Subject(s): CX.07, CX.11, CX.65

Category: Brief Article

Abstract:

This study describes an unsupervised general methodology for partitioning data into clusters. The notion of clusters is based on information-theoretic terms for the entropy of the partitions given the data. The model produces a hierarchy of clustering, where each partition is expressed as a mixture of a set of Gaussian kernel functions. It can deal with cluster of general shapes and sizes.

Retrieve Manuscript

Submit referee report/comment