Using empirical subsumption to reduce the search space in learning

  • Marc Champesme
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 954)


In the traditional learning framework, hypothesis that are not equivalent with respect to the standard subsomption relation can be equivalent from the learning's point of view. We define in this paper a new subsumption relation, called empirical subsumption, that allows to take into account this fact. This new subsomption relation is then used to define a particular kind of search space reduction that do not reduce the class of learnable concepts. Then, we show that theses theoretical results can be applied when the knowledge representation formalism is the conceptual graph formalism.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    Brezellec, P., Soldano, H.: ELENA: a bottom-up learning method. Tenth International Conference on Machine Learning. Amherst (Massachusetts, USA). Morgan Kaufmann. (1993) 9–16.Google Scholar
  2. 2.
    Champesme, M.: Apprentissage par detection de similarites utilisant le formalisme des graphes conceptuels. Phd Thesis, Paris XIII (1993).Google Scholar
  3. 3.
    Champesme, M.: Operateurs de raffinements ideaux pour les graphes conceptuels. Prepublication LIPN (1994).Google Scholar
  4. 4.
    Chein, M., Mugnier, Marie-Laure: Conceptual Graphs: fundamental notions. Revue d'Intelligence Artificielle, 6 (1992) 365–406.Google Scholar
  5. 5.
    Van der Laag, P.R.J., Nienhuys-Cheng, Shan-Hwei: Subsumption and refinement in model inference. ECML-93, European Conference on Machine Learning. Vienna, Austria. Springer-Verlag.(1993) 95–114.Google Scholar
  6. 6.
    Van der Laag, P.R.J., Nienhuys-Cheng, Shan-Hwei: Existence and Nonexistence of Complete Refinement Operators. ECML-94, European Conference on Machine Learning. Springer Verlag. (1994) 307–322.Google Scholar
  7. 7.
    Van der Laag, P.R.J., Nienhuys-Cheng, Shan-Hwei: A note on ideal refinement operators in inductive logic programming. ILP-94, Fourth International Workshop on Inductive Logic Programming. GMD. Bad Honnef/Bonn, Germany. (1994) 247–260.Google Scholar
  8. 8.
    Levinson, R.A.: APS: An architecture for experience-based knowledge acquisition. Workshop on Computational Architectures for Supporting Machine Learning and Knowledge Acquisition at Machine Learning Conference. Aberdeen (Scotland). Morgan Kaufmann. (1992).Google Scholar
  9. 9.
    Liquiere, M.: Apprentissage a partir d'objets structures: Conception et realisation. These de 3eme cycle, Montpellier. (1990).Google Scholar
  10. 10.
    Mineau G.W.: Acquisition d'objets structures destines a la classification symbolique. Premieres Journees Francophones d'Apprentissage et d'explicitation des Connaissances. Dourdan (France). (1992) 131–145.Google Scholar
  11. 11.
    Muggleton, S., Feng, C.: Efficient Induction of Logic Programs. In Muggleton S. (Eds.), Inductive Logic Programming. Academic Press. (1992) 281–298.Google Scholar
  12. 12.
    Plotkin, G.D.: A further note on inductive generalisation. Machine Intelligence, 6 (1971) 101–124.Google Scholar
  13. 13.
    Quinlan, J. R.: Learning logical definitions from relations. Machine Learning, 5, (1990) 239–266.Google Scholar
  14. 14.
    Shapiro, E. Y.: Algorithmic program debugging. MIT Press. (1983).Google Scholar
  15. 15.
    Sowa, J.F.: Conceptual Structures, information processing in mind and machine. Addison Wesley. (1984).Google Scholar
  16. 16.
    Winston, P.H.: Learning structural descriptions from examples. In P.H. Winston (Eds.), The psychology of computer vision. McGraw Hill. (1975).Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1995

Authors and Affiliations

  • Marc Champesme
    • 1
  1. 1.Laboratoire d'Informatique de Paris-Nord (LIPN) CNRS URA 1507 Institut GalileeUniversite Paris-NordVilletaneuse

Personalised recommendations