arXiv (Cornell University) · 2013 · 80 citations · 13 references
Natural SciencesComputer ScienceComputational BiologyComputational GeometryOptimizationApproximation AlgorithmsApproximation MethodsVariational FormulationVariational OptimizationGraphical ModelsMean-field ApproximationMean-field ModelsMean-field TheoryStatistical Potentials
Recently, variational approximations such as the mean field approximation\nhave received much interest. We extend the standard mean field method by using\nan approximating distribution that factorises into cluster potentials. This\nincludes undirected graphs, directed acyclic graphs and junction trees. We\nderive generalized mean field equations to optimize the cluster potentials. We\nshow that the method bridges the gap between the standard mean field\napproximation and the exact junction tree algorithm. In addition, we address\nthe problem of how to choose the graphical structure of the approximating\ndistribution. From the generalised mean field equations we derive rules to\nsimplify the structure of the approximating distribution in advance without\naffecting the quality of the approximation. We also show how the method fits\ninto some other variational approximations that are currently popular.\n
13
Introduction to the theory of neural computation
Neural Networks · 1994 · 6.5K citations
An Introduction to Variational Methods for Graphical Models
Michael I. Jordan, Zoubin Ghahramani, Tommi Jaakkola et al. · Machine Learning · 1999 · 3.8K citations · Full text
Graphical Models in Applied Multivariate Statistics
Colin Goodall, Joe Whittaker · Technometrics · 1991 · 1.8K citations