Efficient experience reuse in non-Markovian environments

Le Tien Dung, Takashi Komeda, Motoki Takagi

研究成果

4 被引用数 (Scopus)

抄録

Learning time is always a critical issue in Reinforcement Learning, especially when Recurrent Neural Networks are used to predict Q values in non-Markovian environments. Experience reuse has been received much attention due to its ability to reduce learning time. In this paper, we propose a new method to efficiently reuse experience. Our method generates new episodes from recorded episodes using an action-pair merger. Recorded episodes and new episodes are replayed after each learning epoch. We compare our method with standard online learning, and learning using experience replay in a vision-based robot problem. The results show the potential of this approach.

本文言語English
ホスト出版物のタイトルProceedings of SICE Annual Conference 2008 - International Conference on Instrumentation, Control and Information Technology
ページ3327-3332
ページ数6
DOI
出版ステータスPublished - 2008
イベントSICE Annual Conference 2008 - International Conference on Instrumentation, Control and Information Technology - Tokyo, Japan
継続期間: 2008 8 202008 8 22

出版物シリーズ

名前Proceedings of the SICE Annual Conference

Conference

ConferenceSICE Annual Conference 2008 - International Conference on Instrumentation, Control and Information Technology
国/地域Japan
CityTokyo
Period08/8/2008/8/22

ASJC Scopus subject areas

  • 制御およびシステム工学
  • コンピュータ サイエンスの応用
  • 電子工学および電気工学

フィンガープリント

「Efficient experience reuse in non-Markovian environments」の研究トピックを掘り下げます。これらがまとまってユニークなフィンガープリントを構成します。

引用スタイル