Stochastic Gradient Descent-like relaxation is equivalent to Metropolis dynamics in discrete optimization and inference problems

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Angelini, Maria Chiara, Cavaliere, Angelo Giorgio, Marino, Raffaele, Ricci-Tersenghi, Federico
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913369617858560
author Angelini, Maria Chiara
Cavaliere, Angelo Giorgio
Marino, Raffaele
Ricci-Tersenghi, Federico
author_facet Angelini, Maria Chiara
Cavaliere, Angelo Giorgio
Marino, Raffaele
Ricci-Tersenghi, Federico
contents Is Stochastic Gradient Descent (SGD) substantially different from Metropolis Monte Carlo dynamics? This is a fundamental question at the time of understanding the most used training algorithm in the field of Machine Learning, but it received no answer until now. Here we show that in discrete optimization and inference problems, the dynamics of an SGD-like algorithm resemble very closely that of Metropolis Monte Carlo with a properly chosen temperature, which depends on the mini-batch size. This quantitative matching holds both at equilibrium and in the out-of-equilibrium regime, despite the two algorithms having fundamental differences (e.g.\ SGD does not satisfy detailed balance). Such equivalence allows us to use results about performances and limits of Monte Carlo algorithms to optimize the mini-batch size in the SGD-like algorithm and make it efficient at recovering the signal in hard inference problems.
format Preprint
id arxiv_https___arxiv_org_abs_2309_05337
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Stochastic Gradient Descent-like relaxation is equivalent to Metropolis dynamics in discrete optimization and inference problems
Angelini, Maria Chiara
Cavaliere, Angelo Giorgio
Marino, Raffaele
Ricci-Tersenghi, Federico
Disordered Systems and Neural Networks
Statistical Mechanics
Machine Learning
Is Stochastic Gradient Descent (SGD) substantially different from Metropolis Monte Carlo dynamics? This is a fundamental question at the time of understanding the most used training algorithm in the field of Machine Learning, but it received no answer until now. Here we show that in discrete optimization and inference problems, the dynamics of an SGD-like algorithm resemble very closely that of Metropolis Monte Carlo with a properly chosen temperature, which depends on the mini-batch size. This quantitative matching holds both at equilibrium and in the out-of-equilibrium regime, despite the two algorithms having fundamental differences (e.g.\ SGD does not satisfy detailed balance). Such equivalence allows us to use results about performances and limits of Monte Carlo algorithms to optimize the mini-batch size in the SGD-like algorithm and make it efficient at recovering the signal in hard inference problems.
title Stochastic Gradient Descent-like relaxation is equivalent to Metropolis dynamics in discrete optimization and inference problems
topic Disordered Systems and Neural Networks
Statistical Mechanics
Machine Learning
url https://arxiv.org/abs/2309.05337