Exploring Robust Features for Improving Adversarial Robustness

Wang, Hong; Deng, Yuefan; Yoo, Shinjae; Lin, Yuewei

doi:10.1109/tcyb.2024.3380437

Exploring Robust Features for Improving Adversarial Robustness

Journal Article · Tue Apr 09 00:00:00 EDT 2024 · IEEE Transactions on Cybernetics

DOI:https://doi.org/10.1109/tcyb.2024.3380437· OSTI ID:2368822

^[1]; Deng, Yuefan ^[2]; ^[1]; ^[1]

Brookhaven National Laboratory (BNL), Upton, NY (United States)
Stony Brook Univ., NY (United States)

While deep neural networks (DNNs) have revolutionized many fields, their fragility to carefully designed adversarial attacks impedes the usage of DNNs in safety-critical applications. In this article, we strive to explore the robust features that are not affected by the adversarial perturbations, that is, invariant to the clean image and its adversarial examples (AEs), to improve the model’s adversarial robustness. Specifically, we propose a feature disentanglement model to segregate the robust features from nonrobust features and domain-specific features. Here, the extensive experiments on five widely used datasets with different attacks demonstrate that robust features obtained from our model improve the model’s adversarial robustness compared to the state-of-the-art approaches. Moreover, the trained domain discriminator is able to identify the domain-specific features from the clean images and AEs almost perfectly. This enables AE detection without incurring additional computational costs. With that, we can also specify different classifiers for clean images and AEs, thereby avoiding any drop in clean image accuracy.

View Accepted Manuscript (DOE)

Research Organization:: Brookhaven National Laboratory (BNL), Upton, NY (United States)

Sponsoring Organization:: USDOE Office of Science (SC), Advanced Scientific Computing Research (ASCR)

Grant/Contract Number:: SC0012704

OSTI ID:: 2368822

Report Number(s):: BNL--225592-2024-JAAM

Journal Information:: IEEE Transactions on Cybernetics, Journal Name: IEEE Transactions on Cybernetics Journal Issue: 9 Vol. 54; ISSN 2168-2267

Publisher:: IEEECopyright Statement

Country of Publication:: United States

Language:: English

References (37)

Square Attack: A Query-Efficient Black-Box Adversarial Attack via Random Search Andriushchenko, Maksym; Croce, Francesco; Flammarion, Nicolas Computer Vision – ECCV 2020 https://doi.org/10.1007/978-3-030-58592-1_29	book	January 2020
Identity Mappings in Deep Residual Networks He, Kaiming; Zhang, Xiangyu; Ren, Shaoqing Computer Vision – ECCV 2016 https://doi.org/10.1007/978-3-319-46493-0_38	book	January 2016
DRIT++: Diverse Image-to-Image Translation via Disentangled Representations Lee, Hsin-Ying; Tseng, Hung-Yu; Mao, Qi International Journal of Computer Vision, Vol. 128, Issue 10-11 https://doi.org/10.1007/s11263-019-01284-z	journal	February 2020
DeepFool: A Simple and Accurate Method to Fool Deep Neural Networks Moosavi-Dezfooli, Seyed-Mohsen; Fawzi, Alhussein; Frossard, Pascal 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) https://doi.org/10.1109/CVPR.2016.282	conference	June 2016
Robust Physical-World Attacks on Deep Learning Visual Classification Eykholt, Kevin; Evtimov, Ivan; Fernandes, Earlence 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition https://doi.org/10.1109/CVPR.2018.00175	conference	June 2018
Geometric Robustness of Deep Networks: Analysis and Improvement Kanbak, Can; Moosavi-Dezfooli, Seyed-Mohsen; Frossard, Pascal 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition https://doi.org/10.1109/CVPR.2018.00467	conference	June 2018
Boosting Adversarial Attacks with Momentum Dong, Yinpeng; Liao, Fangzhou; Pang, Tianyu 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition https://doi.org/10.1109/CVPR.2018.00957	conference	June 2018
Feature Denoising for Improving Adversarial Robustness Xie, Cihang; Wu, Yuxin; Maaten, Laurens van der 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) https://doi.org/10.1109/CVPR.2019.00059	conference	June 2019
Universal Physical Camouflage Attacks on Object Detectors Huang, Lifeng; Gao, Chengying; Zhou, Yuyin 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) https://doi.org/10.1109/CVPR42600.2020.00080	conference	June 2020
Cross-Domain Face Presentation Attack Detection via Multi-Domain Disentangled Representation Learning Wang, Guoqing; Han, Hu; Shan, Shiguang 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) https://doi.org/10.1109/CVPR42600.2020.00671	conference	June 2020
Learning to Learn Single Domain Generalization Qiao, Fengchun; Zhao, Long; Peng, Xi 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) https://doi.org/10.1109/CVPR42600.2020.01257	conference	June 2020
DRANet: Disentangling Representation and Adaptation Networks for Unsupervised Cross-Domain Adaptation Lee, Seunghun; Cho, Sunghyun; Im, Sunghoon 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) https://doi.org/10.1109/CVPR46437.2021.01500	conference	June 2021
LAS-AT: Adversarial Training with Learnable Attack Strategy Jia, Xiaojun; Zhang, Yong; Wu, Baoyuan 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) https://doi.org/10.1109/CVPR52688.2022.01304	conference	June 2022
Enhancing Adversarial Training with Second-Order Statistics of Weights Jin, Gaojie; Yi, Xinping; Huang, Wei 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) https://doi.org/10.1109/CVPR52688.2022.01484	conference	June 2022
Grad-CAM: Visual Explanations from Deep Networks via Gradient-Based Localization Selvaraju, Ramprasaath R.; Cogswell, Michael; Das, Abhishek 2017 IEEE International Conference on Computer Vision (ICCV) https://doi.org/10.1109/ICCV.2017.74	conference	October 2017
Adversarial Learning With Margin-Based Triplet Embedding Regularization Zhong, Yaoyao; Deng, Weihong 2019 IEEE/CVF International Conference on Computer Vision (ICCV) https://doi.org/10.1109/ICCV.2019.00665	conference	October 2019
AGKD-BML: Defense Against Adversarial Attack by Attention Guided Knowledge Distillation and Bi-directional Metric Learning Wang, Hong; Deng, Yuefan; Yoo, Shinjae 2021 IEEE/CVF International Conference on Computer Vision (ICCV) https://doi.org/10.1109/ICCV48922.2021.00756	conference	October 2021
Learnable Boundary Guided Adversarial Training Cui, Jiequan; Liu, Shu; Wang, Liwei 2021 IEEE/CVF International Conference on Computer Vision (ICCV) https://doi.org/10.1109/ICCV48922.2021.01543	conference	October 2021
Adversarial Finetuning with Latent Representation Constraint to Mitigate Accuracy-Robustness Tradeoff Suzuki, Satoshi; Yamaguchi, Shin’ya; Takeda, Shoichiro 2023 IEEE/CVF International Conference on Computer Vision (ICCV) https://doi.org/10.1109/ICCV51070.2023.00405	conference	October 2023
Towards Evaluating the Robustness of Neural Networks Carlini, Nicholas; Wagner, David 2017 IEEE Symposium on Security and Privacy (SP) https://doi.org/10.1109/SP.2017.49	conference	May 2017
Adversarial Examples for Hamming Space Search Yang, Erkun; Liu, Tongliang; Deng, Cheng IEEE Transactions on Cybernetics, Vol. 50, Issue 4 https://doi.org/10.1109/TCYB.2018.2882908	journal	April 2020
ROSA: Robust Salient Object Detection Against Adversarial Attacks Li, Haofeng; Li, Guanbin; Yu, Yizhou IEEE Transactions on Cybernetics, Vol. 50, Issue 11 https://doi.org/10.1109/TCYB.2019.2914099	journal	November 2020
Toward a Controllable Disentanglement Network Song, Zengjie; Koyejo, Oluwasanmi; Zhang, Jiangshe IEEE Transactions on Cybernetics, Vol. 52, Issue 4 https://doi.org/10.1109/TCYB.2020.3000480	journal	April 2022
Daedalus: Breaking Nonmaximum Suppression in Object Detection via Adversarial Examples Wang, Derui; Li, Chaoran; Wen, Sheng IEEE Transactions on Cybernetics, Vol. 52, Issue 8 https://doi.org/10.1109/TCYB.2020.3041481	journal	August 2022
Adversarial CAPTCHAs Shi, Chenghui; Xu, Xiaogang; Ji, Shouling IEEE Transactions on Cybernetics, Vol. 52, Issue 7 https://doi.org/10.1109/TCYB.2021.3071395	journal	July 2022
Point Adversarial Self-Mining: A Simple Method for Facial Expression Recognition Liu, Ping; Lin, Yuewei; Meng, Zibo IEEE Transactions on Cybernetics, Vol. 52, Issue 12 https://doi.org/10.1109/TCYB.2021.3085744	journal	December 2022
D-BIN: A Generalized Disentangling Batch Instance Normalization for Domain Adaptation Chen, Yurong; Zhang, Hui; Wang, Yaonan IEEE Transactions on Cybernetics, Vol. 53, Issue 4 https://doi.org/10.1109/TCYB.2021.3110128	journal	April 2023
Joint Adversarial Example and False Data Injection Attacks for State Estimation in Power Systems Tian, Jiwei; Wang, Buhong; Wang, Zhen IEEE Transactions on Cybernetics, Vol. 52, Issue 12 https://doi.org/10.1109/TCYB.2021.3125345	journal	December 2022
Adversarial Learning With Cost-Sensitive Classes Shen, Haojing; Chen, Sihong; Wang, Ran IEEE Transactions on Cybernetics, Vol. 53, Issue 8 https://doi.org/10.1109/TCYB.2022.3146388	journal	August 2023
A Feature Space-Restricted Attention Attack on Medical Deep Learning Systems Wang, Zizhou; Shu, Xin; Wang, Yan IEEE Transactions on Cybernetics, Vol. 53, Issue 8 https://doi.org/10.1109/TCYB.2022.3209175	journal	August 2023
Defense against Adversarial Cloud Attack on Remote Sensing Salient Object Detection Sun, Huiming; Fu, Lan; Li, Jinlong 2024 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) https://doi.org/10.1109/WACV57701.2024.00816	conference	January 2024
Adversarially Robust Distillation Goldblum, Micah; Fowl, Liam; Feizi, Soheil Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 34, Issue 04 https://doi.org/10.1609/aaai.v34i04.5816	journal	April 2020
Adversarial Robustness through Disentangled Representations Yang, Shuo; Guo, Tianyu; Wang, Yunhe Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 35, Issue 4 https://doi.org/10.1609/aaai.v35i4.16424	journal	May 2021
Efficient Robust Training via Backward Smoothing Chen, Jinghui; Cheng, Yu; Gan, Zhe Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 36, Issue 6 https://doi.org/10.1609/aaai.v36i6.20571	journal	June 2022
Generalizing to Unseen Domains: A Survey on Domain Generalization Wang, Jindong; Lan, Cuiling; Liu, Chang Proceedings of the Thirtieth International Joint Conference on Artificial Intelligence https://doi.org/10.24963/ijcai.2021/628	conference	August 2021
Towards Deep Learning Models Resistant to Adversarial Attacks Madry, Aleksander; Makelov, Aleksandar; Schmidt, Ludwig arXiv https://doi.org/10.48550/ARXIV.1706.06083	preprint	January 2017
Manitest: Are classifiers really invariant? Fawzi, Alhussein; Frossard, Pascal Procedings of the British Machine Vision Conference 2015 https://doi.org/10.5244/C.29.106	conference	January 2015

Similar Records

XploreNAS: Explore Adversarially Robust and Hardware-efficient Neural Architectures for Non-ideal Xbars

Journal Article · Sun Jul 23 20:00:00 EDT 2023 · ACM Transactions on Embedded Computing Systems · OSTI ID:2422212

Towards Query-Efficient Black-Box Adversary with Zeroth-Order Natural Gradient Descent

Conference · Fri Apr 03 00:00:00 EDT 2020 · Proceedings of the AAAI Conference on Artificial Intelligence · OSTI ID:1958810

The Effects of Compounded Model Size Reductions on Adversarial Robustness

Conference · Tue Apr 01 00:00:00 EDT 2025 · OSTI ID:3002675

Related Subjects

97 MATHEMATICS AND COMPUTING
Adversarial attacks
Disentangle
Robustness

Exploring Robust Features for Improving Adversarial Robustness

Citation Formats

References (37)

Similar Records

Related Subjects