Learning Adversarially Robust Representations via Worst-Case Mutual Information Maximization

LearningAdversariallyRobustRepresentationsviaWorst-CaseMutualInformationMaximizationSichengZhu1XiaoZhang1DavidEvans1Abstractfailtoproducerobustmodels,evenforsimpleclassificationtasksonCIFAR-10(Kriz...