A Two-Operator Hybrid DE for Global Numerical Optimization

Solving single objective real-parameter problem is still a challenging task. In this paper, an effective and efficient self-adaptation framework is proposed, called ToHDE, which is hybrid with CMA-ES to improve the performance. The algorithm uses two mutation strategies with linear weighted parameter to balance the exploration and exploitation. Moreover, a two-stage population size reduction and a local research are used to increase the capability of ToHDE. We evaluated the performance of ToHDE on the IEEE CEC2014 benchmark suite and compared it with six state-of-the-art peer DE variants. The statistical results show that ToHDE is competitive with the compared methods.