Defending Against Patch-based Backdoor Attacks on Self-Supervised Learning

被引:5
作者
Tejankar, Ajinkya [1 ]
Sanjabi, Maziar [2 ]
Wang, Qifan [2 ]
Wang, Sinong [2 ]
Firooz, Hamed [2 ]
Pirsiavash, Hamed [1 ]
Tan, Liang [2 ]
机构
[1] Univ Calif Davis, Davis, CA 95616 USA
[2] Meta AI, Delaware, OH USA
来源
2023 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR) | 2023年
关键词
D O I
10.1109/CVPR52729.2023.01178
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Recently, self-supervised learning (SSL) was shown to be vulnerable to patch-based data poisoning backdoor attacks. It was shown that an adversary can poison a small part of the unlabeled data so that when a victim trains an SSL model on it, the final model will have a backdoor that the adversary can exploit. This work aims to defend self-supervised learning against such attacks. We use a three-step defense pipeline, where we first train a model on the poisoned data. In the second step, our proposed defense algorithm (PatchSearch) uses the trained model to search the training data for poisoned samples and removes them from the training set. In the third step, a final model is trained on the cleaned-up training set. Our results show that PatchSearch is an effective defense. As an example, it improves a model's accuracy on images containing the trigger from 38.2% to 63.7% which is very close to the clean model's accuracy, 64.6%. Moreover, we show that PatchSearch outperforms baselines and state-of-the-art defense approaches including those using additional clean, trusted data. Our code is available at https://github.com/UCDvision/PatchSearch
引用
收藏
页码:12239 / 12249
页数:11
相关论文
共 56 条
  • [1] Abbasi Koohpayegani S., 2020, P ADV NEUR INF PROC, P12980
  • [2] Abbasi Koohpayegani Soroush, 2021, INT C COMP VIS ICCV
  • [3] STRONG DATA AUGMENTATION SANITIZES POISONING AND BACKDOOR ATTACKS WITHOUT AN ACCURACY TRADEOFF
    Borgnia, Eitan
    Cherepanova, Valeriia
    Fowl, Liam
    Ghiasi, Amin
    Geiping, Jonas
    Goldblum, Micah
    Goldstein, Tom
    Gupta, Arjun
    [J]. 2021 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING (ICASSP 2021), 2021, : 3855 - 3859
  • [4] Bossard L., 2014, EUR C COMP VIS ECCV
  • [5] Brown T., 2017, ADVERSARIAL PATCH
  • [6] Carlini Nicholas, 2022, INT C LEARN REPR ICL
  • [7] Caron M, 2020, ADV NEUR IN, V33
  • [8] Chen T, 2020, PR MACH LEARN RES, V119
  • [9] Safe Model-Free Optimal Voltage Control via Continuous-Time Zeroth-Order Methods
    Chen, Xin
    Poveda, Jorge, I
    Li, N.
    [J]. 2021 60TH IEEE CONFERENCE ON DECISION AND CONTROL (CDC), 2021, : 4064 - 4070
  • [10] Ferroptosis: machinery and regulation
    Chen, Xin
    Li, Jingbo
    Kang, Rui
    Klionsky, Daniel J.
    Tang, Daolin
    [J]. AUTOPHAGY, 2021, 17 (09) : 2054 - 2081