LowerboundsforSmoothNonconvexFinite-SumOptimizationDongruoZhou1QuanquanGu1Abstractoffirst-orderalgorithmsforsolving(1.1),whichhaveac-cessestotheIncrementalFirst-orderOracle(IFO)(AgarwalSmoothfinite...
TighterVariationalboundsareNotNecessarilyBetterTomRainforth1AdamR.Kosiorek12TuanAnhLe2ChrisJ.Maddison1MaximilianIgl2FrankWood3YeeWhyeTeh1Abstract&Kamp,1988;Hinton&Zemel,1994;Gregoretal.,2016;Chenet...
TightRegretboundsforBayesianOptimizationinOneDimensionJonathanScarlett1Abstract2010),whoconsiderthecumulativeregret:WeconsidertheproblemofBayesianoptimiza-Ttion(BO)inonedimension,underaGaussianproc...
StrongerGeneralizationboundsforDeepNetsviaaCompressionApproachSanjeevArora1RongGe2BehnamNeyshabur3YiZhang1AbstractfueledresearchinthisareabyshowingexperimentallythatstandardarchitecturesusingSGDand...
ProblemDependentReinforcementLearningboundsWhichCanIdentifyBanditStructureinMDPsAndreaZanette1EmmaBrunskill1Abstract(MDPs)andpartiallyobservableMDPs(POMDPs).Ban-ditsassumethattheactionstakendonotim...
PartialOptimalityandFastLowerboundsforWeightedCorrelationClusteringJan-HendrikLange12AndreasKarrenbauer1BjoernAndres134Abstractsolution.Weightedcorrelationclusteringingeneralgraphsishardtosolveexac...
ImprovedRegretboundsforThompsonSamplinginLinearQuadraticControlProblemsMarcAbeille1AlessandroLazaric2Abstracthasbeenmostlyaddressedfollowingtwomainapproaches:optimism-in-face-of-uncertainty(OFU)and...
DropoutTraining,Data-dependentRegularization,andGeneralizationboundsWenlongMou1YuchenZhou2JunGao3LiweiWang34Abstractexperimentalstudieshaveshowntheregularizationeffectincurredbydropout,asthegapbetw...
boundsontheApproximationPowerofFeedforwardNeuralNetworksMohammadMehrabi1AslanTchamkerten2MansoorI.Yousefi2AbstractDeterminingthecapacityofaneuralnetworkswithapiece-wiselinearactivationfunctiontypic...
UniformDeviationboundsfork-MeansClusteringOlivierBachem1MarioLucic1S.HamedHassani1AndreasKrause1AbstractRelatedwork.TraditionalVapnik-Chervonenkistheoryprovidestoolstoobtainuniformdeviationboundsfo...
TightboundsforApproximateCarathéodoryandBeyondVahabMirrokni1RenatoPaesLeme1AdrianVladu2SamChiu-waiWong3Abstractversionstatesthatifoneiswillingtotolerateanerrorof✏in`norm,OD2p/✏2verticessufficeto...
RiskboundsforTransferringRepresentationsWithandWithoutFine-TuningDanielMcNamara1Maria-FlorinaBalcan2Abstractneuralnetworks.1Underthis‘representation-as-a-service’model,ausermayexpecttoaccessthere...
MinimaxRegretboundsforReinforcementLearningMohammadGheshlaghiAzar1IanOsband1RémiMunos1AbstractThemostcommonapproachtothislearningproblemistoseparatetheprocessofestimationandoptimization.Weconsider...
Estimatingindividualtreatmenteffect:generalizationboundsandalgorithmsUriShalit1FredrikD.Johansson2DavidSontag23Abstractdatawhichcontainspastactions,theiroutcomes,andpos-siblymorecontext,butwithoutd...