Boosting the Transferability of Adversarial Attacks with Semantic-Invariance and Low-Gradient Replacement
Mathematics · 04.07.2026 · 14 (13) · 2392
DOI: 10.3390/math14132392 ↗Kurz gefasst
Fachartikel in der Zeitschrift Mathematics, online erschienen am 04.07.2026. Verfasst von Yunqi Zhang und 4 weiteren Autor:innen. Die Arbeit nennt 57 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Natural Science Foundation of China, Yunnan Fundamental Research Projects, Youth Project of the Yunnan Xingdian Talent Program.
Abstract (Originalsprache)
Deepneural networks are highly vulnerable to adversarial examples, which are generated by introducing subtle perturbations to input data to mislead model classification. Currently, transfer-based attacks are prevalent in adversarial example generation and can be categorized into input transformation-based and gradient-based methods. However, most input transformation-based methods tend to produce augmented replicas that are semantically inconsistent with the original inputs, while gradient-based methods often leave low-gradient regions unperturbed within the model’s critical attention areas. These limitations constrain further improvements in adversarial transferability. In this work, we propose a Semantic-Invariance and Low-Gradient Replacement Method (SLRM) to address these challenges. Our framework integrates semantically consistent augmentation and gradient replacement as follows: (1) a feature extractor captures semantic features from original inputs and a reconstructor generates augmented replicas that preserve semantic fidelity to enhance input diversity, and (2) low-gradient regions in adversarial examples are systematically replaced with corresponding regions from augmented replicas to eliminate the under-perturbed areas critical for model robustness. Comprehensive empirical evaluations on ImageNet demonstrate that SLRM significantly enhances the transferability of baseline methods and seamlessly integrates with state-of-the-art approaches to further improve their performance. Moreover, SLRM substantially improves the robustness of baseline methods against defended models, achieving superior attack success rates under advanced adversarial defenses.
Förderung
- National Natural Science Foundation of China
- Yunnan Fundamental Research Projects
- Youth Project of the Yunnan Xingdian Talent Program
Zitieren
APA
Zhang, Y., Chen, W., Zhao, L., Wu, L., & Su, T. (2026). Boosting the Transferability of Adversarial Attacks with Semantic-Invariance and Low-Gradient Replacement. Mathematics, 14(13), 2392. https://doi.org/10.3390/math14132392
Vancouver
Zhang Y, Chen W, Zhao L, Wu L, Su T. Boosting the Transferability of Adversarial Attacks with Semantic-Invariance and Low-Gradient Replacement. Mathematics. 2026;14(13):2392. doi:10.3390/math14132392
BibTeX
@article{Zhang2026,
author = {Zhang, Yunqi and Chen, Wei and Zhao, Lei and Wu, Liwen and Su, Tong},
title = {Boosting the Transferability of Adversarial Attacks with Semantic-Invariance and Low-Gradient Replacement},
journal = {Mathematics},
year = {2026},
volume = {14},
number = {13},
pages = {2392},
doi = {10.3390/math14132392}
}
EUVIMED