arXiv cs.AI· Viktor Valadi, Lucas Beerens, Mattias {\AA}kesson, Johan \"Ostman, Fazeleh Hoseini, Salman Toor, Andreas Hellander·· 10 天前AI 评分28
联邦学习中的梯度反转攻击在实际部署中是否可行？arXiv 研究给出评估
Practical Feasibility of Gradient Inversion Attacks in Federated Learning
AI 导读
一项 arXiv 研究系统评估了图像联邦学习中梯度反转攻击的实际可行性，覆盖图像分类与目标检测任务及当代分辨率的经典视觉架构。结果显示，现代性能优化模型在视觉上始终能抵御有意义的图像重建，许多已报告的攻击成功依赖推理模式运行或架构简化等上界设定。在诚实但好奇的服务器假设下，高保真图像重建不构成生产级联邦学习系统的关键隐私风险。
正文 · 原文
Abstract:Gradient inversion attacks are often presented as a serious privacy threat in federated learning, with recent work reporting increasingly strong reconstructions under favorable experimental settings. However, it remains unclear whether such attacks are feasible in modern, performance-optimized systems deployed in practice. In this work, we evaluate the practical feasibility of gradient inversion for image-based federated learning. We conduct a systematic study across multiple datasets and tasks, including image classification and object detection, using canonical vision architectures at contemporary resolutions. Our results show that while gradient inversion remains possible for certain legacy or transitional designs under highly restrictive assumptions, modern, performance-optimized models consistently resist meaningful reconstruction visually. We further demonstrate that many reported successes rely on upper-bound settings, such as inference mode operation or architectural simplifications which do not reflect realistic training pipelines. Taken together, our findings indicate that, under an honest-but-curious server assumption, high-fidelity image reconstruction via gradient inversion does not constitute a critical privacy risk in production-optimized federated learning systems, and that practical risk assessments must carefully distinguish diagnostic attack settings from real-world deployments.
| Comments: | v3: revised manuscript; expanded experiments; added new feasibility probe; |
| Subjects: | Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) |
| Cite as: | arXiv:2508.19819 [cs.CR] |
| (or arXiv:2508.19819v3 [cs.CR] for this version) | |
| https://doi.org/10.48550/arXiv.2508.19819 arXiv-issued DOI via DataCite |
Submission history
From: Viktor Valadi [view email]
[v1]
Wed, 27 Aug 2025 12:07:23 UTC (1,043 KB)
[v2]
Mon, 9 Feb 2026 14:36:04 UTC (2,035 KB)
[v3]
Tue, 6 Oct 2026 13:02:26 UTC (2,234 KB)
来源：arXiv cs.AI · arxiv.org