Distributed and Parallelled EM Algorithm for Distributed Cluster Ensemble

The paper introduces base clusterings distributed cluster ensemble which can handle the problems of privacy preservation, distributed computing and knowledge reuse. First, the latent variables in latent Dirichlet location model for cluster ensemble (LDA-CE) are defined and some terminologies are defined. Second, Variational approximation inference for LDA-CE is stated in detail. Third, base on the variational approximation inference, we design a distributed and paralleled EM algorithm for cluster ensemble (DPEM). Finally, some datasets from UCI are chosen for experiment, Compared with cluster-based similarity partitioning algorithm (CSPA), hyper-graph partitioning algorithm(HGPA) and meta-clustering algorithm(MCLA), the results show DPEM algorithm does work better and DPEM can work distributed and paralleled, so DPEM can protect privacy information more and can save time.