论文信息 - Piecewise hierarchical clustering

Piecewise hierarchical clustering

We consider two or more ultrametric distance matrices defined over different, possibly overlapping, subsets. These matrices are merged into one ultrametric matrix defined over the whole set. Necessary and sufficient conditions for uniqueness of the merging are established. when these conditions are not satisfied, consistent algorithms are given.ResumeEtant données au moins deux matrices de distances ultramétriques définies sur des sous-ensembles disjoints ou non, on les fusionne pour obtenir une ultramétrique définie sur l'ensemble tout entier. On donne des conditions nécessaires et suffisantes pour que l'extension soit unique. Quand ces conditions ne sont pas satisfaites on donne des algorithmes dont on montre qu'ils sont consistants.

G. Brossier

[1] C. Coombs. A theory of data. , 1965, Psychology Review.

[2] M. Jambu,et al. Cluster analysis and data analysis , 1985 .

[3] Brian Everitt,et al. Clustering of large data sets , 1983 .

[4] George W. Furnas,et al. The estimation of ultrametric and path length trees from rectangular proximity data , 1984 .

[5] A. D. Gordon. Consensus supertrees: The synthesis of rooted trees containing overlapping sets of labeled leaves , 1986 .

[6] J. Barthelemy,et al. On the use of ordered sets in problems of comparison and consensus of classifications , 1986 .

[7] Gildas Brossier. Étude des matrices de proximité rectangulaires en vue de la classification , 1986 .

[8] F. Roush. Les arbres et les representations des proximites : J.-P. Barthelemy and A. Guenoche, Paris: Masson, 1988, 236 pages, 160 francs. , 1989 .