Applied Evolutionary Algorithms in Java by Robert Ghanea-Hercock

Author:Robert Ghanea-Hercock
Language: eng
Format: epub
Publisher: Springer-Verlag Wien 2012
Published: 2015-03-26T16:00:00+00:00

6.3 Speciation and Distributed EA Methods

Because the solution space of most optimisation problems contains multiple fitness peaks, any EA can experience convergence to a suboptimal solution. A number of methods have been proposed to address this, which rely on dividing the evolving population into smaller subpopulations. The basic idea is to create distinct species of individuals, which then perform a parallel search of each potential local optima. In order to achieve this we need a mechanism to maintain several subpopulations or “niches”. This therefore requires a restriction on which individuals are allowed to recombine with each other (i.e., no free sex!). One mating restriction scheme was developed by Deb (1989), based on measures of the phenotypic and genotypic distance between individuals. Hence when a possible mating pair using genotypic separation is selected, the Hamming distance is calculated, (or Euclidean distance in the phenotypic case). If the separation value is less than some threshold value sigma, the individuals are allowed to undergo crossover.

An alternative species method simply appends a sequence of tags to each individual which identify each subgroup (Spears, 1994). Tagging has the advantage that distance metrics do not need to be computed, which therefore saves processing time.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.