DeepGenerativeLearningviaVariationalGradientFlowYuanGao1YulingJiao2YangWang3YaoWang4CanYang3ShunkangZhang3Abstract1.IntroductionWeproposeaframeworktolearndeepgenerativeLearningthegenerativemodel,i....
ConditionalGradientMethodsviaStochasticPath-IntegratedDifferentialEstimatorAlpYurtsever1SuvritSra2VolkanCevher1AbstractarelatedworkbyHazan&Luo(2016)),constraineddeeplearningproblems(e.g.,Ravietal.(...
CompressingGradientOptimizersviaCount-SketchesRyanSpring1AnastasiosKyrillidis1VijaiMohan2AnshumaliShrivastava12AbstractTraininglarge-scalemodelsefficientlyisachallengingtask.Therearenumerouspublica...
ARSM:Augment-REINFORCE-Swap-MergeEstimatorforGradientBackpropagationThroughCategoricalVariablesMingzhangYin1YuguangYue1MingyuanZhou2Abstractzk∈{1,2,...,C}asaunivariateC-waycategoricalvari-able,and...
AdaptiveStochasticNaturalGradientMethodforOne-ShotNeuralArchitectureSearchYouheiAkimoto1ShinichiShirakawa2NozomuYoshinari2KentoUchida2ShotaSaito23KouheiNishida4AbstractWorkpublishedbefore2017oftenf...
ATail-IndexAnalysisofStochasticGradientNoiseinDeepNeuralNetworksUmutS¸ims¸ekli1LeventSagun2MertGu¨rbu¨zbalaban3Abstractmanyapplicationdomains(LeCunetal.,2015;Krizhevskyetal.,2012;Hintonetal.,20...
ACompositeRandomizedIncrementalGradientMethodJunyuZhang1LinXiao2Abstractwhereeachfj:Rp→Rissmoothandcanbenonconvex.SuchproblemsoftenariseasfinitesampleapproximationsWeconsidertheproblemofminimizing...
Zeno:DistributedStochasticGradientDescentwithSuspicion-basedFault-toleranceCongXie1OluwasanmiKoyejo1IndranilGupta1Abstractvarianceandmagnitude,makingthemhardtodistinguish.Itisalsopossiblethatindiff...
ABaselineforAnyOrderGradientEstimationinStochasticComputationGraphsJingkaiMao1JakobFoerster2TimRockta¨schel3MaruanAl-Shedivat4GregoryFarquhar2ShimonWhiteson2Abstract1.IntroductionByenablingcorrect...
StochasticVariance-ReducedPolicyGradientMatteoPapini1DamianoBinaghi1GiuseppeCanonaco1MatteoPirotta2MarcelloRestelli1Abstractavaluefunction,ordirectlyapolicydefiningtheagent’sbehaviour.Furthermore,...
SteinVariationalGradientDescentWithoutGradientJunHan1QiangLiu2AbstractMCMCandVI.Byleveraginganewtypeoffunctionalgra-dientdescentofKLdivergenceonthespaceofdistributions,SteinvariationalGradientdecen...
SADAGRAD:StronglyAdaptiveStochasticGradientMethodsZaiyiChen12YiXu2EnhongChen1TianbaoYang2Abstractiterations.Ithasreceivedtremendousinterestsforsolvingbigdatalearningproblems(e.g.,see(Deanetal.,2012...
RiemannianStochasticRecursiveGradientAlgorithmHiroyukiKasai1HiroyukiSato2BamdevMishra3AbstractmannianGradientdescentmethod,whichcalculatestheR1ie∑mnannianfullGradientestimation,i.e.,gradf(w)=Stoch...
Projection-FreeOnlineOptimizationwithStochasticGradient:FromConvexitytoSubmodularityLinChen12ChristopherHarshaw13HamedHassani4AminKarbasi12Abstract1.IntroductionOnlineoptimizationhasbeenasuccessful...
PolicyOptimizationasWassersteinGradientFlowsRuiyiZhang1ChangyouChen2ChunyuanLi1LawrenceCarin1Abstractwiththeenvironment.Policyoptimizationisacorecomponentofrein-Astandardtechniqueforpolicylearningi...
OptimalDistributedLearningwithMulti-passStochasticGradientMethodsJunhongLin1VolkanCevher1AbstractTheclassicalalgorithmstoperformlearningtaskareregular-izedalgorithms,suchasKRR,kernelprincipalcompon...
NoisyNaturalGradientasVariationalInferenceGuodongZhang12ShengyangSun12DavidDuvenaud12RogerGrosse12Abstract&Welling,2017),butfittingsuchmodelscanbeexpensivewithoutfurtherapproximations.VariationalBa...
Non-convexConditionalGradientSlidingChaoQu1YanLi2HuanXu2AbstractBesidesthisgeneralform,wealsoconsiderastochasticWeinvestigateaprojectionfreeoptimizationsettingandafinite-sumsetting.Inthestochastics...
MessagePassingSteinVariationalGradientDescentJingweiZhuo1ChangLiu1JiaxinShi1JunZhu1NingChen1BoZhang1Abstractparametricfamiliesascommonlydoneintraditionalvari-ationalinference(VI)methods.Besides,SVG...
LearningtoExploreviaMeta-PolicyGradientTianbingXu1QiangLiu2LiangZhao1JianPeng3Abstractalgorithmtothecontinuousactionspaces,exploitspreviousexperienceoroff-policydatafromareplaybufferandoftenTheperf...