Learning stable concepts in a changing world

  • Michael Harries
  • Kim Horn
Inducing Complex Representations
Part of the Lecture Notes in Computer Science book series (LNCS, volume 1359)


Concept drift due to hidden changes in context complicates learning in many domains including financial prediction, medical diagnosis, and network performance. Existing machine learning approaches to this problem use an incremental learning, on-line paradigm. Batch, off-line learners tend to be ineffective in domains with hidden changes in context as they assume that the training set is homogeneous.

We present an off-line method for identifying hidden context. This method uses an existing batch learner to identify likely context boundaries then performs a form of clustering called contextual clustering. The resulting data sets can then be used to produce context specific, locally stable concepts. The method is evaluated in a simple domain with hidden changes in context.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    P. Clark and T. Niblett. The CN2 induction algorithm. Machine Learning, 3:261–283, 1989.Google Scholar
  2. 2.
    Pedro Domingos. Context-sensitive feature selection for lazy learners. Artificial Intelligence Review, 11:227–253, 1997. Special issue on lazy learning, edited by David Aha.CrossRefGoogle Scholar
  3. 3.
    M. Harries and K. Horn. Detecting concept drift in financial time series prediction using symbolic machine learning. In Xin Yao, editor, Eighth Australian Joint Conference on Artificial Intelligence, pages 91–98, Singapore, 1995. World Scientific Publishing.Google Scholar
  4. 4.
    F. Kilander and C. G. Jansson. COBBIT — a control procedure for COBWEB in the presence of concept drift. In Pavel B. Brazdil, editor, European Conference on Machine Learning, pages 244–261, Berlin, 1993. Springer-Verlag.Google Scholar
  5. 5.
    M. Kubat. Floating approximation in time-varying knowledge bases. Pattern Recognition Letters, 10:223–227, 1989.MATHCrossRefGoogle Scholar
  6. 6.
    M. Kubat and G. Widmer. Adapting to drift in continuous domains. In Proceedings of the 8th European Conference on Machine Learning, pages 307–310, Berlin, 1995. Springer.Google Scholar
  7. 7.
    J. R. Quinlan. Learning logical definitions from relations. Machine Learning, 5:239–266, 1990.Google Scholar
  8. 8.
    J. R. Quinlan. C4.5: Programs for Machine Learning. Morgan Kaufmann Publishers Inc., San Mateo, California, 1993.Google Scholar
  9. 9.
    M. Salganicoff. Density adaptive learning and forgetting. In Machine Learning: Proceedings of the Tenth International Conference, pages 276–283, San Mateo, California, 1993. Morgan Kaufmann Publishers.Google Scholar
  10. 10.
    Jeffory Schlimmer and Richard Granger, Jr. Incremental learning from noisy data. Machine Learning, 1(3):317–354, 1986.Google Scholar
  11. 11.
    G. Widmer. Recognition and exploitation of contextual clues via incremental meta-learning. Technical Report oefai-96-O1, Austrian Research Institute for Artificial Intelligence, 1996.Google Scholar
  12. 12.
    G. Widmer and M. Kubat. Effective learning in dynamic environments by explicit concept tracking. In Pavel B. Brazdil, editor, European Conference on Machine Learning, pages 227–243, Berlin, 1993. Springer-Verlag.Google Scholar
  13. 13.
    G. Widmer and M. Kubat. Learning in the presence of concept drift and hidden contexts. Machine Learning, 23:69–101, 1996.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1998

Authors and Affiliations

  • Michael Harries
    • 1
  • Kim Horn
    • 2
  1. 1.Department of Artificial Intelligence School of Computer Science and EngineeringUniversity of NSWAustralia
  2. 2.Predictive Strategies UnitAustralian Gilt Securities LimitedAustralia

Personalised recommendations