Convergence Behaviour of Some Gradient-Based Methods on Bilinear Zero-Sum Games

by   Guojun Zhang, et al.

Min-max formulations have attracted great attention in the ML community due to the rise of deep generative models and adversarial methods, and understanding the dynamics of (stochastic) gradient algorithms for solving such formulations has been a grand challenge. As a first step, we restrict ourselves to bilinear zero-sum games and give a systematic analysis of popular gradient updates, for both simultaneous and alternating versions. We provide exact conditions for their convergence and find the optimal parameter setup and convergence rates. In particular, our results offer formal evidence that alternating updates converge "better" than simultaneous ones.


Convergence Behaviour of Some Gradient-Based Methods on Bilinear Games

Min-max optimization has attracted much attention in the machine learnin...

Complex Momentum for Learning in Games

We generalize gradient descent with momentum for learning in differentia...

Stochastic Hamiltonian Gradient Methods for Smooth Games

The success of adversarial formulations in machine learning has brought ...

Stochastic Primal-Dual Algorithms with Faster Convergence than O(1/√(T)) for Problems without Bilinear Structure

Previous studies on stochastic primal-dual algorithms for solving min-ma...

Alternating Mirror Descent for Constrained Min-Max Games

In this paper we study two-player bilinear zero-sum games with constrain...

Average-case Acceleration for Bilinear Games and Normal Matrices

Advances in generative modeling and adversarial learning have given rise...

Discretization Drift in Two-Player Games

Gradient-based methods for two-player games produce rich dynamics that c...