This repository provides the code for our paper "When Explanations Lie: Why Many Modified BP Attributions Fail" under review for ICML.
To create an enviroment with the correct package versions:
$ conda create --name wel --file ./conda_list_output.txt
Additionally, you need to install the deeplift and innvestigate packages.
After you activated the wel conda enviroment:
$ cd <root dir>
$ pip install ./repos/deeplift ./repos/innvestigate
Finally, you have to specify the path of the ImageNet dataset in imagenet_dir.json.
Sources of the images for the two_classes notebook, both are under a free license:
Elefant & Zebra (Free for commercial use, No attribution required): https://pixabay.com/photos/zebra-elephant-africa-safari-3742242/
Dog & Cat (Creative Commons Zero - CC0): https://www.pxfuel.com/en/free-photo-emext