Table of Contents: :: Library Catalog

Cover Image

Saved in:

Bibliographic Details
Main Authors:	Huang, Zhishen, Becker, Stephen
Format:	Preprint
Published:	2021
Subjects:	Machine Learning Optimization and Control
Online Access:	https://arxiv.org/abs/2102.06759
Tags:	Add Tag No Tags, Be the first to tag this record!

Table of Contents:

Stochastic gradient Langevin dynamics (SGLD) has gained the attention of optimization researchers due to its global optimization properties. This paper proves an improved convergence property to local minimizers of nonconvex objective functions using SGLD accelerated by variance reductions. Moreover, we prove an ergodicity property of the SGLD scheme, which gives insights on its potential to find global minimizers of nonconvex objectives.