Sota Nishiyama / 西山 颯大

Ph.D. student at UTokyo / neural networks / statistical physics

prof_pic.jpg

I am currently a first-year Ph.D. student at the Department of Basic Science, Graduate School of Arts and Sciences, The University of Tokyo, where I am advised by Prof. Masaaki Imaizumi. I also work as a Research Associate at the High-Dimensional Structure Theory Team, RIKEN Center for Advanced Intelligence Project (AIP).

I received my Master’s degree in Arts and Sciences from The University of Tokyo under the supervision by Prof. Masaaki Imaizumi. Prior to that, I received my Bachelor’s degree in Engineering from Tohoku University, where I was supervised by Prof. Kazuyuki Tanaka and Prof. Masayuki Ohzeki.

My research interests focus on the theory of machine learning and neural networks, particularly the learning dynamics of neural networks. I am also interested in statistical physics of disordered systems and applying its tools to study neural networks.

News

2026-04-01 I am starting as a PhD student at UTokyo and a Research Associate at RIKEN.
2026-02-10 New preprint: Dichotomy of Feature Learning and Unlearning: Fast-Slow Analysis on Neural Networks with Stochastic Gradient Descent.
2026-02-09 New preprint: High-Dimensional Limit of Stochastic Gradient Flow via Dynamical Mean-Field Theory.