Tabu-based exploratory evolutionary algorithm for effective multi-objective optimization

Abstract

This paper proposes an exploratory multi-objective evolutionary algorithm (EMOEA) that makes use of the integrated features of tabu search and evolutionary algorithms for effective multi-objective optimization. It incorporates a tabu list and tabu constraint for individual examination and preservation to enhance the evolutionary search diversity in multi-objective optimization, which subsequently helps to avoid the search from trapping in local optima and at the same time, promotes the evolution towards the global Pareto-front. A novel method of lateral interference is also suggested, which is capable of distributing non-dominated individuals uniformly along the discovered Pareto-front at each generation. Unlike existing niching/sharing methods, lateral interference can be performed without the need of any parameter setting and can be flexibly applied in either parameter or objective domain depending on the nature of the optimization problem involved. The proposed features are experimented in order to illustrate their behavior and usefulness in the algorithm.