Empirical Inference Conference Paper 2009

Efficient Sample Reuse in EM-Based Policy Search

Direct policy search is a promising reinforcement learning framework in particular for controlling in continuous, high-dimensional systems such as anthropomorphic robots. Policy search often requires a large number of samples for obtaining a stable policy update estimator due to its high flexibility. However, this is prohibitive when the sampling cost is expensive. In this paper, we extend a EM-based policy search method so that previously collected samples can be efficiently reused. The usefulness of the proposed method, called Reward-weighted Regression with sample Reuse, is demonstrated through a robot learning experiment.

Author(s): Hachiya, H. and Peters, J. and Sugiyama, M.
Book Title: 16th European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases
Journal: Machine Learning and Knowledge Discovery in Databases: European Conference ECML PKDD 2009
Pages: 469-484
Year: 2009
Month: September
Day: 0
Editors: Buntine, W. , M. Grobelnik, D. Mladenic, J. Shawe-Taylor
Publisher: Springer
Bibtex Type: Conference Paper (inproceedings)
Address: Berlin, Germany
DOI: 10.1007/978-3-642-04180-8_48
Event Name: ECML PKDD 2009
Event Place: Bled, Slovenia
Electronic Archiving: grant_archive
Language: en
Organization: Max-Planck-Gesellschaft
School: Biologische Kybernetik
Links:

BibTex

@inproceedings{6068,
  title = {Efficient Sample Reuse in EM-Based Policy Search},
  journal = {Machine Learning and Knowledge Discovery in Databases: European Conference ECML PKDD 2009},
  booktitle = {16th European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases},
  abstract = {Direct policy search is a promising reinforcement learning framework in particular for controlling in continuous, high-dimensional systems such as anthropomorphic robots. Policy search often requires a large number of samples for obtaining a stable policy update estimator due to its high flexibility. However, this is prohibitive when the sampling cost is expensive. In this paper, we extend a EM-based policy search method so that previously collected samples can be efficiently reused. The usefulness of the proposed method, called Reward-weighted Regression with sample Reuse, is demonstrated through a robot learning experiment.},
  pages = {469-484},
  editors = {Buntine, W. , M. Grobelnik, D. Mladenic, J. Shawe-Taylor},
  publisher = {Springer},
  organization = {Max-Planck-Gesellschaft},
  school = {Biologische Kybernetik},
  address = {Berlin, Germany},
  month = sep,
  year = {2009},
  slug = {6068},
  author = {Hachiya, H. and Peters, J. and Sugiyama, M.},
  month_numeric = {9}
}