Enhanced Opposition-based Differential Evolution Using Dynamic Optimum for Function Optimization

Abstract

Opposition-based Differential Evolution (ODE) is one of the effective various DE variants. It is faster in convergence speed and robust in search abilities. The concept of opposition is utilized in the algorithm. During the algorithm, it evaluates an estimate and its corresponding opposite estimate at the same time. Many evaluation results on function optimization problems have shown the selection of the symmetry point between the estimate and the opposite estimate which will greatly influence variants’ performance. In this paper, we propose a new type of ODE variant, Oppositional Differential Evolution using the Dynamic Optimum (ODEDO). ODEDO uses the optimum in the current population to learn the opposite estimates, and employs a similar strategy as traditional ODE in population initialization and generation jumping. The approach is validated using 5 benchmark functions. The results show a better performance when compare ODEDO to classical DE and ODE.