Hidden Markov Models as Self-Organizing Maps to Exploit Time Dependencies in Data Clustering

We propose a novel hidden Markov model which acts as a self-organizing map to exploit temporal dependencies in data clustering. The proposed technique is able to automatically identify the number of clusters contained in the data in an unsupervised manner. It also makes it possible to cluster together sequences that are shifted and scaled versions of each other, a problem that to the best of our knowledge has not been systematically addressed in the literature.