Selection strategies and random perturbations in differential evolution

Differential evolution is a simple algorithm for global optimization. Basically it consists of three operations: mutation, crossover and selection. Despite many research papers dealing with the first two, hardly any attention has been paid to the third one nor is there a place for this operation in the algorithm basic naming scheme. In the paper we show that employing different selection strategies combined with some random perturbation of population vectors notably improves performance in high-dimensional problems. Further analysis of results shows that the improvement is statistically significant.