Unsupervised Bayesian feature selection based on k-means clustering

Bayesian methods have been successfully used for feature selection in many supervised learning tasks; while relatively little work has been done on feature selection for clustering. In this paper, the adaptation of such methods for unsupervised learning (clustering) is investigated. Basically, a Bayesian network is generated from a dataset, and then the Markov Blanket of the class variable is used to the feature subset selection task. We adopt an algorithm that iterates between clustering (assuming that the number of clusters is unknown a priori) and feature selection. From this standpoint, one Bayesian approach for feature selection is addressed: Markov Blanket Filter (MBF) obtained from the construction of Bayesian Networks.