Integrating numerical and syntactic learning models for pattern recognition

  • Terry Caelli
Invited Talks
Part of the Lecture Notes in Computer Science book series (LNCS, volume 1451)


In this paper we consider how the recognition, interpretation of image structures, patterns, objects can be posed in terms of “Inductive Bayesian Networks” (IBN) which combine syntactic domain models with the numerical/statistical characteristics of what is sensed. The net result of this formulation is the production of contextual and relational rules which can be used to summarize, generalize structural descriptions from examples in ways which are consistent with domain knowledge. In this approach the associated algorithms are also, constrained by principles o: Minimum Description Length (MDL) which endeavor to produce structural descriptions which generalize over numerical data attribute while specializing over symbolic description length. Examples in pattern and object recognition are discussed.


Pattern Recognition Macine Learning Image Annotation 


  1. 1.
    W. Bischof and T. Caelli. Learning structural descriptions of patterns: A new technique for conditional clustering and rule generation. Pattern Recognition, 27(5):689–697, 1994.CrossRefGoogle Scholar
  2. 2.
    W. Bischof and T. Caelli. Sure: Scene understanding by rule evaluation. IEEE: Transactions on Pattern Analysis and Machine Intelligence, 19(11):1284–1289, 1997.CrossRefGoogle Scholar
  3. 3.
    T. Caelli and W. Bischof. Machine Learning and Image Interpretation. Plenum, 1997.Google Scholar
  4. 4.
    C. Dillon and T. Caelli. Cite: A scene understanding and object recognition system. In Asian Conference on Computer Vision: ACCV95, volume I, pages 214–218, Singapore, Dec 1995.Google Scholar
  5. 5.
    B. McCane and T. Caelli. A fuzzy machine learning approach to recognising 3D objects in 2D s cenes. In Asian Conference on Computer Vision:ACCV95, volume III, pages 106–111, Singapore, Dec 1995.Google Scholar
  6. 6.
    B.T. Messmer and H. Bunke. Fast error-correcting graph isomorphism based on model precompilation. Technical Report IAM-96-012, University of Bern, Sep 1996.Google Scholar
  7. 7.
    T. Mitchell. Machine Learning. McGraw-Hill, 1997.Google Scholar
  8. 8.
    A. Pearce and T. Caelli. On the effiency of learning in spatial domains and relational evidence theory. volume 3, pages 290–294, Singapore, Dec 1995.Google Scholar
  9. 9.
    Prantl M. Ganster H. Pinz, A. and H. Kopp-Borot Schnig. Active fusion — a new method applied to remote sensing image interpretation. Pattern Recognition Letters, 17:1349–1359, 1996.CrossRefGoogle Scholar
  10. 10.
    T. Poggio and F. Girosi. Regularization algorithms for learning that are equivalent to multilay er networks. Science, 247:978–982, 1990.Google Scholar
  11. 11.
    J. R. Quinlan. Improved use of continuous attributes in c4.5. Journal of Artificial Intelligence Research, 1996.Google Scholar
  12. 12.
    S. Sarkar and K. Boyer. Using perceptual inference networks to manage vision processes. Computer Vision and Image Understanding, 62(1):27–46, 1995.CrossRefGoogle Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1998

Authors and Affiliations

  • Terry Caelli
    • 1
  1. 1.Center for MappingThe Ohio State UniversityUSA

Personalised recommendations