A Many-Objective Evolutionary Algorithm with Pareto Front Estimation and Angle-Based Selection

Abstract

Evolutionary algorithms have been gaining increasing attention from the evolutionary computation research community. However, the performance of the algorithms deteriorates progressively in handling many-objective optimization problems due to the sensitivity of the curve of the Pareto front, which is usually hard to obtain beforehand. Convergence and diversity strongly depend on the geometry of the Pareto front. This paper proposes a novel algorithm consisting of an angle-based selection strategy and Pareto front estimation method. These two strategies are employed in the environment selection to select promising solutions. The proposed algorithm is compared with five representative algorithms on nine test problems. The experiment results show that the proposed algorithm outperforms state-of-the-art compared algorithms.