Bypassing Backdoor Detection Algorithms in Deep Learning

被引:72
|
作者
Tan, Te Juin Lester [1 ]
Shokri, Reza [1 ]
机构
[1] Natl Univ Singapore NUS, Dept Comp Sci, Singapore, Singapore
关键词
D O I
10.1109/EuroSP48549.2020.00019
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
Deep learning models are vulnerable to various adversarial manipulations of their training data, parameters, and input sample. In particular, an adversary can modify the training data and model parameters to embed backdoors into the model, so the model behaves according to the adversary's objective if the input contains the backdoor features, referred to as the backdoor trigger (e.g., a stamp on an image). The poisoned model's behavior on clean data, however, remains unchanged. Many detection algorithms are designed to detect backdoors on input samples or model parameters, through the statistical difference between the latent representations of adversarial and clean input samples in the poisoned model. In this paper, we design an adversarial backdoor embedding algorithm that can bypass the existing detection algorithms including the state-of-the-art techniques. We design an adaptive adversarial training algorithm that optimizes the original loss function of the model, and also maximizes the indistinguishability of the hidden representations of poisoned data and clean data. This work calls for designing adversary-aware defense mechanisms for backdoor detection.
引用
收藏
页码:175 / 183
页数:9
相关论文
共 50 条
  • [31] Reverse Backdoor Distillation: Towards Online Backdoor Attack Detection for Deep Neural Network Models
    Yao, Zeming
    Zhang, Hangtao
    Guo, Yicheng
    Tian, Xin
    Peng, Wei
    Zou, Yi
    Zhang, Leo Yu
    Chen, Chao
    IEEE TRANSACTIONS ON DEPENDABLE AND SECURE COMPUTING, 2024, 21 (06) : 5098 - 5111
  • [32] Bypassing Stationary Points in Training Deep Learning Models
    Jung, Jaeheun
    Lee, Donghun
    IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS, 2024, 35 (12) : 18859 - 18871
  • [33] Smart Pothole Detection System using Deep Learning Algorithms
    Savita Chougule
    Alka Barhatte
    International Journal of Intelligent Transportation Systems Research, 2023, 21 : 483 - 492
  • [34] Development and optimization of image fire detection on deep learning algorithms
    Yang, Yi
    Pan, Mengyi
    Li, Pu
    Wang, Xuefeng
    Tsai, Yun-Ting
    JOURNAL OF THERMAL ANALYSIS AND CALORIMETRY, 2023, 148 (11) : 5089 - 5095
  • [35] Detection and Classification of Fabric Defects Using Deep Learning Algorithms
    Geze, Recep Ali
    Akbas, Ayhan
    JOURNAL OF POLYTECHNIC-POLITEKNIK DERGISI, 2024, 27 (01):
  • [36] Design of Objects Detection System using Deep Learning Algorithms
    Saidani, Taoufik
    Said, Yahia Fahem
    INTERNATIONAL JOURNAL OF COMPUTER SCIENCE AND NETWORK SECURITY, 2020, 20 (03): : 223 - 228
  • [37] A Review on Deep Learning Algorithms in the Detection of Autism Spectrum Disorder
    Lamani, Manjunath Ramanna
    Benadit, P. Julian
    FOURTH CONGRESS ON INTELLIGENT SYSTEMS, VOL 3, CIS 2023, 2024, 865 : 283 - 297
  • [38] Weakly Supervised Deep Learning for the Detection of Domain Generation Algorithms
    Yu, Bin
    Pan, Jie
    Gray, Daniel
    Hu, Jiaming
    Choudhary, Chhaya
    Nascimento, Anderson C. A.
    De Cock, Martine
    IEEE ACCESS, 2019, 7 : 51542 - 51556
  • [39] Intelligent phishing detection scheme using deep learning algorithms
    Adebowale, Moruf Akin
    Lwin, Khin T.
    Hossain, M. A.
    JOURNAL OF ENTERPRISE INFORMATION MANAGEMENT, 2023, 36 (03) : 747 - 766
  • [40] Lung Diseases Detection Using Various Deep Learning Algorithms
    Jasmine Pemeena Priyadarsini M.
    Kotecha K.
    Rajini G.K.
    Hariharan K.
    Utkarsh Raj K.
    Bhargav Ram K.
    Indragandhi V.
    Subramaniyaswamy V.
    Pandya S.
    Journal of Healthcare Engineering, 2023, 2023