Solving ill-posed inverse problems using iterative deep neural networks

@article{Adler2017SolvingII,
  title={Solving ill-posed inverse problems using iterative deep neural networks},
  author={J. Adler and O. {\"O}ktem},
  journal={ArXiv},
  year={2017},
  volume={abs/1704.04058}
}
We propose a partially learned approach for the solution of ill posed inverse problems with not necessarily linear forward operators. The method builds on ideas from classical regularization theory and recent advances in deep learning to perform learning while making use of prior information about the inverse problem encoded in the forward operator, noise model and a regularizing functional. The method results in a gradient-like iterative scheme, where the "gradient" component is learned using… Expand
253 Citations
Data-consistent neural networks for solving nonlinear inverse problems
  • 2
  • PDF
Learned Primal-Dual Reconstruction
  • J. Adler, O. Öktem
  • Computer Science, Medicine
  • IEEE Transactions on Medical Imaging
  • 2018
  • 257
  • PDF
Neural Networks-based Regularization of Large-Scale Inverse Problems in Medical Imaging
  • 1
  • Highly Influenced
  • PDF
Image Reconstruction by Splitting Deep Learning Regularization from Iterative Inversion
  • 8
  • PDF
One Network to Solve Them All — Solving Linear Inverse Problems Using Deep Projection Models
  • 214
  • PDF
NETT: Solving Inverse Problems with Deep Neural Networks
  • 65
  • PDF
Regularization of Inverse Problems by Neural Networks
  • 2
  • PDF
An inner-loop free solution to inverse problems using deep neural networks
  • 16
  • PDF
Graph Convolutional Networks for Model-Based Learning in Nonlinear Inverse Problems
  • Highly Influenced
  • PDF
...
1
2
3
4
5
...

References

SHOWING 1-10 OF 50 REFERENCES
One Network to Solve Them All — Solving Linear Inverse Problems Using Deep Projection Models
  • 214
  • PDF
Deep Convolutional Neural Network for Inverse Problems in Imaging
  • 911
  • PDF
Deep Learning Computed Tomography
  • 81
  • PDF
Iterative image reconstruction : a point of view
  • 46
  • PDF
ADMM-Net: A Deep Learning Approach for Compressive Sensing MRI
  • 55
  • PDF
Deep ADMM-Net for Compressive Sensing MRI
  • 485
  • PDF
Fast Tomographic Reconstruction From Limited Data Using Artificial Neural Networks
  • 70
  • PDF
Feature reconstruction in inverse problems
  • 48
  • PDF
A Deep Learning Architecture for Limited-Angle Computed Tomography Reconstruction
  • 43
  • PDF
A First-Order Primal-Dual Algorithm for Convex Problems with Applications to Imaging
  • A. Chambolle, T. Pock
  • Mathematics, Computer Science
  • Journal of Mathematical Imaging and Vision
  • 2010
  • 3,231
  • PDF
...
1
2
3
4
5
...