2020 (modified: 04 Nov 2022)ICML 2020Readers: Everyone
Abstract:Adversarial robustness has become a fundamental requirement in modern machine learning applications. Yet, there has been surprisingly little statistical understanding so far. In this paper, we prov...