Cluster analysis using a deviation contrast to independence
Résumé
In this paper, Maximum Entropy Principle is used as a general measure of directed divergence or discrepancy between two distributions to measure the deviation to independence of a set of variables. Potential applications of such study are problems of pattern recognition and in particular any problem for which distance between distribution is needed such as cluster analysis, independent component analysis, etc.
Origine : Fichiers produits par l'(les) auteur(s)
Loading...