UTILISATION DES CARTES D'AUTO-ORGANISATION DANS LA CLASSIFICATION, ILLUSTRATION PAR DES EXEMPLES.

The article presents a paper on the possible uses of self-organizing maps, especially in classification and data-set visualization purposes, through various examples and approaches. From the point of view of applicability and of statistical problems, the diversity of the examples proposes different characteristic ownership of the self-organizing maps in reference to more traditional methods. The statistical algorithm is presented as a general implementation, which introduces a notion of neighborhood between the classes of the algorithm of the mobile centers. The algorithms of classification take into account at the same time the effects of big distance and local effects. Apart from clustering, self-organizing maps are also useful as a visualization tool for the description of data-set intrinsic structure.