Presentation 2000/3/21
Adaptive GP-Automata for the maze problem
Hiroaki Kataoka, Tomoharu Nagao,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Ganerally, it is difficult for Genetic Programming; GP to solve problem which requires several different output to the same input signal of an agent. This paper, we improve GP-Automata, a modified version of GP previously proposed by Ashlock for such problems, and propose a new method named Adaptive GP-Automata; AGPA. Since AGPA automatically obtains the adequate total number of states of the agent for solving a given task successfully through evolutionary iterations, complex and laborious trial and error for tuning it is not necessary in AGPA. Several experiments to investigate the performance of AGPA were executed using maze problems, and their results suggested the advantage of AGPA compared with the original version of GP-Automata.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Genetic Programming / GP-Automata / maze problem
Paper # AI99-92,KBSE99-51
Date of Issue

Conference Information
Committee AI
Conference Date 2000/3/21(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) Adaptive GP-Automata for the maze problem
Sub Title (in English)
Keyword(1) Genetic Programming
Keyword(2) GP-Automata
Keyword(3) maze problem
1st Author's Name Hiroaki Kataoka
1st Author's Affiliation Imaging Science and Engineering Laboratory, Tokyo Institute of Technology()
2nd Author's Name Tomoharu Nagao
2nd Author's Affiliation Imaging Science and Engineering Laboratory, Tokyo Institute of Technology
Date 2000/3/21
Paper # AI99-92,KBSE99-51
Volume (vol) vol.99
Number (no) 717
Page pp.pp.-
#Pages 8
Date of Issue