A Study on the influence of domination control techniques and leader selection methods in Many-Objective Problems

Pareto Based Multi Objective Optimization Algorithms, including Multi-Objective Particle Swarm Optimization (MOPSO), face several problems when applied to Multi Objective Optimization Problems with a large number of objective functions, especially the deterioration of the search ability. In the literature some techniques were proposed to overcome these limitations, among them the modification of the domination relation and leader selection methods. In this work, it is investigated the influence of the domination control techniques, like CDAS and S-CDAS, on the leader selection methods. This study aims to find which combination of leader selection mechanism and dominance control technique present the better performance and to analyze if the alteration in the domination control technique influences the leader selection behavior. The combinations of leader selection and domination control methods are compared through experiments in terms of convergence and diversity in many-objective scenarios. The results are analyzed through different quality indicators and statistical tests.