Hierarchical Approach Based Evolutionary Algorithm for Many-Objective Optimization
Pareto-dominance multi-objective evolutionary algorithms (PDMOEAs) are extensively employed in the literature to handle multi-objective problems (MOPs) effectively. However, the performance of PDMOEAs drastically reduces for the problems with higher objectives termed as the many-objective problems (MaOPs) due to the inefficiency of the Pareto-dominance to segregate the solutions. Hence, in this paper, we propose a hierarchical approach for the PDMOEAs to solve the MaOPs. The proposed approach employs Pareto-dominance along with approximate nondominated sorting and Shift-based density estimation in the mating and environmental selections to select and preserve better solutions respectively. To demonstrate the effectiveness of our algorithm, we have conducted the experiments on 16 benchmark problems with 64 test instances. The experimental results demonstrate that the proposed approach performs competitively with the state-of-art algorithms. Keywords - Hierarchical Approach, Evolutionary Algorithms, Pareto-dominance, Shift-based Density Estimation.