Article
Keywords:
factorizability; directed graph
Summary:
Four notions of factorizability over arbitrary directed graphs are examined. For acyclic graphs they coincide and are identical with the usual factorization of probability distributions in Markov models. Relations between the factorizations over circuits are described in detail including nontrivial counterexamples. Restrictions on the cardinality of state spaces cause that a factorizability with respect to some special cyclic graphs implies the factorizability with respect to their, more simple, strict edge-subgraphs. This gives sometimes the possibility to break circuits and get back to the acyclic, well-understood case.
References:
[1] Berman A., Plemmons R. J.:
Nonnegative Matrices in the Mathematical Sciences. Academic Press, New York – San Francisco – London 1979
MR 0544666 |
Zbl 0815.15016
[2] Cox D. R., Wermuth N.:
Multivariate Dependencies. (Monographs on Statistics and Applied Probability 67.) Chapman & Hall, London 1996
MR 1456990 |
Zbl 0880.62124
[7] Spirtes P.: Directed cyclic graphical representation of feedback models. In: Proceedings of the Eleventh Conference on Uncertainty and Artificial Inteligence (P. Besnard and S. Hanks, eds.), Morgan Kaufman Publ. Inc., San Mateo 1995
[8] Strohmeier B.: Cyclical Causal Networks and Knowledge Integration. Ph.D. Dissertation. Fakultät für Wirtschaftswissenschaften, Universität Bielefeld 1996
[9] Vajda I.:
Theory of Statistical Inference and Information. Kluwer Academic Publishers, Dordrecht – Boston – London 1989
Zbl 0711.62002
[10] Whittaker J.:
Graphical Models in Applied Multivariate Statistics. J. Wiley, New York 1990
MR 1112133 |
Zbl 1151.62053