Yingjie Bi Ph.D. '20 wins student paper prize

Yingjie Bi Ph.D. '20 is the winner of the 2020 INFORMS Optimization Society Student Paper Prize. The paper, titled "Duality gap estimation via a refined Shapley-Folkman lemma," addresses an important question in optimization: estimating the duality gap for optimization problems with a nonconvex, separable objective.

The paper was co-authored by Bi’s Cornell advisor, A. Kevin Tang, professor of electrical and computer engineering, and formed the first part of Bi's Ph.D. dissertation. It introduces new concepts, such as the k-th convex hull, which considerably refine the Shapley-Folkman lemma. 

"This paper is about estimating the duality gap for nonconvex optimization problems," Bi said, "which could be valuable for the analysis of many approximation algorithms solving important nonconvex problems that occur in fields such as engineering and machine learning."

Bi, who is now a postdoctoral scholar at the University of California, Berkeley, reflected on his work with Professor Tang. "Working with Kevin, and the Ph.D. education at Cornell ECE, provided me with opportunities in a wide range of areas, including theoretical ones such as control and optimization as well as experimental ones such as cloud computing and computer networking."

The INFORMS judging committee was impressed by the potential the paper represents. They stated, "We expect this work to endure and to find exciting applications in domains including operations research, machine learning, and engineering where nonconvexities are prevalent."