Abstract:Machine learning approaches relying on such criteria as adversarial robustness or multi-agent settings have raised the need for solving game-theoretic equilibrium problems. Of particular relevance to these applications are methods targeting finite-sum structure, which generically arises in empirical variants of learning problems in these contexts. Further, methods with computable approximation errors are highly desirable, as they provide verifiable exit criteria. Motivated by these applications, we study finite-sum monotone inclusion problems, which model broad classes of equilibrium problems. Our main contributions are variants of the classical Halpern iteration that employ variance reduction to obtain improved complexity guarantees in which $n$ component operators in the finite sum are ``on average'' either cocoercive or Lipschitz continuous and monotone, with parameter $L$. The resulting oracle complexity of our methods, which provide guarantees for the last iterate and for a (computable) operator norm residual, is $\widetilde{\mathcal{O}}( n + \sqrt{n}L\varepsilon^{-1})$, which improves upon existing methods by a factor up to $\sqrt{n}$. This constitutes the first variance reduction-type result for general finite-sum monotone inclusions and for more specific problems such as convex-concave optimization when operator norm residual is the optimality measure. We further argue that, up to poly-logarithmic factors, this complexity is unimprovable in the monotone Lipschitz setting; i.e., the provided result is near-optimal.

Stochastic Proximal Point Methods for Monotone Inclusions under Expected Similarity

A strongly convergent inertial inexact proximal-point algorithm for monotone inclusions with applications to variational inequalities

Stochastic Monotone Inclusion with Closed Loop Distributions

A Unified Theory of Stochastic Proximal Point Methods without Smoothness

On the Bregman-proximal iterative algorithm for the monotone inclusion problem in Banach spaces

On the Bredies-Chenchene-Lorenz-Naldi algorithm

Solution of Mismatched Monotone+Lipschitz Inclusion Problems

Stochastic proximal gradient methods for nonconvex problems in Hilbert spaces

Convergence rate of inexact proximal point algorithms for operator with holder metric subregularity

Monotone vector fields and the proximal point algorithm on Hadamard manifolds

Rosenthal’s Inequalities for Independent and Negatively Dependent Random Variables under Sub-Linear Expectations with Applications

Variance Reduced Halpern Iteration for Finite-Sum Monotone Inclusions

Stochastic Variance-Reduced Majorization-Minimization Algorithms

Tight Sublinear Convergence Rate of the Proximal Point Algorithm for Maximal Monotone Inclusion Problems

Iterative methods for solving monotone variational inclusions without prior knowledge of the Lipschitz constant of the single-valued operator

New nonasymptotic convergence rates of stochastic proximal pointalgorithm for convex optimization problems

A Semismooth Newton Stochastic Proximal Point Algorithm with Variance Reduction

Stochastic Approximation for Expectation Objective and Expectation Inequality-Constrained Nonconvex Optimization

Stochastic forward-backward-half forward splitting algorithm with variance reduction

Constant Step Stochastic Approximations Involving Differential Inclusions: Stability, Long-Run Convergence and Applications