USENET News comp.ai.alife

Säie: Interpreting Kitano

Edellinen säie: C++ alife design principles?
[Muut säikeet] [Muut uutisryhmät]

From: Marko Gronroos <magi AT iki PISTE fi>
Newsgroups: comp.ai.alife,comp.ai.genetic
Subject: Interpreting Kitano
Date: 13 Jan 1998 17:14:16 +0200

I'd like to make sure that I have understood something correctly in
Hiroaki Kitano's "classic" 1990 paper "Designing Neural Networks Using
Genetic Algorithms with Graph Generation System". Page 467 says:

"For the variable region, symbols are generated for each position of
the chromosome in range between "A" to "p." The first position of the
chromosome, however, is fixed to be "S" (an initial graph), so that
the rule for the first cell division can always be created."

The question is: what does "A" to "p" mean? Does it mean:

1) A-Z + a-p
            - rules like "Q->BfTd" are possible
2) A-P + a-p, where "A" is distinct from "a"
            - rules like "D->fEnO" are possible
3) A-P, where A=a, etc. at the last rewriting step
            - only rules like "G->AHED" or "G->ogeh" are possible

My choice was 3, since I suppose it generates less "broken" rules
(only symbols of the same type (terminal or nonterminal) on right
side). Any other opinions?

-- Marko Grönroos, magi AT iki PISTE fi (http://www.iki.fi/~magi/)

Edellinen säie: C++ alife design principles?
[Muut säikeet] [Muut uutisryhmät]