LP-based Construction of DC Decompositions for Efficient Inference of Markov Random Fields.
Chaitanya MurtiDhruva KashyapChiranjib BhattacharyyaPublished in: AISTATS (2024)
Keyphrases
- efficient inference
- markov random field
- conditional random fields
- probabilistic inference
- higher order
- graph cuts
- parameter estimation
- energy minimization
- belief propagation
- pairwise
- maximum a posteriori
- fully connected
- approximate inference
- random fields
- potential functions
- human pose estimation
- energy function
- image restoration
- hidden variables
- exact inference
- structured prediction
- image segmentation
- np hard
- computer vision
- message passing
- graphical models
- graph structure
- least squares
- probabilistic graphical models