Best Member Detection and Using as Differential Evolution Crossover Operator in Decomposition-based Multiobjective Optimization Algorithm

Abstract

Decomposition is a method to distributes a mutliobjective problems to the many single objective problems like scalarization. Multi-objective Evolutionary Algorithm based on Decomposition (MOEA/D) is one of the many algorithms uses decomposition method. In MOEA/D algorithm genetic operators are preferred to alter the population. As one of the genetic operators, the crossover is an important element in the algorithm. Hence it is possible to propose new possible methods instead of well-known SBX method. Differential Evolution (DE) which is a single objective optimization algorithm can be used as crossover operator in MOEA/D. However, in DE the best member needed to be detected in the population. Even it is relatively easy in single objective, systematic methods are needed for this purpose. Therefore, in this research three different best member detection methodology will be compared in DE assist MOEA/D algorithm. These methods will be compared on benchmark problems with many objectives.