An Improved Scalarization-based Dominance Evolutionary Algorithm for Many-Objective Optimization

Many-objective optimization problems (MaOPs) pose a multitude of challenges for existing multi-objective evolutionary algorithms. One of the key challenges is the poor selection pressure for optimization problems involving a high-dimensional objective space. To overcome this challenge, this paper extends the scalarization-based dominance evolutionary algorithm (SDEA) to improve its convergence rate. Inspired by the neighborhood information sharing scheme between the subproblems in the decomposition-based multi-objective evolutionary algorithm (MOEA/D), a selection mechanism is proposed for enhancing the SDEA in tackling MaOPs. The improved SDEA model is evaluated using different MaOP instances, which include DTLZ and WFG. The results indicate the effectiveness of the enhanced SDEA model in undertaking MaOPs.