An improved genetic algorithm performance with benchmark functions

Abstract

In a previous paper by the authors (1998) it was shown that a new genetic algorithm (GA) performed better than a basic GA. In the present paper we focus on an improved genetic algorithm which uses the theory of cosets. The new technique was tested using some benchmark functions for which the basic GA performance was not good enough. The main purpose of this paper is to suggest how to use linear codes properties to guide the GA search.