A Bio-inspired Genetic Algorithm with a Self-Organizing Genome: The RBF-Gene Model

Virginie Lefort, Carole Knibbe, Guillaume Beslon, and Joël Favrel

Abstract. Although Genetic Algorithms (GAs) are directly inspired by Darwinian principles, they use an over-simplistic model of chromosome and genotype to phenotype mapping. This simplification leads to a lack of performance, mainly because the chromosome structure directly constrains the evolution process.

In biology, the structure of the chromosome is free to evolve. The main feature permitting it is the presence of an intermediate layer (the proteins) between genotype and phenotype: Whatever the size and the locus of a gene, it is translated into a protein and all the proteins are combined to “produce” the phenotype.

Some authors, like Goldberg [1], have tried to introduce some independence between the genotype and the phenotype in GAs but none have really introduced the “protein level”. Thus, they do not really part the two levels. We propose a new model of GA introducing such an intermediate level in order to permit evolvability during and by the evolutionary process to improve convergence.

LNCS 3103, p. 406 f.

Full article in PDF

lncs@springer.de
© Springer-Verlag Berlin Heidelberg 2004