Selection functions enable Evolutionary Algorithms (EAs) to apply selection pressure to a population of individuals, by regulating the probability that an individual's genes survive, typically based on fitness. Various conventional fitness based selection methods exist, each providing a unique relationship between the fitnesses of individuals in a population and their chances of selection. However, the full space of selection algorithms is only limited by max algorithm size, and each possible selection algorithm is optimal for some EA configuration applied to a particular problem class. Therefore, improved performance may be expected by tuning an EA's selection algorithm to the problem at hand, rather than employing a conventional selection method. The objective of this paper is to investigate the extent to which performance can be improved by tuning selection algorithms, employing a Hyper-heuristic to explore the space of search algorithms which encode the relationships between the fitnesses of individuals and their probability of selection. We show the improved performance obtained versus conventional selection functions on fixed instances from a benchmark problem class, including separate testing instances to show generalization of the improved performance.
- Genetic Programming,
Available at: http://works.bepress.com/daniel-tauritz/73/