Скачать или смотреть Research Seminar: "Privacy & Adversarial Robustness in Statistical Estimation" by Prof. Linjun Zhang

Fall 2020 SIP Seminar Series: November 18, 2020
[http://www.inspirelab.us/seminars/]

Speaker: Prof. Linjun Zhang
Title: The cost of privacy and adversarial robustness in statistical estimation

Abstract: Privacy-preserving data analysis is a rising challenge in contemporary statistics, as the privacy guarantees of statistical methods are often achieved at the expense of accuracy. In this talk, we investigate the tradeoff between statistical accuracy and privacy in generalized linear models, under both the classical low- dimensional and modern high-dimensional settings. A primary focus is to establish minimax optimality for statistical estimation with the (\epsilon, \delta)-differential privacy constraint. To this end, we find that classical lower bound arguments fail to yield sharp results, and new technical tools are called for. Another challenge in contemporary statistics is adversarial robustness. However, little research has been done to quantify how robust optimization changes the optimizers and the prediction losses comparing to standard training. In this talk, inspired by the influence function in robust statistics, we introduce the Adversarial Influence Function (AIF) as a tool to investigate the solution produced by robust optimization. The proposed AIF enjoys a closed-form and can be calculated efficiently.

Biography: Linjun Zhang is an Assistant Professor in the Department of Statistics at Rutgers University. He earned his Ph.D. from the Wharton School, University of Pennsylvania. His research focuses on high-dimensional statistics, adversarial robustness, and differential privacy.