Genetic programming operators applied to genetic algorithms

Like other learning paradigms, the performance of the genetic algorithms (GAs) is dependent on the parameter choice, on the problem representation, and on the fitness landscape. Accordingly, a GA can show good or weak results even when applied on the same problem. Following this idea, the crossover operator plays an important role, and its study is the object of the present paper. A mathematical analysis has led us to construct a new form of crossover operator inspired from genetic programming (GP) that we have already applied in field of information retrieval. In this paper we extend the previous results and compare the new operator with several known crossover operators under various experimental conditions.