The effect of extensive use of the mutation operator on generalization in genetic programming using sparse data sets

  • Wolfgang Banzhaf
  • Frank D. Francone
  • Peter Nordin
Modifications and Extensions of Evolutionary Algorithms Genetic Operators and Problem Representation
Part of the Lecture Notes in Computer Science book series (LNCS, volume 1141)


Ordinarily, Genetic Programming uses little or no mutation. Crossover is the predominant operator. This study tests the effect of a very aggressive use of the mutation operator on the generalization performance of our Compiling Genetic Programming System (‘CPGS’). We ran our tests on two benchmark classification problems on very sparse training sets. In all, we performed 240 complete runs of population 3000 for each of the problems, varying mutation rate between 5% and 80%. We found that increasing the mutation rate can significantly improve the generalization capabilities of GP. The mechanism by which mutation affects the generalization capability of GP is not entirely clear. What is clear is that changing the balance between mutation and crossover effects the course of GP training substantially — for example, increasing mutation greatly extends the number of generations for which the GP system can train before the population converges.


Mutation Rate Genetic Program Mutation Operator High Mutation Rate Generalization Capability 
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.
    T. Bäck: Optimal Mutation Rates in Genetic Search. In: S. Forrest (Ed.): Proc. 5th Int. Conference on Genetic Algorithms, ICGA-93. San Mateo: Morgan Kaufmann 1993Google Scholar
  2. 2.
    D. Fogel, L. Stayton: On the Effectiveness of Crossover in Simulated Evolutionary Optimization. Biosystems 32 171–182 (1994)CrossRefPubMedGoogle Scholar
  3. 3.
    D. Goldberg: Genetic Algorithms in Search Optimization & Learning, Reading: Addison-Wesley 1989Google Scholar
  4. 4.
    J.R. Koza: Genetic Programming. Cambridge (USA): MIT Press 1992Google Scholar
  5. 5.
    K. Lang: Hill Climbing Beats Genetic Search on a Boolean Circuit Synthesis Problem of Koza's. In: A. Prieditis, S. Russell (Eds.): Proc. 12th Int. Conference on Machine Learning. San Mateo: Morgan Kaufmann 1995Google Scholar
  6. 6.
    T. Masters: Advanced Algorithms for Neural Networks. New York: Wiley 1995Google Scholar
  7. 7.
    J.P. Nordin, J.P.: A Compiling Genetic Programming System that Directly Manipulates the Machine Code. In: K. Kinnear, Jr. (ed.). Advances in Genetic Programming. Cambridge MA: MIT Press 1994Google Scholar
  8. 8.
    J.P. Nordin, W. Banzhaf: Complexity Compression and Evolution. In:L. Eshelman (Ed.): Proc. 6th Int. Conference on Genetic Algorithms, ICGA-95. San Mateo: Morgan Kaufmann 1995Google Scholar
  9. 9.
    J.P. Nordin, W. Banzhaf: Evolving Turing Complete Programs for a Register Machine with Self Modifying Code. In: L. Eshelman (Ed.): Proc. 6th Int. Conference on Genetic Algorithms, ICGA-95. San Mateo: Morgan Kaufmann 1995Google Scholar
  10. 10.
    J.P. Nordin, F.D. Francone, W. Banzhaf: Explicitly Defined Introns and Destructive Crossover in Genetic Programming. In: K. Kinnear, Jr., P. Angeline (eds.): Advances in Genetic Programming 2. Cambridge MA: MIT Press 1996, in pressGoogle Scholar
  11. 11.
    I. Rechenberg: Evolutionsstrategie 94. Stuttgart: Holzmann-Froboog 1994 (2nd. ed.)Google Scholar
  12. 12.
    H.P. Schwefel: Evolution and Optimum Seeking. New York: Wiley 1995 (2nd. ed.)Google Scholar
  13. 13.
    The ELENA Partners: C. Jutten, Project Coordinator: ESPRIT Basic Research Project Number 6891, Document Number R3-B1-P. Available via ftp at either or at, 1995Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1996

Authors and Affiliations

  • Wolfgang Banzhaf
    • 1
  • Frank D. Francone
    • 2
  • Peter Nordin
    • 1
  1. 1.Department of Computer ScienceDortmund UniversityDortmundGermany
  2. 2.Law Office of Frank D. FranconeLos AngelesUSA

Personalised recommendations