Physics-constrained Unsupervised Learning of Partial Differential Equations using Meshes.
Mike Yan MichelisRobert K. KatzschmannPublished in: CoRR (2022)
Keyphrases
- partial differential equations
- unsupervised learning
- anisotropic diffusion
- image denoising
- numerical solution
- image processing
- supervised learning
- level set
- multiresolution
- fourth order
- mathematical morphology
- numerical algorithms
- image enhancement
- numerical methods
- multiscale
- energy functional
- finite difference
- problems in image processing
- differential equations
- diffusion equation
- numerical scheme
- object recognition
- curve evolution
- machine learning
- expectation maximization
- semi supervised
- d mesh
- boundary value problem
- high order
- denoising
- conservation laws
- boundary element
- total variation
- natural images
- text classification
- dimensionality reduction
- higher order
- pattern recognition
- computer vision