e-01 -8.20349038e-01 -6.67500079e-01 1.14024198e+00 5.66417456e-01 5.87262750e-01 -1.14841485e+00 -7.32773125e-01 1.17736198e-01 -3.88926417e-01 -1.31558144e+00 -4.15763646e-01 -2.26669937e-01 -7.70409226e-01 -1.29385376e+00 -8.63875806e-01 -1.04348004e+00 9.50921237e-01 -4.45725858e-01 5.85477471e-01 -2.21715756e-02 -2.15036646e-01 5.98352551e-01 -8.00365388e-01 -2.49493450e-01 -8.93357813e-01 -2.27472618e-01 -3.98017429e-02 5.96144982e-02 4.89321530e-01 -5.69811225e-01 -1.50482029e-01 1.86121643e-01 -1.00405383e+00 1.40622824e-01 1.01262271e+00 1.06634831e+00 2.14895532e-01 -9.24635291e-01 2.85718590e-01 -5.53294957e-01 8.16630602e-01 2.90199280e-01 -1.92746624e-01 3.34747374e-01 -3.81485879e-01 3.11573863e-01 2.99252808e-01 -6.02238715e-01 -1.01228058e+00 4.29422468e-01 -2.50651032e-01 1.50667161e-01 -1.70920908e-01 4.22569782e-01 -2.30731100e-01 -1.93243638e-01 6.90580368e-01 5.54532409e-01 3.13778847e-01 -4.36114997e-01 7.92831361e-01 1.16673994e+00 9.50890839e-01 -4.88836318e-01 6.47670090e-01 4.80539322e-01 1.36736304e-01 -9.01133120e-01 -5.18682711e-02 -3.55900556e-01 -1.00910521e+00 -4.10692602e-01 6.12268209e-01 -5.23018420e-01 -7.42937028e-01 4.50368822e-01 -1.11260402e+00 7.59896860e-02 -5.73897600e-01 9.56368029e-01 -7.15566993e-01 6.61685646e-01 -3.25826913e-01 -8.81685138e-01 -2.87723750e-01 -8.94319415e-01 9.89446282e-01 -3.78580898e-01 -6.07491553e-01 -5.04554391e-01 4.10456322e-02 4.80654329e-01 4.08148974e-01 4.13895041e-01 9.40278888e-01 -5.97422183e-01 -1.62932068e-01 -6.41396582e-01 -2.22121879e-01 5.87206185e-01 3.95243973e-01 -1.55801326e-01 -1.00014091e+00 1.27013788e-01 -2.47914299e-01 -4.50239718e-01 6.14000857e-01 -1.59945831e-01 3.77159297e-01 -2.88990140e-01 6.32764921e-02 2.45088875e-01 1.02486777e+00 2.25674823e-01 6.97305143e-01 1.24738909e-01 5.49331665e-01 5.69344819e-01 5.38217723e-01 2.67166018e-01 2.69130528e-01 7.24896550e-01 -2.72815768e-02 -2.34725788e-01 -5.38164973e-01 -4.00199145e-01 2.82673657e-01 1.15356445e+00 -5.23340106e-01 4.94758040e-02 -7.66206980e-01 5.53793311e-01 -1.62038600e+00 -8.71577978e-01 -1.86425447e-01 1.98714995e+00 9.96194720e-01 -2.58779675e-01 1.37261644e-01 4.69363481e-01 3.31385344e-01 -2.82823056e-01 -1.65108442e-01 -4.20125335e-01 -5.46666443e-01 1.74634725e-01 4.27870870e-01 5.87809980e-01 -9.42041039e-01 9.00844693e-01 6.35968208e+00 4.57944274e-01 -1.22677827e+00 4.57423441e-02 -1.48760140e-01 -4.93058935e-02 -2.64379352e-01 -4.09220427e-01 -3.20797086e-01 3.88143152e-01 5.48456192e-01 7.95103982e-03 2.90698558e-01 5.79938054e-01 3.95458132e-01 -5.43918228e-03 -1.22459900e+00 1.12029719e+00 4.28452283e-01 -9.00407851e-01 3.28613013e-01 -9.85383764e-02 4.06141669e-01 -1.05558030e-01 -3.05868477e-01 2.51320660e-01 -1.40691593e-01 -7.56572664e-01 6.99586153e-01 5.40869296e-01 1.22865474e+00 -6.45807460e-02 1.05988812e+00 3.02012097e-02 -9.53773081e-01 6.96659461e-02 4.31435168e-01 -7.64864981e-02 4.19971794e-01 -1.89450160e-01 -9.21080410e-01 5.02412081e-01 7.26571083e-02 3.51296693e-01 -4.26028371e-01 8.88562143e-01 -4.95190531e-01 4.60173607e-01 -6.20538950e-01 -2.77557164e-01 2.44486029e-03 -3.19460891e-02 5.64347267e-01 1.26599276e+00 2.21515566e-01 1.78868309e-01 -2.07888544e-01 5.31687915e-01 -6.64673895e-02 7.52463102e-01 -7.51444042e-01 -3.38086069e-01 2.46075317e-01 7.62992740e-01 -5.74232817e-01 -4.48168248e-01 -4.29125458e-01 1.12597001e+00 -2.61819601e-01 3.06195647e-01 -5.69535673e-01 -2.99628705e-01 2.80515134e-01 -1.49455637e-01 -3.04122478e-01 -2.36333430e-01 -4.93087888e-01 -1.19340944e+00 5.05405664e-01 -9.42026734e-01 2.31646538e-01 -6.15165472e-01 -9.87938046e-01 7.31228292e-01 5.88154234e-02 -1.71498215e+00 -5.03406286e-01 -7.41327345e-01 7.44331852e-02 6.78739309e-01 -1.23224008e+00 -1.72300673e+00 -4.34474915e-01 7.11269677e-01 2.23926082e-01 -3.34085852e-01 1.08624256e+00 6.38040721e-01 8.23356658e-02 9.45481718e-01 -2.57012188e-01 3.55881900e-01 7.30272293e-01 -8.14357102e-01 -3.72425430e-02 8.33832443e-01 2.00667664e-01 5.90005875e-01 6.17575765e-01 -6.14237130e-01 -1.28040469e+00 -4.54911441e-01 1.39587808e+00 -3.89259398e-01 6.58251107e-01 -1.60696089e-01 -4.20272440e-01 6.00716233e-01 -1.00515321e-01 -4.64847952e-01 7.11676002e-01 -2.58663386e-01 -2.00519875e-01 2.67205775e-01 -1.30894196e+00 8.03275228e-01 1.35944748e+00 -5.92628539e-01 -1.01279402e+00 2.47534111e-01 3.59605372e-01 -4.19369042e-01 -6.30147815e-01 3.73801678e-01 1.04542458e+00 -3.68174702e-01 4.61131692e-01 -8.53972852e-01 4.95587975e-01 -1.96175382e-01 -3.95558655e-01 -8.57339621e-01 4.13120061e-01 -4.95236009e-01 2.43331969e-01 1.15770257e+00 5.37474036e-01 -6.04957819e-01 6.75392509e-01 9.23340559e-01 -1.34014368e-01 -3.50504249e-01 -1.12256157e+00 -7.77496040e-01 -4.70747113e-01 -7.37704456e-01 4.60406631e-01 1.03754437e+00 5.43503046e-01 1.49981817e-02 -7.31757104e-01 -3.29691052e-01 2.77735323e-01 -8.47405270e-02 9.78147805e-01 -9.11997676e-01 -1.51318833e-01 -5.47644794e-01 -7.57858455e-01 -7.00464129e-01 3.92023064e-02 -1.10259974e+00 -8.04200619e-02 -1.41131449e+00 6.67515583e-03 -8.55539273e-03 -5.13034835e-02 1.08085859e+00 1.81397408e-01 6.52829051e-01 4.85822886e-01 2.35952154e-01 -6.23626634e-02 5.68808198e-01 1.43136144e+00 -1.34543285e-01 -4.01273042e-01 -2.65294671e-01 -3.85269940e-01 7.75360227e-01 5.82279563e-01 -1.14591733e-01 -1.28855228e-01 -2.84066767e-01 -5.45654036e-02 -3.01909894e-01 2.53745764e-01 -6.34648681e-01 -3.21047716e-02 6.13804348e-02 -9.02947560e-02 -3.61711591e-01 4.07267451e-01 -1.12444866e+00 -1.16923735e-01 6.22933328e-01 -5.01040161e-01 -2.52300471e-01 1.34146422e-01 -7.18979016e-02 -5.20039916e-01 -1.42451571e-02 4.98733908e-01 1.07576689e-02 -9.37169909e-01 -2.02347577e-01 -3.97307068e-01 -8.52241963e-02 7.67832875e-01 -5.00510871e-01 -9.55858380e-02 -4.91357177e-01 -1.01578772e+00 2.67967377e-02 2.70023793e-01 5.62481999e-01 6.40675843e-01 -1.41827524e+00 -7.02805221e-01 5.94134450e-01 5.23507357e-01 -6.48079753e-01 -6.88615590e-02 1.04495084e+00 -5.70862889e-01 4.34323490e-01 -5.71644962e-01 -2.90286392e-01 -1.71917069e+00 -3.75419520e-02 1.05935253e-01 9.81078297e-02 -7.55656600e-01 5.77181518e-01 -4.33816224e-01 -4.62229580e-01 4.02269840e-01 -8.83377910e-01 -2.60548562e-01 1.68426484e-01 3.58987629e-01 1.88690245e-01 -1.13021426e-01 -9.41751897e-01 -3.93166602e-01 7.26443112e-01 3.93079072e-01 -5.51725507e-01 1.02717960e+00 -7.19653890e-02 -1.44973338e-01 3.96612793e-01 8.59518170e-01 2.93021113e-01 -4.04898018e-01 -2.43777141e-01 6.16352335e-02 -4.36406255e-01 -3.76463026e-01 -1.30119407e+00 -6.43250287e-01 5.86112857e-01 9.14531827e-01 -4.93454993e-01 1.34452260e+00 2.86694262e-02 6.51453376e-01 6.35487854e-01 6.14836395e-01 -1.11919785e+00 -3.67535025e-01 3.75668198e-01 1.20524144e+00 -1.34956515e+00 -3.12589973e-01 -3.22383910e-01 -8.44925165e-01 9.60680068e-01 1.86125115e-01 3.35991353e-01 2.10379422e-01 2.47098282e-01 6.91488862e-01 -1.16916616e-02 -1.53240442e-01 -5.34921110e-01 4.68473047e-01 6.63934946e-01 5.74990630e-01 2.02753693e-01 -1.14263964e+00 6.42636359e-01 -5.96426427e-01 5.29240251e-01 4.59610492e-01 9.72383618e-01 -8.09518471e-02 -1.39257026e+00 -4.31435496e-01 2.78358519e-01 -1.38347432e-01 -7.98306242e-02 -7.85816729e-01 1.31748450e+00 2.88169235e-01 7.01013803e-01 -3.89986545e-01 -3.93731683e-01 7.81748235e-01 4.97352391e-01 7.91170239e-01 -2.25661337e-01 -4.30354744e-01 -6.01694435e-02 4.02168661e-01 -5.24249911e-01 -8.32747281e-01 -6.55300856e-01 -1.45214152e+00 1.36646509e-01 4.71143797e-02 -1.21345423e-01 1.03097045e+00 1.23100340e+00 -1.54834852e-01 4.41845387e-01 8.74283165e-02 -5.40984571e-01 -5.21627367e-01 -1.31425631e+00 -4.36170042e-01 9.39382195e-01 1.72636986e-01 -3.59801799e-01 -1.80647567e-01 4.23501849e-01]
[9.141382217407227, -6.446753978729248]
560ecd73-66f3-48f2-b825-972f1d135543
multimodal-machine-translation-with
1805.02356
null
http://arxiv.org/abs/1805.02356v1
http://arxiv.org/pdf/1805.02356v1.pdf
Multimodal Machine Translation with Reinforcement Learning
Multimodal machine translation is one of the applications that integrates computer vision and language processing. It is a unique task given that in the field of machine translation, many state-of-the-arts algorithms still only employ textual information. In this work, we explore the effectiveness of reinforcement learning in multimodal machine translation. We present a novel algorithm based on the Advantage Actor-Critic (A2C) algorithm that specifically cater to the multimodal machine translation task of the EMNLP 2018 Third Conference on Machine Translation (WMT18). We experiment our proposed algorithm on the Multi30K multilingual English-German image description dataset and the Flickr30K image entity dataset. Our model takes two channels of inputs, image and text, uses translation evaluation metrics as training rewards, and achieves better results than supervised learning MLE baseline models. Furthermore, we discuss the prospects and limitations of using reinforcement learning for machine translation. Our experiment results suggest a promising reinforcement learning solution to the general task of multimodal sequence to sequence learning.
['Xin Qian', 'Jieli Zhou', 'Ziyi Zhong']
2018-05-07
null
null
null
null
['multimodal-machine-translation']
['natural-language-processing']
[ 4.37858254e-01 -8.01472664e-02 -6.09707952e-01 -3.04092854e-01 -1.63677645e+00 -6.59992099e-01 9.41074252e-01 -2.20185339e-01 -6.46666229e-01 9.81715620e-01 2.45009929e-01 -6.67971671e-01 4.37669665e-01 -2.12558225e-01 -1.06994331e+00 -5.16451836e-01 6.46924794e-01 8.43705058e-01 -5.59305131e-01 -3.18915337e-01 6.97029531e-02 6.40858337e-02 -9.05132294e-01 6.05529189e-01 8.35314155e-01 8.02875340e-01 8.59417841e-02 7.88154781e-01 -1.45406321e-01 5.62608600e-01 -2.30650619e-01 -8.54491413e-01 1.77513316e-01 -7.77466714e-01 -1.05837941e+00 1.43025011e-01 5.83795488e-01 -2.75689840e-01 -1.90465763e-01 1.01217115e+00 7.69583702e-01 -1.47845700e-01 8.11766207e-01 -1.46340311e+00 -1.02240789e+00 6.15231454e-01 -6.09041989e-01 -1.46860853e-01 6.67532921e-01 1.49982795e-01 9.30759251e-01 -1.26462293e+00 8.23180974e-01 1.32116401e+00 1.61846712e-01 7.82790005e-01 -1.11256945e+00 -4.53583211e-01 -1.36797354e-01 4.49900895e-01 -9.27401304e-01 -5.35961866e-01 5.47422945e-01 -4.08406556e-01 9.81579959e-01 8.67265984e-02 2.67362922e-01 1.59239721e+00 3.15350890e-01 1.25771308e+00 1.53535557e+00 -8.70159566e-01 -7.38160536e-02 3.12573969e-01 -3.89000058e-01 6.36550367e-01 -4.87579048e-01 1.12600930e-01 -6.65405333e-01 -7.71285966e-02 5.47731459e-01 -3.99031848e-01 -2.01135911e-02 -3.96532208e-01 -1.93193543e+00 9.65570152e-01 5.58209457e-02 1.15445875e-01 -2.12735176e-01 3.68938804e-01 5.30747473e-01 6.12348080e-01 3.65495265e-01 3.15775573e-01 -4.36793298e-01 -3.44180703e-01 -7.06011117e-01 5.16539700e-02 6.01799369e-01 1.18523669e+00 8.47719550e-01 -1.33882388e-01 -2.90646613e-01 9.37115848e-01 4.81254905e-01 1.16214752e+00 5.56961536e-01 -9.28323507e-01 9.04406548e-01 4.33775723e-01 2.12373927e-01 -4.39319015e-01 -2.56817061e-02 7.39832520e-02 -7.27190316e-01 -1.52516916e-01 2.42733136e-01 -3.70430738e-01 -7.29927480e-01 1.46874583e+00 1.47451401e-01 -2.90279508e-01 5.00480294e-01 9.89341319e-01 8.09510410e-01 8.37809920e-01 9.99846309e-02 -4.14260179e-01 1.08297944e+00 -1.44118643e+00 -9.44326997e-01 -1.17526069e-01 6.51793838e-01 -1.29471564e+00 1.09876943e+00 1.01200556e-02 -1.13836980e+00 -4.18754995e-01 -6.97124660e-01 -7.58886710e-03 -4.50166702e-01 5.23729801e-01 5.35470605e-01 2.86965907e-01 -1.09982538e+00 2.34022394e-01 -6.01709723e-01 -6.64203227e-01 5.04708057e-03 5.09810865e-01 -4.45295215e-01 -1.16282627e-01 -1.28609657e+00 1.41651118e+00 2.27834925e-01 1.60457298e-01 -9.61864293e-01 1.44488448e-02 -8.07352960e-01 -3.40585947e-01 2.16768488e-01 -7.63843834e-01 1.59186578e+00 -1.65382302e+00 -1.89707685e+00 1.04042006e+00 -4.11370873e-01 -3.94035488e-01 6.27169192e-01 -1.61038265e-01 -4.02791470e-01 2.46791750e-01 1.41026914e-01 1.06720221e+00 1.03958786e+00 -1.20599222e+00 -5.98303378e-01 -2.39813626e-01 -2.06738472e-01 4.68383640e-01 -1.28416866e-01 2.67403215e-01 -2.89891988e-01 -2.86536068e-01 -4.40081149e-01 -1.20469713e+00 -1.17225274e-01 -3.88000220e-01 -6.09430410e-02 -3.18067312e-01 6.21288180e-01 -8.05472672e-01 8.66355002e-01 -1.98560321e+00 7.17708647e-01 -1.97267041e-01 -4.12098020e-01 7.85109401e-02 -5.82029760e-01 8.07021737e-01 9.73133221e-02 -4.85648178e-02 -3.15521628e-01 -4.34280306e-01 1.74573436e-01 2.53505379e-01 -5.19257843e-01 3.33683342e-01 4.52147156e-01 1.57641041e+00 -9.20278013e-01 -6.80650175e-01 1.08543478e-01 2.17519656e-01 3.98351960e-02 3.66469592e-01 -3.80089700e-01 5.85009754e-01 -3.84197593e-01 9.03472126e-01 4.24243897e-01 -1.44935668e-01 1.15556389e-01 -2.12821260e-01 -1.48588851e-01 -1.90791070e-01 -4.79321569e-01 2.30033660e+00 -4.53326821e-01 7.63330042e-01 -1.90486252e-01 -9.79559064e-01 8.03273916e-01 6.13278568e-01 4.20940846e-01 -9.78563607e-01 1.28536373e-01 5.50808072e-01 -2.68477678e-01 -8.82488251e-01 7.40647256e-01 -1.08487077e-01 -2.28590891e-01 5.95839202e-01 3.02646250e-01 -1.41658857e-01 1.22081943e-01 1.00106467e-02 6.13321245e-01 7.87826777e-01 1.95715770e-01 2.47652099e-01 6.27155662e-01 3.37750286e-01 4.00975607e-02 5.40325701e-01 -4.53767896e-01 4.12695229e-01 2.72932500e-01 -2.81801790e-01 -1.34698701e+00 -8.61728609e-01 2.91085303e-01 1.32943773e+00 7.71887675e-02 6.66822493e-02 -8.28425586e-01 -8.42918336e-01 -2.35480994e-01 5.63955426e-01 -5.01466751e-01 -1.87496543e-02 -6.33396387e-01 -6.98012471e-01 7.07888901e-01 2.56965816e-01 4.94344920e-01 -1.18832350e+00 -1.94372058e-01 1.15364902e-01 -7.50192702e-01 -1.46511436e+00 -6.51494980e-01 -1.05258532e-01 -7.64931083e-01 -5.97965062e-01 -1.01705003e+00 -1.12753463e+00 4.55693275e-01 1.46982118e-01 1.04066646e+00 -4.47484463e-01 1.30374864e-01 8.58570516e-01 -6.14839554e-01 -2.91772276e-01 -1.05076075e+00 2.74579912e-01 -7.29175657e-02 1.81603253e-01 4.67496037e-01 1.30395805e-02 -2.43545726e-01 3.30875874e-01 -7.03810096e-01 2.87746966e-01 1.00023532e+00 1.11918640e+00 5.65261364e-01 -9.60607529e-01 7.55510330e-01 -3.65042061e-01 8.36440265e-01 -1.38807207e-01 -4.68309492e-01 8.64532173e-01 -5.91647923e-01 1.92555323e-01 3.10493201e-01 -6.24945819e-01 -9.81339157e-01 4.96025652e-01 2.16571316e-01 -5.01061678e-01 -7.80614763e-02 6.23987198e-01 7.12437555e-02 -9.30274278e-02 4.15962219e-01 6.45180702e-01 9.87360701e-02 -2.27170531e-02 6.68715477e-01 9.78351295e-01 3.68911266e-01 -5.88082314e-01 5.86386621e-01 1.39778957e-01 -1.15526743e-01 -5.27881742e-01 -3.45417202e-01 -4.03046668e-01 -7.94705689e-01 -4.45413530e-01 1.17517936e+00 -1.05327952e+00 -6.68705702e-01 1.40209407e-01 -1.50786006e+00 -2.40441978e-01 1.47340596e-01 7.41150677e-01 -1.15001786e+00 3.56388420e-01 -5.65148950e-01 -7.13971972e-01 -4.93653923e-01 -1.49879527e+00 1.48476040e+00 3.64055075e-02 7.09956065e-02 -9.92033243e-01 4.23403502e-01 7.71032751e-01 2.48714820e-01 9.20237303e-02 8.75596821e-01 -5.06865859e-01 -6.23837411e-01 3.59145440e-02 -1.32126212e-01 3.73471439e-01 -1.09099589e-01 -1.42403528e-01 -7.86055982e-01 -3.51923436e-01 -3.91394228e-01 -9.44334507e-01 8.07746947e-01 1.26242071e-01 4.05314922e-01 -1.12764515e-01 -2.37244871e-02 6.42569810e-02 1.42592394e+00 1.21930942e-01 6.46959126e-01 4.59521174e-01 5.90349793e-01 6.23757243e-01 9.63337123e-01 -1.66305006e-02 5.73191762e-01 7.38411009e-01 5.38465202e-01 -3.39710444e-01 2.75571756e-02 -3.37814718e-01 8.71097982e-01 1.17180812e+00 -4.97287139e-02 -2.33405992e-01 -9.32989717e-01 5.38655221e-01 -2.35472393e+00 -8.97363842e-01 6.98296204e-02 1.93724263e+00 8.92568409e-01 -5.52830160e-01 3.74002568e-02 -4.91243035e-01 7.56587565e-01 -1.39415473e-01 -3.70196790e-01 -8.90365243e-01 -3.21495920e-01 -6.74366504e-02 4.80254114e-01 5.40717959e-01 -1.12215459e+00 1.39736176e+00 5.76116562e+00 8.07587147e-01 -1.06648898e+00 3.28502983e-01 5.23522794e-01 3.12261224e-01 -1.86710447e-01 -2.39080377e-02 -6.47830665e-01 4.67016958e-02 1.15484464e+00 1.15515046e-01 8.06805968e-01 4.27707613e-01 2.72382438e-01 -1.21871181e-01 -1.42053068e+00 1.21567941e+00 4.64209080e-01 -1.15214229e+00 3.99942189e-01 1.75280049e-02 9.88028109e-01 2.03116894e-01 3.23494971e-01 4.56276804e-01 1.24363236e-01 -1.18523717e+00 5.88160455e-01 4.76573914e-01 9.55027938e-01 -6.86806738e-01 7.94904292e-01 2.99406648e-01 -5.79147696e-01 3.56257260e-02 -4.19486374e-01 3.26771557e-01 2.15412214e-01 -2.29804978e-01 -1.15888476e+00 7.02732861e-01 2.75737733e-01 7.63971090e-01 -4.08472300e-01 6.51215494e-01 -2.18315393e-01 4.77131218e-01 3.54152888e-01 -3.20990086e-01 5.39640129e-01 -4.89261955e-01 3.16415995e-01 1.41721928e+00 3.74563217e-01 -3.85007054e-01 2.26702347e-01 5.06176174e-01 -4.36780661e-01 5.70407212e-01 -7.55637705e-01 -5.59017301e-01 -1.79223698e-02 1.26299834e+00 -3.11545551e-01 -4.11894172e-01 -6.72432005e-01 1.35251355e+00 3.30891967e-01 6.68465614e-01 -9.12313521e-01 8.99886042e-02 1.35872811e-01 -4.88636494e-01 1.90884948e-01 -2.48055786e-01 -1.31929666e-01 -1.49008179e+00 -4.76967134e-02 -1.50456655e+00 3.40770371e-02 -8.91515374e-01 -1.23049450e+00 5.68506598e-01 -2.11743847e-01 -1.58038521e+00 -5.14637411e-01 -6.47057891e-01 4.89660837e-02 7.63764024e-01 -1.63421094e+00 -1.85845017e+00 2.11920038e-01 6.88993812e-01 8.69899154e-01 -5.90781629e-01 1.13509750e+00 3.12186092e-01 -1.75320968e-01 6.89225078e-01 5.26483238e-01 2.90839523e-01 1.32489049e+00 -1.01143229e+00 1.62411600e-01 4.05783445e-01 3.93296570e-01 2.59542286e-01 5.78865707e-01 -4.93080676e-01 -2.06032825e+00 -9.47361231e-01 1.27228463e+00 -5.71165442e-01 8.44593585e-01 -2.97669113e-01 -3.09708089e-01 8.81373227e-01 1.14050281e+00 -4.51462179e-01 6.54177725e-01 -4.33482379e-01 -2.48156413e-01 2.94880215e-02 -8.34520936e-01 6.74718916e-01 4.04697299e-01 -6.84706986e-01 -4.75354314e-01 6.17639005e-01 7.95010149e-01 -5.66375792e-01 -8.97094548e-01 2.91297793e-01 5.48341572e-01 -2.23022416e-01 7.26025820e-01 -9.75204945e-01 1.03011501e+00 -2.34145060e-01 -5.33094108e-01 -1.52472460e+00 1.06512003e-01 -7.49023914e-01 5.75203076e-02 9.72220302e-01 9.35015321e-01 -1.65680945e-01 3.34935069e-01 1.91688463e-01 4.30040732e-02 -6.31858468e-01 -1.13431525e+00 -5.19315064e-01 5.67953467e-01 -1.69668928e-01 2.91940480e-01 9.64173496e-01 1.46796733e-01 7.92578459e-01 -9.50053096e-01 -1.00675210e-01 5.74739218e-01 2.60406584e-01 9.08614039e-01 -4.84194517e-01 -3.35962415e-01 -3.55577737e-01 -9.00796726e-02 -8.80004585e-01 5.58356822e-01 -1.15837014e+00 3.64475213e-02 -1.58935070e+00 5.94860792e-01 4.11346346e-01 -6.32948475e-03 4.53776002e-01 1.85712259e-02 3.42517197e-01 2.89110869e-01 2.28431582e-01 -8.82542551e-01 6.85346484e-01 1.60500634e+00 -6.79470599e-01 4.47163470e-02 -2.18866512e-01 -2.44787380e-01 1.10197470e-01 8.12181532e-01 -2.99032211e-01 -4.49851640e-02 -8.64767313e-01 3.47171158e-01 4.52921510e-01 2.54820973e-01 -1.79176256e-01 1.98467806e-01 -4.18665767e-01 4.24137086e-01 -5.67483366e-01 2.98645526e-01 -9.34982836e-01 -2.46040776e-01 3.59927803e-01 -7.89183915e-01 5.55261910e-01 1.65436134e-01 4.59191084e-01 -4.82205838e-01 -1.46599159e-01 5.96931994e-01 -2.07350761e-01 -5.33879936e-01 1.22987904e-01 -5.98301828e-01 -8.65173414e-02 9.73903775e-01 2.50426143e-01 -3.91188532e-01 -6.02705657e-01 -6.84851587e-01 3.88209671e-01 3.17195654e-01 7.26083755e-01 8.03445816e-01 -1.72327256e+00 -1.04969966e+00 -2.39509538e-01 3.81090105e-01 -7.84407377e-01 -1.91016361e-01 1.03915274e+00 -3.13401461e-01 6.01861596e-01 -3.42250913e-01 -8.16237628e-01 -1.36029530e+00 7.22190380e-01 1.92784414e-01 -3.28559875e-01 3.25605832e-02 2.82475203e-01 -2.04048306e-01 -8.47827196e-01 1.44081935e-01 1.81256190e-01 -1.56699955e-01 -8.10266882e-02 2.73070335e-01 2.73151964e-01 -3.85450646e-02 -9.13950682e-01 -1.92910820e-01 6.81822658e-01 -1.97480991e-01 -7.92814672e-01 1.07183158e+00 -3.53092372e-01 -3.42808813e-01 5.37143946e-01 1.27737176e+00 -4.00969118e-01 -7.27172613e-01 -4.52501267e-01 1.71354771e-01 -6.50909543e-02 -4.50810939e-01 -1.14936280e+00 -5.14785290e-01 1.20313823e+00 9.60065782e-01 -4.18076843e-01 9.22234595e-01 2.66161952e-02 7.81748533e-01 7.75704503e-01 3.56795579e-01 -1.49083209e+00 2.66608626e-01 7.31413901e-01 1.07686102e+00 -1.90401840e+00 -2.41676047e-01 9.69960317e-02 -1.10558474e+00 1.39243174e+00 4.94136930e-01 2.05770031e-01 -1.19823352e-01 -2.20048577e-01 4.81547654e-01 3.04015458e-01 -8.01629663e-01 -1.12373143e-01 3.57076228e-01 4.72357988e-01 5.98642647e-01 2.31748343e-01 -6.55451119e-01 6.36483729e-02 8.03392231e-02 8.22520703e-02 4.82123762e-01 8.38520229e-01 -2.64060497e-01 -1.54210639e+00 -4.79417235e-01 -2.19791055e-01 -3.84908110e-01 -2.10288733e-01 -9.47488487e-01 5.82543194e-01 -1.93948686e-01 1.03220332e+00 -4.96582180e-01 -3.47155124e-01 1.49367288e-01 3.22897822e-01 7.92782962e-01 -1.08371772e-01 -6.91403747e-01 3.37417841e-01 4.83319946e-02 -3.74636531e-01 -9.64488685e-01 -6.74812376e-01 -1.01582229e+00 -6.60161674e-02 -1.24449112e-01 2.20911980e-01 1.27798557e+00 9.86531198e-01 2.90205419e-01 8.91093537e-02 7.42985845e-01 -6.90556765e-01 -8.12957287e-01 -1.08646452e+00 5.45400791e-02 3.39221001e-01 3.84848982e-01 -1.19081818e-01 1.05043367e-01 4.43338245e-01]
[11.488062858581543, 1.5275534391403198]
8bfbdad0-c8e1-4ad3-9f1c-0204df28fece
a-convolutional-approach-to-reflection
1609.05257
null
http://arxiv.org/abs/1609.05257v1
http://arxiv.org/pdf/1609.05257v1.pdf
A convolutional approach to reflection symmetry
We present a convolutional approach to reflection symmetry detection in 2D. Our model, built on the products of complex-valued wavelet convolutions, simplifies previous edge-based pairwise methods. Being parameter-centered, as opposed to feature-centered, it has certain computational advantages when the object sizes are known a priori, as demonstrated in an ellipse detection application. The method outperforms the best-performing algorithm on the CVPR 2013 Symmetry Detection Competition Database in the single-symmetry case. Code and a new database for 2D symmetry detection is available.
['Davi Geiger', 'Vighnesh Birodkar', 'Marcelo Cicconet', 'Michael Werman', 'Mads Lund']
2016-09-17
null
null
null
null
['symmetry-detection']
['computer-vision']
[-1.97706204e-02 2.11787131e-02 1.05332062e-01 -2.53260523e-01 -6.43331170e-01 -6.26998007e-01 6.01587832e-01 -1.43593289e-02 -3.41472507e-01 2.18846947e-01 4.47608471e-01 -2.83816367e-01 -5.18195271e-01 -6.52963459e-01 -4.29168820e-01 -3.38819742e-01 -7.29436457e-01 4.50435966e-01 1.95890307e-01 -1.51826233e-01 6.79556251e-01 1.08046734e+00 -1.30142069e+00 1.49399668e-01 9.44122151e-02 1.11532819e+00 -5.19404292e-01 5.08662283e-01 5.81564486e-01 1.50187137e-02 -3.57229337e-02 -6.66999340e-01 6.29276454e-01 7.79724121e-02 -5.77011228e-01 -2.97294408e-02 9.68532264e-01 -4.48111780e-02 -3.98278534e-01 9.05067086e-01 7.70110667e-01 2.44326200e-02 9.75821257e-01 -1.11773288e+00 -5.54436207e-01 2.90086806e-01 -9.43663418e-01 3.20252448e-01 4.27709281e-01 -5.42225778e-01 1.25736260e+00 -1.49164212e+00 1.05427623e+00 9.64872897e-01 1.52235281e+00 1.25791728e-01 -1.12031984e+00 -5.49131274e-01 -4.12496299e-01 -1.28735676e-01 -1.84426737e+00 -6.33216381e-01 8.14721823e-01 -2.94570982e-01 1.28693354e+00 1.17952958e-01 7.45444894e-01 7.88475752e-01 2.14156970e-01 7.09114373e-01 6.08527482e-01 -5.12731373e-01 -1.58566788e-01 -4.51229364e-01 -3.86074692e-01 9.33352530e-01 3.90465617e-01 2.89519817e-01 -6.63681030e-01 -4.99761879e-01 1.12974691e+00 -5.14178336e-01 -3.51484239e-01 -1.02364123e+00 -1.26100111e+00 8.45069349e-01 2.48660415e-01 1.98449448e-01 -6.25060797e-02 9.58245620e-02 4.72579330e-01 3.62939745e-01 7.13319242e-01 4.39996004e-01 -3.83339912e-01 -1.45616993e-01 -1.33365834e+00 4.60704476e-01 6.99983656e-01 1.00511348e+00 3.18886250e-01 -2.08570540e-01 1.51276402e-02 5.62117398e-01 2.72692621e-01 8.01833495e-02 -8.72232094e-02 -8.69759738e-01 3.11076075e-01 3.06906730e-01 5.35717756e-02 -1.15151834e+00 -8.21709871e-01 -6.02102697e-01 -7.74501562e-01 5.40779352e-01 5.79012215e-01 7.23363310e-02 -4.82137531e-01 1.41072702e+00 6.78217560e-02 3.34341228e-01 -1.91623628e-01 7.76732743e-01 1.09439921e+00 -3.07910174e-01 -7.54186213e-01 1.94972992e-01 1.25700164e+00 -3.91357571e-01 -2.57594526e-01 2.84333616e-01 6.30628109e-01 -1.41670525e+00 -6.90056756e-03 4.09121037e-01 -1.13158631e+00 -1.15481928e-01 -1.29709196e+00 -8.62814188e-02 -3.40646446e-01 5.66331565e-01 9.47484434e-01 8.20215225e-01 -1.26834691e+00 6.90451920e-01 -6.42062426e-01 -4.61323559e-01 6.67457700e-01 3.41365635e-01 -7.49107599e-01 3.06872457e-01 -7.78902113e-01 8.29698503e-01 7.57519389e-03 -4.68594506e-02 -6.47818893e-02 -9.19870913e-01 -1.14644468e+00 5.02820276e-02 -1.25175297e-01 -7.37764418e-01 1.28142333e+00 -1.47996843e-01 -1.09095669e+00 1.42757559e+00 -4.05095555e-02 -5.14670789e-01 1.06566763e+00 1.25651851e-01 -3.59358490e-01 1.95924640e-01 -1.72412768e-01 6.41729236e-01 9.86665726e-01 -6.99960530e-01 -4.22481775e-01 -1.91196904e-01 -2.76348054e-01 2.20070019e-01 -1.25685502e-02 5.31130850e-01 -3.23307842e-01 -6.85764968e-01 6.45020902e-01 -6.36004508e-01 3.53658721e-02 5.14795542e-01 -2.74413586e-01 -3.61987591e-01 1.04331481e+00 -4.54378963e-01 6.64021194e-01 -2.23620343e+00 -2.64773965e-01 5.76699615e-01 5.30763030e-01 -2.33816490e-01 -7.78308278e-03 5.77434719e-01 -6.99238658e-01 1.15549125e-01 3.00226118e-02 -5.91328561e-01 1.80735707e-01 -6.86319768e-01 5.81336766e-02 1.09073532e+00 3.68724048e-01 7.89961636e-01 -5.40653467e-01 -4.78074610e-01 2.61454672e-01 7.76135087e-01 -6.07954621e-01 -5.16338706e-01 6.42578959e-01 -1.42056242e-01 1.48215406e-02 7.25584090e-01 1.16678178e+00 -2.64626324e-01 -2.80069530e-01 -4.56979573e-01 -2.44604647e-01 2.58706212e-01 -1.62552106e+00 1.68991649e+00 -2.93713480e-01 1.18059480e+00 2.03678594e-03 -9.16925550e-01 1.08189511e+00 4.47309405e-01 7.11647272e-01 -2.37696484e-01 1.09428301e-01 3.17437589e-01 1.72049299e-01 1.50045887e-01 5.09274423e-01 1.88757077e-01 2.07342774e-01 4.51616764e-01 1.33300811e-01 -4.41808462e-01 3.36527705e-01 2.38560796e-01 1.12730551e+00 1.98755145e-01 4.99098897e-01 -5.14634788e-01 3.25736552e-01 -3.56190532e-01 3.94717127e-01 4.73056912e-01 -1.17153442e-02 1.20239365e+00 4.72896457e-01 -6.74977303e-01 -1.23578274e+00 -1.14704430e+00 -8.46508741e-01 6.73235729e-02 1.90674961e-02 -8.85211408e-01 -3.67629945e-01 -3.75676125e-01 1.99396640e-01 1.10428914e-01 -7.76078641e-01 1.21056527e-01 -3.69495511e-01 -4.30834353e-01 6.63120210e-01 6.41095519e-01 6.34361744e-01 -4.73541290e-01 -5.75060308e-01 -1.94253489e-01 1.88382208e-01 -1.19815767e+00 -8.08866799e-01 -5.31382896e-02 -6.24413729e-01 -1.55007923e+00 -8.88676643e-01 -9.25260782e-01 6.21046424e-01 2.76054144e-01 1.23214102e+00 -1.04025327e-01 -8.69130135e-01 7.21563160e-01 1.56515669e-02 -3.18172693e-01 1.24896742e-01 -3.47001851e-01 7.72381872e-02 -1.37055829e-01 6.81641757e-01 -6.36362314e-01 -5.26123822e-01 3.50068420e-01 -1.00143388e-01 -1.19040579e-01 4.57758904e-01 8.61424327e-01 2.93661058e-01 3.74626577e-01 -1.40089199e-01 -3.34004939e-01 9.94630277e-01 1.04239404e-01 -8.66603494e-01 9.36002433e-02 -7.30240569e-02 -1.68952480e-01 -1.35592863e-01 -1.48655772e-01 -8.13814580e-01 3.49619985e-01 1.08485527e-01 -5.97407758e-01 -1.99123248e-01 4.18185323e-01 4.86189723e-01 -9.57188308e-01 6.56271815e-01 1.49360344e-01 3.28655839e-02 -1.98725536e-01 1.44486120e-02 2.95538038e-01 5.05220592e-01 -5.07968247e-01 8.53136003e-01 6.21103346e-01 5.11705160e-01 -1.01456654e+00 -3.58910374e-02 -6.67875886e-01 -7.99761415e-01 -4.92845066e-02 5.47766149e-01 -9.90189970e-01 -9.87599850e-01 6.48251295e-01 -1.32731891e+00 3.82709742e-01 -2.98474342e-01 7.60622621e-01 -9.55759168e-01 5.30399561e-01 -3.21296066e-01 -7.13663340e-01 -2.48824641e-01 -7.90125310e-01 1.40354264e+00 5.69766909e-02 -3.71454358e-01 -9.74193096e-01 1.64413482e-01 -3.78648341e-01 4.82125372e-01 7.99397603e-02 6.27450943e-01 -8.31175983e-01 -2.88623840e-01 -7.48127162e-01 -6.12253785e-01 4.25368585e-02 -1.65362567e-01 2.73864329e-01 -7.50548959e-01 -2.04041302e-01 -3.27020913e-01 -1.22345656e-01 9.36592460e-01 6.76824749e-01 9.54682112e-01 5.10176301e-01 -4.29729283e-01 7.00708091e-01 1.02391458e+00 -4.55798954e-01 4.59539592e-01 1.97007239e-01 5.88054908e-03 3.42789829e-01 2.70520717e-01 5.70202470e-01 2.45095845e-02 7.99128652e-01 2.84997642e-01 -2.94673055e-01 -3.74105573e-01 -1.11946292e-01 -2.92719841e-01 -7.35701844e-02 -6.35120571e-01 5.44367284e-02 -9.66515541e-01 3.29031467e-01 -1.77997530e+00 -1.27868259e+00 -2.14427143e-01 2.17262936e+00 8.64245296e-02 2.43289918e-01 2.24491850e-01 2.71248341e-01 8.45292687e-01 8.64900798e-02 -7.47729419e-03 -2.22015575e-01 -4.13850844e-01 5.29148698e-01 7.08395839e-01 3.03357899e-01 -1.58581388e+00 5.20063996e-01 7.88002968e+00 6.78948224e-01 -7.80652165e-01 -2.54854739e-01 4.10911664e-02 1.81511059e-01 -5.40328771e-02 -6.81125447e-02 -8.20730805e-01 -2.76068784e-02 -9.64026228e-02 -1.50089458e-01 -8.44693780e-02 7.13563919e-01 9.06697884e-02 7.36871511e-02 -1.12638617e+00 1.45484340e+00 2.23049849e-01 -1.88018465e+00 -4.95325536e-01 3.08727145e-01 5.87947965e-01 3.25931579e-01 1.89167663e-01 -3.46935451e-01 8.21765140e-02 -8.97115469e-01 5.09111822e-01 3.94622684e-01 8.51878583e-01 -9.75212038e-01 7.67601073e-01 -1.52314559e-01 -1.52789962e+00 3.00700128e-01 -2.54047781e-01 1.42393515e-01 6.96513131e-02 6.08678997e-01 -8.07850718e-01 6.05534434e-01 9.55688179e-01 9.57213819e-01 -4.22881901e-01 1.93767977e+00 -9.25159827e-02 1.41582862e-02 -9.61838067e-01 1.03128523e-01 2.50849593e-02 -2.81358898e-01 9.15680468e-01 1.27958727e+00 5.27113557e-01 -1.49793133e-01 -1.37131676e-01 9.30779696e-01 -5.09548485e-02 1.85375750e-01 -9.89445627e-01 9.72090587e-02 1.81597650e-01 1.49836624e+00 -1.00408399e+00 3.11154813e-01 -6.23640835e-01 7.54182041e-01 1.26108602e-01 2.16572896e-01 -5.31058609e-01 -8.49821031e-01 5.91112494e-01 1.96548745e-01 6.84897602e-01 -5.89476168e-01 -1.74414605e-01 -1.26262116e+00 1.44670978e-01 -3.98953229e-01 4.16961312e-01 -8.94711971e-01 -1.31534266e+00 2.74489671e-01 2.48654615e-02 -1.52766407e+00 9.41107720e-02 -1.13824236e+00 -1.08780229e+00 8.30388367e-01 -1.17765570e+00 -1.40857160e+00 -7.29268119e-02 5.20836115e-01 9.53755826e-02 -5.58348477e-01 1.01548791e+00 2.17065752e-01 1.04379185e-01 8.23518097e-01 -9.69909504e-02 5.39290011e-01 7.51384914e-01 -1.08771777e+00 7.46582329e-01 8.60136390e-01 5.83841562e-01 5.65088272e-01 5.62230885e-01 -4.30588216e-01 -1.04313850e+00 -2.40299612e-01 1.27409601e+00 -3.01117361e-01 7.55776346e-01 -4.02409464e-01 -2.96532661e-01 7.00879931e-01 1.97436452e-01 3.54074180e-01 7.76577175e-01 5.47692180e-01 -6.75851762e-01 5.19406617e-01 -1.24812162e+00 5.68609536e-01 1.56178522e+00 -3.97803128e-01 -6.74229503e-01 6.05339944e-01 -1.85977370e-01 -8.68839800e-01 -9.45748866e-01 8.16831768e-01 7.02955902e-01 -1.22295070e+00 1.15451896e+00 -6.43542111e-01 8.13263282e-02 -1.07873827e-01 -1.21593907e-01 -1.03027594e+00 -5.05828977e-01 -9.51398432e-01 -3.28007489e-02 5.72766006e-01 4.67097819e-01 -6.24702930e-01 1.32178235e+00 6.89923987e-02 -2.46823967e-01 -5.66861689e-01 -1.45136750e+00 -9.89003122e-01 -7.05027282e-02 -4.57749277e-01 3.16026658e-01 8.65137637e-01 9.00456682e-02 -2.53307104e-01 -1.78767927e-02 1.81469426e-01 8.69697332e-01 3.66214007e-01 7.63031423e-01 -1.62240100e+00 -8.51068199e-02 -1.11308432e+00 -1.22256637e+00 -1.01628816e+00 1.88955918e-01 -1.00486434e+00 -6.38182223e-01 -1.00793040e+00 1.04021594e-01 -1.95251465e-01 3.00036967e-02 3.65063101e-01 7.43003309e-01 7.22133517e-01 -2.93835640e-01 -3.57549131e-01 -2.19679356e-01 4.17958945e-01 1.00960445e+00 -1.38314560e-01 2.09511206e-01 5.08843958e-01 -3.73106033e-01 8.15300465e-01 7.22958565e-01 -2.18541488e-01 -1.45755038e-01 3.53474505e-02 6.06488287e-01 -3.52691799e-01 5.38652062e-01 -9.05821800e-01 5.56609094e-01 5.11585057e-01 8.31865489e-01 -9.93595421e-01 5.98524928e-01 -6.85917258e-01 -9.09549147e-02 2.47342795e-01 1.12373888e-01 2.53214836e-01 3.91534746e-01 3.68334174e-01 -3.04393083e-01 -2.74414599e-01 7.01532483e-01 -2.94400919e-02 -5.63479006e-01 4.78533894e-01 -3.28834176e-01 6.21488914e-02 9.46362197e-01 -7.44872212e-01 -1.04295976e-01 -6.27033472e-01 -8.58591914e-01 -9.08375755e-02 4.48064446e-01 4.79130179e-01 1.04867160e+00 -1.54152346e+00 -1.00929272e+00 6.48085237e-01 3.83540928e-01 -4.10480678e-01 5.18196747e-02 1.19299901e+00 -7.80010462e-01 4.30573314e-01 -3.41847271e-01 -7.10946023e-01 -1.56401765e+00 2.26067156e-01 6.43393934e-01 2.49132872e-01 -8.05369616e-01 1.08500600e+00 -3.35633934e-01 -4.94120479e-01 3.07791531e-01 -2.60107014e-02 5.42315133e-02 -5.28066792e-02 3.91630471e-01 4.41176802e-01 4.68233585e-01 -6.57810748e-01 -5.28664410e-01 1.10226607e+00 -2.76100598e-02 -4.10155505e-02 1.45789170e+00 3.16092104e-01 -1.30537853e-01 -3.57846588e-01 1.23520601e+00 4.45482522e-01 -7.68890917e-01 -2.41962045e-01 -1.01345047e-01 -6.60964847e-01 2.58688331e-01 -4.44519430e-01 -9.31909740e-01 7.16457427e-01 4.28689629e-01 8.64347816e-02 6.96806192e-01 1.24709411e-02 1.92420691e-01 6.14015698e-01 9.73573551e-02 -1.03823507e+00 -3.04030597e-01 5.34798980e-01 1.20590436e+00 -1.21470881e+00 3.35365862e-01 -7.16711342e-01 -8.97776112e-02 1.79065681e+00 3.00123572e-01 -5.00856996e-01 1.31342280e+00 5.79946756e-01 -1.85468763e-01 -5.13178468e-01 -1.48290113e-01 -1.10948130e-01 8.77754271e-01 8.40911150e-01 5.08731902e-01 -2.29877889e-01 -2.97137082e-01 1.46863118e-01 -3.89828682e-01 -3.18691999e-01 3.13768297e-01 5.79204917e-01 -1.63726807e-01 -9.79751825e-01 -2.01407000e-01 2.27192208e-01 -3.22970539e-01 -2.94371098e-01 -7.37811744e-01 1.24227929e+00 -9.52150524e-02 4.43088949e-01 5.81018686e-01 -1.13112949e-01 2.23909214e-01 -9.55741256e-02 9.40022349e-01 -3.72100145e-01 -2.02797219e-01 1.48161333e-02 3.99843961e-01 -6.36050582e-01 -6.05760634e-01 -1.03024292e+00 -8.66089702e-01 -4.17943388e-01 -4.58783358e-01 -1.06393449e-01 7.71929324e-01 5.63025951e-01 4.26936746e-01 -1.10689022e-01 4.49360877e-01 -1.09630895e+00 -5.54807425e-01 -7.32511044e-01 -7.21095920e-01 1.32371664e-01 1.45093992e-01 -1.05022001e+00 -6.14560962e-01 -7.33512044e-01]
[8.607162475585938, -2.1008033752441406]
f9e7aff5-2721-4bf8-8923-dc411f49d746
text-driven-visual-synthesis-with-latent
2302.0851
null
https://arxiv.org/abs/2302.08510v2
https://arxiv.org/pdf/2302.08510v2.pdf
Text-driven Visual Synthesis with Latent Diffusion Prior
There has been tremendous progress in large-scale text-to-image synthesis driven by diffusion models enabling versatile downstream applications such as 3D object synthesis from texts, image editing, and customized generation. We present a generic approach using latent diffusion models as powerful image priors for various visual synthesis tasks. Existing methods that utilize such priors fail to use these models' full capabilities. To improve this, our core ideas are 1) a feature matching loss between features from different layers of the decoder to provide detailed guidance and 2) a KL divergence loss to regularize the predicted latent features and stabilize the training. We demonstrate the efficacy of our approach on three different applications, text-to-3D, StyleGAN adaptation, and layered image editing. Extensive results show our method compares favorably against baselines.
['Jia-Bin Huang', 'Badour AlBahar', 'Yao-Chih Lee', 'Yiran Xu', 'Songwei Ge', 'Ting-Hsuan Liao']
2023-02-16
null
null
null
null
['text-to-3d']
['computer-vision']
[ 5.67430377e-01 3.75688225e-01 -2.75038570e-01 -3.15421700e-01 -8.20570469e-01 -5.07790685e-01 1.19392574e+00 -6.46137059e-01 9.00576562e-02 5.68853021e-01 6.75308526e-01 -1.53210506e-01 5.65653980e-01 -4.38793659e-01 -7.87258267e-01 -6.40080333e-01 4.12041843e-01 2.89050132e-01 1.77758127e-01 -1.69125259e-01 4.05992448e-01 5.63774228e-01 -1.18442655e+00 5.09404004e-01 6.68255866e-01 7.58055687e-01 3.55490655e-01 8.22395802e-01 -1.55104101e-01 8.49504173e-01 -4.71095234e-01 -4.73081768e-01 5.34603417e-01 -6.57509029e-01 -4.22872961e-01 5.27643323e-01 5.67364573e-01 -7.66291916e-01 -3.88180763e-01 8.48547101e-01 6.13020301e-01 -3.82233337e-02 9.20076013e-01 -1.05383420e+00 -1.04543042e+00 3.13632011e-01 -8.04082572e-01 -3.18110943e-01 3.29102635e-01 5.69216847e-01 8.19952011e-01 -1.08291090e+00 1.05445933e+00 1.51848364e+00 4.93648827e-01 9.58720863e-01 -1.38858163e+00 -4.88536537e-01 1.19518712e-01 -4.44039822e-01 -1.04470372e+00 -9.67116237e-01 6.93005264e-01 -5.94848275e-01 1.06420982e+00 -1.74284575e-03 5.27517438e-01 1.39012587e+00 3.88580471e-01 9.70997691e-01 9.36911404e-01 -5.65518498e-01 -1.27116948e-01 2.05507040e-01 -7.64431298e-01 9.22278404e-01 -2.83874512e-01 1.75901562e-01 -8.42361450e-01 -3.92257199e-02 1.27958465e+00 -4.50465202e-01 -1.06069826e-01 -2.65424669e-01 -1.21791673e+00 8.46675515e-01 3.44966538e-02 -2.10561842e-01 -2.92285293e-01 6.56863868e-01 1.02879658e-01 3.56852233e-01 9.54415977e-01 3.69107306e-01 -1.88067555e-01 2.81622056e-02 -1.13125956e+00 3.65920931e-01 5.71173787e-01 1.38653123e+00 6.78341448e-01 4.78875250e-01 -7.06652820e-01 8.37606788e-01 5.74134707e-01 6.03108227e-01 2.53457248e-01 -1.34743345e+00 5.34926236e-01 -3.49254981e-02 2.51042962e-01 -7.76367843e-01 3.32971632e-01 -7.42743835e-02 -5.89464366e-01 5.58568954e-01 1.03903510e-01 -2.17734769e-01 -1.32471013e+00 1.68122900e+00 2.32632354e-01 1.23417541e-01 -9.10377130e-02 6.42978191e-01 5.19614220e-01 9.88914192e-01 -1.42046362e-01 -4.35802108e-03 8.17960739e-01 -1.20161474e+00 -6.75984204e-01 -2.88535923e-01 3.10807377e-01 -1.22177124e+00 1.07835174e+00 1.94022395e-02 -1.61714053e+00 -4.43275273e-01 -9.65396523e-01 -3.82751793e-01 -3.47731123e-03 1.19177654e-01 5.19637108e-01 5.21280110e-01 -1.42700124e+00 6.57540679e-01 -7.24596262e-01 -2.13585928e-01 5.63871264e-01 2.87891507e-01 -1.73225418e-01 -3.81058529e-02 -7.39664912e-01 9.06627357e-01 -1.57515481e-01 -2.64962763e-01 -1.11091685e+00 -8.07094574e-01 -8.03573966e-01 -8.97822529e-02 -2.74033844e-02 -1.20257783e+00 1.32899165e+00 -1.21031165e+00 -2.14603305e+00 1.03105605e+00 -2.70479918e-01 -3.39469969e-01 1.01919472e+00 -2.93641269e-01 2.05818206e-01 4.19207737e-02 2.11436599e-01 1.31611478e+00 1.50499725e+00 -1.23304987e+00 -3.65031153e-01 3.18218708e-01 -2.72492051e-01 3.99555445e-01 -1.23597562e-01 4.48975973e-02 -8.19626689e-01 -1.07426476e+00 -2.24933311e-01 -8.79818916e-01 -3.12710702e-01 5.48267722e-01 -5.20247340e-01 1.13411270e-01 9.20664907e-01 -6.23311639e-01 5.93062520e-01 -2.11524057e+00 3.74785066e-01 8.00466910e-02 1.70373768e-01 -9.33760428e-04 -4.62596655e-01 4.33980078e-01 1.03849612e-01 2.02418804e-01 -1.16283514e-01 -8.53528976e-01 1.00932047e-01 7.11019859e-02 -6.30662143e-01 2.87831962e-01 4.71491963e-01 1.18361664e+00 -6.77823186e-01 -4.45976913e-01 2.64245838e-01 7.29085565e-01 -7.72676587e-01 2.87249446e-01 -6.56760335e-01 7.95513749e-01 -3.11976969e-01 4.88578081e-01 4.58101183e-01 -3.73345226e-01 -1.21583104e-01 -1.17084324e-01 -1.24571107e-01 1.64655089e-01 -7.76622474e-01 1.92094040e+00 -4.28217590e-01 9.94330764e-01 1.43407509e-01 -3.08396578e-01 7.28560269e-01 3.89779270e-01 2.67334282e-01 -3.51115644e-01 -5.41246645e-02 1.05198897e-01 -4.87480402e-01 -2.20387071e-01 5.58190942e-01 -1.08699843e-01 2.75682718e-01 5.31019509e-01 1.41834930e-01 -8.64054501e-01 4.10089605e-02 4.20055777e-01 7.82956600e-01 7.54763901e-01 -2.39596277e-01 -1.94645479e-01 5.15904315e-02 -1.39302552e-01 2.40820691e-01 7.21289158e-01 2.51883179e-01 1.14584613e+00 4.94228542e-01 -1.13547266e-01 -1.67743742e+00 -1.10257721e+00 3.56250226e-01 7.68675685e-01 -1.52962282e-01 -4.11275327e-01 -7.32755840e-01 -7.36033559e-01 -1.62156112e-02 9.25028682e-01 -5.92480421e-01 4.53111231e-02 -4.44991499e-01 -3.50229651e-01 5.28488457e-01 3.59464765e-01 3.58415484e-01 -8.57984543e-01 -1.27517611e-01 1.42314479e-01 -6.33091852e-02 -1.08846307e+00 -9.56414819e-01 -1.94826394e-01 -9.29747939e-01 -2.82099098e-01 -1.13250244e+00 -7.50845194e-01 8.70487571e-01 2.10172132e-01 1.02977777e+00 -8.17814320e-02 -2.03350261e-01 4.66490358e-01 7.51191229e-02 -3.06919575e-01 -8.73649597e-01 -9.51033682e-02 -2.65841093e-02 -1.39063345e-02 -4.16093409e-01 -6.00866258e-01 -6.74968839e-01 2.33070895e-01 -9.08068478e-01 6.22693896e-01 6.54882789e-01 8.16818833e-01 3.92928660e-01 -6.72135353e-01 3.74610186e-01 -9.34107482e-01 7.14916348e-01 -2.97858149e-01 -6.90900505e-01 1.97184920e-01 -6.39295220e-01 2.93279260e-01 3.35510135e-01 -6.16594791e-01 -1.41753769e+00 1.77984625e-01 -2.54439175e-01 -6.48280144e-01 2.86191195e-01 1.01666451e-01 -5.47138639e-02 -2.50955015e-01 6.76941752e-01 2.32787535e-01 2.27334782e-01 -1.86875746e-01 8.90660405e-01 3.78784627e-01 2.74846226e-01 -6.56169891e-01 1.07656574e+00 4.96272862e-01 -1.06489003e-01 -8.63376498e-01 -6.61948204e-01 2.50170946e-01 -4.36252981e-01 -6.41560405e-02 9.41954970e-01 -1.18419576e+00 -9.23263580e-02 4.51513588e-01 -1.40448558e+00 -9.03244853e-01 -5.70912898e-01 2.59318650e-01 -9.70629215e-01 3.07857096e-01 -9.32558060e-01 -4.40698624e-01 -2.73370355e-01 -1.35886848e+00 1.44763756e+00 1.04322329e-01 -3.21722269e-01 -1.08622193e+00 -1.20441150e-02 2.02946514e-01 7.31469750e-01 1.89836338e-01 9.01409805e-01 1.52949214e-01 -1.01430130e+00 1.67234942e-01 -2.54926264e-01 3.73385936e-01 1.60124138e-01 3.39776695e-01 -9.17596161e-01 -2.18030781e-01 -3.08563888e-01 -4.90469068e-01 8.77278984e-01 5.97950459e-01 8.41592968e-01 -2.14357808e-01 -4.07332510e-01 9.85622048e-01 1.04783034e+00 -1.50857955e-01 7.91625917e-01 -1.39907464e-01 8.29734445e-01 3.88210505e-01 1.34754300e-01 4.52838004e-01 1.86459377e-01 5.96067786e-01 7.59844556e-02 -3.73239458e-01 -8.27429891e-01 -5.04724383e-01 6.86809182e-01 8.81273091e-01 2.85515822e-02 -3.71467441e-01 -4.57058758e-01 5.18271327e-01 -1.66271746e+00 -8.79759789e-01 1.87780917e-01 1.72574103e+00 9.73014057e-01 1.33059114e-01 -3.18169981e-01 -6.44503951e-01 6.95154667e-01 5.00198960e-01 -6.93088174e-01 -4.73992765e-01 -1.97652310e-01 1.64900333e-01 4.97672141e-01 8.18009079e-01 -7.71827281e-01 1.51898897e+00 7.55185938e+00 9.09169853e-01 -1.09395099e+00 1.19830489e-01 8.46793175e-01 -2.08205029e-01 -8.51293087e-01 1.20548919e-01 -7.62595952e-01 2.10872695e-01 4.68544185e-01 -3.13533843e-03 5.24629772e-01 5.27607083e-01 3.81119400e-01 2.11895749e-01 -1.06723762e+00 8.55906606e-01 1.93228424e-01 -1.79882598e+00 4.96842235e-01 1.41844481e-01 1.39191198e+00 5.50997145e-02 6.44117236e-01 -1.25915125e-01 7.69354999e-01 -8.94598186e-01 1.01772845e+00 5.52137733e-01 1.43236470e+00 -3.54181916e-01 -1.89472392e-01 6.12660348e-02 -7.72574782e-01 3.04143757e-01 -2.30869621e-01 2.86781698e-01 4.30552572e-01 4.97302532e-01 -8.17316353e-01 1.46222621e-01 1.79442674e-01 8.74690413e-01 -1.00482367e-01 4.98682171e-01 -5.56710362e-01 4.67761099e-01 -2.42388204e-01 2.83804983e-01 3.05245280e-01 -1.47607327e-01 6.86335742e-01 1.16703057e+00 6.43432796e-01 -1.35556355e-01 -1.62363097e-01 1.20243895e+00 -2.60383636e-01 -7.30860680e-02 -9.38179612e-01 -1.90089881e-01 1.25758201e-01 9.51059461e-01 -7.51539886e-01 -4.68640745e-01 -5.62637806e-01 1.65508044e+00 2.08130553e-01 5.83281577e-01 -8.87870610e-01 -2.02768177e-01 7.14893341e-01 1.80817753e-01 5.64319611e-01 -5.46672523e-01 -3.41814309e-01 -1.40305829e+00 -2.52690315e-01 -7.32438743e-01 -3.14032584e-01 -1.15529227e+00 -1.30346334e+00 3.68406385e-01 -1.50729820e-01 -9.72181499e-01 -4.42794710e-01 -5.08396924e-01 -5.90687156e-01 1.04151845e+00 -1.50136185e+00 -1.53301549e+00 7.59949982e-02 5.80498219e-01 8.96652997e-01 -2.95538723e-01 5.79380810e-01 2.78847702e-02 -1.04605094e-01 4.89442736e-01 1.90738142e-01 -1.64686188e-01 1.11399329e+00 -1.08689582e+00 1.09244692e+00 9.60023105e-01 1.15978415e-03 4.37972695e-01 6.23716056e-01 -1.03830123e+00 -1.42723322e+00 -9.33282971e-01 6.59490705e-01 -6.60116553e-01 4.23380852e-01 -5.10357559e-01 -2.55792558e-01 1.00796282e+00 6.85449243e-01 -2.71298468e-01 3.13805997e-01 -3.47246110e-01 -2.95025557e-01 2.02208400e-01 -9.62806344e-01 1.04723704e+00 1.12794363e+00 -5.65199852e-01 -5.63080497e-02 4.79032189e-01 7.94454694e-01 -6.31965935e-01 -6.14333928e-01 8.35769996e-02 6.61581993e-01 -8.09851527e-01 9.16922927e-01 -3.58374298e-01 9.09172416e-01 -1.45005971e-01 -4.02549617e-02 -1.36351740e+00 -3.19931448e-01 -1.37293839e+00 -1.04982220e-01 1.19885838e+00 6.15574658e-01 -3.03487539e-01 8.20596576e-01 7.91803718e-01 -2.74707496e-01 -4.26987827e-01 -4.42125142e-01 -3.45294714e-01 1.20392822e-01 -2.86972940e-01 4.00418520e-01 8.44254017e-01 -4.79766697e-01 4.01664883e-01 -8.92112553e-01 -1.32564679e-01 5.45661926e-01 -5.96351512e-02 1.00294447e+00 -6.99657559e-01 -4.20824111e-01 -5.68788230e-01 5.23835048e-02 -1.88271296e+00 1.20415069e-01 -8.38982463e-01 1.54802769e-01 -1.66421449e+00 7.72525370e-02 -3.39430600e-01 3.46357346e-01 3.21610481e-01 -1.41423762e-01 3.25808823e-01 3.28299850e-01 2.58923173e-01 -3.19795281e-01 7.19006658e-01 1.65258217e+00 -1.24225512e-01 -2.31510609e-01 -1.52814463e-01 -6.93438649e-01 6.02570653e-01 5.51143289e-01 -4.34765249e-01 -6.36921704e-01 -1.02307832e+00 2.94344723e-01 9.06486660e-02 8.85808244e-02 -6.20225430e-01 9.81217325e-02 -2.80111879e-01 6.95394397e-01 -2.67718703e-01 6.20069861e-01 -4.97250736e-01 3.91058810e-02 2.13895559e-01 -4.29882288e-01 -3.78327444e-02 9.47577283e-02 5.53990722e-01 9.85220298e-02 -9.01876576e-03 8.66127372e-01 -2.33034059e-01 -3.99374425e-01 4.91724640e-01 -4.71698850e-01 2.28448752e-02 8.46651912e-01 -2.95425504e-01 -2.11645633e-01 -9.49882686e-01 -6.99725807e-01 -1.16739031e-02 7.46921182e-01 4.79568809e-01 7.88541317e-01 -1.44227004e+00 -9.66025651e-01 4.33889717e-01 -2.07403377e-01 -1.27332985e-01 -7.43122920e-02 6.35311902e-01 -7.22611487e-01 1.18496716e-01 -2.42303163e-01 -6.96979403e-01 -1.04220366e+00 2.94643164e-01 1.90192148e-01 -1.08499505e-01 -6.86650276e-01 1.05122626e+00 5.73900521e-01 -2.13315725e-01 9.34155732e-02 -1.39265209e-01 3.89403492e-01 -2.04064369e-01 4.10523236e-01 -2.35929489e-02 -4.08862680e-01 -4.99452025e-01 1.38895243e-01 8.31822217e-01 -1.98419541e-01 -7.85356641e-01 1.24432492e+00 -4.70485598e-01 1.79356664e-01 1.96980238e-01 1.00343096e+00 3.94674987e-01 -2.22893333e+00 -1.84505776e-01 -4.48463529e-01 -6.59964204e-01 -8.83750152e-03 -5.63281417e-01 -1.12825263e+00 9.65349197e-01 3.34765553e-01 -1.15244642e-01 8.92457008e-01 -1.50893301e-01 8.57052326e-01 3.87986414e-02 5.82212210e-02 -1.00401723e+00 3.79473746e-01 5.27039528e-01 9.69786048e-01 -1.02572668e+00 4.19398211e-02 -2.61913925e-01 -9.77259815e-01 1.13728523e+00 3.58412474e-01 -2.49849871e-01 5.89558125e-01 5.06633043e-01 2.41783291e-01 1.43525302e-01 -1.03066981e+00 3.78084667e-02 3.98287147e-01 6.27208650e-01 5.12386203e-01 -4.27218854e-01 6.64599985e-02 -2.52912343e-01 1.39360815e-01 1.39253046e-02 4.40007508e-01 7.80853987e-01 -1.28585547e-01 -1.28449512e+00 -6.61392659e-02 2.61207610e-01 -6.17190003e-01 -4.18980122e-01 -3.32098246e-01 4.15774435e-01 -2.03114554e-01 6.91500068e-01 -8.75868946e-02 -2.38189772e-01 -1.89413112e-02 2.79651079e-02 8.75808120e-01 -7.03569114e-01 -3.76354873e-01 5.16045690e-01 6.14928007e-02 -6.70378149e-01 -4.23861086e-01 -6.50638700e-01 -8.71590495e-01 -4.27302927e-01 -2.25164503e-01 -2.95457572e-01 8.70813727e-01 7.76952446e-01 6.77144170e-01 3.48449141e-01 6.11435115e-01 -1.26640582e+00 -3.93252373e-01 -7.62725294e-01 -2.56723583e-01 2.90614843e-01 4.03739870e-01 -4.11589295e-01 -2.41907671e-01 5.76785386e-01]
[11.3384370803833, -0.24147245287895203]
f0203051-8780-4f4f-b96b-7858fc149fc8
sparse-and-low-bias-estimation-of-high
null
null
https://openreview.net/forum?id=vK4ta9RgKMg
https://openreview.net/pdf?id=vK4ta9RgKMg
Sparse and Low-bias Estimation of High Dimensional Vector Autoregressive Models
Vector autoregressive ($VAR$) models are widely used for causal discovery and forecasting in multivariate time series analysis. In the high-dimensional setting, which is increasingly common in fields such as neuroscience and econometrics, model parameters are inferred by $L_1$-regularized maximum likelihood (RML). A well-known feature of RML inference is that in general the technique produces a trade-off between sparsity and bias that depends on the choice of the regularization hyperparameter. In the context of multivariate time series analysis, sparse estimates are favorable for causal discovery and low-bias estimates are favorable for forecasting. However, owing to a paucity of research on hyperparameter selection methods, practitioners must rely on \textit{ad-hoc} methods such as cross-validation (or manual tuning). The particular balance that such approaches achieve between the two goals --- causal discovery and forecasting --- is poorly understood. Our paper investigates this behavior and proposes a method ($UoI_{VAR}$) that achieves a better balance between sparsity and bias when the underlying causal influences are in fact sparse. We demonstrate through simulation that RML with a hyperparameter selected by cross-validation tends to overfit, producing relatively dense estimates. We further demonstrate that $UoI_{VAR}$ much more effectively approximates the correct sparsity pattern with only a minor compromise in model fit, particularly so for larger data dimensions, and that the estimates produced by $UoI_{VAR}$ exhibit less bias. We conclude that our method achieves improved performance especially well-suited to applications involving simultaneous causal discovery and forecasting in high-dimensional settings.
['Kristofer Bouchard', 'Mahesh Balasubramanian', 'Sharmodeep Bhattacharyya', 'Trevor Ruiz']
2020-06-08
null
null
null
l4dc-2020-6
['econometrics']
['miscellaneous']
[ 1.48666203e-01 1.01710698e-02 -5.49317062e-01 -4.60199058e-01 -9.51532066e-01 -3.02038074e-01 5.68652570e-01 -1.44735742e-02 -1.42826110e-01 1.02388632e+00 5.47048151e-01 -4.72831219e-01 -6.10430479e-01 -6.41522050e-01 -8.10455322e-01 -8.42063129e-01 -6.12262368e-01 2.47413769e-01 -3.98150146e-01 7.31310472e-02 2.43250743e-01 1.38686553e-01 -1.33717787e+00 -2.63953835e-01 9.31502581e-01 6.91542029e-01 -1.10928610e-01 3.93731475e-01 1.30927503e-01 4.91751254e-01 -3.08638006e-01 -3.16405565e-01 1.46216795e-01 -6.24891222e-01 -3.31513137e-01 -2.04483762e-01 2.93682367e-01 3.44010182e-02 -1.36039525e-01 9.05332446e-01 4.72726017e-01 8.80693346e-02 8.54771972e-01 -1.09922528e+00 -3.46818447e-01 8.24201226e-01 -8.08697701e-01 4.92283404e-01 1.90096125e-01 2.22844124e-01 9.98303533e-01 -7.92116821e-01 4.44952309e-01 1.51651275e+00 7.73142934e-01 6.33938238e-02 -1.69169366e+00 -1.02491117e+00 3.97807837e-01 -2.62451619e-01 -1.48730409e+00 -6.66095555e-01 7.09137201e-01 -6.93519890e-01 6.12501979e-01 2.21630991e-01 2.71968454e-01 1.10576367e+00 4.71542805e-01 3.53973061e-01 1.02417219e+00 -2.91275293e-01 5.66018581e-01 -2.40905434e-02 1.76944472e-02 4.76683646e-01 4.07274157e-01 2.94518024e-01 -6.82436705e-01 -7.53256500e-01 1.16573107e+00 -1.78664446e-01 -3.27413470e-01 -1.65770337e-01 -1.26350868e+00 1.17054498e+00 -9.69749615e-02 2.95182467e-01 -5.79340696e-01 5.27079701e-01 2.60553598e-01 3.56271207e-01 6.65295601e-01 4.96012211e-01 -4.59950894e-01 -1.47107989e-01 -1.07008135e+00 3.88635963e-01 7.15579569e-01 7.07551837e-01 3.35629731e-01 5.63200533e-01 -1.37769692e-02 9.68024969e-01 3.00115645e-01 7.60865450e-01 2.48552248e-01 -9.81494486e-01 2.84567177e-01 1.67593062e-01 4.04601216e-01 -1.28972900e+00 -4.17028964e-01 -5.92944145e-01 -1.14822030e+00 -9.83924419e-02 5.32808781e-01 -4.77581471e-01 -5.41336775e-01 2.22529578e+00 3.36872578e-01 5.06159604e-01 -2.02595174e-01 8.47818732e-01 4.00462776e-01 5.98794520e-01 5.97164571e-01 -9.59215164e-01 1.19232690e+00 3.74860317e-02 -7.91028023e-01 -2.16483638e-01 5.74061334e-01 -5.90230584e-01 1.11840570e+00 3.22271228e-01 -1.14809465e+00 -7.66259357e-02 -6.21536672e-01 5.53228796e-01 1.29857525e-01 -4.67994213e-02 1.00693190e+00 6.69794619e-01 -7.73170710e-01 3.62274796e-01 -7.72718966e-01 -2.54171580e-01 2.67749667e-01 3.50201011e-01 -9.22990888e-02 4.04323973e-02 -1.34895504e+00 5.90323091e-01 -1.07384883e-01 8.72191787e-02 -9.07917321e-01 -1.18453526e+00 -5.75741589e-01 8.45975429e-02 2.17057601e-01 -8.58027101e-01 8.33410203e-01 -8.07109535e-01 -1.04807723e+00 4.73527551e-01 -4.79453087e-01 -4.49273676e-01 3.07763010e-01 -2.86675785e-02 -5.44342518e-01 -2.46048018e-01 1.02125131e-01 2.35922575e-01 1.06893754e+00 -1.09158826e+00 -3.57386142e-01 -5.06653845e-01 -3.15364689e-01 -1.26898438e-01 -1.18369259e-01 7.13751391e-02 -1.29739434e-01 -9.48947012e-01 2.33418122e-01 -8.79163742e-01 -6.56162858e-01 -5.06276727e-01 -3.46332222e-01 -2.23035201e-01 3.25385153e-01 -3.23817402e-01 1.74993825e+00 -1.95981550e+00 -2.38387398e-02 5.65099955e-01 1.97726473e-01 -1.75907165e-01 -2.84656044e-03 4.57646042e-01 -4.61528152e-01 2.55207807e-01 -2.66940862e-01 4.23444100e-02 -2.78842539e-01 1.21971089e-02 -5.59398651e-01 8.10921848e-01 1.20011039e-01 5.50880313e-01 -8.85967612e-01 -3.27256113e-01 1.17006004e-02 4.93698865e-01 -6.94418430e-01 6.23773374e-02 -2.42491573e-01 5.68190157e-01 -6.23218656e-01 4.57250386e-01 3.30951840e-01 -7.37344325e-01 4.52986300e-01 4.88028862e-03 -2.61932641e-01 -7.69406036e-02 -1.56220758e+00 1.11925721e+00 -3.80120754e-01 4.84664083e-01 3.51811945e-02 -1.15071118e+00 9.21767890e-01 4.45356637e-01 6.90172911e-01 -2.99109638e-01 5.28853759e-02 1.82801619e-01 -7.63464272e-02 -5.39296746e-01 6.80755228e-02 -3.35931093e-01 -1.30761582e-02 3.89125526e-01 -3.47776741e-01 1.59026533e-01 9.06366557e-02 3.04286890e-02 1.17162478e+00 -1.84433803e-01 4.45613235e-01 -5.84800363e-01 7.24011511e-02 1.30976632e-01 7.62184262e-01 1.00788593e+00 1.74655125e-01 2.98077554e-01 8.93269598e-01 -3.14184785e-01 -1.01230741e+00 -8.90379190e-01 -5.53315878e-01 9.36382890e-01 -3.56686950e-01 -1.36082515e-01 -4.78232831e-01 -3.90239470e-02 2.52711028e-01 9.48068678e-01 -7.34162211e-01 2.52865069e-02 -6.92293823e-01 -1.46144235e+00 4.08621311e-01 3.79700035e-01 -1.37426749e-01 -6.78155065e-01 -4.73969519e-01 4.66003507e-01 3.85388061e-02 -5.74965954e-01 -2.84076959e-01 1.07702151e-01 -1.15189445e+00 -9.10896719e-01 -7.37287164e-01 -1.14534803e-01 7.27694094e-01 -4.49935831e-02 1.28041840e+00 -3.63820374e-01 -1.13510199e-01 3.97836596e-01 -4.07689102e-02 -4.34971571e-01 -1.15259230e-01 -2.41173312e-01 2.27937981e-01 7.27940947e-02 2.77786583e-01 -9.30484414e-01 -7.54868448e-01 2.66486824e-01 -8.03793371e-01 -4.52453792e-01 5.38530469e-01 1.00238919e+00 6.72171474e-01 3.40518542e-02 1.09983885e+00 -1.17834759e+00 8.19109499e-01 -1.03399014e+00 -8.35507691e-01 1.27812132e-01 -9.91478384e-01 1.00985445e-01 3.64184320e-01 -6.15373373e-01 -1.07747221e+00 -1.61231264e-01 1.42858416e-01 -6.42071128e-01 2.02977285e-02 1.07430935e+00 2.12957695e-01 2.64656454e-01 9.16135430e-01 -1.59393951e-01 1.90814119e-02 -4.45214242e-01 2.95586824e-01 2.20330015e-01 2.04127237e-01 -5.22649527e-01 3.86315912e-01 5.13926685e-01 2.63908565e-01 -9.27014947e-01 -7.95427501e-01 -2.55071729e-01 -7.16019794e-02 -3.72189172e-02 5.91600895e-01 -1.06436360e+00 -8.94662559e-01 -1.37833551e-01 -7.26720631e-01 -2.89506644e-01 -1.66564554e-01 8.68120015e-01 -6.97013617e-01 -9.98854265e-02 -1.88279182e-01 -1.19712913e+00 -5.56770600e-02 -1.00847018e+00 9.14302826e-01 -2.47761115e-01 -6.18704736e-01 -1.28238785e+00 2.45824009e-01 -9.99187455e-02 2.76745051e-01 3.87333453e-01 1.27141929e+00 -4.71638292e-01 -3.01204443e-01 -1.78369790e-01 -5.79928569e-02 -2.90597737e-01 -5.82949631e-02 1.76438894e-02 -7.57188022e-01 -2.00540602e-01 -4.81265411e-02 1.00276973e-02 7.45399952e-01 1.32431984e+00 1.11231327e+00 -5.27458787e-01 -4.83272791e-01 3.99360567e-01 1.34125328e+00 1.89784378e-01 4.61244226e-01 -1.51997462e-01 4.00949568e-01 6.17417634e-01 5.34259617e-01 9.06346142e-01 2.79491812e-01 4.66628402e-01 2.27927253e-01 -3.59502621e-02 2.45896250e-01 -2.59191841e-01 2.38271262e-02 4.90606695e-01 -1.04240969e-01 -2.60106146e-01 -9.47678268e-01 6.19200468e-01 -1.95202744e+00 -1.20566559e+00 -3.93378109e-01 2.54033828e+00 1.10569954e+00 -5.08009978e-02 2.04952881e-01 -1.54052675e-01 6.25705838e-01 1.43028840e-01 -5.94115257e-01 -1.03805631e-01 -1.98262811e-01 4.01115641e-02 7.02810168e-01 5.62197685e-01 -9.04277623e-01 5.70062757e-01 7.64862776e+00 8.49134982e-01 -9.93126571e-01 3.48122716e-02 8.77766669e-01 -3.51867110e-01 -6.77916825e-01 1.32927194e-01 -9.03509080e-01 5.30867457e-01 1.28436482e+00 -2.10495844e-01 1.79948598e-01 6.02435172e-01 1.02772713e+00 -2.81060070e-01 -1.00478244e+00 9.71796155e-01 -3.85020971e-01 -1.55003655e+00 -1.91689134e-01 1.15302950e-01 8.72803986e-01 -1.87349051e-01 1.39589474e-01 3.09143346e-02 5.87809265e-01 -1.29663575e+00 4.29717183e-01 6.02480173e-01 8.96008134e-01 -6.88805223e-01 3.39783490e-01 2.59428799e-01 -8.25425923e-01 -1.87559441e-01 -3.66504401e-01 -4.26941477e-02 4.10360336e-01 1.43571448e+00 -6.83439612e-01 1.06729224e-01 6.63922071e-01 5.59095800e-01 1.24583468e-01 8.22712600e-01 7.56203532e-02 1.17545033e+00 -4.47558820e-01 -1.64555926e-02 -2.76877470e-02 -2.74413735e-01 7.01016068e-01 1.20193934e+00 4.96145248e-01 4.40445393e-01 2.89088413e-02 9.14758027e-01 2.91067243e-01 1.73906833e-01 -7.29846239e-01 -1.66710407e-01 7.93704093e-01 6.26675427e-01 -3.89727294e-01 -2.35624984e-01 -2.87320077e-01 -3.44332531e-02 1.34936228e-01 6.73113227e-01 -6.50594056e-01 2.03015864e-01 7.69572377e-01 3.64578038e-01 1.38290584e-01 -3.82828772e-01 -6.49001539e-01 -1.04095376e+00 -4.23296273e-01 -9.93333757e-01 7.00742364e-01 -4.92796421e-01 -1.50209475e+00 2.36971140e-01 4.61598933e-01 -1.07927918e+00 -6.76098049e-01 -1.04634956e-01 -3.02538037e-01 1.07467389e+00 -1.13899255e+00 -6.95710421e-01 2.51449615e-01 5.22400975e-01 4.38327074e-01 1.02984056e-01 7.55535662e-01 1.95945144e-01 -7.67732918e-01 4.36768711e-01 2.32308686e-01 -4.03919965e-01 6.56015992e-01 -9.02631223e-01 -2.71719936e-02 5.97873330e-01 -1.84058368e-01 1.00230110e+00 1.30057335e+00 -1.06643331e+00 -1.52554739e+00 -9.37257528e-01 9.48327899e-01 -2.57930279e-01 8.80609095e-01 2.92356368e-02 -8.21942568e-01 6.57710910e-01 -2.18861610e-01 -1.62167475e-01 9.00528491e-01 8.01373482e-01 -3.06500345e-01 -1.04541034e-01 -9.46853936e-01 7.15651393e-01 8.69828820e-01 -1.27429321e-01 -1.20496169e-01 4.86365825e-01 4.42162424e-01 2.29955860e-03 -1.30729461e+00 4.65156108e-01 5.70547521e-01 -7.63782084e-01 9.73636210e-01 -8.04346323e-01 3.66151422e-01 -3.40131782e-02 -1.98410451e-01 -1.30347645e+00 -4.23739076e-01 -8.25457335e-01 1.97480824e-02 1.11615348e+00 5.74318469e-01 -6.19862795e-01 6.68569803e-01 8.94667864e-01 2.97998101e-01 -6.52308822e-01 -8.85388136e-01 -5.77322960e-01 1.29477888e-01 -6.91018283e-01 3.30482662e-01 1.30356550e+00 -6.20792173e-02 2.30346084e-01 -5.64973891e-01 2.85473466e-01 1.01443386e+00 1.59424827e-01 5.96233249e-01 -1.34545815e+00 -2.46220142e-01 -3.65129799e-01 3.23891453e-02 -8.39204550e-01 1.59095213e-01 -4.40075487e-01 -9.56781581e-02 -8.41741145e-01 7.91026279e-02 -9.63423133e-01 -1.80707082e-01 3.30403298e-01 -2.82470107e-01 -9.59600657e-02 -3.92568678e-01 4.76796865e-01 -2.81774234e-02 4.83101398e-01 9.60260272e-01 2.97628134e-01 -6.17271006e-01 7.71608502e-02 -6.80435479e-01 7.59221971e-01 5.97813070e-01 -6.36328042e-01 -6.33769691e-01 -2.26921946e-01 5.06591141e-01 8.48343372e-01 4.71763343e-01 -2.62372077e-01 1.44707784e-01 -6.18037581e-01 2.89619118e-01 -3.13789099e-01 1.23950846e-01 -5.34277797e-01 6.76623523e-01 1.13619179e-01 -6.64224029e-01 2.11514920e-01 -1.11414790e-01 8.77104223e-01 -7.46663585e-02 -6.98171183e-02 6.45993471e-01 -2.50949413e-01 -3.61745924e-01 2.53680438e-01 -3.69615763e-01 2.53923804e-01 6.51874781e-01 9.71642807e-02 -1.35882258e-01 -7.38633931e-01 -8.32474411e-01 1.95322007e-01 -4.93046008e-02 3.62935103e-02 3.76322955e-01 -1.25301731e+00 -9.35826182e-01 1.98134407e-02 -1.29153788e-01 -3.40722144e-01 3.30772489e-01 1.23672593e+00 2.43014917e-01 5.11441052e-01 3.60459208e-01 -6.52386248e-01 -7.62301207e-01 3.67447317e-01 5.76602556e-02 -1.90261081e-01 -4.99466270e-01 7.73641229e-01 3.94446045e-01 1.02304243e-01 2.12539881e-01 7.89034460e-03 -1.05052218e-01 2.68645495e-01 4.81823593e-01 6.15156054e-01 -2.21809596e-01 -3.20897460e-01 -2.22096294e-01 2.42179692e-01 3.03349555e-01 -1.51695222e-01 1.42812228e+00 -2.77940959e-01 -2.44897425e-01 6.70158684e-01 9.63384569e-01 4.98745739e-02 -1.34830952e+00 -7.66964033e-02 6.47871420e-02 -5.43873310e-01 3.10465604e-01 -4.73332852e-01 -9.31294620e-01 4.02890652e-01 4.23645079e-01 4.39900339e-01 8.81574452e-01 -5.61453355e-03 9.74433944e-02 6.84029013e-02 3.42862785e-01 -7.77837217e-01 -2.40711123e-01 1.25161290e-01 1.00249064e+00 -1.15656495e+00 2.27708787e-01 -4.30229366e-01 -4.13176805e-01 6.65351391e-01 7.12240860e-02 -3.29548508e-01 1.03280032e+00 3.56294543e-01 -1.05103508e-01 -3.65042984e-01 -1.04943359e+00 1.29930601e-01 3.29865843e-01 3.55640411e-01 7.19880521e-01 9.93574932e-02 -6.20730221e-01 6.70120120e-01 -1.70524135e-01 1.21568087e-02 3.09891015e-01 6.08991146e-01 -2.89141089e-01 -7.99288273e-01 -6.21509910e-01 7.88811028e-01 -6.18321538e-01 -3.16388220e-01 2.52530277e-01 7.69134164e-01 -3.19121867e-01 1.14976621e+00 1.69008419e-01 1.67448118e-01 1.45241097e-01 5.55813387e-02 1.18525624e-01 -2.80224591e-01 -4.69469041e-01 6.68432117e-01 2.55737275e-01 -7.51846552e-01 -5.56187451e-01 -1.02024543e+00 -9.42733109e-01 -4.41609740e-01 -2.80411363e-01 2.19536543e-01 4.79167700e-01 8.21707726e-01 4.31041598e-01 4.40483689e-01 6.60347581e-01 -4.93361324e-01 -7.42315769e-01 -6.90107644e-01 -8.16076756e-01 1.62789047e-01 2.77060330e-01 -9.03025746e-01 -6.59928381e-01 2.51238555e-01]
[7.734426975250244, 5.118396759033203]
ac636c3c-14ea-462b-aec3-ea378eb73c8c
autonomous-aerial-cinematography-in
1910.06988
null
https://arxiv.org/abs/1910.06988v1
https://arxiv.org/pdf/1910.06988v1.pdf
Autonomous Aerial Cinematography In Unstructured Environments With Learned Artistic Decision-Making
Aerial cinematography is revolutionizing industries that require live and dynamic camera viewpoints such as entertainment, sports, and security. However, safely piloting a drone while filming a moving target in the presence of obstacles is immensely taxing, often requiring multiple expert human operators. Hence, there is demand for an autonomous cinematographer that can reason about both geometry and scene context in real-time. Existing approaches do not address all aspects of this problem; they either require high-precision motion-capture systems or GPS tags to localize targets, rely on prior maps of the environment, plan for short time horizons, or only follow artistic guidelines specified before flight. In this work, we address the problem in its entirety and propose a complete system for real-time aerial cinematography that for the first time combines: (1) vision-based target estimation; (2) 3D signed-distance mapping for occlusion estimation; (3) efficient trajectory optimization for long time-horizon camera motion; and (4) learning-based artistic shot selection. We extensively evaluate our system both in simulation and in field experiments by filming dynamic targets moving through unstructured environments. Our results indicate that our system can operate reliably in the real world without restrictive assumptions. We also provide in-depth analysis and discussions for each module, with the hope that our design tradeoffs can generalize to other related applications. Videos of the complete system can be found at: https://youtu.be/ookhHnqmlaU.
['Sebastian Scherer', 'Erdal Kayacan', 'Cherie Ho', 'Wenshan Wang', 'Sanjiban Choudhury', 'Mirko Gschwindt', 'Aayush Ahuja', 'Rogerio Bonatti', 'Efe Camci']
2019-10-15
null
null
null
null
['occlusion-estimation']
['computer-vision']
[ 3.42705220e-01 -2.57817864e-01 -6.22555837e-02 6.97748214e-02 -4.90325153e-01 -1.05555511e+00 3.62670064e-01 -1.65974889e-02 -1.83051199e-01 5.25359988e-01 -2.85835624e-01 -3.84934813e-01 -2.12537169e-01 -6.11179054e-01 -4.62321609e-01 -4.43901300e-01 -3.23337048e-01 2.95186162e-01 6.73551917e-01 -3.82167369e-01 4.12163854e-01 7.21641183e-01 -1.45519781e+00 -3.54653716e-01 5.14132380e-01 1.10278010e+00 2.90759265e-01 1.05822957e+00 5.46557784e-01 9.41172123e-01 -5.52077174e-01 -2.04513371e-01 6.79381907e-01 -3.22051615e-01 -2.58630127e-01 4.73478943e-01 4.00462747e-01 -5.93874693e-01 -4.35171217e-01 8.78848255e-01 2.44606405e-01 4.27984238e-01 2.76591539e-01 -1.25943136e+00 8.18833113e-02 -2.39397600e-01 -5.96821249e-01 3.83152366e-01 4.87072855e-01 4.50090557e-01 5.47977865e-01 -5.36622882e-01 5.10465741e-01 7.42358685e-01 5.27721584e-01 7.99889565e-02 -7.44503558e-01 -4.56303269e-01 1.89386636e-01 2.23606408e-01 -1.42233360e+00 -5.01486123e-01 8.46697688e-01 -5.51620901e-01 5.09950399e-01 4.47817981e-01 8.55130613e-01 6.98039770e-01 4.01217073e-01 3.14720064e-01 6.57389760e-01 -1.96209118e-01 2.86500365e-01 -1.26856761e-02 -5.86427331e-01 8.59369099e-01 1.31937921e-01 3.49240422e-01 -4.25767571e-01 -1.38644546e-01 9.49393988e-01 8.89770165e-02 -5.87309003e-01 -7.60042608e-01 -1.32944167e+00 7.68426061e-01 1.08664460e-01 -2.43170913e-02 -3.02232981e-01 2.50790626e-01 8.89741257e-02 2.75040179e-01 2.87067741e-01 6.86886787e-01 3.73526476e-02 -4.67420310e-01 -1.03233445e+00 2.55672067e-01 7.03788817e-01 1.05330825e+00 5.20009577e-01 3.96081239e-01 2.73289502e-01 2.47793674e-01 -1.30837306e-01 6.30092323e-01 -9.57602262e-02 -1.56314611e+00 3.87520075e-01 1.83935180e-01 6.89080000e-01 -1.46008277e+00 -1.80765003e-01 -2.68026352e-01 -3.41788530e-01 5.35778642e-01 3.15851390e-01 -4.87125129e-01 -4.19674397e-01 1.28832626e+00 6.03828669e-01 4.47396070e-01 -1.64548412e-01 1.33318770e+00 1.25257656e-01 7.51936138e-01 -5.11071563e-01 -2.97610015e-01 9.82408106e-01 -8.15108657e-01 -6.11347556e-01 -4.76367086e-01 3.30168039e-01 -9.16570187e-01 7.98033118e-01 4.49500412e-01 -9.48894501e-01 -4.71659273e-01 -1.18460476e+00 3.71321470e-01 3.04219835e-02 1.68941796e-01 3.94076258e-01 5.64714074e-01 -7.41984010e-01 4.92687553e-01 -9.37661171e-01 -3.84853989e-01 -1.21086515e-01 2.97815681e-01 -1.88290060e-01 -8.99109319e-02 -9.70437944e-01 8.05467367e-01 1.42702935e-02 1.20096281e-01 -1.12131250e+00 -5.25792837e-01 -9.40778196e-01 -1.34613186e-01 8.96429956e-01 -5.13188004e-01 1.22252047e+00 -9.86525416e-01 -1.58338332e+00 4.75591004e-01 8.70871544e-02 -3.58423144e-01 5.63017011e-01 -3.57482374e-01 -3.09507877e-01 4.55868036e-01 -2.36648209e-02 3.91361177e-01 7.42681563e-01 -1.28853369e+00 -9.49913979e-01 -1.82158172e-01 6.62360728e-01 6.79037988e-01 -4.33730297e-02 -3.11700284e-01 -4.77236092e-01 -5.64326227e-01 6.64465735e-03 -1.10417914e+00 -3.40953112e-01 2.37345055e-01 -6.09435774e-02 6.60101652e-01 1.00248373e+00 -4.54615146e-01 1.11829984e+00 -2.01490569e+00 2.09818140e-01 -7.00468123e-02 2.53817290e-02 2.45333299e-01 1.92494825e-01 6.81726277e-01 3.82555574e-01 -3.69362414e-01 -1.41984925e-01 -1.23279445e-01 -2.93053299e-01 -8.13957825e-02 -6.36417091e-01 7.29141951e-01 -3.13375592e-01 4.77063775e-01 -1.06541550e+00 -4.86834615e-01 7.11605251e-01 2.16982141e-01 -3.71724159e-01 1.61590502e-01 -8.89683291e-02 4.35695320e-01 -4.31011707e-01 8.23128760e-01 3.00858200e-01 1.04100384e-01 1.28082350e-01 -9.28148478e-02 -5.13313115e-01 -2.37688228e-01 -1.36775804e+00 1.57156861e+00 -4.23004448e-01 9.11837697e-01 5.00382543e-01 -6.45509601e-01 7.88403809e-01 1.70654058e-01 6.92949533e-01 -3.43379855e-01 2.60224342e-01 -6.52994663e-02 -3.07698756e-01 -5.19512534e-01 9.01809216e-01 -5.63570075e-02 -2.18503982e-01 2.71643519e-01 -4.06624824e-01 -5.05483806e-01 1.74543902e-01 1.86766818e-01 1.29668713e+00 1.27177030e-01 1.62905782e-01 1.13102660e-01 2.43358582e-01 5.60631812e-01 5.69236040e-01 4.86643940e-01 -3.44010234e-01 5.62244117e-01 6.83276728e-02 -5.82903743e-01 -8.97041857e-01 -9.49776530e-01 3.49444717e-01 5.48088908e-01 9.15868342e-01 -2.81357020e-01 -5.12922227e-01 -3.21211725e-01 -2.25978956e-01 7.76052535e-01 -2.67280668e-01 -3.31138447e-02 -5.62756896e-01 -1.20669849e-01 1.40918195e-01 2.11868629e-01 3.78083676e-01 -5.46004176e-01 -1.50725317e+00 2.12392569e-01 -2.52939284e-01 -1.32034779e+00 -6.54560864e-01 -1.67160362e-01 -6.74912691e-01 -1.21734047e+00 -5.46970010e-01 -3.27094615e-01 5.31083345e-01 9.61592376e-01 6.87482059e-01 8.06148909e-03 -3.15325499e-01 8.76320541e-01 -3.92908037e-01 -1.01070426e-01 -1.07093982e-01 -3.73037755e-01 2.52777040e-01 -3.04233618e-02 -1.91771135e-01 -4.90334541e-01 -5.93933582e-01 7.73424447e-01 -6.64304197e-01 2.12410763e-01 2.88647801e-01 4.17228758e-01 5.18807292e-01 4.13721144e-01 -1.40135899e-01 -3.66387039e-01 2.44630635e-01 -2.89897114e-01 -1.07269263e+00 1.95025072e-01 -7.47621730e-02 -7.20122159e-01 5.44625640e-01 -4.97193158e-01 -7.66240120e-01 4.03798550e-01 3.30450028e-01 -1.03152776e+00 -6.35940433e-02 3.68410110e-01 2.42214724e-02 -4.01053309e-01 5.20239234e-01 1.84876397e-01 -2.47104526e-01 1.32188946e-01 1.57647431e-01 4.38624024e-01 6.19457424e-01 -2.30788738e-01 1.08576739e+00 7.55120933e-01 1.86582550e-01 -1.10864532e+00 -6.83817923e-01 -4.41269815e-01 -6.63733780e-01 -8.19562972e-01 6.96482301e-01 -9.04608727e-01 -7.40423262e-01 1.50237098e-01 -9.80675101e-01 -5.92487633e-01 -2.11385518e-01 7.06776202e-01 -7.14487791e-01 5.77564120e-01 -2.71882325e-01 -9.79771972e-01 1.19189799e-01 -1.09272969e+00 1.03892803e+00 2.21531481e-01 -1.17494434e-01 -1.04839277e+00 1.79870263e-01 3.95465612e-01 1.93230093e-01 6.93469346e-01 9.84559879e-02 2.65833110e-01 -9.63604510e-01 -4.61260885e-01 1.97828755e-01 -1.72952369e-01 1.56893402e-01 -1.17806849e-04 -5.91204286e-01 -4.12837863e-01 1.68365240e-01 -4.41397987e-02 4.62435514e-01 5.01977563e-01 6.04586601e-01 -2.32091442e-01 -3.76434505e-01 7.50479341e-01 1.29410100e+00 5.58812499e-01 3.23662698e-01 2.23284811e-01 6.61289454e-01 6.76185548e-01 1.39182150e+00 7.13475704e-01 2.86859363e-01 1.14015973e+00 8.03205848e-01 -1.50881708e-02 2.65224099e-01 -3.33024830e-01 5.12988508e-01 3.67320418e-01 -2.11230293e-01 -5.71462989e-01 -7.48127878e-01 5.84906280e-01 -1.87394905e+00 -1.14974892e+00 5.26219346e-02 2.40751791e+00 1.10489782e-03 1.10737756e-01 1.14078350e-01 2.54970174e-02 7.58700252e-01 3.17629218e-01 -6.14520252e-01 -1.01032652e-01 3.26394469e-01 -5.10161400e-01 8.98657799e-01 7.36733794e-01 -1.20330262e+00 9.86223280e-01 5.44614601e+00 6.55793607e-01 -1.21850073e+00 -2.91260816e-02 3.13258469e-01 -3.61238599e-01 2.84489114e-02 3.73276442e-01 -5.38907766e-01 2.57981628e-01 5.31374216e-01 -2.40553036e-01 5.72876275e-01 9.31841731e-01 3.09619278e-01 -5.69441915e-01 -6.55811012e-01 1.10637593e+00 2.84315825e-01 -1.32208765e+00 -4.01009649e-01 1.97976738e-01 4.64520603e-01 -2.59954244e-01 3.98565866e-02 -1.50894150e-01 1.69632778e-01 -5.87804079e-01 1.03410268e+00 4.08306628e-01 6.99807823e-01 -8.32908213e-01 3.94405395e-01 4.91614372e-01 -1.37566519e+00 -1.54419407e-01 -3.96347284e-01 -2.85030693e-01 7.00300694e-01 3.05890262e-01 -7.06198812e-01 5.62976420e-01 4.15712357e-01 6.13199949e-01 -3.13257128e-01 1.33422482e+00 -2.08264500e-01 4.29475754e-01 -3.33434552e-01 1.08081155e-01 3.20770323e-01 -2.85431564e-01 9.96474743e-01 7.02367783e-01 7.84143090e-01 5.59260428e-01 6.09518826e-01 5.14657795e-01 6.23427212e-01 -3.09888601e-01 -9.51117218e-01 -1.29987337e-02 3.46660823e-01 1.31072366e+00 -1.06833196e+00 -3.81163992e-02 -1.40444726e-01 1.06420696e+00 -2.74361640e-01 3.04585993e-01 -1.20964670e+00 -4.36407626e-01 6.57798231e-01 5.28977036e-01 2.93413073e-01 -8.22503388e-01 4.19537984e-02 -1.12741780e+00 -7.66043663e-02 -5.17952025e-01 1.90805852e-01 -1.00785398e+00 -4.70808953e-01 8.16028059e-01 9.76201519e-02 -1.77739060e+00 -4.38230664e-01 -2.95745045e-01 -5.62513411e-01 3.68336499e-01 -1.11310637e+00 -9.55723822e-01 -4.79844064e-01 5.62893569e-01 8.66101146e-01 -7.37926736e-02 5.19682050e-01 1.68611839e-01 -3.82876158e-01 -1.43589944e-01 -2.91776747e-01 -1.99356124e-01 5.46829343e-01 -8.13820839e-01 3.04804355e-01 1.18804145e+00 2.58995980e-01 1.25390664e-01 1.01603353e+00 -8.28193128e-01 -1.66630805e+00 -9.66517687e-01 2.84419566e-01 -5.05154908e-01 6.33286238e-01 -3.47178698e-01 -3.46576065e-01 5.84668338e-01 5.89583302e-03 -1.34090453e-01 2.93128729e-01 -3.44440490e-01 4.06045079e-01 -1.03703201e-01 -8.97048831e-01 7.05067813e-01 8.84765625e-01 -7.68337771e-02 -5.71706109e-02 4.23499316e-01 5.75211167e-01 -8.70382965e-01 -5.29398978e-01 2.90004671e-01 6.17099702e-01 -1.19020915e+00 8.09788644e-01 1.31388884e-02 6.86278120e-02 -6.90561235e-01 -2.58420765e-01 -1.11141038e+00 -1.93300799e-01 -8.63214314e-01 3.22075151e-02 6.68851793e-01 1.98943596e-02 -2.72660226e-01 8.09537530e-01 5.88496745e-01 -2.84964532e-01 -6.27738535e-01 -9.17169571e-01 -8.40436995e-01 -7.53153682e-01 -6.13205075e-01 2.00188279e-01 7.98128843e-01 -6.92259297e-02 1.57163590e-01 -8.45243394e-01 7.07348645e-01 6.06543183e-01 3.10982496e-01 1.13900876e+00 -8.80309284e-01 -4.24794018e-01 -7.10369572e-02 -5.52343786e-01 -1.14076877e+00 -8.18721429e-02 -7.22538456e-02 1.58739179e-01 -1.37961853e+00 -4.05902505e-01 -3.34379196e-01 4.06056851e-01 -5.40466681e-02 2.59171724e-01 2.47705474e-01 3.42151999e-01 3.88784587e-01 -7.80248225e-01 5.39254844e-01 1.15249002e+00 1.49092421e-01 -2.81434149e-01 3.04167211e-01 -1.15376428e-01 9.44294393e-01 6.70982599e-01 -3.37120265e-01 -7.97940731e-01 -4.47016656e-01 1.24492630e-01 7.88332760e-01 5.40967882e-01 -1.57236648e+00 4.89373893e-01 -6.74387574e-01 2.04149365e-01 -4.07833725e-01 1.05176353e+00 -1.09062028e+00 4.10442978e-01 4.56241667e-01 1.58306941e-01 3.32320482e-01 2.78192669e-01 6.89936697e-01 -9.55773890e-02 -1.34903818e-01 5.85870087e-01 -2.02063695e-01 -1.01924145e+00 3.18555921e-01 -5.93426406e-01 -8.65548253e-02 1.49329317e+00 -6.09795332e-01 -2.49065027e-01 -9.97079372e-01 -5.52065969e-01 2.75157809e-01 1.00790417e+00 3.26882809e-01 7.86702693e-01 -1.02234983e+00 -1.94868907e-01 8.47673267e-02 -2.79761031e-02 -2.08391160e-01 5.13623953e-01 8.22666049e-01 -9.12288547e-01 3.57060403e-01 -1.67471439e-01 -5.49606800e-01 -1.23423660e+00 6.73932552e-01 3.09213310e-01 1.41907567e-02 -5.00221789e-01 4.30449456e-01 2.83817470e-01 1.36126503e-02 1.09788746e-01 1.61881503e-02 -3.37669030e-02 -2.59341747e-01 6.65964544e-01 4.76094484e-01 -2.97571719e-01 -8.78286242e-01 -2.00221181e-01 8.73110473e-01 4.30631429e-01 -5.54567516e-01 1.07127750e+00 -6.00059390e-01 3.97329688e-01 3.74917567e-01 5.12841284e-01 2.80470461e-01 -1.80634093e+00 1.87270746e-01 -3.84085983e-01 -9.71743882e-01 3.19018811e-01 -3.54173392e-01 -1.03721023e+00 8.11917007e-01 5.01362562e-01 1.12107024e-01 1.17286491e+00 -4.32996482e-01 7.73279428e-01 4.83443320e-01 8.22796404e-01 -9.57841158e-01 1.63637564e-01 2.77963340e-01 7.45430529e-01 -9.11833704e-01 1.92525640e-01 -5.39364994e-01 -1.02170372e+00 1.11739755e+00 6.02033973e-01 -2.56840825e-01 3.44579905e-01 2.92628050e-01 1.38622835e-01 -3.02686870e-01 -7.08461225e-01 -2.38972157e-01 1.58099197e-02 5.80423415e-01 -1.93212062e-01 -1.01570368e-01 2.78570324e-01 7.04832450e-02 -1.04767002e-01 -1.87757030e-01 9.28482234e-01 1.16889727e+00 -7.19517767e-01 -7.01923013e-01 -6.74771726e-01 -1.09669633e-01 -3.94945703e-02 2.13612214e-01 -2.95028597e-01 8.77886772e-01 1.16994426e-01 1.14211023e+00 1.02608623e-02 -6.07660294e-01 4.68738616e-01 -5.57626665e-01 4.76036698e-01 -5.73451161e-01 -1.26518682e-01 1.64341956e-01 2.37147719e-01 -7.06327975e-01 -3.25823605e-01 -6.86121702e-01 -7.96720266e-01 -4.78671551e-01 -2.35938430e-01 1.10721923e-01 5.80682158e-01 6.65016830e-01 4.02129680e-01 3.13431531e-01 6.99632287e-01 -1.24631763e+00 -1.34988725e-01 -4.73103344e-01 -6.15422606e-01 -1.95522979e-01 3.29643190e-01 -8.58607352e-01 -3.11086625e-01 2.19574764e-01]
[7.376079082489014, -1.450825572013855]
6d656383-f8ab-4012-9ec1-0f9b691a2853
learning-to-generate-realistic-lidar-point
2209.03954
null
https://arxiv.org/abs/2209.03954v2
https://arxiv.org/pdf/2209.03954v2.pdf
Learning to Generate Realistic LiDAR Point Clouds
We present LiDARGen, a novel, effective, and controllable generative model that produces realistic LiDAR point cloud sensory readings. Our method leverages the powerful score-matching energy-based model and formulates the point cloud generation process as a stochastic denoising process in the equirectangular view. This model allows us to sample diverse and high-quality point cloud samples with guaranteed physical feasibility and controllability. We validate the effectiveness of our method on the challenging KITTI-360 and NuScenes datasets. The quantitative and qualitative results show that our approach produces more realistic samples than other generative models. Furthermore, LiDARGen can sample point clouds conditioned on inputs without retraining. We demonstrate that our proposed generative model could be directly used to densify LiDAR point clouds. Our code is available at: https://www.zyrianov.org/lidargen/
['Shenlong Wang', 'Xiyue Zhu', 'Vlas Zyrianov']
2022-09-08
null
null
null
null
['point-cloud-generation']
['computer-vision']
[-1.43179491e-01 -3.09409499e-01 6.23394325e-02 -2.10619211e-01 -1.06595337e+00 -9.11470175e-01 6.85886323e-01 -3.86669546e-01 2.31024995e-01 6.79350734e-01 -2.94254124e-01 -2.47473251e-02 -9.93682370e-02 -1.27244020e+00 -9.74934220e-01 -6.36956155e-01 2.07390264e-01 9.11794722e-01 7.45219761e-04 -4.21265373e-03 1.28619865e-01 1.08905208e+00 -2.04367447e+00 -2.38153324e-01 1.36170185e+00 7.54677713e-01 5.01945436e-01 7.36661553e-01 1.42535167e-02 4.21263613e-02 -2.43280396e-01 -1.96048960e-01 5.65007269e-01 9.10828486e-02 -6.79370239e-02 1.47642076e-01 7.37956226e-01 -3.69616598e-01 -7.63492659e-02 9.99715090e-01 5.66741586e-01 1.60887301e-01 8.72092247e-01 -1.25429952e+00 -5.83730519e-01 -2.28946796e-03 -4.54947948e-01 -3.71474206e-01 5.00567444e-02 6.44776642e-01 8.60534489e-01 -1.42307842e+00 6.74668908e-01 1.26048648e+00 6.37298584e-01 2.41332918e-01 -1.52400470e+00 -1.03050303e+00 -1.35851964e-01 -3.00167799e-01 -1.57535458e+00 -3.82467896e-01 8.48558962e-01 -5.74144363e-01 5.22997379e-01 2.96166390e-01 1.01245010e+00 1.21146119e+00 1.28057256e-01 6.03500009e-01 8.78833115e-01 2.60250270e-02 4.68954772e-01 -1.42364353e-01 -2.27511197e-01 5.11996329e-01 4.88966137e-01 7.72212505e-01 -6.30616724e-01 -4.24921334e-01 9.05030012e-01 2.05303833e-01 -1.20452642e-01 -6.92632914e-01 -9.12099719e-01 8.84376943e-01 4.93689358e-01 -5.24994850e-01 -4.58990425e-01 6.62704110e-01 -3.53498697e-01 -1.98848411e-01 6.31139934e-01 1.47507712e-01 1.18303575e-01 -1.85198590e-01 -1.21283162e+00 5.96577466e-01 4.85927224e-01 1.52845848e+00 9.24943089e-01 3.81462723e-01 1.14509471e-01 5.36553323e-01 5.62848926e-01 1.62917483e+00 -5.30553997e-01 -1.54396641e+00 3.02904040e-01 1.91351503e-01 3.93596262e-01 -6.19952142e-01 2.13273272e-01 -4.04435426e-01 -7.10093796e-01 6.76770091e-01 -2.23081306e-01 -1.79233905e-02 -1.08216774e+00 1.33309126e+00 3.69799823e-01 5.27282894e-01 4.69451211e-02 8.12846899e-01 6.64480269e-01 7.29856670e-01 -1.92678437e-01 1.41544253e-01 6.94211662e-01 -3.13501656e-01 -3.46421748e-01 6.02197647e-03 -1.86862618e-01 -7.52769232e-01 1.21921873e+00 4.50075001e-01 -1.31246233e+00 -4.41476792e-01 -9.76263821e-01 -1.50839649e-02 9.10163224e-02 -3.91697884e-03 6.79221213e-01 4.82981682e-01 -9.75826383e-01 7.90722013e-01 -1.27960455e+00 -1.04574248e-01 6.38810694e-01 7.27017969e-02 1.71837285e-01 -1.90704584e-01 -4.79839981e-01 4.96347725e-01 -8.86880159e-02 9.23138559e-02 -1.21212935e+00 -1.01357245e+00 -7.44710028e-01 -3.47683616e-02 1.43470436e-01 -1.10000527e+00 1.33439052e+00 -1.31763816e-01 -1.47590232e+00 5.26371896e-01 -5.28299630e-01 -2.68729061e-01 5.53422928e-01 -1.91009283e-01 5.75727373e-02 1.98131427e-01 1.42426267e-01 9.74030972e-01 8.15594018e-01 -1.94646871e+00 -4.12172616e-01 -1.97364286e-01 -3.68353575e-01 1.26242965e-01 3.29237968e-01 -6.13366008e-01 -2.78872907e-01 -4.50640768e-01 1.99970245e-01 -1.02897716e+00 -2.59219080e-01 3.44850928e-01 -4.57908362e-01 1.71465769e-01 8.19124222e-01 -9.79087036e-03 4.50428396e-01 -1.99756074e+00 7.65433684e-02 4.22448486e-01 1.62960693e-01 -1.79215282e-01 -1.36737153e-01 5.89253366e-01 2.65589625e-01 3.90273601e-01 -4.54113036e-01 -6.22083366e-01 1.41760945e-01 4.05412287e-01 -8.12895000e-01 2.54367322e-01 4.47384626e-01 1.04404402e+00 -9.94095027e-01 -3.06639045e-01 7.21087635e-01 7.49930024e-01 -6.03315234e-01 4.16824259e-02 -4.63660747e-01 5.61403453e-01 -4.69403833e-01 1.18299663e+00 1.10176468e+00 2.54776999e-02 -2.88175344e-01 -9.26174819e-02 -1.28878295e-01 -1.10149272e-01 -1.09395504e+00 1.70998263e+00 -5.25133133e-01 4.16005701e-01 2.99499899e-01 8.77576298e-04 1.01713848e+00 -1.74468368e-01 5.54763973e-01 -1.34513855e-01 -7.68149048e-02 2.94893295e-01 -4.70933408e-01 -1.02313757e-01 7.93217123e-01 -3.56906176e-01 1.62252575e-01 2.26506159e-01 -3.05332541e-02 -1.23782957e+00 -1.15926050e-01 2.30195358e-01 6.34638906e-01 4.78444099e-01 -1.88857019e-01 -2.28377983e-01 -1.82404399e-01 1.81523100e-01 4.25932437e-01 6.68524802e-01 2.88308978e-01 1.01763904e+00 -1.90640181e-01 1.40478015e-01 -1.14181578e+00 -2.07030678e+00 -2.90100962e-01 1.03492616e-02 2.76866138e-01 -3.03650200e-01 -3.46330881e-01 -2.17243701e-01 5.96430957e-01 1.11090469e+00 -1.86006740e-01 1.24456349e-03 -1.90383643e-01 -3.34082186e-01 3.21028620e-01 5.42810619e-01 1.35150015e-01 -6.78035200e-01 -3.37043136e-01 -9.97520760e-02 5.11836112e-02 -7.59623468e-01 -1.47187889e-01 -6.54314160e-02 -1.22009850e+00 -8.50087762e-01 -2.11992458e-01 -1.68546334e-01 6.10184133e-01 5.28906047e-01 1.34660375e+00 -8.88793916e-02 -1.88032240e-01 7.16300488e-01 4.35666367e-03 -7.32505858e-01 -3.32962602e-01 -5.74546568e-02 1.85624659e-01 -3.69392037e-01 1.48348600e-01 -1.06900764e+00 -5.88601112e-01 1.65603012e-01 -8.35903823e-01 4.75894623e-02 2.47061700e-01 4.33008403e-01 1.25419283e+00 -1.30110174e-01 2.47509837e-01 -4.54965174e-01 4.51752663e-01 -5.37743390e-01 -1.16476285e+00 -2.27401644e-01 -5.78148782e-01 -4.06769738e-02 2.39844292e-01 -1.82731912e-01 -8.11920524e-01 3.63296002e-01 -1.43171251e-01 -1.15446925e+00 -8.00454468e-02 1.81431577e-01 -3.76820087e-01 -2.23173201e-01 5.79582274e-01 2.16059536e-01 3.25415051e-03 -4.01623696e-01 6.81248128e-01 3.13597798e-01 7.80083835e-01 -1.15770936e+00 1.60649025e+00 1.00297916e+00 3.17253232e-01 -1.04326797e+00 -2.54964858e-01 -3.21870506e-01 -5.56420624e-01 -5.44133246e-01 5.84826469e-01 -1.07926381e+00 -6.89641714e-01 3.38871181e-01 -1.11060452e+00 -4.11986023e-01 -7.14467824e-01 3.19840521e-01 -7.56027043e-01 2.16248408e-01 -1.94470286e-01 -1.15950775e+00 -2.74469763e-01 -1.10359418e+00 1.83094001e+00 1.33137673e-01 1.38487697e-01 -7.91496038e-01 1.79612964e-01 5.86905330e-02 2.01706722e-01 5.75622320e-01 3.77492189e-01 2.87145764e-01 -1.59314573e+00 -8.79939198e-02 4.67844792e-02 2.43285060e-01 7.85034373e-02 8.82344842e-01 -1.10644376e+00 -1.36183172e-01 -1.12910345e-01 -3.50374401e-01 8.01156223e-01 4.37703550e-01 1.15451145e+00 1.41327724e-01 -3.57755989e-01 1.01224875e+00 1.58123255e+00 -1.14120856e-01 7.16768622e-01 -2.45543584e-01 6.72641456e-01 2.63590723e-01 6.53733015e-01 4.45929199e-01 3.63493830e-01 5.24963200e-01 8.75871301e-01 2.90555842e-02 -3.16193588e-02 -6.60101354e-01 1.60836563e-01 8.52939665e-01 -2.54012227e-01 -3.87589276e-01 -9.82641637e-01 5.39935172e-01 -1.56447363e+00 -1.17761862e+00 -4.19231683e-01 2.17229962e+00 5.02019823e-01 -1.43960655e-01 -2.92708039e-01 -3.14593166e-01 4.92985994e-01 2.75736339e-02 -7.28297114e-01 -1.73724946e-02 -5.59036434e-02 7.43021667e-01 7.08112180e-01 7.19835758e-01 -7.18227148e-01 9.49538648e-01 6.64921618e+00 8.65166843e-01 -9.20693934e-01 1.05418697e-01 9.56683308e-02 -5.18583000e-01 -9.95740414e-01 5.65809384e-02 -8.91284645e-01 4.53310728e-01 8.70239973e-01 -4.41162109e-01 5.32017589e-01 8.53034377e-01 5.89685738e-01 -1.03068717e-01 -9.28346157e-01 9.61039603e-01 -3.53098154e-01 -1.71980035e+00 1.29097745e-01 4.23085660e-01 7.44603813e-01 4.58477050e-01 3.45022947e-01 3.16515937e-02 9.15704072e-01 -9.86007333e-01 9.68765438e-01 9.42498565e-01 1.07221675e+00 -9.53138828e-01 7.70706832e-02 4.87854749e-01 -1.30756879e+00 4.39719528e-01 -5.48919797e-01 1.73140034e-01 5.41862428e-01 9.25129354e-01 -7.23101616e-01 9.47079360e-01 5.99874139e-01 8.52797687e-01 -3.42993319e-01 1.06715441e+00 -3.53818357e-01 7.33400404e-01 -7.88542271e-01 2.36709118e-01 -1.26420811e-01 -8.25174332e-01 8.52782130e-01 7.27021217e-01 9.21216428e-01 -2.45201513e-02 2.86541969e-01 1.58673835e+00 -1.21300273e-01 -2.90206134e-01 -1.03497374e+00 8.53557661e-02 9.28019226e-01 1.18404996e+00 -3.77130091e-01 -2.19603240e-01 1.89720288e-01 5.54752827e-01 9.33978558e-02 5.46302199e-01 -1.00314319e+00 -2.18102057e-02 1.10840201e+00 2.34791726e-01 3.22357446e-01 -5.83612919e-01 -7.28011012e-01 -1.27824879e+00 1.18458673e-01 -4.29961443e-01 -3.82476419e-01 -1.35325181e+00 -1.54034829e+00 3.05684835e-01 3.52542520e-01 -1.58627141e+00 -3.43682617e-01 -4.75393087e-01 -7.50025094e-01 1.09426928e+00 -1.45134008e+00 -1.35254323e+00 -7.50282347e-01 3.20398569e-01 3.03229898e-01 1.21141411e-01 6.60758018e-01 -5.14611090e-03 -1.87724322e-01 -1.14084333e-01 2.75473893e-01 -5.22941291e-01 3.64724576e-01 -1.28528559e+00 7.81540692e-01 9.91780937e-01 1.06560215e-01 5.96636653e-01 7.68195510e-01 -9.25193369e-01 -1.50200677e+00 -1.51709580e+00 8.88555869e-02 -7.99972594e-01 5.27680278e-01 -5.45851409e-01 -7.08384633e-01 6.01157606e-01 1.10524125e-01 5.20510189e-02 2.82239407e-01 -3.49615306e-01 -6.56867446e-03 -1.74916103e-01 -1.42926180e+00 5.66360116e-01 1.41334057e+00 -4.50034142e-01 -3.77613902e-01 4.80959833e-01 6.94057465e-01 -6.43037617e-01 -8.60693812e-01 5.74423313e-01 3.31665665e-01 -9.87311780e-01 1.28901160e+00 -7.09266514e-02 4.64141130e-01 -5.38304925e-01 -4.34174597e-01 -1.42892933e+00 -1.19138367e-01 -5.35050750e-01 -2.62569904e-01 1.17765391e+00 1.88016877e-01 -7.81225979e-01 8.84616435e-01 3.28891546e-01 -3.52376997e-01 -5.24273157e-01 -9.44545686e-01 -1.14868319e+00 3.29997689e-01 -8.20657134e-01 8.94421458e-01 5.33908904e-01 -8.75770509e-01 -4.10451293e-02 -3.32489349e-02 6.90188587e-01 1.17207313e+00 5.26005685e-01 1.14823985e+00 -1.47814655e+00 -1.68590561e-01 -1.99270621e-01 -1.49270505e-01 -9.44548011e-01 1.95342034e-01 -1.07791436e+00 1.40366569e-01 -1.60503769e+00 -1.45642757e-01 -7.90565729e-01 5.33880055e-01 1.17439009e-01 -1.47774850e-03 2.24021807e-01 3.52087975e-01 3.62309456e-01 1.79485902e-01 1.04795980e+00 1.20611262e+00 -1.04499727e-01 -5.89325055e-02 1.53847024e-01 -4.12254333e-01 5.33097506e-01 9.14808095e-01 -4.70982552e-01 -7.28362799e-01 -4.90809858e-01 1.65924862e-01 -2.34345585e-01 8.15326214e-01 -1.16207469e+00 -1.48077458e-01 -6.24759734e-01 3.42331827e-01 -1.22854078e+00 8.74077082e-01 -8.06105077e-01 8.14658523e-01 2.34536976e-01 3.85732144e-01 7.18198568e-02 3.23024482e-01 5.67666411e-01 6.35574460e-02 -4.20701355e-02 7.86482811e-01 4.22338881e-02 -2.07021102e-01 7.82763243e-01 -7.54752904e-02 -9.38622057e-02 9.80253696e-01 -2.08005995e-01 -2.84876585e-01 -4.43983227e-01 -4.86835510e-01 4.06129897e-01 1.25340497e+00 7.86452070e-02 9.64109123e-01 -1.60013592e+00 -7.36359894e-01 2.61552691e-01 1.61090091e-01 6.52359366e-01 1.30400220e-02 3.51175368e-01 -7.27759063e-01 -9.25747529e-02 1.91698834e-01 -1.25911248e+00 -9.01688695e-01 3.12263876e-01 4.72585797e-01 1.86498880e-01 -6.79138184e-01 5.62593818e-01 5.25759300e-03 -8.65685582e-01 -2.48439372e-01 -6.97034478e-01 6.87419116e-01 -2.54440784e-01 3.61141898e-02 4.09215480e-01 -4.62467410e-02 -3.82835925e-01 -2.02657387e-01 6.99887335e-01 6.48157775e-01 -4.33680922e-01 1.44377744e+00 6.62473813e-02 1.53546065e-01 5.82983851e-01 7.04740942e-01 4.54010725e-01 -1.62308717e+00 2.21933216e-01 -5.65597951e-01 -8.40534627e-01 4.57658619e-02 -4.46149886e-01 -1.10249698e+00 9.77873683e-01 4.70211148e-01 -1.52713463e-01 7.59363651e-01 -1.67853944e-02 7.05670714e-01 2.41916612e-01 8.69457304e-01 -6.29068673e-01 -2.28577405e-01 2.94939488e-01 1.05908954e+00 -9.73964751e-01 7.57163167e-02 -7.05364525e-01 -4.71491933e-01 8.42958808e-01 3.77300054e-01 -4.48933184e-01 5.78699708e-01 6.07380509e-01 -6.27837479e-02 -4.51568156e-01 -8.22286308e-01 -2.03248426e-01 1.54746160e-01 9.07830238e-01 -2.05224901e-01 2.77814895e-01 4.66185570e-01 1.32247940e-01 -5.77397645e-01 1.69223323e-01 5.51556051e-01 9.37690973e-01 -5.43167651e-01 -1.11232698e+00 -6.75187349e-01 4.50476885e-01 4.60982651e-01 -1.61925957e-01 -2.09824950e-01 7.56455958e-01 8.17393363e-02 7.62838602e-01 2.64646053e-01 -3.96588892e-01 3.73040706e-01 2.34994804e-03 5.99036396e-01 -7.67926633e-01 -1.45812228e-01 1.69504046e-01 -1.66612998e-01 -7.94190884e-01 -2.83676237e-01 -9.90661442e-01 -1.27533281e+00 -7.25138187e-01 -2.61337399e-01 2.22535920e-03 8.05410981e-01 2.40237400e-01 7.25317419e-01 2.73031265e-01 7.94169962e-01 -1.62510872e+00 -5.76528609e-01 -6.67227030e-01 -4.47064191e-01 1.41590796e-02 1.89097181e-01 -1.02507961e+00 -4.88412142e-01 -3.36998962e-02]
[8.879133224487305, -3.656184434890747]
697f82ec-5e2e-4c96-84b4-12ef254856f5
mobilesal-extremely-efficient-rgb-d-salient
2012.13095
null
https://arxiv.org/abs/2012.13095v3
https://arxiv.org/pdf/2012.13095v3.pdf
MobileSal: Extremely Efficient RGB-D Salient Object Detection
The high computational cost of neural networks has prevented recent successes in RGB-D salient object detection (SOD) from benefiting real-world applications. Hence, this paper introduces a novel network, MobileSal, which focuses on efficient RGB-D SOD using mobile networks for deep feature extraction. However, mobile networks are less powerful in feature representation than cumbersome networks. To this end, we observe that the depth information of color images can strengthen the feature representation related to SOD if leveraged properly. Therefore, we propose an implicit depth restoration (IDR) technique to strengthen the mobile networks' feature representation capability for RGB-D SOD. IDR is only adopted in the training phase and is omitted during testing, so it is computationally free. Besides, we propose compact pyramid refinement (CPR) for efficient multi-level feature aggregation to derive salient objects with clear boundaries. With IDR and CPR incorporated, MobileSal performs favorably against state-of-the-art methods on six challenging RGB-D SOD datasets with much faster speed (450fps for the input size of 320 $\times$ 320) and fewer parameters (6.5M). The code is released at https://mmcheng.net/mobilesal.
['Yu-Chao Gu', 'Ming-Ming Cheng', 'Jia-Wang Bian', 'Jun Xu', 'Yun Liu', 'Yu-Huan Wu']
2020-12-24
null
null
null
null
['rgb-d-salient-object-detection']
['computer-vision']
[ 1.75092071e-01 3.17982845e-02 -7.63993636e-02 -1.34746283e-01 -2.89683670e-01 -1.48102149e-01 1.32972524e-01 -1.40051916e-01 -5.86057544e-01 4.25335318e-01 7.20111579e-02 -3.22606683e-01 -6.27471209e-02 -8.91624928e-01 -7.19898641e-01 -7.42263258e-01 -9.50407311e-02 -4.40832525e-01 7.41100788e-01 -4.12010640e-01 1.58286378e-01 7.22221971e-01 -2.04450560e+00 1.84648246e-01 7.92603910e-01 1.33400989e+00 6.24529600e-01 5.37410796e-01 6.20807111e-02 9.19962049e-01 -3.67098331e-01 -2.62230009e-01 5.84349632e-01 -1.22531354e-02 -5.86310208e-01 -4.72389087e-02 5.03828466e-01 -6.58786595e-01 -6.28680229e-01 1.05190957e+00 7.40987182e-01 1.87898129e-01 2.26491839e-01 -1.34892738e+00 -6.45883560e-01 3.15369844e-01 -9.65798616e-01 3.85070801e-01 -7.14811534e-02 6.10506274e-02 8.88834417e-01 -1.08661854e+00 5.00438929e-01 1.01523376e+00 6.24568820e-01 4.15683717e-01 -7.21050203e-01 -5.01591206e-01 1.26851842e-01 4.09294873e-01 -1.30135965e+00 -3.31488818e-01 1.22226179e+00 1.78869978e-01 9.13246989e-01 2.50371963e-01 9.44158375e-01 5.58692038e-01 -1.25316828e-01 1.14008987e+00 7.55187750e-01 -3.09423804e-01 1.32938519e-01 -1.90927073e-01 -8.59985799e-02 1.10931945e+00 3.96383345e-01 -1.07716255e-01 -9.04054761e-01 2.61752278e-01 1.02719808e+00 4.43724662e-01 -3.02199632e-01 -4.94388849e-01 -8.57298732e-01 7.66657650e-01 1.02146041e+00 1.50190949e-01 -5.20755827e-01 2.95623481e-01 1.69326365e-01 3.39197479e-02 4.79424298e-01 4.63114865e-02 -3.95944893e-01 -1.02812275e-01 -8.48926783e-01 2.07662776e-01 1.00709841e-01 9.41471994e-01 8.31301272e-01 2.49647290e-01 1.34254605e-01 7.17712045e-01 1.84914485e-01 6.00915730e-01 3.46522063e-01 -1.19888270e+00 4.98396337e-01 9.46447134e-01 1.58412680e-01 -1.29863322e+00 -6.17370963e-01 -5.46722651e-01 -9.41536963e-01 4.00923073e-01 3.26659858e-01 -8.13409593e-03 -9.63981092e-01 1.31295896e+00 5.33376336e-01 -2.27914713e-02 7.47713596e-02 1.19510782e+00 1.14364135e+00 5.20425260e-01 -2.82206774e-01 1.63644344e-01 1.26582456e+00 -9.65555668e-01 -3.70698661e-01 -3.24427903e-01 6.06503606e-01 -5.10127008e-01 1.27623284e+00 3.06188554e-01 -1.30761623e+00 -5.36081910e-01 -1.19691002e+00 -5.59693456e-01 -3.54522228e-01 2.84493804e-01 9.69812691e-01 6.26341105e-01 -1.19754064e+00 5.86778581e-01 -9.35926616e-01 -1.11035313e-02 8.58910978e-01 6.03678226e-01 -1.49061128e-01 -2.34274954e-01 -1.05105793e+00 5.51598012e-01 1.27600208e-01 3.31879318e-01 -4.74201441e-01 -6.79054737e-01 -8.69936883e-01 -3.69782820e-02 3.61788839e-01 -5.00014901e-01 1.10022807e+00 -7.97809005e-01 -1.27923465e+00 8.18247199e-01 -1.35511130e-01 -4.98522729e-01 4.71199661e-01 -3.54497075e-01 -3.96150351e-02 6.87154770e-01 -2.60027759e-02 9.90338564e-01 1.08490109e+00 -1.24713421e+00 -8.70200157e-01 -4.02140677e-01 3.77338886e-01 3.44262838e-01 -5.95241249e-01 -1.16715625e-01 -5.98067582e-01 -5.60173571e-01 5.01846015e-01 -6.57584369e-01 -1.45291924e-01 4.95035291e-01 -3.54706526e-01 -1.10584997e-01 1.07838750e+00 -6.27212703e-01 9.28309858e-01 -2.28912330e+00 -1.21283777e-01 1.03179440e-02 4.94981050e-01 5.40520251e-01 -3.08852214e-02 -1.39571816e-01 2.80213714e-01 -8.57534185e-02 -2.59193540e-01 -5.97584963e-01 2.04335637e-02 1.76315099e-01 -1.96088701e-01 5.78794658e-01 2.48084724e-01 1.04947639e+00 -6.49303079e-01 -6.32213056e-01 4.15290862e-01 7.90349782e-01 -5.97798467e-01 -1.08331867e-01 2.36512408e-01 -8.79915804e-02 -5.11657834e-01 1.17197287e+00 9.98470366e-01 -3.24157625e-01 -3.90345573e-01 -4.13671017e-01 -3.11136931e-01 8.32686052e-02 -1.10845244e+00 1.68259978e+00 -2.30852604e-01 7.38340378e-01 1.20207623e-01 -8.97166312e-01 9.50203657e-01 -2.26766557e-01 4.86321807e-01 -9.70801294e-01 2.93684900e-01 2.35338926e-01 -2.75576711e-01 -3.13973337e-01 8.04303527e-01 2.69358754e-01 1.76456168e-01 2.78483510e-01 -2.11456090e-01 -7.48140439e-02 6.23127408e-02 2.04594463e-01 9.56463933e-01 9.41390097e-02 4.88703363e-02 -7.78217092e-02 3.02342117e-01 -4.64650057e-03 6.52883410e-01 7.27976918e-01 -4.68219429e-01 6.82619810e-01 2.45750219e-01 -6.29140317e-01 -7.45490849e-01 -1.09960675e+00 3.46685504e-03 1.10326922e+00 7.48428226e-01 -2.56132603e-01 -5.49930096e-01 -5.93929231e-01 1.68234557e-02 2.90228911e-02 -6.06146753e-01 -1.89192981e-01 -6.87754393e-01 -6.87776327e-01 4.36711073e-01 7.73726761e-01 1.13989878e+00 -1.16346836e+00 -1.45470285e+00 1.09477885e-01 -2.28176370e-01 -1.09162295e+00 -1.55574620e-01 3.92606229e-01 -1.17692578e+00 -9.04555202e-01 -9.62038398e-01 -9.08615410e-01 6.96421802e-01 9.29840922e-01 8.80124569e-01 2.31495216e-01 -3.37526292e-01 1.51088521e-01 -4.66440111e-01 -4.94311422e-01 3.68860602e-01 1.59121186e-01 -1.04375474e-01 -4.07212317e-01 2.45997250e-01 -6.78642869e-01 -1.10237324e+00 2.22360358e-01 -9.74676609e-01 4.22058225e-01 8.26853096e-01 5.65617085e-01 5.34808278e-01 4.99631800e-02 3.47201794e-01 -2.02903375e-01 1.74086705e-01 -2.02812590e-02 -3.80915761e-01 -7.90565684e-02 -1.59993857e-01 -1.93231508e-01 4.47508216e-01 -3.13128084e-01 -9.31209028e-01 1.38793066e-01 -2.63361901e-01 -4.06037986e-01 -4.58009653e-02 2.11307541e-01 -5.11617102e-02 -4.43253189e-01 4.30555850e-01 3.29856366e-01 -4.91665006e-02 -4.94422883e-01 1.27434641e-01 5.65437019e-01 5.42539239e-01 -8.09923708e-02 8.67857337e-01 8.88543010e-01 9.43023786e-02 -9.40398574e-01 -8.81361306e-01 -3.61249298e-01 -5.20047903e-01 -1.78596318e-01 5.87195098e-01 -1.02052259e+00 -7.46673703e-01 8.03466320e-01 -9.61035788e-01 -4.85072076e-01 -2.85277545e-01 2.84366280e-01 -3.77027005e-01 2.29915887e-01 -5.89163423e-01 -7.00975060e-01 -6.27539575e-01 -9.45802093e-01 1.11835492e+00 5.79631686e-01 2.81555772e-01 -6.09231710e-01 -6.01935804e-01 2.50368506e-01 5.20431936e-01 3.11419487e-01 4.25077528e-01 2.27892011e-01 -8.19833398e-01 -3.55476816e-03 -7.89219081e-01 2.43361384e-01 1.54808596e-01 -1.17094018e-01 -1.09517884e+00 -1.76094472e-01 -6.91508725e-02 -2.20152617e-01 1.03732538e+00 6.03075981e-01 1.37337470e+00 -1.67543024e-01 -1.55973986e-01 9.16320562e-01 1.45218670e+00 -1.14407856e-02 5.90846956e-01 8.06992352e-01 9.47638929e-01 4.10584956e-01 7.91878939e-01 6.30203843e-01 5.25698543e-01 3.64481270e-01 9.22871828e-01 -7.91657567e-01 -2.39115253e-01 -6.12323657e-02 2.57845700e-01 4.70162451e-01 -3.74620765e-01 -7.91869909e-02 -8.42965603e-01 5.49918890e-01 -1.67304480e+00 -6.49658859e-01 -1.34524196e-01 1.66430140e+00 7.61197686e-01 2.72244602e-01 1.45820707e-01 4.89356160e-01 5.55668175e-01 3.88902158e-01 -6.94614470e-01 -7.79486224e-02 -4.37029839e-01 2.02218890e-01 6.19276404e-01 1.21284276e-01 -1.15892422e+00 8.44091833e-01 4.68849134e+00 7.74327457e-01 -1.22355950e+00 -1.23561649e-02 6.81033075e-01 -2.55683720e-01 -8.23979080e-02 -4.48450387e-01 -8.10852051e-01 2.83188224e-01 2.50052691e-01 2.35787526e-01 6.73866794e-02 1.07944250e+00 2.91510284e-01 -4.40310508e-01 -3.83708358e-01 1.23735082e+00 3.68672498e-02 -1.50091946e+00 -1.30479142e-01 -7.27311298e-02 6.52144492e-01 2.73493230e-01 3.24659616e-01 -1.37971872e-02 -1.17054202e-01 -5.36780477e-01 9.30270255e-01 1.52766958e-01 6.46414280e-01 -1.04540527e+00 8.21609974e-01 1.15696676e-01 -1.39721763e+00 -2.81069547e-01 -6.84417069e-01 -1.50871783e-01 1.45975109e-02 6.89147592e-01 -5.02885580e-01 4.54299271e-01 1.31673276e+00 9.36709464e-01 -7.70710051e-01 9.83405292e-01 -1.81351796e-01 2.21407950e-01 -4.88798469e-01 -8.77834260e-02 4.64931428e-01 1.97017729e-01 3.94632161e-01 9.48777437e-01 3.80446613e-01 8.22070763e-02 -1.21769488e-01 5.32900870e-01 -1.53393775e-01 -7.34716579e-02 -4.01100099e-01 3.65830123e-01 3.10533255e-01 1.23273337e+00 -1.14879835e+00 -1.80102006e-01 -3.07791322e-01 9.58290458e-01 2.13720575e-01 3.39880526e-01 -9.07975018e-01 -6.15994155e-01 6.48730993e-01 1.80582836e-01 7.86347270e-01 -3.16630691e-01 -3.27926129e-01 -9.55528498e-01 2.27216616e-01 -4.91921842e-01 1.54021308e-01 -9.60471272e-01 -7.99567878e-01 5.95895648e-01 -2.57852644e-01 -1.42000854e+00 3.11489552e-01 -6.82276070e-01 -4.20232713e-01 5.29308140e-01 -2.10606122e+00 -1.20162952e+00 -6.70502901e-01 8.90975833e-01 3.54402155e-01 2.87831366e-01 2.39666387e-01 2.71834671e-01 -4.22321796e-01 5.27245939e-01 -1.43846065e-01 1.53764635e-01 3.63822609e-01 -1.04204357e+00 4.32927072e-01 1.06245363e+00 -9.45636779e-02 4.45593715e-01 5.47932148e-01 -3.85319829e-01 -1.54974854e+00 -1.20823848e+00 5.55312574e-01 9.43609774e-02 4.19855237e-01 -3.16495985e-01 -7.13338256e-01 9.08858255e-02 -7.98918530e-02 4.82378215e-01 3.55628371e-01 -4.78690803e-01 -2.42107604e-02 -2.56004065e-01 -1.25619066e+00 6.59221411e-01 1.40493381e+00 -4.46110159e-01 -3.42815936e-01 -1.60364583e-01 9.99770641e-01 -5.38760006e-01 -6.84850991e-01 4.73949641e-01 5.04293263e-01 -1.30721450e+00 1.28300917e+00 1.08468354e-01 5.57820082e-01 -5.23406625e-01 -2.45794699e-01 -7.00381577e-01 -5.11357076e-02 -4.26209837e-01 -5.22667706e-01 8.93805146e-01 5.97531684e-02 -5.20205438e-01 1.10717940e+00 3.56725991e-01 -3.91802877e-01 -1.21964002e+00 -9.51075613e-01 -5.48582435e-01 -4.99440491e-01 -5.12110174e-01 5.81443906e-01 6.99807763e-01 -3.49041313e-01 -2.21226901e-01 -2.88123280e-01 2.33442485e-01 6.84479952e-01 4.09415990e-01 6.22176945e-01 -1.16540956e+00 6.88499585e-02 -3.61705065e-01 -5.38977921e-01 -1.21962273e+00 -3.93143952e-01 -3.83524776e-01 -9.13744122e-02 -1.62175965e+00 -1.06947392e-01 -6.23195946e-01 -2.94049412e-01 7.88849592e-01 -2.26700395e-01 8.46209884e-01 3.67238373e-01 5.48877493e-02 -7.32184350e-01 8.79739523e-01 1.52977169e+00 -6.72652200e-02 -4.73373771e-01 -1.12454429e-01 -8.83660078e-01 9.49534535e-01 1.20358098e+00 -3.40489477e-01 -4.17011440e-01 -4.86580044e-01 2.05555782e-01 -2.28559464e-01 7.18953490e-01 -1.17009199e+00 2.96266973e-01 1.10273659e-01 6.59133613e-01 -1.04267681e+00 6.07304215e-01 -6.11906171e-01 -4.82342273e-01 6.00641668e-01 1.36759430e-01 9.13696140e-02 3.61917436e-01 2.85844922e-01 -2.17875004e-01 4.86238394e-03 6.57433331e-01 -8.00932050e-02 -1.18944907e+00 4.76689249e-01 -1.93545431e-01 -2.45846882e-02 7.97979295e-01 -7.86468327e-01 -4.17313695e-01 -1.73358202e-01 -4.05949473e-01 -2.92576253e-02 5.52153349e-01 2.97814190e-01 1.13206112e+00 -1.15443611e+00 -2.53956974e-01 2.88276821e-01 -1.09194152e-01 4.64404553e-01 5.15653849e-01 9.39444065e-01 -6.65056586e-01 1.70149148e-01 -3.33222777e-01 -7.51124680e-01 -1.33483648e+00 3.01524520e-01 1.24482751e-01 6.54905364e-02 -9.37847614e-01 1.13683677e+00 1.94119662e-02 -1.11180127e-01 3.93581659e-01 -5.73349237e-01 -1.80560380e-01 6.73900247e-02 5.29421270e-01 3.98354203e-01 1.11568183e-01 -4.86986428e-01 -4.50189292e-01 6.15179479e-01 -7.90697113e-02 1.56547531e-01 1.68778753e+00 -3.54414552e-01 4.92612273e-02 4.49177176e-02 1.21248031e+00 -2.45754614e-01 -1.76513290e+00 -2.14121833e-01 -2.98152447e-01 -7.05897629e-01 4.58579451e-01 -2.85636276e-01 -1.38774860e+00 8.33443642e-01 8.99333358e-01 8.43444243e-02 1.58938134e+00 -7.41996318e-02 9.86663818e-01 5.64786017e-01 3.73298556e-01 -1.12730682e+00 3.48647445e-01 3.64442170e-01 7.59843171e-01 -1.31564939e+00 1.41459525e-01 -4.37748909e-01 -5.22940874e-01 9.45449173e-01 6.94193482e-01 -2.35312834e-01 5.03454626e-01 2.96247393e-01 -1.87049732e-02 -2.27220714e-01 -2.37707943e-01 -4.84743446e-01 1.82907179e-01 6.99106455e-01 -3.03744879e-02 -2.65848279e-01 2.24223942e-01 3.57488632e-01 -1.90537736e-01 -1.12669148e-01 4.93592769e-01 1.38204372e+00 -7.22669840e-01 -5.80893517e-01 -2.90932328e-01 3.76373947e-01 -4.59217757e-01 -2.52729326e-01 -7.83305839e-02 9.55350637e-01 3.55222374e-01 7.72727549e-01 9.26736463e-03 -4.49769318e-01 2.26521656e-01 -5.29600620e-01 3.59143078e-01 -1.43454239e-01 -3.65796536e-01 -9.84422043e-02 -1.42127603e-01 -8.55929673e-01 -8.61748278e-01 -4.70054477e-01 -1.40050638e+00 -4.18682426e-01 -2.51192331e-01 -2.23426133e-01 5.84804237e-01 6.78672433e-01 4.48854774e-01 5.36439002e-01 5.58998048e-01 -1.31737089e+00 6.66724285e-03 -5.67194641e-01 -5.30187964e-01 6.30388316e-03 6.05291784e-01 -8.97179961e-01 -3.36298794e-01 -5.96506782e-02]
[9.605786323547363, -0.813919186592102]
1b13073a-50a0-4688-bcdf-b1e5c7caa047
skin-lesion-analysis-toward-melanoma-2
1605.01397
null
http://arxiv.org/abs/1605.01397v1
http://arxiv.org/pdf/1605.01397v1.pdf
Skin Lesion Analysis toward Melanoma Detection: A Challenge at the International Symposium on Biomedical Imaging (ISBI) 2016, hosted by the International Skin Imaging Collaboration (ISIC)
In this article, we describe the design and implementation of a publicly accessible dermatology image analysis benchmark challenge. The goal of the challenge is to sup- port research and development of algorithms for automated diagnosis of melanoma, a lethal form of skin cancer, from dermoscopic images. The challenge was divided into sub-challenges for each task involved in image analysis, including lesion segmentation, dermoscopic feature detection within a lesion, and classification of melanoma. Training data included 900 images. A separate test dataset of 379 images was provided to measure resultant performance of systems developed with the training data. Ground truth for both training and test sets was generated by a panel of dermoscopic experts. In total, there were 79 submissions from a group of 38 participants, making this the largest standardized and comparative study for melanoma diagnosis in dermoscopic images to date. While the official challenge duration and ranking of participants has concluded, the datasets remain available for further research and development.
['Emre Celebi', 'David Gutman', 'Brian Helba', 'Michael Marchetti', 'Allan Halpern', 'Nabin Mishra', 'Noel C. F. Codella']
2016-05-04
null
null
null
null
['melanoma-diagnosis']
['computer-vision']
[ 7.34442174e-01 1.10294223e-01 -1.90327600e-01 -4.48342055e-01 -8.85910928e-01 -7.73247302e-01 5.20724595e-01 4.15570199e-01 -6.25993431e-01 3.18735063e-01 -2.95404047e-01 -3.49207222e-01 9.89548191e-02 -4.83647257e-01 -2.11056903e-01 -6.70672059e-01 1.75446853e-01 3.00563335e-01 2.53554016e-01 1.25498578e-01 2.79750884e-01 4.49380994e-01 -1.22839093e+00 7.90433109e-01 1.04997945e+00 7.62516081e-01 9.50189959e-03 1.32910204e+00 2.12256581e-01 4.70575809e-01 -5.64708233e-01 -7.70805895e-01 3.05207551e-01 -3.70026529e-01 -1.17708492e+00 6.60290658e-01 1.09207320e+00 -5.19985974e-01 -7.87241980e-02 8.39515507e-01 7.47959197e-01 -6.29692137e-01 8.01437855e-01 -7.16741145e-01 -3.96885395e-01 -2.80533850e-01 -7.74277151e-01 4.64350879e-01 3.08252335e-01 2.97230363e-01 5.72478354e-01 -3.56481522e-01 1.24227548e+00 6.03097618e-01 8.52248073e-01 9.60630536e-01 -1.10924220e+00 -9.83205214e-02 -1.84664890e-01 1.98914334e-02 -1.14977825e+00 -2.39240259e-01 -1.90072209e-01 -7.19581544e-01 8.17533970e-01 7.84748852e-01 9.70378637e-01 8.72121036e-01 7.87488967e-02 7.96824515e-01 1.61999285e+00 -5.64908445e-01 3.41797583e-02 4.73220915e-01 2.24478155e-01 8.45003068e-01 2.27394909e-01 4.14663963e-02 -1.56070173e-01 -2.35159457e-01 7.26222396e-01 -4.49303687e-01 -1.64310426e-01 6.40547350e-02 -7.14020610e-01 5.74455678e-01 2.38729507e-01 -1.92360282e-01 -3.81412208e-01 -2.91890889e-01 4.00336444e-01 3.97823423e-01 7.20070720e-01 3.59911680e-01 1.14675999e-01 1.99152425e-01 -9.65539992e-01 2.35466167e-01 8.19162667e-01 2.05888420e-01 5.70376292e-02 -7.40169942e-01 -2.73151040e-01 1.04972053e+00 5.17994398e-03 3.60142708e-01 2.19832227e-01 -5.72280467e-01 6.41753450e-02 8.02728295e-01 -1.89260796e-01 -4.46595937e-01 -3.78145963e-01 1.91557750e-01 -4.65609103e-01 4.57274705e-01 8.83264303e-01 -3.67317915e-01 -1.62653601e+00 8.18379045e-01 5.54397047e-01 -4.30772677e-02 -2.32079580e-01 1.05308616e+00 1.04278982e+00 1.98020115e-02 1.99306712e-01 2.45660737e-01 1.45824468e+00 -9.32774723e-01 -3.60485703e-01 7.35710487e-02 6.30626798e-01 -1.20733368e+00 6.53492808e-01 6.43076956e-01 -1.17560399e+00 -8.81533548e-02 -8.42803955e-01 2.63741706e-02 -3.93439323e-01 3.51397812e-01 8.10740292e-01 9.30616200e-01 -1.23410439e+00 1.91624641e-01 -7.80260742e-01 -1.08037102e+00 8.02198529e-01 3.85356754e-01 -7.27730215e-01 -2.97278136e-01 -6.56818509e-01 1.16126800e+00 -2.27230683e-01 -1.56680662e-02 -7.38898873e-01 -9.44570661e-01 -3.91271323e-01 -1.19267058e+00 -2.67961547e-02 -5.63540995e-01 1.40114474e+00 -1.21429026e+00 -1.00229168e+00 2.03475904e+00 -1.52689591e-01 -5.28559923e-01 8.67287755e-01 2.86839128e-01 -6.53387308e-01 6.12724721e-01 -7.79772997e-02 7.78849423e-01 7.86976337e-01 -1.15251446e+00 -1.18552327e+00 -4.63810444e-01 1.45173728e-01 2.83252507e-01 -4.73979376e-02 2.60016590e-01 -5.90928793e-01 -2.76814222e-01 -5.00476480e-01 -1.13127542e+00 -3.22399586e-01 2.24238425e-01 -8.09460044e-01 4.83163595e-02 3.49018067e-01 -1.12612331e+00 9.58477795e-01 -1.85768425e+00 -3.65725070e-01 4.27016973e-01 5.83460450e-01 6.55030608e-01 -2.70459294e-01 4.68534529e-01 -1.24505445e-01 4.73351747e-01 1.93483811e-02 -2.23047897e-01 -5.01138568e-01 -3.32458973e-01 1.84325278e-01 6.44498944e-01 5.42840421e-01 8.93927336e-01 -6.36227012e-01 -5.80421329e-01 3.61017287e-01 2.31678590e-01 1.76596954e-01 -1.59171924e-01 -6.35142550e-02 1.27344787e-01 -2.18418866e-01 1.10690975e+00 9.02832687e-01 -3.53938639e-01 2.02588335e-01 -3.19085062e-01 7.90351555e-02 -2.07271308e-01 -7.16839910e-01 1.48149347e+00 -1.75174579e-01 7.48235822e-01 4.22195435e-01 -9.24433172e-02 2.72271097e-01 2.41626337e-01 5.00202298e-01 -4.87082839e-01 -1.09539563e-02 2.26255462e-01 1.02633908e-01 -9.37210619e-01 -2.48705037e-02 7.30507225e-02 4.35224831e-01 4.54217017e-01 -1.12412147e-01 -2.41864890e-01 7.74642289e-01 1.37356043e-01 1.40759504e+00 -2.78917193e-01 1.86717644e-01 2.92112380e-02 3.25729191e-01 8.14919591e-01 -2.78398335e-01 4.76288259e-01 -5.48384190e-01 8.15455019e-01 5.21495283e-01 -5.89817822e-01 -1.10243118e+00 -1.39880502e+00 -7.42014885e-01 7.09295988e-01 1.55224517e-01 -2.03379855e-01 -1.17608190e+00 -1.01502740e+00 2.42032051e-01 -1.84784725e-01 -1.16911292e+00 5.25194705e-01 3.93348709e-02 -1.26673925e+00 7.67696798e-01 8.99454653e-02 1.86104894e-01 -7.38570273e-01 -3.90733778e-01 -1.83083028e-01 5.36328368e-02 -9.04124737e-01 -4.31622952e-01 -1.44600958e-01 -5.34290552e-01 -1.94343209e+00 -1.00793123e+00 -1.01691985e+00 1.17833090e+00 2.15444744e-01 8.92346919e-01 3.18155169e-01 -1.71058488e+00 4.63364631e-01 -9.79821756e-02 -7.62203217e-01 -4.84608531e-01 6.88029677e-02 -4.99352872e-01 -3.01574320e-02 6.72620296e-01 3.50311846e-01 -8.74920964e-01 1.84274539e-01 -1.02703965e+00 -2.64116675e-02 9.18841541e-01 8.76622140e-01 9.07811582e-01 -2.88415849e-01 2.87503004e-01 -1.36652946e+00 9.47137594e-01 -3.32087576e-01 -2.78813839e-01 6.09727323e-01 -3.54191184e-01 -7.34229743e-01 -3.74886952e-02 -3.79475328e-04 -9.73363459e-01 2.03438416e-01 -2.28441134e-02 2.19632864e-01 -4.94580060e-01 3.02153260e-01 7.71557331e-01 -6.36298180e-01 1.10458362e+00 -1.56088606e-01 7.49025881e-01 -1.92567050e-01 -3.47477347e-02 8.58417749e-01 4.28038895e-01 1.90527216e-01 4.36282575e-01 7.39568174e-01 5.79454415e-02 -1.03219676e+00 -7.67431200e-01 -8.89055014e-01 -6.28940105e-01 -2.18788266e-01 7.31713295e-01 -8.31125975e-01 -1.35734826e-01 1.10561335e+00 -5.91009855e-01 -5.56314111e-01 -1.65897399e-01 6.44510686e-02 -3.88098508e-02 4.45313632e-01 -8.70278358e-01 -5.39900541e-01 -5.77995837e-01 -9.78964806e-01 1.09641564e+00 5.04524291e-01 -4.85318780e-01 -1.51523495e+00 4.75879639e-01 8.40364158e-01 1.06040776e-01 8.22964609e-01 5.42457342e-01 -4.01615322e-01 -3.24232094e-02 -8.15155983e-01 -3.58391017e-01 4.28706408e-01 3.76527667e-01 8.26712132e-01 -1.21587849e+00 -3.71093154e-01 -5.84129691e-01 -6.92168891e-01 1.19635451e+00 5.76186419e-01 1.07585490e+00 2.75344819e-01 -5.39245188e-01 5.98579466e-01 1.55230761e+00 -2.02779785e-01 7.51253307e-01 1.69849530e-01 3.75868052e-01 1.03210783e+00 8.28281879e-01 -2.16935048e-04 2.80568957e-01 1.54071599e-01 4.06317264e-01 -8.48308802e-01 -5.47382176e-01 1.75092429e-01 -1.98159441e-01 -1.25068232e-01 -5.47323704e-01 -7.30940923e-02 -1.05112314e+00 8.02612841e-01 -1.13011241e+00 -6.13160133e-01 -4.99627501e-01 2.23573613e+00 8.29092979e-01 -2.03721762e-01 6.14877939e-01 -1.81649566e-01 9.20158327e-01 -1.08224832e-01 -5.79824567e-01 -5.11361122e-01 -6.37521222e-02 5.22826433e-01 6.89263701e-01 4.59510535e-01 -1.43771470e+00 5.89006066e-01 7.79195070e+00 9.58021224e-01 -1.38330090e+00 -2.64980704e-01 1.00774550e+00 -3.67444247e-01 -1.52006119e-01 -5.02035260e-01 -6.11678004e-01 2.47673392e-01 7.66516924e-01 -4.68547754e-02 1.81667492e-01 2.72724420e-01 2.49927808e-02 -6.93872750e-01 -8.57445300e-01 5.68819344e-01 9.45042074e-02 -1.47490740e+00 -1.49960056e-01 3.89714003e-01 9.19235945e-01 1.74958944e-01 5.10707974e-01 -5.03542066e-01 1.12001069e-01 -1.59412026e+00 -1.96365088e-01 6.27204478e-01 1.45983028e+00 -5.32012284e-01 9.73556221e-01 -6.78950101e-02 -6.21794939e-01 3.68481070e-01 -9.15014446e-02 3.60620409e-01 -3.67265612e-01 4.97496217e-01 -1.56948864e+00 4.23150003e-01 3.48012120e-01 5.78045547e-01 -1.17468858e+00 1.42372417e+00 -2.35818084e-02 6.87253773e-01 -1.48741469e-01 -2.96802185e-02 3.01026762e-01 -3.97207178e-02 1.92319393e-01 1.44138813e+00 -3.48144114e-01 -3.67217243e-01 -1.64373353e-01 3.73900384e-01 1.53599933e-01 2.04410091e-01 -3.62984389e-01 -3.48807037e-01 2.25165635e-01 1.93197000e+00 -7.08080173e-01 1.06676377e-01 -1.95229262e-01 9.97906089e-01 6.51192442e-02 3.04571569e-01 -4.71211106e-01 -3.73326570e-01 5.51122367e-01 5.22938788e-01 -2.88584769e-01 4.59282249e-01 -4.34753329e-01 -5.12902975e-01 5.77314049e-02 -8.71499598e-01 6.46219790e-01 -2.45297670e-01 -1.73459852e+00 3.83712351e-01 -3.95867884e-01 -7.85793960e-01 -1.32431120e-01 -1.04118538e+00 -7.87570298e-01 1.29758680e+00 -1.64536083e+00 -1.58000612e+00 -8.97823095e-01 4.00374234e-01 1.69515118e-01 -1.81916982e-01 1.18772125e+00 -1.25699475e-01 -7.53429115e-01 7.42395461e-01 8.91116634e-02 1.73036560e-01 1.29300678e+00 -1.73776340e+00 5.08615732e-01 3.82618994e-01 -2.00111836e-01 4.03792351e-01 1.10022768e-01 -7.72052109e-01 -1.10283291e+00 -1.19356608e+00 5.47838807e-01 -8.25454950e-01 6.94385290e-01 -1.83717962e-02 -2.88523227e-01 2.97855645e-01 3.89591128e-01 1.67851690e-02 1.34465837e+00 -6.27519712e-02 -8.76789615e-02 4.39100936e-02 -1.63883615e+00 4.51939851e-01 4.85431641e-01 -5.32184064e-01 3.41217257e-02 9.26833212e-01 -2.63879836e-01 -8.80007327e-01 -1.26327074e+00 1.83403373e-01 7.66673923e-01 -7.52543449e-01 7.79969215e-01 -6.85811281e-01 7.63975620e-01 1.07858293e-01 5.27026296e-01 -1.11950624e+00 9.20429453e-02 -4.46799517e-01 2.11895347e-01 8.02959919e-01 6.00085437e-01 -4.68159497e-01 1.32812476e+00 4.28256333e-01 1.07499726e-01 -1.33049476e+00 -5.53027332e-01 -9.85233635e-02 1.90337241e-01 1.64950371e-01 -1.54595971e-01 6.49949551e-01 -1.29435569e-01 -1.86317503e-01 2.97570437e-01 -1.53922230e-01 8.33871007e-01 -1.15339674e-01 7.67765760e-01 -8.31976056e-01 7.71690011e-02 -3.97901356e-01 -6.26368701e-01 -2.30554834e-01 -3.94757748e-01 -9.87176418e-01 -4.05847222e-01 -1.96589458e+00 4.47716594e-01 -3.79885554e-01 -4.46354412e-02 5.16056955e-01 -5.59280157e-01 8.39562535e-01 -1.88734248e-01 1.09117538e-01 -3.89770091e-01 -8.29932213e-01 1.68525863e+00 -3.17088991e-01 1.13340467e-01 2.42006332e-01 -7.97203124e-01 5.02114177e-01 8.76130223e-01 1.03134759e-01 -3.99354309e-01 -1.69946030e-01 -1.85742959e-01 -5.06011665e-01 7.25529432e-01 -8.17816675e-01 2.59023219e-01 -2.02329785e-01 7.96049416e-01 -4.71886605e-01 2.99335361e-01 -1.78477287e-01 -9.00020897e-02 6.61430717e-01 -2.33810723e-01 -4.84964877e-01 2.26843894e-01 4.54563677e-01 -2.94477761e-01 -2.16489941e-01 1.04677427e+00 -2.67433256e-01 -1.01383030e+00 4.09551859e-01 -4.65731919e-01 7.63955265e-02 1.79020619e+00 -6.63138330e-01 -8.55731606e-01 3.24241929e-02 -9.80265617e-01 3.27622026e-01 7.69494772e-01 2.44767070e-02 5.67904353e-01 -7.69849658e-01 -1.19763219e+00 -1.35704726e-01 4.10895467e-01 -1.21417314e-01 8.22567225e-01 1.14741838e+00 -1.23595643e+00 2.77323574e-01 -4.70162600e-01 -6.40412629e-01 -1.81251454e+00 -6.28635958e-02 8.21692824e-01 -4.75422055e-01 -2.14666352e-01 1.29360855e+00 -3.13738018e-01 -4.62962955e-01 -9.10452604e-02 1.27270184e-02 -7.79848769e-02 6.86169951e-04 8.29049110e-01 4.40435827e-01 4.26574588e-01 -2.02015832e-01 -2.79103756e-01 3.42039555e-01 -6.50068581e-01 1.86775208e-01 8.37086678e-01 1.72139868e-01 -2.06060782e-01 2.72274595e-02 8.46587181e-01 -9.15639196e-03 -1.00576639e+00 3.05843949e-01 -3.23594600e-01 -5.07386804e-01 -2.36062631e-02 -1.48907232e+00 -7.72223115e-01 7.34863222e-01 1.00224459e+00 4.48346645e-01 1.10588682e+00 -3.25995713e-01 6.63903117e-01 -7.11693391e-02 -1.03446366e-02 -1.43917525e+00 -1.04285419e-01 -1.56475119e-02 6.62185371e-01 -1.44821179e+00 1.18479058e-01 -1.00760555e+00 -9.17445779e-01 1.16715407e+00 7.96289980e-01 -2.24011213e-01 4.67889607e-01 4.39585567e-01 7.34291255e-01 -4.11524445e-01 -6.53380156e-01 -2.17918500e-01 6.48567140e-01 1.21811926e+00 5.92188835e-01 2.05904320e-01 -4.07815635e-01 1.23166583e-01 5.34430593e-02 9.13386643e-02 4.53538477e-01 9.98566866e-01 -1.64406359e-01 -1.32928789e+00 -9.56278220e-02 1.28120971e+00 -7.61200905e-01 4.09771837e-02 -1.29510987e+00 1.16558278e+00 3.93601358e-01 9.45359349e-01 -4.02363576e-03 -2.98610091e-01 2.74853855e-01 -4.21812445e-01 9.65344965e-01 -9.45069671e-01 -1.02075362e+00 -2.86943883e-01 5.81157684e-01 -4.96672541e-01 -3.87568265e-01 -6.06332541e-01 -8.53078544e-01 -3.22102726e-01 -2.07390279e-01 -3.21898282e-01 9.56257343e-01 5.79321325e-01 6.96770400e-02 4.88280565e-01 5.90463698e-01 -1.72412187e-01 -5.99351466e-01 -8.91913831e-01 -9.93666708e-01 4.46746498e-01 1.87085330e-01 -4.56755906e-02 -1.14342086e-01 3.42385471e-01]
[15.703741073608398, -3.001753330230713]
141e0dc3-8a02-4487-8db0-01219663bee5
divisive-language-and-propaganda-detection
null
null
https://aclanthology.org/D19-5014
https://aclanthology.org/D19-5014.pdf
Divisive Language and Propaganda Detection using Multi-head Attention Transformers with Deep Learning BERT-based Language Models for Binary Classification
On the NLP4IF 2019 sentence level propaganda classification task, we used a BERT language model that was pre-trained on Wikipedia and BookCorpus as team ltuorp ranking {\#}1 of 26. It uses deep learning in the form of an attention transformer. We substituted the final layer of the neural network to a linear real valued output neuron from a layer of softmaxes. The backpropagation trained the entire neural network and not just the last layer. Training took 3 epochs and on our computation resources this took approximately one day. The pre-trained model consisted of uncased words and there were 12-layers, 768-hidden neurons with 12-heads for a total of 110 million parameters. The articles used in the training data promote divisive language similar to state-actor-funded influence operations on social media. Twitter shows state-sponsored examples designed to maximize division occurring across political lines, ranging from {``}Obama calls me a clinger, Hillary calls me deplorable, ... and Trump calls me an American{''} oriented to the political right, to Russian propaganda featuring {``}Black Lives Matter{''} material with suggestions of institutional racism in US police forces oriented to the political left. We hope that raising awareness through our work will reduce the polarizing dialogue for the betterment of nations.
['Norman Mapes', 'Sumeet Dua', 'Radhika Medury', 'Anna White']
2019-11-01
null
null
null
ws-2019-11
['propaganda-detection']
['natural-language-processing']
[-8.12832937e-02 7.23691761e-01 -3.92653137e-01 -1.77099735e-01 -6.64107800e-01 -4.95472431e-01 1.11064827e+00 6.38235658e-02 -7.79968798e-01 1.03381503e+00 9.62020338e-01 -1.18993819e+00 -9.07745212e-02 -8.51584017e-01 -8.17116737e-01 -5.65652668e-01 7.99899846e-02 6.46391213e-01 -3.24423254e-01 -8.39498341e-01 5.85162044e-01 2.51204699e-01 -5.78523457e-01 3.72075856e-01 6.96627796e-01 4.87889618e-01 -3.50579232e-01 6.60976589e-01 -1.84139326e-01 1.79540503e+00 -1.07813704e+00 -6.20599985e-01 2.68754572e-01 -2.10922986e-01 -1.02798200e+00 -5.72676122e-01 4.79895413e-01 -4.11248416e-01 -7.16928422e-01 9.41668868e-01 4.64030743e-01 9.58634242e-02 7.41773665e-01 -5.83489001e-01 -9.28618312e-01 1.55508876e+00 -5.60771406e-01 6.24116898e-01 -1.34491965e-01 1.38036031e-02 1.17600536e+00 -7.74245381e-01 8.92202020e-01 1.46798515e+00 5.60624540e-01 3.57779980e-01 -1.12478292e+00 -8.91679645e-01 2.45472595e-01 -1.34607777e-01 -7.83926010e-01 -2.78185248e-01 5.83683312e-01 -7.39800036e-01 1.34123278e+00 1.84479237e-01 8.24568391e-01 1.46827400e+00 5.07462442e-01 3.72085989e-01 1.10637319e+00 -1.88905239e-01 -3.46898347e-01 1.68590754e-01 3.15309197e-01 7.26673961e-01 3.04960936e-01 -1.66681558e-01 -2.69371629e-01 -5.38534701e-01 5.91257036e-01 -3.08448404e-01 1.02175511e-01 6.24348044e-01 -1.19521499e+00 1.36202729e+00 7.16273785e-01 5.60482919e-01 -2.33788967e-01 4.69290078e-01 4.37924922e-01 3.27047199e-01 6.52751088e-01 8.04608941e-01 -3.63462299e-01 -8.66782442e-02 -8.94219279e-01 2.72533417e-01 8.49415421e-01 2.78490096e-01 4.60018218e-01 2.66171873e-01 -9.92884114e-02 7.32886076e-01 1.31290674e-01 5.03295243e-01 1.10208884e-01 -8.33269775e-01 7.78842449e-01 2.88645893e-01 -6.71863258e-02 -1.26889729e+00 -8.77504170e-01 -7.55521774e-01 -8.14940333e-01 1.75772980e-01 6.35763824e-01 -5.79736352e-01 -9.70732152e-01 1.62791145e+00 -3.78436565e-01 -6.36511683e-01 -1.59772843e-01 7.44541526e-01 7.89541960e-01 1.16362119e+00 2.89385736e-01 -1.20226011e-01 1.26037085e+00 -8.85687768e-01 -5.48798442e-01 -5.12713850e-01 7.49376416e-01 -6.04266524e-01 8.83648515e-01 6.48654178e-02 -1.18477988e+00 2.61362810e-02 -8.75941396e-01 -2.92621464e-01 -4.97468531e-01 -1.68087512e-01 5.26873529e-01 3.37606251e-01 -8.56031835e-01 5.23317575e-01 -4.74560708e-01 -2.56696194e-01 6.95478439e-01 1.12576321e-01 -2.16921598e-01 5.26904047e-01 -1.62728608e+00 1.38086712e+00 4.14168507e-01 3.09134752e-01 -7.75195122e-01 -4.25419539e-01 -7.29912341e-01 1.29392713e-01 3.51810843e-01 -4.81947035e-01 8.28153491e-01 -1.13803267e+00 -1.24587262e+00 9.95147049e-01 4.24935251e-01 -6.49529815e-01 4.71928328e-01 -1.36092931e-01 -5.28480768e-01 -6.58462048e-02 3.79916340e-01 5.12769997e-01 6.21199191e-01 -8.58383179e-01 -2.76760310e-01 -5.62697873e-02 3.18482995e-01 1.70171544e-01 -2.27230281e-01 6.57761693e-01 5.21508992e-01 -8.15688848e-01 -1.38323635e-01 -7.29180515e-01 -9.78240967e-02 -8.86305630e-01 -9.02398527e-01 -2.52827048e-01 5.02309561e-01 -1.02923143e+00 1.39781857e+00 -1.75981891e+00 5.98436408e-02 3.85193616e-01 3.29821438e-01 -6.10424438e-03 1.10911384e-01 6.47698998e-01 -2.12145343e-01 7.86891818e-01 -3.79198492e-02 4.49126571e-01 9.21489298e-02 -8.09671581e-02 -5.52614748e-01 6.71095133e-01 1.30272210e-01 8.02310467e-01 -8.79376829e-01 -2.07313806e-01 -3.20457131e-01 2.46857166e-01 -5.41056275e-01 -4.16743398e-01 -2.22072572e-01 1.81806967e-01 -2.60241032e-01 4.19846743e-01 2.74292916e-01 -4.98357743e-01 2.50884414e-01 2.34834269e-01 -4.87125784e-01 8.66585076e-01 -2.94274837e-01 1.19535339e+00 -1.66737244e-01 1.16474068e+00 3.85876983e-01 -9.01797533e-01 6.65797889e-01 3.12549733e-02 1.70146763e-01 -5.67805648e-01 6.87453687e-01 2.52931565e-01 5.12957096e-01 -3.23876590e-01 6.19489074e-01 -3.59642416e-01 -5.42451322e-01 6.40440762e-01 -1.52011305e-01 -1.74896672e-01 1.24362782e-01 7.00646400e-01 1.20752883e+00 -2.19880119e-01 1.23867586e-01 -7.18983471e-01 2.14311808e-01 2.76846528e-01 3.88520390e-01 8.63815486e-01 -7.30545968e-02 1.15324371e-01 1.13863790e+00 -9.14806724e-01 -1.16390598e+00 -8.84125829e-01 -1.72077030e-01 1.49195552e+00 -5.15286267e-01 -3.51217568e-01 -4.93066937e-01 -5.44899881e-01 -2.50108123e-01 1.06489360e+00 -8.93530846e-01 1.43130913e-01 -9.17102218e-01 -8.39598536e-01 7.30820358e-01 -5.25721796e-02 5.26306391e-01 -1.31584489e+00 -5.72554886e-01 1.73411682e-01 -2.02349052e-01 -6.22055888e-01 -2.76029557e-01 5.09483218e-01 -4.10809219e-01 -8.07608187e-01 -7.04752505e-01 -7.23042965e-01 5.42189837e-01 -5.02012789e-01 1.17143953e+00 3.73789035e-02 -5.68935648e-02 -4.14873719e-01 4.27791327e-02 -6.05587006e-01 -6.56513989e-01 4.11775768e-01 -1.89713668e-02 -6.01048827e-01 1.20311521e-01 -2.91624844e-01 -2.62905985e-01 -4.24694091e-01 -4.65039641e-01 3.63592833e-01 4.56047177e-01 1.01501572e+00 -2.96385437e-01 -4.40062284e-01 7.65022278e-01 -1.25921118e+00 9.21330094e-01 -6.43207133e-01 -4.27741945e-01 -8.30281675e-02 -2.22568810e-01 -1.70340445e-02 5.58639765e-01 -8.38094801e-02 -8.54263604e-01 -8.52231503e-01 -1.87486857e-01 2.10597843e-01 3.11663628e-01 7.50299811e-01 4.18519676e-01 5.16094863e-01 1.21684241e+00 -1.49754927e-01 -8.47378448e-02 -2.04648435e-01 4.89310533e-01 7.03869164e-01 5.47188759e-01 -3.22194666e-01 7.84051061e-01 2.90382802e-01 -4.51090306e-01 -6.79365039e-01 -9.63128209e-01 2.47949466e-01 -6.89438283e-02 -1.31476253e-01 1.08116150e+00 -9.95703697e-01 -6.51946902e-01 4.60989892e-01 -1.32545078e+00 -5.97938240e-01 -2.48081014e-01 3.69716734e-01 -2.32265234e-01 -1.66244686e-01 -1.15368164e+00 -7.17423141e-01 -5.47410011e-01 -6.13584399e-01 2.84095615e-01 2.00585589e-01 -2.13052481e-01 -9.51904535e-01 1.35772359e-02 4.84724343e-01 5.96778333e-01 2.87450820e-01 1.30577433e+00 -9.90099847e-01 -3.37366574e-02 -4.59560603e-02 -3.90900433e-01 2.05114916e-01 -7.58610368e-02 1.10848792e-01 -6.53690457e-01 -1.04470104e-01 -2.93726921e-01 -4.70564932e-01 9.28058803e-01 4.51185733e-01 7.63408363e-01 -1.06022072e+00 -7.57815540e-02 9.66137201e-02 9.61168945e-01 4.00978737e-02 5.67125797e-01 8.59532773e-01 5.76170385e-01 6.19764209e-01 1.40747055e-03 3.13645959e-01 4.59146857e-01 6.62585050e-02 4.00034666e-01 -3.27079564e-01 1.01481453e-01 -2.85098910e-01 6.86149240e-01 7.19465911e-01 -2.92117000e-01 -1.83224186e-01 -1.27875936e+00 4.20633227e-01 -1.48298323e+00 -1.47626936e+00 -6.19329661e-02 1.74837661e+00 9.83703196e-01 5.76142430e-01 9.69968662e-02 -4.25153017e-01 6.45950258e-01 6.78118944e-01 -1.73459798e-01 -1.12766695e+00 -3.03525478e-01 2.33311236e-01 7.16230929e-01 9.42838192e-01 -1.07994556e+00 1.30740380e+00 6.46647644e+00 6.97212934e-01 -1.25565004e+00 2.60687262e-01 9.85039949e-01 -5.48970938e-01 -4.78646457e-01 2.40232740e-02 -5.99655151e-01 5.62798619e-01 1.00918019e+00 -3.54870319e-01 4.80227828e-01 5.80749631e-01 3.46257687e-01 -2.48107146e-02 -3.34917963e-01 4.72127140e-01 1.24038376e-01 -1.90519440e+00 -9.30030197e-02 3.73304665e-01 9.16357756e-01 6.32246077e-01 2.06760257e-01 5.52562296e-01 8.63339365e-01 -1.43302190e+00 1.04944718e+00 3.77981484e-01 7.46647239e-01 -8.17253172e-01 7.11177886e-01 6.09391212e-01 -2.12914377e-01 -4.33491290e-01 -1.90057695e-01 -5.69313884e-01 3.49770695e-01 6.39246166e-01 -7.07897246e-01 4.69429493e-02 5.43849051e-01 3.14271510e-01 -2.04602331e-01 7.35599324e-02 -5.69384992e-01 8.26666713e-01 -4.15492535e-01 -3.19647580e-01 9.23335552e-01 -1.29695326e-01 8.27882171e-01 1.33633697e+00 -8.59543011e-02 2.31182605e-01 -1.34016678e-01 6.65414751e-01 -4.84346211e-01 2.18564466e-01 -7.21788406e-01 -3.12302023e-01 2.81463891e-01 1.23373473e+00 -6.03177011e-01 -5.34095526e-01 -5.01498580e-02 3.40963423e-01 5.29038310e-01 3.63800466e-01 -9.91939604e-01 -5.20171881e-01 3.55233788e-01 6.52198493e-01 -2.37870798e-01 -2.31709659e-01 -2.84675270e-01 -1.04064727e+00 -6.48126423e-01 -8.42529058e-01 2.55252272e-01 -4.82588768e-01 -1.23497701e+00 7.00701058e-01 -1.06150530e-01 -3.12327385e-01 -2.45686546e-01 -5.55903733e-01 -7.73018777e-01 9.57852423e-01 -8.67784441e-01 -1.13516223e+00 4.50960726e-01 -2.88879126e-02 3.21862936e-01 -3.66099685e-01 4.19541776e-01 9.91417244e-02 -5.03163397e-01 2.64412552e-01 -6.20981790e-02 5.09777129e-01 6.32737279e-01 -9.89510119e-01 3.05418432e-01 6.70950651e-01 -6.84684962e-02 8.05822134e-01 8.22578192e-01 -6.98177397e-01 -7.08679855e-01 -6.56489491e-01 1.52336431e+00 -4.90031660e-01 1.36940229e+00 -5.08664310e-01 -6.19177699e-01 9.25587177e-01 8.19447577e-01 -5.97224891e-01 5.37633896e-01 3.40081990e-01 -3.63780588e-01 3.66872847e-01 -8.49341691e-01 6.13327324e-01 6.83360755e-01 -5.16104341e-01 -7.93524981e-01 8.45849752e-01 5.72101176e-01 -3.46376926e-01 -5.20970702e-01 -5.49399368e-02 5.94327927e-01 -5.24985135e-01 6.62069857e-01 -1.14977205e+00 9.75147426e-01 1.72428355e-01 1.39365435e-01 -1.25048506e+00 -8.02923918e-01 -5.90822995e-01 3.08480442e-01 7.61178911e-01 1.05126774e+00 -8.23310435e-01 5.39221883e-01 4.49804395e-01 -3.77073258e-01 -7.32499599e-01 -1.06610644e+00 -2.94490993e-01 8.30871701e-01 -2.84550861e-02 -1.37839302e-01 1.30405462e+00 3.82888466e-01 7.44793773e-01 -7.10402071e-01 -2.52519757e-01 2.66699791e-01 -3.17596972e-01 4.35960352e-01 -9.19250906e-01 -2.08643422e-01 -9.22755361e-01 7.75566548e-02 -7.12028265e-01 2.24592656e-01 -1.24814355e+00 -3.07825416e-01 -1.89570403e+00 3.83743763e-01 -3.74283671e-01 -1.12407673e-02 7.59350359e-01 1.06835000e-01 5.68413734e-02 3.68787467e-01 1.58010826e-01 -1.24893755e-01 9.34246257e-02 1.05002940e+00 -5.24053097e-01 -9.57780257e-02 -4.05661732e-01 -1.32747185e+00 8.94680321e-01 9.55606997e-01 -5.74810147e-01 1.17686972e-01 -7.18161702e-01 9.99071479e-01 -2.06185713e-01 3.55321586e-01 -5.73901951e-01 3.02641153e-01 -3.14581543e-01 3.58899385e-01 -4.94624496e-01 2.54039854e-01 -2.51056254e-01 6.41585439e-02 8.66719544e-01 -4.59186614e-01 -3.85005376e-03 8.44387561e-02 1.04857542e-01 9.24399570e-02 -1.68383390e-01 7.77510285e-01 -3.18782151e-01 2.52162404e-02 -9.72517878e-02 -8.79764378e-01 1.18800230e-01 6.04357481e-01 2.82048613e-01 -1.21359193e+00 -6.61656976e-01 -7.09460318e-01 2.33325049e-01 2.30515629e-01 3.25586885e-01 -4.55846190e-02 -1.06576824e+00 -1.22820985e+00 -3.31718385e-01 -5.14168143e-01 -2.72351265e-01 8.02890211e-02 9.61574495e-01 -8.97224963e-01 4.90250826e-01 -1.48879901e-01 1.54075325e-01 -7.54200876e-01 1.99391991e-01 3.37801576e-01 -5.87368369e-01 -6.77082658e-01 1.09903204e+00 5.24829999e-02 -4.28853303e-01 -6.77884594e-02 -5.41307852e-02 -3.92556906e-01 5.16878784e-01 5.23356795e-01 4.59452450e-01 -3.42133850e-01 -7.95329571e-01 -5.24187684e-01 -5.85087538e-02 -3.14021289e-01 -3.75375897e-01 1.41021776e+00 3.09990346e-01 -5.04467189e-01 4.86177355e-01 1.06291926e+00 4.53154325e-01 -6.33123934e-01 1.10234730e-01 -1.31082907e-01 -5.58433011e-02 5.33950962e-02 -1.28298140e+00 -7.14052022e-01 7.36240745e-01 -8.94581228e-02 4.83690947e-01 3.77303153e-01 1.72222704e-01 4.95265037e-01 4.56434339e-01 8.18089619e-02 -1.55433071e+00 -3.95199507e-02 1.09619164e+00 1.16108012e+00 -9.27670777e-01 2.82452762e-01 1.77187875e-01 -5.84642231e-01 9.31591630e-01 3.66362870e-01 -3.07959318e-01 5.26931286e-01 2.55971730e-01 3.86843756e-02 -5.31155467e-01 -9.05254304e-01 3.70125860e-01 -2.54857510e-01 -4.79982421e-02 6.22718096e-01 3.06290865e-01 -8.63565445e-01 5.23952663e-01 -4.94202852e-01 -4.55502570e-01 6.93290174e-01 6.34612083e-01 -8.11249316e-01 -6.40123308e-01 -4.17236388e-01 8.59730244e-01 -9.06943202e-01 -6.28704727e-01 -6.44445777e-01 8.98089886e-01 1.02847062e-01 8.62302780e-01 2.22736791e-01 -1.72139689e-01 -8.71867090e-02 -2.13628188e-02 3.11665833e-02 -5.11171460e-01 -1.17453313e+00 2.28801116e-01 7.54448712e-01 -5.20606339e-02 5.14112115e-02 -4.65713829e-01 -1.36243689e+00 -9.67606425e-01 -2.39364296e-01 2.66857266e-01 8.29607010e-01 7.28255033e-01 3.94384377e-02 4.65947390e-01 5.06679833e-01 -6.85489357e-01 -7.17502892e-01 -1.37037134e+00 -4.84551340e-01 -3.46524306e-02 1.05249681e-01 -1.96776599e-01 -5.63055098e-01 -4.06884104e-01]
[8.488813400268555, 10.630599021911621]
f87ef4f4-2551-4c28-b5e8-4f1fbf52c418
pixel2mesh-generating-3d-mesh-models-from
1804.01654
null
http://arxiv.org/abs/1804.01654v2
http://arxiv.org/pdf/1804.01654v2.pdf
Pixel2Mesh: Generating 3D Mesh Models from Single RGB Images
We propose an end-to-end deep learning architecture that produces a 3D shape in triangular mesh from a single color image. Limited by the nature of deep neural network, previous methods usually represent a 3D shape in volume or point cloud, and it is non-trivial to convert them to the more ready-to-use mesh model. Unlike the existing methods, our network represents 3D mesh in a graph-based convolutional neural network and produces correct geometry by progressively deforming an ellipsoid, leveraging perceptual features extracted from the input image. We adopt a coarse-to-fine strategy to make the whole deformation procedure stable, and define various of mesh related losses to capture properties of different levels to guarantee visually appealing and physically accurate 3D geometry. Extensive experiments show that our method not only qualitatively produces mesh model with better details, but also achieves higher 3D shape estimation accuracy compared to the state-of-the-art.
['Yu-Gang Jiang', 'yinda zhang', 'Zhuwen Li', 'Yanwei Fu', 'Wei Liu', 'Nanyang Wang']
2018-04-05
pixel2mesh-generating-3d-mesh-models-from-1
http://openaccess.thecvf.com/content_ECCV_2018/html/Nanyang_Wang_Pixel2Mesh_Generating_3D_ECCV_2018_paper.html
http://openaccess.thecvf.com/content_ECCV_2018/papers/Nanyang_Wang_Pixel2Mesh_Generating_3D_ECCV_2018_paper.pdf
eccv-2018-9
['3d-object-reconstruction']
['computer-vision']
[-2.25699827e-01 6.69297501e-02 3.56798053e-01 -8.41220841e-02 -5.15734196e-01 -7.32256532e-01 2.87215143e-01 -8.02638233e-02 1.27881542e-01 3.39686722e-01 -2.74386406e-01 -3.24903041e-01 2.28127450e-01 -1.26120114e+00 -1.14826965e+00 -2.27745190e-01 -1.19065687e-01 7.30283678e-01 1.37693137e-01 -1.77168086e-01 1.70135796e-01 1.08980191e+00 -1.14568651e+00 1.49996057e-01 9.12874162e-01 1.45529270e+00 -2.44147837e-01 4.45063502e-01 -2.24922508e-01 8.60764608e-02 -2.20899060e-01 -6.91805065e-01 7.17795908e-01 -2.02057511e-01 -8.30488741e-01 2.09348947e-01 7.12607086e-01 -6.39485180e-01 -1.81482360e-01 1.05277431e+00 4.40833688e-01 -1.59300297e-01 9.81163859e-01 -9.33937013e-01 -1.19207501e+00 7.59828463e-02 -7.87618816e-01 -6.93433344e-01 3.25735092e-01 1.35837719e-01 6.53509796e-01 -1.17618084e+00 6.90442622e-01 1.61821198e+00 1.02068579e+00 3.37333232e-01 -1.44774556e+00 -5.91970444e-01 7.71771073e-02 -5.36101997e-01 -1.36737478e+00 8.74187201e-02 1.30266798e+00 -4.99229997e-01 4.34626937e-01 1.57749042e-01 1.00000727e+00 7.18661666e-01 4.23287779e-01 5.08797109e-01 8.61143231e-01 -2.02434599e-01 1.83994636e-01 -4.47788060e-01 -6.06193721e-01 1.07135820e+00 1.95273057e-01 6.89361766e-02 5.26166745e-02 -1.66571051e-01 1.68073118e+00 4.32179682e-02 -1.40890583e-01 -8.80831182e-01 -1.14910436e+00 5.53889871e-01 8.88628125e-01 -1.50388524e-01 -3.87545645e-01 5.24525404e-01 1.09739564e-01 2.39868447e-01 8.73034298e-01 3.39255631e-01 -1.84570357e-01 9.90918279e-02 -8.73414695e-01 4.55022573e-01 5.56788146e-01 9.12241757e-01 8.74138057e-01 3.98645878e-01 1.37943283e-01 4.85170156e-01 3.36089492e-01 6.94348752e-01 -3.25960338e-01 -1.32512140e+00 1.38371617e-01 9.49946821e-01 3.94961774e-01 -1.16131222e+00 -3.10869902e-01 -2.72778988e-01 -1.12851548e+00 9.99615371e-01 4.24023837e-01 -3.72473970e-02 -1.10748851e+00 1.26288199e+00 5.60096800e-01 9.95489955e-02 -3.62583458e-01 1.10746789e+00 9.45744097e-01 5.54404497e-01 -2.44729221e-01 2.63106167e-01 9.84260499e-01 -6.12610996e-01 -2.20870122e-01 2.92139560e-01 1.35465205e-01 -6.60580039e-01 1.21540356e+00 4.14682508e-01 -1.57796025e+00 -6.89275026e-01 -8.48788977e-01 -3.80763859e-01 -1.86401099e-01 1.82795182e-01 5.16984522e-01 3.43866259e-01 -1.28499568e+00 9.86701012e-01 -7.34883487e-01 1.50064647e-01 7.32982337e-01 8.66816267e-02 -2.44914219e-01 2.13080689e-01 -8.99748623e-01 5.59802592e-01 -4.79038991e-02 1.31989136e-01 -7.39510655e-01 -1.02641141e+00 -9.92358148e-01 2.05358099e-02 1.25730289e-02 -9.97411966e-01 1.00734973e+00 -9.45509017e-01 -1.79342639e+00 9.85683918e-01 2.06040636e-01 -7.10621069e-04 1.01250160e+00 -1.00541234e-01 2.23049164e-01 1.62349820e-01 -2.77491868e-01 6.61673486e-01 1.13222575e+00 -1.84694517e+00 1.02108665e-01 -7.01633394e-02 4.31993067e-01 1.70014113e-01 1.75353318e-01 -2.91832507e-01 -5.41483462e-01 -8.52763593e-01 3.07605535e-01 -6.88800573e-01 -2.06937596e-01 1.01091695e+00 -4.27170843e-01 -2.08594233e-01 8.11862171e-01 -5.05511999e-01 6.59712195e-01 -2.06742740e+00 2.14634776e-01 1.74281046e-01 5.20859540e-01 8.97219628e-02 -1.31748155e-01 3.93169135e-01 1.76972933e-02 4.40283686e-01 -4.29443508e-01 -6.06547296e-01 1.57045722e-01 -7.61225894e-02 -2.69006968e-01 4.74066079e-01 5.45888364e-01 1.10455084e+00 -1.00059676e+00 -5.26448786e-01 5.02972782e-01 7.98459172e-01 -6.83517456e-01 1.81739360e-01 -3.12155634e-01 4.81510341e-01 -5.98957360e-01 8.73102367e-01 1.17443085e+00 -4.10969079e-01 -1.98357299e-01 -5.21493793e-01 -3.03120781e-02 -2.57707536e-01 -1.14134133e+00 1.92845583e+00 -5.55365324e-01 1.56844988e-01 4.09567535e-01 -7.16337562e-01 1.07751703e+00 7.37186447e-02 5.87579250e-01 -4.75142628e-01 2.34659180e-01 3.06430459e-01 -3.63296181e-01 -1.62491798e-01 3.55438650e-01 -1.06450826e-01 -1.15680419e-01 2.71424800e-01 -4.89183307e-01 -8.08676839e-01 -4.02665555e-01 -1.19094603e-01 4.76994842e-01 6.16326690e-01 -1.87256694e-01 -3.21915925e-01 3.02300543e-01 3.66887897e-02 3.63813788e-01 2.97200024e-01 1.06764257e-01 1.22842956e+00 5.07987022e-01 -8.69202375e-01 -1.34275973e+00 -1.47870708e+00 -3.10451031e-01 4.66089636e-01 4.40874338e-01 1.23724751e-01 -7.54865706e-01 -4.10065353e-01 3.40281725e-01 2.69521952e-01 -7.67099202e-01 -4.98813926e-04 -8.26227903e-01 -2.17919752e-01 2.42933974e-01 7.37232983e-01 5.67262590e-01 -9.92133915e-01 -4.76022750e-01 1.90687597e-01 3.82938415e-01 -8.18115711e-01 -6.27118886e-01 -3.97483289e-01 -9.41252291e-01 -1.09058499e+00 -1.08837402e+00 -9.50601876e-01 9.36293006e-01 -1.08948946e-01 1.42368555e+00 3.39452475e-01 -9.12154689e-02 1.90945819e-01 -2.27854669e-01 -1.61374360e-01 -5.61832309e-01 -2.42333993e-01 -9.62993279e-02 8.97238702e-02 -5.62979996e-01 -7.96944737e-01 -9.69257474e-01 5.76768257e-02 -8.47743273e-01 4.17593092e-01 4.64885920e-01 5.43168485e-01 1.02562809e+00 -1.00706816e-02 2.96122313e-01 -6.66853428e-01 5.22222757e-01 -6.41764933e-03 -7.62812316e-01 5.59114218e-02 -3.57756346e-01 -4.69626486e-02 9.67020750e-01 -3.95496517e-01 -6.77961230e-01 2.14310020e-01 -2.05469236e-01 -1.13315988e+00 -4.27135490e-02 2.01109543e-01 -1.13919321e-02 -5.75255454e-01 2.97358841e-01 1.64658725e-01 7.62029365e-02 -6.62654340e-01 5.38707614e-01 1.65738925e-01 4.50642020e-01 -8.38822246e-01 1.16311467e+00 7.24757254e-01 3.43602568e-01 -3.41199964e-01 -5.63904643e-01 3.09309661e-01 -8.59230816e-01 -3.92177939e-01 8.48998070e-01 -7.27942586e-01 -9.64079738e-01 7.98417807e-01 -1.36940432e+00 -6.10930026e-01 -2.81177968e-01 -1.11723945e-01 -6.30364120e-01 3.62385899e-01 -7.26399601e-01 -6.59663916e-01 -6.34243846e-01 -1.22932422e+00 1.63422465e+00 -5.56500331e-02 1.43539101e-01 -8.60767961e-01 -7.58904368e-02 -1.09099254e-01 4.92192507e-01 1.01133287e+00 1.11327708e+00 3.72437060e-01 -6.88802719e-01 -1.36157542e-01 -5.53641617e-01 1.64797843e-01 5.06691188e-02 5.10551155e-01 -6.97558522e-01 -3.26561540e-01 -4.58689362e-01 -4.11375016e-01 6.48017526e-01 4.20489311e-01 1.69674623e+00 -2.15928704e-01 -6.58509582e-02 1.03537261e+00 1.57579339e+00 -6.60908595e-02 5.64946890e-01 -1.65827632e-01 1.12212682e+00 1.73258975e-01 1.12519469e-02 2.72210658e-01 5.08735001e-01 5.55394828e-01 8.89883697e-01 -5.43136120e-01 -3.63183081e-01 -2.60020316e-01 -4.96180914e-02 6.19018555e-01 -2.72209048e-01 3.68429460e-02 -7.57299125e-01 3.15859288e-01 -1.45731401e+00 -4.65308130e-01 -1.40319631e-01 2.09595680e+00 7.98598826e-01 2.24186048e-01 1.29519165e-01 -9.85694751e-02 4.98476297e-01 1.88000247e-01 -8.48274410e-01 -5.72100163e-01 1.70431107e-01 3.75370413e-01 2.18317747e-01 4.68998909e-01 -9.66248453e-01 8.19121838e-01 6.26073360e+00 7.30586290e-01 -1.43517256e+00 -2.67624319e-01 8.56936157e-01 1.15124092e-01 -7.45364368e-01 -3.10723454e-01 -6.20002346e-03 4.33474034e-01 1.27050444e-01 3.70270275e-02 4.24518853e-01 8.04749727e-01 1.47038728e-01 4.12750453e-01 -1.12123251e+00 1.07672524e+00 -3.01476568e-01 -1.67799175e+00 5.15366137e-01 6.57301843e-02 8.55898321e-01 -1.09129004e-01 8.50791857e-02 5.26937954e-02 3.69409800e-01 -1.27497149e+00 1.29601443e+00 9.09819901e-01 1.41513896e+00 -9.00202334e-01 2.39383474e-01 1.20847814e-01 -1.54515243e+00 5.31165957e-01 -4.46426839e-01 4.59581800e-02 9.62627381e-02 6.33106530e-01 -1.50317922e-01 7.59779930e-01 8.51899981e-01 6.14105701e-01 -3.16349924e-01 1.02837467e+00 1.66908503e-01 2.14790791e-01 -2.33207181e-01 5.19630611e-02 1.84002504e-01 -4.87772673e-01 4.28330511e-01 8.03753078e-01 3.47190291e-01 8.80641043e-02 4.28590924e-01 1.57615447e+00 -5.45953751e-01 3.81988958e-02 -6.01108432e-01 1.95934400e-01 5.37666798e-01 1.13340247e+00 -7.15339601e-01 -1.90308914e-01 -2.67945051e-01 1.05761933e+00 5.38315713e-01 3.69563341e-01 -7.29016900e-01 -4.90292221e-01 4.69296724e-01 3.56686294e-01 4.37316567e-01 -5.35866737e-01 -5.20505309e-01 -9.64548349e-01 2.00331897e-01 -4.13771063e-01 -3.03467065e-01 -1.12213194e+00 -1.61249471e+00 7.54464626e-01 -1.75412640e-01 -1.42261577e+00 1.72478169e-01 -7.98366189e-01 -9.32084441e-01 1.03042126e+00 -1.55212760e+00 -1.35217810e+00 -4.08909261e-01 3.11746746e-01 2.26370677e-01 3.68764281e-01 7.23006546e-01 -3.06093954e-02 -1.45388231e-01 5.79486907e-01 1.02603488e-01 4.26782906e-01 3.83900732e-01 -1.38889635e+00 8.61783624e-01 5.07104039e-01 -3.04926306e-01 2.60250986e-01 2.18529671e-01 -7.17009604e-01 -1.66919279e+00 -1.29033291e+00 1.79685101e-01 -4.07867193e-01 2.69863993e-01 -4.03755546e-01 -1.10099304e+00 2.49506265e-01 7.96469003e-02 4.44711566e-01 -1.42681330e-01 -4.53370839e-01 -3.83423090e-01 1.84177468e-03 -1.41622615e+00 7.40590751e-01 1.24367225e+00 -3.85412157e-01 -3.36879343e-01 9.64789651e-03 9.09494042e-01 -9.96017635e-01 -1.12543392e+00 5.14040351e-01 5.47482014e-01 -9.18247700e-01 1.22482109e+00 -5.21516681e-01 7.68090129e-01 -3.76713336e-01 2.12157935e-01 -1.44922364e+00 -5.74140370e-01 -8.59226465e-01 -1.79370657e-01 7.46109724e-01 1.21685371e-01 -4.41850364e-01 9.35430765e-01 5.40635586e-01 -4.50332969e-01 -1.25346494e+00 -9.58190978e-01 -6.99375629e-01 8.33654106e-01 -2.65403748e-01 1.05906153e+00 9.39569473e-01 -5.98949552e-01 -3.22593778e-01 -1.21132523e-01 1.62972584e-01 7.32694209e-01 8.08402598e-01 6.41594946e-01 -1.60368705e+00 1.92707017e-01 -7.77356267e-01 -2.35293999e-01 -1.12552369e+00 5.87579571e-02 -8.58579934e-01 -1.66943848e-01 -1.75210202e+00 -1.48259059e-01 -6.85172081e-01 1.21474378e-02 4.19470966e-01 5.37202135e-02 6.62215948e-01 3.32136080e-02 -1.15775308e-02 -1.78814322e-01 7.52862751e-01 2.13636422e+00 -1.99937120e-01 -1.17753603e-01 -2.80295968e-01 -5.16220570e-01 8.60301375e-01 4.96864289e-01 -4.79515754e-02 -1.94165304e-01 -8.75127077e-01 4.39594895e-01 2.09732190e-01 7.19158113e-01 -7.77635336e-01 -2.37039834e-01 -3.22952867e-01 9.07664478e-01 -8.49702537e-01 3.67563397e-01 -8.97295773e-01 1.51563004e-01 2.40176946e-01 -7.88049772e-02 1.26697376e-01 3.57271135e-01 4.29907978e-01 2.34043792e-01 1.01911031e-01 1.00257230e+00 -3.22648376e-01 -1.30152538e-01 9.16851044e-01 9.84479934e-02 1.55470431e-01 7.44514763e-01 -2.91721851e-01 8.53705034e-02 -3.03258121e-01 -5.07862151e-01 2.55319029e-02 1.15369594e+00 2.12091923e-01 8.91021669e-01 -1.89405787e+00 -7.57622838e-01 3.39823157e-01 -2.25281313e-01 7.13452876e-01 1.04436569e-01 2.66051203e-01 -1.15739095e+00 -2.08214819e-01 -2.27904379e-01 -6.04898572e-01 -4.94309574e-01 3.82214338e-01 7.46783197e-01 1.16377614e-01 -9.90735173e-01 7.83699632e-01 2.95442432e-01 -6.65827692e-01 2.29138255e-01 -8.34366560e-01 2.78387308e-01 -4.92038518e-01 8.13111663e-02 8.73343870e-02 5.78924417e-02 -5.32616675e-01 -3.39081019e-01 1.15955126e+00 2.67699569e-01 2.38344669e-01 1.34494996e+00 1.28655270e-01 -4.78901528e-02 8.64927471e-02 1.21794808e+00 1.25338987e-01 -1.76534629e+00 6.36200421e-03 -8.07576656e-01 -5.12604713e-01 1.40051901e-01 -5.94386697e-01 -1.57608581e+00 1.02317798e+00 3.38693559e-01 4.02228147e-01 9.00809705e-01 -1.23389125e-01 1.20484567e+00 1.63740650e-01 4.12709922e-01 -6.33773565e-01 1.30941689e-01 5.21667659e-01 1.44848728e+00 -1.11817384e+00 1.31461054e-01 -5.51512957e-01 -1.38825715e-01 1.40435839e+00 6.43540680e-01 -7.56671071e-01 9.52806413e-01 1.93481430e-01 -3.37803513e-02 -4.68582243e-01 -1.21456809e-01 1.57455549e-01 8.59444439e-01 5.88751197e-01 2.39311516e-01 5.86311854e-02 2.33389601e-01 3.92172039e-01 -2.29236022e-01 -1.13232568e-01 3.82610410e-01 5.47709942e-01 -1.22172490e-01 -8.68440032e-01 -2.88950562e-01 3.42449129e-01 -1.50254712e-01 7.52028823e-02 -4.22388375e-01 8.41495156e-01 1.18224047e-01 2.87369937e-01 4.23967153e-01 -2.36883655e-01 4.58692044e-01 -2.89341986e-01 6.65201128e-01 -4.41987932e-01 -3.88648182e-01 1.74398404e-02 -2.37407759e-01 -8.03499758e-01 -1.46830678e-01 -2.77054638e-01 -1.40956080e+00 -5.30057609e-01 6.84410706e-02 -3.77728283e-01 3.28964680e-01 5.82667351e-01 6.44678295e-01 4.56748545e-01 8.61163437e-01 -1.64794111e+00 -4.10412580e-01 -5.74324906e-01 -4.65808839e-01 6.67915523e-01 4.11338687e-01 -9.32111621e-01 -1.24077387e-01 -3.66582349e-02]
[8.709382057189941, -3.600011110305786]
492ae2d3-c4ce-4cab-96cc-31978c24d00f
sfcnext-a-simple-fully-convolutional-network
2305.18771
null
https://arxiv.org/abs/2305.18771v1
https://arxiv.org/pdf/2305.18771v1.pdf
SFCNeXt: a simple fully convolutional network for effective brain age estimation with small sample size
Deep neural networks (DNN) have been designed to predict the chronological age of a healthy brain from T1-weighted magnetic resonance images (T1 MRIs), and the predicted brain age could serve as a valuable biomarker for the early detection of development-related or aging-related disorders. Recent DNN models for brain age estimations usually rely too much on large sample sizes and complex network structures for multi-stage feature refinement. However, in clinical application scenarios, researchers usually cannot obtain thousands or tens of thousands of MRIs in each data center for thorough training of these complex models. This paper proposes a simple fully convolutional network (SFCNeXt) for brain age estimation in small-sized cohorts with biased age distributions. The SFCNeXt consists of Single Pathway Encoded ConvNeXt (SPEC) and Hybrid Ranking Loss (HRL), aiming to estimate brain ages in a lightweight way with a sufficient exploration of MRI, age, and ranking features of each batch of subjects. Experimental results demonstrate the superiority and efficiency of our approach.
['Cheng Zhuo', 'Meng Niu', 'Tianbai Yu', 'Yalin Wang', 'Shunjie Dong', 'Yanyan Huang', 'Yu Fu']
2023-05-30
null
null
null
null
['age-estimation', 'age-estimation']
['computer-vision', 'miscellaneous']
[-8.70529935e-02 3.34906913e-02 7.29825720e-02 -8.34722102e-01 -3.19918752e-01 1.89196080e-01 4.05033857e-01 2.85329372e-01 -9.44630325e-01 7.05480337e-01 5.49307913e-02 -1.63601711e-02 -2.07975596e-01 -7.47539878e-01 -5.44268847e-01 -5.93685389e-01 -6.50223434e-01 7.26408720e-01 8.84458870e-02 1.50986761e-01 1.08600467e-01 4.76480842e-01 -1.40913129e+00 -3.11132371e-01 1.10465920e+00 1.42583084e+00 3.08434695e-01 4.54578340e-01 1.96253806e-01 5.50718129e-01 -2.56157160e-01 -5.24141848e-01 1.15385383e-01 1.23716407e-01 -6.11888111e-01 -4.47103351e-01 5.37383676e-01 -7.76692033e-01 -4.57943112e-01 9.66053009e-01 1.00715971e+00 -1.73849791e-01 6.49647415e-01 -1.25885272e+00 -4.55740273e-01 9.90078986e-01 -6.75001204e-01 4.64923650e-01 -3.71817768e-01 3.14840041e-02 5.30476153e-01 -7.08820641e-01 2.62641460e-01 9.07922983e-01 9.21214163e-01 9.42532182e-01 -7.81216085e-01 -1.04349232e+00 5.06327935e-02 5.07797360e-01 -1.09056759e+00 -3.04257184e-01 4.85013604e-01 -6.17830157e-01 2.65810311e-01 -2.03360081e-01 9.83465493e-01 1.26633334e+00 4.41286206e-01 5.36468565e-01 1.06037784e+00 1.48577601e-01 2.37434521e-01 -6.44055724e-01 2.08550349e-01 7.10351288e-01 3.81586730e-01 1.47229850e-01 -5.25072217e-01 3.88895310e-02 9.27901328e-01 6.18141815e-02 1.04314566e-01 -3.52056086e-01 -1.31389415e+00 5.65563500e-01 8.22586596e-01 2.26769820e-01 -5.67509234e-01 2.98241138e-01 6.21364713e-01 1.72116861e-01 8.42718244e-01 3.14270645e-01 -5.83626807e-01 9.65257132e-05 -1.17449939e+00 4.73003209e-01 -1.33603215e-02 5.42744756e-01 2.75182188e-01 9.22040492e-02 -3.27128768e-01 9.02956128e-01 -1.05925158e-01 4.53463346e-01 7.84287035e-01 -7.03913331e-01 2.28519440e-01 6.52868569e-01 -3.79390776e-01 -4.43844110e-01 -1.07816315e+00 -6.95315897e-01 -1.26637340e+00 4.69756983e-02 6.31528735e-01 -3.22176635e-01 -1.04285908e+00 2.08405972e+00 3.59241128e-01 -1.09824702e-01 -5.30586302e-01 9.16363180e-01 7.40317643e-01 4.16816138e-02 4.33333427e-01 -8.78240019e-02 1.41590416e+00 -6.86423063e-01 -1.56497926e-01 -4.17235523e-01 6.04570568e-01 -9.85586345e-02 7.17182279e-01 3.78911912e-01 -1.39520454e+00 -5.01703978e-01 -7.66831338e-01 -1.71119720e-01 -2.75960356e-01 3.31137061e-01 1.10076809e+00 5.98044157e-01 -1.17944169e+00 1.04798281e+00 -1.00435472e+00 -4.22664285e-02 9.82507110e-01 7.49790907e-01 -5.07606745e-01 -2.49170527e-01 -1.37709236e+00 6.80133939e-01 3.31248105e-01 4.65669155e-01 -1.19346821e+00 -1.34263408e+00 -6.25650406e-01 5.69686964e-02 -4.45948867e-03 -9.52139914e-01 9.73740518e-01 -7.44725466e-01 -7.97994971e-01 8.88148248e-01 2.95696706e-01 -6.62957191e-01 6.58821046e-01 -5.99767983e-01 -3.02237391e-01 2.54721195e-01 1.02185555e-01 1.02190042e+00 7.58614302e-01 -3.72891217e-01 -4.13907528e-01 -1.04070890e+00 -2.49591246e-01 -2.32276190e-02 -6.60820067e-01 3.66706043e-01 1.77338094e-01 -7.22834229e-01 7.48584569e-02 -4.03334022e-01 -5.64550638e-01 4.08699393e-01 -1.21833801e-01 -2.30603144e-01 1.80408210e-01 -1.18966985e+00 1.11506712e+00 -1.65796006e+00 1.90546602e-01 -3.60566825e-02 8.70617926e-01 1.37039512e-01 -2.40592092e-01 -3.40643376e-01 -3.69480669e-01 -2.05541402e-01 3.22158188e-02 -1.69227630e-01 -2.71716863e-01 -4.43165153e-01 4.66350913e-01 4.35142159e-01 6.87370822e-02 8.63005102e-01 -9.03260887e-01 -6.51416779e-01 -1.83388248e-01 5.30919373e-01 -4.19505566e-01 1.26411140e-01 1.14410967e-01 5.00810742e-01 -4.25148040e-01 6.04009151e-01 6.25551522e-01 -8.84812232e-03 -2.05219597e-01 -2.51806289e-01 -1.66280908e-04 -1.95521206e-01 -4.06639874e-01 1.84869707e+00 -4.12642479e-01 2.31241673e-01 -2.81614746e-04 -9.98883903e-01 1.02409720e+00 1.19609550e-01 7.83863485e-01 -8.66128385e-01 4.00665015e-01 3.19571793e-01 3.48231345e-01 -2.25687206e-01 -7.49923065e-02 4.31521870e-02 7.09926635e-02 4.66743737e-01 2.50326127e-01 3.73738348e-01 4.60328609e-01 1.39763981e-01 1.17760766e+00 -2.33815357e-01 -1.78627327e-01 -2.96936095e-01 5.93649447e-01 -7.31642425e-01 8.84041071e-01 4.27367210e-01 -5.04141271e-01 4.04007882e-01 5.87994635e-01 -1.03078020e+00 -1.48294711e+00 -1.04223990e+00 -1.35256484e-01 1.08107269e+00 -5.85771441e-01 -1.28455088e-01 -1.08818424e+00 -6.99645102e-01 -1.93590537e-01 1.80552498e-01 -8.25657308e-01 -4.87275124e-01 -5.85191667e-01 -1.16434824e+00 5.51644981e-01 9.44030404e-01 4.39670444e-01 -9.72947776e-01 -4.26995426e-01 4.42524515e-02 1.79835428e-02 -7.96627641e-01 -4.88873780e-01 3.33222337e-02 -1.27588654e+00 -9.70612168e-01 -1.39121115e+00 -7.09194422e-01 1.00844276e+00 -4.32122678e-01 1.06580913e+00 1.26702219e-01 -5.66790938e-01 2.24091094e-02 -1.44043356e-01 -4.42093045e-01 -9.34780017e-02 5.40296972e-01 4.75599557e-01 -1.19495742e-01 2.42179707e-01 -1.17406857e+00 -1.22854221e+00 5.70466258e-02 -5.63413560e-01 1.35649309e-01 9.62473869e-01 7.91953683e-01 2.58878976e-01 -2.93312848e-01 9.10524964e-01 -6.26068354e-01 3.59180033e-01 -3.89649332e-01 -5.32502234e-01 3.03617656e-01 -9.40229833e-01 5.43197468e-02 5.36116898e-01 -6.41820014e-01 -6.95504069e-01 -3.96688841e-02 -3.93324435e-01 -1.86853260e-01 -8.25272594e-03 3.45374256e-01 -8.63559246e-02 -5.32991886e-02 4.96439934e-01 2.04207093e-01 2.09376752e-01 -5.90078056e-01 5.66406883e-02 2.81935185e-01 6.08902037e-01 -7.00322688e-01 5.22034645e-01 4.59561013e-02 3.04807633e-01 -4.66000289e-01 -8.75694156e-01 4.28493842e-02 -8.44633400e-01 -4.10582542e-01 6.65610671e-01 -1.04635501e+00 -7.25072443e-01 1.13419676e+00 -8.94352913e-01 -4.61516589e-01 5.48509993e-02 6.82894707e-01 -2.79514432e-01 2.84954518e-01 -9.27715898e-01 -4.12456334e-01 -1.11898386e+00 -8.59684944e-01 8.93660486e-01 3.21642667e-01 4.97486778e-02 -7.93988347e-01 -1.24320619e-01 3.93660247e-01 6.63808584e-01 1.86413154e-01 1.25483346e+00 -5.85884690e-01 6.39821030e-03 -3.43034267e-01 -5.71209908e-01 3.65599036e-01 -3.22172314e-01 -4.10822809e-01 -6.35799110e-01 -3.82510334e-01 -4.18111205e-01 -5.23494542e-01 8.59158516e-01 9.22846735e-01 1.72197425e+00 3.90832573e-02 -2.42632173e-04 6.88053548e-01 1.14869130e+00 -1.55149564e-01 6.21826768e-01 1.38132736e-01 8.42152536e-01 5.91686308e-01 2.66780257e-01 6.17108464e-01 6.16246819e-01 1.38174146e-01 5.38202107e-01 -1.53963536e-01 6.98456634e-03 3.12224865e-01 1.15012497e-01 8.58142912e-01 -6.16099596e-01 4.27043229e-01 -1.01206589e+00 7.25740135e-01 -1.31927669e+00 -6.45066559e-01 -4.32239398e-02 2.19347382e+00 9.57438171e-01 2.48055011e-01 4.01578724e-01 -3.62171121e-02 8.75566185e-01 8.43013600e-02 -9.90831494e-01 -1.51825294e-01 3.26209905e-04 3.01971942e-01 5.85286200e-01 -3.79894763e-01 -9.83437121e-01 3.44961017e-01 6.24549341e+00 6.49333358e-01 -1.16367793e+00 2.68407106e-01 1.13360989e+00 -3.23985249e-01 5.37458137e-02 -4.14237082e-01 -6.51393592e-01 5.80338895e-01 1.04024327e+00 -2.78096795e-01 3.48397195e-01 8.64274859e-01 1.48053169e-01 -5.44854738e-02 -1.05339992e+00 9.71843362e-01 -2.90034413e-01 -9.55255806e-01 -7.83179477e-02 -3.99665497e-02 4.43284094e-01 1.51913792e-01 2.36617789e-01 3.20650697e-01 1.05541330e-02 -1.00739610e+00 6.54788375e-01 9.21387732e-01 1.01093662e+00 -8.82373095e-01 7.85714805e-01 1.90724447e-01 -8.93880486e-01 -4.59939063e-01 -3.95775378e-01 -1.14533894e-01 8.74812454e-02 1.19124448e+00 -4.26352203e-01 2.66961515e-01 1.01165617e+00 4.74806160e-01 -1.02124202e+00 1.14882636e+00 7.77711943e-02 3.37130994e-01 -7.43288994e-02 1.50532842e-01 2.16561761e-02 1.06139025e-02 -4.88433465e-02 5.31555593e-01 6.25972152e-01 -1.45556271e-01 -2.17726961e-01 7.39905894e-01 -2.96703517e-01 1.12001829e-01 6.86788708e-02 -1.24697298e-01 2.52593577e-01 1.60547841e+00 -7.08584905e-01 -1.20635711e-01 -1.62909284e-01 4.73960578e-01 5.72069228e-01 -2.23454535e-01 -5.59753478e-01 -2.54723370e-01 5.43054640e-01 4.35693085e-01 8.41037780e-02 -2.57561415e-01 -1.91287354e-01 -8.98970664e-01 -6.92279190e-02 -5.78952849e-01 2.09922418e-01 -7.91925848e-01 -1.43718159e+00 5.74094534e-01 -1.92299590e-01 -6.67491794e-01 -1.05634227e-01 -6.11318648e-01 -7.43097425e-01 7.74678588e-01 -1.26566184e+00 -1.35888851e+00 -3.29457581e-01 3.92095417e-01 2.07496852e-01 -2.54208982e-01 5.89878738e-01 7.55828738e-01 -9.01491284e-01 5.45087755e-01 -9.61557478e-02 3.88051718e-01 8.27201426e-01 -1.41156149e+00 4.66190696e-01 7.31021583e-01 -7.49476433e-01 5.46857178e-01 4.18149650e-01 -7.90456653e-01 -8.17863047e-01 -1.21130037e+00 1.08234286e+00 -9.35839042e-02 6.45410955e-01 -4.86941814e-01 -7.26441920e-01 3.48179311e-01 -3.33097756e-01 2.35297903e-01 4.97246444e-01 4.38597530e-01 -3.35839599e-01 -5.19300163e-01 -1.02579033e+00 4.99877483e-01 1.20705211e+00 -8.35678503e-02 -1.68330163e-01 2.84304440e-01 6.70772135e-01 -2.38997683e-01 -1.43103147e+00 6.25290751e-01 9.36906934e-01 -9.43151772e-01 1.06961763e+00 -5.77191591e-01 8.34359288e-01 2.88251698e-01 5.03604949e-01 -1.16646314e+00 -3.37585926e-01 -1.37494048e-02 -2.62796700e-01 1.16205513e+00 1.02784008e-01 -1.50375590e-01 9.55841064e-01 8.14381242e-01 -1.98553190e-01 -1.16452265e+00 -7.84117639e-01 -5.95892549e-01 4.24121648e-01 -2.29439288e-01 9.57061350e-01 5.06204426e-01 -5.91918111e-01 1.44894913e-01 -3.37894499e-01 2.86070220e-02 9.62193847e-01 -4.35453206e-01 1.90755367e-01 -1.74916399e+00 3.70814115e-01 -5.96121550e-01 -6.44805312e-01 -1.76015884e-01 1.49285957e-01 -6.14756525e-01 -1.59209698e-01 -1.44954348e+00 5.20465255e-01 -5.47843397e-01 -7.97121644e-01 2.97501355e-01 -2.10007504e-01 1.66126385e-01 -3.24857652e-01 -3.09549451e-01 -5.24762869e-01 6.06309175e-01 1.24409223e+00 -2.78072774e-01 3.01605105e-01 2.06615344e-01 -5.14977396e-01 7.25425303e-01 8.56548905e-01 -4.71211702e-01 -4.54781711e-01 -4.62886751e-01 4.77035940e-01 -1.52406111e-01 5.80024064e-01 -1.30698013e+00 1.33644119e-01 1.37824461e-01 1.05782890e+00 -6.14568532e-01 -2.73175426e-02 -3.56070817e-01 -7.39240870e-02 6.37674212e-01 -3.06086749e-01 1.11577436e-01 -3.05977613e-01 1.10510625e-01 3.55195068e-02 -1.12411343e-01 7.70892382e-01 -2.25054830e-01 -6.20336652e-01 1.20564413e+00 -1.80811435e-01 4.79791686e-02 7.26932526e-01 -4.98628765e-02 -1.38179794e-01 9.58766565e-02 -8.17503572e-01 4.53926533e-01 2.21101835e-01 3.96366805e-01 5.82935154e-01 -1.38268864e+00 -7.34369695e-01 -3.28379869e-02 4.73309383e-02 5.43625429e-02 9.11837578e-01 1.18169403e+00 -5.16702294e-01 4.31569070e-02 -9.31123078e-01 -2.14454904e-01 -1.02501273e+00 5.46798825e-01 3.10417026e-01 -6.72923803e-01 -4.63503838e-01 1.07763445e+00 3.20197165e-01 -4.54490006e-01 3.37191284e-01 -2.77392745e-01 -4.79414731e-01 2.18492642e-01 7.78463662e-01 5.85769713e-01 9.33589637e-02 -3.26448083e-01 -2.59641677e-01 3.04189026e-01 -5.50040364e-01 2.44835079e-01 1.92784202e+00 -5.98553047e-02 -2.84881055e-01 1.14792399e-01 7.99050033e-01 -8.16057503e-01 -1.44565880e+00 -2.42256328e-01 -2.19050739e-02 1.60334766e-01 2.97072053e-01 -7.15089500e-01 -1.65314424e+00 1.04461396e+00 1.09722161e+00 -5.07925570e-01 1.30451405e+00 -6.36539012e-02 1.14168060e+00 1.62679985e-01 4.15893883e-01 -1.28865802e+00 4.90733944e-02 1.62232399e-01 7.47296870e-01 -1.04731739e+00 1.82663143e-01 1.96976691e-01 -2.60385215e-01 1.22425103e+00 8.64338160e-01 3.75040360e-02 6.76861525e-01 -7.74730518e-02 -1.61391333e-01 -2.40646213e-01 -5.55120885e-01 1.23068959e-01 2.63283730e-01 8.21486831e-01 4.46250558e-01 2.84433402e-02 -5.79135835e-01 1.25700200e+00 -1.97568819e-01 1.52174711e-01 2.00507849e-01 4.64852661e-01 -3.79208535e-01 -1.04680371e+00 -8.37269947e-02 1.14286757e+00 -6.86365664e-01 -1.99462041e-01 1.03323404e-02 4.45287555e-01 4.04567659e-01 1.14488989e-01 1.51914150e-01 -2.57763177e-01 -1.04918117e-02 6.29566982e-02 6.38837576e-01 -2.81270683e-01 -4.91289288e-01 -4.52536076e-01 -4.28431965e-02 -6.42545164e-01 -3.24626774e-01 -7.58399248e-01 -9.41638350e-01 -2.94603527e-01 -7.23060817e-02 -2.10587502e-01 8.87856185e-01 8.42525363e-01 2.04662398e-01 6.87938273e-01 6.40610337e-01 -8.97529483e-01 -3.85102272e-01 -1.18542349e+00 -1.01800084e+00 1.67423487e-01 4.88470010e-02 -8.57140005e-01 1.43270105e-01 -1.15248166e-01]
[14.099427223205566, -1.5465842485427856]
a76b8f1a-5d96-4053-982e-be9ddfb5dd8b
large-scale-pedestrian-retrieval-competition
1903.02137
null
http://arxiv.org/abs/1903.02137v1
http://arxiv.org/pdf/1903.02137v1.pdf
Large-Scale Pedestrian Retrieval Competition
The Large-Scale Pedestrian Retrieval Competition (LSPRC) mainly focuses on person retrieval which is an important end application in intelligent vision system of surveillance. Person retrieval aims at searching the interested target with specific visual attributes or images. The low image quality, various camera viewpoints, large pose variations and occlusions in real scenes make it a challenge problem. By providing large-scale surveillance data in real scene and standard evaluation methods that are closer to real application, the competition aims to improve the robust of related algorithms and further meet the complicated situations in real application. LSPRC includes two kinds of tasks, i.e., Attribute based Pedestrian Retrieval (PR-A) and Re-IDentification (ReID) based Pedestrian Retrieval (PR-ID). The normal evaluation index, i.e., mean Average Precision (mAP), is used to measure the performances of the two tasks under various scale, pose and occlusion. While the method of system evaluation is introduced to evaluate the person retrieval system in which the related algorithms of the two tasks are integrated into a large-scale video parsing platform (named ISEE) combing with algorithm of pedestrian detection.
['Zhang Zhang', 'Da Li']
2019-03-06
null
null
null
null
['person-retrieval']
['computer-vision']
[-1.28791153e-01 -9.49553549e-01 1.80337340e-01 -4.02008921e-01 -9.80306268e-01 -5.25988042e-01 8.05422187e-01 1.35375530e-01 -8.61964107e-01 4.66287494e-01 4.77984659e-02 2.71329165e-01 -2.17432063e-02 -8.00004721e-01 -5.36540866e-01 -7.54793406e-01 9.16904807e-02 4.29478824e-01 8.23136449e-01 -1.88943759e-01 1.62563771e-01 4.56046343e-01 -1.83449435e+00 4.23443288e-01 3.65461648e-01 8.12742829e-01 3.64123583e-01 9.08026993e-01 1.08912975e-01 3.25112164e-01 -9.15501118e-01 -7.40492761e-01 2.26596177e-01 8.00583437e-02 -7.23643184e-01 4.98445518e-02 6.18649125e-01 -5.10117888e-01 -2.53845483e-01 1.21429312e+00 8.87155771e-01 9.70055610e-02 6.79816782e-01 -1.48561144e+00 -5.61076939e-01 -1.55242965e-01 -6.15583897e-01 5.50041616e-01 6.69845700e-01 2.12403879e-01 3.48963678e-01 -7.05352724e-01 6.68207645e-01 1.88984549e+00 4.76815581e-01 5.81369340e-01 -4.13779438e-01 -3.06317359e-01 5.30378260e-02 5.76260269e-01 -1.52345347e+00 -2.08060130e-01 3.53582412e-01 -4.47404861e-01 5.54954946e-01 4.41406339e-01 4.27508712e-01 9.70867574e-01 6.92016184e-02 8.80443037e-01 7.39237010e-01 -2.99764574e-01 1.04855932e-02 4.78048593e-01 7.03177989e-01 4.92875457e-01 5.50416112e-01 3.59773487e-02 -8.12684745e-02 -8.67833942e-02 3.87673944e-01 6.88045248e-02 -1.88480020e-01 -3.03611517e-01 -1.26739335e+00 5.59602857e-01 3.04780573e-01 1.66553363e-01 -3.75461072e-01 -1.19431235e-01 7.63122678e-01 2.06428319e-01 -3.58628556e-02 -1.16190903e-01 -3.33775073e-01 2.84294248e-01 -4.42496955e-01 6.05017841e-01 5.73719144e-01 1.27517152e+00 2.99210221e-01 -4.19969946e-01 -6.79093063e-01 9.35477138e-01 4.89823639e-01 1.05048418e+00 4.92180794e-01 -7.94568539e-01 4.62977499e-01 5.39015830e-01 4.96646285e-01 -1.20850122e+00 -2.91596353e-01 -1.61040083e-01 -6.56661630e-01 -1.29720256e-01 5.21206081e-01 2.11210810e-02 -7.41296709e-01 1.56700134e+00 6.42514825e-01 -2.83179522e-01 2.60352135e-01 1.32105756e+00 1.69168389e+00 8.90514791e-01 7.36556232e-01 -1.82678863e-01 2.14558220e+00 -1.09810388e+00 -5.06221056e-01 -2.18073651e-01 1.57070667e-01 -1.12140501e+00 6.39281690e-01 -1.05960980e-01 -8.12946200e-01 -1.04588270e+00 -7.95038640e-01 -3.98658738e-02 -7.70995438e-01 3.24345827e-01 1.46850199e-01 8.01406741e-01 -9.99489009e-01 -2.76772857e-01 -1.10163987e-01 -1.07628334e+00 3.92632000e-02 2.99370080e-01 -6.45684719e-01 -4.30872500e-01 -1.28822148e+00 9.37145472e-01 3.34400833e-01 5.89278229e-02 -7.52641261e-01 -1.94803536e-01 -5.49512804e-01 -2.36135945e-01 3.56278181e-01 -9.98830736e-01 9.24880087e-01 -6.96230412e-01 -8.21667612e-01 1.46300209e+00 -1.03527777e-01 -2.05102488e-01 6.42784894e-01 -2.20891058e-01 -5.62166929e-01 3.53960782e-01 5.43719113e-01 9.35653031e-01 7.55246937e-01 -1.13134396e+00 -9.72821355e-01 -6.84803665e-01 4.00408395e-02 4.49315250e-01 8.94241631e-02 8.84172916e-01 -1.16109943e+00 -2.74776965e-01 -3.84940922e-01 -7.46132791e-01 -1.53756076e-02 -1.56233400e-01 -2.86612217e-03 -6.20250046e-01 9.95407820e-01 -6.87324584e-01 7.69268513e-01 -1.92491436e+00 -8.37064087e-02 -8.13684165e-02 -2.12353677e-01 7.47777224e-01 -3.27107251e-01 1.43844262e-01 2.96855152e-01 -1.94180667e-01 2.94510275e-01 -1.11033492e-01 -1.77092984e-01 -1.33917689e-01 3.15335363e-01 3.76260430e-01 8.09838176e-02 8.64988387e-01 -7.60468185e-01 -1.15043247e+00 3.86042088e-01 3.77387285e-01 4.76110727e-02 2.94866890e-01 1.91373631e-01 1.95183784e-01 -7.44156003e-01 9.88938808e-01 9.15690184e-01 1.17304936e-01 -4.85430062e-01 -5.37911117e-01 -1.56770438e-01 -8.02774251e-01 -1.28849244e+00 1.29226542e+00 1.38705388e-01 4.93431509e-01 1.94636226e-01 -8.29701245e-01 8.05387437e-01 2.78831214e-01 1.40293062e-01 -7.15010226e-01 2.58689284e-01 7.30318874e-02 -2.99193203e-01 -8.56168866e-01 7.23536670e-01 5.95229208e-01 -2.42103055e-01 -2.31857568e-01 -5.09167323e-03 4.06074792e-01 5.39369106e-01 3.38996649e-01 7.74105608e-01 3.31806019e-02 3.82160515e-01 -4.98546392e-01 1.18087101e+00 1.96884990e-01 4.23655689e-01 9.44840014e-01 -8.19723606e-01 5.14400482e-01 6.67531341e-02 -7.18315244e-01 -1.05769718e+00 -1.00236118e+00 -2.51925826e-01 1.08158123e+00 6.54846549e-01 -1.38554841e-01 -9.71325159e-01 -4.91476774e-01 -1.54218540e-01 -1.04166381e-01 -3.48195523e-01 1.18954919e-01 -6.17089570e-01 -1.09054899e+00 5.76863885e-01 2.58683294e-01 1.15459096e+00 -1.18200076e+00 -8.36473286e-01 -1.10057175e-01 -4.16873008e-01 -1.48783183e+00 -4.99168724e-01 -8.50844443e-01 -2.43438706e-01 -1.52221286e+00 -1.26112652e+00 -1.21132016e+00 5.35582662e-01 6.55548871e-01 1.25991845e+00 2.43303135e-01 -6.24631464e-01 9.79638696e-01 -5.09270370e-01 -4.56528425e-01 -8.69128257e-02 -2.23232552e-01 1.97802950e-02 1.01277903e-02 8.53288829e-01 4.10161853e-01 -7.65562296e-01 7.22156525e-01 -7.31415451e-01 -3.14301789e-01 5.93179643e-01 6.61815047e-01 5.02096891e-01 1.21477731e-01 3.71739388e-01 -3.86606872e-01 6.53020680e-01 -1.65669382e-01 -7.57434428e-01 8.43493104e-01 -5.46372868e-02 -4.53283668e-01 1.14117488e-01 -3.82198960e-01 -1.02784467e+00 -1.76486164e-01 1.13955028e-01 -7.24534169e-02 -5.50479412e-01 -1.20719761e-01 -6.19843662e-01 -5.24850786e-02 4.33323860e-01 2.81249642e-01 -2.32677981e-01 -1.33959949e-01 4.07619812e-02 8.63703370e-01 7.24655092e-01 -6.21466339e-01 5.71143210e-01 6.82880357e-02 3.75443771e-02 -1.01779783e+00 -5.78450918e-01 -1.03955662e+00 -5.31943500e-01 -5.60351253e-01 1.30067098e+00 -1.11599278e+00 -8.47459137e-01 9.32838917e-01 -1.51688147e+00 3.89332086e-01 3.57600302e-01 4.13089424e-01 -1.10292323e-01 5.69301963e-01 -4.46702331e-01 -9.99489307e-01 -7.73959517e-01 -1.32756531e+00 1.53502190e+00 6.51807249e-01 4.77827966e-01 -4.83090639e-01 -3.07059050e-01 6.20150447e-01 2.61220843e-01 2.00800404e-01 5.39355516e-01 -6.96380854e-01 -7.93599486e-01 -5.83920062e-01 -7.88601398e-01 1.84707969e-01 -3.65033716e-01 1.29588507e-02 -7.56068349e-01 -2.89292663e-01 -2.76709199e-01 -7.05113560e-02 6.66825891e-01 4.99475539e-01 6.83954954e-01 4.40833792e-02 -5.88715315e-01 1.38415202e-01 1.47962010e+00 4.92250562e-01 8.75828207e-01 6.35342836e-01 3.87809157e-01 9.57609951e-01 1.06095159e+00 -2.65387557e-02 4.13831651e-01 9.58639324e-01 -1.01010064e-02 1.10690787e-01 -2.86121249e-01 8.94006565e-02 1.19778171e-01 2.48769373e-01 -2.59554923e-01 -5.02949119e-01 -8.89072657e-01 3.95348579e-01 -2.10693574e+00 -1.18180335e+00 -3.63341630e-01 2.29833555e+00 1.07735172e-01 -1.18384205e-01 3.27525645e-01 -1.55113786e-01 1.33550608e+00 -8.77109095e-02 -1.27313331e-01 2.68453788e-02 -2.69541740e-01 -4.58262026e-01 3.80313545e-01 1.54955253e-01 -1.57307804e+00 9.68377531e-01 5.85405159e+00 1.03434014e+00 -5.03946722e-01 1.32115960e-01 8.06114614e-01 4.62279350e-01 5.86744547e-01 -3.92387360e-01 -1.33883882e+00 6.30904913e-01 6.67912543e-01 9.28902552e-02 -6.81509897e-02 8.30523670e-01 -7.62138935e-03 -3.40743154e-01 -8.59460235e-01 1.34919715e+00 3.35494995e-01 -7.43855178e-01 4.03484136e-01 -3.63796115e-01 1.96371898e-01 -3.22074771e-01 -3.65429223e-02 5.54064453e-01 -1.32448390e-01 -5.12262464e-01 5.29443860e-01 6.92620754e-01 4.55832303e-01 -6.61756337e-01 1.44698024e+00 2.94980437e-01 -1.48390615e+00 5.12579903e-02 -7.67913938e-01 4.74294335e-01 2.40852535e-01 -1.38601050e-01 -1.90993875e-01 6.22034013e-01 1.30343413e+00 4.43216026e-01 -1.03299475e+00 1.46390009e+00 1.96387976e-01 -1.31974652e-01 -1.47395775e-01 -3.38019431e-01 7.72222355e-02 -1.77237317e-01 7.70758271e-01 1.49340868e+00 9.40319225e-02 3.64324898e-01 3.30243081e-01 2.00525060e-01 1.47615865e-01 4.01899546e-01 -5.72402298e-01 5.61863720e-01 5.35988331e-01 1.39584374e+00 -6.20067477e-01 -5.29463530e-01 -3.96218479e-01 8.85978162e-01 -2.58553833e-01 3.18895876e-01 -8.27909052e-01 -2.48410195e-01 3.35668296e-01 -5.98888546e-02 1.56142563e-01 1.15896985e-01 5.78183413e-01 -9.72975492e-01 2.29036674e-01 -8.51022780e-01 6.20015740e-01 -9.66854632e-01 -1.36894357e+00 6.91233873e-01 4.60427076e-01 -1.23371100e+00 -4.54300605e-02 -7.76747763e-01 -5.62183142e-01 8.27869952e-01 -1.45547235e+00 -1.60066199e+00 -7.30936408e-01 9.72332656e-01 8.68891120e-01 -7.41337180e-01 6.55534685e-01 6.40210450e-01 -8.19776177e-01 6.98795497e-01 -1.60096005e-01 3.41551036e-01 8.94095838e-01 -7.61427760e-01 5.26940729e-03 9.80462015e-01 -5.14056683e-01 4.33095723e-01 7.59898245e-01 -5.50527513e-01 -1.39105392e+00 -1.14869189e+00 6.87948346e-01 -6.94418788e-01 1.50446281e-01 -3.56759615e-02 -4.68740135e-01 1.64509326e-01 5.18049784e-02 1.27040103e-01 3.72832984e-01 -3.00149590e-01 -2.28027120e-01 -4.37890589e-01 -1.47576618e+00 2.94984698e-01 8.43506694e-01 -1.63325608e-01 -4.89184946e-01 4.45398748e-01 7.41360962e-01 -2.72305101e-01 -6.70382917e-01 4.68808174e-01 6.53603435e-01 -8.03613186e-01 1.68724370e+00 -5.91029942e-01 5.93269207e-02 -5.32556653e-01 -5.39133310e-01 -3.92024219e-01 -2.26798952e-01 7.55129978e-02 4.39620197e-01 1.62676716e+00 -1.00839529e-02 -4.62823510e-01 3.71663123e-01 8.26724350e-01 2.70171613e-01 -1.30679116e-01 -7.72018254e-01 -7.64050186e-01 -3.14097345e-01 5.54498173e-02 5.31473100e-01 3.38745713e-01 -7.19287872e-01 3.69811326e-01 -5.16305208e-01 5.24433792e-01 1.11012673e+00 -1.85307503e-01 9.96282399e-01 -1.28412080e+00 5.76898344e-02 -2.71463245e-01 -1.05037773e+00 -8.24847996e-01 -1.48246929e-01 -2.15758294e-01 2.88111828e-02 -1.56533289e+00 7.12509573e-01 -2.03069255e-01 -2.55513757e-01 -1.95717171e-01 -4.09828573e-01 2.41781011e-01 4.69614685e-01 5.21463215e-01 -1.34313309e+00 1.55173928e-01 1.03887653e+00 -4.41934913e-01 5.68270013e-02 2.30127946e-01 -3.19122583e-01 5.96989334e-01 5.21223783e-01 -2.35700831e-01 -8.68829936e-02 -5.36672831e-01 -2.94633150e-01 2.27093443e-01 8.22921872e-01 -1.11836004e+00 4.74879533e-01 1.42425880e-01 6.91660702e-01 -8.44194651e-01 4.67124909e-01 -8.21403503e-01 -2.79811379e-02 5.10697901e-01 -5.65535612e-02 4.79350924e-01 7.57791921e-02 6.72820330e-01 -3.81010830e-01 -4.55809921e-01 6.98576152e-01 -3.48112315e-01 -1.32020795e+00 4.06555563e-01 -1.49511725e-01 -1.48122206e-01 1.28035438e+00 -2.56999701e-01 -7.32703030e-01 -2.05992699e-01 -3.94404143e-01 5.34388840e-01 2.52714813e-01 6.59565687e-01 7.32450306e-01 -1.34610701e+00 -1.03557360e+00 -1.32863536e-01 4.68356103e-01 -2.44500130e-01 3.98076743e-01 4.50504869e-01 -7.35350370e-01 7.52204657e-01 -3.14279377e-01 -7.80548871e-01 -2.03032923e+00 8.24671566e-01 2.16941580e-01 -2.87462890e-01 -2.93356031e-01 6.08195364e-01 3.13249499e-01 -2.33490899e-01 3.52415681e-01 3.88408452e-01 -8.27177048e-01 -2.04979386e-02 9.47812021e-01 4.17226970e-01 -2.58335084e-01 -1.08381164e+00 -6.50329232e-01 9.44266856e-01 -8.84115547e-02 1.32185206e-01 8.51135075e-01 -3.19753319e-01 -2.57997394e-01 -1.76114544e-01 1.09455156e+00 -4.33799952e-01 -6.82880223e-01 -3.00764948e-01 -4.35195416e-02 -5.60570419e-01 -2.84802854e-01 -7.48542011e-01 -8.29128265e-01 6.47646010e-01 1.29577243e+00 1.32458275e-02 1.07323766e+00 5.67604601e-02 7.43320346e-01 7.30167627e-01 6.80154622e-01 -1.12231398e+00 -1.77576557e-01 2.26745307e-01 8.75750601e-01 -1.79251814e+00 8.43593180e-02 -4.47527856e-01 -6.78668380e-01 8.82777452e-01 7.75436938e-01 4.82917242e-02 5.01584768e-01 -2.84621775e-01 1.35978028e-01 -1.25323474e-01 -3.26481879e-01 -4.23432678e-01 4.87208903e-01 9.29843247e-01 3.99452671e-02 5.76518141e-02 -5.32129169e-01 4.72106695e-01 1.81297168e-01 -3.12950224e-01 -2.20266581e-02 6.66947901e-01 -6.64794922e-01 -1.00229633e+00 -9.75362122e-01 1.48894668e-01 -4.06016320e-01 2.29939878e-01 -2.56841630e-01 1.01114547e+00 3.10939133e-01 1.30123293e+00 -5.72176874e-02 -1.56558603e-01 5.50562382e-01 -2.31547162e-01 3.38593721e-01 5.97134270e-02 -4.96098310e-01 -1.01561122e-01 2.38863841e-01 -4.03510064e-01 -9.73093867e-01 -7.43519485e-01 -5.38401723e-01 -2.42929593e-01 -1.92959681e-01 2.17891559e-01 6.20290399e-01 8.16477537e-01 -2.40170471e-02 1.65672585e-01 2.74589270e-01 -5.51929593e-01 -1.81337625e-01 -8.08779478e-01 -2.66162097e-01 8.01554739e-01 5.13891689e-02 -6.28509521e-01 1.48290798e-01 6.33973479e-02]
[14.703261375427246, 0.8472479581832886]
41aa639a-656d-4113-80be-17af6dc1873d
capturing-pragmatic-knowledge-in-article
null
null
https://aclanthology.org/C16-1247
https://aclanthology.org/C16-1247.pdf
Capturing Pragmatic Knowledge in Article Usage Prediction using LSTMs
We examine the potential of recurrent neural networks for handling pragmatic inferences involving complex contextual cues for the task of article usage prediction. We train and compare several variants of Long Short-Term Memory (LSTM) networks with an attention mechanism. Our model outperforms a previous state-of-the-art system, achieving up to 96.63{\%} accuracy on the WSJ/PTB corpus. In addition, we perform a series of analyses to understand the impact of various model choices. We find that the gain in performance can be attributed to the ability of LSTMs to pick up on contextual cues, both local and further away in distance, and that the model is able to solve cases involving reasoning about coreference and synonymy. We also show how the attention mechanism contributes to the interpretability of the model{'}s effectiveness.
['Jad Kabbara', 'Jackie Chi Kit Cheung', 'Yulan Feng']
2016-12-01
capturing-pragmatic-knowledge-in-article-1
https://aclanthology.org/C16-1247
https://aclanthology.org/C16-1247.pdf
coling-2016-12
['grammatical-error-detection']
['natural-language-processing']
[ 2.60480344e-01 4.88432407e-01 -4.19578642e-01 -4.21094328e-01 -1.06343830e+00 -2.52363205e-01 5.78966737e-01 7.79565424e-02 -6.38418138e-01 6.88304424e-01 9.41939235e-01 -7.63103724e-01 -4.69871551e-01 -4.93010223e-01 -6.17942989e-01 -3.10507119e-01 1.85856193e-01 4.61625308e-01 -1.35410473e-01 -5.20992756e-01 4.52148080e-01 2.81647354e-01 -1.16793907e+00 7.68221915e-01 4.38967317e-01 8.78913343e-01 4.30693060e-01 4.65337664e-01 -1.33094102e-01 1.10515511e+00 -8.24137688e-01 -5.64897656e-01 -2.43664458e-01 3.28829512e-02 -1.33086729e+00 -5.63117683e-01 2.38818482e-01 -2.67653435e-01 -3.72497380e-01 6.29685521e-01 5.68159044e-01 2.76076168e-01 5.02085686e-01 -5.19513190e-01 -7.49057591e-01 1.13561571e+00 -2.65887558e-01 8.30937862e-01 4.47442353e-01 1.15029067e-01 1.65045273e+00 -5.76049089e-01 8.31007123e-01 1.34424078e+00 7.97943473e-01 5.31336427e-01 -9.54157650e-01 -2.22227409e-01 3.86729568e-01 4.65817779e-01 -8.47385347e-01 -7.53775477e-01 5.84230602e-01 -2.62033582e-01 1.67306328e+00 2.64775157e-01 3.78391176e-01 1.55275989e+00 1.55831665e-01 9.02163923e-01 9.30141509e-01 -6.34925127e-01 -1.55475855e-01 -1.60749719e-01 6.90479875e-01 4.53053236e-01 -1.49154305e-01 1.32148698e-01 -6.81263924e-01 -3.97177100e-01 3.77830863e-01 -4.03904676e-01 -3.98697734e-01 4.57174510e-01 -1.19223690e+00 8.36943150e-01 6.39604926e-01 7.38598466e-01 -5.70159733e-01 2.12133095e-01 4.92709756e-01 1.75670862e-01 2.39848495e-01 1.05786335e+00 -7.90292621e-01 -4.65148807e-01 -6.81546748e-01 1.28586873e-01 8.68296802e-01 7.27535605e-01 -1.92898083e-02 -1.14546023e-01 -5.16954482e-01 1.17433608e+00 2.63296872e-01 1.01217225e-01 7.59881318e-01 -1.26272285e+00 7.04771161e-01 1.21851213e-01 4.74099889e-02 -7.74839997e-01 -6.41585231e-01 -3.98424596e-01 2.47587706e-03 -4.81378585e-01 5.34238160e-01 -2.42836043e-01 -3.76259565e-01 2.00739193e+00 -4.00273979e-01 -3.16080302e-01 1.73960179e-01 6.39766276e-01 9.04937744e-01 5.77351749e-01 3.75211924e-01 -1.88078597e-01 1.50585139e+00 -8.31093788e-01 -9.47347760e-01 -6.55090868e-01 9.19026017e-01 -7.17222154e-01 1.23061574e+00 -1.00918502e-01 -1.26839650e+00 -2.56230444e-01 -9.20671582e-01 -4.36010838e-01 -2.51249403e-01 7.85270985e-03 6.16226554e-01 1.26745597e-01 -9.00632739e-01 8.12630057e-01 -3.98004532e-01 -4.83286798e-01 3.74433875e-01 2.92084724e-01 2.10764661e-01 3.90338987e-01 -1.51762748e+00 1.49398148e+00 1.92698613e-01 4.55097824e-01 -1.29217729e-01 -4.42562014e-01 -6.75319076e-01 2.87713736e-01 1.31233379e-01 -7.42964804e-01 1.62650287e+00 -9.31670606e-01 -1.54837084e+00 9.09563661e-01 -4.12347406e-01 -8.07199538e-01 1.40948623e-01 -1.99964628e-01 -4.24496680e-01 2.89512295e-02 2.86762603e-02 5.06248415e-01 3.54481548e-01 -8.08526337e-01 -4.60008651e-01 -2.01965615e-01 3.13684642e-01 2.93739706e-01 -2.11922869e-01 3.96704376e-01 7.08244927e-03 -6.67188942e-01 -1.61085621e-01 -9.75092530e-01 1.17691010e-01 -4.88162637e-01 -5.58520734e-01 -6.69435143e-01 3.71424168e-01 -9.18050706e-01 1.32373357e+00 -2.12971735e+00 1.36536926e-01 1.61977988e-02 -9.63150933e-02 1.24456458e-01 -1.37443498e-01 2.87823349e-01 -5.16207814e-02 4.61645007e-01 -2.64571887e-02 -3.54576290e-01 1.48115024e-01 2.04482540e-01 -5.20616174e-01 9.64272767e-02 1.71852186e-01 1.16044879e+00 -5.70223093e-01 -3.05968851e-01 -1.87574755e-02 3.17586452e-01 -3.85121912e-01 -2.18171448e-01 -3.83887976e-01 1.03040829e-01 -3.67371559e-01 3.58392268e-01 -1.58108428e-01 -3.43117207e-01 5.57434678e-01 3.20015959e-02 2.45231390e-02 1.41896582e+00 -3.73498738e-01 1.45527685e+00 -7.22348571e-01 1.07427514e+00 -5.12076691e-02 -6.45034611e-01 4.34464484e-01 5.00006258e-01 -9.58451703e-02 -8.89650762e-01 3.66440952e-01 5.23852348e-01 6.50375605e-01 -5.34889817e-01 4.97401386e-01 -1.95794165e-01 -1.80007935e-01 4.79570955e-01 -1.76504731e-01 3.33369583e-01 -1.74491942e-01 -1.11181006e-01 9.47134137e-01 1.13314942e-01 7.55272657e-02 -6.15645945e-01 2.45358273e-01 -2.15009883e-01 5.70268989e-01 1.02290571e+00 -8.45057666e-02 2.04417378e-01 6.52702749e-01 -4.40055579e-01 -6.71075046e-01 -4.93156999e-01 -3.65060449e-01 1.52084279e+00 -2.30025277e-01 -2.98165917e-01 -6.34401977e-01 -5.11589170e-01 -8.92616883e-02 1.53382432e+00 -7.03375459e-01 -2.25779638e-01 -9.38071311e-01 -7.30311096e-01 6.42381251e-01 8.11229110e-01 2.19386190e-01 -1.65122485e+00 -9.40765142e-01 2.75986701e-01 -4.79846418e-01 -1.24191022e+00 -1.46334186e-01 3.15635383e-01 -7.75464237e-01 -9.10556555e-01 -1.11359708e-01 -6.65421724e-01 -1.52578205e-01 -2.19739646e-01 1.27054989e+00 2.86456972e-01 2.91964561e-01 2.82768849e-02 -8.95006731e-02 -3.62214714e-01 -4.51460421e-01 5.49283862e-01 -2.77711779e-01 -6.86006129e-01 7.34241486e-01 -3.91022921e-01 -2.49566674e-01 -4.45981026e-02 -3.66854310e-01 -1.43742546e-01 4.88644481e-01 1.14185846e+00 3.88481878e-02 -7.12059140e-01 6.75073743e-01 -1.04829252e+00 1.14122951e+00 -4.68116313e-01 -1.37716308e-01 2.87529379e-01 -5.48963010e-01 3.26568365e-01 1.75015152e-01 -2.72439122e-01 -1.35333478e+00 -7.43653476e-01 -4.38661784e-01 1.14622727e-01 -3.82003076e-02 7.68112183e-01 5.17866649e-02 2.57140249e-01 4.58108068e-01 -3.14721346e-01 -1.20122619e-01 -3.43778640e-01 2.11108446e-01 8.34642649e-01 9.90648270e-02 -7.25736856e-01 -2.12614298e-01 -8.01571587e-05 -3.79823297e-01 -9.23881352e-01 -1.31762612e+00 6.61552623e-02 -3.00048351e-01 1.72905520e-01 1.01854336e+00 -5.05768776e-01 -8.67576063e-01 1.18777573e-01 -1.47788227e+00 -7.20018625e-01 -1.72675714e-01 4.04983044e-01 -6.35741472e-01 -2.62948293e-02 -1.32069254e+00 -6.00873649e-01 -3.54618818e-01 -1.24755549e+00 8.47326219e-01 -1.05668426e-01 -1.05113065e+00 -1.20011580e+00 -4.62287515e-01 5.89598715e-01 5.34401178e-01 -5.77798225e-02 1.50163257e+00 -1.13505876e+00 -2.07611799e-01 1.29228219e-01 -1.03747413e-01 -3.24716941e-02 -2.88001359e-01 -9.91219729e-02 -1.18008864e+00 2.16734737e-01 2.41898909e-01 -2.25512043e-01 1.00914228e+00 6.39958501e-01 1.18960655e+00 -4.70172793e-01 -4.62945819e-01 3.38433921e-01 9.42753971e-01 1.56612307e-01 5.76462090e-01 6.95796728e-01 4.39840138e-01 6.61691189e-01 4.30043161e-01 3.07905599e-02 4.79529142e-01 7.69542277e-01 -4.76642996e-02 1.96123615e-01 -4.26705778e-02 -1.38856545e-01 1.03614308e-01 7.37614214e-01 1.57922003e-02 -2.44477004e-01 -1.18119919e+00 6.03085101e-01 -1.94416595e+00 -1.04721737e+00 2.11647861e-02 1.81320405e+00 7.82434165e-01 5.95730782e-01 -9.33272988e-02 -2.51596887e-02 6.03859842e-01 4.46241498e-01 -2.51230627e-01 -1.10769999e+00 -2.78238416e-01 2.66666025e-01 4.05770093e-01 9.19639051e-01 -6.79326773e-01 1.13994181e+00 7.38822603e+00 6.35380924e-01 -9.92985845e-01 1.43420264e-01 6.87899947e-01 -2.36347377e-01 -4.25341398e-01 -2.25864545e-01 -6.63811505e-01 4.45933491e-01 1.27619720e+00 1.82592750e-01 2.88906097e-01 4.42658007e-01 5.84648028e-02 -4.48276550e-02 -1.25471544e+00 4.40465093e-01 9.20698494e-02 -1.48192990e+00 -3.42742912e-02 5.85518591e-02 3.46120864e-01 3.60048145e-01 4.11542393e-02 5.13224244e-01 1.46057054e-01 -1.10607791e+00 7.32770085e-01 5.08481026e-01 3.06746989e-01 -5.53291082e-01 9.14765656e-01 3.77669036e-01 -5.05413115e-01 -4.79085922e-01 -9.82595831e-02 -3.75470728e-01 2.29223877e-01 3.08708176e-02 -8.03838491e-01 -5.58398888e-02 6.69727087e-01 4.29047376e-01 -4.10432190e-01 4.64692891e-01 -5.11490107e-01 6.32198155e-01 -2.10896775e-01 -5.21817029e-01 4.94906396e-01 1.08456813e-01 7.17567623e-01 1.37246549e+00 3.58264660e-03 2.60784656e-01 -3.62420380e-01 8.86811912e-01 -2.37998858e-01 -9.25435722e-02 -5.10398149e-01 -2.73449346e-03 8.17108512e-01 7.07901537e-01 -2.59653062e-01 -3.29512537e-01 -2.49466941e-01 7.55656779e-01 7.34191179e-01 5.10940194e-01 -7.30411112e-01 -1.98869511e-01 4.86157537e-01 -1.54631421e-01 3.84802520e-01 2.14604009e-02 -5.60268164e-01 -8.20854902e-01 1.31506279e-01 -8.37280810e-01 6.05156779e-01 -9.98661637e-01 -1.17506838e+00 7.91477442e-01 -1.42716303e-01 -2.89299458e-01 -4.43634659e-01 -8.15994084e-01 -6.12831056e-01 8.87110889e-01 -1.55217552e+00 -9.99099374e-01 3.90736073e-01 6.36810139e-02 6.82789266e-01 -5.35152890e-02 1.09722102e+00 -8.89737904e-03 -5.74367523e-01 5.06722987e-01 -6.69968948e-02 8.24261308e-02 5.60101092e-01 -1.14393342e+00 3.99654210e-01 4.63944286e-01 5.25557064e-02 1.14974022e+00 8.45130563e-01 -3.34175318e-01 -8.56293440e-01 -4.70392793e-01 1.36748290e+00 -6.70462787e-01 8.10217559e-01 -9.02309716e-02 -7.82458663e-01 1.19728231e+00 5.98544180e-01 -4.00858313e-01 9.24603820e-01 9.34213579e-01 -4.28828210e-01 1.68268234e-01 -9.82108891e-01 7.42665946e-01 1.10910141e+00 -9.06657398e-01 -1.34254241e+00 2.83660263e-01 8.41992915e-01 -4.45685536e-01 -9.24486220e-01 3.99951577e-01 6.13897622e-01 -9.00525451e-01 7.59664834e-01 -9.54956949e-01 6.94248974e-01 5.16093552e-01 -2.33687997e-01 -1.15264571e+00 -3.53439361e-01 -5.78434229e-01 -7.78278410e-02 8.91477168e-01 1.04403389e+00 -6.96823001e-01 2.00625136e-01 9.32899475e-01 -2.80568808e-01 -1.08047414e+00 -1.08341146e+00 -2.35096797e-01 4.65179980e-01 -4.94400561e-01 5.98512888e-01 9.30256844e-01 4.83887881e-01 8.56998503e-01 -7.27685839e-02 -2.60433912e-01 -4.59480546e-02 2.71103024e-01 -1.19161859e-01 -1.08509433e+00 -3.21373552e-01 -9.18970942e-01 1.53282672e-01 -1.02139795e+00 8.53354156e-01 -6.80832088e-01 -4.64973375e-02 -1.68902910e+00 -1.06893154e-02 -3.60172451e-01 -3.95414174e-01 5.13216496e-01 -1.73972890e-01 -2.15738714e-01 3.21012229e-01 2.43411526e-01 -4.63709801e-01 2.84728467e-01 8.62899005e-01 6.81577250e-02 -1.63246974e-01 -6.85764477e-02 -9.47198570e-01 9.34829712e-01 9.95190084e-01 -2.62187630e-01 1.21160865e-01 -9.39282060e-01 4.92048621e-01 2.91226834e-01 1.58097133e-01 -4.75177407e-01 -2.75739655e-02 3.59745394e-03 1.62972435e-01 -1.94280401e-01 6.83182657e-01 -3.39963824e-01 -4.66946751e-01 3.73448133e-01 -1.11983097e+00 2.15166986e-01 4.38602000e-01 3.27628344e-01 8.31083208e-02 -4.30728078e-01 4.68935579e-01 -2.52867222e-01 -5.56409061e-01 -5.15387177e-01 -8.38967979e-01 1.90935716e-01 3.48883480e-01 1.87089831e-01 -6.81413114e-01 -3.98418695e-01 -7.89096653e-01 1.52504101e-01 8.08836445e-02 6.04036570e-01 2.73303539e-01 -1.07116783e+00 -2.86811948e-01 -2.49681190e-01 -2.87648030e-02 -4.65379834e-01 -1.01488382e-01 9.18930590e-01 -2.00579703e-01 7.98312724e-01 2.04726588e-02 -2.13145062e-01 -9.76268709e-01 3.90757143e-01 6.33889735e-01 -1.84618965e-01 -5.61160803e-01 9.39333081e-01 -3.70107681e-01 -4.68379259e-01 3.33555043e-01 -4.99462664e-01 -4.04993981e-01 1.29709706e-01 1.01974465e-01 2.28410676e-01 3.43728513e-02 -6.70248032e-01 -4.52956259e-01 1.30594611e-01 -1.46621019e-01 -1.98178008e-01 1.32406986e+00 -4.41624112e-02 -1.48265556e-01 9.13818538e-01 1.11996067e+00 1.16497405e-01 -5.74921548e-01 -2.61433125e-01 5.87403595e-01 1.24457151e-01 9.41279903e-02 -1.02810216e+00 -6.36664033e-01 8.14930022e-01 6.48447052e-02 2.00283453e-01 4.13201630e-01 6.13990538e-02 9.80544269e-01 8.03475857e-01 -1.43713802e-02 -1.43926942e+00 -2.07580209e-01 8.50726664e-01 8.51961374e-01 -9.28786516e-01 -1.45688921e-01 -7.62720481e-02 -6.81913972e-01 1.18251014e+00 4.70131189e-01 -5.44751510e-02 3.78900498e-01 1.38647124e-01 1.27558127e-01 -4.93578523e-01 -9.74556565e-01 4.32032831e-02 1.60824642e-01 -6.85272156e-04 8.96569371e-01 6.81017265e-02 -4.83740062e-01 6.61811054e-01 -6.39804900e-01 -3.94133061e-01 4.58687276e-01 6.12809241e-01 -2.94917792e-01 -8.13401341e-01 6.91795275e-02 6.98224068e-01 -7.54041076e-01 -4.66741860e-01 -5.86351633e-01 8.89288723e-01 -2.52780437e-01 9.81234789e-01 2.05976427e-01 4.30727005e-02 2.13566005e-01 4.28277493e-01 2.81181812e-01 -5.68141997e-01 -9.08855021e-01 -7.70593733e-02 8.49514723e-01 -6.29789889e-01 -4.26452368e-01 -7.84078538e-01 -1.16755903e+00 -8.60728994e-02 -5.02921402e-01 1.70288086e-01 3.83026779e-01 1.40563321e+00 5.14259934e-01 8.02186012e-01 6.30707517e-02 -3.89236569e-01 -9.11052465e-01 -1.31231439e+00 -1.45748198e-01 4.02352184e-01 4.16210711e-01 -5.67500710e-01 -3.83847386e-01 -5.65449715e-01]
[10.754324913024902, 9.063986778259277]
8738270d-cccb-47e3-8b69-c4d7561ada96
deep-learning-methods-for-sar-image
2012.05508
null
https://arxiv.org/abs/2012.05508v2
https://arxiv.org/pdf/2012.05508v2.pdf
Deep Learning Methods For Synthetic Aperture Radar Image Despeckling: An Overview Of Trends And Perspectives
Synthetic aperture radar (SAR) images are affected by a spatially-correlated and signal-dependent noise called speckle, which is very severe and may hinder image exploitation. Despeckling is an important task that aims at removing such noise, so as to improve the accuracy of all downstream image processing tasks. The first despeckling methods date back to the 1970's, and several model-based algorithms have been developed in the subsequent years. The field has received growing attention, sparkled by the availability of powerful deep learning models that have yielded excellent performance for inverse problems in image processing. This paper surveys the literature on deep learning methods applied to SAR despeckling, covering both the supervised and the more recent self-supervised approaches. We provide a critical analysis of existing methods with the objective to recognize the most promising research lines, to identify the factors that have limited the success of deep models, and to propose ways forward in an attempt to fully exploit the potential of deep learning for SAR despeckling.
['Luisa Verdoliva', 'Diego Valsesia', 'Giuseppe Scarpa', 'Giovanni Poggi', 'Enrico Magli', 'Giulia Fracastoro']
2020-12-10
null
null
null
null
['sar-image-despeckling']
['computer-vision']
[ 5.67904890e-01 -4.32051659e-01 3.46108973e-01 -4.96419638e-01 -7.49801993e-01 -3.12788039e-01 6.82082593e-01 -3.46954346e-01 -4.86939818e-01 3.59505713e-01 3.92881185e-01 -1.81025222e-01 -5.64735353e-01 -5.60284674e-01 -1.29265323e-01 -1.05541730e+00 -1.53726846e-01 2.13937119e-01 -2.31154710e-01 -3.73900533e-01 3.32766712e-01 9.20735538e-01 -1.30107975e+00 2.74727225e-01 5.55966258e-01 1.04066515e+00 3.40679705e-01 5.94660580e-01 5.03406227e-02 9.26564455e-01 -5.71223080e-01 3.78415138e-02 4.44121569e-01 -6.34391725e-01 -2.81329930e-01 1.87195390e-01 3.62660646e-01 -3.44511956e-01 -5.95221758e-01 1.07569933e+00 8.50934207e-01 -1.51649518e-02 4.68875527e-01 -4.04128730e-01 -5.73335171e-01 2.69270420e-01 -5.76159358e-01 5.96003115e-01 -4.68668729e-01 2.15008900e-01 5.12406707e-01 -9.49319363e-01 4.01853412e-01 9.10601556e-01 7.93277442e-01 3.31615329e-01 -9.06890869e-01 -4.94688988e-01 -5.01249611e-01 3.16660851e-01 -1.11508071e+00 -7.93844521e-01 9.41623390e-01 -5.65361440e-01 7.07374394e-01 5.17206527e-02 5.22740781e-01 7.26122737e-01 5.31985521e-01 7.05541372e-01 1.51973641e+00 -5.98280013e-01 2.25293189e-01 -4.44483817e-01 1.61854684e-01 1.64624721e-01 1.94768324e-01 6.34465396e-01 -4.16171938e-01 1.87794849e-01 6.76604450e-01 -1.38915896e-01 -2.57577807e-01 -2.66980916e-01 -1.05086219e+00 1.07147646e+00 6.59783661e-01 6.56177402e-01 -4.79221940e-01 -1.27629563e-01 2.39836901e-01 4.39325809e-01 6.55750692e-01 7.42833555e-01 -2.95310736e-01 2.07780033e-01 -1.44177985e+00 2.52259284e-01 3.74257773e-01 2.25099191e-01 5.95421314e-01 7.02558160e-01 -1.11710645e-01 8.39232862e-01 1.80295616e-01 8.74305964e-01 1.82873592e-01 -9.02261615e-01 8.59354511e-02 4.48400192e-02 1.29459307e-01 -1.05168247e+00 -5.34610033e-01 -8.91316116e-01 -1.31835508e+00 6.17248714e-01 1.69463158e-01 -3.76701862e-01 -1.27408719e+00 1.08213460e+00 -3.14237088e-01 -5.58285750e-02 3.92370462e-01 1.06844938e+00 8.16093922e-01 5.76134622e-01 -5.37217930e-02 -2.76319563e-01 8.59281301e-01 -5.92814445e-01 -7.61462510e-01 -7.75159001e-01 3.10088187e-01 -1.06039202e+00 3.06827098e-01 3.68521392e-01 -6.88548625e-01 -6.12519324e-01 -1.13786578e+00 2.46418700e-01 -1.21550031e-01 2.11849198e-01 5.76705456e-01 6.60430133e-01 -8.64109337e-01 6.08742118e-01 -7.70784438e-01 -8.09865668e-02 7.87888646e-01 9.79917720e-02 -7.61800185e-02 -4.29761589e-01 -1.08772123e+00 1.17057455e+00 1.92423183e-02 8.49794626e-01 -8.65875244e-01 -4.63075101e-01 -7.05889821e-01 -1.43475324e-01 1.50056452e-01 -4.59923953e-01 9.08605456e-01 -1.12574291e+00 -1.07226765e+00 9.53924298e-01 2.58137379e-02 -9.00874436e-01 3.18084598e-01 -4.16431397e-01 -7.02819824e-01 9.76194888e-02 3.70031595e-02 2.72277534e-01 1.29358733e+00 -1.10877872e+00 -3.10676306e-01 -6.08977735e-01 -5.30046284e-01 8.52226540e-02 7.37928376e-02 1.69590056e-01 1.10541753e-01 -7.50823081e-01 2.79552937e-01 -7.25658536e-01 -5.96927106e-01 -3.36028337e-01 9.93534550e-03 3.61857295e-01 9.47703242e-01 -6.56650841e-01 8.56772065e-01 -2.41540742e+00 -1.05837241e-01 -7.14911297e-02 1.62141025e-01 9.54468369e-01 -2.72538185e-01 4.31466013e-01 -2.24923283e-01 -3.12540323e-01 -6.49502039e-01 1.21492170e-01 -5.60073733e-01 2.44999956e-02 -5.84170580e-01 6.04228735e-01 2.56679207e-01 9.67283726e-01 -6.18280351e-01 8.73757154e-02 5.85568368e-01 4.29580450e-01 3.15259695e-02 1.74288124e-01 -1.03368104e-01 6.11951292e-01 -4.07084763e-01 5.23538113e-01 9.35881793e-01 6.00806884e-02 -1.12232290e-01 -4.80367243e-01 -4.99470800e-01 -1.50874808e-01 -8.24238062e-01 1.03178203e+00 -2.70272821e-01 1.12672532e+00 4.34550345e-01 -1.33795989e+00 1.16746819e+00 1.04309414e-02 5.30218959e-01 -9.60622072e-01 2.87046134e-01 3.60551268e-01 1.61988407e-01 -6.68398738e-01 3.31811339e-01 -6.13659680e-01 1.76626980e-01 1.49264798e-01 -3.13784212e-01 -4.20041293e-01 -1.73528224e-01 -1.29509091e-01 1.00300717e+00 -2.03184977e-01 3.98811668e-01 -1.44232586e-01 5.06507277e-01 4.82003957e-01 3.97627145e-01 9.97901738e-01 -2.48605043e-01 7.11267889e-01 -2.09744483e-01 -8.87449503e-01 -8.85609090e-01 -8.15547109e-01 -1.68378323e-01 3.73768508e-01 -6.09597638e-02 3.49337786e-01 -3.54246676e-01 -2.22376987e-01 -2.82488763e-01 5.67162395e-01 -3.63585055e-01 -8.63058213e-03 -6.95984960e-01 -1.14243114e+00 3.67997646e-01 3.86061877e-01 1.08088815e+00 -1.18753982e+00 -7.15856969e-01 3.93877149e-01 -3.06331608e-02 -1.13200152e+00 1.92309305e-01 3.64890039e-01 -1.12043214e+00 -9.63316858e-01 -9.36020374e-01 -6.53416216e-01 3.63255858e-01 5.53563416e-01 9.01239336e-01 -2.90962398e-01 -6.03700936e-01 2.57655829e-01 -3.95692259e-01 -5.49497843e-01 -3.17565888e-01 -2.55609035e-01 -1.16150320e-01 3.26525494e-02 7.39372492e-01 -3.89439791e-01 -4.43107426e-01 -6.15547150e-02 -1.10654664e+00 -2.12021172e-01 1.09591615e+00 1.03365362e+00 2.59240896e-01 2.58860916e-01 6.57169342e-01 -8.21425378e-01 4.40303773e-01 -1.15614660e-01 -7.29009390e-01 -9.85155851e-02 -4.85746920e-01 -5.93330152e-02 4.89134669e-01 9.27536041e-02 -1.15253067e+00 1.20762214e-01 -3.41238648e-01 -1.50034845e-01 -4.43129808e-01 8.81461978e-01 1.89921275e-01 -5.61061561e-01 8.78524482e-01 4.66380566e-01 2.03819498e-01 -5.47813892e-01 1.12662494e-01 7.80757368e-01 5.45335174e-01 2.82541543e-01 1.02309573e+00 7.20933437e-01 2.40926981e-01 -1.55383694e+00 -1.25117099e+00 -4.89207447e-01 -5.90764880e-01 -2.97913134e-01 8.17419052e-01 -1.01136959e+00 1.15348116e-01 8.74255419e-01 -7.97434509e-01 -3.66279721e-01 -2.24703163e-01 6.30954981e-01 -3.93372118e-01 4.62174058e-01 -4.06747162e-01 -7.37398207e-01 -4.29300249e-01 -8.69081020e-01 7.19081163e-01 3.33057493e-01 1.79069638e-01 -9.03642714e-01 1.56169280e-01 5.67219973e-01 9.34725463e-01 9.41981226e-02 7.41684139e-01 -2.83406764e-01 -6.59416676e-01 -2.22810701e-01 -4.58233804e-01 9.02797341e-01 6.38345331e-02 -6.21187329e-01 -9.97623861e-01 -3.39455813e-01 6.07805073e-01 -1.94444478e-01 1.33132863e+00 9.21887934e-01 6.61427796e-01 1.35290712e-01 -1.28439009e-01 8.07873309e-01 1.60125887e+00 2.68208832e-01 9.32806134e-01 5.40928543e-01 2.87231952e-01 4.09797579e-01 6.59671605e-01 1.87727585e-01 -4.97774065e-01 3.33476484e-01 5.20299673e-01 -5.75354099e-01 -4.23162133e-01 4.41214412e-01 5.59369363e-02 5.40679932e-01 -1.36111110e-01 -8.24308246e-02 -8.63782763e-01 6.23236835e-01 -1.44420290e+00 -1.09810495e+00 -3.14995795e-01 1.68234551e+00 1.93880409e-01 -1.43746778e-01 -5.75434268e-01 1.56299114e-01 4.24672961e-01 7.69139051e-01 -3.89889389e-01 -2.18794774e-03 -5.27594507e-01 5.69737971e-01 8.10359716e-01 4.04234201e-01 -1.37719858e+00 9.91826653e-01 7.27057600e+00 5.62773824e-01 -1.31886947e+00 -2.56533742e-01 4.53568429e-01 3.74361396e-01 1.60603806e-01 -2.37879068e-01 -5.71529150e-01 1.29538581e-01 7.03404307e-01 8.78874809e-02 1.94639042e-01 6.50221348e-01 6.13347650e-01 -3.44258636e-01 -4.08322692e-01 9.37316716e-01 2.89342076e-01 -1.56908119e+00 -1.52318791e-01 -6.57654777e-02 8.61653745e-01 4.51429695e-01 2.75476247e-01 1.07458673e-01 1.06721692e-01 -1.00596380e+00 4.14604247e-02 6.10154867e-01 5.99516392e-01 -5.68726957e-01 1.14143157e+00 2.97696471e-01 -5.61478674e-01 -1.89103618e-01 -5.26962578e-01 -4.06897634e-01 3.08162361e-01 1.15188146e+00 -6.08138263e-01 5.79358637e-01 6.16196573e-01 9.78767991e-01 -3.99457037e-01 1.15188003e+00 -1.97552428e-01 7.19325304e-01 -2.80769356e-02 2.85808206e-01 5.66317141e-01 -3.82501036e-01 8.08482051e-01 1.17465210e+00 3.83393258e-01 3.27981591e-01 -8.77971053e-02 5.84268391e-01 3.52048129e-01 -3.76901269e-01 -6.81388736e-01 -3.68464768e-01 -4.47813496e-02 1.08649659e+00 -6.62340462e-01 1.64312273e-02 -4.36585546e-01 7.04437077e-01 -3.72833222e-01 4.26588118e-01 -3.62708598e-01 -3.12682092e-01 6.04339957e-01 2.48621464e-01 4.57845271e-01 -8.15279961e-01 -5.57129025e-01 -8.59742224e-01 -3.22821170e-01 -9.87615228e-01 1.60493344e-01 -8.59200835e-01 -1.42173839e+00 6.72456205e-01 -1.94567591e-01 -1.21471381e+00 -4.59460802e-02 -7.62078404e-01 -3.89803678e-01 1.02841818e+00 -1.84902275e+00 -1.02367842e+00 -4.38176841e-01 3.50983649e-01 6.35856628e-01 -6.44208014e-01 8.01154077e-01 2.68443882e-01 -1.43438596e-02 -3.44491392e-01 4.99618381e-01 3.46888363e-01 6.47862315e-01 -5.62817156e-01 3.81879807e-01 1.19539034e+00 1.17750004e-01 1.73196197e-01 9.92328405e-01 -5.65618038e-01 -1.24428642e+00 -1.00818825e+00 8.87842417e-01 -9.48065743e-02 5.32401443e-01 1.36812478e-01 -7.32281685e-01 5.43134093e-01 2.38150675e-02 3.49458456e-02 4.10653174e-01 -3.55600983e-01 9.34300423e-02 -3.62387747e-01 -9.05505478e-01 3.63582402e-01 5.59991121e-01 -1.88713327e-01 -8.26636493e-01 2.15646937e-01 -9.50834528e-02 -1.70150802e-01 -1.73615292e-01 7.98714995e-01 1.56450108e-01 -1.26500010e+00 1.14291120e+00 -1.79323763e-01 4.25364912e-01 -3.27074111e-01 -1.10404491e-01 -1.33539057e+00 -6.80419147e-01 -2.35542595e-01 3.05560678e-01 5.87449729e-01 7.43625984e-02 -5.69098771e-01 9.04290020e-01 -7.26864710e-02 -3.16226929e-01 -3.12923759e-01 -8.65887702e-01 -8.26941192e-01 -3.79373096e-02 -2.17406273e-01 3.54192145e-02 7.13266373e-01 -9.59158063e-01 3.66641462e-01 -7.47912943e-01 2.76740223e-01 1.03592551e+00 2.76873052e-01 5.54784775e-01 -1.33723164e+00 4.61663194e-02 -2.58276582e-01 -4.45396960e-01 -1.13946974e+00 -8.67571086e-02 -5.96353769e-01 3.04611146e-01 -1.71486390e+00 -1.83492780e-01 -2.69451022e-01 -3.03881685e-03 1.40533403e-01 1.53510004e-01 6.07070386e-01 5.06408662e-02 4.56633568e-01 -1.76414214e-02 4.53154594e-01 1.13258600e+00 -3.01364660e-01 -6.64724112e-02 3.06146353e-01 -7.26942658e-01 8.32803071e-01 1.04683125e+00 -4.09785986e-01 -4.06569690e-02 -8.32244396e-01 7.29396343e-02 -9.73656178e-02 5.59219182e-01 -1.38395393e+00 3.59740287e-01 -1.07162513e-01 7.54646182e-01 -7.80830503e-01 2.73097873e-01 -8.70475411e-01 4.98678908e-02 4.00432378e-01 -8.30877945e-02 -3.91018540e-01 1.40349090e-01 5.95636368e-01 -7.19043970e-01 -2.69423515e-01 1.29769528e+00 -3.34234238e-01 -1.21883261e+00 5.50407618e-02 -7.61556745e-01 -5.87826073e-02 7.27792442e-01 -1.99306816e-01 1.53802102e-02 -4.01352912e-01 -7.73902297e-01 -2.09378138e-01 -9.16886106e-02 1.73555836e-01 7.33121812e-01 -7.56930053e-01 -8.87479544e-01 4.36654985e-01 -1.66799814e-01 -3.39528263e-01 5.26651740e-01 8.41441870e-01 -8.07211757e-01 5.49876630e-01 -3.72463256e-01 -4.78101283e-01 -1.23318887e+00 3.72712463e-01 5.03729284e-01 -2.76786715e-01 -9.22555506e-01 7.47969449e-01 -6.42960891e-02 5.44267613e-03 -6.20387979e-02 2.42640570e-01 -3.24564934e-01 1.83887407e-02 7.48673320e-01 4.40627903e-01 3.58457953e-01 -5.87234139e-01 -3.32363844e-01 8.13657403e-01 -6.47441065e-03 -1.97272748e-02 1.77739549e+00 -8.27575997e-02 -1.12554938e-01 6.99667782e-02 8.26662302e-01 -1.31788850e-01 -1.04969537e+00 -4.65783715e-01 7.05872327e-02 -4.32908475e-01 7.55920649e-01 -9.47115183e-01 -1.05131710e+00 1.06348085e+00 8.18630397e-01 1.07079804e-01 1.55050528e+00 -2.55714685e-01 6.94996536e-01 3.67925525e-01 1.95115119e-01 -1.04771936e+00 -4.84529287e-02 8.68503630e-01 8.60028565e-01 -1.43222582e+00 2.69543201e-01 -1.87892690e-01 -4.36421394e-01 1.34357357e+00 -1.03486538e-01 -2.54700422e-01 7.83357441e-01 5.57814360e-01 5.70099771e-01 -5.98571956e-01 -4.84905913e-02 -4.34054017e-01 2.82921013e-03 7.67026484e-01 3.97622168e-01 -1.47674546e-01 -2.79324949e-01 1.64454177e-01 1.56170741e-01 3.46465886e-01 4.29663062e-01 8.57738435e-01 -9.68156219e-01 -9.04394746e-01 -6.40134215e-01 6.36789441e-01 -3.75484496e-01 -2.28359759e-01 -2.69185036e-01 6.50743544e-01 7.08460510e-02 9.96542096e-01 -8.29430968e-02 -6.07337654e-02 1.87710315e-01 -1.00436136e-01 3.22892785e-01 -4.86351818e-01 -3.88407439e-01 2.41816491e-01 -1.57442596e-02 -1.68016106e-01 -7.24694312e-01 -6.03672385e-01 -5.72395980e-01 -6.13766909e-02 -1.66331664e-01 4.24049720e-02 5.73227525e-01 1.13262486e+00 1.21666141e-01 4.70098883e-01 5.46932757e-01 -9.60616052e-01 -5.78933477e-01 -9.59877431e-01 -8.54339361e-01 -1.76678430e-02 6.30643845e-01 -2.72322536e-01 -7.47438788e-01 1.84577063e-01]
[10.412603378295898, -2.218968152999878]
400737f9-1ddb-4526-a4e5-af14df52fbe8
visual-policy-learning-through-multi-camera
2303.07026
null
https://arxiv.org/abs/2303.07026v1
https://arxiv.org/pdf/2303.07026v1.pdf
Visual-Policy Learning through Multi-Camera View to Single-Camera View Knowledge Distillation for Robot Manipulation Tasks
The use of multi-camera views simultaneously has been shown to improve the generalization capabilities and performance of visual policies. However, the hardware cost and design constraints in real-world scenarios can potentially make it challenging to use multiple cameras. In this study, we present a novel approach to enhance the generalization performance of vision-based Reinforcement Learning (RL) algorithms for robotic manipulation tasks. Our proposed method involves utilizing a technique known as knowledge distillation, in which a pre-trained ``teacher'' policy trained with multiple camera viewpoints guides a ``student'' policy in learning from a single camera viewpoint. To enhance the student policy's robustness against camera location perturbations, it is trained using data augmentation and extreme viewpoint changes. As a result, the student policy learns robust visual features that allow it to locate the object of interest accurately and consistently, regardless of the camera viewpoint. The efficacy and efficiency of the proposed method were evaluated both in simulation and real-world environments. The results demonstrate that the single-view visual student policy can successfully learn to grasp and lift a challenging object, which was not possible with a single-view policy alone. Furthermore, the student policy demonstrates zero-shot transfer capability, where it can successfully grasp and lift objects in real-world scenarios for unseen visual configurations.
['Wu Ya', 'Alp Tekirdağ', 'Kuluhan Binici', 'Cihan Acar']
2023-03-13
null
null
null
null
['robot-manipulation']
['robots']
[-3.59140001e-02 -2.68197179e-01 -2.34679341e-01 5.35460822e-02 -3.24452788e-01 -9.01160717e-01 3.92938495e-01 -1.84906334e-01 -4.37168360e-01 8.37996781e-01 -5.82558513e-01 -1.23206891e-01 -4.83397773e-04 -3.56653422e-01 -1.17409265e+00 -1.07026732e+00 1.54042035e-01 1.51851237e-01 4.15524244e-01 -6.67053312e-02 3.43810529e-01 8.51128042e-01 -1.64687145e+00 4.13727500e-02 5.98261952e-01 7.53543079e-01 9.75186288e-01 7.23518431e-01 3.79145503e-01 7.49002755e-01 -5.37575662e-01 1.60272673e-01 6.91477001e-01 1.24320343e-01 -2.95849085e-01 3.74261469e-01 6.40067279e-01 -7.25812793e-01 -4.23535198e-01 1.06036079e+00 5.10658443e-01 5.47792852e-01 3.83773863e-01 -1.39603400e+00 -7.05625594e-01 1.53219681e-02 -7.74309158e-01 2.24065542e-01 3.34270805e-01 5.87515116e-01 5.61816514e-01 -7.48239815e-01 4.97945547e-01 1.29545939e+00 1.93788782e-01 9.11064565e-01 -1.10516226e+00 -6.51158392e-01 5.72518766e-01 1.46715328e-01 -9.37874258e-01 -1.33345217e-01 8.15101206e-01 -4.44173425e-01 8.81615222e-01 -1.40373319e-01 7.08852232e-01 1.21655786e+00 1.96022332e-01 1.01818907e+00 1.29276037e+00 -3.86369169e-01 2.30432764e-01 3.82881254e-01 -4.21269953e-01 7.92188168e-01 3.00766975e-01 7.34080791e-01 -3.20650995e-01 3.32774892e-02 1.10668778e+00 1.33381844e-01 -5.43137431e-01 -1.16566122e+00 -1.12888694e+00 6.43913686e-01 6.48430049e-01 -1.86486199e-01 -4.34586287e-01 1.85531050e-01 2.25631014e-01 2.28325650e-01 -1.14155389e-01 5.67900419e-01 -4.36314732e-01 9.04096663e-02 -2.67479599e-01 2.20369101e-01 4.26325023e-01 1.40595818e+00 4.50381130e-01 5.15127838e-01 2.09383234e-01 6.87644184e-01 2.05111250e-01 7.36982763e-01 4.12103236e-01 -1.07256889e+00 6.57366812e-01 4.83927041e-01 4.55847561e-01 -8.29949379e-01 -1.92783803e-01 -2.05302462e-01 -3.18664312e-01 1.00440419e+00 3.61182749e-01 -2.16061771e-01 -9.84278679e-01 1.78063214e+00 5.60947478e-01 -2.19561383e-02 4.17159379e-01 1.25104392e+00 3.63014132e-01 5.37195981e-01 -4.39992808e-02 -2.33559936e-01 9.34959233e-01 -1.02158821e+00 -3.42848927e-01 -4.87905741e-01 6.13702945e-02 -7.53416717e-01 1.22738886e+00 6.54645443e-01 -1.11821938e+00 -7.21489727e-01 -1.05129349e+00 4.51436549e-01 -7.02930540e-02 3.29481333e-01 4.49624687e-01 2.40578741e-01 -7.02505052e-01 4.00515676e-01 -1.02674317e+00 -4.34679508e-01 2.29705781e-01 5.42141736e-01 -5.58757186e-01 -2.61168182e-01 -6.16726696e-01 1.15742695e+00 6.02951646e-01 -7.25558400e-02 -1.57416403e+00 -4.38508183e-01 -6.21617854e-01 -2.52784863e-02 8.54730010e-01 -4.71877486e-01 1.23199177e+00 -8.49293947e-01 -1.85987210e+00 3.84388745e-01 1.69126317e-01 -5.03709130e-02 4.44789320e-01 -4.90626663e-01 1.65586174e-01 4.60656106e-01 -1.55315489e-01 7.93663800e-01 1.11120164e+00 -1.93607724e+00 -8.76782537e-01 -4.53405321e-01 3.75855207e-01 7.44726419e-01 -1.71825364e-01 -4.06390399e-01 -1.88151628e-01 -4.90852147e-01 -1.93998218e-02 -1.24680352e+00 -3.57810631e-02 1.55995637e-01 1.48448631e-01 -9.13178101e-02 1.28264928e+00 -2.24041060e-01 3.37909907e-01 -2.00189090e+00 4.57969189e-01 -4.95168157e-02 -2.81392932e-01 5.87775469e-01 1.19407773e-02 4.94748861e-01 1.23829648e-01 -4.38927025e-01 3.60662282e-01 2.73650855e-01 -4.67385739e-01 5.10372221e-01 -4.75776523e-01 5.55425763e-01 -2.80694887e-02 5.50235748e-01 -1.04145920e+00 -6.74134269e-02 4.63916421e-01 2.50007957e-01 -6.36138976e-01 5.96801519e-01 -3.63748997e-01 8.57818842e-01 -4.94978875e-01 4.74894881e-01 3.94114584e-01 -2.06860173e-02 1.41402826e-01 -3.63351218e-02 -1.59742720e-02 -5.25267124e-01 -1.11389983e+00 1.61106873e+00 -5.74770391e-01 4.49849278e-01 2.97042251e-01 -1.02007580e+00 1.02293324e+00 2.53213048e-01 3.00694048e-01 -2.95829624e-01 1.04252204e-01 -2.76852027e-03 -4.78642294e-03 -8.11261356e-01 2.88715482e-01 -1.20968528e-01 2.08594486e-01 9.50370207e-02 1.89871058e-01 -4.00682837e-01 -1.32323757e-01 -1.29910395e-01 6.47649467e-01 5.91606677e-01 3.87935638e-01 1.79912020e-02 5.37719250e-01 5.11168353e-02 4.91689056e-01 7.08296955e-01 -2.71976531e-01 7.47115165e-02 -1.41870007e-01 -3.71224642e-01 -1.10214484e+00 -1.37117445e+00 3.20918411e-01 1.16132593e+00 5.75245082e-01 3.58183771e-01 -2.88982183e-01 -7.55762339e-01 3.93173844e-01 6.14082515e-01 -2.57669955e-01 -4.59261477e-01 -7.62903929e-01 -1.16179109e-01 1.19081795e-01 8.28803062e-01 5.04894733e-01 -1.01092184e+00 -1.34319365e+00 3.64050362e-03 1.24490120e-01 -1.32682014e+00 -3.20725769e-01 7.78300241e-02 -9.11158979e-01 -1.28749239e+00 -6.58942342e-01 -9.59650517e-01 9.66570139e-01 7.37816691e-01 4.21388537e-01 -2.06224084e-01 -3.10031742e-01 9.38194275e-01 -3.39322507e-01 -4.39046830e-01 -3.58756393e-01 -5.05554974e-01 5.59600770e-01 -2.85784036e-01 -7.27588162e-02 -4.51095343e-01 -6.12437725e-01 4.34211880e-01 -6.89898133e-01 -2.22002313e-01 8.48669291e-01 1.05506659e+00 3.29441756e-01 -1.41409993e-01 4.88949478e-01 -1.67999730e-01 6.32265985e-01 -1.23073258e-01 -1.07281995e+00 3.36479247e-01 -5.75270116e-01 -3.61852124e-02 9.54639435e-01 -1.04064596e+00 -1.10754907e+00 3.35643381e-01 6.45673037e-01 -1.17479110e+00 -3.21494639e-01 3.80550250e-02 5.86817004e-02 -4.54900682e-01 6.11924112e-01 4.41760004e-01 2.10197538e-01 -1.65714011e-01 4.27017242e-01 6.67935789e-01 6.77326381e-01 -8.71833563e-01 8.41712117e-01 4.73821640e-01 8.05856958e-02 -8.60277891e-01 -5.94525814e-01 -4.99433786e-01 -6.51708603e-01 -6.00694478e-01 6.49883330e-01 -1.06399357e+00 -1.27771902e+00 3.95365655e-01 -9.24083114e-01 -3.57444316e-01 -1.66588593e-02 8.52807403e-01 -9.44799781e-01 4.09678280e-01 -2.58140415e-01 -9.41178799e-01 -7.29243234e-02 -1.36619103e+00 7.82569528e-01 5.89157522e-01 3.33797574e-01 -8.03721547e-01 -3.44737440e-01 4.02734160e-01 1.59191847e-01 1.65201277e-01 7.61308670e-01 -4.04599160e-01 -6.53573811e-01 -4.04221676e-02 1.00487590e-01 5.33434927e-01 2.51344889e-01 -2.22866714e-01 -7.97094584e-01 -1.21155500e+00 9.12116617e-02 -7.33803451e-01 3.60776126e-01 1.78852677e-01 9.67158675e-01 -2.50580311e-01 -3.51899445e-01 3.88929576e-01 1.52235329e+00 8.01196814e-01 1.27708599e-01 3.39067966e-01 5.83167791e-01 2.14930534e-01 8.55707645e-01 3.16580951e-01 5.28448783e-02 6.55525446e-01 8.09302509e-01 2.39466995e-01 1.51192337e-01 -3.92282248e-01 5.52741647e-01 4.79567677e-01 -2.78165400e-01 -1.55334011e-01 -6.88499451e-01 4.31647897e-01 -1.91522241e+00 -9.36802566e-01 5.91337681e-01 2.38666868e+00 4.51149493e-01 -7.22532254e-03 -3.70807536e-02 -1.91744655e-01 6.90764189e-01 -2.17278138e-01 -1.02666926e+00 -3.79604906e-01 3.48627716e-01 -2.80108213e-01 5.96282601e-01 3.67180616e-01 -8.82971466e-01 1.08442509e+00 5.65703344e+00 3.28472942e-01 -1.42620122e+00 -2.11415723e-01 -2.87415653e-01 -2.72959471e-01 4.89654601e-01 -9.30030346e-02 -7.57520437e-01 2.20050529e-01 1.90648109e-01 3.67232300e-02 8.14836264e-01 1.25033343e+00 1.61304787e-01 -4.80937809e-01 -1.16231370e+00 9.27237451e-01 2.78351367e-01 -8.20740044e-01 1.36056840e-01 -9.25180763e-02 7.44010687e-01 -1.37692153e-01 2.60138214e-01 4.44003850e-01 5.09306788e-01 -5.61900079e-01 5.35739064e-01 2.57732511e-01 5.98347843e-01 -7.20872700e-01 3.07726383e-01 8.83697033e-01 -8.63331616e-01 -8.07619214e-01 -5.37279844e-01 2.22476106e-02 -3.89552228e-02 -4.02411133e-01 -1.20250297e+00 4.62311149e-01 8.23513031e-01 4.02010918e-01 -1.53345644e-01 1.01849508e+00 -4.17480797e-01 1.05181754e-01 -9.79929268e-02 -1.37192398e-01 4.26486969e-01 -1.01185545e-01 8.47548783e-01 5.55615664e-01 1.79173201e-01 2.64200866e-01 8.05780947e-01 4.80371863e-01 2.64161557e-01 -2.55469948e-01 -1.00925863e+00 2.60941833e-01 4.81795758e-01 9.61186528e-01 -3.44736516e-01 -1.57658547e-01 -4.07068878e-01 8.50660741e-01 6.49702013e-01 5.92095792e-01 -8.51022065e-01 -4.34845462e-02 3.99037749e-01 -2.56257147e-01 5.99737167e-01 -4.88385379e-01 3.41730356e-01 -1.10336339e+00 1.36090219e-01 -8.73370051e-01 1.03220887e-01 -1.14404559e+00 -1.04092824e+00 2.94991732e-01 3.58278066e-01 -1.48913085e+00 -1.60765052e-01 -8.70425045e-01 -3.12312573e-01 5.47141314e-01 -1.46798897e+00 -1.10136628e+00 -4.94187206e-01 7.68832624e-01 8.56891215e-01 -6.68083608e-01 5.89143157e-01 -2.81640649e-01 -3.10907066e-01 3.90482694e-01 2.69664794e-01 -1.14801101e-01 7.31988668e-01 -1.17570221e+00 -4.54069048e-01 6.75738096e-01 -7.52296820e-02 5.73962092e-01 6.68026030e-01 -5.24286032e-01 -1.75946474e+00 -8.66534948e-01 -4.85694319e-01 -9.77050290e-02 4.19754922e-01 -1.04772605e-01 -7.73696125e-01 7.98388004e-01 2.69150257e-01 1.67966187e-01 1.75959721e-01 -4.03005600e-01 -1.12601064e-01 -2.51695305e-01 -1.29209065e+00 6.68035030e-01 6.62915409e-01 -1.83183834e-01 -8.22107077e-01 2.57782161e-01 4.58998352e-01 -7.97995865e-01 -7.21648753e-01 5.28642893e-01 6.33916974e-01 -6.47921681e-01 1.09960568e+00 -1.01474130e+00 3.04863989e-01 -3.21000278e-01 -1.31615415e-01 -1.72364593e+00 -2.40877122e-01 -4.78223145e-01 -2.00407222e-01 6.45343423e-01 -8.81725736e-03 -5.74493468e-01 7.30907679e-01 1.89347133e-01 -1.12001702e-01 -8.18847716e-01 -8.85458767e-01 -1.18440354e+00 1.15757130e-01 2.58689791e-01 -3.71690914e-02 5.70173264e-01 1.43586457e-01 1.72208592e-01 -4.89906698e-01 7.06028819e-01 5.03924429e-01 2.43778974e-01 9.82997239e-01 -8.44633818e-01 -4.03722465e-01 3.20144085e-04 -3.61483127e-01 -1.13940239e+00 2.19132558e-01 -7.20329881e-01 2.91372597e-01 -1.28709233e+00 1.97138026e-01 -4.14667517e-01 -1.75825104e-01 3.90422195e-01 -2.06994534e-01 -4.20122892e-01 6.19114935e-01 3.11937958e-01 -2.55542517e-01 6.60738826e-01 1.82368302e+00 -4.65025119e-02 -3.05423409e-01 2.91789889e-01 -1.38620675e-01 7.31545746e-01 9.61804211e-01 -2.27886707e-01 -9.08904791e-01 -7.04843104e-01 -2.87554681e-01 4.25337464e-01 5.80708265e-01 -1.08215761e+00 3.05246919e-01 -5.97589433e-01 5.43322444e-01 -2.33562231e-01 6.99728429e-01 -1.23822463e+00 -3.85129869e-01 7.49040484e-01 -2.30968058e-01 3.37278336e-01 5.38832068e-01 1.01865828e+00 1.36466548e-01 -2.21698627e-01 1.10094512e+00 -4.58340555e-01 -8.02276194e-01 2.93899067e-02 -3.78102809e-01 -1.40553758e-01 1.81107330e+00 -1.40271887e-01 -4.34030652e-01 -3.65136445e-01 -7.39627123e-01 4.24426973e-01 5.24196327e-01 6.64626181e-01 1.01128232e+00 -1.15641975e+00 -1.88335821e-01 2.65711010e-01 -4.81956750e-02 5.93102723e-02 2.71118637e-02 5.62186360e-01 -3.75891894e-01 4.41368073e-01 -6.60993934e-01 -8.77528608e-01 -1.47122669e+00 1.17261159e+00 2.25129485e-01 2.29722813e-01 -7.35489905e-01 5.67107916e-01 1.75251186e-01 -3.78260642e-01 5.58708906e-01 -2.61803448e-01 -9.67385247e-02 -5.42312086e-01 1.63015217e-01 2.78020501e-01 -3.68578494e-01 -2.23599896e-01 -9.29977298e-02 7.11614609e-01 -2.19895869e-01 -1.39870778e-01 1.18360853e+00 -6.66524544e-02 4.04392421e-01 3.18478435e-01 7.38054156e-01 -2.27768928e-01 -1.85734046e+00 -1.88304976e-01 -4.20407951e-01 -7.84562171e-01 -1.03830159e-01 -9.16244686e-01 -1.07901287e+00 1.00383103e+00 6.55496657e-01 -3.11434388e-01 9.49349225e-01 -2.27698475e-01 3.92317325e-01 8.64418089e-01 7.45935500e-01 -1.17573893e+00 6.80481434e-01 3.20197254e-01 1.10202730e+00 -1.39304316e+00 -2.94532161e-02 -2.49664485e-01 -9.98282135e-01 1.20536518e+00 1.37729383e+00 -3.58615458e-01 1.88792601e-01 9.97026414e-02 9.74806547e-02 5.18585928e-02 -8.33309710e-01 3.06759179e-02 4.16848771e-02 7.85458744e-01 -4.47644323e-01 -8.35253000e-02 3.55501086e-01 -1.41706735e-01 1.78943485e-01 -3.74877825e-02 7.20600724e-01 1.46254981e+00 -8.04061472e-01 -6.13744020e-01 -5.19428372e-01 -7.07606003e-02 -9.28219333e-02 5.38509548e-01 -1.49796784e-01 1.15148926e+00 -3.93251628e-02 8.68763983e-01 -2.49676615e-01 -4.04774487e-01 6.40279591e-01 -1.50344580e-01 1.03368580e+00 -6.17679656e-01 -3.04893583e-01 1.04995593e-01 -4.60271150e-01 -5.03808975e-01 -3.92963439e-01 -5.55162191e-01 -1.17347527e+00 1.04036108e-01 -4.92320687e-01 -1.16951622e-01 5.96775949e-01 8.07066441e-01 1.06548816e-01 3.97628248e-01 8.18949640e-01 -1.01092935e+00 -1.33730435e+00 -6.63117707e-01 -2.07546711e-01 3.66729885e-01 4.31386083e-01 -1.17074263e+00 -2.71935284e-01 4.77799736e-02]
[4.6678266525268555, 0.6970402002334595]
977504e7-b87c-4791-b8df-6583fcd921b7
zet-speech-zero-shot-adaptive-emotion
2305.13831
null
https://arxiv.org/abs/2305.13831v1
https://arxiv.org/pdf/2305.13831v1.pdf
ZET-Speech: Zero-shot adaptive Emotion-controllable Text-to-Speech Synthesis with Diffusion and Style-based Models
Emotional Text-To-Speech (TTS) is an important task in the development of systems (e.g., human-like dialogue agents) that require natural and emotional speech. Existing approaches, however, only aim to produce emotional TTS for seen speakers during training, without consideration of the generalization to unseen speakers. In this paper, we propose ZET-Speech, a zero-shot adaptive emotion-controllable TTS model that allows users to synthesize any speaker's emotional speech using only a short, neutral speech segment and the target emotion label. Specifically, to enable a zero-shot adaptive TTS model to synthesize emotional speech, we propose domain adversarial learning and guidance methods on the diffusion model. Experimental results demonstrate that ZET-Speech successfully synthesizes natural and emotional speech with the desired emotion for both seen and unseen speakers. Samples are at https://ZET-Speech.github.io/ZET-Speech-Demo/.
['Eunho Yang', 'Sung Ju Hwang', 'Wooseok Han', 'Minki Kang']
2023-05-23
null
null
null
null
['text-to-speech-synthesis', 'speech-synthesis']
['speech', 'speech']
[-3.85384001e-02 6.72535419e-01 2.26267755e-01 -6.17401600e-01 -9.42478776e-01 -6.37113750e-01 5.97833633e-01 -5.35970986e-01 4.76617999e-02 5.36806464e-01 2.60493279e-01 -3.68123323e-01 7.50567555e-01 -4.49476510e-01 -3.87769789e-01 -6.12191141e-01 3.01348448e-01 5.01324117e-01 -3.02601218e-01 -4.74702448e-01 -5.54201484e-01 2.73117512e-01 -1.20129335e+00 1.40312359e-01 9.76200104e-01 7.19808817e-01 -3.27128619e-02 1.16966784e+00 -1.49220660e-01 8.22033167e-01 -1.16667414e+00 -6.16521001e-01 -2.94040814e-02 -7.28479445e-01 -5.48898101e-01 1.30787864e-01 2.79746838e-02 -3.50976437e-01 -4.60492581e-01 1.08331144e+00 7.96676576e-01 5.79354227e-01 8.22281480e-01 -1.60216463e+00 -1.03131557e+00 4.50888097e-01 1.01751737e-01 -1.84798792e-01 5.71573436e-01 5.29101968e-01 6.29558563e-01 -1.04304981e+00 5.84455669e-01 1.53611493e+00 1.36367574e-01 1.57682896e+00 -9.80023801e-01 -8.24287951e-01 1.06840946e-01 -2.36266792e-01 -9.95372295e-01 -1.01967525e+00 8.89711320e-01 -2.36108512e-01 8.19487512e-01 5.97832739e-01 5.30213952e-01 1.96437526e+00 -2.94430684e-02 1.07847035e+00 1.05021250e+00 -3.27368379e-01 6.01586163e-01 4.44377810e-01 -3.27441305e-01 3.21541131e-01 -9.09486592e-01 3.09353709e-01 -4.11439478e-01 -1.10192038e-01 1.25185415e-01 -4.55410719e-01 -2.15851292e-01 2.38507763e-01 -6.94462597e-01 9.35973108e-01 -8.12770650e-02 1.98316112e-01 -4.28524852e-01 -1.21734418e-01 6.27197087e-01 6.07861996e-01 9.47504878e-01 3.26877505e-01 -1.31258965e-01 -4.80117410e-01 -7.06194103e-01 2.36871820e-02 9.21829462e-01 1.08780873e+00 2.22771451e-01 9.32368040e-01 -3.70531231e-01 1.08914149e+00 2.33087055e-02 1.03073978e+00 5.85359573e-01 -1.09111476e+00 1.32451177e-01 -7.74292201e-02 1.74389914e-01 -4.51729923e-01 8.55175108e-02 -5.81086054e-02 -7.69526362e-01 3.41318488e-01 -1.50395915e-01 -9.89651322e-01 -1.00669467e+00 1.92440832e+00 5.15777767e-01 2.89059132e-01 8.47997308e-01 8.92768383e-01 1.07321024e+00 1.26313865e+00 2.05381855e-01 -2.82002270e-01 1.05984890e+00 -1.14293110e+00 -1.14705670e+00 -5.27947247e-01 3.43682557e-01 -7.22205937e-01 1.58831239e+00 1.47624820e-01 -1.14688182e+00 -3.00558925e-01 -7.80567646e-01 1.84448436e-01 -3.18586081e-01 -2.59113431e-01 1.36582181e-01 8.49757373e-01 -1.09191573e+00 -8.90310854e-03 -6.27665162e-01 -1.24614432e-01 1.73614286e-02 5.03872335e-02 -2.62093525e-02 2.96614885e-01 -1.67674410e+00 8.35371256e-01 -8.97061378e-02 -1.40610993e-01 -1.24400699e+00 -3.17609400e-01 -1.13554478e+00 1.50451615e-01 2.89720088e-01 -4.68418717e-01 1.91853452e+00 -1.61613226e+00 -2.29588294e+00 5.41287720e-01 -3.62548739e-01 -4.04602587e-01 5.26470363e-01 4.54483218e-02 -7.90065289e-01 3.10669780e-01 -1.04568966e-01 8.91782045e-01 1.36389971e+00 -1.48984957e+00 -1.88983008e-01 1.33518755e-01 -2.67437667e-01 4.54049140e-01 -5.29401720e-01 1.75350726e-01 -8.01810548e-02 -8.54723692e-01 -7.59755850e-01 -1.01109147e+00 -1.48399547e-01 -2.23043531e-01 -6.58957064e-01 -4.34798509e-01 1.07962835e+00 -6.56761765e-01 8.19381773e-01 -2.43461847e+00 8.66934583e-02 -3.62803563e-02 -1.42628983e-01 5.29313982e-01 -3.92228156e-01 4.15785819e-01 -1.17479794e-01 1.49529472e-01 -1.86372429e-01 -7.68920720e-01 3.89710128e-01 2.94587970e-01 -6.79247618e-01 4.53977957e-02 2.98782319e-01 8.62132728e-01 -1.03931212e+00 -3.68798107e-01 4.22015578e-01 6.84929669e-01 -3.50231618e-01 6.46711528e-01 -5.46877980e-01 7.52328694e-01 -5.96103728e-01 4.56209987e-01 3.07960629e-01 3.72489333e-01 -2.22282901e-01 4.60748076e-01 1.34101644e-01 2.24130750e-01 -5.26190698e-01 1.32482159e+00 -8.35736752e-01 6.94045663e-01 3.28477025e-01 -4.55060363e-01 9.88460183e-01 9.41328645e-01 1.43011242e-01 -5.88640332e-01 4.24617171e-01 1.19932212e-01 -1.40025467e-01 -5.78591406e-01 5.90715587e-01 -7.10230410e-01 -3.93760413e-01 4.10923779e-01 5.07882014e-02 -5.42744219e-01 -2.93449342e-01 2.79073864e-01 7.27963030e-01 -3.40026200e-01 -1.16215013e-01 2.72032440e-01 2.26301417e-01 -4.30343747e-01 3.06182683e-01 3.84583831e-01 -5.07787585e-01 4.05020863e-01 2.81657636e-01 3.16653907e-01 -1.03231061e+00 -1.09161758e+00 3.27126682e-01 1.21944642e+00 -5.19517884e-02 7.24945813e-02 -1.20108247e+00 -5.85795462e-01 -4.65057135e-01 1.57597494e+00 -5.51906586e-01 -5.25140285e-01 -4.22231816e-02 9.07383934e-02 9.03601468e-01 8.36481974e-02 7.09788948e-02 -1.58241057e+00 -2.42306709e-01 2.44955629e-01 -3.78689200e-01 -9.64101553e-01 -1.07600570e+00 -1.96154434e-02 -1.80578083e-01 -1.56622842e-01 -1.07564604e+00 -7.30408788e-01 5.27442992e-01 -1.50892019e-01 7.27905810e-01 -3.53796601e-01 1.70452550e-01 6.76099241e-01 -5.25132000e-01 -5.04371822e-01 -1.34402955e+00 -1.00303538e-01 2.18630239e-01 2.71063298e-01 2.04665571e-01 -4.51654524e-01 -3.07254404e-01 1.79007232e-01 -9.24360335e-01 1.54343657e-02 1.12159923e-01 9.07021046e-01 9.86042991e-02 -9.79175568e-02 1.12559986e+00 -5.88672042e-01 1.22371101e+00 -4.94499177e-01 -1.87011082e-02 3.10948491e-01 -1.73288897e-01 -1.38736874e-01 1.06262994e+00 -9.72912312e-01 -1.46062863e+00 -2.58706897e-01 -7.24721014e-01 -9.18855906e-01 -4.38698113e-01 1.57218322e-01 -4.32890922e-01 3.51726353e-01 6.81339264e-01 3.85863066e-01 -3.01897954e-02 1.73072532e-01 7.64597654e-01 1.36307085e+00 5.57435155e-01 -5.78418672e-01 7.69287825e-01 1.33682052e-02 -8.26657236e-01 -1.04514456e+00 -6.42890990e-01 -9.71626788e-02 1.50635660e-01 -6.32340848e-01 8.73140275e-01 -9.19736505e-01 -6.15621626e-01 5.70668876e-01 -1.16012156e+00 -8.79823625e-01 -5.42565167e-01 2.23913923e-01 -7.60339379e-01 1.51558399e-01 -7.36143887e-01 -1.38480031e+00 -6.77646697e-01 -1.13936198e+00 1.17333269e+00 4.23404455e-01 -6.57031178e-01 -9.18977201e-01 1.23452898e-02 3.02834839e-01 3.59568238e-01 1.59273878e-01 4.18616742e-01 -8.37717354e-01 1.91010232e-03 -8.82463679e-02 5.75248778e-01 7.57367253e-01 1.76546320e-01 1.71573862e-01 -1.16114533e+00 -2.03882352e-01 2.33059123e-01 -8.27920735e-01 1.79135665e-01 1.25712380e-01 7.87276924e-01 -7.56055117e-01 1.36133894e-01 3.42826486e-01 6.01038456e-01 8.15431535e-01 4.53114122e-01 -2.97493070e-01 3.28383863e-01 7.72934258e-01 8.15034509e-01 5.02526104e-01 7.15771690e-02 3.58567506e-01 1.46973610e-01 -2.13359177e-01 -1.22197762e-01 -4.78704542e-01 8.05084407e-01 1.10703421e+00 6.72207654e-01 -9.14639056e-01 -5.48757493e-01 6.09172463e-01 -1.39953125e+00 -1.20304966e+00 3.14480484e-01 1.79041839e+00 1.02880824e+00 3.34095396e-02 -6.34649247e-02 -2.04941452e-01 8.74097645e-01 4.20373827e-01 -1.03444541e+00 -1.04637814e+00 -7.01083615e-02 2.18031898e-01 -9.95974988e-02 9.98823702e-01 -7.40651369e-01 1.33660305e+00 5.30401468e+00 1.12827063e+00 -1.38528657e+00 2.30364993e-01 9.05167103e-01 -2.66981632e-01 -6.67449057e-01 -5.35055459e-01 -3.41608346e-01 5.87017119e-01 1.42383671e+00 -6.73498511e-01 6.78926528e-01 9.65642571e-01 6.46494269e-01 5.19523859e-01 -8.38264644e-01 6.95000470e-01 1.62191346e-01 -6.04675055e-01 -8.33795741e-02 -2.75161088e-01 7.46053338e-01 -1.83095321e-01 5.27473748e-01 7.03638792e-01 5.79977393e-01 -9.76336956e-01 8.58624578e-01 2.61951029e-01 1.27491999e+00 -9.74997401e-01 1.99306205e-01 4.90402222e-01 -6.85287118e-01 1.69715390e-01 1.33658797e-01 3.75038683e-01 4.76480603e-01 1.52263150e-01 -1.04190755e+00 5.97199425e-02 3.00916314e-01 2.42067426e-01 2.01866537e-01 1.22367427e-01 -4.37841266e-01 9.08402622e-01 -2.40857467e-01 -4.00565028e-01 3.62044692e-01 -1.09660000e-01 9.41333950e-01 1.21758664e+00 5.90417147e-01 3.74190599e-01 9.91019085e-02 7.59318411e-01 -4.07095522e-01 9.63040069e-02 -8.03865910e-01 -3.95894408e-01 6.98391438e-01 1.11500180e+00 -1.47817045e-01 -6.03043795e-01 -1.22593619e-01 1.53824341e+00 3.13443914e-02 6.55715168e-01 -1.02726126e+00 -5.26106358e-01 8.77339005e-01 -2.97183037e-01 2.72316206e-03 5.14854752e-02 2.95164995e-02 -1.08211541e+00 -1.58746541e-01 -1.35383546e+00 -2.14749902e-01 -1.24298251e+00 -1.39373684e+00 1.06491196e+00 -4.01876003e-01 -1.12672579e+00 -7.85313427e-01 -1.02432080e-01 -8.74609828e-01 9.64520693e-01 -1.11560881e+00 -1.02546871e+00 1.12471707e-01 7.00878143e-01 1.12180722e+00 -2.93581188e-01 9.84863281e-01 -3.02946996e-02 -5.48866510e-01 8.46813679e-01 2.27830082e-01 1.22063115e-01 9.39035237e-01 -1.23819196e+00 7.93840647e-01 7.15620816e-01 -1.99608743e-01 4.18275595e-01 1.28611910e+00 -5.69059908e-01 -1.00368297e+00 -1.23506773e+00 7.31814981e-01 -1.36443049e-01 6.74360216e-01 -6.15937412e-01 -8.90242398e-01 6.98439538e-01 8.51720572e-01 -2.58129478e-01 8.46819460e-01 -2.87512273e-01 -7.53468648e-02 3.39851618e-01 -1.38132107e+00 1.08725810e+00 4.70956355e-01 -7.90482640e-01 -5.07171154e-01 1.81881636e-01 1.22940552e+00 -5.03826857e-01 -7.27573156e-01 -5.13918251e-02 3.12468708e-01 -6.37996614e-01 6.90750122e-01 -5.98434091e-01 3.16874623e-01 2.78189499e-02 2.69284770e-02 -1.93061244e+00 3.34705591e-01 -1.18368459e+00 -9.30544063e-02 1.38854146e+00 5.46850681e-01 -7.92233407e-01 5.52597940e-01 8.14000845e-01 -4.78510886e-01 -5.31173408e-01 -1.02745533e+00 -8.86134326e-01 2.51065761e-01 -4.31402355e-01 5.59325397e-01 1.05293453e+00 1.62332073e-01 6.87723219e-01 -8.56554925e-01 3.55158418e-01 2.91302949e-01 -2.09837899e-01 7.76845098e-01 -5.52857399e-01 -3.11932325e-01 -3.57987314e-01 3.36023301e-01 -1.12703502e+00 8.39762092e-01 -6.31408751e-01 6.37912393e-01 -1.29117095e+00 -4.04576868e-01 -1.32105919e-02 2.23208904e-01 4.28097904e-01 -2.64820457e-01 -2.67925173e-01 2.89610803e-01 -3.36461186e-01 -3.09990764e-01 1.33805299e+00 1.41824186e+00 -2.67720252e-01 -2.38188207e-01 1.14927091e-01 -4.81214523e-01 4.98952448e-01 1.07753146e+00 -5.72944164e-01 -8.06480229e-01 -3.73203382e-02 -3.51043344e-01 6.29661560e-01 1.68636620e-01 -5.89154184e-01 3.87654603e-02 -3.00154775e-01 -1.46165371e-01 -1.58322141e-01 9.15808618e-01 -6.78319097e-01 -6.34311214e-02 2.71363705e-01 -6.72373831e-01 -4.07575786e-01 3.10045302e-01 4.33854073e-01 -2.34419599e-01 -3.88915360e-01 9.16799903e-01 8.53577405e-02 -2.34202877e-01 4.41469073e-01 -9.53666627e-01 3.27004075e-01 1.07726240e+00 1.27054542e-01 -3.60896766e-01 -1.26572061e+00 -1.09208596e+00 3.42551649e-01 3.24758440e-01 6.97640657e-01 8.40141475e-01 -1.31600046e+00 -7.29644954e-01 6.64617792e-02 6.29053861e-02 -1.63759619e-01 6.13419592e-01 1.63009986e-01 5.22253476e-02 7.24026188e-02 2.16049179e-01 -1.19862825e-01 -1.50046051e+00 6.44965589e-01 5.13278067e-01 1.66608483e-01 -4.33100253e-01 8.21791232e-01 3.28916997e-01 -7.47265756e-01 3.02842557e-01 4.36620526e-02 1.76832855e-01 -1.64061725e-01 3.94091100e-01 -5.63330017e-02 -4.64475483e-01 -7.39050925e-01 1.87016353e-02 -7.89892748e-02 1.60834163e-01 -8.62968266e-01 9.43064213e-01 -3.07915092e-01 5.11504769e-01 6.96363509e-01 1.08748174e+00 1.88511997e-01 -1.45216858e+00 7.80584589e-02 -5.11376619e-01 -1.65930316e-01 -1.12493061e-01 -9.10238266e-01 -9.02820349e-01 1.14025998e+00 4.18172538e-01 4.94999915e-01 1.01415694e+00 -6.24162965e-02 1.31549358e+00 2.69946456e-01 -3.53271626e-02 -1.07382405e+00 2.90411234e-01 5.50535798e-01 1.09244490e+00 -1.10643327e+00 -1.02970374e+00 -2.18794167e-01 -1.40887392e+00 7.31919348e-01 6.46680117e-01 2.68648505e-01 4.04618979e-01 4.16391730e-01 6.78469837e-01 9.73590687e-02 -1.15421331e+00 -6.07925951e-02 -5.75326458e-02 8.36212456e-01 3.07331115e-01 3.66073608e-01 3.25747728e-01 4.83650774e-01 -5.65711617e-01 -2.40650102e-01 5.48245430e-01 5.66165090e-01 -3.78553033e-01 -1.07157469e+00 -2.97486395e-01 -6.04824647e-02 -4.32609320e-01 -2.94956893e-01 -7.10281968e-01 3.29493940e-01 -4.09175843e-01 1.41977787e+00 -7.14116767e-02 -4.02843237e-01 4.30502176e-01 5.14039278e-01 -2.28740033e-02 -5.52724600e-01 -6.30412221e-01 2.07509428e-01 4.27401811e-01 -2.48272698e-02 6.28375076e-03 -5.46121061e-01 -1.48788524e+00 -3.39017421e-01 -1.96293406e-02 4.71254617e-01 7.01802969e-01 7.30500638e-01 3.19606602e-01 7.13769197e-01 1.20385242e+00 -6.28790975e-01 -6.97311223e-01 -1.15113592e+00 -5.12345910e-01 4.56879437e-01 5.71433783e-01 -9.22892392e-02 -6.70305967e-01 1.53270453e-01]
[14.769095420837402, 6.4517717361450195]
e7c3d9fa-9b24-4ba4-8737-46836cc885e2
a-survey-of-toxic-comment-classification
2112.06412
null
https://arxiv.org/abs/2112.06412v1
https://arxiv.org/pdf/2112.06412v1.pdf
A Survey of Toxic Comment Classification Methods
While in real life everyone behaves themselves at least to some extent, it is much more difficult to expect people to behave themselves on the internet, because there are few checks or consequences for posting something toxic to others. Yet, for people on the other side, toxic texts often lead to serious psychological consequences. Detecting such toxic texts is challenging. In this paper, we attempt to build a toxicity detector using machine learning methods including CNN, Naive Bayes model, as well as LSTM. While there has been numerous groundwork laid by others, we aim to build models that provide higher accuracy than the predecessors. We produced very high accuracy models using LSTM and CNN, and compared them to the go-to solutions in language processing, the Naive Bayes model. A word embedding approach is also applied to empower the accuracy of our models.
['Hongjun Wu', 'Jiaxi Yang', 'Kehan Wang']
2021-12-13
null
null
null
null
['toxic-comment-classification']
['natural-language-processing']
[-4.88431659e-03 -7.11795017e-02 1.08594373e-01 -3.63970101e-01 -2.53932804e-01 -4.08293784e-01 7.25889087e-01 6.05754852e-01 -6.54674470e-01 8.59063983e-01 3.04298967e-01 -4.46648538e-01 -4.67887484e-02 -1.16266000e+00 -2.77622283e-01 -2.94365019e-01 1.38196483e-01 2.91271776e-01 8.34000707e-02 -3.45823556e-01 4.31283504e-01 4.90660340e-01 -1.08840346e+00 2.15869308e-01 5.90316951e-01 6.36870861e-01 -4.49646622e-01 4.93176222e-01 -1.81448460e-01 1.24576950e+00 -6.87918723e-01 -1.03663182e+00 -1.61569804e-01 -1.93382353e-01 -7.91166961e-01 -2.62568623e-01 1.65201351e-01 -5.07650733e-01 -4.19351220e-01 1.25253081e+00 5.66570818e-01 1.33062705e-01 8.65106404e-01 -1.07851934e+00 -1.03556585e+00 8.89647961e-01 -3.93708080e-01 1.26733512e-01 7.92813241e-01 3.55485678e-01 7.39695668e-01 -3.25201124e-01 3.65485370e-01 1.32163870e+00 1.03091586e+00 6.41982555e-01 -8.24363887e-01 -7.43619740e-01 -5.34345657e-02 7.88610056e-02 -1.05421424e+00 -1.20278314e-01 5.64613879e-01 -5.44533968e-01 1.06080496e+00 1.06478646e-01 5.44023693e-01 1.81008077e+00 6.75591409e-01 6.03316724e-01 1.10530293e+00 -3.66266787e-01 1.70710817e-01 5.48557580e-01 4.35170859e-01 5.54258049e-01 6.48754418e-01 2.95701474e-02 -5.04235148e-01 -4.58136290e-01 2.18673237e-02 2.97802299e-01 -1.51603833e-01 4.10155356e-01 -6.58985734e-01 1.04368877e+00 1.56652018e-01 9.17616904e-01 -4.13776219e-01 1.45596936e-01 7.36459434e-01 3.20120096e-01 5.95408559e-01 5.10069549e-01 -4.09866244e-01 -7.64869452e-02 -8.81972611e-01 3.17224920e-01 1.13408101e+00 4.17069018e-01 3.15333873e-01 -1.70620561e-01 -3.97765070e-01 5.13075233e-01 4.49529052e-01 1.54372647e-01 6.42955542e-01 -3.34217608e-01 2.58399010e-01 5.52280962e-01 -3.24941576e-02 -1.68640482e+00 -5.45638263e-01 -3.11206877e-01 -8.67660880e-01 2.22692847e-01 3.92262042e-01 -3.30644667e-01 -5.01174450e-01 1.47263682e+00 -1.76753387e-01 -2.07002804e-01 -2.18136832e-01 5.25163352e-01 5.24658382e-01 6.69007719e-01 5.72194636e-01 -3.01252514e-01 1.15323532e+00 -4.64279860e-01 -1.08916688e+00 -9.05125663e-02 1.14057636e+00 -5.05043507e-01 1.05005574e+00 8.16486716e-01 -7.85045981e-01 -9.28067416e-02 -9.98464286e-01 -6.69242674e-03 -7.55327106e-01 -5.14892220e-01 8.69650900e-01 1.34189332e+00 -9.62996364e-01 1.14493918e+00 -4.16349232e-01 -7.64893651e-01 4.79554683e-01 3.06373090e-01 -2.63887942e-01 1.53818771e-01 -1.80912578e+00 1.36202347e+00 3.28352600e-01 -1.39074236e-01 -5.55508494e-01 -3.22942644e-01 -6.70306325e-01 7.39263818e-02 7.69228861e-03 -6.44765139e-01 1.20286739e+00 -7.67009377e-01 -1.21857631e+00 7.64490962e-01 2.24877626e-01 -6.06581211e-01 6.69124722e-01 -1.79589137e-01 -5.04340768e-01 -3.56208712e-01 -1.35690957e-01 2.22410977e-01 4.13999885e-01 -6.82241261e-01 -2.08523914e-01 -6.16302907e-01 1.49894759e-01 -1.91768333e-01 -1.08924353e+00 5.67613125e-01 3.10694635e-01 -3.86780500e-01 -4.59324598e-01 -4.13467586e-01 -4.18884635e-01 7.84451962e-02 -6.48444951e-01 -6.37721539e-01 5.92850387e-01 -7.09469140e-01 1.48001158e+00 -1.89895284e+00 -6.32834911e-01 5.83056621e-02 4.18397486e-01 3.77324611e-01 1.64844304e-01 7.09652126e-01 -8.80577788e-02 7.41413176e-01 -1.06568962e-01 -2.26888493e-01 3.00589561e-01 -3.42880189e-02 -4.70760614e-01 6.72152400e-01 1.06043279e-01 7.16308475e-01 -7.89333880e-01 -4.39331859e-01 1.54860899e-01 6.79689884e-01 -3.21621329e-01 -2.11040109e-01 -8.62511992e-02 -3.06661278e-01 -5.88878155e-01 4.82354581e-01 5.08392215e-01 4.19849530e-02 1.29287019e-02 3.24049324e-01 3.21963280e-02 5.14268041e-01 -7.97510028e-01 1.12283349e+00 -2.30513573e-01 8.07933092e-01 -4.37025428e-01 -7.92625964e-01 8.18504512e-01 3.84510547e-01 6.42334297e-02 -7.74232566e-01 6.79492295e-01 -6.43933490e-02 2.99232975e-02 -8.33344042e-01 4.63016957e-01 -4.32974279e-01 -1.16868153e-01 7.84786403e-01 -1.99329421e-01 1.69668317e-01 2.04220891e-01 3.06899607e-01 1.52947450e+00 -2.84966052e-01 5.51419616e-01 1.86231875e-04 4.26129609e-01 -3.48918214e-02 3.57909411e-01 8.40267479e-01 -3.89992714e-01 2.58724451e-01 6.31748855e-01 -7.70388544e-01 -7.63225675e-01 -5.78036308e-01 -5.89372218e-02 1.04500747e+00 -4.12470400e-01 -5.74078441e-01 -1.08295596e+00 -5.92503846e-01 -1.97807193e-01 1.17633533e+00 -4.87858921e-01 -4.17488098e-01 1.99775994e-02 -1.06184983e+00 9.98187184e-01 1.85942978e-01 4.72156286e-01 -1.15276694e+00 -4.65452820e-01 4.28579330e-01 2.11327553e-01 -6.58292294e-01 -1.38359711e-01 2.66594380e-01 -7.53648281e-01 -7.90741146e-01 -2.59286255e-01 -1.39792472e-01 1.07265361e-01 -2.49210224e-02 1.00656855e+00 3.91562194e-01 -3.76484394e-01 5.67294806e-02 -4.59045053e-01 -7.19517291e-01 -3.99527550e-01 -2.25123405e-01 3.68571669e-01 -1.82783738e-01 1.00455344e+00 -6.56663775e-01 -1.73532575e-01 -3.19247723e-01 -1.05037248e+00 -6.27792120e-01 4.12627161e-01 2.21234545e-01 -2.96500921e-01 3.76522005e-01 4.09541398e-01 -1.24677253e+00 1.26178038e+00 -6.23012900e-01 -2.89912000e-02 -1.95041634e-02 -7.73373127e-01 -2.72679955e-01 7.42817044e-01 -5.94975710e-01 -7.33061194e-01 -8.22810754e-02 -6.45049930e-01 1.40542731e-01 -5.29491425e-01 5.34492791e-01 -2.29084969e-01 1.29678160e-01 1.02999854e+00 4.70725894e-02 -4.35461462e-01 -4.69571531e-01 3.08126926e-01 1.00231671e+00 -1.18322171e-01 -9.57375169e-02 7.32358336e-01 1.35731176e-01 -4.12717551e-01 -9.91014659e-01 -9.06198978e-01 -1.04533017e-01 -4.38031346e-01 -1.91275731e-01 8.71467948e-01 -4.80969489e-01 -9.55050349e-01 5.11989057e-01 -1.67837048e+00 7.61517137e-02 1.30426198e-01 3.18558246e-01 -8.86402130e-02 6.19333923e-01 -9.13692117e-01 -1.17809224e+00 -3.66164625e-01 -5.81471145e-01 3.38710934e-01 -1.25484556e-01 -9.00379360e-01 -1.32028890e+00 2.81337023e-01 3.61162841e-01 6.07776165e-01 2.48531207e-01 9.97170389e-01 -1.30535436e+00 7.55011216e-02 -5.75317681e-01 -1.32538378e-02 2.83945680e-01 -9.87704396e-02 2.30108485e-01 -1.24310398e+00 -6.27907589e-02 4.85582173e-01 -4.08826023e-01 8.75025332e-01 9.29367840e-02 1.12386262e+00 -6.44602239e-01 -3.83671135e-01 -1.84966121e-02 1.29720938e+00 4.08465534e-01 9.41856027e-01 2.73899406e-01 4.14030343e-01 8.13882470e-01 4.42526042e-02 4.28423941e-01 1.51868761e-01 3.09512734e-01 2.57103920e-01 1.77013412e-01 9.53969732e-02 -3.66708010e-01 7.05658913e-01 6.70398414e-01 2.22708523e-01 -6.25268936e-01 -1.10106111e+00 2.26459533e-01 -1.46300721e+00 -1.22031510e+00 -7.07533181e-01 1.90324330e+00 8.31853986e-01 5.50267339e-01 2.10698647e-03 5.71905315e-01 6.74558103e-01 -3.67613174e-02 -1.40784517e-01 -9.62012887e-01 -2.54492331e-02 1.65934339e-02 4.05332565e-01 3.31318319e-01 -1.00658751e+00 8.40535462e-01 7.33428383e+00 9.48267639e-01 -1.18830359e+00 3.39496076e-01 8.54660034e-01 6.31615296e-02 -3.34537596e-01 -2.20959306e-01 -8.02113116e-01 6.41789258e-01 1.30885518e+00 -2.09210962e-01 1.58334747e-01 8.60055089e-01 4.45886254e-01 -6.06977046e-02 -1.13361835e+00 8.27914655e-01 3.43602747e-01 -9.36322629e-01 2.05195308e-01 5.58149032e-02 3.13375801e-01 -2.02536583e-01 -1.08848222e-01 3.78547519e-01 4.49614108e-01 -1.33415329e+00 5.86432874e-01 4.77615118e-01 6.04895130e-02 -7.79076695e-01 8.05576682e-01 7.78408408e-01 -1.08777702e-01 -1.24692507e-01 -6.07270956e-01 -8.72026742e-01 -1.72520459e-01 1.14992166e+00 -7.66735256e-01 7.98957646e-02 6.64087415e-01 5.97893059e-01 -6.79115415e-01 8.75265598e-01 -3.85624290e-01 6.60269499e-01 -3.79192494e-02 -8.82453561e-01 2.90902436e-01 -1.23474851e-01 1.87499180e-01 1.33044028e+00 1.56516999e-01 1.86863579e-02 -1.64695531e-01 9.33238685e-01 -1.48472905e-01 4.15912896e-01 -1.35544026e+00 -6.15835547e-01 2.64513381e-02 1.22901213e+00 -7.15729773e-01 -1.99846402e-01 -4.47851449e-01 9.70557451e-01 2.73961008e-01 4.79684323e-02 -6.76538169e-01 -4.86039609e-01 3.46299708e-01 2.61200100e-01 -5.33299446e-01 5.89062534e-02 -6.04985058e-01 -9.48921323e-01 -1.28259778e-01 -7.41894543e-01 3.00191820e-01 -7.66801655e-01 -1.80028856e+00 5.46921253e-01 -5.64669907e-01 -8.12748313e-01 -1.13963298e-01 -8.88335288e-01 -9.15001273e-01 6.07109189e-01 -9.01131928e-01 -8.47075880e-01 3.32787931e-01 3.56706023e-01 3.68869901e-01 -4.60698046e-02 9.60988820e-01 4.47905481e-01 -5.56512594e-01 3.64826828e-01 -1.61548063e-01 2.99840599e-01 6.39511108e-01 -9.19882655e-01 1.95109025e-01 6.39137149e-01 1.73727691e-01 8.63786995e-01 8.80155563e-01 -8.16612244e-01 -1.11621702e+00 -8.99721801e-01 1.69425678e+00 -7.53561854e-01 9.98390019e-01 -4.99865025e-01 -8.05848241e-01 5.82897067e-01 5.20806253e-01 -7.40790606e-01 9.32091475e-01 2.58043438e-01 -4.27797765e-01 2.67727524e-01 -1.34088242e+00 8.09026062e-01 7.99740016e-01 -6.29284739e-01 -8.73988390e-01 7.47286558e-01 6.54485703e-01 5.67772090e-01 -4.65675622e-01 -1.47680864e-01 3.55738223e-01 -1.12377572e+00 6.87519252e-01 -9.79163766e-01 6.97669983e-01 1.23384252e-01 -2.47766972e-02 -1.27359211e+00 -4.19801354e-01 -3.43587011e-01 3.30066353e-01 1.47194970e+00 7.14323044e-01 -5.89523494e-01 6.29831016e-01 1.11788082e+00 3.15994322e-01 -5.98999918e-01 -6.59443140e-01 -7.63077796e-01 5.06652296e-01 -9.42915797e-01 2.00231895e-01 1.44169414e+00 4.27980959e-01 6.83921933e-01 -4.13200021e-01 -1.14204153e-01 4.86875176e-01 -6.63057625e-01 2.58933306e-01 -1.44471121e+00 -9.07554664e-03 -5.55448472e-01 -4.96748924e-01 -3.19386750e-01 3.22696239e-01 -8.72203469e-01 -1.98139191e-01 -1.66397226e+00 6.69856191e-01 2.47312505e-02 -2.99484432e-01 6.92112803e-01 1.52416095e-01 4.12652373e-01 2.58255750e-02 -2.98193336e-01 -6.13827050e-01 1.91259548e-01 7.58629680e-01 -3.86494130e-01 1.19670585e-01 -7.77848661e-02 -1.23548317e+00 1.10592902e+00 1.05425131e+00 -8.63029659e-01 -1.43421531e-01 -2.64964014e-01 9.51593220e-01 -1.69589445e-01 2.44893983e-01 -1.08507609e+00 5.07711530e-01 -9.31578595e-03 6.15932643e-01 -4.24668312e-01 1.31990984e-01 -6.52260423e-01 4.70785005e-03 7.14354038e-01 -5.51753044e-01 -1.78164750e-01 -5.38920565e-03 4.16654319e-01 1.17889926e-01 -7.09647417e-01 7.40750968e-01 -4.30707127e-01 -1.16538525e-01 2.23235533e-01 -1.09521794e+00 -2.51746893e-01 8.45216274e-01 -1.68968946e-01 -2.99792707e-01 -6.62168622e-01 -6.09800398e-01 2.47807838e-02 3.55604351e-01 3.34199250e-01 6.62044346e-01 -1.13754416e+00 -4.61692393e-01 -1.67000771e-01 -1.21156476e-01 -8.39338958e-01 -5.29485419e-02 7.04993188e-01 -4.99507189e-01 5.19466519e-01 -2.24348113e-01 1.99147359e-01 -1.10934889e+00 9.14223552e-01 2.05222487e-01 -3.25691789e-01 -3.41629535e-01 6.97177410e-01 -1.64483517e-01 -5.01791716e-01 5.41458607e-01 -1.27838254e-01 -6.32426739e-01 2.99073160e-01 1.05413949e+00 4.96254593e-01 1.55200690e-01 -2.73586661e-01 -2.69953549e-01 4.27517258e-02 -2.77064860e-01 4.45379689e-02 1.42832553e+00 1.68716699e-01 -4.86827701e-01 5.44665039e-01 1.11593103e+00 -1.15354359e-01 -2.71476746e-01 1.60717279e-01 4.30778682e-01 -4.90098059e-01 2.51860619e-01 -1.08201218e+00 -5.98845720e-01 1.28758657e+00 4.95502651e-01 6.44445419e-01 6.80134892e-01 -3.64181519e-01 8.79513443e-01 8.99061024e-01 3.85414660e-01 -1.04925919e+00 3.44468132e-02 6.16885304e-01 5.96791029e-01 -1.04022968e+00 3.14625576e-02 -2.09420204e-01 -5.24111032e-01 1.14339101e+00 2.82001168e-01 -6.55707642e-02 6.57310724e-01 3.84166896e-01 -9.30194482e-02 -2.15451971e-01 -8.47611487e-01 1.29563719e-01 -3.09971601e-01 6.31484807e-01 8.50783944e-01 -1.19217955e-01 -8.19963634e-01 8.20680261e-01 -1.21000595e-01 1.98944002e-01 9.37583387e-01 5.63975453e-01 -6.82034612e-01 -1.16900253e+00 -3.68685126e-01 6.23631120e-01 -1.12284493e+00 -3.31476718e-01 -1.02163911e+00 5.31502604e-01 1.46295503e-01 1.33382511e+00 -2.56239355e-01 -7.58747280e-01 2.18208537e-01 3.20290029e-01 4.55224477e-02 -6.45424962e-01 -9.72505271e-01 -5.08951664e-01 4.26020086e-01 -3.92488211e-01 -7.26160631e-02 -6.35442019e-01 -9.42489088e-01 -9.80870008e-01 -1.71267733e-01 -6.64430782e-02 5.81433713e-01 1.08356655e+00 -1.32532403e-01 2.64041305e-01 3.11466962e-01 -4.66765672e-01 -7.01621830e-01 -1.16060197e+00 -8.81550193e-01 4.01237667e-01 -1.36789769e-01 -2.58075863e-01 -5.10681689e-01 -1.00262165e-01]
[8.882262229919434, 10.51397705078125]
e1a0e296-680f-4d74-8eb1-d6f149dbcfa6
synergistic-graph-fusion-via-encoder
2303.18051
null
https://arxiv.org/abs/2303.18051v1
https://arxiv.org/pdf/2303.18051v1.pdf
Synergistic Graph Fusion via Encoder Embedding
In this paper, we introduce a novel approach to multi-graph embedding called graph fusion encoder embedding. The method is designed to work with multiple graphs that share a common vertex set. Under the supervised learning setting, we show that the resulting embedding exhibits a surprising yet highly desirable "synergistic effect": for sufficiently large vertex size, the vertex classification accuracy always benefits from additional graphs. We provide a mathematical proof of this effect under the stochastic block model, and identify the necessary and sufficient condition for asymptotically perfect classification. The simulations and real data experiments confirm the superiority of the proposed method, which consistently outperforms recent benchmark methods in classification.
['Ha Trinh', 'Jonathan Larson', 'Carey E. Priebe', 'Cencheng Shen']
2023-03-31
null
null
null
null
['stochastic-block-model']
['graphs']
[ 2.59053260e-01 2.96306103e-01 -6.99002981e-01 -1.38865307e-01 -6.15942836e-01 -3.08713853e-01 4.49877888e-01 4.73591328e-01 1.42464206e-01 7.59948075e-01 1.34372711e-01 -3.76811147e-01 -4.15163428e-01 -7.79290795e-01 -7.41645694e-01 -9.34207201e-01 -2.01585561e-01 2.29497463e-01 -1.77019760e-01 -3.53326052e-02 -4.85478416e-02 3.12541664e-01 -1.19804740e+00 -4.03105654e-02 6.94615483e-01 7.33438134e-01 -1.53076932e-01 7.75820076e-01 2.28958547e-01 7.25735962e-01 -3.63684714e-01 -8.03152740e-01 4.77527440e-01 -3.27491075e-01 -6.90840006e-01 2.67604887e-01 3.80343318e-01 9.28129721e-03 -6.97098255e-01 1.27744401e+00 4.88455564e-01 -1.06631942e-01 6.82019949e-01 -1.74163866e+00 -1.01299465e+00 9.66518760e-01 -6.61818504e-01 1.72336787e-01 3.73218179e-01 -3.94696981e-01 1.64304316e+00 -6.44802511e-01 4.94728446e-01 1.19155991e+00 7.06736863e-01 4.42166418e-01 -1.35962307e+00 -4.92958844e-01 1.43116459e-01 -4.94892187e-02 -1.76870441e+00 -1.63140073e-02 1.12130034e+00 -3.60579580e-01 6.24355018e-01 2.32614473e-01 5.62043667e-01 8.79967809e-01 5.78464806e-01 9.05843318e-01 1.01810324e+00 -5.59640527e-01 -9.98644158e-02 2.95385420e-01 5.35845697e-01 1.15725672e+00 7.66489565e-01 1.49866924e-01 -3.17757934e-01 -5.90372741e-01 5.34811974e-01 1.69523060e-01 -3.92076135e-01 -6.38404906e-01 -1.12742305e+00 1.15369606e+00 4.57078487e-01 2.68088490e-01 -1.84518158e-01 4.34680730e-01 4.60100591e-01 6.05335057e-01 5.66654503e-01 1.26793638e-01 -1.73610449e-02 3.28265369e-01 -2.80091614e-01 -1.25405699e-01 7.77890027e-01 1.15000629e+00 6.46680534e-01 4.80104089e-02 -1.01620751e-02 4.91575271e-01 3.34747195e-01 4.49092627e-01 2.27750987e-01 -3.77190679e-01 4.82848078e-01 6.82858109e-01 -3.49572688e-01 -1.25975573e+00 -2.48956293e-01 -7.40752339e-01 -1.22863889e+00 -2.95512438e-01 -2.76077364e-04 2.19513196e-02 -5.42607248e-01 1.76347816e+00 2.13335007e-01 5.44655800e-01 2.41404101e-01 4.64429408e-01 7.35976577e-01 2.66908437e-01 -1.97904766e-01 -3.75906020e-01 1.09830368e+00 -9.48292494e-01 -9.13720131e-01 4.62876745e-02 9.72365975e-01 -2.90799290e-01 7.58703470e-01 -4.57756314e-03 -8.60878825e-01 -2.69410521e-01 -1.40624714e+00 4.00647491e-01 -2.79494524e-01 1.02238782e-01 1.08906555e+00 9.10110354e-01 -1.21133518e+00 5.70076466e-01 -3.66203725e-01 -2.73791969e-01 2.97854573e-01 5.11518598e-01 -6.87818348e-01 -1.07235491e-01 -1.19305015e+00 5.16522944e-01 4.72173005e-01 -8.99351165e-02 -5.94212711e-01 -3.63203228e-01 -1.08099174e+00 6.88238814e-02 2.80923873e-01 -7.34219074e-01 7.10977733e-01 -8.15391123e-01 -1.04281843e+00 6.53042018e-01 -2.00378269e-01 -6.60482705e-01 1.49410695e-01 1.55843467e-01 -4.80200827e-01 2.67499059e-01 -6.35222271e-02 1.38937011e-01 9.19253528e-01 -1.30059719e+00 -4.18700248e-01 -3.36540222e-01 3.26361030e-01 1.19910553e-01 -6.95116103e-01 -2.36532018e-01 -2.00503185e-01 -8.09695661e-01 -8.70310590e-02 -9.78274405e-01 -3.93452853e-01 -3.52411568e-01 -2.99685508e-01 -2.22887620e-01 6.62513852e-01 -2.16667622e-01 1.37045062e+00 -2.17909670e+00 3.81692201e-01 5.06993532e-01 8.85065794e-01 -4.18756790e-02 -9.94064212e-02 6.87436759e-01 -4.46666360e-01 3.44803214e-01 -3.10986526e-02 -2.90437222e-01 -9.86036733e-02 1.81672499e-01 -1.63660839e-01 7.94825613e-01 2.84601506e-02 9.67542827e-01 -1.05925596e+00 -4.82490420e-01 2.63700843e-01 4.31405276e-01 -4.43277597e-01 -8.94713327e-02 5.23857951e-01 -9.84088182e-02 -4.42710221e-01 6.32245004e-01 5.71210027e-01 -8.81840050e-01 5.06211579e-01 -3.17445666e-01 7.74856329e-01 -9.41902772e-02 -1.37264681e+00 1.17890573e+00 -2.43242934e-01 3.89965296e-01 -7.27281049e-02 -1.20385718e+00 6.93569124e-01 4.79295522e-01 6.62656009e-01 -6.58353046e-02 3.20675194e-01 4.94186543e-02 -8.74293223e-02 1.00992598e-01 3.54490668e-01 -9.30316374e-02 -1.05602473e-01 4.90977257e-01 1.98547244e-01 3.79370481e-01 -2.01591030e-02 5.55307508e-01 1.23634148e+00 -6.24305010e-01 8.42068672e-01 -3.46707970e-01 5.97712576e-01 -4.70956832e-01 3.69097382e-01 7.30554283e-01 -3.24248523e-01 1.61973864e-01 6.94781065e-01 -2.68815160e-01 -8.46533060e-01 -8.94883513e-01 -3.08164638e-02 5.97764730e-01 2.87238866e-01 -8.36327195e-01 -5.73259234e-01 -9.46805418e-01 4.30457950e-01 1.37290746e-01 -9.53311741e-01 -5.77277184e-01 -4.24534567e-02 -9.65039909e-01 5.26363194e-01 6.13551974e-01 1.96934104e-01 -1.81225881e-01 2.20375851e-01 5.89825101e-02 1.50641724e-01 -1.22702193e+00 -5.13282061e-01 7.49162808e-02 -8.46044242e-01 -1.17937624e+00 -5.77241600e-01 -9.48138654e-01 8.62618148e-01 7.52546430e-01 7.64976978e-01 3.68528873e-01 -3.44234854e-02 6.01117194e-01 -4.54415858e-01 6.70684725e-02 -6.40982568e-01 1.03542551e-01 2.69147247e-01 2.90355623e-01 1.46049231e-01 -4.52262521e-01 -2.79201448e-01 1.58715501e-01 -7.77109206e-01 2.34717384e-01 6.61296070e-01 1.12335563e+00 4.81665015e-01 2.14247674e-01 7.12955058e-01 -9.27424133e-01 8.36400270e-01 -6.73034847e-01 -4.46328074e-01 5.79873323e-01 -1.16724873e+00 2.69343555e-01 6.35432780e-01 -2.88212448e-01 -3.57993633e-01 6.70055673e-02 1.18870333e-01 -6.75209641e-01 4.59251195e-01 6.89549387e-01 -1.14592776e-01 -5.45709789e-01 2.21312687e-01 2.16948882e-01 1.79781586e-01 -2.92025805e-01 5.14438033e-01 7.15065837e-01 1.77376464e-01 -3.11984837e-01 9.19195712e-01 4.85188305e-01 4.67009276e-01 -5.95519543e-01 -4.91564274e-01 -4.79923278e-01 -5.83974004e-01 -1.59757122e-01 3.97705704e-01 -1.00185120e+00 -9.46470141e-01 -6.33437093e-03 -1.00896347e+00 2.69620091e-01 1.30396150e-02 5.71943104e-01 -4.91995305e-01 6.78263485e-01 -6.52643383e-01 -8.55675340e-01 -3.99776369e-01 -1.04334104e+00 9.63493228e-01 -1.57092661e-01 2.60392010e-01 -1.53228951e+00 4.07275669e-02 2.14183480e-02 -1.78198636e-01 3.00578803e-01 9.55944955e-01 -8.08519244e-01 -3.22570026e-01 -5.28279662e-01 -4.88114595e-01 4.16104794e-01 4.28023815e-01 -1.09592259e-01 -6.77862406e-01 -6.40985847e-01 -3.74987543e-01 1.53949648e-01 8.53790462e-01 1.26119331e-01 1.06043029e+00 -2.23080695e-01 -7.79135942e-01 5.93493402e-01 1.76027906e+00 -9.54063088e-02 1.76313668e-01 -4.38214600e-01 1.13776577e+00 2.83563763e-01 3.20809066e-01 3.86833787e-01 5.31609654e-01 6.48473680e-01 3.26162696e-01 -1.21992886e-01 -1.00259393e-01 -2.95786589e-01 2.94480145e-01 1.40275347e+00 -2.71252003e-02 -4.95760709e-01 -6.36991322e-01 5.43909371e-01 -2.06846952e+00 -8.48777413e-01 -2.20513701e-01 2.34830737e+00 4.76095527e-01 -2.35382877e-02 5.86994775e-02 4.87826318e-01 9.83704209e-01 2.40310594e-01 -6.45558611e-02 -4.22936529e-01 -3.76483262e-01 1.05108052e-01 1.03873408e+00 7.44143784e-01 -1.08255422e+00 7.13860393e-01 7.91005898e+00 9.02198613e-01 -6.83447361e-01 1.86178505e-01 3.73135656e-01 3.80610824e-01 -5.36012173e-01 4.95063514e-02 -7.04127848e-01 3.29619884e-01 1.04064906e+00 -7.52003968e-01 4.25959140e-01 7.01688826e-01 -3.90625596e-01 4.60038275e-01 -1.09855998e+00 1.16664994e+00 3.37083131e-01 -1.60581875e+00 3.49353760e-01 2.51612425e-01 7.00714350e-01 -1.15317278e-01 -4.49497104e-02 3.33957076e-02 4.56626534e-01 -9.10102725e-01 1.54933870e-01 2.18255118e-01 8.17189395e-01 -9.05868351e-01 7.26617277e-01 -1.37206744e-02 -1.66084123e+00 -2.00575337e-01 -3.40491384e-01 -1.50448903e-01 -1.71694476e-02 6.18282974e-01 -7.87719727e-01 1.17581010e+00 1.14487223e-01 1.11782885e+00 -7.98236787e-01 8.49541426e-01 -1.26279950e-01 4.71365958e-01 -4.10466082e-02 -2.06190720e-01 1.23484306e-01 -1.55661479e-01 6.17478848e-01 1.03924417e+00 5.21581173e-02 -1.11125067e-01 1.82244703e-01 3.02302688e-01 -2.66659826e-01 2.43862540e-01 -1.20076740e+00 -3.99955511e-01 4.73056197e-01 1.15942454e+00 -7.98501730e-01 -5.03239751e-01 -7.14492500e-01 1.06113303e+00 6.28377616e-01 2.88793057e-01 -9.22513783e-01 -3.87625754e-01 5.36687195e-01 -3.37739557e-01 2.78998166e-01 -5.18014841e-02 -2.42260829e-01 -1.29236746e+00 1.17639333e-01 -7.28943229e-01 6.33915603e-01 -2.79589593e-01 -1.24210680e+00 4.51434284e-01 -2.30357461e-02 -1.27252507e+00 -6.77262545e-02 -7.11230814e-01 -3.93494129e-01 4.45046604e-01 -1.28260386e+00 -1.29767954e+00 -2.92207617e-02 5.94425499e-01 -1.33844495e-01 -3.20700318e-01 9.06631708e-01 6.00095332e-01 -5.42407334e-01 1.13950288e+00 4.00841087e-01 1.18444145e-01 3.73372287e-01 -1.33034062e+00 4.05560702e-01 6.57808900e-01 4.74919647e-01 5.02812147e-01 6.48107827e-01 -7.11588144e-01 -1.85538614e+00 -1.14351976e+00 9.09350336e-01 -4.04241234e-01 7.82686174e-01 -5.05948722e-01 -6.60749316e-01 9.78856266e-01 8.25835839e-02 2.05963016e-01 9.36524212e-01 3.09981972e-01 -5.21737158e-01 -9.14450437e-02 -1.10314405e+00 6.22926354e-01 1.21912539e+00 -7.79961228e-01 -4.66390282e-01 6.94444299e-01 1.03534055e+00 -1.94421440e-01 -1.18053031e+00 4.42997664e-01 4.69625980e-01 -3.73438984e-01 1.01519573e+00 -7.69239604e-01 1.50721475e-01 -7.34609663e-02 -4.22241986e-01 -1.35071611e+00 -4.56890106e-01 -8.57928813e-01 -5.12886524e-01 9.54282522e-01 3.74472469e-01 -1.13992727e+00 6.36917472e-01 -3.87590341e-02 2.19652727e-01 -1.00611162e+00 -9.56892967e-01 -1.17287576e+00 -1.03803314e-01 -2.47927323e-01 7.64470637e-01 1.11850345e+00 3.14627767e-01 4.74155009e-01 -7.06158578e-01 3.12459141e-01 8.31765056e-01 1.00429989e-01 6.55364215e-01 -1.21962297e+00 -3.64595681e-01 -4.01562274e-01 -1.11661124e+00 -8.03103507e-01 4.57010239e-01 -1.39127254e+00 -6.29509091e-01 -1.38551342e+00 4.91730511e-01 -3.20644557e-01 -7.48010814e-01 2.24083036e-01 -3.68829399e-01 2.50725776e-01 6.78476989e-02 3.88405621e-02 -6.38657570e-01 6.14440560e-01 1.01940358e+00 -1.61530599e-01 2.24621147e-01 -7.19966367e-02 -9.98066127e-01 3.06917906e-01 5.20156145e-01 -3.17550123e-01 -5.35576105e-01 -1.10578217e-01 3.41364592e-01 7.35603347e-02 3.39166194e-01 -7.44686484e-01 5.37478514e-02 1.31337449e-01 -7.12566152e-02 -3.17201525e-01 2.09968299e-01 -8.63347590e-01 2.23429680e-01 6.34657323e-01 -2.87244529e-01 2.61095405e-01 -1.26110718e-01 1.14274597e+00 -1.29330844e-01 8.61303974e-03 5.61428308e-01 5.14537275e-01 -3.75521839e-01 6.21454835e-01 5.37942052e-02 -2.06696942e-01 1.36420095e+00 -8.83296207e-02 -3.02906275e-01 -5.85337579e-01 -5.77579379e-01 3.21195871e-01 3.73148948e-01 5.15094817e-01 7.04575956e-01 -1.99803126e+00 -7.79191494e-01 2.47489110e-01 5.40593624e-01 -8.58799279e-01 -4.90017571e-02 9.85033095e-01 -9.67303216e-02 4.80537832e-01 1.61409259e-01 -4.60963577e-01 -1.46934271e+00 1.03985167e+00 1.66777954e-01 -3.70555639e-01 -7.22455025e-01 6.15922093e-01 1.17582642e-01 -2.03073278e-01 1.81838404e-02 -2.67066926e-01 4.67150994e-02 -7.47662559e-02 3.42357397e-01 3.03102046e-01 2.03454904e-02 -7.75663614e-01 -3.83932799e-01 3.99274260e-01 -1.79033563e-01 1.46733508e-01 9.68039989e-01 -1.83183983e-01 -2.32868455e-02 5.50248563e-01 1.60616088e+00 -1.66411623e-02 -7.13860154e-01 -6.07003152e-01 -2.63461262e-01 -7.24744260e-01 7.57624879e-02 -6.32955655e-02 -1.19298780e+00 5.86633027e-01 3.21463525e-01 5.52274644e-01 8.44440997e-01 2.91705847e-01 5.76455355e-01 3.38546395e-01 5.93755305e-01 -7.73546398e-01 -1.60540372e-01 1.49183795e-01 5.95148444e-01 -1.29599118e+00 2.58208096e-01 -8.11195076e-01 -5.23575068e-01 8.38456869e-01 2.00582638e-01 -3.33653897e-01 8.62809479e-01 1.54719297e-02 -5.70117295e-01 -3.21998358e-01 -9.92604196e-01 -2.03417972e-01 4.03079718e-01 5.89454293e-01 2.88105547e-01 5.59655368e-01 -5.54038823e-01 6.26937509e-01 -7.90240429e-03 -3.83314878e-01 5.30838788e-01 7.97516704e-01 -1.34227172e-01 -1.07028973e+00 -6.20474294e-02 7.19613492e-01 -4.01272058e-01 -6.51198775e-02 -5.73757887e-01 8.94204021e-01 -4.55229819e-01 8.55515897e-01 -1.17698185e-01 -8.08236003e-01 -4.18714546e-02 -1.14130490e-01 7.72816360e-01 -4.73043412e-01 -3.28466505e-01 -1.84361532e-01 7.88159668e-03 -3.41456324e-01 -4.40928817e-01 -4.89828795e-01 -1.02261341e+00 -5.72520316e-01 -8.20808232e-01 2.88230211e-01 3.95944864e-01 8.38654160e-01 5.25171995e-01 6.69928074e-01 1.00351238e+00 -3.26612681e-01 -5.54031491e-01 -6.76390290e-01 -7.86269903e-01 3.67291003e-01 4.44805115e-01 -9.12771821e-01 -7.59792149e-01 -3.22230965e-01]
[7.155426502227783, 6.058928489685059]
41b9217b-914a-4fce-95e6-f575856c62ab
ontology-matching-with-knowledge-rules
1507.03097
null
http://arxiv.org/abs/1507.03097v1
http://arxiv.org/pdf/1507.03097v1.pdf
Ontology Matching with Knowledge Rules
Ontology matching is the process of automatically determining the semantic equivalences between the concepts of two ontologies. Most ontology matching algorithms are based on two types of strategies: terminology-based strategies, which align concepts based on their names or descriptions, and structure-based strategies, which exploit concept hierarchies to find the alignment. In many domains, there is additional information about the relationships of concepts represented in various ways, such as Bayesian networks, decision trees, and association rules. We propose to use the similarities between these relationships to find more accurate alignments. We accomplish this by defining soft constraints that prefer alignments where corresponding concepts have the same local relationships encoded as knowledge rules. We use a probabilistic framework to integrate this new knowledge-based strategy with standard terminology-based and structure-based strategies. Furthermore, our method is particularly effective in identifying correspondences between complex concepts. Our method achieves substantially better F-score than the previous state-of-the-art on three ontology matching domains.
['Shangpu Jiang', 'Dejing Dou', 'Daniel Lowd']
2015-07-11
null
null
null
null
['ontology-matching']
['knowledge-base']
[ 1.49589509e-01 -1.17615145e-02 -3.32987100e-01 -5.82235932e-01 -2.77497590e-01 -5.19059658e-01 5.97692907e-01 6.03153586e-01 -2.82835871e-01 4.70351726e-01 2.72919565e-01 -2.01904118e-01 -7.40272343e-01 -1.10907364e+00 -1.33118555e-01 -1.36594966e-01 -1.77610647e-02 9.70896363e-01 7.01481938e-01 -3.23033839e-01 4.67729360e-01 3.12424958e-01 -1.96447277e+00 3.70534301e-01 9.73689914e-01 8.11372697e-01 -3.26942839e-02 -4.75347154e-02 -8.05570483e-01 5.81485152e-01 -2.01113641e-01 -6.14101827e-01 8.90348554e-02 -2.83097684e-01 -1.29695678e+00 -4.69527483e-01 2.31278077e-01 1.89914584e-01 -2.54511505e-01 1.44605935e+00 -7.72512630e-02 2.07948357e-01 7.16213226e-01 -1.42766666e+00 -3.76443326e-01 5.06351769e-01 -1.95843875e-01 1.39039546e-01 9.60740983e-01 -7.31646597e-01 1.52279663e+00 -5.71455777e-01 5.40306628e-01 1.64116693e+00 5.82577467e-01 3.35791230e-01 -1.13996506e+00 -8.59412134e-01 1.86011165e-01 7.55421877e-01 -1.64761209e+00 -3.43074948e-01 3.77986759e-01 -6.75232232e-01 1.06027472e+00 1.86960965e-01 4.00354803e-01 4.13537532e-01 8.54219962e-03 3.27926457e-01 8.33577216e-01 -9.52976346e-01 2.89774865e-01 1.96227521e-01 3.32979232e-01 7.43299127e-01 4.20674145e-01 -8.12433288e-02 -5.59344828e-01 -7.38966525e-01 3.72991681e-01 -5.42338192e-02 -1.14145130e-01 -5.11626422e-01 -9.51676905e-01 8.82728934e-01 1.43189281e-01 6.19888544e-01 -1.28037453e-01 -2.01555938e-01 1.53196141e-01 1.99609533e-01 6.46788999e-02 7.64070451e-01 -5.67285597e-01 2.10086718e-01 -6.94014132e-01 3.17478240e-01 1.07744789e+00 1.28435540e+00 1.09216547e+00 -8.77837956e-01 3.23164314e-01 1.03723979e+00 8.34609807e-01 -3.00989840e-02 6.47661150e-01 -1.15698278e+00 1.87425330e-01 9.87344861e-01 9.50292572e-02 -1.19797540e+00 -2.69932717e-01 2.50744790e-01 -1.15622714e-01 -1.00775741e-01 2.54825920e-01 5.03047407e-01 -7.02531874e-01 1.81604981e+00 4.12194908e-01 1.80415884e-01 1.91728339e-01 4.40301955e-01 7.30203092e-01 2.04904392e-01 2.12775394e-01 1.21103540e-01 1.70967603e+00 -4.77910340e-01 -6.63081110e-01 -1.86425090e-01 6.55033171e-01 -8.78685534e-01 6.09607875e-01 -3.29728536e-02 -6.25044286e-01 -2.20897928e-01 -1.00789034e+00 6.88966513e-02 -7.94478357e-01 -5.53597748e-01 7.80541301e-01 7.39969492e-01 -7.81672001e-01 7.67295241e-01 -5.50835729e-01 -8.62850547e-01 1.14241950e-01 2.79863030e-01 -4.81700093e-01 -2.65034795e-01 -1.65357399e+00 1.25528169e+00 9.36557829e-01 -5.67973137e-01 7.04837823e-03 -5.22397876e-01 -1.07867765e+00 1.49149939e-01 5.49434185e-01 -8.03158462e-01 1.07720184e+00 -6.84796810e-01 -1.15885544e+00 1.07381821e+00 -4.31431442e-01 -2.13090181e-01 -8.28214586e-02 8.41838270e-02 -7.80978978e-01 8.99157226e-02 4.81249243e-01 4.57279533e-01 7.22118467e-02 -1.02866340e+00 -1.20899653e+00 -4.36084211e-01 3.55204761e-01 1.51880667e-01 -3.00684482e-01 4.45483088e-01 -6.88599527e-01 -2.80796021e-01 6.21040642e-01 -8.63384485e-01 -7.62163252e-02 1.63680002e-01 -1.21938802e-01 -7.50170708e-01 4.79617983e-01 -3.89604956e-01 1.35012484e+00 -1.86097586e+00 2.75028008e-03 7.51532018e-01 1.09691083e-01 -5.46908863e-02 6.35426641e-02 4.59825516e-01 -1.01994216e-01 9.56635848e-02 -3.66618335e-01 2.31348321e-01 2.82072484e-01 6.26220465e-01 -2.33457133e-01 -6.25075325e-02 -3.77742425e-02 3.89560133e-01 -1.09545112e+00 -8.71070981e-01 5.03060073e-02 -3.05547807e-02 -5.14969885e-01 7.18597397e-02 -2.23969609e-01 -2.54349131e-03 -6.40085578e-01 5.60236692e-01 2.97122478e-01 -1.38490081e-01 9.69956815e-01 -1.96786046e-01 1.66112378e-01 7.08909929e-01 -1.33256745e+00 1.70090783e+00 -3.41457993e-01 2.95590997e-01 -4.03825641e-01 -1.21369028e+00 1.10433245e+00 4.64871317e-01 6.54330969e-01 -4.38471317e-01 -1.87645867e-01 4.48724747e-01 5.26393354e-02 -5.93145669e-01 3.79441261e-01 -2.19823316e-01 -2.80447938e-02 2.98412651e-01 1.61861405e-01 -5.08227013e-02 5.30725121e-01 1.37480358e-02 1.11619925e+00 4.03148711e-01 8.28586876e-01 -4.67030972e-01 6.34638429e-01 3.16726714e-02 9.40870225e-01 6.45562470e-01 1.07372021e-02 -5.41621237e-04 3.07151109e-01 -3.99205446e-01 -7.54568458e-01 -1.06129265e+00 -4.70619649e-01 1.04508388e+00 4.52242285e-01 -9.86013055e-01 -3.80916387e-01 -6.97296381e-01 2.42210612e-01 7.68429637e-01 -3.22944850e-01 -1.73303679e-01 -1.95946366e-01 -4.51998264e-01 5.67765892e-01 5.17589211e-01 3.26246530e-01 -6.75636709e-01 -3.45620811e-01 4.12459254e-01 -4.22694534e-01 -1.25400865e+00 -8.69695023e-02 1.69992447e-03 -7.99360991e-01 -1.49242151e+00 9.10002366e-02 -6.85266674e-01 4.88653004e-01 4.05125655e-02 1.37205541e+00 2.92927802e-01 8.96984804e-03 2.11501062e-01 -3.80781353e-01 -3.89701694e-01 -4.69660670e-01 -7.26479813e-02 1.55270129e-01 -1.92848042e-01 1.17222774e+00 -7.52758920e-01 -9.49844271e-02 8.75621200e-01 -8.77798736e-01 -3.25514227e-01 2.68641770e-01 5.95564127e-01 3.56181413e-01 5.67589343e-01 1.62059709e-01 -7.92270541e-01 4.77188200e-01 -5.77808499e-01 -5.70391774e-01 7.68705666e-01 -8.89327586e-01 5.12977481e-01 8.15332979e-02 -2.50982285e-01 -9.04440761e-01 7.22834468e-02 1.10333785e-01 -4.86754589e-02 -3.23375911e-01 8.69238913e-01 -3.20994973e-01 -6.02166131e-02 6.03553355e-01 -2.35943690e-01 -3.91589254e-01 -6.21747673e-01 2.14145437e-01 9.63997006e-01 5.40041506e-01 -1.30934131e+00 6.78426862e-01 4.10870016e-01 -2.81289741e-02 -4.42930937e-01 -1.06218743e+00 -9.76943195e-01 -7.78194547e-01 1.75498560e-01 7.28579164e-01 -6.32713795e-01 -5.97737491e-01 -1.28482878e-01 -1.07445943e+00 4.44730520e-01 1.18600875e-01 6.52626634e-01 -3.43344957e-01 6.64204180e-01 -1.97457284e-01 -6.11722231e-01 5.44297174e-02 -9.72511232e-01 7.88093269e-01 2.49422416e-01 -7.36383796e-01 -1.04377592e+00 2.71068960e-01 4.59470600e-01 1.14401288e-01 -3.51392850e-02 1.48190200e+00 -1.26520431e+00 -1.66646451e-01 -2.66223162e-01 -2.54382521e-01 -8.61932635e-02 4.55173224e-01 -1.20032832e-01 -5.93922198e-01 -1.24598071e-01 -5.23356438e-01 1.19681261e-01 4.98247862e-01 -2.55587250e-01 8.13691437e-01 -7.01333955e-02 -8.84340227e-01 1.53655529e-01 1.32855165e+00 5.13040662e-01 6.57255590e-01 5.78519762e-01 3.87176126e-01 1.07817793e+00 6.35782599e-01 1.20124444e-01 6.99339330e-01 1.14718306e+00 -3.71308513e-02 3.18961024e-01 3.68280768e-01 -2.24900484e-01 -7.28309676e-02 6.78627193e-01 -2.02022046e-01 1.24823945e-02 -1.38286030e+00 5.91601133e-01 -2.20183802e+00 -1.03629112e+00 4.30036746e-02 2.31369543e+00 1.12546682e+00 1.34672642e-01 -1.04223989e-01 8.06448832e-02 1.09766066e+00 -3.39387774e-01 -1.33205250e-01 -1.81188926e-01 1.26763597e-01 3.88022870e-01 1.85419425e-01 4.48492825e-01 -9.92998242e-01 1.00737917e+00 6.85825920e+00 6.01204813e-01 -3.72740746e-01 6.16991296e-02 -3.27291608e-01 4.67534542e-01 -3.90461594e-01 6.35648608e-01 -8.62075090e-01 4.75510389e-01 7.33660936e-01 -4.40931678e-01 2.29999840e-01 7.12411344e-01 -3.72051179e-01 -1.39847323e-01 -1.41768110e+00 7.72024453e-01 -2.31995853e-03 -1.25127745e+00 2.62537330e-01 1.29111022e-01 5.11937737e-01 -2.18162701e-01 -8.36644769e-01 2.11732611e-01 9.51407611e-01 -7.79202461e-01 5.23900568e-01 6.90490425e-01 4.19258147e-01 -5.89524031e-01 9.11251783e-01 -8.77969116e-02 -1.50654078e+00 -1.62825882e-02 -4.80437279e-01 1.64088968e-03 -3.73368077e-02 5.78378201e-01 -6.04993522e-01 1.08925104e+00 9.15857196e-01 7.23638117e-01 -2.55252510e-01 1.27968383e+00 -5.70402443e-01 1.43274739e-01 -4.78542984e-01 1.73864290e-01 -1.00301750e-01 -3.37717295e-01 4.21537936e-01 1.08565807e+00 2.68775612e-01 4.41514440e-02 4.66991842e-01 8.44958663e-01 1.92877859e-01 1.71951964e-01 -5.03577530e-01 5.26808277e-02 9.99961197e-01 8.04698110e-01 -5.79752624e-01 -4.75246936e-01 -6.97527170e-01 4.88046318e-01 1.72833696e-01 1.87183674e-02 -3.63740563e-01 -6.65908575e-01 1.11138678e+00 -1.17267966e-02 2.73755006e-02 -9.28583071e-02 1.48458809e-01 -1.10696518e+00 -4.24123195e-04 -8.51667941e-01 1.10528851e+00 -7.59493709e-01 -1.60074914e+00 4.25800234e-01 5.16377628e-01 -1.10696900e+00 -3.01826596e-01 -7.43442357e-01 -3.08906019e-01 7.74894536e-01 -1.53200471e+00 -8.25010836e-01 -2.50298411e-01 6.07640326e-01 -3.13660428e-02 -2.07620293e-01 1.45577574e+00 4.25871670e-01 -1.42990589e-01 3.25260609e-01 -1.59352552e-03 2.63896853e-01 8.52992237e-01 -1.15183485e+00 1.92846522e-01 6.65346444e-01 2.82043993e-01 1.08119631e+00 6.58739209e-01 -6.93567276e-01 -5.88125765e-01 -6.73866093e-01 1.34572744e+00 -4.15072143e-01 9.11831021e-01 3.19259800e-02 -1.15900087e+00 7.17496395e-01 -1.93580344e-01 -3.07426691e-01 1.27274811e+00 6.29810929e-01 -1.06105888e+00 -1.73208676e-02 -1.11789966e+00 4.75803047e-01 1.29213655e+00 -7.84209907e-01 -1.44925940e+00 2.89665312e-01 5.72739840e-01 -8.42103735e-02 -1.21487343e+00 5.68327665e-01 8.34030569e-01 -6.85954094e-01 9.13189530e-01 -1.01339459e+00 7.35860318e-02 -6.75667107e-01 -6.97197139e-01 -1.06467950e+00 -4.61640954e-01 -3.19730014e-01 1.73667237e-01 1.22301126e+00 6.26479626e-01 -9.36152458e-01 4.76547450e-01 5.75225890e-01 2.27810994e-01 -1.85926661e-01 -9.39768970e-01 -1.03732288e+00 -1.52816415e-01 -3.00409555e-01 1.14482808e+00 1.64144671e+00 5.10190070e-01 2.83173472e-01 2.52065778e-01 4.99758124e-01 5.60863256e-01 3.31577450e-01 3.42305273e-01 -2.13095617e+00 -3.91628891e-02 -5.62427819e-01 -1.11417031e+00 -5.11192679e-01 5.68386912e-01 -1.07393944e+00 -2.10291464e-02 -1.58410990e+00 3.20744187e-01 -7.64548004e-01 -5.30815542e-01 9.46466148e-01 -1.78909734e-01 -2.38642722e-01 -1.54589146e-01 4.18790549e-01 -5.61713636e-01 2.75598496e-01 4.04338151e-01 -7.52666816e-02 1.09813353e-02 -2.14005068e-01 -6.83383107e-01 8.65164578e-01 6.50609910e-01 -9.11295772e-01 -3.39243054e-01 -1.03285693e-01 3.69947731e-01 -2.67991543e-01 -3.21830586e-02 -9.76846635e-01 5.63465893e-01 -4.94328082e-01 -1.89753801e-01 -1.60331175e-01 1.07902780e-01 -9.20789897e-01 4.23573077e-01 3.86493027e-01 -3.02458704e-01 3.84155437e-02 6.30958602e-02 6.09428942e-01 -3.98703068e-01 -6.39758110e-01 7.05145001e-01 -2.22597465e-01 -1.05199969e+00 -8.72857776e-03 -3.34725648e-01 7.10193142e-02 7.19595313e-01 -2.46567085e-01 -7.38690868e-02 -1.65755391e-01 -7.04150915e-01 3.60157222e-01 5.00073552e-01 7.27067411e-01 3.01770806e-01 -1.40933526e+00 -4.86493438e-01 -1.67665347e-01 7.58166373e-01 -3.29005003e-01 -5.38901925e-01 4.29737359e-01 -1.41937599e-01 4.98837173e-01 -2.91750997e-01 -4.53067511e-01 -1.47452462e+00 3.30331951e-01 5.70138931e-01 -2.64794886e-01 -3.24592501e-01 5.00186086e-01 1.44372983e-02 -9.18207645e-01 6.78941086e-02 8.66174623e-02 -4.20739323e-01 2.11069616e-03 5.14272809e-01 1.37803778e-01 1.63834959e-01 -6.35588646e-01 -8.41015756e-01 8.31693411e-01 -1.74733251e-01 -1.20202199e-01 1.04432833e+00 4.90195081e-02 -6.02990985e-01 1.79046944e-01 7.41429806e-01 -1.55576169e-01 -2.77155459e-01 -9.71774161e-01 9.50284004e-01 -7.63983846e-01 -1.80617161e-02 -6.76635087e-01 -6.19449675e-01 2.33357444e-01 3.33308309e-01 2.20697939e-01 8.00165653e-01 2.39448264e-01 3.19223672e-01 6.53992355e-01 6.98043346e-01 -1.04800045e+00 -3.34626675e-01 5.79192579e-01 4.82566595e-01 -1.04548264e+00 1.75565004e-01 -9.87126231e-01 -4.87172157e-02 1.24363077e+00 4.78899866e-01 3.80087018e-01 6.86545730e-01 5.69195561e-02 -1.04575910e-01 -4.88818169e-01 -6.47408664e-01 -6.27495110e-01 6.17694676e-01 6.58635676e-01 5.02209365e-01 -1.19290343e-02 -4.55717981e-01 3.59456241e-01 -2.35425055e-01 -1.14293844e-01 2.81424802e-02 1.15602982e+00 -6.36774302e-01 -1.64134097e+00 -3.44208091e-01 3.32809806e-01 -2.71542400e-01 -1.82288557e-01 -5.06794393e-01 6.51936054e-01 2.64705718e-01 1.21947837e+00 2.51859486e-01 -4.26288426e-01 4.79660064e-01 5.31599045e-01 5.71303546e-01 -9.18603361e-01 -1.43768027e-01 -4.37971443e-01 4.74615514e-01 -6.24292552e-01 -7.60859489e-01 -6.62556946e-01 -1.34898102e+00 -2.69090235e-01 -7.23110795e-01 4.75561351e-01 5.28379858e-01 1.60600972e+00 3.61576468e-01 1.52191728e-01 2.10619956e-01 -1.29361406e-01 -3.03821981e-01 -6.23719990e-01 -4.63394076e-01 6.05437100e-01 -4.70746338e-01 -1.12271416e+00 -1.33169040e-01 4.85332422e-02]
[9.238590240478516, 8.20124626159668]
8e9936db-9422-40d5-9511-4ec7b1f36fcc
a-case-study-on-record-matching-of
2302.07784
null
https://arxiv.org/abs/2302.07784v1
https://arxiv.org/pdf/2302.07784v1.pdf
A Case Study on Record Matching of Individuals in Historical Archives of Indigenous Databases
Digitization of historical records has produced a significant amount of data for analysis and interpretation. A critical challenge is the ability to relate historical information across different archives to allow for the data to be framed in the appropriate historical context. This paper presents a real-world case study on historical information integration and record matching with the goal to improve the historical value of archives containing data in the period 1800 to 1920. The archives contain unique information about M\'etis and Indigenous people in Canada and interactions with European settlers. The archives contain thousands of records that have increased relevance when relationships and interconnections are discovered. The contribution is a record linking approach suitable for historical archives and an evaluation of its effectiveness. Experimental results demonstrate potential for discovering historical linkage with high precision enabling new historical discoveries.
['Ramon Lawrence', 'Matthew Currie']
2023-02-15
null
null
null
null
['record-linking']
['natural-language-processing']
[ 2.97194924e-02 -1.98309958e-01 -2.52369255e-01 -4.39310819e-01 -7.87581027e-01 -8.75204742e-01 1.00080562e+00 7.49689221e-01 -5.77728748e-01 9.67234731e-01 7.99657404e-01 -4.34340954e-01 -7.57194221e-01 -1.03138328e+00 -4.32108521e-01 -1.40171930e-01 -6.55197024e-01 6.02283537e-01 2.79656768e-01 -4.26530659e-01 4.06106770e-01 8.35799515e-01 -1.32193017e+00 9.49250683e-02 3.16995949e-01 3.71106863e-01 3.51640671e-01 5.99591613e-01 -2.62643635e-01 2.65701324e-01 -9.43483889e-01 -5.53802609e-01 2.68025011e-01 -1.48341566e-01 -1.09410870e+00 -5.94769895e-01 6.90458596e-01 -9.10426304e-02 -3.98498893e-01 7.29864299e-01 5.42409956e-01 1.07482551e-02 2.19240695e-01 -9.07751143e-01 -5.36036134e-01 1.03807020e+00 -3.36402595e-01 8.52083623e-01 5.93448222e-01 -4.29807454e-01 9.01282310e-01 -4.26152915e-01 1.19361889e+00 1.00856614e+00 1.00996268e+00 -7.09948689e-02 -1.13968968e+00 -5.86281419e-01 -3.33166778e-01 2.05103770e-01 -1.58913553e+00 -7.85472393e-01 3.41812372e-01 -4.23989654e-01 9.34485912e-01 5.18257380e-01 9.81656790e-01 3.20866853e-01 1.08940259e-01 -2.18432993e-01 3.53747666e-01 -7.30051637e-01 -1.34829059e-01 -1.79510847e-01 1.92880332e-01 6.59330711e-02 4.45985228e-01 -4.21382263e-02 -7.75949299e-01 -6.37978017e-01 7.76284933e-01 -1.58852875e-01 -2.48235926e-01 5.63466251e-01 -1.01611352e+00 4.61881369e-01 3.40516686e-01 9.00190115e-01 -2.02310413e-01 2.34337479e-01 4.15728986e-01 3.70146453e-01 2.56069422e-01 4.80157405e-01 -1.17747150e-01 -2.44918436e-01 -1.16807008e+00 7.69683957e-01 7.73958683e-01 1.16074955e+00 5.55600107e-01 -2.29036838e-01 6.94761038e-01 7.54586339e-01 3.57734472e-01 3.61162722e-01 1.15526006e-01 -8.22985888e-01 7.40611911e-01 5.66374600e-01 1.42354801e-01 -1.20530200e+00 -5.89280307e-01 -1.57349110e-01 -2.61671007e-01 1.06650218e-02 4.77807432e-01 1.80441767e-01 -4.17789847e-01 1.48458374e+00 3.48867327e-01 -3.26108523e-02 3.56729835e-01 4.00002629e-01 6.36356473e-01 6.19528949e-01 6.55938610e-02 -2.77366638e-01 1.36105287e+00 2.57654011e-01 -6.19144022e-01 4.74458709e-02 3.42051923e-01 -9.71563220e-01 2.96400070e-01 -1.51137665e-01 -1.13461030e+00 -3.52382421e-01 -9.44435239e-01 -1.32019252e-01 -4.22241271e-01 -4.59122181e-01 7.15678334e-01 2.56439596e-01 -1.00776291e+00 5.95375478e-01 -7.85408795e-01 -8.75776470e-01 1.68679580e-01 3.85246962e-01 -4.39854383e-01 4.60616380e-01 -1.21681869e+00 1.01971316e+00 8.15122008e-01 1.61211178e-01 -2.15424746e-01 -1.07090449e+00 -6.25256658e-01 -1.92771628e-01 -2.19884962e-01 -2.90540665e-01 1.05526757e+00 -4.95488316e-01 -1.90616950e-01 7.25029349e-01 5.10798097e-02 -5.99185526e-01 4.60280597e-01 5.04390858e-02 -1.21111155e+00 7.63165280e-02 6.70128167e-01 -5.10185957e-02 -3.42107832e-01 -9.43002462e-01 -1.25178754e+00 -5.73461115e-01 -4.07314628e-01 -2.23088428e-01 -2.68204093e-01 5.72085202e-01 -4.25573617e-01 -9.97307122e-01 5.50938360e-02 -5.01298368e-01 -2.01933105e-02 -2.42317855e-01 3.75750154e-01 6.85173720e-02 6.66436493e-01 -1.19004917e+00 1.65344679e+00 -1.69739926e+00 -6.71935230e-02 5.92513561e-01 -2.49309957e-01 1.04913242e-01 2.74009079e-01 1.22486341e+00 5.45885460e-03 3.76257211e-01 -4.46291298e-01 1.69296384e-01 -5.77216208e-01 6.04276657e-01 -6.67106509e-01 5.27845621e-01 -2.73829639e-01 4.31844145e-01 -1.04313195e+00 -7.41178572e-01 -4.69735079e-02 2.90789783e-01 4.31133434e-03 -1.57123983e-01 8.22053030e-02 2.63266206e-01 -4.58495706e-01 7.39215851e-01 3.18556041e-01 3.03302467e-01 4.73768294e-01 8.40285420e-02 -8.90716195e-01 3.58407706e-01 -1.14030981e+00 1.86001205e+00 -5.82041554e-02 1.06459105e+00 -1.67679101e-01 -2.33262658e-01 1.18958211e+00 5.64471424e-01 4.48103130e-01 -7.26533234e-01 -3.35177183e-01 5.19537032e-01 -1.52854130e-01 -5.91377735e-01 1.21646559e+00 -1.73649862e-01 -2.88651854e-01 3.93117338e-01 -1.68312103e-01 1.89920262e-01 3.94038200e-01 2.92085558e-01 9.53098178e-01 6.59874082e-02 4.11810338e-01 -4.49813455e-01 3.06194395e-01 6.77201509e-01 6.97585404e-01 4.96531129e-01 4.64163870e-01 3.26546669e-01 -7.75422454e-02 -8.75483096e-01 -1.66020262e+00 -9.16581333e-01 -8.09007764e-01 5.07515013e-01 -9.36299115e-02 -6.09336257e-01 -2.87086274e-02 2.32810885e-01 8.87373984e-02 6.26738191e-01 -6.33274078e-01 3.23508918e-01 -8.94674361e-01 -9.69393551e-01 1.02586842e+00 5.18659651e-01 1.92033529e-01 -7.89250672e-01 -8.24507892e-01 6.85632348e-01 -2.23743081e-01 -5.20013332e-01 5.99065311e-02 -2.84988642e-01 -1.02506292e+00 -8.67477775e-01 -3.26374114e-01 -5.51874280e-01 4.72412288e-01 -1.53890625e-01 8.25538337e-01 2.06928924e-01 -4.12695110e-01 2.08853230e-01 -4.13551271e-01 -3.32225561e-01 -5.90045631e-01 3.21290374e-01 1.50798440e-01 -5.66834092e-01 1.32774383e-01 -8.45552266e-01 -1.48670584e-01 3.09631586e-01 -1.00644124e+00 -4.71001834e-01 4.58306298e-02 3.38030279e-01 2.02424273e-01 7.88144618e-02 9.65817988e-01 -5.99597514e-01 3.51231307e-01 -7.53356934e-01 -8.26922059e-01 8.89542997e-01 -4.80713844e-01 -1.99744374e-01 2.33679712e-01 3.81726511e-02 -1.28390372e+00 -6.51955754e-02 -1.42285287e-01 4.71277267e-01 1.98781535e-01 1.04446745e+00 1.96258470e-01 1.78244412e-01 7.57884920e-01 -1.78398907e-01 -3.74430686e-01 -1.03062820e+00 2.15366706e-01 8.71084452e-01 1.51580179e+00 -8.50546718e-01 5.31752467e-01 5.05986035e-01 -2.36385927e-01 -9.62488890e-01 -5.24078198e-02 -4.69574213e-01 -1.20464671e+00 -3.59855354e-01 3.95870745e-01 -9.30936575e-01 -2.79384464e-01 1.33830279e-01 -1.17495751e+00 2.11125985e-01 -2.99406886e-01 5.38617671e-01 -4.81061600e-02 2.29183987e-01 -2.67454565e-01 -8.09953809e-01 -3.85086358e-01 -2.24385366e-01 4.62736100e-01 4.32967842e-01 -7.61375368e-01 -1.16603541e+00 8.15098226e-01 -6.20899089e-02 -5.18520586e-02 7.52703369e-01 8.42314601e-01 -6.78397477e-01 -3.24520111e-01 -4.18359697e-01 4.81549576e-02 -7.36835659e-01 3.97334754e-01 4.89017934e-01 -7.31409311e-01 -2.60416418e-01 -5.20832121e-01 5.21142423e-01 3.59829664e-01 -1.89662557e-02 -1.66818410e-01 -2.94033438e-01 -6.46235049e-01 4.12631273e-01 1.59847462e+00 7.48952389e-01 8.08258712e-01 1.11477995e+00 4.15905923e-01 7.05837965e-01 2.36141056e-01 4.53717142e-01 4.47901100e-01 7.29279935e-01 -2.24698901e-01 1.49172068e-01 6.68328479e-02 -2.20512435e-01 -2.00403959e-01 6.37299836e-01 -5.07242262e-01 -1.06721416e-01 -1.66336870e+00 1.18572295e+00 -2.11373234e+00 -1.39428532e+00 -5.46482503e-01 2.27190304e+00 1.00117433e+00 -1.45053133e-01 1.90591887e-01 -4.27363999e-02 8.57377231e-01 -1.22952089e-01 -6.28798157e-02 -1.72333881e-01 -2.65020609e-01 2.96549369e-02 9.05861497e-01 5.06848574e-01 -6.34990752e-01 6.31890297e-01 7.85916042e+00 1.38022095e-01 -5.18857956e-01 -2.23813429e-01 -3.83801349e-02 -1.94586545e-01 -4.79346067e-01 4.94930714e-01 -1.11093187e+00 3.73967022e-01 1.26330471e+00 -7.12234974e-01 -4.00430448e-02 1.15761094e-01 1.90691307e-01 -2.96963960e-01 -7.45305717e-01 3.09304774e-01 -1.55038252e-01 -1.96978533e+00 1.08747117e-01 2.76922017e-01 4.97356266e-01 4.25570644e-02 -7.15401649e-01 -6.46104574e-01 5.62853277e-01 -4.92926210e-01 1.09786761e+00 1.20656860e+00 6.76594436e-01 -1.07209694e+00 5.75333178e-01 -3.08254540e-01 -1.45933127e+00 -6.40044659e-02 -4.18808758e-01 -1.66126758e-01 7.22201824e-01 1.03718720e-01 -8.77466500e-01 1.10617554e+00 9.12316442e-01 7.96364009e-01 -7.59883761e-01 1.41747499e+00 1.50141880e-01 4.31472480e-01 -7.04402387e-01 3.31896603e-01 2.02760965e-01 -3.72057021e-01 6.72804892e-01 1.22170234e+00 6.15508914e-01 4.37480986e-01 -2.63801187e-01 4.16582167e-01 1.30694687e-01 -5.74400686e-02 -6.40771329e-01 5.40542006e-02 1.08647346e+00 7.18963206e-01 -7.64682949e-01 -2.79713482e-01 -3.86686087e-01 5.97081110e-02 1.53578147e-01 7.14149373e-03 -2.37367153e-01 -5.83057761e-01 5.55057287e-01 4.12723362e-01 -1.64666548e-01 -3.70001644e-01 -1.36352450e-01 -5.96620500e-01 1.63266271e-01 -3.72593403e-01 1.03962433e+00 -5.98937690e-01 -9.81129706e-01 5.41054010e-01 5.08146942e-01 -9.61550295e-01 -4.74239171e-01 2.28804216e-01 -5.80390096e-01 1.01475096e+00 -8.05371523e-01 -1.47761643e+00 1.27303109e-01 1.70050174e-01 6.15353286e-02 -2.68511951e-01 6.96090519e-01 6.54680014e-01 -3.61842036e-01 1.89796641e-01 6.78390861e-01 4.09272105e-01 5.30237973e-01 -9.52157915e-01 1.05700600e+00 1.08259904e+00 5.75367995e-02 8.90683115e-01 8.08308005e-01 -1.35214865e+00 -1.05579102e+00 -8.82426977e-01 1.20592010e+00 -5.94607413e-01 1.06614196e+00 -8.24048743e-02 -1.34009063e+00 9.96865749e-01 2.47788411e-02 -6.09150171e-01 9.47997689e-01 1.41711563e-01 -1.34563684e-01 -2.71388948e-01 -9.75313306e-01 3.91495436e-01 1.09279478e+00 -6.26385152e-01 -1.10614669e+00 -7.02348799e-02 2.83380300e-01 2.77416799e-02 -1.61234999e+00 -2.11301725e-02 1.17322969e+00 -2.07575902e-01 1.01176071e+00 -3.72857839e-01 1.00052170e-01 -5.97497702e-01 -2.25058347e-01 -7.98193157e-01 -4.93679121e-02 -7.36282110e-01 7.80180216e-01 2.18009186e+00 5.32410920e-01 -3.98871124e-01 1.09333374e-01 9.85905349e-01 -1.51316240e-01 5.06828785e-01 -1.09502769e+00 -8.71583700e-01 -6.52748272e-02 -3.11137289e-01 1.11571300e+00 1.28738403e+00 2.40059093e-01 -2.80926377e-02 -3.59015167e-01 4.64380205e-01 5.53605258e-01 9.28307697e-02 6.13928080e-01 -1.64918029e+00 1.44428313e-01 -3.13149303e-01 -7.78063715e-01 6.79202229e-02 -4.41640645e-01 -8.94470036e-01 -4.64374244e-01 -1.68285346e+00 -3.09999347e-01 -8.96298289e-01 1.88505322e-01 5.98801911e-01 2.06255883e-01 1.56564564e-01 5.58758639e-02 9.05047238e-01 2.44458646e-01 -1.45781934e-01 2.73227364e-01 1.80841714e-01 -4.77976888e-01 -2.50534892e-01 -4.32987541e-01 4.53070343e-01 6.26810789e-01 -8.48559916e-01 3.06946002e-02 -5.49416065e-01 6.00013375e-01 2.81623065e-01 3.00372839e-01 -1.07849061e+00 7.27251053e-01 -4.80450429e-02 5.30190110e-01 -8.83326113e-01 -1.48242399e-01 -9.36529040e-01 1.48798811e+00 4.49150950e-01 -2.22737595e-01 5.34866869e-01 5.30647635e-01 5.58824062e-01 -1.30409911e-01 -5.44356465e-01 2.59047240e-01 -1.51618287e-01 -1.17391849e+00 -1.71163470e-01 -3.75947833e-01 -2.48151734e-01 9.27007675e-01 -2.52862602e-01 -6.72164917e-01 1.85444027e-01 -8.14529061e-01 4.88367766e-01 8.54519010e-01 4.36803997e-01 1.66522697e-01 -1.35852635e+00 -9.11216438e-01 -1.02772443e-02 3.18114460e-01 -1.90299064e-01 -2.23183259e-01 1.05905496e-01 -1.04674053e+00 1.02810234e-01 -7.12949157e-01 -2.15392873e-01 -1.42631435e+00 3.20824772e-01 2.16734469e-01 2.70885140e-01 -1.04388368e+00 3.48180413e-01 -7.87383854e-01 -1.36268273e-01 -1.75904483e-02 8.52082446e-02 -3.07577789e-01 7.34852493e-01 1.09532130e+00 6.76681519e-01 9.11308601e-02 -1.01629078e+00 -4.58549798e-01 6.38310194e-01 -7.08644688e-02 -7.57741332e-01 1.93402529e+00 -4.53525543e-01 -4.36641157e-01 7.18783796e-01 9.11433816e-01 8.22783336e-02 -8.04939508e-01 -3.50882888e-01 7.82860100e-01 -6.80204391e-01 -1.49117529e-01 -7.34102011e-01 -7.75420487e-01 1.26201928e-01 5.42388797e-01 2.71508247e-01 8.13627422e-01 2.60562032e-01 4.52240437e-01 4.48659778e-01 5.39238334e-01 -9.86867189e-01 -1.26087475e+00 -9.26309079e-02 9.89948153e-01 -5.18641114e-01 6.90133452e-01 1.37685269e-01 -2.90987771e-02 1.44598508e+00 -1.05892010e-01 3.74396533e-01 5.59308708e-01 2.84305006e-01 7.35066086e-02 -5.15909731e-01 -3.37680995e-01 -2.23834381e-01 4.19507548e-02 5.33901215e-01 4.18036342e-01 -2.22456455e-01 -7.31263578e-01 6.64057210e-02 -5.88049173e-01 2.06689313e-02 4.94921923e-01 1.41074824e+00 -4.48275447e-01 -1.36478782e+00 -8.25732470e-01 2.95123369e-01 -4.98709142e-01 -6.98942924e-04 -5.67102969e-01 1.08533323e+00 1.78579286e-01 7.67597437e-01 7.10267782e-01 3.28281745e-02 4.78115857e-01 -8.97732526e-02 1.94591880e-01 -2.86941618e-01 -1.18039751e+00 2.27348298e-01 7.00398088e-01 1.95181772e-01 -8.27926636e-01 -1.23233891e+00 -1.43367708e+00 -7.19487846e-01 -2.56054878e-01 6.65927172e-01 7.45553136e-01 7.06244946e-01 2.13979945e-01 2.17099056e-01 1.75907642e-01 -4.36004341e-01 4.40255374e-01 -6.18847311e-01 -6.88496768e-01 1.97062716e-02 2.39117533e-01 -1.89684853e-01 1.67882562e-01 6.54987156e-01]
[9.903570175170898, 9.716473579406738]
ae01ff2f-f7e1-47ef-8f2b-3f5e426c3a84
distilling-self-supervised-vision-1
2307.03407
null
https://arxiv.org/abs/2307.03407v1
https://arxiv.org/pdf/2307.03407v1.pdf
Distilling Self-Supervised Vision Transformers for Weakly-Supervised Few-Shot Classification & Segmentation
We address the task of weakly-supervised few-shot image classification and segmentation, by leveraging a Vision Transformer (ViT) pretrained with self-supervision. Our proposed method takes token representations from the self-supervised ViT and leverages their correlations, via self-attention, to produce classification and segmentation predictions through separate task heads. Our model is able to effectively learn to perform classification and segmentation in the absence of pixel-level labels during training, using only image-level labels. To do this it uses attention maps, created from tokens generated by the self-supervised ViT backbone, as pixel-level pseudo-labels. We also explore a practical setup with ``mixed" supervision, where a small number of training images contains ground-truth pixel-level labels and the remaining images have only image-level labels. For this mixed setup, we propose to improve the pseudo-labels using a pseudo-label enhancer that was trained using the available ground-truth pixel-level labels. Experiments on Pascal-5i and COCO-20i demonstrate significant performance gains in a variety of supervision settings, and in particular when little-to-no pixel-level labels are available.
['Naila Murray', 'Minsu Cho', 'Piotr Koniusz', 'Dahyun Kang']
2023-07-07
distilling-self-supervised-vision
http://openaccess.thecvf.com//content/CVPR2023/html/Kang_Distilling_Self-Supervised_Vision_Transformers_for_Weakly-Supervised_Few-Shot_Classification__Segmentation_CVPR_2023_paper.html
http://openaccess.thecvf.com//content/CVPR2023/papers/Kang_Distilling_Self-Supervised_Vision_Transformers_for_Weakly-Supervised_Few-Shot_Classification__Segmentation_CVPR_2023_paper.pdf
cvpr-2023-1
['few-shot-image-classification', 'pseudo-label']
['computer-vision', 'miscellaneous']
[ 7.16965497e-01 4.38855261e-01 -5.28742433e-01 -7.21346676e-01 -1.00354624e+00 -4.83248353e-01 5.22039056e-01 -1.35326162e-01 -5.60298979e-01 6.18693292e-01 -1.35940805e-01 6.12434894e-02 6.15394652e-01 -6.89290702e-01 -1.20336699e+00 -7.08194196e-01 3.42790753e-01 6.08154178e-01 2.93512851e-01 5.41665964e-02 -1.14186041e-01 -2.57828623e-01 -1.71597183e+00 3.19313228e-01 7.55993187e-01 1.18078148e+00 3.85540098e-01 7.57201254e-01 -7.49674663e-02 1.30977643e+00 -3.92746389e-01 -2.66295910e-01 3.76511037e-01 -5.48591912e-01 -1.07450128e+00 6.54649496e-01 6.57794535e-01 -3.46919656e-01 -1.30816996e-01 1.17721581e+00 3.86921801e-02 6.56180969e-03 5.13316631e-01 -1.11789179e+00 -7.14017987e-01 5.90362787e-01 -5.39533854e-01 -4.07366008e-02 -6.10690191e-02 6.26226485e-01 1.29035568e+00 -6.64557338e-01 7.86355376e-01 9.84763801e-01 5.41719854e-01 7.83269882e-01 -1.44049609e+00 -4.36549991e-01 1.50221422e-01 -2.24914197e-02 -1.06502008e+00 -3.06275070e-01 7.85003424e-01 -4.09510702e-01 6.89627111e-01 -3.68857652e-01 5.33882916e-01 1.13320458e+00 -2.66805261e-01 1.04609287e+00 1.28520930e+00 -5.83154500e-01 3.63356322e-01 3.55635494e-01 3.03642511e-01 9.57530022e-01 -2.47002140e-01 -1.22185647e-01 -5.95195353e-01 3.18784624e-01 7.33760059e-01 7.02921972e-02 -8.88786465e-02 -2.59039372e-01 -1.19709301e+00 7.24112809e-01 5.07312059e-01 2.74286956e-01 -2.88789570e-01 4.60065246e-01 3.10536027e-01 1.10934466e-01 6.23313665e-01 3.04729342e-01 -4.99943882e-01 1.05301894e-01 -1.25156617e+00 -3.02239656e-01 7.02340901e-01 1.15758634e+00 1.34936166e+00 1.13266446e-01 -4.62158024e-01 8.86816502e-01 4.23486829e-02 4.54893798e-01 4.98716235e-01 -1.26526022e+00 3.01826209e-01 3.36534441e-01 8.09151605e-02 -1.97967723e-01 -1.85562707e-02 -4.25844759e-01 -6.39726639e-01 2.17551529e-01 3.97619367e-01 -2.03040272e-01 -1.74280095e+00 1.95129323e+00 2.66794473e-01 4.62358177e-01 1.41182080e-01 8.44429195e-01 5.77159584e-01 5.41779101e-01 1.94947258e-01 2.58007217e-02 1.16515088e+00 -1.44841218e+00 -5.20138860e-01 -6.30013168e-01 6.59259081e-01 -3.70677799e-01 1.24458694e+00 2.17255112e-02 -1.08146608e+00 -8.40775192e-01 -1.12415636e+00 -2.07520917e-01 -3.97477239e-01 1.92169875e-01 3.38372082e-01 3.81150186e-01 -1.31221998e+00 6.74293578e-01 -8.30585241e-01 -1.75608024e-01 9.25367594e-01 1.33938134e-01 -1.25154898e-01 -3.05558801e-01 -1.07523060e+00 7.03735292e-01 4.13616270e-01 -1.01610996e-01 -1.48523843e+00 -6.96064472e-01 -1.15716588e+00 -3.20015475e-03 5.88250041e-01 -4.27909911e-01 1.26892114e+00 -1.39366651e+00 -1.39678252e+00 1.31361175e+00 -1.44519970e-01 -6.06533408e-01 4.91036355e-01 -1.68986563e-02 1.69908851e-01 4.09613311e-01 5.79795539e-01 1.28947222e+00 9.74345505e-01 -1.42067230e+00 -7.24843025e-01 -2.08197623e-01 2.63476580e-01 1.84113860e-01 -1.01958610e-01 -3.74712080e-01 -6.64152443e-01 -2.79370308e-01 -9.44512337e-02 -1.00527036e+00 -3.94453555e-01 6.49416670e-02 -7.31471062e-01 -6.91407621e-02 7.10005581e-01 -4.23319995e-01 3.00321937e-01 -2.06770062e+00 -1.10510863e-01 -7.91271329e-02 6.38261214e-02 3.18460435e-01 -3.47034723e-01 -1.23567328e-01 1.06183323e-03 -2.51719467e-02 -6.04885221e-01 -9.81821656e-01 -2.12201923e-01 6.17364705e-01 -1.81864753e-01 3.80636781e-01 6.83816433e-01 1.13023233e+00 -1.20217025e+00 -7.47902870e-01 4.83158380e-01 2.91319072e-01 -3.22264522e-01 4.04353350e-01 -8.64332438e-01 5.72341084e-01 -1.75819620e-01 8.50494027e-01 3.69134307e-01 -5.44856131e-01 -1.41304657e-01 -3.49849416e-03 7.59245306e-02 6.44180402e-02 -7.13056266e-01 2.03981113e+00 -6.53025925e-01 6.58134937e-01 2.00028971e-01 -1.30518758e+00 7.07673907e-01 1.44684285e-01 3.98754060e-01 -6.18172884e-01 2.43557036e-01 1.70339495e-02 -4.86702532e-01 -3.84865075e-01 1.53850049e-01 -2.08484367e-01 -3.71107310e-02 4.94417965e-01 7.22329974e-01 -3.28376591e-01 2.92539775e-01 3.54556978e-01 1.24517810e+00 4.15632844e-01 -4.69046347e-02 1.81256086e-02 2.83428371e-01 2.34682947e-01 6.72371507e-01 9.56555188e-01 -3.16799700e-01 9.81097162e-01 5.19094050e-01 -8.61685202e-02 -1.22241330e+00 -8.43778729e-01 -8.44502673e-02 1.33682919e+00 2.05011025e-01 -1.28214121e-01 -9.88822877e-01 -8.63438189e-01 -2.52740651e-01 7.08562493e-01 -1.00787389e+00 -1.78468432e-02 -1.49896279e-01 -4.14055139e-01 3.94556761e-01 6.31370068e-01 6.49507165e-01 -1.30048525e+00 -6.97935879e-01 2.13779211e-01 -1.05301976e-01 -1.57079530e+00 -3.70046020e-01 8.39086950e-01 -5.90438068e-01 -1.00323284e+00 -8.55045021e-01 -1.02291727e+00 8.10782075e-01 2.37218499e-01 1.08523369e+00 5.78798652e-02 -4.87814516e-01 4.09679234e-01 -3.60643297e-01 -2.61767209e-01 -3.93840969e-01 5.95549075e-03 -5.25101900e-01 2.36852631e-01 7.58032724e-02 -4.07218099e-01 -5.12262642e-01 2.58344322e-01 -8.47682476e-01 3.29386413e-01 5.90540886e-01 1.20877087e+00 9.13604915e-01 -2.10646376e-01 4.03002709e-01 -1.45046091e+00 -3.54211360e-01 -3.66835982e-01 -4.01817709e-01 2.84192801e-01 -4.27358687e-01 7.60453343e-02 7.69805133e-01 -3.46593559e-01 -1.13156772e+00 5.12166440e-01 -7.41099268e-02 -9.33527946e-01 -4.21508938e-01 1.98159605e-01 -1.49331763e-01 3.98222096e-02 6.27032876e-01 3.02450091e-01 2.69880220e-02 -3.11568171e-01 8.15784216e-01 6.69414222e-01 9.50738668e-01 -4.57261235e-01 6.30493522e-01 7.31008351e-01 -3.30355912e-01 -6.49636030e-01 -1.65217936e+00 -7.67091453e-01 -8.84364724e-01 -1.87453613e-01 1.35409725e+00 -1.18170679e+00 -7.11232498e-02 6.98102295e-01 -1.03559721e+00 -9.99021649e-01 -8.23160172e-01 8.32779035e-02 -9.09336686e-01 -1.75934639e-02 -8.49529207e-01 -6.48595333e-01 -2.20372230e-01 -1.25006151e+00 1.33682537e+00 3.50161076e-01 2.45473310e-01 -1.08256042e+00 -1.45343214e-01 5.96613109e-01 1.71830535e-01 2.94639856e-01 5.26219010e-01 -5.63564241e-01 -6.48704708e-01 -8.32803845e-02 -4.06023979e-01 7.54414499e-01 6.95684254e-02 -3.00520569e-01 -1.56302035e+00 -1.39166629e-02 -1.27538860e-01 -1.04971325e+00 1.20388043e+00 3.74415040e-01 1.27593887e+00 -5.41634336e-02 -1.37019947e-01 7.47169018e-01 1.42470574e+00 -2.87734121e-01 4.19982165e-01 -8.04657303e-03 1.12004364e+00 7.44564474e-01 6.46354079e-01 1.74594969e-01 5.94271839e-01 3.72682065e-01 4.55476254e-01 -5.89097738e-01 -6.34540737e-01 -3.70817542e-01 2.16164470e-01 4.71003175e-01 3.10684711e-01 -6.65582716e-02 -7.76511431e-01 8.75777543e-01 -1.83083844e+00 -7.49711871e-01 2.18716070e-01 1.87950039e+00 1.28583300e+00 3.67252588e-01 -4.15793732e-02 -1.00597516e-01 8.14233840e-01 3.43866557e-01 -1.01675427e+00 -4.75730859e-02 -6.66109771e-02 3.47001851e-01 9.18933988e-01 3.91795754e-01 -1.45002258e+00 1.42824006e+00 5.84731722e+00 5.57271123e-01 -1.12248325e+00 5.37196994e-01 1.15866292e+00 1.69954654e-02 -1.53732464e-01 1.67915210e-01 -7.29789674e-01 4.47183162e-01 9.75747108e-01 3.07289511e-01 2.13019863e-01 1.08696795e+00 -1.65959045e-01 -3.70528519e-01 -1.32336605e+00 7.79706299e-01 3.13205481e-01 -1.30145967e+00 -3.37798119e-01 -1.58110216e-01 1.23708403e+00 3.67365152e-01 1.77779645e-02 3.92710447e-01 7.46763170e-01 -8.05383384e-01 7.99020946e-01 2.89664775e-01 1.22513330e+00 -1.90885499e-01 6.84056818e-01 3.09461445e-01 -9.58553851e-01 -2.78373789e-02 -2.83930212e-01 5.00523858e-02 1.31905600e-01 5.85433304e-01 -7.54572868e-01 2.21991032e-01 6.72290325e-01 1.18153989e+00 -4.51273739e-01 6.21415198e-01 -6.26753867e-01 9.62607682e-01 -2.79967934e-01 3.82620305e-01 6.04959846e-01 -1.13710754e-01 6.11213176e-03 1.17018735e+00 -2.90965229e-01 -2.96524651e-02 4.64114457e-01 1.17360735e+00 -3.19499701e-01 -3.05977821e-01 -3.74284536e-01 5.84581494e-02 2.41503701e-01 1.25915873e+00 -1.01798582e+00 -7.91249871e-01 -5.27223706e-01 1.29869938e+00 4.05324340e-01 6.05972052e-01 -6.76713943e-01 -1.37321115e-01 2.66666472e-01 -3.22452448e-02 4.91561830e-01 1.52759731e-01 -5.14869809e-01 -1.13949239e+00 -1.39734760e-01 -4.76573825e-01 2.70980924e-01 -9.91819322e-01 -1.21494269e+00 3.30703348e-01 -9.46084857e-02 -8.94194841e-01 -2.77252585e-01 -4.44955885e-01 -8.27922523e-01 6.04656398e-01 -1.90848362e+00 -1.59797776e+00 -3.88823152e-01 4.80172753e-01 8.92703950e-01 1.71579063e-01 5.96495271e-01 -1.36593347e-02 -5.88927627e-01 5.88120162e-01 7.64940353e-03 4.40106302e-01 6.74790025e-01 -1.60004306e+00 3.81629229e-01 8.01740646e-01 3.80847573e-01 2.85013225e-02 6.08583272e-01 -4.59813207e-01 -9.27704573e-01 -1.49098563e+00 4.79857206e-01 -3.59914958e-01 5.84595323e-01 -6.13524079e-01 -8.84831429e-01 8.81924748e-01 1.55650869e-01 5.53382814e-01 4.40532833e-01 -1.50414929e-01 -3.92844319e-01 3.15653272e-02 -1.13917947e+00 1.86929047e-01 9.78083253e-01 -7.54990101e-01 -5.02479076e-01 7.06766009e-01 9.70878541e-01 -3.69531721e-01 -5.06933570e-01 6.54194877e-02 1.25808865e-01 -6.96817577e-01 8.30322921e-01 -1.89303026e-01 6.58929884e-01 -1.08082972e-01 -2.84356344e-02 -1.24536884e+00 2.03935623e-01 -3.19921285e-01 2.21120447e-01 1.32990217e+00 4.61002052e-01 -2.79210359e-01 1.05736148e+00 5.30560195e-01 -2.63680905e-01 -7.16794848e-01 -8.03736687e-01 -5.45962989e-01 -7.63567686e-02 -4.79189008e-01 1.01502888e-01 9.23049092e-01 -3.17524552e-01 4.20953184e-01 -5.20066977e-01 -1.39620170e-01 9.12228644e-01 8.85952637e-02 6.84101641e-01 -9.37492013e-01 -4.43869323e-01 -1.58712436e-02 -4.50356036e-01 -1.04985535e+00 5.96679688e-01 -9.48174477e-01 7.20391810e-01 -1.46401572e+00 4.13987309e-01 -6.18017972e-01 -3.50692302e-01 1.06014442e+00 -1.50324941e-01 7.21469402e-01 2.11879089e-01 2.49251992e-01 -1.08560562e+00 5.04778922e-01 1.24854136e+00 -4.90781844e-01 3.36607620e-02 -1.33008033e-01 -4.48972672e-01 7.22123027e-01 5.42424262e-01 -5.62306941e-01 -4.81413841e-01 -3.74775350e-01 -3.32302779e-01 -1.63815711e-02 4.11876827e-01 -1.08705354e+00 2.96125770e-01 1.23594366e-02 3.23662668e-01 -4.61665630e-01 4.64682370e-01 -4.72405255e-01 -4.89089459e-01 1.65945455e-01 -8.08848262e-01 -6.63181186e-01 -2.33436495e-01 7.41776645e-01 -1.82687715e-01 -4.01766181e-01 1.06723380e+00 -4.12609071e-01 -9.82951105e-01 3.56701881e-01 -1.24535665e-01 4.20168400e-01 1.15414202e+00 -1.66191936e-01 -1.40649065e-01 -2.97981203e-01 -9.11238730e-01 6.02852166e-01 6.59982860e-01 1.33342177e-01 4.34024632e-01 -1.05973840e+00 -4.88952190e-01 2.31165558e-01 3.27984810e-01 4.71314490e-01 1.69328421e-01 5.40586829e-01 -2.46948466e-01 1.95692882e-01 -2.21080273e-01 -9.64484453e-01 -8.39577019e-01 5.31495154e-01 3.34048033e-01 -1.35138795e-01 -6.67635500e-01 1.21244645e+00 3.43669027e-01 -3.79203022e-01 3.61972958e-01 -4.36820030e-01 1.42090037e-01 1.60429850e-02 4.17924255e-01 -1.02203280e-01 -1.59694314e-01 -7.12047398e-01 -7.71962386e-03 4.95345414e-01 -1.30309045e-01 -3.02191854e-01 1.27960539e+00 -2.25138590e-01 1.72432810e-01 8.41080606e-01 1.41167772e+00 -6.61295891e-01 -1.91977954e+00 -6.87873542e-01 -8.89374092e-02 -2.42738515e-01 2.31918007e-01 -8.17607760e-01 -1.32165968e+00 1.16513550e+00 4.68509704e-01 -2.91718125e-01 8.52773309e-01 3.00116658e-01 8.52243125e-01 3.30298454e-01 3.82163346e-01 -1.32043684e+00 4.70378160e-01 4.57934737e-01 5.75254671e-02 -1.67544007e+00 -4.67516482e-01 -3.44588429e-01 -9.29364383e-01 6.18381739e-01 7.55860865e-01 -2.50751406e-01 4.74430114e-01 1.97860271e-01 2.76933402e-01 -1.04718231e-01 -7.74156153e-01 -7.46547043e-01 -5.53137325e-02 8.30371916e-01 1.48301914e-01 3.03018726e-02 3.68798494e-01 3.27200502e-01 1.52300000e-01 7.69398212e-02 5.79644740e-01 9.16999280e-01 -4.63843018e-01 -9.71210361e-01 -1.63376048e-01 7.12684274e-01 -3.06096762e-01 -1.63309023e-01 -1.82105467e-01 3.26657325e-01 3.50575060e-01 9.60052729e-01 3.51378918e-01 -2.85012156e-01 -1.66986063e-01 2.64357209e-01 2.83351362e-01 -1.13352644e+00 -3.40595335e-01 6.33643121e-02 -1.58824041e-01 -6.42103195e-01 -6.79959953e-01 -4.84097809e-01 -1.28170598e+00 3.28811049e-01 -3.16614449e-01 -1.70235708e-02 7.32580006e-01 1.14750171e+00 -2.10178215e-02 6.22105658e-01 6.92593992e-01 -1.04471540e+00 -4.63146865e-01 -9.08745885e-01 -6.33496523e-01 7.24250972e-01 5.38366377e-01 -5.46158075e-01 -4.61975902e-01 5.75083137e-01]
[9.67390251159668, 0.8227362632751465]
cdbf3d6f-8cc6-4442-8557-aa39dce23d97
pretrained-language-encoders-are-natural
2208.09617
null
https://arxiv.org/abs/2208.09617v1
https://arxiv.org/pdf/2208.09617v1.pdf
Pretrained Language Encoders are Natural Tagging Frameworks for Aspect Sentiment Triplet Extraction
Aspect Sentiment Triplet Extraction (ASTE) aims to extract the spans of aspect, opinion, and their sentiment relations as sentiment triplets. Existing works usually formulate the span detection as a 1D token tagging problem, and model the sentiment recognition with a 2D tagging matrix of token pairs. Moreover, by leveraging the token representation of Pretrained Language Encoders (PLEs) like BERT, they can achieve better performance. However, they simply leverage PLEs as feature extractors to build their modules but never have a deep look at what specific knowledge does PLEs contain. In this paper, we argue that instead of further designing modules to capture the inductive bias of ASTE, PLEs themselves contain "enough" features for 1D and 2D tagging: (1) The token representation contains the contextualized meaning of token itself, so this level feature carries necessary information for 1D tagging. (2) The attention matrix of different PLE layers can further capture multi-level linguistic knowledge existing in token pairs, which benefits 2D tagging. (3) Furthermore, with simple transformations, these two features can also be easily converted to the 2D tagging matrix and 1D tagging sequence, respectively. That will further boost the tagging results. By doing so, PLEs can be natural tagging frameworks and achieve a new state of the art, which is verified by extensive experiments and deep analyses.
['Yongqi Tong', 'Chunxu Shen', 'Yong Dai', 'Lingqiao Liu', 'Yinjie Lei', 'Yanjie Gou']
2022-08-20
null
null
null
null
['aspect-sentiment-triplet-extraction']
['natural-language-processing']
[-3.40959817e-01 1.98336765e-01 -4.49464470e-01 -4.57599014e-01 -4.77880955e-01 -6.49709344e-01 4.38766271e-01 2.69640654e-01 -2.10895240e-01 2.29654595e-01 5.60935915e-01 -1.80544570e-01 3.31149042e-01 -9.44061697e-01 -5.43031991e-01 -4.62103635e-01 1.69030949e-01 1.40947863e-01 -7.15791853e-03 -4.89070326e-01 -8.05491433e-02 7.12916814e-03 -1.16626942e+00 4.31377143e-01 6.74960732e-01 1.51796269e+00 -1.41000897e-01 5.18432483e-02 -7.17146158e-01 8.95725310e-01 -5.46663105e-01 -7.51359344e-01 1.40180752e-01 -2.57456601e-01 -8.01020086e-01 1.34522676e-01 -4.68860148e-03 -3.69148254e-02 -5.86306155e-02 1.11991990e+00 1.80723861e-01 -2.20156580e-01 5.60592651e-01 -1.00639141e+00 -9.39220428e-01 1.01875055e+00 -6.13952994e-01 -3.49299520e-01 3.11550885e-01 8.69089067e-02 1.68259180e+00 -9.65183079e-01 2.12394580e-01 1.14106882e+00 7.99003661e-01 3.06367069e-01 -6.71035886e-01 -5.37988067e-01 4.62299824e-01 -1.74753800e-01 -9.74409699e-01 -1.08531237e-01 1.02802813e+00 -3.92538577e-01 1.05493343e+00 4.56239581e-02 1.12062347e+00 1.14740825e+00 -1.60412595e-01 1.34313631e+00 9.53596711e-01 -2.60830700e-01 -1.15270793e-01 3.54370028e-01 4.44809020e-01 7.22403228e-01 3.56973380e-01 -2.52384692e-01 -6.07644737e-01 2.70092040e-01 6.12660348e-01 1.68363810e-01 2.07051039e-02 -8.34955722e-02 -1.20037520e+00 7.51458347e-01 5.31201661e-01 6.21123731e-01 -2.74312288e-01 -8.79990533e-02 6.55859172e-01 2.62480497e-01 6.18217587e-01 6.02880239e-01 -8.88242185e-01 -9.02810097e-02 -5.19121170e-01 -2.63447225e-01 7.83296764e-01 1.08022451e+00 1.15390623e+00 1.21384254e-02 -4.20213282e-01 8.37120652e-01 5.25862217e-01 5.94675720e-01 6.68777347e-01 -1.81741372e-01 4.95781004e-01 1.28632319e+00 -1.08145364e-01 -9.80413854e-01 -5.91100812e-01 -6.60203695e-01 -8.06860507e-01 -5.16425967e-01 -1.56358838e-01 -4.31970924e-01 -6.99722648e-01 1.84915864e+00 2.22533599e-01 -1.86804309e-01 2.72801697e-01 6.32735729e-01 9.75427628e-01 3.21179152e-01 -4.11244556e-02 -1.70327984e-02 1.94504440e+00 -1.04467010e+00 -6.43789411e-01 -5.70663333e-01 9.43662822e-01 -8.18706572e-01 1.20451677e+00 -1.33023664e-01 -7.25367367e-01 -5.89894950e-01 -9.97203112e-01 -1.46214142e-01 -5.35001457e-01 3.86638045e-01 1.16818964e+00 5.52462041e-01 -7.01244473e-01 7.95311034e-02 -6.32344246e-01 -1.62592128e-01 4.46243674e-01 2.75428057e-01 -2.89804012e-01 1.06938872e-02 -1.56194675e+00 6.69301867e-01 3.87185246e-01 2.03721166e-01 -3.15417737e-01 -7.00581729e-01 -1.21032798e+00 2.70198345e-01 2.79881716e-01 -7.98958778e-01 1.08984244e+00 -1.00825918e+00 -1.26515210e+00 9.24703240e-01 -2.87010282e-01 -4.51059826e-02 -1.46265432e-01 -2.60956049e-01 -4.67279822e-01 -1.36618376e-01 3.78788859e-01 4.59025562e-01 5.28945565e-01 -1.05036438e+00 -7.80021369e-01 -3.73682320e-01 6.87284052e-01 1.60630405e-01 -9.10099089e-01 -4.27077077e-02 -5.67056060e-01 -6.37862027e-01 6.73786849e-02 -6.77790701e-01 -2.72277117e-01 -5.44518948e-01 -7.10850477e-01 -4.19394195e-01 5.51736474e-01 -2.16486618e-01 1.27963805e+00 -2.31999278e+00 -3.61312687e-01 1.61948353e-01 3.16088170e-01 3.61428529e-01 -3.01148295e-01 3.62221450e-01 2.89913975e-02 2.61482775e-01 1.62781477e-02 -5.10063171e-01 3.95027131e-01 1.86068624e-01 -3.70628744e-01 -1.85687050e-01 6.25861883e-01 1.41745627e+00 -9.68689859e-01 -4.29305524e-01 -1.81683570e-01 2.70111591e-01 -6.88185751e-01 3.90934348e-01 -2.57742733e-01 -1.05677135e-02 -7.13460445e-01 5.87727845e-01 5.90413809e-01 -2.97633827e-01 1.91804841e-01 -6.54889464e-01 3.78074646e-02 7.69615531e-01 -9.43690479e-01 1.47955394e+00 -7.57624567e-01 3.63947928e-01 -2.21958518e-01 -1.06003642e+00 1.20747650e+00 3.48286629e-01 6.17576778e-01 -6.49176896e-01 2.01613247e-01 4.31245891e-03 -2.31220767e-01 -4.98860478e-01 5.73825955e-01 -4.47120607e-01 -5.74333549e-01 6.07009768e-01 2.19623923e-01 1.63246781e-01 1.91756591e-01 1.11280315e-01 1.05643511e+00 -3.06611508e-02 1.84565127e-01 -5.58884777e-02 4.70286965e-01 -1.87251493e-01 9.51383531e-01 3.63940328e-01 5.83117902e-02 2.66065150e-01 9.97461677e-01 -4.35234874e-01 -7.16355562e-01 -6.02442980e-01 -5.57596087e-02 7.49664426e-01 2.52003342e-01 -8.54253113e-01 -4.94601130e-01 -1.15314364e+00 -1.58368006e-01 2.88950264e-01 -8.19506884e-01 -3.69533718e-01 -3.54100645e-01 -9.28308606e-01 5.36551297e-01 8.81960750e-01 6.73356771e-01 -9.62117910e-01 -1.74777821e-01 1.58291906e-01 -2.51037270e-01 -1.44295216e+00 -6.46639466e-01 5.35965621e-01 -5.29210806e-01 -1.01768410e+00 -2.80164033e-01 -9.52797115e-01 6.56332910e-01 3.69233072e-01 1.36752808e+00 1.70091271e-01 3.56799662e-01 2.22039014e-01 -7.50603199e-01 -5.29890478e-01 1.45863323e-02 4.56985950e-01 5.40530756e-02 2.05298886e-01 9.51590121e-01 -8.04617584e-01 -5.29062986e-01 2.25050315e-01 -9.38883841e-01 2.30306145e-02 8.85349989e-01 9.12011027e-01 5.78674197e-01 7.18513876e-03 4.71258193e-01 -1.28026283e+00 6.11812890e-01 -3.48116398e-01 -2.73766071e-01 1.39836654e-01 -5.06515503e-01 2.17784271e-01 7.73328900e-01 -2.90967822e-01 -1.04602075e+00 -2.13823244e-01 -4.93170649e-01 -2.51957178e-01 -1.18970533e-03 9.28628802e-01 -6.51078701e-01 3.98817331e-01 1.41995415e-01 1.72326431e-01 -3.81579071e-01 -6.50207281e-01 3.46305460e-01 7.02841580e-01 -3.86376381e-02 -6.81900680e-01 7.38056004e-01 3.02619755e-01 -2.42190570e-01 -3.51935267e-01 -1.82583547e+00 -5.28529704e-01 -7.66214073e-01 2.51219273e-01 7.99179614e-01 -1.17937970e+00 -6.55507684e-01 4.23283070e-01 -1.11876476e+00 6.09807074e-02 -5.68271101e-01 3.58104527e-01 -1.32286340e-01 2.48560622e-01 -7.04386890e-01 -6.01103723e-01 -4.03816521e-01 -9.61713374e-01 1.33969533e+00 3.45764846e-01 -1.33976385e-01 -1.27779520e+00 5.22753820e-02 1.81905597e-01 3.32852095e-01 6.04613312e-02 1.07920825e+00 -8.51479709e-01 -3.95314783e-01 -3.16072524e-01 -2.79929340e-01 5.93941748e-01 4.09720927e-01 -1.84057981e-01 -1.06880641e+00 3.76373492e-02 1.27672121e-01 -3.50005895e-01 9.25206780e-01 -9.96371433e-02 7.57287204e-01 -4.38421786e-01 -8.35281089e-02 6.35166943e-01 1.38687241e+00 -6.77570403e-02 4.04068172e-01 3.13824803e-01 1.13402200e+00 4.42802995e-01 5.44067800e-01 2.70263940e-01 1.01942766e+00 5.39424896e-01 3.61739904e-01 -4.58219558e-01 -2.15510070e-01 -6.38648272e-01 8.05226088e-01 1.55412555e+00 2.01736048e-01 1.98949203e-01 -6.83583021e-01 6.98845506e-01 -1.81503844e+00 -7.40394950e-01 1.70264058e-02 1.68540704e+00 9.65032578e-01 2.96887189e-01 -1.23567268e-01 2.54368410e-02 4.04933065e-01 5.34407914e-01 -4.80900377e-01 -4.31640565e-01 -4.55618829e-01 1.20833434e-01 8.15466493e-02 1.71537727e-01 -1.11244369e+00 1.18351269e+00 5.28992128e+00 8.51594210e-01 -1.08474386e+00 6.82277745e-03 4.23170567e-01 1.95987269e-01 -8.94207060e-01 2.73789883e-01 -1.12044537e+00 3.98483902e-01 5.31342983e-01 -5.62602468e-02 -1.68628529e-01 9.00878131e-01 -2.01951966e-01 2.79563397e-01 -1.15979362e+00 7.34015822e-01 -1.42701669e-03 -1.09925985e+00 3.11256528e-01 7.31602162e-02 8.45724642e-01 -7.25815818e-02 -9.36953072e-03 7.37030149e-01 3.73249918e-01 -6.96133614e-01 5.85323513e-01 3.67428958e-01 6.97287202e-01 -7.45267153e-01 1.24122870e+00 1.66429520e-01 -1.59332144e+00 6.06911909e-03 -6.07740521e-01 -1.09138362e-01 6.59697875e-02 1.21264470e+00 -4.98259842e-01 9.25085843e-01 5.10076702e-01 1.27240980e+00 -6.11537278e-01 6.11265361e-01 -7.91705072e-01 6.96424365e-01 -1.97434083e-01 -2.00982049e-01 4.92626756e-01 -1.39270484e-01 2.72946090e-01 1.30881357e+00 1.95036203e-01 -1.76654965e-01 2.64436245e-01 6.36123776e-01 -3.44424963e-01 2.83804029e-01 -4.60469693e-01 -1.94048688e-01 3.35568905e-01 1.50718164e+00 -5.69770217e-01 -4.25295591e-01 -7.80602396e-01 6.41929090e-01 6.06303036e-01 2.07579061e-01 -5.67303836e-01 -5.52368820e-01 1.07656026e+00 -2.47495309e-01 6.88319027e-01 2.19044201e-02 -3.91456485e-01 -1.81099582e+00 3.42350125e-01 -7.69320548e-01 2.05765098e-01 -4.71793741e-01 -1.69185615e+00 7.25062728e-01 -5.55075586e-01 -1.37462091e+00 9.84542593e-02 -7.79064536e-01 -8.02569389e-01 5.87573588e-01 -1.83861876e+00 -1.62261748e+00 5.54931629e-03 4.75908786e-01 2.46780723e-01 -1.09487489e-01 1.00166643e+00 3.35132182e-01 -7.68858373e-01 8.87339056e-01 -3.47541690e-01 7.71573126e-01 6.51993036e-01 -1.38453484e+00 3.91393691e-01 7.54287064e-01 4.23504353e-01 8.68983448e-01 3.63964438e-02 -3.14233571e-01 -1.27900577e+00 -1.17219770e+00 1.23982310e+00 -6.13079727e-01 9.14273083e-01 -4.94512439e-01 -7.34265625e-01 7.97682285e-01 1.36278063e-01 -1.27719089e-01 1.05668020e+00 8.78624856e-01 -7.89907753e-01 -3.61404240e-01 -5.81896901e-01 4.43747044e-01 1.05839312e+00 -8.70772243e-01 -7.37884879e-01 1.71895504e-01 1.12877083e+00 -1.64032489e-01 -1.10062683e+00 4.92001474e-01 4.77160782e-01 -9.93685484e-01 6.93618178e-01 -7.82841325e-01 8.36703658e-01 -2.84438640e-01 -1.63234279e-01 -1.32170200e+00 -1.98161498e-01 -2.95650959e-01 -1.05190575e-01 1.85621703e+00 7.40955591e-01 -5.49460113e-01 6.10236645e-01 4.07197982e-01 -2.66359061e-01 -1.20279741e+00 -5.32999575e-01 -5.89345574e-01 1.73180588e-02 -6.90940917e-01 1.20966077e+00 1.12002087e+00 3.29738826e-01 1.02095687e+00 -3.59869659e-01 4.20262739e-02 7.08910450e-02 6.63397253e-01 5.05535424e-01 -1.20016396e+00 -2.27187157e-01 -5.26752293e-01 -4.62490559e-01 -1.42479563e+00 3.60986561e-01 -1.13259423e+00 -1.68757126e-01 -1.58297789e+00 3.56572539e-01 -6.86066866e-01 -5.98231196e-01 9.66343284e-01 -5.98129094e-01 1.36568755e-01 1.21755421e-01 8.59270990e-02 -8.96907985e-01 8.52712274e-01 1.40003324e+00 -2.41774440e-01 1.21982604e-01 6.53112754e-02 -1.36518478e+00 8.63570929e-01 6.88945353e-01 -4.20193851e-01 -2.90879846e-01 -6.83389664e-01 8.56015444e-01 -3.59096915e-01 -6.08895943e-02 -5.34116089e-01 1.16672806e-01 6.82457089e-02 3.14404666e-01 -4.21035796e-01 1.50448501e-01 -8.74216616e-01 -5.71723759e-01 4.24573794e-02 -2.09344417e-01 6.06945567e-02 5.53989038e-02 2.11446270e-01 -7.04788089e-01 -2.22379059e-01 2.68052489e-01 -3.22740614e-01 -6.15880609e-01 5.85026562e-01 -1.64291784e-01 3.08487713e-01 4.95525867e-01 1.53695971e-01 -1.56659514e-01 -1.58058584e-01 -3.92461270e-01 4.02690798e-01 4.73343343e-01 5.20708919e-01 2.66871125e-01 -1.71493602e+00 -3.44776094e-01 3.98757339e-01 3.33610862e-01 2.51376569e-01 1.90727174e-01 7.93368042e-01 3.13751251e-01 3.26771349e-01 2.34548002e-02 -4.35251802e-01 -7.31470585e-01 5.61763823e-01 1.38496727e-01 -8.53039980e-01 -3.04344684e-01 9.74957585e-01 4.18708593e-01 -6.53618872e-01 -1.38248190e-01 -3.44128400e-01 -5.64471126e-01 6.15965188e-01 4.64954078e-01 -4.89916474e-01 1.22844614e-01 -4.85568255e-01 -3.70784611e-01 8.11710954e-01 -1.05378374e-01 2.61176229e-01 1.57098842e+00 -2.07857966e-01 -4.30117935e-01 4.83515948e-01 1.33217454e+00 4.82004136e-01 -9.12857890e-01 -5.54332376e-01 1.47391558e-01 -1.04658514e-01 -2.86340386e-01 -5.31427026e-01 -1.47806954e+00 1.11848462e+00 -1.75581425e-01 4.41458672e-01 1.19907665e+00 1.62721023e-01 1.12577760e+00 3.38214248e-01 1.69925690e-01 -8.85429442e-01 1.42482668e-01 1.00793874e+00 3.28527510e-01 -1.18518758e+00 -3.92377675e-01 -3.59450340e-01 -8.65976870e-01 9.43161905e-01 6.82644606e-01 6.21076636e-02 8.69435072e-01 4.12710994e-01 3.95277292e-01 -4.56785858e-01 -8.73982728e-01 -7.22311914e-01 3.90547842e-01 5.73103130e-01 6.22693777e-01 6.34243339e-02 -1.52337477e-01 1.48582554e+00 -5.95585287e-01 -3.82646143e-01 1.59890845e-01 5.59938371e-01 -2.67399132e-01 -1.44845366e+00 2.45745927e-01 5.70528626e-01 -3.05747122e-01 -5.12579978e-01 -3.97667259e-01 4.83164519e-01 4.25413489e-01 8.00862253e-01 -1.59426317e-01 -7.48000979e-01 5.92153430e-01 1.36511460e-01 -2.04903204e-02 -9.48782086e-01 -8.79784942e-01 -6.82901442e-02 2.42344484e-01 -4.21104908e-01 -5.73712349e-01 -4.14934546e-01 -1.11249709e+00 2.35909270e-03 -3.62217754e-01 5.45988917e-01 4.19717312e-01 1.27543819e+00 4.19419289e-01 6.68880463e-01 1.01574266e+00 -2.51371592e-01 -3.17367613e-01 -9.32806134e-01 -6.12405300e-01 6.12129390e-01 1.90244883e-01 -5.05809247e-01 -1.15441196e-01 -1.30250484e-01]
[11.476585388183594, 6.617456436157227]