Is Approximation Universally Defensive Against Adversarial Attacks in Deep Neural Networks?

被引:0
|
作者
Siddique, Ayesha [1 ]
Hoque, Khaza Anuarul [1 ]
机构
[1] Univ Missouri, Dept Elect Engn & Comp Sci, Columbia, MO 65211 USA
关键词
Adversarial Attacks; Adversarial Robustness; Approximate Computing; Deep Neural Networks;
D O I
暂无
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
Approximate computing is known for its effectiveness in improvising the energy efficiency of deep neural network (DNN) accelerators at the cost of slight accuracy loss. Very recently, the inexact nature of approximate components, such as approximate multipliers have also been reported successful in defending adversarial attacks on DNNs models. Since the approximation errors traverse through the DNN layers as masked or unmasked, this raises a key research question-can approximate computing always offer a defense against adversarial attacks in DNNs, i.e., are they universally defensive? Towards this, we present an extensive adversarial robustness analysis of different approximate DNN accelerators (AxDNNs) using the state-of-the-art approximate multipliers. In particular, we evaluate the impact of ten adversarial attacks on different AxDNNs using the MNIST and CIFAR-10 datasets. Our results demonstrate that adversarial attacks on AxDNNs can cause 53% accuracy loss whereas the same attack may lead to almost no accuracy loss (as low as 0.06%) in the accurate DNN. Thus, approximate computing cannot be referred to as a universal defense strategy against adversarial attacks.
引用
收藏
页码:364 / 369
页数:6
相关论文
共 50 条
  • [1] Defensive Dropout for Hardening Deep Neural Networks under Adversarial Attacks
    Wang, Siyue
    Wang, Xiao
    Zhao, Pu
    Wen, Wujie
    Kaeli, David
    Chin, Peter
    Lin, Xue
    [J]. 2018 IEEE/ACM INTERNATIONAL CONFERENCE ON COMPUTER-AIDED DESIGN (ICCAD) DIGEST OF TECHNICAL PAPERS, 2018,
  • [2] Defending Against Adversarial Attacks in Deep Neural Networks
    You, Suya
    Kuo, C-C Jay
    [J]. ARTIFICIAL INTELLIGENCE AND MACHINE LEARNING FOR MULTI-DOMAIN OPERATIONS APPLICATIONS, 2019, 11006
  • [3] A survey on the vulnerability of deep neural networks against adversarial attacks
    Andy Michel
    Sumit Kumar Jha
    Rickard Ewetz
    [J]. Progress in Artificial Intelligence, 2022, 11 : 131 - 141
  • [4] Adversarial Attacks and Defenses Against Deep Neural Networks: A Survey
    Ozdag, Mesut
    [J]. CYBER PHYSICAL SYSTEMS AND DEEP LEARNING, 2018, 140 : 152 - 161
  • [5] A survey on the vulnerability of deep neural networks against adversarial attacks
    Michel, Andy
    Jha, Sumit Kumar
    Ewetz, Rickard
    [J]. PROGRESS IN ARTIFICIAL INTELLIGENCE, 2022, 11 (02) : 131 - 141
  • [6] Evolving Hyperparameters for Training Deep Neural Networks against Adversarial Attacks
    Liu, Jia
    Jin, Yaochu
    [J]. 2019 IEEE SYMPOSIUM SERIES ON COMPUTATIONAL INTELLIGENCE (IEEE SSCI 2019), 2019, : 1778 - 1785
  • [7] MRobust: A Method for Robustness against Adversarial Attacks on Deep Neural Networks
    Liu, Yi-Ling
    Lomuscio, Alessio
    [J]. 2020 INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS (IJCNN), 2020,
  • [8] Efficacy of Defending Deep Neural Networks against Adversarial Attacks with Randomization
    Zhou, Yan
    Kantarcioglu, Murat
    Xi, Bowei
    [J]. ARTIFICIAL INTELLIGENCE AND MACHINE LEARNING FOR MULTI-DOMAIN OPERATIONS APPLICATIONS II, 2020, 11413
  • [9] Detect Adversarial Attacks Against Deep Neural Networks With GPU Monitoring
    Zoppi, Tommaso
    Ceccarelli, Andrea
    [J]. IEEE ACCESS, 2021, 9 : 150579 - 150591
  • [10] Watermarking-based Defense against Adversarial Attacks on Deep Neural Networks
    Li, Xiaoting
    Chen, Lingwei
    Zhang, Jinquan
    Larus, James
    Wu, Dinghao
    [J]. 2021 INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS (IJCNN), 2021,