Discrete gradient flow approximations of high dimensional evolution partial differential equations via deep neural networks.
Emmanuil H. GeorgoulisMichail LoulakisAsterios TsiourvasPublished in: Commun. Nonlinear Sci. Numer. Simul. (2023)
Keyphrases
- partial differential equations
- gradient flow
- difference equations
- neural network
- high dimensional
- energy functional
- image denoising
- level set
- numerical methods
- evolution equation
- anisotropic diffusion
- numerical solution
- image processing
- image enhancement
- differential equations
- curve evolution
- pattern recognition
- multiscale
- discrete data
- diffusion equation
- high order
- parameter space
- denoising
- global optimization
- active contours
- natural images
- pairwise
- variational formulation
- feature space
- image segmentation
- computer vision