ML Anthology

Newscast EM

Wojtek Kowalczyk, Nikos Vlassis
NeurIPS 2004 pp. 713-720
/neurips/2004/kowalczyk2004neurips-newscast/

Abstract

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 com- bine them by (weighted) averaging. We provide theoretical evidence and demonstrate experimentally that, under this protocol, nodes converge ex- ponentially fast to the correct estimates in each M-step of the EM algo- rithm.

PDF NeurIPS Semantic Scholar

Cite

Text

Kowalczyk and Vlassis. "Newscast EM." Neural Information Processing Systems, 2004.

Markdown

[Kowalczyk and Vlassis. "Newscast EM." Neural Information Processing Systems, 2004.](https://mlanthology.org/neurips/2004/kowalczyk2004neurips-newscast/)

BibTeX

@inproceedings{kowalczyk2004neurips-newscast,
  title     = {{Newscast EM}},
  author    = {Kowalczyk, Wojtek and Vlassis, Nikos},
  booktitle = {Neural Information Processing Systems},
  year      = {2004},
  pages     = {713-720},
  url       = {https://mlanthology.org/neurips/2004/kowalczyk2004neurips-newscast/}
}