|| Robust Machine Learning via Privacy/Rate-Distortion Theory
||Ye Wang, Mitsubishi Electric Research Laboratories, United States; Shuchin Aeron, Tufts University, United States; Adnan Siraj Rakin, Arizona State University, United States; Toshiaki Koike-Akino, Mitsubishi Electric Research Laboratories, United States; Pierre Moulin, University of Illinois at Urbana-Champaign, United States|
||D3-S5-T3: Privacy & Learning
||Wednesday, 14 July, 23:20 - 23:40
||Wednesday, 14 July, 23:40 - 00:00
Robust machine learning formulations have emerged to address the prevalent vulnerability of deep neural networks to adversarial examples. Our work draws the connection between optimal robust learning and the privacy-utility tradeoff problem, which is a generalization of the rate-distortion problem. The saddle point of the game between a robust classifier and an adversarial perturbation can be found via the solution of a maximum conditional entropy problem. This information-theoretic perspective sheds light on the fundamental tradeoff between robustness and clean data performance, which ultimately arises from the geometric structure of the underlying data distribution and perturbation constraints.