A Pattern Based Evolutionary Approach to Prediction Computation in XCSF

  • Ali Hamzeh
  • Adel Rahmani
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 4221)


XCSF is a new version of XCS with the ability of computing the environmental rewards. In the early versions of XCSF, this computation was done by approximating the coefficients of the associated polynomial reward functions. However, recent researches show that this approximation method suffers from some significant drawbacks such as input range dependency. In this paper, we propose a new method to approximate the reward functions using Genetic Algorithms. Our proposed method uses a new representation scheme for chromosomes and some newly introduced pattern based operators which are adapted for function approximation.


Benchmark Problem Reward Function Mean Absolute Error Input Range Good Chromosome 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    Wilson, S.W.: Classifiers that Approximate Functions. Journal of Natural Computating 1, 211–234 (2002)MATHCrossRefGoogle Scholar
  2. 2.
    Wilson, S.W.: Classifier Fitness Based on Accuracy. Evolutionary Computation 3(2), 149–175 (1995)CrossRefGoogle Scholar
  3. 3.
    Lanzi, P.L., Loiacono D., Wilson, S.W., Goldberg, D.E.: XCS with Computable Prediction for the Learning of Boolean Functions. Technical Report 2005007, Illinois Genetic Algorithms Laboratory (2005)Google Scholar
  4. 4.
    Wilson, S.W.: Function Approximation with a Classifier System. In: Proceedings of the GECCO-2001, San Francisco, California, USA, pp. 974–981 (2001)Google Scholar
  5. 5.
    Hamzeh, A., Rahmani, A.: An Evolutionary Function Approximation Approach to Compute Prediction in XCSF. In: Proceedings of 16th European Conference of Machine Learning, Porto, Portugal (2005)Google Scholar
  6. 6.
    Hamzeh, A., Rahmani, A.: Extending XCSFG beyond Linear Approximation. In: The proceedings of the IEEE world congress on Evolutionary Computation (CEC 2006), Vancouver, Canada (to appear, 2006)Google Scholar
  7. 7.
    Lanzi, P.L., Loiacono, D., Wilson, S.W., Goldberg, D.E.: Generalization in the XCSF Classifier System: Analysis, Improvement, and Extension. Technical Report 2005012, Illinois Genetic Algorithms Laboratory (2005)Google Scholar
  8. 8.
    Lanzi, P.L., Loiacono, D., Wilson, S.W., Goldberg, D.E.: XCS with Computable Prediction in Multistep Environments. Technical Report 2005008, Illinois Genetic Algorithms Laboratory (2005) Google Scholar
  9. 9.
    Lanzi, P.L., Loiacono D., Wilson S.W., Goldberg, D.E.: Extending XCSF beyond Linear Approximation. Illinois Genetic Algorithms Laboratory University of Illinois at Urbana-Champaign (2005) Google Scholar
  10. 10.
    Hamzeh, A., Rahmani, A.: A New Selection Method for Genetic Algorithms based on Genotypic Information of the Population. In: Proceedings of 10th Annual Conference of Computer Society of Iran, Tehran (2005)Google Scholar
  11. 11.
    Hamzeh, A., Rahmani, A.: An Adaptive Pattern-Based Uniform Crossover for Genetic Algorithms. In: Proceedings of 9th Iranian Computer Society, Computer Conference, Tehran, Iran (2004)Google Scholar
  12. 12.
    Holland, J.H.: Adaptation in Natural and Artificial Systems. University of Michigan Press, Ann Arbor (1975), Republished by the MIT press (1992)Google Scholar
  13. 13.
    Kanji, G.: 100 Statistical Tests. SAGE Publications, Thousand Oaks (1994)Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2006

Authors and Affiliations

  • Ali Hamzeh
    • 1
  • Adel Rahmani
    • 1
  1. 1.Department of Computer EngineeringIran University of Science and TechnologyTehranIran

Personalised recommendations