mlsanityData to 7 Oct 2026

On Distinctive Properties of Universal Perturbations

arXiv 2112.15329 ↗First posted 31 Dec 2021cs.LG cs.CVcited by 1

Not on the map. Why?

Abstract (arXiv)

We identify properties of universal adversarial perturbations (UAPs) that distinguish them from standard adversarial perturbations. Specifically, we show that targeted UAPs generated by projected gradient descent exhibit two human-aligned properties: semantic locality and spatial invariance, which standard targeted adversarial perturbations lack. We also demonstrate that UAPs contain significantly less signal for generalization than standard adversarial perturbations -- that is, UAPs leverage non-robust features to a smaller extent than standard adversarial perturbations.

Key facts. Cited by 1 paper in the map (data through 7 Oct 2026). Builds on 33 of its 47 references that are in the map. First posted on arXiv 31 Dec 2021.

Builds on

33 of 47 references.

References · 33

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  2. 2014Very Deep Convolutional Networks for Large-Scale Image Recognitioncited by 15,267
  3. 2014ImageNet Large Scale Visual Recognition Challengecited by 8,441
  4. 2014Explaining and Harnessing Adversarial Examplescited by 7,287
  5. 2017Towards Deep Learning Models Resistant to Adversarial Attackscited by 6,326
  6. 2013Intriguing properties of neural networkscited by 5,114
  7. 2016"Why Should I Trust You?": Explaining the Predictions of Any Classifiercited by 5,040
  8. 2017Axiomatic Attribution for Deep Networkscited by 3,350
  9. 2016Towards Evaluating the Robustness of Neural Networkscited by 2,976
  10. 2013Deep Inside Convolutional Networks: Visualising Image Classification Models and Saliency Mapscited by 2,348
  11. 2018Sanity Checks for Saliency Mapscited by 1,149
  12. 2019Adversarial Examples Are Not Bugs, They Are Featurescited by 950
21 more
  1. 2018Robustness May Be at Odds with Accuracycited by 873
  2. 2016Universal adversarial perturbationscited by 774
  3. 2018Certified Robustness to Adversarial Examples with Differential Privacycited by 462
  4. 2019On Evaluating Adversarial Robustnesscited by 442
  5. 2016Transferability in Machine Learning: from Phenomena to Black-Box Attacks using Adversarial Samplescited by 439
  6. 2018Adversarially Robust Generalization Requires More Datacited by 318
  7. 2017Generative Adversarial Perturbationscited by 153
  8. 2017Universal Adversarial Perturbations Against Semantic Image Segmentationcited by 92
  9. 2016Robustness of classifiers: from adversarial to random noisecited by 87
  10. 2018Universal Adversarial Trainingcited by 85
  11. 2017Fast Feature Fool: A data independent approach to universal adversarial perturbationscited by 65
  12. 2020Understanding Adversarial Examples from the Mutual Influence of Images and Perturbationscited by 59
  13. 2017NAG: Network for Adversary Generationcited by 51
  14. 2017Art of singular vectors and universal adversarial perturbationscited by 49
  15. 2020CD-UAP: Class Discriminative Universal Adversarial Perturbationcited by 27
  16. 2018Defending Against Universal Perturbations With Shared Adversarial Trainingcited by 25
  17. 2020Double Targeted Universal Adversarial Perturbationscited by 23
  18. 2018With Friends Like These, Who Needs Adversaries?cited by 18
  19. 2017Robustness of classifiers to universal perturbations: a geometric perspectivecited by 14
  20. 2019Universal, transferable and targeted adversarial attackscited by 7
  21. 2019A Method for Computing Class-wise Universal Adversarial Perturbationscited by 5

Built on by

Cited by year

Citing papers per year: 2022: 1.

1 paper.

  1. 2022Universal Adversarial Directionscited by 1