Advertisement

Machine Learning

, Volume 5, Issue 1, pp 71–99 | Cite as

Boolean feature discovery in empirical learning

  • Giulia Bagallo
  • David Haussler
Article

Abstract

We investigate the problem of learning Boolean functions with a short DNF representation using decision trees as a concept description language. Unfortunately, Boolean concepts with a short description may not have a small decision tree representation when the tests at the nodes are limited to the primitive attributes. This representational shortcoming may be overcome by using Boolean features at the decision nodes. We present two new methods that adaptively introduce relevant features while learning a decision tree from examples. We show empirically that these methods outperform a standard decision tree algorithm for learning small random DNF functions when the examples are drawn at random from the uniform distribution.

Keywords

concept learning dynamic bias DNF functions decision trees decision lists 

References

  1. Beach, L.R. (1964). Cue probabilism and inference behavior.Psychological Monographs, Whole, 582.Google Scholar
  2. BlumerA., EhrenfeuchtA., HausslerD. & WarmuthM. (1987). Occam's razor.Information Processing Letters.24:377–380.Google Scholar
  3. Breiman, L., Friedman, J.H., Olsen, R.A., & Stone, C.J. (1984).Classification and Regression Trees. Wadsworth Statistic/Probability Series.Google Scholar
  4. Carbonell, J., Michalski, R., & Mitchell, T. (1983). An overview of machine learning. InMachine Learning: An Artificial Intelligence Approach, 3–24 Morgan Kaufmann.Google Scholar
  5. ClarkP. & NiblettT. (1989). The CN2 induction algorithm.Machine Learning,3:251–283.Google Scholar
  6. Flann, N. & Dietterich, T. (1986). Selecting appropriate representation for learning from examples.Proceedings of AAAI-86, (pp. 460–466). Morgan Kaufmann.Google Scholar
  7. Haussler,D. (1988). Quantifying inductive bias: AI learning algorithms and Valiant's learning framework.Artificial Intelligence, 36: 177–221.Google Scholar
  8. Kearns, M., Li, M. & Valiant, L. (1987). Recent results on Boolean concept learning.Proceedings of 4th International Workshop on Machine Learning, (pp. 337–352). Morgan Kaufmann.Google Scholar
  9. Matheus, C. (1989).Feature Construction: An Analytic Framework and An Application to Decision Trees. Ph.D. thesis, University of Illinois, December 1989. In preparation.Google Scholar
  10. Muggleton, S. (1987). Duce, an oracle-based approach to constructive induction.Proceedings of IJCAI-87, (pp. 287–292). Morgan Kaufmann.Google Scholar
  11. Pagallo, G. and Haussler, D. (1989). A greedy method for learning μDNF functions under the uniform distribution. (Technical Report UCSC-CRL-89–12). Santa Cruz: Dept. of Computer and Information Sciences, University of California at Santa Cruz.Google Scholar
  12. Pagallo, G. (1989). Learning DNF by decision trees.Proceedings of IJCAI-89, (pp. 639–644). Morgan Kaufmann.Google Scholar
  13. Quinlan, J.R. (1986). Induction of decision trees.Machine Learning, 1:81–106.Google Scholar
  14. Quinlan, J.R. (1987a). Generating production rules from decision trees.Proceedings of IJCAI-87, 1: (pp. 304–307). Morgan Kaufmann.Google Scholar
  15. Quinlan, J.R. (1987b). Simplifying decision trees.International Journal of Man-machine Studies, 27:221–234.Google Scholar
  16. Quinlan, J.R. (1988). An empirical comparison of genetic and decision tree classifiers.Proceedings of the 5th International Conference on Machine Learning (pp. 135–141). Morgan Kaufmann.Google Scholar
  17. Rivest, R. (1987). Learning decision lists.Machine Learning, 2:229–246.Google Scholar
  18. Schlimmer, J.C. (1986). Concept acquisition through representational adjustment.Machine Learning, 1:81–106.Google Scholar
  19. Utgoff, P. and Mitchell, T.M. (1982). Acquisition of appropriate bias for inductive concept learning.Proceedings of AAAI-82, (pp. 414–417). Morgan Kaufmann.Google Scholar
  20. Valiant, L.G. (1984). A theory of the learnable.Communications of ACM,27:1134–1142.Google Scholar
  21. Valiant, L.G. (1985). Learning disjunctions of conjunctions.Proceedings of IJCAI-85, (pp. 560–566). Morgan Kaufmann.Google Scholar
  22. Vapnik, V.N. (1987).Estimation of Dependeneies Based on Empirical Data. Springer Verlag.Google Scholar
  23. Wilson, S.W. (1987). Classifler systems and the animat problem.Machine Learning, 2:199–228, 1987.Google Scholar

Copyright information

© Kluwer Academic Publishers 1990

Authors and Affiliations

  • Giulia Bagallo
    • 1
  • David Haussler
    • 1
  1. 1.Department of Computer and Information SciencesUniversity of CaliforniaSanta Cruz

Personalised recommendations