A Genetic Algorithm for Nonstationary Function Optimization Problems

概要

論文の詳細を見る
An extended genetic algorithm for solving nonstationary function optimization problems is presented. When using a standard genetic algorithm, it is difficut to deal with those problems due to brittleness caused by the fact that the population tends to stay where it believes to be optimal. In order to overcome this unwanted phenomenon, a new string representation associated with inactive regions for being capable of adopting various types of neutral mutations is presented. Genetic operations except for point mutations are prepared to be effective for all the genes in as a subset of genes that which is also an object of other genetic opetations. Further, using this extension, two 0/1 knapsack problems are examined to discuss the dynamics of the extended genetic algorithm. In the experiments, the population shows the behavior asserted in the neutral theory of molecular evolution and the directed evolution emerged as the result of acquiring two adaptive strategies.
システム制御情報学会の論文