Bayesian Immigrant Diploid Genetic Algorithm for Dynamic Environments

Abstract

In dynamic environments, the main aim of an optimization algorithm is to track the changes and to adapt the search process. In this paper, we propose an approach called the Bayesian Immigrant Diploid Genetic Algorithm (BIDGA). BIDGA uses implicit memory in the form of diploid chromosomes, combined with the Bayesian Optimization Algorithm (BOA), which is a form of Estimation of Distribution Algorithms (EDAs). Through the use of BOA, BIDGA is able to take into account epistasis in the form of binary relationships between the variables. Experiments show that the proposed approach is efficient and also indicates that exploiting interactions between variables is important to adapt to the newly formed environments.