Robust Proxy: Improving Adversarial Robustness by Robust Proxy Learning

被引:1
|
作者
Lee, Hong Joo [1 ]
Ro, Yong Man [1 ]
机构
[1] Korea Adv Inst Sci & Technol KAIST, Sch Elect Engn, Image & Video Syst Lab, Daejeon 34141, South Korea
关键词
Robust perturbation; class-wise robust perturbation; robust proxy learning;
D O I
10.1109/TIFS.2023.3288672
中图分类号
TP301 [理论、方法];
学科分类号
081202 ;
摘要
Recently, it has been widely known that deep neural networks are highly vulnerable and easily broken by adversarial attacks. To mitigate the adversarial vulnerability, many defense algorithms have been proposed. Recently, to improve adversarial robustness, many works try to enhance feature representation by imposing more direct supervision on the discriminative feature. However, existing approaches lack an understanding of learning adversarially robust feature representation. In this paper, we propose a novel training framework called Robust Proxy Learning. In the proposed method, the model explicitly learns robust feature representations with robust proxies. To this end, firstly, we demonstrate that we can generate class-representative robust features by adding class-wise robust perturbations. Then, we use the class representative features as robust proxies. With the class-wise robust features, the model explicitly learns adversarially robust features through the proposed robust proxy learning framework. Through extensive experiments, we verify that we can manually generate robust features, and our proposed learning framework could increase the robustness of the DNNs.
引用
收藏
页码:4021 / 4033
页数:13
相关论文
共 50 条
  • [1] Exploring Robust Features for Improving Adversarial Robustness
    Wang, Hong
    Deng, Yuefan
    Yoo, Shinjae
    Lin, Yuewei
    IEEE TRANSACTIONS ON CYBERNETICS, 2024, 54 (09) : 5141 - 5151
  • [2] On the Adversarial Robustness of Robust Estimators
    Lai, Lifeng
    Bayraktar, Erhan
    IEEE TRANSACTIONS ON INFORMATION THEORY, 2020, 66 (08) : 5097 - 5109
  • [3] Robust Bayesian inference in proxy SVARs
    Giacomini, Raffaella
    Kitagawa, Toru
    Read, Matthew
    JOURNAL OF ECONOMETRICS, 2022, 228 (01) : 107 - 126
  • [4] Adversarial Robust Deep Reinforcement Learning Requires Redefining Robustness
    Korkmaz, Ezgi
    THIRTY-SEVENTH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE, VOL 37 NO 7, 2023, : 8369 - 8377
  • [5] Is pollen size a robust proxy for moisture availability?
    Jardine, Phillip E.
    Lomax, Barry H.
    REVIEW OF PALAEOBOTANY AND PALYNOLOGY, 2017, 246 : 161 - 166
  • [6] IDENTIFYING ROBUST PROXY VARIABLES OF LATRINE USE: EXAMINING ACCESS TO IMPROVED SANITATION AS A PROXY
    Lopez, Velma
    Clarke, Philippa
    West, Brady
    Eisenberg, Joseph
    AMERICAN JOURNAL OF TROPICAL MEDICINE AND HYGIENE, 2017, 97 (05): : 27 - 27
  • [7] Robust shortcut and disordered robustness: Improving adversarial training through adaptive smoothing
    Li, Lin
    Spratling, Michael
    PATTERN RECOGNITION, 2025, 163
  • [8] Proxy-based robust deep metric learning in the presence of label noise
    Mohammed Neamah, Farah
    Aghdasi, Hadi S.
    Salehpour, Pedram
    Sokhandan Sorkhabi, Alireza
    PHYSICA SCRIPTA, 2024, 99 (07)
  • [9] Robust Graph Meta-Learning via Manifold Calibration with Proxy Subgraphs
    Wang, Zhenzhong
    Cao, Lulu
    Lin, Wanyu
    Jiang, Min
    Tan, Kay Chen
    THIRTY-SEVENTH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE, VOL 37 NO 12, 2023, : 15224 - 15232
  • [10] A robust proxy for production well placement optimization problems
    Pouladi, Behzad
    Keshavarz, Sahar
    Sharifi, Mohammad
    Ahmadi, Mohammad Ali
    FUEL, 2017, 206 : 467 - 481