Abstract
The creation of the accuracy-based classifier system XCS (Wilson, 1995) can be considered a milestone in classifier system research. XCS addresses the general LCS challenges in a very distributed fashion. The problem of generalization is approached by niche reproduction in conjunction with panmictic (population-wide) deletion. The problem of strong overgenerals is solved by deriving classifier fitness from the estimated accuracy of reward predictions instead of from the reward predictions themselves. In effect, XCS is designed to not only evolve a representation of the best solution for all possible problem instances but rather to evolve a complete and accurate payoff map of all possible solutions for all possible problem instances.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
Author information
Authors and Affiliations
Rights and permissions
Copyright information
© 2006 Springer
About this chapter
Cite this chapter
Butz, M.V. (2006). The XCS Classifier System. In: Rule-Based Evolutionary Online Learning Systems. Studies in Fuzziness and Soft Computing, vol 191. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-31231-5_4
Download citation
DOI: https://doi.org/10.1007/3-540-31231-5_4
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-25379-2
Online ISBN: 978-3-540-31231-4
eBook Packages: EngineeringEngineering (R0)