Header logo is

Predictive Representations for Policy Gradient in POMDPs

2009

Conference Paper

ei


We consider the problem of estimating the policy gradient in Partially Observable Markov Decision Processes (POMDPs) with a special class of policies that are based on Predictive State Representations (PSRs). We compare PSR policies to Finite-State Controllers (FSCs), which are considered as a standard model for policy gradient methods in POMDPs. We present a general Actor- Critic algorithm for learning both FSCs and PSR policies. The critic part computes a value function that has as variables the parameters of the policy. These latter parameters are gradually updated to maximize the value function. We show that the value function is polynomial for both FSCs and PSR policies, with a potentially smaller degree in the case of PSR policies. Therefore, the value function of a PSR policy can have less local optima than the equivalent FSC, and consequently, the gradient algorithm is more likely to converge to a global optimal solution.

Author(s): Boularias, A. and Chaib-Draa, B.
Book Title: ICML 2009
Journal: Proceedings of the 26th International Conference on Machine Learning (ICML 2009)
Pages: 65-72
Year: 2009
Month: June
Day: 0
Editors: Danyluk, A. , L. Bottou, M. Littman
Publisher: ACM Press

Department(s): Empirical Inference
Bibtex Type: Conference Paper (inproceedings)

DOI: 10.1145/1553374.1553383
Event Name: 26th International Conference on Machine Learning
Event Place: Montreal, Canada

Address: New York, NY, USA
Digital: 0
Language: en
Organization: Max-Planck-Gesellschaft
School: Biologische Kybernetik

Links: PDF
Web

BibTex

@inproceedings{6827,
  title = {Predictive Representations for Policy Gradient in POMDPs},
  author = {Boularias, A. and Chaib-Draa, B.},
  journal = {Proceedings of the 26th International Conference on Machine Learning (ICML 2009)},
  booktitle = {ICML 2009},
  pages = {65-72},
  editors = {Danyluk, A. , L. Bottou, M. Littman},
  publisher = {ACM Press},
  organization = {Max-Planck-Gesellschaft},
  school = {Biologische Kybernetik},
  address = {New York, NY, USA},
  month = jun,
  year = {2009},
  month_numeric = {6}
}