English

A Hessian-Aware Stochastic Differential Equation for Modelling SGD

Machine Learning 2025-06-04 v3 Machine Learning Optimization and Control

Abstract

Continuous-time approximation of Stochastic Gradient Descent (SGD) is a crucial tool to study its escaping behaviors from stationary points. However, existing stochastic differential equation (SDE) models fail to fully capture these behaviors, even for simple quadratic objectives. Built on a novel stochastic backward error analysis framework, we derive the Hessian-Aware Stochastic Modified Equation (HA-SME), an SDE that incorporates Hessian information of the objective function into both its drift and diffusion terms. Our analysis shows that HA-SME achieves the order-best approximation error guarantee among existing SDE models in the literature, while significantly reducing the dependence on the smoothness parameter of the objective. Empirical experiments on neural network-based loss functions further validate this improvement. Further, for quadratic objectives, under mild conditions, HA-SME is proved to be the first SDE model that recovers exactly the SGD dynamics in the distributional sense. Consequently, when the local landscape near a stationary point can be approximated by quadratics, HA-SME provides a more precise characterization of the local escaping behaviors of SGD. With the enhanced approximation guarantee, we further conduct an escape time analysis using HA-SME, showcasing how it can be employed to analytically study the escaping behavior of SGD for general function classes.

Keywords

Cite

@article{arxiv.2405.18373,
  title  = {A Hessian-Aware Stochastic Differential Equation for Modelling SGD},
  author = {Xiang Li and Zebang Shen and Liang Zhang and Niao He},
  journal= {arXiv preprint arXiv:2405.18373},
  year   = {2025}
}
R2 v1 2026-06-28T16:44:23.818Z