Abstract
We propose a directed acyclic hypergraph framework for a probabilistic graphical model that we call Bayesian hypergraphs. The space of directed acyclic hypergraphs is much larger than the space of chain graphs. Hence Bayesian hypergraphs can model much finer factorizations than Bayesian networks or LWF chain graphs and provide simpler and more computationally efficient procedures for factorizations and interventions. Bayesian hypergraphs also allow a modeler to represent causal patterns of interaction such as Noisy-OR graphically (without additional annotations). We introduce global, local and pairwise Markov properties of Bayesian hypergraphs and prove under which conditions they are equivalent. We also extend the causal interpretation of LWF chain graphs to Bayesian hypergraphs and provide corresponding formulas and a graphical criterion for intervention.
Similar content being viewed by others
References
Andersson, S.A., Madigan, D., Perlman, M.D.: Alternative Markov properties for chain graphs. Scand. J. Stat. 28, 33–85 (2001)
Bloemeke, M., Valtorta, M.: A hybrid algorithm to compute marginal and joint beliefs in bayesian networks and its complexity. In: Cooper, G. F., Moral, S. (eds.) Uncertainty in Artificial Intelligence: Proceedings of the Fourteenth Conference, pp 16–23. Morgan-Kaufmann, San Francisco (1998)
Cox, D.R., Wermuth, N.: Linear dependencies represented by chain graphs (with discussion). Statist. Sci. 8, 204–218 (1993). 247–277
Cox, D.R., Wermuth, N.: Multivariate Dependencies. Chapman & Hall, London (1996)
Darroch, J.N., Lauritzen, S.L., Speed, T.P.: Markov fields and log-linear interaction models for contingency tables. Ann. Statist. 8, 522–539 (1980)
Dawid, A.P.: Conditional independence for statistical operations. Ann. Statist. 8, 598–617 (1980)
Drton, M.: Discrete chain graph models. Bernoulli 15, 736–753 (2009)
Frydenberg, M.: The chain graph Markov property. Scand. J. Statist. 17, 333–353 (1990)
Ghosh, J.K., Valtorta, M.: Building a bayesian network model of heart disease. In: Proceedings of the 38th Annual ACM Southeastern Conference, Clemson, South Carolina, pp. 239–240. (Extended version at https://cse.sc.edu/mgv/reports/tr1999-11.pdf) (2000)
Hammersley, J.M., Clifford, P.E: Markov fields on finite graphs and lattices. Unpublished manuscript (1971)
Isham, V.: An introduction to spatial point processes and Markov random fields. Internat. Statist. Rev. 49, 21–43 (1981)
Hajek, P., Havranek, T., Jirousek, R.: Uncertain Information Processing In Expert Systems. CRC Press, Boca Raton (1992)
Jensen, F.V., Nielsen, T.D.: Bayesian Networks and Decision Graphs, 2nd edn. Springer, New York (2007)
Kiiveri, H., Speed, T.P., Carlin, J.B.: Recursive causal models. J. Aust. Math. Soc. Ser. A 36, 30–52 (1984)
Koster, J.T.A.: On the validity of the Markov interpretation of path diagrams of Gaussian structural equation systems with correlated errors. Scand. J. Statist. 26, 413–431 (1999)
Kschischang, F.R., Frey, B.J., Loeliger, H.: Factor graphs and the sum-product algorithm. IEEE Trans. Inf. Theory 47(2), 498–519 (2001)
Lauritzen, S.L., Spiegelhalter, D.J.: Local computations with probabilities on graphical structures and their application to expert systems. J. Roy. Statist. Soc. Ser. B 50, 157–224 (1988)
Lauritzen, S.L., Dawid, A.P., Larsen, B.N., Leimer, H.G.: Independence properties of directed Markov fields. Networks 20, 491–505 (1990)
Lauritzen, S.L.: Graphical models. Clarendon, New York (1996)
Lauritzen, S. L.: Causal inference from graphical models. In: Bardorff-Nielsen, O.E., Cox, D.R., Kliippelberg, C. (eds.) Complex Stochastic Systems, pp 63–107. Chapman and Hall, Boca Raton (2000)
Lauritzen, S.L., Jensen, F.V.: Local computations with valuations from a commutative semigroup. Ann. Math. Artif. Intell. 21, 51–69 (1997)
Lauritzen, S.L., Richardson, T.S.: Chain graph models and their causal interpretations (with discussion). J. Roy. Stat. Soc. Ser B 64, 321–361 (2002)
Lauritzen, S.L., Wermuth, N.: Graphical models for association between variables, some of which are qualitative and some quantitative. Ann. Stat. 17, 31–57 (1989)
Loeliger, H.: An introduction to factor graphs. IEEE Signal Process. Mag. 21(1), 28–41 (2004)
Neyman, J.: On the application of probability theory to agricultural experiments: essay on principles. (in Polish) (Engl. transl D. Dabrowska and T. P. Speed. Statist. Sci. 5(1990), 465–480 (1923)
Pearl, J., Paz, A.: Graphoids. A graph-based logic for reasoning about relevancy relations. In: Proceedings of 7th European Conference on Artificial Intelligence, Brighton, United Kingdom (1986)
Pearl, J.: Probabilistic Reasoning in Intelligent Systems. Morgan-Kaufmann, San Mateo (1988)
Pearl, J.: Graphical models, causality and intervention. Statist. Sci. 8, 266–269 (1993)
Pearl, J.: Causal diagrams for empirical research. Biometrika 82, 669–710 (1995)
Pearl, J.: Causality: Models, Reasoning, and Inference, 2nd edn. Cambridge University Press, Cambridge (2009)
Pena, J.M.: Marginal AMP chain graphs. Internat. J. Approx. Reason. 55, 1185–1206 (2014)
Richardson, T.S.: Markov properties for acyclic directed mixed graphs. Scand. J. Statist. 30, 145–157 (2003)
Richardson, T.S.: Personal communication (2018)
Richardson, T.S., Spirtes, P.: Ancestral graph Markov models. Ann. Statist. 30, 962–1030 (2002)
Rubin, D.B.: Estimating causal effects of treatments in randomized and non-randomized studies. J. Educ. Psychol. 66, 688–701 (1974)
Speed, T.P.: A note on nearest-neighhbour Gibbs and Markov probabilities. Sankhya A 41, 184–197 (1979)
Sperner, E.: Ein Satz über Untermengen einer endlichen Menge. Math. Z. 27, 544–548 (1928)
Spirtes, P., Glymour, C., Scheines, R.: Causation, Prediction and Search. Springer, New York (1993)
Studený, M.: Conditional independence relations have no finite complete characterization. In: Kubík, S., Á. Víšek, J. (eds.) Information Theory, Statistical Decision Functions and Random Processes: Proceedings of the 11th Prague Conference - B, Kluwer, Dordrecht (also Academia, Prague), vol. 15, pp 377–396 (1992)
Studený, M., Roverato, A., Štěpánová, Š.: Two operations of merging and splitting components in a chain graph. Kybernetika (Prague) 45(2), 208–248 (2009)
Wermuth, N., Cox, D.R.: Joint response graphs and separation induced by triangular systems. J. R. Stat. Soc. Ser. B Stat. Methodol. 66, 687–717 (2004)
Wermuth, N., Cox, D.R., Pearl, J.: Explanation for Multivariate Structures Derived from Univariate Recursive Regressions Technical Report No. 94(1). Univ, Mainz, Germany (1994)
Wermuth, N., Lauritzen, S.L.: Graphical and recursive models for contingency tables. Biometrika 70, 537–552 (1983)
Acknowledgments
This work is primarily supported by Office on Naval Research grant ONR N00014-17-1-2842. The causal interpretation of Bayesian hypergraphs described in Section 6 was presented at the AAAI Spring Symposium “Beyond Curve Fitting: Causation, Counterfactuals, and Imagination-based AI (AAAI-WHY 2019),” Palo Alto. CA, March 25-27, 2019. Comments by reviewers and symposium participants are gratefully acknowledged.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Linyuan Lu was supported in part by NSF grant DMS-1600811 and ONR grant N00014-17-1-2842.
Marco Valtorta was supported in part by ONR grant N00014-17-1-2842.
Rights and permissions
About this article
Cite this article
Javidian, M.A., Wang, Z., Lu, L. et al. On a hypergraph probabilistic graphical model. Ann Math Artif Intell 88, 1003–1033 (2020). https://doi.org/10.1007/s10472-020-09701-7
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10472-020-09701-7
Keywords
- Graphical model
- Hypergraph
- Chain graph
- Bayesian network
- Intervention
- Markov property
- Factorization
- Structural equation model