A painless gradient-assisted multi-objective memetic mechanism for solving continuous bi-objective optimization problems
Adriana Lara, Carlos A. Coello Coello, Oliver Schütze · 2010
In this work we present a simple way to introduce gradient-based information as a means to improve the search performed by a multi-objective evolutionary algorithm (MOEA). Our proposal can be easily incorporated into any MOEA, and is able to improve its performance when solving continuous bi-objective problems. We propose a novel mechanism to control the balance between the local search, and the global search performed by a MOEA. We discuss the advantages of the proposed method and its possible use when dealing with more objectives. Finally, we provide some guidelines regarding the use of our proposed approach.