Geometric Robustness of Deep Networks: Analysis and Improvement
Abstract
Deep convolutional neural networks have been shown to be vulnerable to arbitrary geometric transformations. However, there is no systematic method to measure the invariance properties of deep networks to such transformations. We propose ManiFool as a simple yet scalable algorithm to measure the invariance of deep networks. In particular, our algorithm measures the robustness of deep networks to geometric transformations in a worst-case regime as they can be problematic for sensitive applications. Our extensive experimental results show that ManiFool can be used to measure the invariance of fairly complex networks on high dimensional datasets and these values can be used for analyzing the reasons for it. Furthermore, we build on ManiFool to propose a new adversarial training scheme and we show its effectiveness on improving the invariance properties of deep neural networks.
Cite
Text
Kanbak et al. "Geometric Robustness of Deep Networks: Analysis and Improvement." Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2018. doi:10.1109/CVPR.2018.00467Markdown
[Kanbak et al. "Geometric Robustness of Deep Networks: Analysis and Improvement." Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2018.](https://mlanthology.org/cvpr/2018/kanbak2018cvpr-geometric/) doi:10.1109/CVPR.2018.00467BibTeX
@inproceedings{kanbak2018cvpr-geometric,
title = {{Geometric Robustness of Deep Networks: Analysis and Improvement}},
author = {Kanbak, Can and Moosavi-Dezfooli, Seyed-Mohsen and Frossard, Pascal},
booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
year = {2018},
doi = {10.1109/CVPR.2018.00467},
url = {https://mlanthology.org/cvpr/2018/kanbak2018cvpr-geometric/}
}