GEP Book   Home   News   Author   Q&A   Tutorials   Downloads   GEP Biblio   Contacts   Visit Gepsoft

C. FERREIRA In J. M. Benitez, O. Cordon, F. Hoffmann, and R. Roy, eds., Advances in Soft Computing: Engineering Design and Manufacturing, pages 257-266, Springer-Verlag, 2003. Function Finding and the Creation of Numerical Constants in Gene Expression Programming Genetic Programming   As simple replicators, the ramified structures of GP are tied up in their own complexity: on the one hand, bigger, more complex structures are more difficult to handle and, on the other, the introduction of genetic variation can only be done at the tree level and, therefore, must be done carefully so that valid structures are created. A special kind of tree crossover is practically the only source of genetic variation used in GP for it allows the exchange of sub-trees and, therefore, always produces valid structures. Indeed, the implementation of high-performing operators, like the equivalent of natural point mutation, is unproductive as most mutations would have resulted in syntactically invalid structures. Understandingly, the other genetic operators described by Koza (1992) – mutation and permutation – also operate at the tree level. Home | Contents | Previous | Next