Smith, J. Q., 1953- and Daneshkhah, Alireza (2008) Large incomplete sample robustness in Bayesian networks. Working Paper. University of Warwick. Centre for Research in Statistical Methodology, Coventry.

Preview

PDF WRAP_Smith_08-12w.pdf - Published Version - Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader Download (351Kb)

Abstract

Under local DeRobertis (LDR) separation measures, the posterior distances between two densities is the same as between the prior densities. Like Kullback - Leibler separation they also are additive under factorization. These two properties allow us the prove that the precise specification of the prior will not be critical with respect to the variation distance on the posteriors under the following conditions. The genuine and approximating prior need to be similarly rough, the approximating prior has concentrated on a small ball on the margin of interest, not on the boundary of the probability space, and the approximating prior has similar or fatter tails to the genuine prior. Robustness then follows for all likelihoods, even ones that are misspecified. Furthermore, the variation distances can be bounded explicitly by a easy to calculate function the prior LDR and simple summary statistics of the functioning posterior. In this paper we apply these results to study the robustness of prior specification to learning Bayesian Networks.

Item Type:	Working or Discussion Paper (Working Paper)
Subjects:	Q Science > QA Mathematics
Divisions:	Faculty of Science > Statistics
Library of Congress Subject Headings (LCSH):	Robust statistics, Bayesian statistical decision theory
Series Name:	Working papers
Publisher:	University of Warwick. Centre for Research in Statistical Methodology
Place of Publication:	Coventry
Date:	2008
Volume:	Vol.2008
Number:	No.12
Number of Pages:	15
Status:	Not Peer Reviewed
Access rights to Published version:	Open Access
References:	[1] J. A. A. Andrade and Anthony O’Hagan. 2006. Bayesian robustness modelling using regularly varying distributions. Bayesian Analysis, 1:169–188. [2] R. G. Cowell, A. P. Dawid, S. L. Lauritzen and D. J. Spiegelhalter. 2000. Probabilistic Networks and Expert Systems. Springer Verlag. [3] L. DeRobertis. 1978. The use of partial prior knowledge in Bayesian inference. Ph.D. dissertation, Yale Univ. [4] P. Gustafson and L. Wasserman. 1995. Local sensitivity diagnostics for Bayesiain inference. Annals Statist, 23:2153–2167. [5] S. French and D. Rios Insua. 2000. Statistical Decision Theory. Kendall’s Library of Statistics 9 Arnold. [6] J. B. Kadane and D. T. Chuang. 1978. Stable decision problems. Ann. Statist, 6:1095–1111. [7] S. L. Lauritzen. 1996. Graphical Models. Oxford University Press. [8] A. O’Hagan. 1979. On outlier rejection phenomena in Bayesian inference. J. R. Statist. Soc. B, 41:358–367. [9] A. O’Hagan and J. Forster. 2004. Bayesian Inference. Kendall’s Advanced Theory of Statistics, Arnold. [10] D. J. Spiegelhalter and S. L. Lauritzen .1990. Sequential updating of conditional probabilities on directed graphical structures. Networks, 20:579–605. [11] J. Q. Smith. 2007. Local Robustness of Bayesian Parametric Inference and Observed Likelihoods. CRiSM Res Rep, 07-08. [12] J. Q. Smith and F. Rigat. 2008. Isoseparation and Robustness in Finitre Parameter Bayesian Inference. CRiSM Res Rep, 07-22. [13] J. Q. Smith and J. Croft. 2003. Bayesian networks for discrete multivariae data: an algebraic approach to inference. J of Multivariate Analysis, 84(2):387–402. [14] Y. L. Tong. 1980. Probability Inequalities in Multivariate Distributions. Academic Press New York.
URI:	http://wrap.warwick.ac.uk/id/eprint/35487

Request changes to a record

Actions (login required)

View Item