PASCAL - Pattern Analysis, Statistical Modelling and Computational Learning

Cooled and Relaxed Survey Propagation for MRFs
Hai Leong Chieu, Wee Sun Lee and Yee Whye Teh
In: NIPS 2007, 03 Dec - 08 Dec 2007, Vancouver, Canada.


We describe a new algorithm, Relaxed Survey Propagation (RSP), for finding MAP configurations in Markov random fields. We compare its performance with state-of-the-art algorithms including the max-product belief propagation, its sequential tree-reweighted variant, residual (sum-product) belief propagation, and tree-structured expectation propagation. We show that it outperforms all approaches for Ising models with mixed couplings, as well as on a web disambiguation task formulated as a supervised clustering problem.

EPrint Type:Conference or Workshop Item (Paper)
Project Keyword:Project Keyword UNSPECIFIED
Subjects:Learning/Statistics & Optimisation
Theory & Algorithms
ID Code:3792
Deposited By:Yee Whye Teh
Deposited On:25 February 2008