An improved NSGA-III algorithm based on objective space decomposition for many-objective optimization
Soft Comput.(2016)
摘要
Maintaining balance between convergence and diversity is of great importance for many-objective evolutionary algorithms. The recently suggested non-dominated sorting genetic algorithm III could obtain a fair diversity but the convergence is unsatisfactory. For this purpose, an improved NSGA-III algorithm based on objective space decomposition (we call it NSGA-III-OSD) is proposed for enhancing the convergence of NSGA-III. Firstly, the objective space is decomposed into several subspaces by clustering the weight vectors uniformly distributed in the whole objective space and each subspace has its own population. Secondly, individual information is exchanged between subspaces in the mating selection phase. Finally, the convergence information is added in the environmental selection phase by the penalty-based boundary intersection distance. The proposed NSGA-III-OSD is tested on a number of many-objective optimization problems with three to fifteen objectives and compared with six state-of-the-art algorithms. Experimental results show that NSGA-III-OSD is competitive with the chosen state-of-the-art designs in convergence and diversity.
更多查看译文
关键词
Many-objective optimization, NSGA-III, Convergence, Objective space decomposition, PBI distance
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络