A survey of model-based clustering algorithms for sequential data

Clustering is a fundamental and widely applied method in understanding and exploring a data set. Interest in clustering has increased recently due to the emergence of several new areas of applications including data mining, bioinformatics, web use data analysis, image analysis and so on. Model-based clustering is one of the most important and widely used clustering methods. This paper presents some existing learning algorithms for model-based clustering. Firstly, different kinds of stochastic models are introduced, followed by finite mixture models for sequential data. Then, we review learning algorithms for finite mixture models, where expectation-maximization (EM) algorithm is described in detail. The component number estimation and model initialization issues for finite mixture models will then be presented respectively. Finally, we will discuss some possible future research issues.