Identifying Meaningful Solution Structures in Massive Multiobjective Evolutionary Optimization

Abstract

Massive Multiobjective Optimization Problems (MMOPs) involve over three objectives, which make them very challenging to the conventional methods of optimization. With the escalation in the number of objectives, evolutionary algorithms may not be able to preserve convergence, diversity, and readability of solutions. This paper is aimed at finding meaningful solutions structures in Massive Multiobjective Evolutionary Optimization (MMOEO). The study will improve the knowledge of solutions and decision making by ensuring objective correlations and determination of relationship among decision variables are analyzed to generate a more effective solution. Structural properties identified in the study include, sparsity, clustering and dominance resistance, which can be utilized in enhancing performance of optimization. The results would serve to improve the algorithm structure and real-life applicability of multiobjective evolutionary optimization in complicated real problems