next up previous contents
Next: Data-sets Up: Cluster Ensembles Previous: Discussion and Comparison   Contents

Cluster Ensemble Applications and Experiments

Consensus functions enable a variety of new approaches to several problems. After introducing the data-sets used and the evaluation methodology, in subsection 5.4.3 we illustrate how robustness of clustering can be increased through combining a set of clusterings. In subsection 5.4.4, we discuss how distributed clustering can be performed when each entity only has access to a very limited subset of features. Finally, subsection 5.4.5 shows how one can operate on several very limited subset of objects and combine them afterwards thereby enabling distributed clustering.


Alexander Strehl 2002-05-03