Probabilistic selection in cellular genetic algorithm

Abstract

In this paper, we introduce a new selection operator, namely, a Probabilistic Selection operator which allows us to control the selection pressure in cellular genetic algorithms through reducing the effective neighborhood radius. One advantage for having probabilistic selection is that, once we have our probability density function in hand, we can apply it on any type of neighborhoods. The main idea of this selection operator is that, as we move away from the center of the neighborhood, the probability of an individual is selected as parent will get lower. We will first discuss the general idea of how we implement this selection algorithm into the cellular genetic algorithm. We then conduct experiments on several combinatorial optimization benchmark problems in order to show its performance. Finally, we will briefly discuss about our further work on self-adaptive capability.