HOME   ::  Conference List   ::   Conference Paper
O'Neill, M., Ryan, C., Keijzer, M., and Cattolico, M. (2001) Crossover in Grammatical Evolution: The Search Continues. In EuroGP 2001, pages 337--347.
Bookmark:  

Full-text
   URL: http://shine.csis.ul.ie/papers/eurogp2001.ps.gz
   Cached: PDF-173K    PS-182K    PS.gz-65K   
   SAVE AS an easy-to-recall long filename:
      Filename format: author--year--title   PDF-173K    PS-65K    :: About GZip'd PS
      Filename format: author--year--title--journal|proceedings|...--pages   PDF-173K    PS-65K   

Related links
   CiteSeer: http://citeseer.ist.psu.edu/483107.html
  Web search: Google Web Search   ::   Google Scholar
  Within this site: Cited by (2)    References (17)

Paper at a Glance 

Crossover in Grammatical Evolution: The
Search Continues
Michael O'Neill 1 , Conor Ryan 1 , Maarten Keijzer 2 , and Mike Cattolico 3
1 Dept. Of Computer Science And Information Systems,
University of Limerick, Ireland.
fMichael.ONeilljConor.Ryang@ul.ie
2 Danish Hydraulic Institute, Denmark. mak@dhi.dk
3 Tiger Mountain Scientic. mike@tigerscience.com
Abstract. Grammatical Evolution is an evolutionary automatic pro- gramming algorithm that can produce code in any language, requiring as inputs a BNF grammar de nition describing the output language, and the tness function. The utility of crossover in GP systems has been hotly debated for some time, and this debate has also arisen with respect to Grammatical Evolution. This paper serves to continue an analysis of the crossover operator in Grammatical Evolution by looking at the result of turning o crossover, and by exchanging randomly generated blocks in a headless chicken-like crossover. Results show that crossover in Gram- matical Evolution is essential on the problem domains examined. The mechanism of one-point crossover in Grammatical Evolution is discussed, resulting in the discovery of some interesting properties that could yield an insight into the operator's success.
1 Introduction While crossover is generally accepted as an explorative operator in string based G.A.s [4] the bene t or otherwise of employing crossover in tree based Genetic Programming hasn't been fully established. Work such as [2] went as far as to dismiss GP as a evolutionary search method due to its use of trees, while [1] presented results which suggested that crossover in GP provides little bene t over randomly generating subtrees. Langdon and Francone et. al. have also addressed this issue, the former on tree based GP and the latter on linear structures, and have both introduced new crossover operators in an attempt to improve exploration [3] [7]. These exploit the idea of a homologous crossover that draws inspiration from the molecular ...
BibTex
@inproceedings{oneill01crossoverIn,
  author={Michael O'Neill and Conor Ryan and Maarten Keijzer and Mike Cattolico},
  title={Crossover in Grammatical Evolution: The Search Continues},
  year={2001},
  pages={337-347},
  booktitle={EuroGP 2001},
  url={http://www.isrl.uiuc.edu/~amag/langev/paper/oneill01crossoverIn.html}
}

 HOME   ::  Conference List   ::   Conference Paper Comments to: junwang4 you-know-at gmail.com Last update: 2/2/08