eduzhai > Applied Sciences > Engineering >

Necessary and Sufficient Conditions for Inverse Reinforcement Learning of Bayesian Stopping Time Problems

  • king
  • (0) Download
  • 20210505
  • Save

... pages left unread,continue reading

Document pages: 43 pages

Abstract: This paper presents an inverse reinforcement learning (IRL) framework forBayesian stopping time problems. By observing the actions of a Bayesiandecision maker, we provide a necessary and sufficient condition to identify ifthese actions are consistent with optimizing a cost function; then we constructset valued estimates of the cost function. To achieve this IRL objective, weuse novel ideas from Bayesian revealed preferences stemming frommicroeconomics. To illustrate our IRL scheme,we consider two important examplesof stopping time problems, namely, sequential hypothesis testing and Bayesiansearch. Finally, for finite datasets, we propose an IRL detection algorithm andgive finite sample bounds on its error probabilities. Also we discuss how toidentify $ epsilon$-optimal Bayesian decision makers and perform IRL.

Please select stars to rate!


0 comments Sign in to leave a comment.

    Data loading, please wait...