Improved population partitioning method in multi-objective shuffled frog leaping algorithm

Population partitioning strategy is one of crucial issues to the performance of shuffled frog leaping algorithm. This paper proposed a new population partitioning strategy in the case of multi-objective optimization. The idea is to divide the closely located non-dominated solutions into the same memeplexes so that different memeplexes evolve toward different potential optimal area of search space in order to prevent the optimizer from prematurity. The dominated individuals are assigned to memeplexes according to their approximation to the non-dominated set. Moreover, adding individuals from other memeplexes is used to promote the diversity of each memeplex. The comparative results with other partitioning methods on test problems involving up to ten objectives show that, the new method further improves the performance of multi-objective shuffled frog leaping algorithm not only in convergence and speed of convergence to Pareto optimal front.