An Improved Multi-Objective Evolution Algorithm Based on Shuffled Frog Leaping

In this paper, we present a meta-heuristic base on improved shuffled frog leaping algorithm (SFLA) to tackle the multi-objective problem (MOP). The SFLA is suitable to solve the single objective problem. For the multi-objective problem, one main issue is that how to evaluate the quality of two optional solutions and select the better one from them. The traditional Pareto dominance cannot generate a strong selection pressure toward the Pareto front when we have many objectives (since almost all solutions in the current population become non-dominated). In our algorithm, we propose a relaxed dominance mechanism to promote the selection pressure in evolution. The comparison between two frogs proposed in this work takes the modified Pareto dominance relations into account. At the same time, the number of frog in each memeplex of SFLA is not too much, we only select two individuals (the best and the worst) to perform the memetic evolution. Therefore, the algorithm has the better selection pressure toward to the Pareto front. The experimental results show that our algorithm processes good performance to solve the MOP.