MOVEMENT OF FLOCKED SUBPOPULATIONS IN DISTRIBUTED GENETIC PROGRAMMING

Abstract

The rules of swarming intelligence can be applied to govern migration in the distributed genetic prog-ramming (DGP) algorithm, but they require modifications. Initial rules are taken from the flocking algorithm and adapted for DGP. As the rule for the alignment of direction is completely discarded and the remaining rules operate on implicit data of subpopulation locations, the resulting joint search technique must be reevaluated. This article presents the pragmatic coupling of flocking and DGP algorithms. The experiment of visualizing the movement of DGP sub-populations through the search space provides a graphic overview of the behavior of DGP subpopulations. The results confirm a typical influence of the modified flocking rules to the flockmates represented by the subpopulations.