Robustness and Exploration of Variational and Machine Learning Approaches to Inverse Problems: An Overview

Alexander Auras¹, Kanchana Vaishnavi Gandikota¹, Hannah Droege²and Michael Moeller¹

¹Institute for Vision and Graphics, University of Siegen
²Institute of Computer Science, Rheinische Friedrich-Wilhelms-University Bonn

Official implementation of Robustness and Exploration of Variational and Machine Learning Approaches to Inverse Problems: An Overview.

Based off of the paper "Solving Inverse Problems With Deep Neural Networks - Robustness Included?" by M. Genzel, J. Macdonald, and M. März.
For their implementation see https://github.com/jmaces/robust-nets.

Installation

git clone https://github.com/AlexanderAuras/GAMM-Overview-23.git
cd GAMM-Overview-23
pip install .

Then simply run the jupyter notebook.

The used operator, data and model weights are available in the repository using git LFS or at https://drive.google.com/drive/folders/1nL_Z6gKyRRp36E58KUwNSZbO7Yj58chL?usp=drive_link.
The downloaded files location must be specified in the jupyter notebook. Alternatively you can use the generate_op.py and generate_data.py scripts (in src/gamm23/operators and src/gamm23/data) to generate a new operator/new data and train the models using the config.yaml file and the train.py script.