×

zbMATH — the first resource for mathematics

Asymptotic results in parameter estimation for Gibbs random fields. (English) Zbl 0962.62092
The author studies asymptotic properties of maximum likelhood (ML) estimators and a class of maximum pseudo-likelihood (MPL) estimators for parameters of stationary Gibbs random fields. The strong consistency of the ML and particular MPL estimators is proved. It is shown that the ML estimator can be approximated by a sequence of MPL estimators for every fixed sample size. The asymptotic normality of MPL estimators is proved provided the true Gibbs distribution is ergodic; for a stationary Gibbs distribution the limiting distribution of MPL estimators is mixed normal. The asymptotic normality and efficiency of ML estimators is proved inside the Dobrushin uniqueness region. In this case MPL estimators are less efficient compared to ML estimators. However, it is shown that the maximum asymptotic efficiency is approached by MPL estimators with growing ranges.

MSC:
62M40 Random fields; image analysis
62F12 Asymptotic properties of parametric estimators
PDF BibTeX XML Cite
Full Text: Link EuDML
References:
[1] J. Besag: Spatial interaction and the statistical analysis of lattice systems. J. Roy. Statist. Soc. B 36 (1974), 192-226. · Zbl 0327.60067
[2] J. Besag: On the statistical analysis of dirty pictures (with discussion). J. Roy. Statist. Soc. Ser. B 48 (1986), 259-302. · Zbl 0609.62150
[3] E. Bolthausen: On the central limit theorem for stationary mixing random fields. Ann. Probab. 10 (1982), 1047-1050. · Zbl 0496.60020 · doi:10.1214/aop/1176993726
[4] F. Comets: On consistency of a class of estimators for exponential families of Markov random fields on a lattice. Ann. Statist. 20 (1992), 455-468. · Zbl 0787.62100 · doi:10.1214/aos/1176348532
[5] R. L. Dobrushin, B. S. Nahapetian: Strong convexity of the pressure for the lattice systems of classical statistical physics. Teor. Mat. Phys. 20 (1974), 223-234. · Zbl 0311.60063
[6] D. Geman, S. Geman: Maximum Entropy and Bayesian Methods in Sciences and Engineering. (C. R. Smith and G. J. Erickson, Kluwer, Dordrecht 1988. · Zbl 0706.62036
[7] S. Geman, C. Graffigne: Markov random field image models and their applications to computer vision. Proc. Internat. Congress Math. (A. M. Gleason, Amer. Math. Soc, Providence, R. I. 1987. · Zbl 0665.68067
[8] H. O. Georgii: Gibbs Measures and Phase Transitions. De Gruyter, Berlin 1988. · Zbl 0657.60122 · doi:10.1515/9783110850147
[9] B. Gidas: Consistency of maximum likelihood and pseudo-likelihood estimators for Gibbs distribution. Stochastic Differential Systems, Stochastic Control Theory, and Application (W. Fleming and P. L. Lions, IMA Vol. Math. Appl. 10). Springer, New York 1988. · Zbl 0648.62099
[10] B. Gidas: Parameter estimation for Gibbs distributions. I. Fully observed data. Markov Random Fields: Theory and Applications (R. Chellapa and R. Jain, Academic Press, New York 1991.
[11] L. Gross: Absence of second-order phase transition in the Dobrushin’s uniqueness region. J. Statist. Phys. 27 (1981), 57-72.
[12] X. Guyon: Estimation d’un champ par pseudo-vraisemblance conditionnelle: Etude asymptotique et application au cas Markovien. Actes de la 6eme Rencontre Franco-Belge de Statisticiens, Bruxelles 1985.
[13] X. Guyon, H. R. Künsch: Asymptotic comparison of estimators in the Ising model. Stochastic Models, Statistical Methods, and Algorithms in Image Analysis (P. Barone, A. Frigessi and M. Piccioni, Lecture Notes in Statistics 74), Springer, Berlin 1992, pp. 177-198. · Zbl 0801.62086
[14] J. Hájek: Local asymptotic minimax and admissibility in estimation. Proc. 6th Berkeley Symposium, Vol. 1, Berkeley, Calif. 1970, pp. 175-194.
[15] F. R. Hampel E. M. Ronchetti P. J. Rousseeuw, W. A. Stahel: Robust Statistics – The Approach Based on Influence Functions. Wiley, New York 1986. · Zbl 0593.62027
[16] M. Janžura: Estimating interactions in binary lattice data with nearest-neighbor property. Kybernetika 23 (1987), 2, 136-142. · Zbl 0668.62059 · eudml:28026
[17] M. Janžura: Statistical analysis of Gibbs random fields. Trans. 10th Prague Conf. on Inform. Theory, Stat. Dec. Functions, Random Processes 1986, Praha, pp. 429-438.
[18] M. Janžura: Asymptotic theory of parameter estimation for Gauss-Markov random fields. Kybernetika 24 (1988), 161-176. · Zbl 0655.62094 · www.kybernetika.cz · eudml:28512
[19] M. Janžura: Local asymptotic normality for Gibbs random fields. Proceedings of the Fourth Prague Symposium on Asymptotic Statistics (P. Mandl, M. Hušková, Charles University, Prague 1989, pp. 275-284. · Zbl 0697.62091
[20] M. Janžura, P. Lachout: A central limit theorem for stationary random fields. Math. Methods Statist. 4 (1995), 463-472. · Zbl 0853.60024
[21] H. R. Künsch: Thermodynamics and statistical analysis of Gaussian random fields. Z. Wahrsch. Verw. Gebiete 58 (1981), 407-421. · Zbl 0458.60053
[22] H. Künsch: Decay of correlations under Dobrushin’s uniqueness condition and its applications. Commun. Math. Phys. 84 (1982), 207-222. · Zbl 0495.60097 · doi:10.1007/BF01208568
[23] H. Künsch: Infinitesimal robustness for autoregressive processes. Ann. Statist. 12 (1984), 843-863. · Zbl 0587.62077 · doi:10.1214/aos/1176346706
[24] C. Preston: Random Fields. (Lecture Notes in Mathematics 534). Springer, Berlin 1976. · Zbl 0335.60074 · doi:10.1007/BFb0080563
[25] D. J. Strauss: Analysing binary lattice data with the nearest-neighbor property. J. Appl. Probab. 12 (1975), 702-712. · Zbl 0322.62072 · doi:10.2307/3212721
[26] L. Younès: Estimation and annealing for Gibbsian fields. Ann. Inst. H. Poincaré Sect. B (N. S.) 24 (1988), 269-294. · Zbl 0651.62091 · numdam:AIHPB_1988__24_2_269_0 · eudml:77326
[27] L. Younès: Parametric inference for imperfectly observed Gibbsian fields. Probab. Theory Related Fields 82 (1989), 625-645. · Zbl 0659.62115 · doi:10.1007/BF00341287
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. It attempts to reflect the references listed in the original paper as accurately as possible without claiming the completeness or perfect precision of the matching.