On the Use of Non-Stationary Policies for Stationary Infinite-Horizon Markov Decision Processes

Abstract : We consider infinite-horizon stationary $\gamma$-discounted Markov Decision Processes, for which it is known that there exists a stationary optimal policy. Using Value and Policy Iteration with some error $\epsilon$ at each iteration, it is well-known that one can compute stationary policies that are $\frac{2\gamma}{(1-\gamma)^2}\epsilon$-optimal. After arguing that this guarantee is tight, we develop variations of Value and Policy Iteration for computing non-stationary policies that can be up to $\frac{2\gamma}{1-\gamma}\epsilon$-optimal, which constitutes a significant improvement in the usual situation when $\gamma$ is close to $1$. Surprisingly, this shows that the problem of ''computing near-optimal non-stationary policies'' is much simpler than that of ''computing near-optimal stationary policies''.
Document type :
Conference papers
NIPS 2012 - Neural Information Processing Systems, Dec 2012, South Lake Tahoe, United States. 2012


https://hal.inria.fr/hal-00758809
Contributor : Bruno Scherrer <>
Submitted on : Thursday, November 29, 2012 - 1:31:31 PM
Last modification on : Tuesday, December 4, 2012 - 10:51:28 AM

Files

nips2012.pdf
fileSource_public_author

Identifiers

  • HAL Id : hal-00758809, version 1
  • ARXIV : 1211.6898

Collections

Citation

Bruno Scherrer, Boris Lesner. On the Use of Non-Stationary Policies for Stationary Infinite-Horizon Markov Decision Processes. NIPS 2012 - Neural Information Processing Systems, Dec 2012, South Lake Tahoe, United States. 2012. <hal-00758809>

Export

Share

Metrics

Consultation de
la notice

117

Téléchargement du document

170