Solving Zero-Sum Games with Fewer Matrix-Vector Products
In this paper we consider the problem of computing an -approximate Nash Equilibrium of a zero-sum game in a payoff matrix with -bounded entries given access to a matrix-vector product oracle for and its transpose . We provide a deterministic algorithm that solves the problem using -oracle queries, where hides factors polylogarithmic in , , and . Our result improves upon the state-of-the-art query complexity of established by [Nemirovski, 2004] and [Nesterov, 2005]. We obtain this result through a general framework that yields improved deterministic query complexities for solving a broader class of minimax optimization problems which includes computing a linear classifier (hard-margin support vector machine) as well as linear regression.
View on arXiv