Submodular Relaxation for Inference in Markov Random Fields.
Anton OsokinDmitry P. VetrovPublished in: IEEE Trans. Pattern Anal. Mach. Intell. (2015)
Keyphrases
- markov random field
- energy minimization
- random fields
- loopy belief propagation
- map inference
- efficient inference
- high order
- graph cuts
- higher order
- piecewise constant functions
- dual decomposition
- belief propagation
- energy function
- partition function
- conditional random fields
- image segmentation
- image restoration
- parameter estimation
- maximum a posteriori
- objective function
- mrf model
- pairwise
- parameter learning
- undirected graphical models
- low level vision
- potential functions
- potts model
- message passing
- map estimation
- higher order cliques
- lp relaxation
- iterative conditional
- factor graphs
- structured prediction
- graphical models
- textured images
- bayesian networks
- discriminative random fields
- computer vision
- max flow
- bayesian inference
- generalized belief propagation
- maximum likelihood
- convex relaxation
- markov networks