TrainSimultaneously,generalizebetter:Stabilityofgradient-basedminimaxlearnersFarzanFarnia1AsumanOzdaglar1Abstract2014)andadversarialtraining(Madryetal.,2017)haveachievedgreatsuccessoverawidearrayof...
BeatingStochasticandAdversarialSemi-banditsOptimallyandSimultaneouslyJulianZimmert1HaipengLuo2Chen-YuWei2Abstracttrary,theminimaxoptimalregretisoforderO(√T)(Aueretal.,2002).Wedevelopthefirstgenera...