A Framework for Cluster Ensemble Based on a Max Metric as Cluster Evaluator

A new criterion for clusters validation is proposed in the paper and based on the new cluster validation criterion a clustering ensemble framework is proposed. The main idea behind the framework is to extract the most stable clusters in terms of the defined criteria. To combine a set of partitions into one consensus partition, hierarchical clustering algorithms can be employed where first the EAC method is applied over the output partitions to convert them into a co-association matrix and then considering it as a new data space bring a consensus partition out of them. But in proposed method due to having a set of clusters instead of a set of partitions, to extract the best representative consensus partition out of the set of chosen clusters the EAC method cannot be employed, and then we turn to a new EAC based method which is called Extended EAC, EEAC. EEAC is applied to construct the co-association matrix from the subset of clusters. Finally employing a simple hierarchical clustering algorithm as final consensus function the final representative partition is produced. Employing this new cluster validation criterion, the obtained ensemble is evaluated on some well-known and standard data sets. The empirical studies show promising results for the ensemble obtained using the proposed criterion comparing with the ensemble obtained using the standard clusters validation criterion.