Accelerating distributed Expectation-Maximization algorithms with frequent updates

Jiangtao Yin, Yanfeng Zhang, Lixin Gao

Abstract: Highlights •Propose two approaches to parallelize EM algorithms with frequent updates in a distributed environment to scale to massive data sets.•Prove that both approaches maintain the convergence properties of the EM algorithms.•Design and implement a distributed framework to support the implementation of frequent updates for the EM algorithms.•Extensively evaluate our framework on both a cluster of local machines and the Amazon EC2 cloud with several popular EM algorithms.•The EM algorithms with frequent updates implemented on our framework can converge much faster than traditional implementations.