TowardsCertifying∞RobustnessusingNeuralNetworkswith∞-distNeuronsBohangZhang1TianleCai23ZhouLu4DiHe5LiweiWang16Abstract1.IntroductionItiswell-knownthatstandardneuralnetworks,Modernneuralnetworksar...
OnCertifyingNon-uniformBoundsagainstAdversarialAttacksChenLiu1RyotaTomioka2VolkanCevher1AbstractCWattack(Carlini&Wagner,2017).MostattackmethodssearchfortheadversarialexamplebyutilizingthegradientTh...