Error estimates between SGD with momentum and underdamped Langevin diffusion

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Guillin, Arnaud, Wang, Yu, Xu, Lihu, Yang, Haoran
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866916449001406464
author Guillin, Arnaud
Wang, Yu
Xu, Lihu
Yang, Haoran
author_facet Guillin, Arnaud
Wang, Yu
Xu, Lihu
Yang, Haoran
contents Stochastic gradient descent with momentum is a popular variant of stochastic gradient descent, which has recently been reported to have a close relationship with the underdamped Langevin diffusion. In this paper, we establish a quantitative error estimate between them in the 1-Wasserstein and total variation distances.
format Preprint
id arxiv_https___arxiv_org_abs_2410_17297
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Error estimates between SGD with momentum and underdamped Langevin diffusion
Guillin, Arnaud
Wang, Yu
Xu, Lihu
Yang, Haoran
Machine Learning
Probability
Stochastic gradient descent with momentum is a popular variant of stochastic gradient descent, which has recently been reported to have a close relationship with the underdamped Langevin diffusion. In this paper, we establish a quantitative error estimate between them in the 1-Wasserstein and total variation distances.
title Error estimates between SGD with momentum and underdamped Langevin diffusion
topic Machine Learning
Probability
url https://arxiv.org/abs/2410.17297