mlsanityData to 7 Oct 2026

Efficient Adversarial Training With Data Pruning

arXiv 2207.00694 ↗First posted 1 Jul 2022cs.LGcited by 7

Not on the map. Why?

Abstract (arXiv)

Neural networks are susceptible to adversarial examples-small input perturbations that cause models to fail. Adversarial training is one of the solutions that stops adversarial examples; models are exposed to attacks during training and learn to be resilient to them. Yet, such a procedure is currently expensive-it takes a long time to produce and train models with adversarial samples, and, what is worse, it occasionally fails. In this paper we demonstrate data pruning-a method for increasing adversarial training efficiency through data sub-sampling.We empirically show that data pruning leads to improvements in convergence and reliability of adversarial training, albeit with different levels of utility degradation. For example, we observe that using random sub-sampling of CIFAR10 to drop 40% of data, we lose 8% adversarial accuracy against the strongest attackers, while by using only 20% of data we lose 14% adversarial accuracy and reduce runtime by a factor of 3. Interestingly, we discover that in some settings data pruning brings benefits from both worlds-it both improves adversarial accuracy and training time.

Key facts. Cited by 7 papers in the map (data through 7 Oct 2026). Builds on 39 of its 48 references that are in the map. First posted on arXiv 1 Jul 2022.

Builds on

39 of 48 references.

References · 39

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  2. 2014Explaining and Harnessing Adversarial Examplescited by 7,287
  3. 2017Towards Deep Learning Models Resistant to Adversarial Attackscited by 6,326
  4. 2013Intriguing properties of neural networkscited by 5,114
  5. 2016Overcoming catastrophic forgetting in neural networkscited by 4,718
  6. 2020Reliable evaluation of adversarial robustness with an ensemble of diverse parameter-free attackscited by 1,442
  7. 2019Adversarial Examples Are Not Bugs, They Are Featurescited by 950
  8. 2018Robustness May Be at Odds with Accuracycited by 873
  9. 2020Fast is better than free: Revisiting adversarial trainingcited by 723
  10. 2017Evasion Attacks against Machine Learning at Test Timecited by 686
  11. 2015Distillation as a Defense to Adversarial Perturbations against Deep Neural Networkscited by 675
  12. 2016Making Deep Neural Networks Robust to Label Noise: a Loss Correction Approachcited by 663
27 more
  1. 2019Adversarial Training for Free!cited by 638
  2. 2017Adversarial Examples Are Not Easily Detected: Bypassing Ten Detection Methodscited by 542
  3. 2019On Evaluating Adversarial Robustnesscited by 442
  4. 2021Pervasive Label Errors in Test Sets Destabilize Machine Learning Benchmarkscited by 388
  5. 2018Feature Denoising for Improving Adversarial Robustnesscited by 312
  6. 2021Unsolved Problems in ML Safetycited by 291
  7. 2018Audio Adversarial Examples: Targeted Attacks on Speech-to-Textcited by 268
  8. 2018Adversarial Attacks and Defences: A Surveycited by 259
  9. 2017Detecting Adversarial Samples from Artifactscited by 244
  10. 2018Adversarial Logit Pairingcited by 224
  11. 2019Adversarial Policies: Attacking Deep Reinforcement Learningcited by 209
  12. 2018Shield: Fast, Practical Defense and Vaccination for Deep Learning using JPEG Compressioncited by 74
  13. 2021Bad Characters: Imperceptible NLP Attackscited by 69
  14. 2016Defensive Distillation is Not Robust to Adversarial Examplescited by 60
  15. 2017Towards Crafting Text Adversarial Samplescited by 55
  16. 2021Active label cleaning for improved dataset quality under resource constraintscited by 44
  17. 2019Semantic Redundancies in Image-Classification Datasets: The 10% You Don't Needcited by 43
  18. 2018Attacks Meet Interpretability: Attribute-steered Detection of Adversarial Samplescited by 38
  19. 2019Identifying Mislabeled Instances in Classification Datasetscited by 26
  20. 2022On the Limitations of Stochastic Pre-processing Defensescited by 21
  21. 2019Blackbox Attacks on Reinforcement Learning Agents Using Approximated Temporal Informationcited by 13
  22. 2019Is AmI (Attacks Meet Interpretability) Robust to Adversarial Examples?cited by 7
  23. 2020Continuous Safety Verification of Neural Networkscited by 5
  24. 2019Sitatapatra: Blocking the Transfer of Adversarial Samplescited by 4
  25. 2018The Taboo Trap: Behavioural Detection of Adversarial Samplescited by 2
  26. 2021Understanding the Error in Evaluating Adversarial Robustnesscited by 2
  27. 2019The Efficacy of SHIELD under Different Threat Modelscited by 1

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 3; 2024: 1; 2025: 2; 2026: 1.

7 papers.

  1. 2023Repeated Random Sampling for Minimizing the Time-to-Accuracy of Learningcited by 26
  2. 2023Less is More: Data Pruning for Faster Adversarial Trainingcited by 13
  3. 2024Large-Scale Dataset Pruning in Adversarial Training through Data Importance Extrapolationcited by 2
  4. 2025AdvSumm: Adversarial Training for Bias Mitigation in Text Summarizationcited by 2
  5. 2025Effective Data Pruning through Score Extrapolationcited by 2
  6. 2023Intrinsic Biologically Plausible Adversarial Robustnesscited by 1
  7. 2026Counterfactual Training: Teaching Models Plausible and Actionable Explanationscited by 0