Submodular relaxation for inference in Markov random fields.
Anton OsokinDmitry P. VetrovPublished in: CoRR (2015)
Keyphrases
- markov random field
- random fields
- energy minimization
- loopy belief propagation
- map inference
- efficient inference
- dual decomposition
- high order
- piecewise constant functions
- belief propagation
- parameter estimation
- higher order
- graph cuts
- partition function
- maximum a posteriori
- energy function
- objective function
- undirected graphical models
- potential functions
- image segmentation
- mrf model
- parameter learning
- pairwise
- higher order cliques
- factor graphs
- conditional random fields
- low level vision
- image restoration
- map estimation
- message passing
- iterative conditional
- graphical models
- probabilistic inference
- image labeling
- lp relaxation
- graphical structure
- potts model
- efficient inference algorithms
- textured images
- bayesian networks
- super resolution
- high level vision
- bayesian inference