Block Belief Propagation for Parameter Learning in Markov Random Fields
This addresses efficiency issues for researchers and practitioners working with large graphical models, though it is incremental as it builds on existing belief propagation methods.
The paper tackles the problem of high iteration complexity scaling with graph size in traditional Markov random field training by proposing block belief propagation learning (BBPL), which uses block-coordinate updates to compute approximate gradients without full inference, and empirically demonstrates scalability improvements.
Traditional learning methods for training Markov random fields require doing inference over all variables to compute the likelihood gradient. The iteration complexity for those methods therefore scales with the size of the graphical models. In this paper, we propose \emph{block belief propagation learning} (BBPL), which uses block-coordinate updates of approximate marginals to compute approximate gradients, removing the need to compute inference on the entire graphical model. Thus, the iteration complexity of BBPL does not scale with the size of the graphs. We prove that the method converges to the same solution as that obtained by using full inference per iteration, despite these approximations, and we empirically demonstrate its scalability improvements over standard training methods.