Corpus ID: 4381036

Detecting Adversarial Examples via Neural Fingerprinting

@article{Dathathri2018DetectingAE,
  title={Detecting Adversarial Examples via Neural Fingerprinting},
  author={Sumanth Dathathri and Stephan Zheng and Richard M. Murray and Yisong Yue},
  journal={ArXiv},
  year={2018},
  volume={abs/1803.03870}
}
  • Sumanth Dathathri, Stephan Zheng, +1 author Yisong Yue
  • Published 2018
  • Computer Science, Mathematics
  • ArXiv
  • Deep neural networks are vulnerable to adversarial examples, which dramatically alter model output using small input changes. We propose Neural Fingerprinting, a simple, yet effective method to detect adversarial examples by verifying whether model behavior is consistent with a set of secret fingerprints, inspired by the use of biometric and cryptographic signatures. The benefits of our method are that 1) it is fast, 2) it is prohibitively expensive for an attacker to reverse-engineer which… CONTINUE READING

    Citations

    Publications citing this paper.

    References

    Publications referenced by this paper.
    SHOWING 1-10 OF 50 REFERENCES

    Characterizing Adversarial Subspaces Using Local Intrinsic Dimensionality

    VIEW 3 EXCERPTS

    Towards Evaluating the Robustness of Neural Networks

    VIEW 2 EXCERPTS

    The Limitations of Deep Learning in Adversarial Settings

    VIEW 1 EXCERPT