Sampling from the Mean-Field Stationary Distribution

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Kook, Yunbum, Zhang, Matthew S., Chewi, Sinho, Erdogdu, Murat A., Li, Mufan Bill
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866916313028362240
author Kook, Yunbum
Zhang, Matthew S.
Chewi, Sinho
Erdogdu, Murat A.
Li, Mufan Bill
author_facet Kook, Yunbum
Zhang, Matthew S.
Chewi, Sinho
Erdogdu, Murat A.
Li, Mufan Bill
contents We study the complexity of sampling from the stationary distribution of a mean-field SDE, or equivalently, the complexity of minimizing a functional over the space of probability measures which includes an interaction term. Our main insight is to decouple the two key aspects of this problem: (1) approximation of the mean-field SDE via a finite-particle system, via uniform-in-time propagation of chaos, and (2) sampling from the finite-particle stationary distribution, via standard log-concave samplers. Our approach is conceptually simpler and its flexibility allows for incorporating the state-of-the-art for both algorithms and theory. This leads to improved guarantees in numerous settings, including better guarantees for optimizing certain two-layer neural networks in the mean-field regime. A key technical contribution is to establish a new uniform-in-$N$ log-Sobolev inequality for the stationary distribution of the mean-field Langevin dynamics.
format Preprint
id arxiv_https___arxiv_org_abs_2402_07355
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Sampling from the Mean-Field Stationary Distribution
Kook, Yunbum
Zhang, Matthew S.
Chewi, Sinho
Erdogdu, Murat A.
Li, Mufan Bill
Statistics Theory
Machine Learning
We study the complexity of sampling from the stationary distribution of a mean-field SDE, or equivalently, the complexity of minimizing a functional over the space of probability measures which includes an interaction term. Our main insight is to decouple the two key aspects of this problem: (1) approximation of the mean-field SDE via a finite-particle system, via uniform-in-time propagation of chaos, and (2) sampling from the finite-particle stationary distribution, via standard log-concave samplers. Our approach is conceptually simpler and its flexibility allows for incorporating the state-of-the-art for both algorithms and theory. This leads to improved guarantees in numerous settings, including better guarantees for optimizing certain two-layer neural networks in the mean-field regime. A key technical contribution is to establish a new uniform-in-$N$ log-Sobolev inequality for the stationary distribution of the mean-field Langevin dynamics.
title Sampling from the Mean-Field Stationary Distribution
topic Statistics Theory
Machine Learning
url https://arxiv.org/abs/2402.07355