mlsanityData to 14 Oct 2026

Understanding Adversarial Examples from the Mutual Influence of Images and Perturbations

arXiv 2007.06189 ↗First posted 13 Jul 2020cs.CV cs.LGcited by 59

Not on the map. Why?

Abstract (arXiv)

A wide variety of works have explored the reason for the existence of adversarial examples, but there is no consensus on the explanation. We propose to treat the DNN logits as a vector for feature representation, and exploit them to analyze the mutual influence of two independent inputs based on the Pearson correlation coefficient (PCC). We utilize this vector representation to understand adversarial examples by disentangling the clean images and adversarial perturbations, and analyze their influence on each other. Our results suggest a new perspective towards the relationship between images and universal perturbations: Universal perturbations contain dominant features, and images behave like noise to them. This feature perspective leads to a new method for generating targeted universal adversarial perturbations using random source images. We are the first to achieve the challenging task of a targeted universal attack without utilizing original training data. Our approach using a proxy dataset achieves comparable performance to the state-of-the-art baselines which utilize the original training dataset.

Key facts. Cited by 59 papers in the map (data through 14 Oct 2026). First posted on arXiv 13 Jul 2020.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2021: 18; 2022: 10; 2023: 10; 2024: 11; 2025: 9; 2026: 1.

59 papers.

Sort:

By area:

  1. 2021Advances in adversarial attacks and defenses in computer vision: A surveycited by 127
  2. 2022Narcissus: A Practical Clean-Label Backdoor Attack with Limited Informationcited by 103
  3. 2021A Survey On Universal Adversarial Attackcited by 59
  4. 2021Adversarial Robustness Comparison of Vision Transformer and MLP-Mixer to CNNscited by 57
  5. 2021Targeted Attack against Deep Neural Networks via Flipping Limited Weight Bitscited by 49
  6. 2021Boosting Transferability of Targeted Adversarial Examples via Hierarchical Generative Networkscited by 32
  7. 2023Robustness of SAM: Segment Anything Under Corruptions and Beyondcited by 31
  8. 2022Enhancing the Self-Universality for Transferable Targeted Attackscited by 30
  9. 2022Investigating Top-k White-Box and Transferable Black-box Attackcited by 26
  10. 2023Continual Face Forgery Detection via Historical Distribution Preservingcited by 23
  11. 2023Enhancing Generalization of Universal Adversarial Perturbation through Gradient Aggregationcited by 22
  12. 2025X-Transfer Attacks: Towards Super Transferable Adversarial Attacks on CLIPcited by 21
38 more
  1. 2022Decoupled Adversarial Contrastive Learning for Self-supervised Adversarial Robustnesscited by 20
  2. 2023A Survey of Robustness and Safety of 2D and 3D Deep Learning Models Against Adversarial Attackscited by 19
  3. 2024CLIP-Guided Generative Networks for Transferable Targeted Adversarial Attackscited by 19
  4. 2021Universal Adversarial Perturbations Through the Lens of Deep Steganography: Towards A Fourier Perspectivecited by 18
  5. 2021Universal Adversarial Training with Class-Wise Perturbationscited by 13
  6. 2021Adversarial Visual Robustness by Causal Interventioncited by 13
  7. 2023Black-box Targeted Adversarial Attack on Segment Anything (SAM)cited by 13
  8. 2021Staircase Sign Method for Boosting Adversarial Attackscited by 12
  9. 2022Imperceptible Adversarial Attack via Invertible Neural Networkscited by 12
  10. 2021Motion-blurred Video Interpolation and Extrapolationcited by 11
  11. 2024Verification of Neural Networks' Global Robustnesscited by 10
  12. 2021A Simple and Strong Baseline for Universal Targeted Attacks on Siamese Visual Trackingcited by 9
  13. 2022Practical No-box Adversarial Attacks with Training-free Hybrid Image Transformationcited by 9
  14. 2024Texture Re-scalable Universal Adversarial Perturbationcited by 9
  15. 2024Scaling Laws for Black box Adversarial Attackscited by 8
  16. 2021Semantically Stealthy Adversarial Attacks against Segmentation Modelscited by 7
  17. 2021Graph-Based Similarity of Neural Network Representationscited by 7
  18. 2026Universal Image Immunization against Diffusion-based Image Editing via Semantic Injectioncited by 7
  19. 2022Consistent Attack: Universal Adversarial Perturbation on Embodied Vision Navigationcited by 6
  20. 2021Multiplicative Reweighting for Robust Neural Network Optimizationcited by 4
  21. 2022FG-UAP: Feature-Gathering Universal Adversarial Perturbationcited by 4
  22. 2023SAM Meets UAP: Attacking Segment Anything Model With Universal Adversarial Perturbationcited by 4
  23. 2025Improving Generalization of Universal Adversarial Perturbation via Dynamic Maximin Optimizationcited by 4
  24. 2025Unveiling and Mitigating Adversarial Vulnerabilities in Iterative Optimizerscited by 4
  25. 2021Evaluating Adversarial Attacks on ImageNet: A Reality Check on Misclassification Classescited by 3
  26. 2023AdaptGuard: Defending Against Universal Attacks for Model Adaptationcited by 3
  27. 2024The Fluorescent Veil: A Stealthy and Effective Physical Adversarial Patch Against Traffic Sign Recognitioncited by 3
  28. 2022Robust Universal Adversarial Perturbationscited by 2
  29. 2024Partner in Crime: Boosting Targeted Poisoning Attacks against Federated Learningcited by 2
  30. 2021Dominant Patterns: Critical Features Hidden in Deep Neural Networkscited by 1
  31. 2021On Distinctive Properties of Universal Perturbationscited by 1
  32. 2023Not So Robust After All: Evaluating the Robustness of Deep Neural Networks to Unseen Adversarial Attackscited by 1
  33. 2024Sparse and Transferable Universal Singular Vectors Attackcited by 1
  34. 2024Towards Assessing the Synthetic-to-Measured Adversarial Vulnerability of SAR ATRcited by 1
  35. 2024When Does Visual Prompting Outperform Linear Probing for Vision-Language Models? A Likelihood Perspectivecited by 1
  36. 2024S4ST: A Strong, Self-transferable, faSt, and Simple Scale Transformation for Transferable Targeted Attackcited by 1
  37. 2025Dual-Flow: Transferable Multi-Target, Instance-Agnostic Attacks via In-the-wild Cascading Flow Optimizationcited by 1
  38. 2025Democratic Training Against Universal Adversarial Perturbationscited by 1

50 most cited of 59 All 59 (JSON)