The Lazy Flipper: MAP Inference in Higher-Order Graphical Models by Depth-limited Exhaustive Search
Björn AndresJörg H. KappesUllrich KötheFred A. HamprechtPublished in: CoRR (2010)
Keyphrases
- map inference
- exhaustive search
- graphical models
- higher order
- markov random field
- conditional random fields
- belief propagation
- probabilistic model
- probabilistic graphical models
- probabilistic inference
- dynamic programming
- solution space
- computational complexity
- approximate inference
- random variables
- search space
- bayesian networks
- simulated annealing
- pairwise
- markov networks
- lower bound
- search methods
- structure learning
- image labeling
- graph structure
- optimal solution
- natural images
- graph cuts
- depth map
- semantic segmentation
- cutting plane
- np hard
- message passing
- branch and bound
- maximum a posteriori
- energy function
- topic models