/robustness-interpretability

Code for the Paper 'On the Connection Between Adversarial Robustness and Saliency Map Interpretability' by C. Etmann, S. Lunz, P. Maass, C.-B. Schönlieb

Primary LanguagePythonMIT LicenseMIT

Description

Code for the Paper 'On the Connection Between Adversarial Robustness and Saliency Map Interpretability' by C. Etmann, S. Lunz, P. Maass, C.-B. Schönlieb, accepted at ICML 2019.

More in-depth documentation to follow.

Execution

It can be run by executing e.g.

python3 adv_reg_experiment.py imagenet_example.ini

This code was tested using the following libraries in Python 3.6:

Tensorflow 1.11

Foolbox 1.8.0

Keras 2.2.2

Conda yml file to come.