Abstract:The optimization problems associated with training generative adversarial neural networks can be largely reduced to certain {\em non-monotone} variational inequality problems (VIPs), whereas existing convergence results are mostly based on monotone or strongly monotone assumptions. In this paper, we propose {\em optimistic dual extrapolation (OptDE)}, a method that only performs {\em one} gradient evaluation per iteration. We show that OptDE is provably convergent to {\em a strong solution} under different coherent non-monotone assumptions. In particular, when a {\em weak solution} exists, the convergence rate of our method is $O(1/{\epsilon^{2}})$, which matches the best existing result of the methods with two gradient evaluations. Further, when a {\em $\sigma$-weak solution} exists, the convergence guarantee is improved to the linear rate $O(\log\frac{1}{\epsilon})$. Along the way--as a byproduct of our inquiries into non-monotone variational inequalities--we provide the near-optimal $O\big(\frac{1}{\epsilon}\log \frac{1}{\epsilon}\big)$ convergence guarantee in terms of restricted strong merit function for monotone variational inequalities. We also show how our results can be naturally generalized to the stochastic setting, and obtain corresponding new convergence results. Taken together, our results contribute to the broad landscape of variational inequality--both non-monotone and monotone alike--by providing a novel and more practical algorithm with the state-of-the-art convergence guarantees.

An extended descent framework for monotone variational inequities.

An extended descent framework for variational inequalities

A Trust Region-Type Method for Solving Monotone Variational Inequality

An Extended Alternating Direction Method for Variational Inequality Problems with Linear Equality and Inequality Constraints

Beyond Monotone Variational Inequalities: Solution Methods and Iteration Complexities

Modified Projection Method For General Variational Inequalities

Adaptive Methods or Variational Inequalities with Relatively Smooth and Reletively Strongly Monotone Operators

Convergence Properties of Feasible Descent Methods for Solving Variational Inequalities in Banach Spaces

Modified Goldstein–Levitin–Polyak Projection Method for Asymmetric Strongly Monotone Variational Inequalities

A new class of projection and contraction methods for solving variational inequality problems

On Two Iterative Methods for Mixed Monotone Variational Inequalities

Revisiting Subgradient Extragradient Methods for Solving Variational Inequalities

On Approximating Solutions to Non-monotone Variational Inequality Problems: An Approach Through the Modified Projection and Contraction Method

Projection and contraction method with double inertial steps for quasi-monotone variational inequalities

Some Adaptive First-order Methods for Variational Inequalities with Relatively Strongly Monotone Operators and Generalized Smoothness

A modified subgradient extragradient method with non-monotonic step sizes for solving quasimonotone variational inequalities

Optimistic Dual Extrapolation for Coherent Non-monotone Variational Inequalities

A NEW PARALLEL SPLITTING DESCENT METHOD FOR STRUCTURED VARIATIONAL INEQUALITIES

Weak Sharp Solutions of Variational Inequalities

Generalized Invex Monotonicity and Its Role in Solving Variational-Like Inequalities

Variational Theory and Algorithms for a Class of Asymptotically Approachable Nonconvex Problems