Comparison of Novel Multi-Objective Self Organizing Migrating Algorithm with Conventional Methods

In the paper, three algorithms for the multi-objective optimization based on the strategy of a self-organized migration are compared. The first two algorithms - Weighted Sum Method and Rotated Weighted Metric Method - transform multiple objectives into a single fitness function. The third method - a novel MOSOMA - combines the principle of the non-dominated sorting of population in the objective space and the survey of the decision space of input variables based on the self-organized migration. All three algorithms are compared on the test problem with the Pareto front, which contains both convex and non-convex parts. Monitored parameters are generational distance, spread of solutions and CPU time.