Adaptive non-uniform mutation based on statistics for genetic algorithms

Abstract

As a meta-heuristic search algorithm based on mechanisms abstracted from population genetics, the genetic algorithm (GA) implicitly maintains the statistics about the search space through the population. This implicit statistics can be explicitly used to enhance GA's performance. In this paper, a statistics-based adaptive non-uniform mutation (SANUM) is proposed. SANUM uses the statistics information of the allele distribution in each locus to adatively adjust the mutation operation. Our preliminary experiments show that SANUM outperforms traditional bit flip mutation across a representative set set of test problems.