Approximate maximum entropy principles via Goemans-Williamson with applications to provable variational methods.
Andrej RisteskiYuanzhi LiPublished in: NIPS (2016)
Keyphrases
- maximum entropy
- variational methods
- maximum entropy principle
- approximate inference
- optic flow
- level set
- conditional random fields
- optical flow
- optical flow estimation
- exponential family
- graphical models
- curve evolution
- probabilistic inference
- pairwise
- belief propagation
- active contours
- motion analysis
- parameter estimation
- least squares
- natural language