Presentation 1996/3/14
An Analysis on the Behavior of Genetic Algorithms by Markov Processes with Rewards
Kazuhiro Matsui, Yukio Kosugi,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) We propose a method to analyze the behavior of Genetic Algorithms (GAs) using Markov processes with rewards, which are extensions of Markov processes by introducing a concept of rewards. We derive the expected maximum and mean fitness values concerning simple GA models. These values are explicitly expressed as functions of generations and can be calculated without simulations, even for the generations at infinity. We analyze a simple count-one problem and a minimal deceptive problem. We then discuss the optimum mutation rate and compare GA and random-search to show the effectiveness of GAs.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Genetic Algorithms / Markov Processes / Rewards / Expected Fitness / Mutation Rate
Paper # AI95-64,KBSE95-52
Date of Issue

Conference Information
Committee AI
Conference Date 1996/3/14(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) An Analysis on the Behavior of Genetic Algorithms by Markov Processes with Rewards
Sub Title (in English)
Keyword(1) Genetic Algorithms
Keyword(2) Markov Processes
Keyword(3) Rewards
Keyword(4) Expected Fitness
Keyword(5) Mutation Rate
1st Author's Name Kazuhiro Matsui
1st Author's Affiliation Interdisciplinary Graduate School of Science and Engineering, Tokyo Institute of Technology()
2nd Author's Name Yukio Kosugi
2nd Author's Affiliation Interdisciplinary Graduate School of Science and Engineering, Tokyo Institute of Technology
Date 1996/3/14
Paper # AI95-64,KBSE95-52
Volume (vol) vol.95
Number (no) 573
Page pp.pp.-
#Pages 8
Date of Issue