An Unsupervised Clustering Method by Entropy Minimization

Abstract

We address the problem of unsupervised clustering using a Bayesian framework. The entropy is considered to define a prior and enables us to overcome the problem of defining a priori the number of clusters and an initialization of their centers. A deterministic algorithm derived from the standard k-means algorithm is proposed and compared with simulated annealing algorithms. The robustness of the proposed method is shown on a magnetic resonance (MR) images database containing 65 volumetric (30) images.