Acceleration of Reinforcement Learning by Efficient Policy Evaluation
-
- SENDA Kei
- Graduate School of Engineering, Kyoto University
-
- HATTORI Suguru
- Graduate School of Engineering, Kyoto University
-
- KOHDA Takehisa
- Graduate School of Engineering, Kyoto University
Bibliographic Information
- Other Title
-
- 強化学習における方策評価の効率化による学習の加速
- キョウカ ガクシュウ ニ オケル ホウサク ヒョウカ ノ コウリツカ ニ ヨル ガクシュウ ノ カソク
Search this article
Abstract
Typical methods for solving reinforcement learning problems iterate two steps, policy evaluation and policy improvement. This study proposes algorithms for the policy evaluation to improve learning efficiency. The proposed algorithms, based on the Krylov Subspace Method (KSM), are tens to hundreds times more efficient than existing algorithms based on the Stationary Iterative Methods (SIM). Algorithms based on KSM are far more efficient than they have been generally expected. This study clarifies what makes algorithms based on KSM makes more efficient with numerical examples and theoretical discussions.
Journal
-
- Transactions of the Society of Instrument and Control Engineers
-
Transactions of the Society of Instrument and Control Engineers 49 (7), 696-702, 2013
The Society of Instrument and Control Engineers
- Tweet
Details 詳細情報について
-
- CRID
- 1390282679479140992
-
- NII Article ID
- 10031188141
-
- NII Book ID
- AN00072392
-
- ISSN
- 18838189
- 04534654
-
- NDL BIB ID
- 024821264
-
- Text Lang
- ja
-
- Data Source
-
- JaLC
- NDL
- Crossref
- CiNii Articles
-
- Abstract License Flag
- Disallowed