- author = "David Andre and Forrest H {Bennett III} and John R. Koza",
- title = "Evolution of Intricate Long-Distance Communication Signals in Cellular Automata using Genetic Programming",
- booktitle = "Artificial Life V: Proceedings of the Fifth International Workshop on the Synthesis and Simulation of Living Systems",
- year = "1996",
- volume = "1",
- address = "Nara, Japan",
- publisher_address = "Cambridge, MA, USA",
- month = "16--18 " # may,
- publisher = "MIT Press",
- keywords = "genetic algorithms, genetic programming",
-
URL = "
http://www.genetic-programming.com/jkpdf/alife1996gkl.pdf",
- size = "10 pages",
-
abstract = "A cellular automata rule for the majority
classification task was evolved using genetic
programming with automatically defined functions. The
genetically evolved rule has an accuracy of 82.326
percent. This level of accuracy exceeds that of the
Gacs-Kurdyumov-Levin (GKL) rule, all other known
human-written rules, and all other rules produced by
known previous automated approaches.
Our genetically evolved rule is qualitatively different from other rules in that it uses a fine-grained internal representation of density information; it employs a large number of different domains and particles; and it uses an intricate set of signals for communicating information over large distances in time and space.",
-
notes = "Alife-5 A longer version of this paper will be
presented at the GP-96 conference. GP gets best
solution to GKL problem
{"}The population size used to evolve the current world's record for the GKL majority classification 1-dimensionall 2-sate 7-neighbor cellular authomata problem was 51,200.
I believe Melanie Mitchell at the Santa Fe Institute has been doing continuing additional work on using GAs to evolve CA rules for various other problems.{"}",
