PDE-GCN: Novel Architectures for Graph Neural Networks Motivated by Partial Differential Equations.
Moshe EliasofEldad HaberEran TreisterPublished in: CoRR (2021)
Keyphrases
- partial differential equations
- neural network
- level set
- image denoising
- anisotropic diffusion
- image processing
- numerical solution
- image enhancement
- fourth order
- pattern recognition
- finite difference
- numerical scheme
- finite difference method
- differential equations
- curve evolution
- numerical algorithms
- multiscale
- numerical methods
- energy functional
- weighted graph
- graph structure
- nonlinear diffusion
- denoising
- reaction diffusion
- heat equation
- diffusion equation
- mathematical morphology
- high order
- natural images
- image smoothing
- difference equations
- hamilton jacobi
- conservation laws
- object recognition
- ambrosio tortorelli