Minimal model complexity search

  • Chris McConnell
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 990)


SURFER is an empirical discovery system that given a set of input data and a modelling vocabulary returns the model that best describes that data. The best model is considered to be the one that minimizes the description length of that model plus the data encoded using that model. The search for models is controlled by the a priori estimate of model likelihoods as encoded in the modelling vocabulary. SURFER includes domain independent mechanisms for identifying redundant models and for finding free parameters. The system is described together with the results of running the system on several different types of problems.


Bayesian Learning discovery explanation-based learning minimum description length 


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    M. M. Kokar. Determining arguments of invariant functional descriptions. Machine Learning, 1:403–422, 1986.Google Scholar
  2. 2.
    John R. Koza. Genetic programming: A paradigm for genetically breeding populations of computer programs to solve problems. Technical Report STAN-CS-90-1314, Stanford University, 1990.Google Scholar
  3. 3.
    Pat Langley and Jan M. Zytkow. Data-driven approaches to empirical discovery. Artificial Intelligence, 40(1):283–312, 1989.CrossRefGoogle Scholar
  4. 4.
    John R. Pierce. An Introduction to Information Theory. Dover Publications, Inc, New York, NY, 1980.Google Scholar
  5. 5.
    William H. Press, Brian P. Flannery, Saul A. Teukolsky, and William T. Vetterling. Numerical Recipes in C. Cambridge University Press, Cambridge, England, 1988.Google Scholar
  6. 6.
    J. Rissanen. Modeling by shortest data description. Automatica, 14:465–471, 1978.CrossRefGoogle Scholar
  7. 7.
    Hanan Samet. The Design and Analysis of Spatial Data Structures. Addison-Wesley, New York, 1990.Google Scholar
  8. 8.
    R. J. Solomonoff. A formal theory of inductive inference. parts i and ii. Information and Control, 7:1–22, 224–254, 1964.CrossRefGoogle Scholar
  9. 9.
    R. J. Solomonoff. A system for incremental learning based on algorithmic probability. In AAAI Symposium on the Theory and Application of Minimal-Length Encoding, pages 140–146, March 1990.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1995

Authors and Affiliations

  • Chris McConnell
    • 1
  1. 1.School of Computer ScienceCarnegie Mellon UniversityPittsburghUSA

Personalised recommendations