A Differentiable Relaxation of Graph Segmentation and Alignment for AMR Parsing.
Chunchuan LyuShay B. CohenIvan TitovPublished in: EMNLP (1) (2021)
Keyphrases
- image parsing
- graph grammars
- objective function
- level set
- min cut max flow
- graph theoretic
- segmentation algorithm
- graph structure
- segmentation accuracy
- random walk
- object segmentation
- graph representation
- medical imaging
- normalized cut
- medical images
- natural language processing
- random walker
- image alignment
- region growing
- probabilistic relaxation
- graph theory
- graph matching
- multiscale
- loss function
- fully automatic
- segmentation method
- image segmentation
- chinese word segmentation
- min cut
- shape prior
- directed graph
- image regions
- connected components
- graph partitioning
- word segmentation
- graph cuts
- graph model
- graphical models
- image analysis
- weighted graph
- clustering algorithm