mlsanityData to 7 Oct 2026

Improving Adversarial Robustness with Self-Paced Hard-Class Pair Reweighting

arXiv 2210.15068 ↗First posted 26 Oct 2022cs.CVcited by 6

Not on the map. Why?

Abstract (arXiv)

Deep Neural Networks are vulnerable to adversarial attacks. Among many defense strategies, adversarial training with untargeted attacks is one of the most effective methods. Theoretically, adversarial perturbation in untargeted attacks can be added along arbitrary directions and the predicted labels of untargeted attacks should be unpredictable. However, we find that the naturally imbalanced inter-class semantic similarity makes those hard-class pairs become virtual targets of each other. This study investigates the impact of such closely-coupled classes on adversarial attacks and develops a self-paced reweighting strategy in adversarial training accordingly. Specifically, we propose to upweight hard-class pair losses in model optimization, which prompts learning discriminative features from hard classes. We further incorporate a term to quantify hard-class pair consistency in adversarial training, which greatly boosts model robustness. Extensive experiments show that the proposed adversarial training method achieves superior robustness performance over state-of-the-art defenses against a wide range of adversarial attacks.

Key facts. Cited by 6 papers in the map (data through 7 Oct 2026). Builds on 44 of its 50 references that are in the map. First posted on arXiv 26 Oct 2022.

Builds on

44 of 50 references.

References · 44

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  2. 2020A Simple Framework for Contrastive Learning of Visual Representationscited by 11,979
  3. 2014Explaining and Harnessing Adversarial Examplescited by 7,287
  4. 2017Towards Deep Learning Models Resistant to Adversarial Attackscited by 6,326
  5. 2013Intriguing properties of neural networkscited by 5,114
  6. 2016Wide Residual Networkscited by 3,057
  7. 2016Towards Evaluating the Robustness of Neural Networkscited by 2,976
  8. 2019Benchmarking Neural Network Robustness to Common Corruptions and Perturbationscited by 2,857
  9. 2015FaceNet: A Unified Embedding for Face Recognition and Clusteringcited by 2,833
  10. 2020Supervised Contrastive Learningcited by 2,803
  11. 2019Theoretically Principled Trade-off between Robustness and Accuracycited by 1,510
  12. 2020Reliable evaluation of adversarial robustness with an ensemble of diverse parameter-free attackscited by 1,442
32 more
  1. 2015DeepFool: a simple and accurate method to fool deep neural networkscited by 1,315
  2. 2018CosFace: Large Margin Cosine Loss for Deep Face Recognitioncited by 968
  3. 2016Adversarial Machine Learning at Scalecited by 934
  4. 2017In Defense of the Triplet Loss for Person Re-Identificationcited by 784
  5. 2014Deep Neural Networks are Easily Fooled: High Confidence Predictions for Unrecognizable Imagescited by 774
  6. 2017SphereFace: Deep Hypersphere Embedding for Face Recognitioncited by 733
  7. 2020Fast is better than free: Revisiting adversarial trainingcited by 723
  8. 2019Adversarial Training for Free!cited by 638
  9. 2016Training Region-based Object Detectors with Online Hard Example Miningcited by 577
  10. 2017Adversarial Examples Are Not Easily Detected: Bypassing Ten Detection Methodscited by 542
  11. 2014Deep metric learning using Triplet networkcited by 421
  12. 2019Unlabeled Data Improves Adversarial Robustnesscited by 415
  13. 2017Mitigating Adversarial Effects Through Randomizationcited by 382
  14. 2019Using Pre-Training Can Improve Model Robustness and Uncertaintycited by 373
  15. 2019Minimally distorted Adversarial Examples with a Fast Adaptive Boundary Attackcited by 333
  16. 2017Detecting Adversarial Samples from Artifactscited by 244
  17. 2017On Detecting Adversarial Perturbationscited by 232
  18. 2018Adversarial Logit Pairingcited by 224
  19. 2020Geometry-aware Instance-reweighted Adversarial Trainingcited by 198
  20. 2017On the (Statistical) Detection of Adversarial Examplescited by 180
  21. 2019Are Labels Required for Improving Adversarial Robustness?cited by 162
  22. 2020Robust Pre-Training by Adversarial Contrastive Learningcited by 144
  23. 2020To be Robust or to be Fair: Towards Fairness in Adversarial Trainingcited by 137
  24. 2020Boosting Adversarial Training with Hypersphere Embeddingcited by 97
  25. 2016Adversarial Examples Detection in Deep Networks with Convolutional Filter Statisticscited by 77
  26. 2019Metric Learning for Adversarial Robustnesscited by 77
  27. 2021When Does Contrastive Learning Preserve Adversarial Robustness from Pretraining to Finetuning?cited by 76
  28. 2019Adversarially Robust Generalization Just Requires More Unlabeled Datacited by 75
  29. 2021Probabilistic Margins for Instance Reweighting in Adversarial Trainingcited by 57
  30. 2022Self-Ensemble Adversarial Training for Improved Robustnesscited by 38
  31. 2021Imbalanced Adversarial Training with Reweightingcited by 11
  32. 2022Adversarial Fine-tune with Dynamically Regulated Adversarycited by 2

Built on by

Cited by year

Citing papers per year: 2023: 2; 2024: 2; 2025: 2.

6 papers.

  1. 2025Exploring the Impact of Temperature Scaling in Softmax for Classification and Adversarial Robustnesscited by 9
  2. 2023Improving Contrastive Learning of Sentence Embeddings with Focal-InfoNCEcited by 7
  3. 2024Adversarial Training: A Surveycited by 6
  4. 2024TIMA: Text-Image Mutual Awareness for Balancing Zero-Shot Adversarial Robustness and Generalization Abilitycited by 2
  5. 2025Alleviating Performance Disparity in Adversarial Spatiotemporal Graph Learning Under Zero-Inflated Distributioncited by 2
  6. 2023Advancing Adversarial Robustness Through Adversarial Logit Updatecited by 0