A Review On Consensus Clustering Methods

Unsupervised learning/clustering is one of the most common, yet computationally intense, data analysis problems in data mining. The plethora of clustering algorithms and performance measures makes the choice of optimal clustering algorithm a challenging task. In order to overcome this shortcoming consensus learning methods have been proposed in the literature. These methods try to optimally combine independently obtained clusterings into a single more robust clustering of improved quality. In this chapter we provide a review of unsupervised consensus learning techniques based on their underlying theoretical principles. We present the exact, approximation, and heuristic approaches, the relation of consensus clustering with other well-studied problems, and discuss relevant applications