Explanations Can Be Manipulated and Geometry Is to Blame
Abstract
Explanation methods aim to make neural networks more trustworthy and interpretable. In this paper, we demonstrate a property of explanation methods which is disconcerting for both of these purposes. Namely, we show that explanations can be manipulated arbitrarily by applying visually hardly perceptible perturbations to the input that keep the network's output approximately constant. We establish theoretically that this phenomenon can be related to certain geometrical properties of neural networks. This allows us to derive an upper bound on the susceptibility of explanations to manipulations. Based on this result, we propose effective mechanisms to enhance the robustness of explanations.
Cite
Text
Dombrowski et al. "Explanations Can Be Manipulated and Geometry Is to Blame." Neural Information Processing Systems, 2019.Markdown
[Dombrowski et al. "Explanations Can Be Manipulated and Geometry Is to Blame." Neural Information Processing Systems, 2019.](https://mlanthology.org/neurips/2019/dombrowski2019neurips-explanations/)BibTeX
@inproceedings{dombrowski2019neurips-explanations,
title = {{Explanations Can Be Manipulated and Geometry Is to Blame}},
author = {Dombrowski, Ann-Kathrin and Alber, Maximillian and Anders, Christopher and Ackermann, Marcel and Müller, Klaus-Robert and Kessel, Pan},
booktitle = {Neural Information Processing Systems},
year = {2019},
pages = {13589-13600},
url = {https://mlanthology.org/neurips/2019/dombrowski2019neurips-explanations/}
}