Re: [UAI] Bayes net structure learning using single link look-ahead.

From: Jae-Hyuck Lee (lee@cis.uoguelph.ca)
Date: Mon Apr 09 2001 - 10:55:05 PDT

  • Next message: nahla.benamor@ihec.rnu.tn: "[UAI] Benchmarks of propagation in multiply connected BN"

    Hi,

    Dr. Y. Xiang is the founder and primary investigator of the theory
    of "pseudo-independent" domains that causes failure of commonly used
    algorithms for learning belief networks from data due to their single-
    link-lookahead search method. You can see the full list of the
    papers dealing with this issue at his homepage where you can download
    most of them:

      http://snowhite.cis.uoguelph.ca/~yxiang/

    The following is the reference to some:

    Y. Xiang, S.K.M Wong and N. Cercone. "Critical Remarks on Single
    Link Search in Learning Belief Networks." In Proc. Uncertainty in
    Artificial Intelligence 96, 564-571, 1996.

    Y. Xiang, "A characterization of single-link search in learning
    belief networks," in Proc. Pacific Rim Knowledge Acquisition
    Workshop, 218-233, Singapore, 1998.

    Y. Xiang, J. Hu, N. Cercone and H. Hamilton, "Learning
    Pseudo-Independent Models: Analytical and Experimental Results",
    Proc. AI 2000, In H. Hamilton, editor, Advances in Artificial
    Intelligence, 227-239, Springer, 2000.

    Hope this helps.

    Jae

    On Fri, 6 Apr 2001 R.G.Cowell@city.ac.uk wrote:

    > Several Bayes net structure learning algorithms work incrementally by
    > testing the addition of a single link to a current model , seeing if
    > it improves some score, such as marginal likelihood, and then
    > selecting the link which increases the score the most. I remember that
    > there were some papers a few years back pointing out that this
    > single-link look-ahead search fails for some special situations and
    > the simultaneous addition of two links should be considered instead,
    > but I cannot recall any details of the papers.
    >
    > I would be grateful if anyone can provide references to papers which
    > deal with this issue.
    >
    > Thanks in advance.
    >
    >
    >
    > Robert Cowell
    > Department of Actuarial Science and Statistics
    > City University
    > Northampton Square
    > London EC1V 0HB
    >
    > tel: (44) 020 7477 - 8454
    > fax (44) 020 7477 - 8838
    > www: http://www.staff.city.ac.uk/~rgc
    >



    This archive was generated by hypermail 2b29 : Mon Apr 09 2001 - 11:05:51 PDT