Analysing the Impact of Niching in Multi-Objective EvolutionaryAlgorithm based on Decompose (MOEA/D)

Recently evolutionary algorithms have been successfully usedfor solving multi-objective optimization problems. Evolutionaryalgorithms are trying to simulate the nature’s behavior in findingthe best solution for problems. Multi-Objective EvolutionaryAlgorithm based on Decomposed (MOEA/D) is an evolutionaryframework for estimating the Pareto optimal points by lesscomputational complexity. This framework has some settingparameters which seem to have important role on achieving finesolutions. Neighboring is such parameter that the best tuning ofits boundary leads to improve the performance of MOEA/D.Since the island model is one of the easiest niching methods thatcan study the manner of evolution in niches, this paper presents aNon-uniform Island MOEA/D to investigate the effect ofneighborhood size on the performance of the solutions in theobjective space. The experimental results show that the Nonuniform Island MOEA/D outperforms the original MOEA/D onthe test instances

Keywords:

Multi-objective optimization , MOEA/D , neighboring , Pareto-set approximation