Two different types of discontinuity of bayesian learning in causal probabilistic networks

  • Ulrich G. Oppel
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 946)


First, we saw that Bayesian learning in causal probabilistic networks (and not only here) is twofold discontinuous and therefore may be risky. In applications we have to be aware of that and have to take precautions. Second, we should look out for situations where Bayesian learning is continuous. This will help us also to develop better procedures for the estimation of the Markov kernels of a CPN from data and expert knowledge.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. [1]
    Billingsley, P.: Convergence of Probability Measures. J. Wiley and Sons: New York-London-Sidney-Toronto, 1968.Google Scholar
  2. [2]
    Håjek, P.; Havránek, T. Jirousèk. R.: Uncertain Information Processing in Expert Systems. CRC Press: Boca Raton-Ann Arbor-London-Tokyo: 1992.Google Scholar
  3. [3]
    Henrion, M.: Propagating uncertainty in Bayesian networks by probabilistic logic sampling. In: Lemmer, F.J.; Kanal, L.N. (eds.): Uncertainty in Artificial Intelligence 2. Elsevier Science Publishers B.V. (North-Holland): 1988.Google Scholar
  4. [4]
    Lauritzen, S.L.; Spiegelhalter, D.: Local Computations with Probabilities on Graphical Structures and Their Application to Expert Systems. J. Roy. Stat. Soc. B, 50 (2) (1988), 157–224.Google Scholar
  5. [5]
    Matthes, R.; Oppel, U.G.: Convergence of causal probabilistic networks. In: Bouchon-Meunier, B., Valverde, L.; Yager, R.R. (ed.): Intelligent Systems with Uncertainty. Elsevier: Amsterdam-London-New York-Tokio, 1993.Google Scholar
  6. [6]
    Neapolitan, R.E.: Probabilistic Reasoning in Expert Systems. Theory and Algorithms. J. Wiley and Sons: New York-Chichester-Brisbane-Toronto-Singapore, 1990.Google Scholar
  7. [7]
    Oppel, U.G.: Every Complex System Can be Determined by a Causal Probabilistic Network without Cycles and Every Such Network Determines a Markov Field. In:Kruse, R.; Siegel, P. (eds.): Symbolic and Quantitative Approaches to Uncertainty. Lecture Notes of Computer Science 548. Springer: Berlin-Heidelberg-New York, 1991.Google Scholar
  8. [8]
    Oppel, U.G.: Causal probabilistic networks and their application to metabolic processes. In: Mammitzsch, V.; Schneeweiß, H.: Proceedings of the Second Gauss Symposium, München, August 2–7, 1993. De Gruyter: Berlin, 1995.Google Scholar
  9. [9]
    Parthasarathy, K.: Probability Measures on Metric Spaces. Academic Press: New York-London, 1969.Google Scholar
  10. [10]
    Pearl, J.: Probabilistic Reasoning in Intelligent Systems: Networks of Plausible Inference. Morgan Kaufmann: San Matteo, 1988.Google Scholar
  11. [11]
    Shachter, R.D.; Peot, M.A.: Simulation approaches to to general probabilistic belief networks. In: Henrion, M.; Shachter, R.D.; Kanal, L.N.; Lemmer, J.F. (eds.): Uncertainty in Artificial Intelligence 5. Elsevier Science Publishers B.V. (North-Holland): 1990.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1995

Authors and Affiliations

  • Ulrich G. Oppel
    • 1
  1. 1.Mathematisches Institut der Ludwig - MaximiliansUniversität MünchenMünchenGermany

Personalised recommendations