Skip to main content

Linear and Parallel Learning for Markov Random Fields

Yariv Dror Mizrahi‚ Misha Denil and Nando de Freitas


We introduce a new embarrassingly parallel parameter learning algorithm for Markov random fields which is efficient for a large class of practical models. Our algorithm parallelizes naturally over cliques and, for graphs of bounded degree, its complexity is linear in the number of cliques. Unlike its competitors, our algorithm is fully parallel and for log-linear models it is also data efficient, requiring only the local sufficient statistics of the data to estimate parameters.

Book Title
International Conference on Machine Learning (ICML)