Clamping Improves TRWand Mean Field Approximations
Proceedings of the 19th International Conference on Artificial Intelligence and Statistics
MetadataShow full item record
Weller, A., & Domke, J. (2016). Clamping Improves TRWand Mean Field Approximations. Proceedings of the 19th International Conference on Artificial Intelligence and Statistics, 38-46. https://doi.org/10.17863/CAM.270
This is the author accepted manuscript. The final version is available from Microtome Publishing via http://jmlr.org/proceedings/papers/v51/weller16a.html.
We examine the effect of clamping variables for approximate inference in undirected graphical models with pairwise relationships and discrete variables. For any number of variable labels, we demonstrate that clamping and summing approximate sub-partition functions can lead only to a decrease in the partition function estimate for TRW, and an increase for the naive mean field method, in each case guaranteeing an improvement in the approximation and bound. We next focus on binary variables, add the Bethe approximation to consideration and examine ways to choose good variables to clamp, introducing new methods. We show the importance of identifying highly frustrated cycles, and of checking the singleton entropy of a variable. We explore the value of our methods by empirical analysis and draw lessons to guide practitioners.
This record's DOI: https://doi.org/10.17863/CAM.270
This record's URL: https://www.repository.cam.ac.uk/handle/1810/256328