New graph polynomials from the Bethe approximation of the Ising partition function.
Yusuke WatanabeKenji FukumizuPublished in: CoRR (2009)
Keyphrases
- partition function
- belief propagation
- markov random field
- approximate inference
- upper bound
- closed form
- importance sampling
- random fields
- upper and lower bounds
- free energy
- message passing
- probability density function
- structured prediction
- bayesian framework
- graphical models
- exact computation
- loopy belief propagation
- maximum likelihood estimation
- lower bound
- parameter estimation
- bethe free energy
- special case
- variational methods
- conditional random fields
- structure learning
- bayesian networks
- directed acyclic graph
- probabilistic inference
- stereo matching
- monte carlo
- least squares
- gaussian process
- closed form solutions
- maximum margin
- latent variables
- regression model
- generative model
- graph cuts
- higher order
- image segmentation