A Fisher-Rao gradient flow for entropic mean-field min-max games

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Lascu, Razvan-Andrei, Majka, Mateusz B., Szpruch, Łukasz
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929502857199616
author Lascu, Razvan-Andrei
Majka, Mateusz B.
Szpruch, Łukasz
author_facet Lascu, Razvan-Andrei
Majka, Mateusz B.
Szpruch, Łukasz
contents Gradient flows play a substantial role in addressing many machine learning problems. We examine the convergence in continuous-time of a \textit{Fisher-Rao} (Mean-Field Birth-Death) gradient flow in the context of solving convex-concave min-max games with entropy regularization. We propose appropriate Lyapunov functions to demonstrate convergence with explicit rates to the unique mixed Nash equilibrium.
format Preprint
id arxiv_https___arxiv_org_abs_2405_15834
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle A Fisher-Rao gradient flow for entropic mean-field min-max games
Lascu, Razvan-Andrei
Majka, Mateusz B.
Szpruch, Łukasz
Optimization and Control
Machine Learning
Probability
Gradient flows play a substantial role in addressing many machine learning problems. We examine the convergence in continuous-time of a \textit{Fisher-Rao} (Mean-Field Birth-Death) gradient flow in the context of solving convex-concave min-max games with entropy regularization. We propose appropriate Lyapunov functions to demonstrate convergence with explicit rates to the unique mixed Nash equilibrium.
title A Fisher-Rao gradient flow for entropic mean-field min-max games
topic Optimization and Control
Machine Learning
Probability
url https://arxiv.org/abs/2405.15834