A new method for weighted ensemble clustering and coupled ensemble selection

Clustering ensemble, also referred to as consensus clustering, has emerged a method of combining an ensemble different clusterings derive final clustering that is better quality and robust than any single in the ensemble. Normally algorithms literature combine all without learning But by one can define merit or even cluster it, forming consensus. In this work, we propose cluster-level surprisal measure reflects both levels agreement well disagreement among clusters. Using proposed merit, devise polynomial heuristics judiciously selects subset from contribute positively We empirically show achieved our performs terms compared well-known on benchmark datasets.