Connections between numerical algorithms for PDEs and neural networks

DOI10.1007/s10851-022-01106-xarXiv2107.14742OpenAlexW3187790213WikidataQ114226029 ScholiaQ114226029MaRDI QIDQ6156049

Tobias Alt, Matthias Augustin, Pascal Peter, Karl Schrader, Joachim Weickert

Publication date: 12 June 2023

Published in: Journal of Mathematical Imaging and Vision (Search for Journal in Brave)

Full work available at URL: https://arxiv.org/abs/2107.14742

zbMATH Keywords

stabilityneural networkspartial differential equationsnonlinear diffusionnumerical algorithms

Mathematics Subject Classification ID

Computer science (68-XX) Information and communication theory, circuits (94-XX)

Related Items

Learning phase field mean curvature flows with neural networks

Cites Work

• Unnamed Item
• Unnamed Item
• Unnamed Item
• Unnamed Item
• Unnamed Item
• Unnamed Item
• Unnamed Item
• Nonlinear total variation based noise removal algorithms
• Equivariant deep learning via morphological and linear scale space PDEs on the space of positions and orientations
• Translating numerical concepts for PDEs into neural architectures
• CLIP: cheap Lipschitz training of neural networks
• A multigrid platform for real-time motion computation with discontinuity-preserving variational methods
• Properties of higher order nonlinear diffusion filtering
• Minimizing total variation flow
• Relations between regularization and diffusion filtering
• Forward stability of ResNet and its variants
• Deep neural networks motivated by partial differential equations
• Residual networks as flows of diffeomorphisms
• Networks for nonlinear diffusion problems in imaging
• Regularisation of neural networks by enforcing Lipschitz continuity
• A theoretical analysis of deep neural networks and parametric PDEs
• Nonlinear approximation and (deep) ReLU networks
• A neural network multigrid solver for the Navier-Stokes equations
• Deep neural network structures solving variational inequalities
• PDE-Net 2.0: learning PDEs from data with a numeric-symbolic hybrid deep network
• Black-box learning of multigrid parameters
• Physics-informed neural networks: a deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations
• MgNet: a unified framework of multigrid and convolutional neural network
• Image compression with anisotropic diffusion
• Visualization and processing of tensor fields.
• Fourth-order partial differential equations for noise removal
• On the Internal Stability of Explicit,m-Stage Runge-Kutta Methods for Largem-Values
• Generalized Du Fort–Frankel Methods for Parabolic Initial-Boundary Value Problems
• Multi-Level Adaptive Solutions to Boundary-Value Problems
• Optimal signalling in attractor neural networks
• Another Look at Neural Multigrid
• A Multigrid Tutorial, Second Edition
• Stable architectures for deep neural networks
• Learning partial differential equations via data discovery and sparse optimization
• Layer-Parallel Training of Deep Residual Neural Networks
• Lipschitz Certificates for Layered Network Structures Driven by Averaged Activation Operators
• What Kinds of Functions Do Deep Neural Networks Learn? Insights from Variational Spline Theory
• On Lipschitz Bounds of General Convolutional Neural Networks
• Neural networks and physical systems with emergent collective computational abilities.
• Some methods of speeding up the convergence of iteration methods
• Image Processing and Analysis
• Multigrid Algorithm for High Order Denoising