On the Sensitivity of Aggregative Multiobjective Optimization Methods

Abstract

In this paper, we present a study on the sensitivity of aggregation methods with respect to the weights associated with multiobjective functions of a multiobjective optimization problem. To do this study, we have developed some measurements, such as the speed metric or the distribution metric. We have performed this study on a set of biobjective optimization test problems : a convex, a non convex, a continuous and a combinatorial test problem. We show that some aggregation methods are more sensitive than others.