The Gene Expression Messy Genetic Algorithm

Introduces the Gene Expression Messy Genetic Algorithm (GEMGA)-a new generation of messy genetic algorithms that directly search for relations among the members of the search space. GEMGA is an O[/spl Lambda//sup k/(l/sup 2/+k)] sample complexity algorithm for the class of order-k delineable problems (problems that can be solved by considering no higher than order-k relations). GEMGA is designed based on an alternate perspective of natural evolution, as proposed by the SEARCH (Search Envisioned As Relation and Class Hierarchizing) framework, that emphasizes the role of gene expression. GEMGA uses the transcription operator to search for relations. This paper also presents test results of the GEMGA for large multimodal order-k delineable problems.