Affiliation:
1. University of Wisconsin-Madison, Madison, Wisconsin, USA
2. Cornell University, Ithaca, New York, USA
Abstract
Motivated by Q-learning, we study nonsmooth contractive stochastic approximation (SA) with constant stepsize. We focus on two important classes of dynamics: 1) nonsmooth contractive SA with additive noise, and 2) synchronous and asynchronous Q-learning, which features both additive and multiplicative noise. For both dynamics, we establish weak convergence of the iterates to a stationary limit distribution in Wasserstein distance. Furthermore, we propose a prelimit coupling technique for establishing steady-state convergence and characterize the limit of the stationary distribution as the stepsize goes to zero. Using this result, we derive that the asymptotic bias of nonsmooth SA is proportional to the square root of the stepsize, which stands in sharp contrast to smooth SA. This bias characterization allows for the use of Richardson-Romberg extrapolation for bias reduction in nonsmooth SA.
Funder
National Science Foundation
Publisher
Association for Computing Machinery (ACM)
Reference11 articles.
1. Bridging the gap between constant step size stochastic gradient descent and Markov chains
2. Alain Durmus, Pablo Jiménez, Eric Moulines, and Salem Said. 2021. On Riemannian Stochastic Approximation Schemes with Fixed Step-Size. In Proceedings of The 24th International Conference on Artificial Intelligence and Statistics. PMLR, 1018--1026. https://proceedings.mlr.press/v130/durmus21a.html
3. F B Hildebrand. 1987. Introduction to numerical analysis (2 ed.). Dover Publications, Mineola, NY.
4. Bias and Extrapolation in Markovian Linear Stochastic Approximation with Constant Stepsizes
5. Effectiveness of Constant Stepsize in Markovian LSA and Statistical Inference