Hybrid evolutionary programming technique using subpopulations with completely different evolution mechanisms

Abstract

A new hybrid technique using multiple subpopulations with completely different evolution mechanisms is proposed to enhance the possibility of quickly finding the global optimum of continuous fitness functions. Three subpopulations, each demonstrating a strength with particular shapes of the fitness function, cooperate, having a synergy effect in which the global performance greatly exceeds any of the local performances. The efficiency and robustness of the proposed technique is verified through benchmark test functions.