Newscast EM

We propose a gossip-based distributed algorithm for Gaussian mixture learning, Newscast EM. The algorithm operates on network topologies where each node observes a local quantity and can communicate with other nodes in an arbitrary point-to-point fashion. The main difference between Newscast EM and the standard EM algorithm is that the M-step in our case is implemented in a decentralized manner: (random) pairs of nodes repeatedly exchange their local parameter estimates and combine them by (weighted) averaging. We provide theoretical evidence and demonstrate experimentally that, under this protocol, nodes converge exponentially fast to the correct estimates in each M-step of the EM algorithm.

• Type: Conference
• Year: 2004
• Where: NIPS
• Authors: Wojtek Kowalczyk, Nikos A. Vlassis
• Keywords: EM Algorithm, Gossip-based Distributed Algorithm, NIPS 2004, NIPS 2007, Standard Em Algorithm