Adaptive Update Range of Solutions in MOEA/D for Multi and Many-Objective Optimization


Abstract

MOEA/D, a representative multi-objective evolutionary algorithm, decomposes a multi-objective optimization problem into a number of single objective optimization problems and tries to approximate Pareto front by simultaneously optimizing each of these single objective problems. MOEA/D has several options to calculate a scalar value from multiple objective function values of a solution. In many-objective optimization problems including four or more objective functions, MOEA/D using the weighted sum scalarizing function achieves high search performance. However, the weighted sum has a serious problem that the entire concave Pareto front cannot be approximated. To overcome this problem of the weighted sum based MOEA/D, in this work we propose a method to adaptively determine update ranges of solutions in the framework of MOEA/D. The experimental results show that the weighted sum based MOEA/D using the proposed solution update method can approximate the entire concave Pareto front and improve the search performance.