Statistically Neutral Promoter based GA for Evolution with Dynamic Fitness Functions

F. Bellas and R.J. Duro (Spain)


Genetic algorithms, artificial neural networks, variable length genotype, phenotype construction, promotergenes.


In this paper we consider the use of promoter genes and introns in the encoding of variable length artificial neural network structures for their evolution. To support these genotypes we present the adaptation of a structured genetic algorithm we have called Promoter Based Genetic Algorithm (PBGA) to contemplate the evolution of the architecture and weight values of artificial neural networks which regulates the expression of the different genes in the chromosome in a statistically neutral manner. Obviously, this leads to a non direct genotype-phenotype transformation which becomes very efficient in dynamic environments. We study some examples where the advantages of using this type of representation over traditional genetic algorithms in problems with changing fitness functions become evident.

Important Links:

Go Back