Deep convolutional framelets: a general deep learning framework for inverse problems
From MaRDI portal
convolutional neural networkdeep learningframeletsinverse problemsperfect reconstruction conditionReLU
Nontrigonometric harmonic analysis involving wavelets and other special systems (42C40) Radon transform (44A12) Numerical methods for wavelets (65T60) Learning and adaptive systems in artificial intelligence (68T05) Biomedical imaging and signal processing (92C55) Image processing (compression, reconstruction, etc.) in information and communication theory (94A08) Signal theory (characterization, reconstruction, filtering, etc.) (94A12)
Abstract: Recently, deep learning approaches with various network architectures have achieved significant performance improvement over existing iterative reconstruction methods in various imaging problems. However, it is still unclear why these deep learning architectures work for specific inverse problems. To address these issues, here we show that the long-searched-for missing link is the convolution framelets for representing a signal by convolving local and non-local bases. The convolution framelets was originally developed to generalize the theory of low-rank Hankel matrix approaches for inverse problems, and this paper further extends the idea so that we can obtain a deep neural network using multilayer convolution framelets with perfect reconstruction (PR) under rectilinear linear unit nonlinearity (ReLU). Our analysis also shows that the popular deep network components such as residual block, redundant filter channels, and concatenated ReLU (CReLU) do indeed help to achieve the PR, while the pooling and unpooling layers should be augmented with high-pass branches to meet the PR condition. Moreover, by changing the number of filter channels and bias, we can control the shrinkage behaviors of the neural network. This discovery leads us to propose a novel theory for deep convolutional framelets neural network. Using numerical experiments with various inverse problems, we demonstrated that our deep convolution framelets network shows consistent improvement over existing deep architectures.This discovery suggests that the success of deep learning is not from a magical power of a black-box, but rather comes from the power of a novel signal representation using non-local basis combined with data-driven local basis, which is indeed a natural extension of classical signal processing theory.
Recommendations
- Deep synthesis network for regularizing inverse problems
- Deep learning for trivial inverse problems
- Deep null space learning for inverse problems: convergence analysis and rates
- NETT: solving inverse problems with deep neural networks
- Solving ill-posed inverse problems using iterative deep neural networks
Cites work
- 10.1162/153244303321897690
- A framelet-based image inpainting algorithm
- A Mathematical Theory of Deep Convolutional Neural Networks for Feature Extraction
- A Review of Image Denoising Algorithms, with a New One
- A tale of two bases: local-nonlocal regularization on image patches with convolution framelets
- Annihilating Filter-Based Low-Rank Hankel Matrix Approach for Image Inpainting
- Beyond a Gaussian Denoiser: Residual Learning of Deep CNN for Image Denoising
- Compressed sensing
- Compressive Sampling Using Annihilating Filter-Based Low-Rank Interpolation
- Convergence analysis of tight framelet approach for missing data recovery
- Convex analysis and monotone operator theory in Hilbert spaces
- Convolutional neural networks analyzed via convolutional sparse coding
- Deep Convolutional Neural Network for Inverse Problems in Imaging
- Framelets: MRA-based constructions of wavelet frames
- Group invariant scattering
- Hankel matrix rank minimization with applications to system identification and realization
- scientific article; zbMATH DE number 1465030 (Why is no real title available?)
- Image restoration: total variation, wavelet frames, and beyond
- Matrix pencil method for estimating parameters of exponentially damped/undamped sinusoids in noise
- Neural Network Learning
- Off-the-grid recovery of piecewise constant images from few Fourier samples
- Robust uncertainty principles: exact signal reconstruction from highly incomplete frequency information
- Sampling signals with finite rate of innovation
- Sparse and Low-Rank Decomposition of a Hankel Structured Matrix for Impulse Noise Removal
- Split Bregman methods and frame based image restoration
- Ten Lectures on Wavelets
Cited in
(24)- Big in Japan: regularizing networks for solving inverse problems
- Networks for nonlinear diffusion problems in imaging
- Deep learning on image denoising: an overview
- Image fusion network for dual-modal restoration
- Universal approximation with quadratic deep networks
- A review on deep learning in medical image reconstruction
- A finite element based deep learning solver for parametric PDEs
- Accelerating Proximal Markov Chain Monte Carlo by Using an Explicit Stabilized Method
- A mathematical framework for deep learning in elastic source imaging
- Wide-band butterfly network: stable and efficient inversion via multi-frequency neural networks
- Shallow neural networks for fluid flow reconstruction with limited sensors
- Solving inverse problems using data-driven models
- Efficient Training of Convolutional Deep Belief Networks in the Frequency Domain for Application to High-Resolution 2D and 3D Images
- A dataset-free deep learning method for low-dose CT image reconstruction
- Bilevel Methods for Image Reconstruction
- Directional mean curvature for textured image demixing
- Estimating a Potential Without the Agony of the Partition Function
- Wavelets in the deep learning era
- Vision graph U-Net: geometric learning enhanced encoder for medical image segmentation and restoration
- Deep unrolling networks with recurrent momentum acceleration for nonlinear inverse problems
- Solving the wide-band inverse scattering problem via equivariant neural networks
- The learned range test method for the inverse inclusion problem
- A learning-based hybrid reconstruction scheme for an inverse boundary value problem
- Fluid sensing using microcantilevers: from physics-based modeling to deep learning
Describes a project that uses
Uses Software
This page was built for publication: Deep convolutional framelets: a general deep learning framework for inverse problems
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4689753)
