Oracle Coached Decision Trees and Lists

  • Ulf Johansson
  • Cecilia Sönströd
  • Tuve Löfström
Part of the Lecture Notes in Computer Science book series (LNCS, volume 6065)


This paper introduces a novel method for obtaining increased predictive performance from transparent models in situations where production input vectors are available when building the model. First, labeled training data is used to build a powerful opaque model, called an oracle. Second, the oracle is applied to production instances, generating predicted target values, which are used as labels. Finally, these newly labeled instances are utilized, in different combinations with normal training data, when inducing a transparent model. Experimental results, on 26 UCI data sets, show that the use of oracle coaches significantly improves predictive performance, compared to standard model induction. Most importantly, both accuracy and AUC results are robust over all combinations of opaque and transparent models evaluated. This study thus implies that the straightforward procedure of using a coaching oracle, which can be used with arbitrary classifiers, yields significantly better predictive performance at a low computational cost.


Decision trees Rule learning Coaching 


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    Quinlan, J.R.: C4.5: programs for machine learning. Morgan Kaufmann Publishers Inc., San Francisco (1993)Google Scholar
  2. 2.
    Breiman, L., Friedman, J., Stone, C.J., Olshen, R.A.: Classification and Regression Trees. Chapman & Hall/CRC (1984)Google Scholar
  3. 3.
    Cohen, W.W.: Fast effective rule induction. In: Proceedings of the 12th International Conference on Machine Learning, pp. 115–123. Morgan Kaufmann, San Francisco (1995)Google Scholar
  4. 4.
    Zhu, X.: Semi-supervised learning literature survey. Technical Report 1530, Computer Sciences, University of Wisconsin-Madison (2005)Google Scholar
  5. 5.
    Joachims, T.: Transductive inference for text classification using support vector machines, pp. 200–209. Morgan Kaufmann, San Francisco (1999)Google Scholar
  6. 6.
    Andrews, R., Diederich, J., Tickle, A.B.: Survey and critique of techniques for extracting rules from trained artificial neural networks. Knowl.-Based Syst. 8(6), 373–389 (1995)CrossRefGoogle Scholar
  7. 7.
    Craven, M.W., Shavlik, J.W.: Extracting tree-structured representations of trained networks. In: Advances in Neural Information Processing Systems, pp. 24–30. MIT Press, Cambridge (1996)Google Scholar
  8. 8.
    Thrun, S., Tesauro, G., Touretzky, D., Leen, T.: Extracting rules from artificial neural networks with distributed representations. In: Advances in Neural Information Processing Systems, vol. 7, pp. 505–512. MIT Press, Cambridge (1995)Google Scholar
  9. 9.
    Zhou, Z.H.: Rule extraction: using neural networks or for neural networks? J. Comput. Sci. Technol. 19(2), 249–253 (2004)CrossRefGoogle Scholar
  10. 10.
    Johansson, U., Niklasson, L.: Evolving decision trees using oracle guides. In: CIDM, pp. 238–244. IEEE, Los Alamitos (2009)Google Scholar
  11. 11.
    Johansson, U., König, R., Niklasson, L.: Rule extraction from trained neural networks using genetic programming. In: ICANN, supplementary proceedings, pp. 13–16 (2003)Google Scholar
  12. 12.
    Witten, I.H., Frank, E.: Data Mining: Practical Machine Learning Tools and Techniques, 2nd edn. Morgan Kaufmann, San Francisco (2005)zbMATHGoogle Scholar
  13. 13.
    Breiman, L.: Random forests. Machine Learning 45(1), 5–32 (2001)zbMATHCrossRefGoogle Scholar
  14. 14.
    Breiman, L.: Bagging predictors. Machine Learning 24(2), 123–140 (1996)zbMATHMathSciNetGoogle Scholar
  15. 15.
    Fawcett, T.: Using rule sets to maximize roc performance. In: IEEE International Conference on Data Mining, ICDM 2001, pp. 131–138. IEEE Computer Society, Los Alamitos (2001)CrossRefGoogle Scholar
  16. 16.
    Asuncion, A., Newman, D.J.: UCI machine learning repository (2007)Google Scholar
  17. 17.
    Demšar, J.: Statistical comparisons of classifiers over multiple data sets. J. Mach. Learn. Res. 7, 1–30 (2006)MathSciNetGoogle Scholar
  18. 18.
    Friedman, M.: The use of ranks to avoid the assumption of normality implicit in the analysis of variance. Journal of American Statistical Association 32, 675–701 (1937)CrossRefGoogle Scholar
  19. 19.
    Nemenyi, P.B.: Distribution-free multiple comparisons. PhD-thesis. Princeton University (1963)Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2010

Authors and Affiliations

  • Ulf Johansson
    • 1
  • Cecilia Sönströd
    • 1
  • Tuve Löfström
    • 1
  1. 1.CSL@BS Research Group School of Business and InformaticsUniversity of BoråsSweden

Personalised recommendations