Can Genome Information be used to Guide Evolutionary Search?

• Main Author: Wold, Håkon Hjelde
• Format: Others
• Language: English
• Published: Norges teknisk-naturvitenskapelige universitet, Institutt for datateknikk og informasjonsvitenskap 2013
• Online Access: http://urn.kb.se/resolve?urn=urn:nbn:no:ntnu:diva-23599

Uniform cellular automata have been evolved as phenotypes from zygotes using an extensive rule table as the genotype. This is used to simulate complex systems which could impact future hardware development and programming. This work falls within the field of Evolutionary and Developmental Systems (EvoDevo). Genome parameters are used in a genetic algorithm to try to reduce the number of generations needed to find a genome with a given complexity. Lambda parameter has been used inside the fitness function and produced promising results. Lambda has also been used to discard genomes before they are developed with poor results. Transition parameters are shown to be similar to lambda in predicting the trajectory length of a developing phenotype, but have yet to produce the same results. The genome usage has been used to control mutation with good results. The results of the work have provided more insight into how genome parameters work and what to not do when using them.