Symmetric Mean-field Langevin Dynamics for Distributional Minimax Problems

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Kim, Juno, Yamamoto, Kakei, Oko, Kazusato, Yang, Zhuoran, Suzuki, Taiji
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916127609716736
author Kim, Juno
Yamamoto, Kakei
Oko, Kazusato
Yang, Zhuoran
Suzuki, Taiji
author_facet Kim, Juno
Yamamoto, Kakei
Oko, Kazusato
Yang, Zhuoran
Suzuki, Taiji
contents In this paper, we extend mean-field Langevin dynamics to minimax optimization over probability distributions for the first time with symmetric and provably convergent updates. We propose mean-field Langevin averaged gradient (MFL-AG), a single-loop algorithm that implements gradient descent ascent in the distribution spaces with a novel weighted averaging, and establish average-iterate convergence to the mixed Nash equilibrium. We also study both time and particle discretization regimes and prove a new uniform-in-time propagation of chaos result which accounts for the dependency of the particle interactions on all previous distributions. Furthermore, we propose mean-field Langevin anchored best response (MFL-ABR), a symmetric double-loop algorithm based on best response dynamics with linear last-iterate convergence. Finally, we study applications to zero-sum Markov games and conduct simulations demonstrating long-term optimality.
format Preprint
id arxiv_https___arxiv_org_abs_2312_01127
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Symmetric Mean-field Langevin Dynamics for Distributional Minimax Problems
Kim, Juno
Yamamoto, Kakei
Oko, Kazusato
Yang, Zhuoran
Suzuki, Taiji
Optimization and Control
Machine Learning
In this paper, we extend mean-field Langevin dynamics to minimax optimization over probability distributions for the first time with symmetric and provably convergent updates. We propose mean-field Langevin averaged gradient (MFL-AG), a single-loop algorithm that implements gradient descent ascent in the distribution spaces with a novel weighted averaging, and establish average-iterate convergence to the mixed Nash equilibrium. We also study both time and particle discretization regimes and prove a new uniform-in-time propagation of chaos result which accounts for the dependency of the particle interactions on all previous distributions. Furthermore, we propose mean-field Langevin anchored best response (MFL-ABR), a symmetric double-loop algorithm based on best response dynamics with linear last-iterate convergence. Finally, we study applications to zero-sum Markov games and conduct simulations demonstrating long-term optimality.
title Symmetric Mean-field Langevin Dynamics for Distributional Minimax Problems
topic Optimization and Control
Machine Learning
url https://arxiv.org/abs/2312.01127