On instabilities of deep learning in image reconstruction and the potential costs of AI
@article{Antun2020OnIO, title={On instabilities of deep learning in image reconstruction and the potential costs of AI}, author={Vegard Antun and Francesco Renna and C. Poon and B. Adcock and A. Hansen}, journal={Proceedings of the National Academy of Sciences}, year={2020}, volume={117}, pages={30088 - 30095} }
Deep learning, due to its unprecedented success in tasks such as image classification, has emerged as a new tool in image reconstruction with potential to change the field. In this paper, we demonstrate a crucial phenomenon: Deep learning typically yields unstable methods for image reconstruction. The instabilities usually occur in several forms: 1) Certain tiny, almost undetectable perturbations, both in the image and sampling domain, may result in severe artefacts in the reconstruction; 2) a… CONTINUE READING
Figures, Tables, and Topics from this paper
69 Citations
Interval Neural Networks as Instability Detectors for Image Reconstructions
- Computer Science, Engineering
- ArXiv
- 2020
- Highly Influenced
- PDF
Solving Inverse Problems With Deep Neural Networks - Robustness Included?
- Computer Science, Mathematics
- ArXiv
- 2020
- 5
- PDF
Improving Robustness of Deep-Learning-Based Image Reconstruction
- Computer Science, Mathematics
- ICML
- 2020
- 4
- Highly Influenced
- PDF
On the existence of stable and accurate neural networks for image reconstruction
- 2019
- 1
- Highly Influenced
- PDF
The troublesome kernel: why deep learning for inverse problems is typically unstable
- Computer Science
- ArXiv
- 2020
- 16
- PDF
Overcoming Measurement Inconsistency in Deep Learning for Linear Inverse Problems: Applications in Medical Imaging
- Computer Science, Engineering
- ArXiv
- 2020
- PDF
Artificial intelligence in image reconstruction: The change is here.
- Computer Science, Medicine
- Physica medica : PM : an international journal devoted to the applications of physics to medicine and biology : official journal of the Italian Association of Biomedical Physics
- 2020
Computed Tomography Reconstruction Using Deep Image Prior and Learned Reconstruction Methods
- Computer Science, Mathematics
- ArXiv
- 2020
- 7
- PDF
References
SHOWING 1-10 OF 54 REFERENCES
The Robustness of Deep Networks: A Geometrical Perspective
- Computer Science
- IEEE Signal Processing Magazine
- 2017
- 86
- PDF
DeepFool: A Simple and Accurate Method to Fool Deep Neural Networks
- Computer Science
- 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)
- 2016
- 1,980
- PDF
DAGAN: Deep De-Aliasing Generative Adversarial Networks for Fast Compressed Sensing MRI Reconstruction
- Medicine, Computer Science
- IEEE Transactions on Medical Imaging
- 2018
- 325
- PDF
Using Deep Neural Networks for Inverse Problems in Imaging: Beyond Analytical Methods
- Computer Science
- IEEE Signal Processing Magazine
- 2018
- 181
- PDF
One Network to Solve Them All — Solving Linear Inverse Problems Using Deep Projection Models
- Mathematics, Computer Science
- 2017 IEEE International Conference on Computer Vision (ICCV)
- 2017
- 202
- PDF
A Deep Cascade of Convolutional Neural Networks for MR Image Reconstruction
- Computer Science
- IPMI
- 2017
- 141
- PDF
MR Image Reconstruction Using Deep Density Priors
- Computer Science, Engineering
- IEEE Transactions on Medical Imaging
- 2019
- 38
- PDF
Deep Convolutional Neural Network for Inverse Problems in Imaging
- Computer Science, Mathematics
- IEEE Transactions on Image Processing
- 2017
- 837
- PDF
Geometric Robustness of Deep Networks: Analysis and Improvement
- Computer Science
- 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition
- 2018
- 63
- PDF