English

Solving Min-Max Optimization with Hidden Structure via Gradient Descent Ascent

Optimization and Control 2021-01-14 v1 Computer Science and Game Theory Machine Learning Machine Learning

Abstract

Many recent AI architectures are inspired by zero-sum games, however, the behavior of their dynamics is still not well understood. Inspired by this, we study standard gradient descent ascent (GDA) dynamics in a specific class of non-convex non-concave zero-sum games, that we call hidden zero-sum games. In this class, players control the inputs of smooth but possibly non-linear functions whose outputs are being applied as inputs to a convex-concave game. Unlike general zero-sum games, these games have a well-defined notion of solution; outcomes that implement the von-Neumann equilibrium of the "hidden" convex-concave game. We prove that if the hidden game is strictly convex-concave then vanilla GDA converges not merely to local Nash, but typically to the von-Neumann solution. If the game lacks strict convexity properties, GDA may fail to converge to any equilibrium, however, by applying standard regularization techniques we can prove convergence to a von-Neumann solution of a slightly perturbed zero-sum game. Our convergence guarantees are non-local, which as far as we know is a first-of-its-kind type of result in non-convex non-concave games. Finally, we discuss connections of our framework with generative adversarial networks.

Keywords

Cite

@article{arxiv.2101.05248,
  title  = {Solving Min-Max Optimization with Hidden Structure via Gradient Descent Ascent},
  author = {Lampros Flokas and Emmanouil-Vasileios Vlatakis-Gkaragkounis and Georgios Piliouras},
  journal= {arXiv preprint arXiv:2101.05248},
  year   = {2021}
}
R2 v1 2026-06-23T22:08:10.755Z