Improving Adversarial Robustness via Decoupled Visual Representation Masking
- 9 June 2025
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Information Forensics and Security
- Vol. 20 (15566013) , 5678-5689
Abstract
Deep neural networks are proven to be vulnerable to finely designed adversarial examples, and adversarial defense algorithms draw more and more attention nowadays. Pre-processing based defense is a major strategy, as well as learning robust feature representation, has been proven an effective way to boost generalization. However, existing defense works lack considering different depth-level visual features in the training process. In this paper, we first highlight two novel properties of robust features from the feature distribution perspective: 1) Diversity (robust features within the same class should maintain appropriate variety). 2) Discriminability (robust features from different classes should be sufficiently separated). We find that state-of-the-art defense methods aim to address both of these mentioned issues well. It motivates us to increase intra-class variance and decrease inter-class discrepancy simultaneously in adversarial training. Specifically, we propose a simple but effective defense based on decoupled visual representation masking. The designed Decoupled Visual Feature Masking (DFM) block can adaptively disentangle visual discriminative features and non-visual features with diverse mask strategies, while the suitable discarding information can disrupt adversarial noise to improve robustness. Our work provides a generic and easy-to-plugin block unit for any former adversarial training algorithm to achieve better protection integrally. Extensive experimental results prove that the proposed method can achieve superior performance compared with state-of-the-art defense approaches. The code is publicly available at https://github.com/chenboluo/Adversarial-defense.Keywords
All Related Versions
Funding Information
- National Natural Science Foundation of China (62306227, 62276198, U22A2035, U22A2096, 62036007)
- Scientific and Technological Innovation Teams in Shaanxi Province (2025RS-CXTD-011)
- Young Elite Scientists Sponsorship Program by CAST (2022QNRC001)
- Shaanxi Province Core Technology Research and Development Project (2024QY2-GJHX-11)
- Open Research Project of Key Laboratory of Artificial Intelligence Ministry of Education (AI202401)
- Overseas Expertise Introduction Center for Discipline Innovation of Food Nutrition and Human Health (B16037)
- CCF-Baidu Open Fund
- Fundamental Research Funds for the Central Universities (QTZX23083, QTZX23042, ZYTS24142)
This publication has 36 references indexed in Scilit:
- FusionMamba: dynamic feature enhancement for multimodal image fusion with MambaVisual Intelligence, 2024
- Towards Defending Multiple $\ell _p$-Norm Bounded Adversarial Perturbations via Gated Batch NormalizationInternational Journal of Computer Vision, 2023
- Hybrid token transformer for deep face recognitionPattern Recognition, 2023
- Data augmentation in natural language processing: a novel text generation approach for long and short text classifiersInternational Journal of Machine Learning and Cybernetics, 2022
- Square Attack: A Query-Efficient Black-Box Adversarial Attack via Random Search2020
- A Survey of Autonomous Driving: Common Practices and Emerging TechnologiesIEEE Access, 2020
- Feature Denoising for Improving Adversarial Robustness2019
- Decoupling Direction and Norm for Efficient Gradient-Based L2 Adversarial Attacks and Defenses2019
- SparseFool: A Few Pixels Make a Big Difference2019
- Barrage of Random Transforms for Adversarially Robust Defense2019