GDE4: The Generalized Differential Evolution with Ordered Mutation

Abstract

Differential Evolution (DE) is one the most popular evolutionary algorithm (EA) to handle optimization problems with an efficient performance. Due to its success and popularity, it has been utilized by researchers in multi-objective optimization, so there are various multi-objective versions of DE. Similar to other population-based algorithms, DE uses a mutation operator to produce the new individual for the next generation. Although the original version of DE randomly selects three candidate solutions from the population without considering any ordering in its mutation scheme, this paper proposes ordering strategy of individuals which influences the performance of the algorithm. An enhanced version (GDE4) of Generalized Differential Evolution (GDE) with ordered mutation operator is designed. GDE is a multi-objective evolutionary algorithm based on DE. The proposed approach orders candidate individuals using popular ranking measures of multi-objective optimization problems to utilize the ordered solutions in mutation operator. The best one of three randomly selected solutions is considered as the parent, and two others are applied as second and third candidate solutions in DE mutation, respectively. Unlike most of the multi-objective methods which consider multi-objectiveness during the selection process, the proposed method improves the performance using a modification on the genetic operator. The standard benchmark functions and measures are adopted to evaluate the performance of GDE4. The conducted experiments are on 5, 10, and 15 objectives for the utilized benchmark set. The comparison results reveal that GDE4 algorithm outperforms GDE3, the last version of GDE.