Skip to yearly menu bar Skip to main content


Minimax-Bayes Reinforcement Learning

Thomas Kleine Buening · Christos Dimitrakakis · Hannes Eriksson · Divya Grover · Emilio Jorge

Auditorium 1 Foyer 167


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.

Live content is unavailable. Log in and register to view live content