An Oracle Based Meta-learner for Function Decomposition

  • R. Syama Sundar Yadav
  • Deepak Khemani
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 3192)

Abstract

Function decomposition is a machine learning algorithm that induces the target concept in the form of a hierarchy of intermediate concepts and their definitions. Though it is effective in discovering the concept structure hidden in the training data, it suffers much from under sampling. In this paper, we propose an oracle based meta learning method that generates new examples with the help of a bagged ensemble to induce accurate classifiers when the training data sets are small. Here the values of new examples to be generated and the number of such examples required are automatically determined by the algorithm. Previous work in this area deals with the generation of fixed number of random examples irrespective of the size of the training set’s attribute space. Experimental analysis on different sized data sets shows that our algorithm significantly improves accuracy of function decomposition and is superior to existing meta-learning method.

Keywords

Machine learning Function decomposition Bagging Meta-learning 

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    Asenhurst, L.: The decomposition of switching functions. Technical Report, Bell laboratories. pp. 541–602 (1952)Google Scholar
  2. 2.
    Breiman, L.: Bagging predictors. In: Machine Learning, pp. 123–140 (1996)Google Scholar
  3. 3.
    Domingos, P.: Knowledge acquisition from examples via multiple models. In: Proc. of the Fourteenth International Conference on Machine Learning, pp. 148–156 (1997)Google Scholar
  4. 4.
    Luba, T.: Decomposition of multiple-valued functions. In: 25th Intl. Symposium on Multiple-Valued Logic, pp. 255–261 (1995)Google Scholar
  5. 5.
    Michie, D.: Problem decomposition and the learning of skill. In: Lavrač, N., Wrobel, S. (eds.) ECML 1995. LNCS, vol. 912, pp. 17–31. Springer, Heidelberg (1995)Google Scholar
  6. 6.
    Michalski, R.S.: Understanding the nature of learning: issues and research directions. In: Machine Learning: An Artificial Intelligence Approach, pp. 3–25. Morgan Kaufmann, San Francisco (1986)Google Scholar
  7. 7.
    Perkowski, M.A., et al.: Unified approach to functional decomposition of switching functions. Technical report, Warsaw University of Technology and Eindhoven University of Technology (1995)Google Scholar
  8. 8.
    Pfahringer, B.: Controlling constructive induction in CiPF. In: Bergadano, F., De Raedt, L. (eds.) ECML 1994. LNCS, vol. 784, pp. 242–256. Springer, Heidelberg (1994)Google Scholar
  9. 9.
    Shapiro, A.D.: Structured induction in expert systems. Turing Institute Press in association with Addison-Wesley Publishing Company (1987)Google Scholar
  10. 10.
    Merz, C.J., Murphy. P.M., Aha, D.W.: UCI repository of machine learning databases. Department of Information and Computer Science, University of California at Irvine, Irvine, CA (1997) Google Scholar
  11. 11.
    Zupan, B., Bohanec, M., Demsar, J., Bratko, I.: Machine Learning by function decomposition. In: Proc. of the Fourteenth International Conference on Machine Learning, pp. 421–429 (1997)Google Scholar
  12. 12.
    Zupan, B., Bohanec, M., Demsar, J., Bratko, I.: Learning by discovering concept hierarchies. Artificial Intelligence 109, 211–242 (1999)MATHCrossRefMathSciNetGoogle Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2004

Authors and Affiliations

  • R. Syama Sundar Yadav
    • 1
  • Deepak Khemani
    • 1
  1. 1.Dept. of Computer Science and EngineeringIndian Institute of Technology MadrasChennaiIndia

Personalised recommendations