Development and applications of an evolutionary multiobjective optimization algorithm for many-objective problems

点击翻译按钮获取中文摘要

In this paper, we examined the ability of evolutionary multiobjective optimization(EMO) algorithms to efficiently search for Pareto-optimal solutions of many-objective problems with a large number of conflicting objectives. We also examined the relation between the search ability of EMO algorithms and the dependency among objectives of many-objective problems. Based on these examinations, we proposed a new crossover operator for diversity maintenance, an approximation method for efficient hypervolume calculation, and the use of multiple scalarizing functions for efficient multiobjective search.