While hybridizing the complementary constituent soft computing techniques has displayed improved efficacy, the hybridization of complementary characteristics of different Differential Evolution (DE) variants (could as well be extended to evolutionary algorithms variants in general) through heterogeneous mixing in a distributed framework also holds a great potential. This chapter proposes to mix competitive DE variants with diverse characteristics in a distributed framework as against the typical distributed (homogeneous) Differential Evolution (dDE) algorithms found in DE literature. After an empirical analysis of 14 classical DE variants on 14 test functions, two heterogeneous dDE frameworks dDE_HeM_best and dDE_HeM_worst obtained by mixing best DE variants and worst DE variants, respectively, have been realized, implemented and tested on the benchmark optimization problems. The simulation results have validated the robustness of the heterogeneous mixing of best variants. The chapter also hybridized DE and dynamic DE variants in a distributed framework. The robustness of the resulting framework has been validated by benchmarking it against the state-of-the-art DE algorithms in the literature.
Dr. Jeyakumar G. and Dr. Shunmuga Velayutham C., “Hybridizing differential evolution variants through heterogeneous mixing in a distributed framework”, Studies in Computational Intelligence, vol. 611, pp. 107-151, 2016.