TITLE: Data-driven robust optimization and regularization in statistical learning

ABSTRACT:

 A central question in statistical learning is to design models and algorithms that not only perform well on training data, but also generalize to new and unseen data. In this talk, we tackle this question by formulating a data-driven distributionally robust optimization (DRO) problem, which seeks a solution that minimizes the worst-case expected loss over a family of distributions that are close to the empirical distribution in Wasserstein distance. We derive a tractable reformulation of the DRO problem via strong duality, and characterize the concise structure of the worst-case distribution. Based on these results, we establish a close connection between DRO and regularization. 

BIO: Rui Gao is a fifth-year Ph.D. candidate in Operations Research in ISyE, working with Anton Kleywegt. His research interests lie in the intersection of data-driven decision-making under uncertainty and statistical learning. Specific application areas include deep learning, revenue management, and power systems design.