Combining Multiple Clustering Systems

Abstract

Abstract. Three methods for combining multiple clustering systems are presented and evaluated, focusing on the problem of finding the correspondence between clusters of different systems. In this work, the clusters of individual systems are represented in a common space and their correspondence estimated by either "clustering clusters" or with Singular Value Decomposition. The approaches are evaluated for the task of topic discovery on three major corpora and eight different clustering algorithms and it is shown experimentally that combination schemes almost always offer gains compared to single systems, but gains from using a combination scheme depend on the underlying clustering systems.

A new link-based method to ensemble clustering and cancer microarray data analysis

“…Each base clustering provides a cluster label as a new feature describing each data point [Figure 2(b)], which is utilised to formulate the final solution (Boulis and Ostendorf, 2004;Nguyen and Caruana, 2007;Topchy et al, 2005). For instance, the technique of Boulis and Ostendorf (2004) makes use of linear programming to find a correspondence between the labels of base clusterings and those of the optimal final-clustering.…”

“…Each base clustering provides a cluster label as a new feature describing each data point [Figure 2(b)], which is utilised to formulate the final solution (Boulis and Ostendorf, 2004;Nguyen and Caruana, 2007;Topchy et al, 2005). For instance, the technique of Boulis and Ostendorf (2004) makes use of linear programming to find a correspondence between the labels of base clusterings and those of the optimal final-clustering. In addition, the aggregation of multiple clustering results has been considered as a maximum likelihood estimation problem, and EM algorithms (Nguyen and Caruana, 2007;Topchy et al, 2004Topchy et al, , 2005 have been proposed for finding the consensus clustering.…”

“…Over the past decade, many techniques have been developed along this line of research. They can be categorised into direct approach (Fischer and Buhmann, 2003;Gionis et al, 2007), feature-based approach (Boulis and Ostendorf, 2004;Nguyen and Caruana, 2007;Topchy et al, 2005), pairwise-similarity approach (Ayad and Kamel, 2003;Fred and Jain, 2005;Monti et al, 2003), and graph-based algorithms (Fern and Brodley, 2004;Strehl and Ghosh, 2002). Despite their theoretical and practical contributions, almost all cluster ensemble methods found in the literature make use of information available in an ensemble only at a coarse level.…”

Clustering aggregation

“…Clustering aggregation has has been previously considered under a variety of names (consensus clustering, clustering ensemble, clustering combination) in a variety of different areas: machine learning [19,12], pattern recognition [14], bio-informatics [13], and data mining [21,5]. The problem of correlation clustering is interesting in its own right, and it has recently attracted a lot of attention in the theoretical computer-science community [2,6,8,10].…”

“…They propose information theoretic distance measures, and they propose genetic algorithms for finding the best aggregation solution. Boulis and Ostendorf [5] use Linear Programming to discover a correspondence between the labels of the individual clusterings and those of an "optimal" meta-clustering. Topchy et al [21] define clustering aggregation as a maximum likelihood estimation problem, and they propose an EM algorithm for finding the consensus clustering.…”