A multiscale neural network based on hierarchical nested bases
arXiv:1808.02376
Abstract
In recent years, deep learning has led to impressive results in many fields. In this paper, we introduce a multi-scale artificial neural network for high-dimensional non-linear maps based on the idea of hierarchical nested bases in the fast multipole method and the -matrices. This approach allows us to efficiently approximate discretized nonlinear maps arising from partial differential equations or integral equations. It also naturally extends our recent work based on the generalization of hierarchical matrices [Fan et al. arXiv:1807.01883] but with a reduced number of parameters. In particular, the number of parameters of the neural network grows linearly with the dimension of the parameter space of the discretized PDE. We demonstrate the properties of the architecture by approximating the solution maps of non-linear Schr{ö}dinger equation, the radiative transfer equation, and the Kohn-Sham map.
arXiv comment21 figures, 25 pages. arXiv admin note: text overlap with arXiv:1807.01883
References in corpus (16)
- Deep Learning in Neural Networks: An Overview
- A Survey on Deep Learning in Medical Image Analysis
- Sequence to Sequence Learning with Neural Networks
- DGM: A deep learning algorithm for solving partial differential equations
- Hidden Physics Models: Machine Learning of Nonlinear Partial Differential Equations
- Deep learning-based numerical methods for high-dimensional parabolic partial differential equations and backward stochastic differential equations
- Deep Image Prior
- A unified deep artificial neural network approach to partial differential equations in complex geometries
- Machine learning approximation algorithms for high-dimensional fully nonlinear partial differential equations and second-order backward stochastic differential equations
- Learning Functions: When Is Deep Better Than Shallow
- A machine learning approach for efficient uncertainty quantification using multiscale methods
- Fast algorithms for integral formulations of steady-state radiative transfer equation
- A multiscale neural network based on hierarchical matrices
- Deep Multiscale Model Learning
- Butterfly-Net: Optimal Function Representation Based on Convolutional Neural Networks
- A fast algorithm for radiative transport in isotropic media