Geodesic
Fuzzy clustering of spatial binary data
Kybernetika, Tome 34 (1998) no. 4, pp. 393-398
Cet article a éte moissonné depuis la source Czech Digital Mathematics Library

Voir la notice de l'article

An iterative fuzzy clustering method is proposed to partition a set of multivariate binary observation vectors located at neighboring geographic sites. The method described here applies in a binary setup a recently proposed algorithm, called Neighborhood EM, which seeks a partition that is both well clustered in the feature space and spatially regular [AmbroiseNEM1996]. This approach is derived from the EM algorithm applied to mixture models [Dempster1977], viewed as an alternate optimization method [Hathaway1986]. The criterion optimized by EM is penalized by a spatial smoothing term that favors classes having many neighbors. The resulting algorithm has a structure similar to EM, with an unchanged M-step and an iterative E-step. The criterion optimized by Neighborhood EM is closely related to a posterior distribution with a multilevel logistic Markov random field as prior [Besag1986,Geman1984]. The application of this approach to binary data relies on a mixture of multivariate Bernoulli distributions [Govaert1990]. Experiments on simulated spatial binary data yield encouraging results.
An iterative fuzzy clustering method is proposed to partition a set of multivariate binary observation vectors located at neighboring geographic sites. The method described here applies in a binary setup a recently proposed algorithm, called Neighborhood EM, which seeks a partition that is both well clustered in the feature space and spatially regular [AmbroiseNEM1996]. This approach is derived from the EM algorithm applied to mixture models [Dempster1977], viewed as an alternate optimization method [Hathaway1986]. The criterion optimized by EM is penalized by a spatial smoothing term that favors classes having many neighbors. The resulting algorithm has a structure similar to EM, with an unchanged M-step and an iterative E-step. The criterion optimized by Neighborhood EM is closely related to a posterior distribution with a multilevel logistic Markov random field as prior [Besag1986,Geman1984]. The application of this approach to binary data relies on a mixture of multivariate Bernoulli distributions [Govaert1990]. Experiments on simulated spatial binary data yield encouraging results.
Classification : 62H30, 62H86, 62M40, 65C60
Keywords: mixture models 
@article{KYB_1998_34_4_a6,
     author = {Dang, M\^o and Govaert, G\'erard},
     title = {Fuzzy clustering of spatial binary data},
     journal = {Kybernetika},
     pages = {393--398},
     year = {1998},
     volume = {34},
     number = {4},
     zbl = {1274.62418},
     language = {en},
     url = {http://geodesic.mathdoc.fr/item/KYB_1998_34_4_a6/}
}
TY  - JOUR
AU  - Dang, Mô
AU  - Govaert, Gérard
TI  - Fuzzy clustering of spatial binary data
JO  - Kybernetika
PY  - 1998
SP  - 393
EP  - 398
VL  - 34
IS  - 4
UR  - http://geodesic.mathdoc.fr/item/KYB_1998_34_4_a6/
LA  - en
ID  - KYB_1998_34_4_a6
ER  - 
%0 Journal Article
%A Dang, Mô
%A Govaert, Gérard
%T Fuzzy clustering of spatial binary data
%J Kybernetika
%D 1998
%P 393-398
%V 34
%N 4
%U http://geodesic.mathdoc.fr/item/KYB_1998_34_4_a6/
%G en
%F KYB_1998_34_4_a6
Dang, Mô; Govaert, Gérard. Fuzzy clustering of spatial binary data. Kybernetika, Tome 34 (1998) no. 4, pp. 393-398. http://geodesic.mathdoc.fr/item/KYB_1998_34_4_a6/

[1] Ambroise C.: Approche probabiliste en classification automatique et contraintes de voisinage. PhD Thesis, Université de Technologie de Compiègne 1996

[2] Ambroise C., Dang M. V., Govaert G.: Clustering of spatial data by the EM algorithm. In: Amílcar Soares (J. Gómez-Hernandez and R. Froidevaux, eds), geoENV I – Geostatistics for Environmental Applications, Kluwer Academic Publisher 1997, pp. 493–504

[3] Ambroise C., Govaert G.: An iterative algorithm for spatial clustering, submitte.

[4] Berry B. J. L.: Essay on Commodity Flows and the Spatial Structure of the Indian Economy. Research paper 111, Departement of Geography, University of Chicago 1966

[5] Besag J. E.: Spatial analysis of dirty pictures. J. Roy. Statist. Soc. 48 (1986), 259–302 | MR

[6] Bezdek J. C., Castelaz P. F.: Prototype classification and feature selection with fuzzy sets. IEEE Trans. Systems Man Cybernet. SMC-7 (1977), 2, 87–92 | DOI | Zbl

[7] Celeux G., Govaert G.: Clustering criteria for discrete data and latent class models. J. Classification 8 (1991), 157–176 | DOI | Zbl

[8] Chalmond B.: An iterative gibbsian technique for reconstruction of m-ary images. Pattern Recognition 22 (1989), 6, 747–761 | DOI

[9] Dempster A. P., Laird N. M., Rubin D. B.: Maximum likelihood from incomplete data via the EM algorithm. J. Roy. Statist. Soc. 39 (1977), 1–38 | MR | Zbl

[10] Geman S., Geman D.: Stochastic relaxation, gibbs distributions, and the bayesian restoration of images. IEEE Trans. Pattern Analysis Machine Intelligence PAMI-6 (1984), 721–741 | DOI | Zbl

[11] Govaert G.: Classification binaire et modéles. Rev. Statist. Appl. 38 (1990), 1, 67–81

[12] Hathaway R. J.: Another interpretation of the EM algorithm for mixture distributions. Statist. Probab. Lett. 4 (1986), 53–56 | DOI | MR | Zbl

[13] Legendre P.: Constrained clustering. Develop. Numerical Ecology. NATO ASI Series G 14 (1987), 289–307 | MR