A Novel Evolutionary Algorithm with Pareto Front Adaption for Many-objective Optimization

Li Li, Avimanyu Sahoo, Liang Chang

Abstract: Evolutionary algorithms have been used to solve a variety of multi-objective optimization problems. However, those algorithms are very sensitive to the curvature of Pareto front, whereas the shape of the front usually hard to obtain beforehand. This paper proposes a new Pareto front estimation based evolutionary algorithm referred to as PaE/EA for many-objective optimization. In this algorithm, the geometric information of Pareto front is estimated by using achievement scalarizing function, which can help to solve the problems more efficiently. The proposed algorithm is compared with four representative algorithms on DTLZ and WFG test suites. It also has been testified by the multi-objective version of traveling salesman problem. The experiment results indicate that the proposed approach has a competitive performance.