Learning from Perfection

A Data Mining Approach to Evaluation Function Learning in Awari
  • Jack van Rijswijck
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 2063)


Automatic tuning of evaluation function parameters for game playing programs, and automatic discovery of the very features that these parameters refer to, are challenging but potentially very powerful tools. While some advances have been made in parameter tuning, the field of feature discovery is still in its infancy. The game ofAwari offers the possibility to achieve both goals. This paper describes the efforts to design an evaluation function without any human expertise as part of the Awari playing program Bambam, as being developed by the Awari team1 at the University of Alberta.


Machine learning heuristic search game playing alpha-beta Awari 


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    Michael Buro. From simple features to sophisticated evaluation functions. In Computers and Games’ 98, pages 126–145. Springer, 1998.Google Scholar
  2. 2.
    Donald Michie and Ivan Bratko. Ideas on Knowledge Synthesis stemming from the KBBKN Endgame. Journal of the International Computer Chess Association, 10(1):3–13, 1987.Google Scholar
  3. 3.
    John Nunn. Secrets of Rook Endings. Batsford, 1992.Google Scholar
  4. 4.
    John Nunn. Extracting Information from Endgame Databases. Journal of the International Computer Chess Association, 16(4):191–200, 1993.Google Scholar
  5. 5.
    John Nunn. Secrets of Pawnless Endings. Batsford, 1994.Google Scholar
  6. 6.
    John Nunn. Secrets of Minor Piece Endings. Batsford, 1995.Google Scholar
  7. 7.
    Paul E. Utgoff and Doina Precup. Constructive Function Approximation. Technical Report 97-04, Department of Computer Science, University of Massachusetts, Amherst, MA, 1997.Google Scholar
  8. 8.
    Paul E. Utgoff and Doina Precup. Constructive Function Approximation. In H. Liu and H. Motoda, editors, Feature Extraction, Construction and Selection: A Data Mining Perspective, volume 453 of The Kluwer International Series in Engineering and Computer Science, Chapter 14. Kluwer Academic Publishers, 1998.Google Scholar
  9. 9.
    Terry van Belle. A New Approach to Genetic-Based Automatic Feature Discovery. Master’s thesis, University of Alberta, 1995.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2001

Authors and Affiliations

  • Jack van Rijswijck
    • 1
  1. 1.Department of Computing ScienceUniversity of AlbertaEdmontonCanada

Personalised recommendations