Matthew Brennan wins Best Student Paper Award at COLT2020

July 30, 2020

It is our great pleasure to share that Matthew Brennan, a member of LIDS and EECS student, won the Best Student Paper Award at the 2020 Conference on Learning Theory (COLT 2020), a leading conference on the theory of machine learning. His paper, entitled “Reducibility and Statistical-Computational Gaps from Secret Leakage,” was co-authored by his supervisor, EECS professor and member of LIDS, Guy Bresler.

This is the second time Matt won the COLT Best Student Paper Award. The first time was in 2018, for related work, with Guy and collaborator Wasim Huleihel.

“The paper relates the computational complexity of a variety of disparate high-dimensional statistics problems, by showing that in fact the problems themselves are at their core essentially the same basic problem. This approach is incredibly appealing but notoriously challenging. Matt really hit it out of the park with this paper,” says Guy. “Indeed, this paper goes far beyond what we imagined was even possible in 2018, in part by shifting perspective and introducing a new way of thinking about these problems. We are optimistic that this new approach will unlock continued further progress in the field.”

The Conference on Learning Theory (COLT) is organized by the Association for Computational Learning (ACL). To learn more, visit: