An explicit exploration strategy for evolutionary algorithms

Bio-inspired algorithms are a trending topic in the field of optimization. In general, evolutionary algorithms start their search process by generating a random population. Although there are some works in the literature that propose a mechanism to generate the initial population of a particular algorithm, there is no general approach to improve exploration in evolutionary algorithms. In this work, a versatile strategy is proposed for improving the exploration of any evolutionary algorithm. An exhaustive set of experiments was designed by applying 4 state-of-the-art algorithms to a well-known benchmark of continuous functions. A comparison based on statistical tests was made to demonstrate the performance of the proposed policy. The results indicate that the explicit exploration strategy is effective.