A State Predictor Based Reinforcement Learning System

• Kobayashi Kunikazu

  Faculty of Engineering, Yamaguchi University

• Nakano Koji

  Faculty of Engineering, Yamaguchi University

• Kuremoto Takashi

  Faculty of Engineering, Yamaguchi University

• Obayashi Masanao

  Faculty of Engineering, Yamaguchi University

Bibliographic Information

Other Title

• 状態予測型強化学習システム
• ジョウタイ ヨソクガタ キョウカ ガクシュウ システム

Abstract

The present paper proposes a new reinforcement learning (RL) system called a state predictor based RL system in order to solve the explosion of state space and create cooperative behaviors in multi-agent systems. The proposed system realizes a predictive function by representing both the present and the next state-action groups with ITPM which is one of incremental topology maps. The proposed system is applied to pursuit problem, and its performance is evaluated by comparing with conventional RL method through computer simulations. The experimental result shows that the proposed system can appropriately learn in a complex environment which is hardly solved by conventional RL. Furthermore, it is confirmed that the proposed system can acquire cooperative strategies.

Journal

• IEEJ Transactions on Electronics, Information and Systems

  IEEJ Transactions on Electronics, Information and Systems 128 (8), 1303-1311, 2008

  The Institute of Electrical Engineers of Japan

Keywords

• reinforcement learning
• multi-agent system
• cooperative behavior
• state predictor
• topology preserving map

Details

• CRID: 1390282679583408000
• NII Article ID: 10024266783
• NII Book ID: AN10065950
• DOI: 10.1541/ieejeiss.128.1303
• ISSN: 13488155
• ISSN: 03854221
• NDL BIB ID: 9600293
• Web Site: http://id.ndl.go.jp/bib/9600293
• Web Site: https://ndlsearch.ndl.go.jp/books/R000000004-I9600293
• Web Site: https://www.jstage.jst.go.jp/article/ieejeiss/128/8/128_8_1303/_pdf
• Text Lang: ja
• Data Source:
  • JaLC
  • NDL
  • Crossref
  • CiNii Articles
  • KAKEN
• Abstract License Flag: Disallowed