English

Competitive Gradient Optimization

Optimization and Control 2022-05-31 v1 Machine Learning

Abstract

We study the problem of convergence to a stationary point in zero-sum games. We propose competitive gradient optimization (CGO ), a gradient-based method that incorporates the interactions between the two players in zero-sum games for optimization updates. We provide continuous-time analysis of CGO and its convergence properties while showing that in the continuous limit, CGO predecessors degenerate to their gradient descent ascent (GDA) variants. We provide a rate of convergence to stationary points and further propose a generalized class of α\alpha-coherent function for which we provide convergence analysis. We show that for strictly α\alpha-coherent functions, our algorithm convergences to a saddle point. Moreover, we propose optimistic CGO (OCGO), an optimistic variant, for which we show convergence rate to saddle points in α\alpha-coherent class of functions.

Keywords

Cite

@article{arxiv.2205.14232,
  title  = {Competitive Gradient Optimization},
  author = {Abhijeet Vyas and Kamyar Azizzadenesheli},
  journal= {arXiv preprint arXiv:2205.14232},
  year   = {2022}
}
R2 v1 2026-06-24T11:31:28.613Z