Presentation 1999/5/27
Acquiring the Protocol of Communication for Task Assignment by Reinforcement Learning
Takefumi Udagawa, Masahiro Murakawa, Kazuyuki Hiraoka, Satoshi Morinaga, Shuji Yoshizawa,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) We propose a model of a multi-agent system which learns communication rule of di-viding works automatically by reinforcement learning. In multi-agent systems, it is important to divide works efficiently. However, efficient division of works needs much communication. The proposed method which communication rules of dividing works are learned by reinforcement learning, offers one of the solution of the two contrary problems: reduction of communication and efficient division of works/ We analyze effects of communication and how agents divide works with acquired rules. In addition, we show that it makes the rules simple and robust to add noise to communication at learning phase.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) multi-agent systems / reinforcement learning / classifier system / task assignment / agent communication / noise
Paper # AI99-9
Date of Issue

Conference Information
Committee AI
Conference Date 1999/5/27(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 Artificial Intelligence and Knowledge-Based Processing (AI)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Acquiring the Protocol of Communication for Task Assignment by Reinforcement Learning
Sub Title (in English)
Keyword(1) multi-agent systems
Keyword(2) reinforcement learning
Keyword(3) classifier system
Keyword(4) task assignment
Keyword(5) agent communication
Keyword(6) noise
1st Author's Name Takefumi Udagawa
1st Author's Affiliation Graduate School of Engineering, University of Tokyo()
2nd Author's Name Masahiro Murakawa
2nd Author's Affiliation Electrotechnical Laboratory
3rd Author's Name Kazuyuki Hiraoka
3rd Author's Affiliation Dept. of Information and Computer Science, Saitama University
4th Author's Name Satoshi Morinaga
4th Author's Affiliation NEC C&C Media Laboratory
5th Author's Name Shuji Yoshizawa
5th Author's Affiliation Graduate School of Engineering, University of Tokyo
Date 1999/5/27
Paper # AI99-9
Volume (vol) vol.99
Number (no) 95
Page pp.pp.-
#Pages 8
Date of Issue