Understanding belief propagation and its generalizations

By Jonathan S. Yedidia, William T. Freeman, Yair Weiss. 2003

1,702 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Directional Message Passing for Molecular Graphs (2020, methodology): “hi = ∑ j∈Ni mji, and update the message mji based on the incoming messages mkj (Yedidia et al., 2003).”
Tracking People by Learning Their Appearance (2007, background): “An attractive strategy is to ignore the loops and pass local messages [37] and to represent those messages with a set of samples [38], [39].”
Statistical physics of inference: thresholds and algorithms (2015, background): “The connection between the Bethe approximation and belief propagation was noticed earlier in connection to error correcting codes by Kabashima and Saad [119], and in a more general setting by [233].”
MAP estimation via agreement on trees: message-passing and linear programming (2005, methodology): “Although the method may perform well in practice, it is no longer guaranteed to output the correct MAP assignment, and it is straightforward to demonstrate problems on which it speciﬁes an incorrect (i.e., non-optimal) assignment.”
Hierarchical Bayesian inference in the visual cortex. (2003, unknown): “Algorithms of this type, although with separate sets of particles for bottom-up and top-down messages, are under active investigation in the computer vision community 19,20,22 and constitute one of the most promising techniques for statistical inference in such large, multilayered domains.”
Collaborative Multiagent Reinforcement Learning by Payoff Propagation (2006, methodology): “The max-product algorithm (Pearl, 1988; Yedidia et al., 2003; Wainwright et al., 2004) is a popular method for computing the maximum a posteriori (MAP) configuration in an (unnormalized ) undirected graphical model.”
Sparse Unsupervised Capsules Generalize Better (2018, background): “, 2011)) of the entities being modelled, it is conceptually similar (but not numerically equivalent) to the concept of inference via message-passing-algorithms such as Belief Propagation (Yedidia et al., 2003), that can output marginal distributions over hidden variables at nodes in a graphical mode”
Face Photo-Sketch Synthesis and Recognition (2009, methodology): “We use belief propagation [20] to do inference.”
A high-bias, low-variance introduction to Machine Learning for physicists (2018, unknown): “, 1999), the more physics oriented discussion in (Yedidia, 2001; Yedidia et al., 2003), as well as David MacKay’s outstanding book (MacKay, 2003).”
Merlin: specification inference for explicit information flow problems (2009, methodology): “There are a wide variety of techniques for performing probabilistic inference on a factor graph and the sum-product algorithm [35] is the most practical algorithm among these.”
Motion estimation via hierarchical block matching and graph cut (2015, background): “Some well-known discrete optimization algorithms, such as graph cut [4] and belief propagation [5], can well handle this labeling problem.”
Fields of Experts (2009, background): “Variational techniques and related ones, such as belief propagation (Yedidia et al., 2003), have also enjoyed enormous popularity, both for MAP inference and computing marginals.”
Residual Splash for Optimally Parallelizing Belief Propagation (2009, methodology): “Nonetheless, BP on " loopy " graphs (often referred to as Loopy BP ) is used extensively with great success as an approximate inference algorithm [McEliece et al., 1998, Sun et al., 2003, Yedidia et al., 2003, Yanover and Weiss, 2002, Yanover et al., 2007].”
Cloud-Assisted Data Fusion and Sensor Selection for Internet of Things (2016, methodology): “In this phase, the base station performs the BP algorithm [8], [9] to infer incrementally the missing data due to the inactive sensor nodes and packet losses (further explained in Section V).”