Thompson Sampling is Asymptotically Optimal in General Environments
Machine Learning
2016-06-06 v2 Artificial Intelligence
Machine Learning
Abstract
We discuss a variant of Thompson sampling for nonparametric reinforcement learning in a countable classes of general stochastic environments. These environments can be non-Markov, non-ergodic, and partially observable. We show that Thompson sampling learns the environment class in the sense that (1) asymptotically its value converges to the optimal value in mean and (2) given a recoverability assumption regret is sublinear.
Cite
@article{arxiv.1602.07905,
title = {Thompson Sampling is Asymptotically Optimal in General Environments},
author = {Jan Leike and Tor Lattimore and Laurent Orseau and Marcus Hutter},
journal= {arXiv preprint arXiv:1602.07905},
year = {2016}
}
Comments
UAI 2016