Evaluating a hybrid DE and BBO with self adaptation on ICSI 2014 benchmark problems

Abstract

The paper presents a new hybrid differential evolution (DE) and biogeography-based optimization (BBO) algorithm and tests its performance on the benchmark set for the ICSI 2014 Competition. The algorithm tends to perform more DE mutations in early search stage and more BBO migrations in later stage, in order to provide a good balance of exploration and exploitation. It also uses a trial-and-error method inspired by the self-adaptive DE (SaDE) to choose appropriate mutation/ migration schemes during the search. Computational experiment shows that the algorithm outperforms DE, SaDE, and blended BBO on the benchmark set.