Summary
Let Y be a Markov random field on S parametrized via its local conditional specifications. We study consistency of coding and pseudo-likelihood estimators. Then we obtain conditional asymptotic normality for the coding estimator and deduce that the difference of coding statistic for two nested hypotheses is, unconditionally, a chi-square. For these results, we do not need regularity of the lattice, translation invariance for the specification or weak dependence for the field.
Résumé
Soit Y un champ de Markov sur S, paramétrisé via ses spécifications conditionnelles locales. On étudie la consistance des estimateurs de codage et de pseudo-vraisemblance. Nous obtenons alors la normalité asymptotique conditionnelle pour l’estimateur de codage et en déduisons que le test asymptotique de différence de codage pour deux hypothèses emboîtées est une statistique du chi 2. Pour établir ces résultats, il n’est pas nécessaire de supposer le lattice S régulier ni la spécification invariante par translation ni le champ Y faiblement dépendant.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
References
Besag J. (1974): Spatial interaction and the statistical analysis of lattice systems. J.R.S.S. b, n 36, p.192–236.
Besag J. (1986): On the statistical analysis of dirty pictures. J.R.S.S. b, n 48, p. 259–302.
Breiman L. (1968): Probability. Addison Wesley.
Comets F. (1989): On consistency for exponential families of Markov random fields on the lattice. Prepub. n.89–30,Université d’Orsay,to appear.
Comets F. et Gidas B. (1988): Parameter estimation for Gibbs distributions from partially observed data. to appear in Ann. Appl. Prob.
Dacunha-Castelle D. et Duflo M. (1983): Probabilités et statistiques, Tome 2: Problèmes à temps mobile. Masson.
Frigessi A. and Piccioni M. (1990): Parameter estimation for two-dimensional Ising fields corrupted by noise. Stoch. Proc. and their Appl.34,p.297–311.
Geman D. et Graffigne C. (1986): Markov random field image models and their application to computer vision. Proceeding of the International congress of math., Ed. A. M. Gleason, A.M.S., Providence.
Gidas B.: Parametric estimation for Gibbs distributions, I: fully observed data. to appear in “Markov random fields: theory and applications”,Acad. Press.
Guyon X. (1987): Estimation d’un champ gaussien par pseudo-vraisemblance conditionnelle: étude asymptotique et application au cas markovien. in Spatial processes and spatial time series analysis,proceeding of,6th,Franco-Belgian meeting of statisticians. Ed. Droesbeke,Pub. Fac. Univ. St Louis,Bruxelles,p.15–62.
Hardouin C. (1991): Processus de contraste fort et applications. Thesis,in progress.
Guyon X. et Kunsch H.R. (1990): Asymptotic comparison of estimators in the Ising model. Same volume.
Janzura M; (1988): Statistical analysis of Gibbs random fields. 10th Prague conference 1986 “Inform. th., Stat.decis. func. and random processes”,Reidel publishing comp.p.429–438.
Jensen J.L. and Moller J. (1989): Pseudo-likelihood for exponential family models of spatial processes. Research report n 203, Dept. of th. Statistics, Univ. of Aarhus.
Strauss D.J. (1975): Analysing binary lattice data with nearest-neighbour property. J.A.P.12,p.702–715.
Yao J.F. (1990): Méthodes bayésiennes en segmentation d’image et estimation par rabotage des modèles spatiaux. Thèse Univ. Orsay.
Younes L. (1988): Parametric inference for imperfectly observed Gibbsian fields. Prob. Th. and Rel. Fields 82,n 4,p.625–645.
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 1992 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Guyon, X., Hardouin, C. (1992). The Chi-Square Coding Test for Nested Markov Random Field Hypotheses. In: Barone, P., Frigessi, A., Piccioni, M. (eds) Stochastic Models, Statistical Methods, and Algorithms in Image Analysis. Lecture Notes in Statistics, vol 74. Springer, New York, NY. https://doi.org/10.1007/978-1-4612-2920-9_11
Download citation
DOI: https://doi.org/10.1007/978-1-4612-2920-9_11
Publisher Name: Springer, New York, NY
Print ISBN: 978-0-387-97810-9
Online ISBN: 978-1-4612-2920-9
eBook Packages: Springer Book Archive