Temporal Pattern Generation using Hidden Markov based Unsupervised Classification

Cen Li and Gautam Biswas

Department of Computer Science, Vanderbilt University
Box 1679, Station B,
Nashville, TN 37235

Advances Intelligent Data Analysis, Lecture notes in Computer Science vol. 1642,
D. Hand, K. Kok, and M. Berthold, eds. Springer, New York, NY, August 1999.

This paper describes a clustering methodology for temporal data using hidden Markov model(HMM) representation. The proposed method improves upon existing HMM based clustering methods in two ways:

it enables HMMs to dynamically change its model structure to obtain a better fit model for data during clstering process
it provides objective criterion function to automatically select the clustering partition

The algorithm is presented in terms of four nested levels of searches:

the search for the number of clusters in a partition
the search for the structure for a fixed sized partion
the search for the HMM structure for each cluster
the search for the parameter values for each HMM

Preliminary experiments with artifically generated data demonstrate the effectiveness of the proposed methodology.

Full Paper (PDF180224 bytes).