Evolutionary algorithm with dynamic population size for multi-objective optimization

Abstract

This paper presents a novel "incremental" multi-objective evolutionary algorithm with dynamic population size that is adaptively computed according to the on-line discovered trade-off surface and its desired population distribution density. It incorporates the method of fuzzy boundary local perturbation with interactive local fine-tuning for broader neighborhood exploration to achieve better convergence as well as discovering any gaps or missing trade-off regions and each generation. The effectiveness of the proposed methodology is validated upon a benchmark multi-objective optimization problem.