English

Minimax-Bayes Reinforcement Learning

Machine Learning 2023-02-22 v1 Machine Learning

Abstract

While the Bayesian decision-theoretic framework offers an elegant solution to the problem of decision making under uncertainty, one question is how to appropriately select the prior distribution. One idea is to employ a worst-case prior. However, this is not as easy to specify in sequential decision making as in simple statistical estimation problems. This paper studies (sometimes approximate) minimax-Bayes solutions for various reinforcement learning problems to gain insights into the properties of the corresponding priors and policies. We find that while the worst-case prior depends on the setting, the corresponding minimax policies are more robust than those that assume a standard (i.e. uniform) prior.

Keywords

Cite

@article{arxiv.2302.10831,
  title  = {Minimax-Bayes Reinforcement Learning},
  author = {Thomas Kleine Buening and Christos Dimitrakakis and Hannes Eriksson and Divya Grover and Emilio Jorge},
  journal= {arXiv preprint arXiv:2302.10831},
  year   = {2023}
}
R2 v1 2026-06-28T08:45:49.194Z