Header logo is

Relative Entropy Policy Search


Conference Paper



Policy search is a successful approach to reinforcement learning. However, policy improvements often result in the loss of information. Hence, it has been marred by premature convergence and implausible solutions. As first suggested in the context of covariant policy gradients (Bagnell and Schneider 2003), many of these problems may be addressed by constraining the information loss. In this paper, we continue this path of reasoning and suggest the Relative Entropy Policy Search (REPS) method. The resulting method differs significantly from previous policy gradient approaches and yields an exact update step. It works well on typical reinforcement learning benchmark problems.

Author(s): Peters, J. and Mülling, K. and Altun, Y.
Journal: Proceedings of the Twenty-Fourth National Conference on Artificial Intelligence
Pages: 1607-1612
Year: 2010
Month: July
Day: 0
Editors: Fox, M. , D. Poole
Publisher: AAAI Press

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

Event Name: Twenty-Fourth National Conference on Artificial Intelligence (AAAI-10)
Event Place: Atlanta, GA, USA

Address: Menlo Park, CA, USA
Institution: Association for the Advancement of Artificial Intelligence
ISBN: 978-1-577-35463-5
Language: en
Organization: Max-Planck-Gesellschaft
School: Biologische Kybernetik

Links: PDF


  title = {Relative Entropy Policy Search},
  author = {Peters, J. and M{\"u}lling, K. and Altun, Y.},
  journal = {Proceedings of the Twenty-Fourth National Conference on Artificial Intelligence},
  pages = {1607-1612},
  editors = {Fox, M. , D. Poole},
  publisher = {AAAI Press},
  organization = {Max-Planck-Gesellschaft},
  institution = {Association for the Advancement of Artificial Intelligence},
  school = {Biologische Kybernetik},
  address = {Menlo Park, CA, USA},
  month = jul,
  year = {2010},
  month_numeric = {7}