The Neural Network Approach to Inverse Problems in Differential Equations

2019/01/23 by Kailai Xu, Xu, Kailai, Eric Darve +1 · 5 citations
Earth and Planetary Sciences · Mathematics · Physics and Astronomy · #FOS: Mathematics #Model Reduction and Neural Networks #Numerical Analysis (math.NA) #Numerical methods in inverse problems #Seismic Imaging and Inversion Techniques

paper · pdf · doi:10.48550/arxiv.1901.07758

openalex publication_date 2019/01/23 · openalex created_date 2019/02/21 · openalex updated_date 2026/07/28

Abstract

We proposed a framework for solving inverse problems in differential equations based on neural networks and automatic differentiation. Neural networks are used to approximate hidden fields. We analyze the source of errors in the framework and derive an error estimate for a model diffusion equation problem. Besides, we propose a way for sensitivity analysis, utilizing the automatic differentiation mechanism embedded in the framework. It frees people from the tedious and error-prone process of deriving the gradients. Numerical examples exhibit consistency with the convergence analysis and error saturation is noteworthily predicted. We also demonstrate the unique benefits neural networks offer at the same time: universal approximation ability, regularizing the solution, bypassing the curse of dimensionality and leveraging efficient computing frameworks.

Citations

• Optical tomography in medical imaging
• ImageNet classification with deep convolutional neural networks
• MXNet: A Flexible and Efficient Machine Learning Library for Heterogeneous Distributed Systems
• Physics Informed Deep Learning (Part I): Data-driven Solutions of Nonlinear Partial Differential Equations
• fPINNs: Fractional Physics-Informed Neural Networks

Cited by

• A Memory Efficient Adjoint Method to Enable Billion Parameter Optimization on a Single GPU in Dynamic Problems
• Physics-informed neural networks for multiphysics data assimilation with application to subsurface transport
• Study of the asteroid Bennu using geodesyANNs and Osiris-Rex data
• Calibrating Multivariate Lévy Processes with Neural Networks
• A Convolutional Neural Network-based Approach to Field Reconstruction
• Solving Irregular and Data-enriched Differential Equations using Deep\n Neural Networks

Related

• Neural network augmented inverse problems for PDEs
• Deep Neural Network Approach to Forward-Inverse Problems
• Solving Inverse Problems in Steady-State Navier-Stokes Equations using Deep Neural Networks
• Neural Network approximation power on homogeneous and heterogeneous reaction-diffusion equations
• Machine Learning Methods for Autonomous Ordinary Differential Equations