Kybernetika 50 no. 3, 363-377, 2014

Variations on undirected graphical models and their relationships

David Heckerman, Christopher Meek and Thomas RichardsonDOI: 10.14736/kyb-2014-3-0363


We compare alternative definitions of undirected graphical models for discrete, finite variables. Lauritzen \cite{Lauritzen:1996} provides several definitions of such models and describes their relationships. He shows that the definitions agree only when joint distributions represented by the models are limited to strictly positive distributions. Heckerman et al. \cite{Heckerman_et_al:2000}, in their paper on dependency networks, describe another definition of undirected graphical models for strictly positive distributions. They show that this definition agrees with those of Lauritzen \cite{Lauritzen:1996} again when distributions are strictly positive. In this paper, we extend the definition of Heckerman et al. \cite{Heckerman_et_al:2000} to arbitrary distributions and show how this definition relates to those of Lauritzen \cite{Lauritzen:1996} in the general case.


graphical model, undirected graph, Markov properties, Gibbs sampler, conditionally specified distributions, dependency network


60E05, 62H99, 68T30


  1. A. Agresti: Categorical Data Analysis. Wiley and Sons, New York 1990.   CrossRef
  2. B. C. Arnold, E. Castillo and J. Sarabia: Conditional Specification of Statistical Models. Springer-Verlag, New York 1999.   CrossRef
  3. M. S. Bartlett: An Introduction to Stochastic Processes. University Press, Cambridge 1955.   CrossRef
  4. J. Besag: Spatial interaction and the statistical analysis of lattice systems. J. Roy. Statist. Soc. Ser. B 36 (1974), 192-236.   CrossRef
  5. D. Brook: On the distinction between the conditional probability and the joint probability approaches in the specification of nearest-neighbor systems. Biometrika 51 (1964), 481-483.   CrossRef
  6. D. Heckerman, D. M. Chickering, C. Meek, R. Rounthwaite and C. Kadie: Dependency networks for inference, collaborative filtering, and data visualization. J. Mach. Learn. Res. 1 (2000), 49-75.   CrossRef
  7. S. L. Lauritzen: Graphical Models. Clarendon Press, Oxford 1996.   CrossRef
  8. P. Lévy: Cha\^{i}nes doubles de {M}arkoff et fonctions aléatoires de deux variables. C. R. Académie des Sciences, Paris 226 (1948), 53-55.   CrossRef
  9. J. Moussouris: Gibbs and Markov random systems with constraints. J. Statist. Phys. 10 (1974), 11-33.   CrossRef
  10. F. Matúš and M. Studený: Conditional independence among four random variables I. Combin. Probab. Comput. 4 (1995), 269-78.   CrossRef
  11. J. R. Norris: Markov {C}hains. Cambridge University Press, Cambridge 1997.   CrossRef
  12. E. Yang, P. Ravikumar, G. I. Allen and Z. Liu: Graphical Models via Generalized Linear Models. In: Advances in Neural Information Processing Systems 25 (2013), Cambridge.   CrossRef