Stochastic Hamiltonian Gradient Methods for Smooth Games
By Nicolas Loizou, Hugo Berard, Alexia Jolicoeur-Martineau, Pascal Vincent, Simon Lacoste-Julien, Ioannis Mitliagkas. 2020
55 citations recorded by the source corpus.
The success of adversarial formulations in machine learning has brought renewed motivation for smooth games. In this work, we focus on the class of stochastic Hamiltonian methods and provide the first convergence guarantees for certain classes of stochastic smooth games. We propose a novel unbiased estimator for the stochastic Hamiltonian gradient descent (SHGD) and highlight its benefits. Using tools from the optimization literature we show that SHGD converges linearly to the neighbourhood of a stationary point. To guarantee convergence to the exact solution, we analyze SHGD with a decreasing step-size and we also present the first stochastic variance reduced Hamiltonian method. Our results provide the first global non-asymptotic last-iterate convergence guarantees for the class of stochastic unconstrained bilinear games and for the more general class of stochastic games that satisfy a "sufficiently bilinear" condition, notably including some non-convex non-concave problems. We supplement our analysis with experiments on stochastic bilinear and sufficiently bilinear games, where our theory is shown to be tight, and on simple adversarial machine learning formulations.
Subfields: Artificial Intelligence, Statistical and Nonlinear Physics, Statistics and Probability, Reinforcement Learning
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
In Statistical and Nonlinear Physics
In Statistics and Probability
In Reinforcement Learning
Citations in the corpus
- Hybrid Variance-Reduced SGD Algorithms For Minimax Problems with Nonconvex-Linear Function (2020, unknown): “, [3, 14, 23, 36] and nonconvex-concave settings [17, 18, 20, 28, 32].”
- Explore Aggressively, Update Conservatively: Stochastic Extragradient Methods with Variable Stepsize Scaling (2020, background): “[17] which appeared on arxiv a few weeks after the submission of our manuscript: it proved that stochastic Hamiltonian methods applied to (sufficiently) bilinear games ensures also a O(1/t) convergence rate.”