MOGA-II PERFORMANCE ON NOISY OPTIMIZATION PROBLEMS

Since the mid-fifties evolutionary algorithms (EAs) have been used in different optimization problems. In the last years their use was extended to the demanding field of multi-objective optimization. For this expansion, EAs themselves had to evolve to more complex forms. The question is whether an algorithm that is adapted to work well with multiple-objectives is still capable to handle single-objective optimization problems. In this paper we present a new EA for multi-objective optimization called MOGA-II. We test it on noisy single-objective problems and compare its performance with two algorithms for single-objective optimization. The results show that MOGA-II is a robust algorithm that can efficiently solve a palette of different optimization problems.