AutoZOOM: Autoencoder-based Zeroth Order Optimization Method for Attacking Black-box Neural Networks

@article{Tu2019AutoZOOMAZ,
  title={AutoZOOM: Autoencoder-based Zeroth Order Optimization Method for Attacking Black-box Neural Networks},
  author={Chun-Chen Tu and Pai-Shun Ting and Pin-Yu Chen and Sijia Liu and Huan Zhang and Jinfeng Yi and Cho-Jui Hsieh and Shin-Ming Cheng},
  journal={ArXiv},
  year={2019},
  volume={abs/1805.11770}
}
  • Chun-Chen Tu, Pai-Shun Ting, +5 authors Shin-Ming Cheng
  • Published 2019
  • Computer Science, Mathematics
  • ArXiv
  • Recent studies have shown that adversarial examples in state-of-the-art image classifiers trained by deep neural networks (DNN) can be easily generated when the target model is transparent to an attacker, known as the white-box setting. [...] Key Method Our framework, AutoZOOM, which is short for Autoencoder-based Zeroth Order Optimization Method, has two novel building blocks towards efficient black-box attacks: (i) an adaptive random gradient estimation strategy to balance query counts and distortion, and (ii…Expand Abstract

    Figures, Tables, and Topics from this paper.

    Citations

    Publications citing this paper.
    SHOWING 1-10 OF 79 CITATIONS, ESTIMATED 95% COVERAGE

    Query-Efficient Black-Box Attack by Active Learning

    QEBA: Query-Efficient Boundary-Based Blackbox Attack

    VIEW 1 EXCERPT
    CITES METHODS

    Parsimonious Black-Box Adversarial Attacks via Efficient Combinatorial Optimization

    VIEW 7 EXCERPTS
    CITES BACKGROUND, METHODS & RESULTS
    HIGHLY INFLUENCED

    Query-efficient Meta Attack to Deep Neural Networks

    VIEW 4 EXCERPTS
    CITES METHODS & BACKGROUND
    HIGHLY INFLUENCED

    On the Design of Black-Box Adversarial Examples by Leveraging Gradient-Free Optimization and Operator Splitting Method

    VIEW 2 EXCERPTS
    CITES BACKGROUND & METHODS

    Black-Box Adversarial Attack with Transferable Model-based Embedding

    VIEW 4 EXCERPTS
    CITES METHODS & BACKGROUND
    HIGHLY INFLUENCED

    FILTER CITATIONS BY YEAR

    2018
    2020

    CITATION STATISTICS

    • 15 Highly Influenced Citations

    • Averaged 26 Citations per year from 2018 through 2020

    • 18% Increase in citations per year in 2020 over 2019

    References

    Publications referenced by this paper.
    SHOWING 1-10 OF 46 REFERENCES

    Adversarial Machine Learning at Scale

    VIEW 1 EXCERPT