The Selection of Surrogate Models in Evolutionary Algorithms

Abstract

In this paper we discuss the problem of the selection of surrogate models for the use with evolutionary algorithms. We compare the types of models selected by two different model selectors and discuss how the models affect the performance of the evolutionary optimizer and how the types of selected models change during the run of the evolution.

Keywords

• Surrogate models
• Evolutionary algorithms
• multiobjective optimization
• meta-learning