Exploring new learning strategies in Differential Evolution algorithm

In the field of evolutionary algorithm, Differential Evolution (DE) has gained a great focus due to its strong global optimization capability and simple implementation. In DE, mutant vector, which plays the role of leading individuals to explore the search space, is generated by combining a base vector and a difference vector. However, these two vectors are selected either randomly or greedily according to the conventional strategies. In this paper, we propose three different learning strategies for conventional DE, one is for selecting the base vector and the other two are for constructing the difference vector. Experimental results on six benchmark functions validate the effectiveness of the proposed strategies.