A Novel Evolutionary Clustering Algorithm Based on Gaussian Mixture Model

Estimating the optimal number of clusters for a dataset is one of the most essential issues in cluster analysis. Traditional clustering algorithms usually predefine the number of clusters via random selection or contend based knowledge. An improper pre-selection for the number of clusters may easily lead to bad clustering outcome. In order to address this issue we propose in this paper a new evolutionary clustering algorithm based on Gaussian Mixture Model. Specifically, the algorithm defines a new entropy-based fitness function, and two new evolutionary operators for splitting and merging clusters. During the evaluation, we conducted two sets of experiments using a synthetic dataset and an existing benchmark for validating our algorithm. The results obtained in the first experiment show that the algorithm can estimate exactly the optimal number of clusters for a set of data. In the second experiment, we computed three major clustering validity indices and compared the corresponding results with those obtained using established clustering techniques, and found that our evolutionary clustering algorithm achieves better clustering structure. Key-Words: Clustering, Gaussian mixture model, EM algorithm, Evolutionary algorithm, Clustering validity