[month] [year]

Prof. Taji Suzuki

December 2022

Prof. Taiji Suzuki, Associate Professor, Department of Mathematical Informatics at the University of Tokyo gave a talk on Convergence of Mean Field Gradient Langevin Dynamics for Optimizing two-layer neural networks on 13 December.

A brief summary of Prof. Taiji Suzuki’s  talk in his words: 

In this talk, we discuss optimization procedures of two-layer neural networks via the gradient Langevin dynamics in a mean field regime. For that purpose, we first introduce a theoretical guarantee of a linear convergence of the mean field gradient Langevin algorithm in the infinite width limit under a uniform log-Sobolev inequality condition. Next, we propose some specific optimization methods for a finite width and discrete time setting. The constructions of those methods are based on the convex optimization techniques for finite dimensional objectives. Finally, we discuss the linear convergence of a vanilla gradient Langevin dynamics without an infinite width limit under a bit stronger condition than the uniform log-Sobolev inequality.

Prof. Taiji Suzuki is currently an Associate Professor in the Department of Mathematical Informatics at the University of Tokyo. He also serves as the team leader of the Deep learning theory team in AIP-RIKEN.

Prof. Taiji Suzuki received his Ph.D in information science and technology from the University of Tokyo in 2009. He has a broad research interest in statistical learning theory on deep learning, kernel methods and sparse estimation, and stochastic optimization for large-scale machine learning problems. He served as area chairs of premier conferences such as NeurIPS, ICML, ICLR, AISTATS and a program chair of ACML.

He received the Outstanding Paper Award at ICLR in 2021, the MEXT Young Scientists’ Prize, Outstanding Achievement Award in 2017 from the Japan Statistical Society, Outstanding Achievement Award in 2016 from the Japan Society for Industrial and Applied Mathematics, and Best Paper Award in 2012 from IBISML.