UnitaryBranchingPrograms:LearnabilityandLowerboundsFidelErnestoD´ıazAndino1MariaKokkou2MateusdeOliveiraOliveira3FarhadVadiee3AbstractdepthBooleancircuitscanbecomputedbywidth-5branch-ingprogramsof...
TowardsTightboundsontheSampleComplexityofAverage-rewardMDPsYujiaJin1AaronSidford1Abstractmakingunderuncertaintyandreinforcementlearning(Puter-man,2014;Sutton&Barto,2018).Itisaprominenttheoret-Wepro...
TowardsPracticalMeanboundsforSmallSamplesMyPhan1PhilipS.Thomas1ErikLearned-Miller1Abstractrefertotwoseparatesettings,theone-endedsupportsetting,inwhichthedistributionisknowntofallintheintervalHisto...
TowardBetterGeneralizationboundswithLocallyElasticStabilityZhunDeng1HangfengHe2WeijieJ.Su3Abstractvarietyofapproachesfromstatisticallearningtheory(Vap-nik,1979;2013;Bartlett&Mendelson,2002;Bousquet...
TighterboundsontheLogMarginalLikelihoodofGaussianProcessRegressionusingConjugateGradientsArtemArtemev12DavidR.Burt3MarkvanderWilk1Abstractdientbasedmethodsinordertoautomaticallyselectmodelhyperpara...
TightboundsontheSmallestEigenvalueoftheNeuralTangentKernelforDeepReLUNetworksQuynhNguyen1MarcoMondelli2GuidoMontufar13AbstractWeassumethatthenetworkhasasingleoutput,namelynL=1andWL∈RnL−1.Forconsi...
SharperGeneralizationboundsforClusteringShaojieLi12YongLiu12Abstractclusteringalgorithmdesignandapplication,thestatisticaltheoryofclusteringmayappeartobenotsufficient.TheExistinggeneralizationanaly...
RiskboundsandRademacherComplexityinBatchReinforcementLearningYaqiDuan1ChiJin2ZhiyuanLi3Abstractalgorithmsincludingsupportvectormachines(Cortes&Vapnik,1995;Suykens&Vandewalle,1999),boosting(Fre-This...
RelativeDeviationMarginboundsCorinnaCortes1MehryarMohri12AnandaTheerthaSuresh1AbstracttheyareusedwithpositivedefinitekernelssuchasGaus-siankernels,forwhichthedimensionofthefeaturespaceWepresentaser...
OnLowerboundsforStandardandRobustGaussianProcessBanditOptimizationXuCai1JonathanScarlett12Abstractgunovicetal.,2016;Wang&Jegelka,2017;Janzetal.,2020),andalgorithm-independentlowerboundshavebeenInth...
Non-ExponentiallyWeightedAggregation:RegretboundsforUnboundedLossFunctionsPierreAlquier1Abstractthesub-g√radientoftcanbeused.SuchstrategiesleadtoregretinTundertheadditionalassumptionthatthetareWet...
LowerboundsonCross-EntropyLossinthePresenceofTest-timeAdversariesArjunNitinBhagoji1DanielCullina2VikashSehwag3PrateekMittal3Abstractonestablishingfundamentalboundsonlearninginthepres-enceoftest-tim...
LearningboundsforOpen-SetLearningZhenFang1JieLu1AnjinLiu1FengLiu1GuangquanZhang1Abstractetal.,2018;Yangetal.,2020).TraditionalsupervisedlearningaimstotrainaHowever,theclosed-setassumptionisnotreali...
InterpretableStabilityboundsforSpectralGraphFiltersHenryKenlay1DorinaThanou2XiaowenDong1Abstractaubiquitoustooldesignedtoprocessgraph-structureddata.Inadditiontovarioussignalprocessingtasks(Shumane...
ImprovedRegretboundsofBilinearBanditsusingActionSpaceAnalysisKyoungseokJang1Kwang-SungJun2Se-YoungYun3WanmoKang1Abstractarrangecouplesbasedontheirexperiencestogetbetterrat-ingsandrewards.Balancinge...
ImprovedConfidenceboundsfortheLinearLogisticModelandApplicationstoBanditsKwang-SungJun1LalitJain2BlakeMason3HoussamNassif4Abstractafeaturevectorforeacharm.Acommonassumptionisthattherewardisanoisyli...
GeneralizationboundsinthePresenceofOutliers:aMedian-of-MeansStudyPierreLaforgue1GuillaumeStaerman2StephanClémençon2Abstractcorruptionoftheobservationsacquired.Fromastatisticalperspective,twoframe...
ExponentialLowerboundsforBatchReinforcementLearning:BatchRLcanbeExponentiallyHarderthanOnlineRLAndreaZanette1AbstractweconsidertwoclassicalbatchRLproblems:1)theoff-policyevaluation(OPE)problem,wher...
EfficientPerformanceboundsforPrimal-DualReinforcementLearningfromDemonstrationsAngelikiKamoutsi1GoranBanjac1JohnLygeros1AbstractInthestandardRLsettingacostsignalisgiventoinstructagentshowtocomplete...
UpperboundsforModel-FreeRow-SparsePrincipalComponentAnalysisGuanyiWang1SantanuDey1AbstractwhereA:=1XX⊤isthesamplecovariancematrix,MSparseprincipalcomponentanalysis(PCA)isandIrdenotesther×ridentit...
