Tuesday, February 1, 2011

The Name RGEP

I chose the name "Robust Gene Expression Programming" for my thesis because my professor once called a neural network technique that made few assumptions about its problem space "robust". I liked the idea, and I incorporated it into my thesis, giving a better name to my variation of GEP then the original BPGEP (as it was originally a union of Binary and Prefix Gene Expression Programming).


While going through some Developmental Genetic Programming papers (DGP is related to GPM, which is a close relative to GEP and therefore relevant to my thesis) I found a reference to a paper about natural genetics in which the author refers to the RNA "secondary structures" as being robust because there are many mappings of genetic code to each RNA structure. It turns out that, as far as I know, RGEP is the first GEP method to map the genotype of symbol lists redundantly into the phenotype of expression trees. This means that among GEP techniques one of the main distinguishing factors is that it is robust in exactly this sense! I didn't at all intend for the name to make sense on this level, but now I think it is the perfect description of what I'm doing.
How odd.

No comments:

Post a Comment