Presentation 2001/3/16
A Model of Prism Adaptation based on Reinforcement Learning and Supervised Learning
Yutaka Sakaguchi, Mitsuo Takano,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) The authors propose a visuo-motor learning model based on reinforcement learning(RL)and supervised learning(SL), in order to explain the human behavior during prism adaptation. In the proposed model, the motor planning module, which works with RL, chooses appropriate motor commands based on the prediction by the internal model modules which work with SL.Moreover,an appropriate internal model is selected by the RL-based context switching module. Utilizing the"reliability of internal model, "the model realizes both parameter modification for a gradual environmental change and context-switch for a sudden environmental change. The behavior of the proposed model is illustrated through a computer simulation. Te relation between the proposed model and other modular learning algorithms is discussed.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) prism adaptation / supervised learning / reinforcement learning / context switching / reaching movement
Paper # NC2000-169
Date of Issue

Conference Information
Committee NC
Conference Date 2001/3/16(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 Model of Prism Adaptation based on Reinforcement Learning and Supervised Learning
Sub Title (in English)
Keyword(1) prism adaptation
Keyword(2) supervised learning
Keyword(3) reinforcement learning
Keyword(4) context switching
Keyword(5) reaching movement
1st Author's Name Yutaka Sakaguchi
1st Author's Affiliation Graduate School of Information Systems, University of Electro-Communications()
2nd Author's Name Mitsuo Takano
2nd Author's Affiliation Graduate School of Information Systems, University of Electro-Communications
Date 2001/3/16
Paper # NC2000-169
Volume (vol) vol.100
Number (no) 688
Page pp.pp.-
#Pages 8
Date of Issue