Progressive-Scale Boundary Blackbox Attack via Projective Gradient Estimation

Jiawei Zhang, Linyi Li, Huichen Li, Xiaolu Zhang, Shuang Yang, Bo Li
Proceedings of the 38th International Conference on Machine Learning, PMLR 139:12479-12490, 2021.

Abstract

Boundary based blackbox attack has been recognized as practical and effective, given that an attacker only needs to access the final model prediction. However, the query efficiency of it is in general high especially for high dimensional image data. In this paper, we show that such efficiency highly depends on the scale at which the attack is applied, and attacking at the optimal scale significantly improves the efficiency. In particular, we propose a theoretical framework to analyze and show three key characteristics to improve the query efficiency. We prove that there exists an optimal scale for projective gradient estimation. Our framework also explains the satisfactory performance achieved by existing boundary black-box attacks. Based on our theoretical framework, we propose Progressive-Scale enabled projective Boundary Attack (PSBA) to improve the query efficiency via progressive scaling techniques. In particular, we employ Progressive-GAN to optimize the scale of projections, which we call PSBA-PGAN. We evaluate our approach on both spatial and frequency scales. Extensive experiments on MNIST, CIFAR-10, CelebA, and ImageNet against different models including a real-world face recognition API show that PSBA-PGAN significantly outperforms existing baseline attacks in terms of query efficiency and attack success rate. We also observe relatively stable optimal scales for different models and datasets. The code is publicly available at https://github.com/AI-secure/PSBA.

Cite this Paper

BibTeX
@InProceedings{pmlr-v139-zhang21l, title = {Progressive-Scale Boundary Blackbox Attack via Projective Gradient Estimation}, author = {Zhang, Jiawei and Li, Linyi and Li, Huichen and Zhang, Xiaolu and Yang, Shuang and Li, Bo}, booktitle = {Proceedings of the 38th International Conference on Machine Learning}, pages = {12479--12490}, year = {2021}, editor = {Meila, Marina and Zhang, Tong}, volume = {139}, series = {Proceedings of Machine Learning Research}, month = {18--24 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v139/zhang21l/zhang21l.pdf}, url = {https://proceedings.mlr.press/v139/zhang21l.html}, abstract = {Boundary based blackbox attack has been recognized as practical and effective, given that an attacker only needs to access the final model prediction. However, the query efficiency of it is in general high especially for high dimensional image data. In this paper, we show that such efficiency highly depends on the scale at which the attack is applied, and attacking at the optimal scale significantly improves the efficiency. In particular, we propose a theoretical framework to analyze and show three key characteristics to improve the query efficiency. We prove that there exists an optimal scale for projective gradient estimation. Our framework also explains the satisfactory performance achieved by existing boundary black-box attacks. Based on our theoretical framework, we propose Progressive-Scale enabled projective Boundary Attack (PSBA) to improve the query efficiency via progressive scaling techniques. In particular, we employ Progressive-GAN to optimize the scale of projections, which we call PSBA-PGAN. We evaluate our approach on both spatial and frequency scales. Extensive experiments on MNIST, CIFAR-10, CelebA, and ImageNet against different models including a real-world face recognition API show that PSBA-PGAN significantly outperforms existing baseline attacks in terms of query efficiency and attack success rate. We also observe relatively stable optimal scales for different models and datasets. The code is publicly available at https://github.com/AI-secure/PSBA.} }
Endnote
%0 Conference Paper %T Progressive-Scale Boundary Blackbox Attack via Projective Gradient Estimation %A Jiawei Zhang %A Linyi Li %A Huichen Li %A Xiaolu Zhang %A Shuang Yang %A Bo Li %B Proceedings of the 38th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2021 %E Marina Meila %E Tong Zhang %F pmlr-v139-zhang21l %I PMLR %P 12479--12490 %U https://proceedings.mlr.press/v139/zhang21l.html %V 139 %X Boundary based blackbox attack has been recognized as practical and effective, given that an attacker only needs to access the final model prediction. However, the query efficiency of it is in general high especially for high dimensional image data. In this paper, we show that such efficiency highly depends on the scale at which the attack is applied, and attacking at the optimal scale significantly improves the efficiency. In particular, we propose a theoretical framework to analyze and show three key characteristics to improve the query efficiency. We prove that there exists an optimal scale for projective gradient estimation. Our framework also explains the satisfactory performance achieved by existing boundary black-box attacks. Based on our theoretical framework, we propose Progressive-Scale enabled projective Boundary Attack (PSBA) to improve the query efficiency via progressive scaling techniques. In particular, we employ Progressive-GAN to optimize the scale of projections, which we call PSBA-PGAN. We evaluate our approach on both spatial and frequency scales. Extensive experiments on MNIST, CIFAR-10, CelebA, and ImageNet against different models including a real-world face recognition API show that PSBA-PGAN significantly outperforms existing baseline attacks in terms of query efficiency and attack success rate. We also observe relatively stable optimal scales for different models and datasets. The code is publicly available at https://github.com/AI-secure/PSBA.
APA
Zhang, J., Li, L., Li, H., Zhang, X., Yang, S. & Li, B.. (2021). Progressive-Scale Boundary Blackbox Attack via Projective Gradient Estimation. Proceedings of the 38th International Conference on Machine Learning, in Proceedings of Machine Learning Research 139:12479-12490 Available from https://proceedings.mlr.press/v139/zhang21l.html.

Related Material