Attribute Clustering in High Dimensional Feature Spaces

In this paper, we will do clustering for the attributes rather than the objects. Like the conventional clustering for objects, the attributes within the same cluster have high similarity, but within different clusters have high dissimilarity. A distance measure for a pair of attributes based on the relative dependency is proposed. An attribute clustering algorithm called Most Neighbors First (MNF) is also proposed to cluster the attributes into a fixed number of groups. An example is also given to illustrate the proposed algorithm.