Improving Fast Adversarial Training via Self-Knowledge Guidance
Not on the map. Why?
Abstract (arXiv)
Adversarial training has achieved remarkable advancements in defending against adversarial attacks. Among them, fast adversarial training (FAT) is gaining attention for its ability to achieve competitive robustness with fewer computing resources. Existing FAT methods typically employ a uniform strategy that optimizes all training data equally without considering the influence of different examples, which leads to an imbalanced optimization. However, this imbalance remains unexplored in the field of FAT. In this paper, we conduct a comprehensive study of the imbalance issue in FAT and observe an obvious class disparity regarding their performances. This disparity could be embodied from a perspective of alignment between clean and robust accuracy. Based on the analysis, we mainly attribute the observed misalignment and disparity to the imbalanced optimization in FAT, which motivates us to optimize different training data adaptively to enhance robustness. Specifically, we take disparity and misalignment into consideration. First, we introduce self-knowledge guided regularization, which assigns differentiated regularization weights to each class based on its training state, alleviating class disparity. Additionally, we propose self-knowledge guided label relaxation, which adjusts label relaxation according to the training accuracy, alleviating the misalignment and improving robustness. By combining these methods, we formulate the Self-Knowledge Guided FAT (SKG-FAT), leveraging naturally generated knowledge during training to enhance the adversarial robustness without compromising training efficiency. Extensive experiments on four standard datasets demonstrate that the SKG-FAT improves the robustness and preserves competitive clean accuracy, outperforming the state-of-the-art methods.
Key facts. Cited by 2 papers in the map (data through 7 Oct 2026). Builds on 50 of its 64 references that are in the map. First posted on arXiv 26 Sep 2024.
Builds on
50 of 64 references.
References · 50
- 2015Deep Residual Learning for Image Recognitioncited by 51,809
- 2021Swin Transformer: Hierarchical Vision Transformer using Shifted Windowscited by 12,082
- 2014Explaining and Harnessing Adversarial Examplescited by 7,287
- 2016Densely Connected Convolutional Networkscited by 6,691
- 2017Towards Deep Learning Models Resistant to Adversarial Attackscited by 6,326
- 2015Rethinking the Inception Architecture for Computer Visioncited by 5,451
- 2016Wide Residual Networkscited by 3,057
- 2016Towards Evaluating the Robustness of Neural Networkscited by 2,976
- 2019Theoretically Principled Trade-off between Robustness and Accuracycited by 1,510
- 2020Reliable evaluation of adversarial robustness with an ensemble of diverse parameter-free attackscited by 1,442
- 2020Fast is better than free: Revisiting adversarial trainingcited by 723
- 2020A Survey on Visual Transformercited by 653
38 more
- 2020Understanding and Improving Fast Adversarial Trainingcited by 213
- 2023Better Diffusion Models Further Improve Adversarial Trainingcited by 207
- 2018Adversarial Attack and Defense on Graph Data: A Surveycited by 166
- 2020Torchattacks: A PyTorch Repository for Adversarial Attackscited by 156
- 2020To be Robust or to be Fair: Towards Fairness in Adversarial Trainingcited by 137
- 2021Towards Robust Vision Transformercited by 137
- 2022Robustness and Accuracy Could Be Reconcilable by (Proper) Definitioncited by 109
- 2020Boosting Adversarial Training with Hypersphere Embeddingcited by 97
- 2022LAS-AT: Adversarial Training with Learnable Attack Strategycited by 94
- 2020Learnable Boundary Guided Adversarial Trainingcited by 93
- 2023Revisiting Adversarial Training for ImageNet: Architectures, Training and Generalization across Threat Modelscited by 79
- 2020Understanding Catastrophic Overfitting in Single-step Adversarial Trainingcited by 75
- 2021Revisiting and Advancing Fast Adversarial Training Through The Lens of Bi-Level Optimizationcited by 75
- 2020Guided Adversarial Attack for Evaluating and Enhancing Adversarial Defensescited by 61
- 2020Fairness Through Robustness: Investigating Robustness Disparity in Deep Learningcited by 55
- 2019Efficient Adversarial Training with Transferable Adversarial Examplescited by 54
- 2023CFA: Class-wise Calibrated Fair Adversarial Trainingcited by 53
- 2023Boosting Adversarial Transferability by Achieving Flat Local Maximacited by 48
- 2022Make Some Noise: Reliable and Efficient Single-Step Adversarial Trainingcited by 42
- 2023An Adaptive Model Ensemble Adversarial Attack for Boosting Adversarial Transferabilitycited by 42
- 2021Subspace Adversarial Trainingcited by 41
- 2021Boosting Fast Adversarial Training with Learnable Adversarial Initializationcited by 40
- 2022Prior-Guided Adversarial Initialization for Fast Adversarial Trainingcited by 34
- 2023Improving Fast Adversarial Training with Prior-Guided Knowledgecited by 34
- 2023Free Lunch for Domain Adversarial Training: Environment Label Smoothingcited by 33
- 2021Reliably fast adversarial training via latent adversarial perturbationcited by 29
- 2023WAT: Improve the Worst-class Robustness in Adversarial Trainingcited by 27
- 2022Fast Adversarial Training with Adaptive Step Sizecited by 24
- 2023On the Over-Memorization During Natural, Robust and Catastrophic Overfittingcited by 19
- 2021Mutual Adversarial Training: Learning together is better than going alonecited by 17
- 2023Feature Separation and Recalibration for Adversarial Robustnesscited by 16
- 2023Adversarial Examples Are Not Real Featurescited by 16
- 2023Revisiting and Exploring Efficient Fast Adversarial Training via LAW: Lipschitz Regularization and Auto Weight Averagingcited by 14
- 2023Fast Adversarial Training with Smooth Convergencecited by 14
- 2023Boosting Black-box Attack to Deep Neural Networks with Conditional Diffusion Modelscited by 12
- 2024DAFA: Distance-Aware Fair Adversarial Trainingcited by 11
- 2022PointCAT: Contrastive Adversarial Training for Robust Point Cloud Recognitioncited by 10
- 2023Semantic-Aware Adversarial Training for Reliable Deep Hashing Retrievalcited by 5
Built on by
Citing papers per year: 2025: 2.
2 papers.