On the design of diploid genetic algorithms for problem optimization in dynamic environments

Abstract

Using diploidy and dominance is one method to enhance the performance of genetic algorithms in dynamic environments. For diploidy genetic algorithms, there are two key design factors: the cardinality of genotypic alleles and the uncertainty in the dominance scheme. This paper investigates the effect of these two factors on the performance of diploidy genetic algorithms in dynamic environments. A generalized diploidy and dominance scheme is proposed for diploidy genetic algorithms, where the cardinality of genotypic alleles and/or the uncertainty in the dominance scheme can be easily tuned and studied. The experimental results show the efficiency of increasing genotypic cardinality rather than introducing uncertainty in the dominance scheme.