Clustering classifiers for knowledge discovery from physically distributed databases

Most distributed classification approaches view data distribution as a technical issue and combine local models aiming at a single global model. This however, is unsuitable for inherently distributed databases, which are often described by more than one classification models that might differ conceptually. In this paper we present an approach for clustering distributed classifiers in order to discover groups of similar classifiers and thus similar databases with respect to a specific classification task. We also show that clustering distributed classifiers as a pre-processing step for classifier combination enhances the achieved predictive performance of the ensemble.