Differential Evolution with Fittest Individual Local Tuning

In this paper we propose Differential Evolution with Fittest Individual Local Tuning (DELT). Differential Evolution (DE) is well-known to be efficient at global optimization in complex search spaces but it is unable to local fine-tuning of candidate solutions. In the proposed method a local search approach has been embedded inside classical DE to add local search capability. By this way, the method gives the best individual in the current population an extra chance to local improvement of its fitness value in each generation. Proposed method speeds up DE without losing of robustness. The method has been tested over 8 well-known benchmark functions which contains both unimodal and multimodal problems. Furthermore, by adding noise to those functions, the performance of new approach has been investigated in noisy environment as well. Achieved results are highly promising.