Non-Nesterov Acceleration Methods and Their Computer-Assisted Discovery Via the Performance Estimation Problem

9/29/22 | 4:15pm | E51-145





Ernest Ryu

Assistant Professor
Seoul National University

Abstract: Since the pioneering work of Nesterov on accelerated gradient methods, finding efficient and optimal first-order methods has been the focus in the study of large-scale optimization. Recently, renewed vitality was injected into this classical line of research by the emergence of computer-assisted proof methodologies. In this talk, we first present recent advances in accelerated first-order optimization algorithms, focusing on the new non-Nesterov's acceleration mechanisms. We then present the performance estimation problem (PEP) methodology, the essential computer-assisted tool for discovering the new acceleration mechanisms. Finally, we conclude by presenting the Branch-and-Bound Performance Estimation Programming (BnB-PEP), which significantly advances the computer-assisted proof methodology.

Bio: Ernest Ryu is an assistant professor in the Department of Mathematical Sciences at Seoul National University. He is an affiliated faculty of the Graduate School of Artificial Intelligence and the Graduate School of Data Science. His current research focus is on optimization and deep learning theory. Professor Ryu received a BS degree in Physics and Electrical engineering with honor at the California Institute of Technology in 2010 and an MS in Statistics and PhD in Computational and Mathematical Engineering with the Gene Golub Best Thesis Award at Stanford University in 2016. In 2016, he joined the Department of Mathematics at the University of California, Los Angeles as an Assistant Adjunct Professor. In 2020 he joined the Department of Mathematical Sciences at Seoul National University as a tenure-track faculty.

Event Time: 

2022 - 16:15