mlsanityData to 7 Oct 2026

Logit Pairing Methods Can Fool Gradient-Based Attacks

arXiv 1810.12042 ↗First posted 29 Oct 2018cs.LG cs.CR stat.MLcited by 27

Not on the map. Why?

Abstract (arXiv)

Recently, Kannan et al. [2018] proposed several logit regularization methods to improve the adversarial robustness of classifiers. We show that the computationally fast methods they propose - Clean Logit Pairing (CLP) and Logit Squeezing (LSQ) - just make the gradient-based optimization problem of crafting adversarial examples harder without providing actual robustness. We find that Adversarial Logit Pairing (ALP) may indeed provide robustness against adversarial examples, especially when combined with adversarial training, and we examine it in a variety of settings. However, the increase in adversarial accuracy is much smaller than previously claimed. Finally, our results suggest that the evaluation against an iterative PGD attack relies heavily on the parameters used and may result in false conclusions regarding robustness of a model.

Key facts. Cited by 27 papers in the map (data through 7 Oct 2026). First posted on arXiv 29 Oct 2018.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 10; 2022: 3; 2023: 5; 2024: 7; 2025: 1; 2026: 1.

27 papers.

Sort:

By area:

  1. 2021Improving Robustness using Generated Datacited by 233
  2. 2021Fixing Data Augmentation to Improve Adversarial Robustnesscited by 201
  3. 2021Data Augmentation Can Improve Robustnesscited by 193
  4. 2021On the effectiveness of adversarial training against common corruptionscited by 71
  5. 2022Sparsity Winning Twice: Better Robust Generalization from More Efficient Trainingcited by 41
  6. 2023DiffAttack: Evasion Attacks Against Diffusion-Based Adversarial Purificationcited by 36
  7. 2021Adversarial Robustness against Multiple and Single lp-Threat Models via Quick Fine-Tuning of Robust Classifierscited by 23
  8. 2023Adversarial Training Should Be Cast as a Non-Zero-Sum Gamecited by 13
  9. 2022Improving Robustness by Enhancing Weak Subnetscited by 12
  10. 2021Understanding Catastrophic Overfitting in Adversarial Trainingcited by 11
  11. 2024Constrained Adaptive Attack: Effective Adversarial Attack Against Deep Neural Networks for Tabular Datacited by 11
  12. 2024Towards Unified Robustness Against Both Backdoor and Adversarial Attackscited by 9
15 more
  1. 2021NoiLIn: Improving Adversarial Training and Correcting Stereotype of Noisy Labelscited by 8
  2. 2023Optimization and Optimizers for Adversarial Robustnesscited by 6
  3. 2021Constrained Gradient Descent: A Powerful and Principled Evasion Attack Against Neural Networkscited by 4
  4. 2024Deep Learning Model Security: Threats and Defensescited by 4
  5. 2022Optimization for Robustness Evaluation beyond ℓp Metricscited by 3
  6. 2021Adversarial Machine Learning in Text Analysis and Generationcited by 2
  7. 2021Understanding the Logit Distributions of Adversarially-Trained Deep Neural Networkscited by 2
  8. 2024Over-parameterization and Adversarial Robustness in Neural Networks: An Overview and Empirical Analysiscited by 2
  9. 2023Breaking Boundaries: Balancing Performance and Robustness in Deep Wireless Traffic Forecastingcited by 1
  10. 2024How to beat a Bayesian adversarycited by 1
  11. 2024Adversarial Robustness Overestimation and Instability in TRADEScited by 1
  12. 2025A General Framework for Property-Driven Machine Learningcited by 1
  13. 2023Cooperation or Competition: Avoiding Player Domination for Multi-Target Robustness via Adaptive Budgetscited by 0
  14. 2024Standard-Deviation-Inspired Regularization for Improving Adversarial Robustnesscited by 0
  15. 2026Quantitative Linear Logic for Neuro-Symbolic Learning and Verificationcited by 0