Robust constructive induction

  • Bernhard Pfahringer
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 861)


We describe how CiPF 2.0, a propositional constructive learner, can cope with both noise and representation mismatch in training examples simultaneously. CiPF 2.0 abilities stem from coupling the robust selective learner C4.5 with a sophisticated constructive induction component. An important new constructive operator incorporated into CiPF 2.0 is the simplified Kramer operator abstracting combinations of two attributes into a single new boolean attribute. The so-called Minimum Description Length (MDL) principle acts as a powerful control heuristic guiding search in the representation space through the abundance of opportunities for constructively adding new attributes. Claims are confirmed empirically by experiments in two artificial domains.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    Angluin D., Laird P.: Learning from Noisy Examples, Machine Learning, 2(4), 343–370, 1987.Google Scholar
  2. 2.
    Bloedorn E., Wnek J., Michalski R.S.: Multistrategy Constructive Induction: AQ17-MCI, in Michalski R.S. and Tecuci G.(eds.), Proceedings of the Second International Workshop on Multistrategy Learning (MSL-93), Harpers Ferry, W.VA., pp.188–206, 1993.Google Scholar
  3. 3.
    Breiman L., Friedman J.H., Olshen R.A., Stone C.J.: Classification and Regression Trees, Wadsworth International Group, Belmont, CA, The Wadsworth Statistics/Probability Series, 1984.Google Scholar
  4. 4.
    Dietterich T.G., Michalski R.S.: Inductive Learning of Structural Descriptions: Evaluation Criteria and Comparative Review of Selected Methods, Artificial Intelligence, 16(3), 257–294, 1981.Google Scholar
  5. 5.
    Fürnkranz J.: A numerical analysis of the KRK domain. Working Note, 1993. Available upon request.Google Scholar
  6. 6.
    Fürnkranz J., Widmer G.: Incremental Reduced Error Pruning. Proceedings of the Eleventh International Conference on Machine Learning (ML-94), New Brunswick, N.J., 1994.Google Scholar
  7. 7.
    Kramer S.: CN2-MCI: Ein zweistufiges Verfahren für konstruktive Induktion, Master's thesis in preparation, Vienna, 1993.Google Scholar
  8. 8.
    Matheus C.J., Rendell L.A.: Constructive Induction On Decision Trees, in Proceedings of the Eleventh International Joint Conference on Artificial Intelligence (IJCAI-89), Morgan Kaufmann, Los Altos, CA, 645–650, 1989.Google Scholar
  9. 9.
    Matheus C.J.: Adding Domain Knowledge to SBL Through Feature Construction, in Proceedings of the Eighth National Conference on Artificial Intelligence (AAAI-90), AAAI Press/MIT Press, Menlo Park, CA, pp.803–808, 1990.Google Scholar
  10. 10.
    Pfahringer B.: Controlling Constructive Induction in CiPF: An MDL Approach, in Proceedings of the European Conference on Machine Learning (ECML94), 1994.Google Scholar
  11. 11.
    Quinlan J.R., Cameron-Jones R.M.: FOIL: A Midterm Report, in Brazdil P.B.(ed.), Machine Learning: ECML-93, Springer, Berlin, pp.3–20, 1993.Google Scholar
  12. 12.
    Quinlan J.R.: C4.5: Programs for Machine Learning, Morgan Kaufmann, San Mateo, CA, 1993.Google Scholar
  13. 13.
    Rissanen J.: Modeling by Shortest Data Description, in Automatica, 14:465–471, 1978.Google Scholar
  14. 14.
    Thrun S.B., The MONK's Problems: A Performance Comparison of Different Learning Algorithms, CMU Tech Report, CMU-CS-91-197, 1991.Google Scholar
  15. 15.
    Watanabe L., Rendell L.: Learning Structural Decision Trees from Examples, in Proceedings of the 12th International Conference on Artificial Intelligence, Morgan Kaufmann, San Mateo, CA, pp.770–776, 1991.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1994

Authors and Affiliations

  • Bernhard Pfahringer
    • 1
  1. 1.Austrian Research Institute for Artificial IntelligenceViennaAustria

Personalised recommendations