Determining the Impact of Spatial Networks and Elitism on\\Coevolutionary Genetic Algorithms
Morgan McLaughlin · The Atrium (University of Guelph) · 2014
It has been demonstrated that spatial networks can improve the behaviour of coevolutionary genetic algorithms (GAs) when optimizing specific problems. We aim to determine whether this result generalizes for more basic problems by applying it to the one-max problem and some variations of that problem, on both cooperative and competitive coevolutionary systems. We confirm that spatial structures do help coevolutionary systems. Finally, we note that adding elitism to the spatial system, a simple but novel idea in the spatial GA world, increases the speed of convergence by an order of magnitude.