DataaugmentationforDeeplearningbasedacceleratedMRIreconstructionwithlimiteddataZalanFabian1ReinhardHeckel2MahdiSoltanolkotabi1Abstracttients.Duetothesubsampling,therearefewerequationsthanunknowns,a...
CooperativeExplorationforMulti-AgentDeepReinforcementLearningIou-JenLiu1UnnatJain1RaymondA.Yeh1AlexanderG.Schwing1Abstract(MADDPG)(Loweetal.,2017),andcounterfactualmulti-agentpolicygradients(COMA)(...
DouZero:MasteringDouDizhuwithSelf-PlayDeepReinforcementLearningDaochenZha1JingruXie2WenyeMa2ShengZhang3XiangruLian2XiaHu1JiLiu2Abstractexample,AlphaGo(Silveretal.,2016),AlphaZero(Sil-veretal.,2018)...
ConsensusControlforDecentralizedDeepLearningLingjingKong1TaoLin1AnastasiaKoloskova1MartinJaggi1SebastianU.Stich1Abstractlocalmini-batchgradientscomputedondifferentsubsetsofthedata,forthelatersynchr...
BayesianDeepLearningviaSubnetworkInferenceErikDaxberger12EricNalisnick3JamesUrquhartAllingham1JavierAntora´n1Jose´MiguelHerna´ndez-Lobato145Abstractpresentsaprincipledwaytocaptureuncertaintyviat...
BackdoorScanningforDeepNeuralNetworksthroughK-ArmOptimizationGuangyuShen1YingqiLiu1GuanhongTao1ShengweiAn1QiulingXu1SiyuanCheng1ShiqingMa2XiangyuZhang1Abstractdoor)attackexploitstheuninterpretabili...
Analyzingthetree-layerstructureofDeepForestsLudovicArnould1ClaireBoyer1ErwanScornet2AbstractDeepForest(DF)algorithm(Zhou&Feng,2017),whichusesRandomForests(RF)(Breiman,2001)asneurons,hasRandomforest...
AdversarialRobustnessGuaranteesforRandomDeepNeuralNetworksGiacomoDePalma123BobakT.Kiani34SethLloyd23Abstractanextremelysmallperturbationcanchangethelabelofacorrectlyclassifedimage.Thispropertyposes...
ActiveDeepProbabilisticSubsamplingHansvanGorp1IrisA.M.Huijben1BastiaanS.Veeling2NicolaPezzotti34RuudJ.G.vanSloun14AbstractFirstproposedbyDonoho(2006),andlaterappliedforMRIbyLustigetal.(2007),CSleve...
ADeepReinforcementLearningApproachtoMarginalizedImportanceSamplingwiththeSuccessorRepresentationScottFujimoto1DavidMeger1DoinaPrecup1Abstractapproachcanhavesignificantlylowervariancethantradi-tiona...
UniquePropertiesofFlatMinimainDeepNetworksRotemMulayoff1TomerMichaeli1Abstractmodels(Gunasekaretal.,2018b)andDeepnonlinearnet-workswithhomogeneousactivationfunctions(Lyu&Li,Itiswellknownthat(stocha...
TrainingDeepEnergy-BasedModelswithf-DivergenceMinimizationLantaoYu1YangSong1JiamingSong1StefanoErmon1Abstractbasedmodels(Dinhetal.,2014;2016;Kingma&Dhariwal,2018)andsum-productnetworks(Poon&Domingo...
TemporalPhenotypingusingDeepPredictiveClusteringofDiseaseProgressionChangheeLee1MihaelavanderSchaar231Abstractunderstandingsuchheterogeneousdiseases.Thisallowsclinicianstoanticipatepatients’progno...
StudentSpecializationinDeepRectifiedNetworksWithFiniteWidthandInputDimensionYuandongTian1Abstract1.IntroductionWeconsideraDeepReLU/LeakyReLUstu-WhileDeepLearninghasachievedgreatsuccessindiffer-dent...
SourceSeparationwithDeepGenerativePriorsVivekJayaramJohnThickstunAbstractWithnofurtherconstraintsorregularization,solvingEqua-tion(1)forxishighlyunderdetermined.Classical“blind”Despitesubstantial...
SimpleandDeepGraphConvolutionalNetworksMingChen1ZheweiWei234ZengfengHuang5BolinDing6YaliangLi6AbstractDespitetheirenormoussuccess,mostofthecurrentGCNmodelsareshallow.Mostoftherecentmodels,suchasGra...
SDE-Net:EquippingDeepNeuralNetworkswithUncertaintyEstimatesLingkaiKong1JimengSun2ChaoZhang1Abstractbilitiesforclassification(Guoetal.,2017).Moreover,theycanmakeerroneousyetwildlyconfidentprediction...
ScalableDeepGenerativeModelingforSparseGraphsHanjunDai1AzadeNazi1YujiaLi2BoDai1DaleSchuurmans1AbstractClassicalgenerativemodelsofgraphs,basedonrandomgraphtheory(Erdo˝s&Re´nyi,1960;Baraba´si&Albe...
SafeDeepSemi-SupervisedLearningforUnseen-ClassUnlabeledDataLan-ZheGuo1Zhen-YuZhang1YuanJiang1Yu-FengLi1Zhi-HuaZhou1AbstractFigure1.Oneexampleofclassdistributionmismatch.Unlabeleddatacontainsclasses...
RIFLE:BackpropagationinDepthforDeepTransferLearningthroughRe-InitializingtheFully-connectedLayErXingjianLi12HaoyiXiong1HaozheAn1ChengzhongXu23DejingDou1Abstract1.IntroductionFine-tuningtheDeepconvo...