Asymmetric pareto-adaptive scheme for multiobjective optimization

Abstract

A core challenge of Multiobjective Evolutionary Algorithms (MOEAs) is to attain evenly distributed Pareto optimal solutions along the Pareto front. In this paper, we propose a novel asymmetric Pareto-adaptive (apa) scheme for the identification of well distributed Pareto optimal solutions based on the geometrical characteristics of the Pareto front. The apa scheme applies to problem with symmetric and asymmetric Pareto fronts. Evaluation on multiobjective problems with Pareto fronts of different forms confirms that apa improves both convergence and diversity of the classical decomposition-based (MOEA/D) and Pareto dominance-based MOEAs (paε-MyDE). © 2011 Springer-Verlag.