Multi-resolution Graph Neural Networks for PDE Approximation.
Wenzhuo LiuMouadh YagoubiMarc SchoenauerPublished in: ICANN (3) (2021)
Keyphrases
- multiresolution
- neural network
- partial differential equations
- graph structure
- hierarchical representation
- genetic algorithm
- error bounds
- directed graph
- structured data
- level set
- wavelet transform
- weighted graph
- fuzzy logic
- random walk
- artificial neural networks
- high order
- graph theory
- graph model
- anisotropic diffusion
- feed forward
- pattern recognition
- connected components
- neural network model
- image fusion
- bipartite graph
- closed form
- numerical scheme
- graph databases
- graph representation
- image magnification
- learning algorithm
- approximation error
- piecewise constant
- curve evolution
- graph theoretic
- graph partitioning
- computer vision
- directed acyclic graph
- training process
- image processing
- special case
- coarse to fine