Fuzzy Clustering with Automated Model Selection: Entropy Penalty Approach
Tao Li, Jinwen Ma · 2018
Fuzzy c-means algorithm is a widely used clustering method. However, its effectiveness depends on the given number of clusters in the dataset, which is unknown in many instances. The existing model selection methods are generally based on certain validity index or Bayesian analysis suffering from high computational cost. Recently, automated model selection methods have been established to make model selection automatically during parameter learning. In this paper, we propose an automated model selection mechanism into fuzzy c-means algorithm through the entropy penalized learning. It is demonstrated by extensive experiments that our proposed entropy penalized fuzzy clustering algorithm can make model selection during the parameter learning. Moreover, this algorithm is successfully applied to unsupervised image segmentation.