Differential evolution with center-based mutation for large-scale optimization
Hanan Hiba, Sedigheh Mahdavi, Shahryar Rahnamayan · 2017
The Differential Evolution (DE) is proved to be a successful approach for solving challenging optimization problems. However, its performance is degraded when solving high dimensional problems. Several significant enhancements of the DE have been proposed in recent years, including a variant of it with modified main operators (i.e., mutation and crossover). This paper proposes a center-based mutation scheme, which is based on the utilization of center of the gravity as a base vector. This mutation scheme aims to generate the candidate solution by using the center of three randomly selected candidate solutions. This new scheme is evaluated on CEC 2013 LSGO benchmark functions on the dimension 1000, as well as on fifteen shifted discrete benchmark functions on dimensions 500 and 1000. Experimental results confirm that the new scheme achieves a great success rate in comparison with the classical DE over the most of the test problems in terms of convergence rate and solution accuracy.