中俄数学研究生讨论班——Stochastic Dynamics Near Critical Points in Stochastic Gradient Descent
报告人：Dmitry Dudukalov (Sobolev Institute of Mathematics)
时间：2025-12-05 15:30-16:30
地点：智华楼四元厅
Abstract:
The talk will focus on limit theorems for stochastic gradient descent (SGD) and on how the choice of step size affects its convergence. We will begin by discussing the convergence of deterministic gradient descent with a constant step size, then review the classical Robbins–Monro result for SGD, and finally turn to several results from our joint work [1] concerning the convergence of SGD with a constant step size that tends to zero.
[1] D. Dudukalov, A. Logachov, V. Lotov, T. Prasolov, E. Prokopenko & A. Tarasenko (2025). Convergence, Sticking and Escape: Stochastic Dynamics Near Critical Points in SGD. arXiv preprint arXiv:2505.18535.
Speaker contact: [email protected]
Online link: //imsoran.ktalk.ru/ogsac0ijeetj
