Presentation 2006-03-21
Evaluating Progress of Reinforcement Learning Using Recurrence Plots
Tetsuya TAKAHASHI, Masaharu ADACHI,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Reinforcement learning is a kind of learning systems which can deal with an unknown environment. In the reinforcement learning, an agent learns the optimal actions by applying a trial-and-error to an environment. Therefore, it is known that it can apply also to a dynamic environment. It is already reported that the method of adjusting specific parameters in the reinforcement learning is effective, when an agent learns a dynamic environment. The method for adjusting the parameters is known as meta-learning in the reinforcement learning. In this article, we propose a novel method for detecting environmental changes using recurrence plots in the reinforcement learning, and present a meta-learning by using a feature of environmental changes obtained from recurrence plots. It is shown that the proposed meta-learning improves the learning performance.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) reinforcement learning / environmental changes / recurrence plots / meta-learning
Paper # NLP2005-155
Date of Issue

Conference Information
Committee NLP
Conference Date 2006/3/14(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Nonlinear Problems (NLP)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Evaluating Progress of Reinforcement Learning Using Recurrence Plots
Sub Title (in English)
Keyword(1) reinforcement learning
Keyword(2) environmental changes
Keyword(3) recurrence plots
Keyword(4) meta-learning
1st Author's Name Tetsuya TAKAHASHI
1st Author's Affiliation Department of Electronic Engineering, Graduate School of Engineering, Tokyo Denki University /()
2nd Author's Name Masaharu ADACHI
2nd Author's Affiliation
Date 2006-03-21
Paper # NLP2005-155
Volume (vol) vol.105
Number (no) 676
Page pp.pp.-
#Pages 6
Date of Issue