English

A Contour Stochastic Gradient Langevin Dynamics Algorithm for Simulations of Multi-modal Distributions

Machine Learning 2022-05-24 v2 Machine Learning Computation

Abstract

We propose an adaptively weighted stochastic gradient Langevin dynamics algorithm (SGLD), so-called contour stochastic gradient Langevin dynamics (CSGLD), for Bayesian learning in big data statistics. The proposed algorithm is essentially a \emph{scalable dynamic importance sampler}, which automatically \emph{flattens} the target distribution such that the simulation for a multi-modal distribution can be greatly facilitated. Theoretically, we prove a stability condition and establish the asymptotic convergence of the self-adapting parameter to a {\it unique fixed-point}, regardless of the non-convexity of the original energy function; we also present an error analysis for the weighted averaging estimators. Empirically, the CSGLD algorithm is tested on multiple benchmark datasets including CIFAR10 and CIFAR100. The numerical results indicate its superiority to avoid the local trap problem in training deep neural networks.

Keywords

Cite

@article{arxiv.2010.09800,
  title  = {A Contour Stochastic Gradient Langevin Dynamics Algorithm for Simulations of Multi-modal Distributions},
  author = {Wei Deng and Guang Lin and Faming Liang},
  journal= {arXiv preprint arXiv:2010.09800},
  year   = {2022}
}

Comments

Accepted by NeurIPS 2020

R2 v1 2026-06-23T19:27:59.489Z