Functional space approach for evolutionary algorithms
Résumé
Artificial life makes an intensive use of genetic algorithms in order to
study the different evolutionary phenomenon, such as speciation or
co-evolution. However, artificial evolution models, as they are commonly
used in computer science, cannot be considered as good models for biological
genetic/evolutionary processes. They suffer from huge simplifications and,
thus, are not useful to study phenomenon as fundamental as evolution of
non-coding sequences, gene tranfers, pseudo-genes, gene regulation, ... In
fact, a simple notion like that of non-coding region has no equivalence in
artificial genetic algorithms.