Presentation 2003/6/20
A Hierarchical Reinforcement Learning by Dividing State Space Based on the Linearity of Dynamics
Norikazu SUGIMOTO, Kazuyuki SAMEJIMA, Kenji DOYA, Mitsuo KAWATO,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) To apply reinforcement learning (RL) for task which have continuous state-space, for example robot control, we can use typical algorithm of reinforcement learning if continuous state-space have been decomposed into discrete state-space. But, critical issues are how to decompose a continuous state-space into discrete state-space. So, we propose a new hierarchical RL method consist of two layers. Bottom layer decompose a continuous state-space based on the linearity of dynamics. Top layer do RL in discrete state-space promoted by bottom layer.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Continuous-Hierarchical-Module reinfocement learning / Non-linear control
Paper # NC2003-16
Date of Issue

Conference Information
Committee NC
Conference Date 2003/6/20(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 Neurocomputing (NC)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) A Hierarchical Reinforcement Learning by Dividing State Space Based on the Linearity of Dynamics
Sub Title (in English)
Keyword(1) Continuous-Hierarchical-Module reinfocement learning
Keyword(2) Non-linear control
1st Author's Name Norikazu SUGIMOTO
1st Author's Affiliation NAra Institute of Science and Technology:ATR, Computational Neuroscience Laboratories:Creating the Brain, CREST, Japan Science and Technology Corporation()
2nd Author's Name Kazuyuki SAMEJIMA
2nd Author's Affiliation ATR, Computational Neuroscience Laboratories:Creating the Brain, CREST, Japan Science and Technology Corporation
3rd Author's Name Kenji DOYA
3rd Author's Affiliation NAra Institute of Science and Technology:ATR, Computational Neuroscience Laboratories:Creating the Brain, CREST, Japan Science and Technology Corporation
4th Author's Name Mitsuo KAWATO
4th Author's Affiliation NAra Institute of Science and Technology:ATR, Computational Neuroscience Laboratories
Date 2003/6/20
Paper # NC2003-16
Volume (vol) vol.103
Number (no) 153
Page pp.pp.-
#Pages 6
Date of Issue