Speaker: Yi Feng (Aarhus University)
Title: Continuous-Time Analysis of Heavy Ball Momentum in Min-Max Games
Abstract: Since Polyak's pioneering work, heavy ball momentum has been widely studied in minimization problems. However, its role in min-max games remains largely unexplored. As a key component of practical min-max algorithms like Adam, this gap limits their effectiveness. In this talk, I will present a continuous-time analysis for heavy ball momentum in min-max games. I will focus on two aspects: convergence to equilibrium and implicit gradient regularization, which describe how momentum shapes the interaction between algorithmic trajectories and the geometry of the min-max loss landscape. We will see that the behavior of heavy ball momentum in min-max games is significantly different from that in minimization.
Bio: Yi Feng is a Postdoctoral researcher at Aarhus University. His research focuses on understanding the learning dynamics in games and their applications in machine learning. His research is supported by a Danish Data Science Academy Postdoctoral Fellowship.
This seminar is jointly organised in collaboration with Trung Tuyen Truong from the Several Complex Variables, Logic and Operator algebras group at UiO.
The seminar is in the interest of SURE-AI - Sustainable, Risk-Averse and Ethical AI.
This series of webinars addresses all interested people in probability, stochastic analysis, control, risk evaluation, statistics, with a view towards applications, in particular to finance, economy, energy markets and production, but also to physics and biology. This series brings together the major research themes of the group. STAR stands for STochastics And Risk! Welcome!