Eigen Crossover in Cooperative Model of Evolutionary Algorithms Applied to CEC 2022 Single Objective Numerical Optimisation

Abstract

In this paper, a cooperative model of four well-performing evolutionary algorithms enhanced by Eigen crossover is proposed and applied to a set of problems CEC 2022. The four adaptive algorithms employed in this model are - Covariance Matrix Adaptation Evolutionary Strategy (CMA-ES), Differen-tial Evolution with Covariance Matrix Learning and Bimodal Distribution Parameter Setting (CoBiDE), an adaptive variant of jSO, and Differential Evolution With an Individual-Dependent Mechanism (IDE). For the higher efficiency of the cooperative model, a linear population-size reduction mechanism is employed. The model was introduced for CEC 2019. Here, Eigen crossover is applied for each cooperating algorithm. The provided results show that the proposed model of four Evolutionary Algorithms with Eigen crossover (EA4eig) is able to solve ten out of 24 optimisation problems. Moreover, comparing EA4eig with four state-of-the-art variants of adaptive Differential Evolution illustrates the superiority of the newly designed optimiser.