Felix Streichert, Gunnar Stein, Holger Ulmer, and Andreas Zell

A Clustering Based Niching Method for Evolutionary Algorithms

Proceedings of the 2003 Genetic and Evolutionary Computation Conference (GECCO-2003)

Abstract

We propose a new Niching Method for Evolutionary Algorithms which is able to identify and track global and local optima in a multimodal search space. To prevent the premature convergence on only few optima we replace the global selection strategy of a standard Evolutionary Algorithm with a multi-population strategy with local selection. A clustering algorithm is used to identify local concentrations of individuals and separate them into demes which do not directly compete with other demes. With additional management of deme size, collisions between demes, convergence behavior and active exploration we are able to use a mixture of the parallel EA strategy and sequential exploration via reinitialization of new individuals from surplus resources.

Download

[pdf], [ps]

BibTeX 

@inproceedings{streichert03evolving,
  author 		= "Felix Streichert and Gunnar Stein and Holger Ulmer and Andreas Zell ",
  title 		= "A Clustering Based Niching Method for Evolutionary Algorithms",
  booktitle 		= "Genetic and Evolutionary Computation -- GECCO-2003",
  editor 		= "E. Cant{\'u}-Paz and J. A. Foster and K. Deb and D. Davis and R. Roy and U.-M. O'Reilly and H.-G. Beyer and R. Standish and G. Kendall and S. Wilson and M. Harman and J. Wegener and D. Dasgupta and M. A. Potter and A. C. Schultz and K. Dowsland and N. Jonoska and J. Miller",
  year 			= "2003",
  pages 		= "644--645",
  address 		= "Chicago",
  publisher_address 	= "Berlin",
  month	 		= "12-16 July",
  volume 		= "2723",
  series 		= "LNCS",
  ISBN 			= {3-540-40602-6},
  publisher 		= "Springer-Verlag",
  keywords 		= {Evolution Strategies/Evolutionary Programming},
  notes	 		=  "GECCO-2003. A joint meeting of the twelfth International Conference on Genetic Algorithms (ICGA-2003) and the eights Annual Genetic Programming Conference (GP-2003)"   
  abstract 		= {We propose a new Niching Method for Evolutionary Algorithms which is able to identify and track 
global and local optima in a multimodal search space. To prevent the premature convergence on only 
few optima we replace the global selection strategy of a standard Evolutionary Algorithm with a 
multi-population strategy with local selection. A clustering algorithm is used to identify local 
concentrations of individuals and separate them into demes which do not directly compete with other 
demes. With additional management of deme size, collisions between demes, convergence behavior and 
active exploration we are able to use a mixture of the parallel EA strategy and sequential exploration 
via reinitialization of new individuals from surplus resources.},
  url			= "http://www.ra.cs.uni-tuebingen.de/mitarb/streiche/publications/streichert03aclustering.html"
}