RobustLearningforDataPoisoningAttacksYunjuanWang1PooryaMianjy1RamanArora1Abstractinsettingswhereanadversarycanaffectanypartofthetrainingdata.Therefore,inthispaper,weareinterestedinWeinvestigatether...
Model-TargetedPoisoningAttackswithProvableConvergenceFnuSuya1SaeedMahloujifar2AnshumanSuri1DavidEvans1YuanTian1AbstractMostworkonPoisoningattackshasconsideredoneoftwoextremalattackerobjectives:indi...
JustHowToxicisDataPoisoning?AUnifiedBenchmarkforBackdoorandDataPoisoningAttacksAviSchwarzschild1MicahGoldblum2ArjunGupta3JohnP.Dickerson2TomGoldstein2AbstractAtthisscale,itisofteninfeasibletoproper...
PolicyTeachingviaEnvironmentPoisoning:Training-timeAdversarialAttacksagainstReinforcementLearningAminRakhsha1GoranRadanovic1RatiDevidze1XiaojinZhu2AdishSingla1Abstractcisions,Poisoningattacksmanipu...
TransferableClean-LabelPoisoningAttacksonDeepNeuralNetsChenZhu1W.RonnyHuang1AliShafahi1HengduoLi1GavinTaylor2ChristophStuder3TomGoldstein1Abstract2017).Incontrasttoevasionattacks(Biggioetal.,2013;S...
DataPoisoningAttacksonStochasticBanditsFangLiu1NessShroff12Abstractismotivatedbymodernindustrialscaleapplicationsofma-chinelearningsystems,wheredatacollectionandpolicyStochasticmulti-armedbanditsfo...
UniversalMulti-PartyPoisoningAttacksSaeedMahloujifar1MohammadMahmoody1AmeerMohammed2Abstractsenagainstthelearner(Valiant,1985;Kearns&Li,1993;Bshoutyetal.,2002).AtightlyrelatedandmorerecentInthiswor...