Presentation 1997/3/17
Learning of Viewpoint Movement by Delayed Reinforcement Signal
Tetsuo Nishino, Katsunari Shibata, Yoichi Okabe,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) When watching something, we move our veiwpoint to it. This veiwpoint movement can be thought as one of actions to achieve recognition. Then it is supposed that a system can learn to move its viewpoint appropriately without teaching how to move its veiwpoint. We propose a learning method using reinforcement learning for a neural network to learn to move its veiwpoint appropriately. The system can obtain the delayed reinforcement signal calculated from the recognition result it made finally. We conclude from simulations that the method enables the system to move its viewpoint appropriately though any information which directly tells how to move its viewpoint isn't given.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) reinforcement learning / active perception / sensor movement / Q-Learning
Paper # NC96-135
Date of Issue

Conference Information
Committee NC
Conference Date 1997/3/17(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) Learning of Viewpoint Movement by Delayed Reinforcement Signal
Sub Title (in English)
Keyword(1) reinforcement learning
Keyword(2) active perception
Keyword(3) sensor movement
Keyword(4) Q-Learning
1st Author's Name Tetsuo Nishino
1st Author's Affiliation Research Center for Advanced Science and Technology,the Univ. of Tokyo()
2nd Author's Name Katsunari Shibata
2nd Author's Affiliation Research Center for Advanced Science and Technology,the Univ. of Tokyo
3rd Author's Name Yoichi Okabe
3rd Author's Affiliation Research Center for Advanced Science and Technology,the Univ. of Tokyo
Date 1997/3/17
Paper # NC96-135
Volume (vol) vol.96
Number (no) 583
Page pp.pp.-
#Pages 8
Date of Issue