Efficient Algorithms for Reinforcement Learning by Linear Programming

• SENDA Kei

  Graduate School of Engineering, Kyoto University

• AMANO Koyu

  Graduate School of Engineering, Kyoto University

Bibliographic Information

Other Title

• 強化学習における線形計画法を用いた効率的解法
• キョウカ ガクシュウ ニ オケル センケイ ケイカクホウ オ モチイタ コウリツテキカイホウ

Abstract

Model-based reinforcement learning includes two steps, estimation of a plant and planning. Planning is formulated as dynamic programming (DP) problem, which is solved by a DP method. This DP problem has an equivalent linear programming (LP) problem that can be solved by LP method, but it is generally less efficient than typical DP method. However, numerical examples show linear programming is more efficient than the typical DP method in problems whose self-transition probabilities are large. The reason is clarified by geometrical discussion of each solution of method approaches to optimal solution.

Journal

• Transactions of the Society of Instrument and Control Engineers

  Transactions of the Society of Instrument and Control Engineers 52 (10), 566-572, 2016

  The Society of Instrument and Control Engineers

Keywords

• reinforcement learning
• planning
• linear programming
• self-transition probability

Details

• CRID: 1390282679485390336
• NII Article ID: 130005432995
• NII Book ID: AN00072392
• DOI: 10.9746/sicetr.52.566
• ISSN: 18838189
• ISSN: 04534654
• HANDLE: 2433/226829
• NDL BIB ID: 027720066
• Web Site: https://ndlsearch.ndl.go.jp/books/R000000004-I027720066
• Web Site: https://www.jstage.jst.go.jp/article/sicetr/52/10/52_566/_pdf
• Text Lang: ja
• Data Source:
  • JaLC
  • IRDB
  • NDL
  • Crossref
  • CiNii Articles
  • KAKEN
• Abstract License Flag: Disallowed