Self-adaptation in genetic algorithms

Within Genetic Algorithms (GAs) the mutation rate is mostly handled as a global, external parameter, which is constant over time or exogeneously changed over time. In this paper a new approach is presented, which transfers a basic idea from Evolution Strategies (ESs) to GAs. Mutation rates are changed into endogeneous items which are adapting during the search process. First experimental results are presented , which indicate that environment{ dependent self{adaptation of appropriate settings for the mutation rate is possible even for GAs. Furthermore, the reduction of the number of external parameters of a GA is seen as a rst step towards achieving a problem{de-pendent self{adaptation of the algorithm.