Certifying Global Optimality of Graph Cuts via Semidefinite Relaxation: A Performance Guarantee for Spectral Clustering.
Shuyang LingThomas StrohmerPublished in: CoRR (2018)
Keyphrases
- spectral clustering
- graph cuts
- semidefinite
- convex relaxation
- globally optimal
- image segmentation
- multi label
- semidefinite programming
- markov random field
- belief propagation
- energy function
- energy minimization
- pairwise
- data clustering
- sufficient conditions
- clustering algorithm
- shape prior
- higher dimensional
- k means
- interior point methods
- finite dimensional
- unsupervised learning
- computer vision
- convex optimization
- linear programming