Randomization of individuals selection in differential evolution

Abstract

This research deals with the hybridization of two computational intelligence fields, which are the chaos theory and evolutionary algorithms. Experiments are focused on the extensive investigation on the different randomization schemes for selection of individuals in differential evolution algorithm (DE). This research is focused on the hypothesis whether the different distribution of different pseudo-random numbers or the similar distribution additionally enhanced with hidden complex chaotic dynamics providing the unique sequencing are more beneficial to the heuristic performance. This paper investigates the utilization of the two-dimensional discrete chaotic systems, which are Burgers and Lozi maps, as the chaotic pseudo-random number generators (CPRNGs) embedded into the DE. Through the utilization of either chaotic systems or equal identified pseudo-random number distribution, it is possible to entirely keep or remove the hidden complex chaotic dynamics from the generated pseudo random data series. This research utilizes set of 4 selected simple benchmark functions, and five different randomizations schemes; further results are compared against canonical DE.