arXiv：cs.LG· Agnivo Ghosh, Saumik Bhattacharya·· 3 小时前AI 评分33
PI-SME：面向联邦学习多步梯度反转的路径积分代理模型
A Path Integral Surrogate for Multi-Step Gradient Inversion in Federated Learning
AI 导读
研究者提出路径积分代理模型扩展 PI-SME，将 FedAvg 下客户端累积的模型更新视为梯度场的路径积分，并通过可学习 Bézier 路径上的 Gauss-Legendre 求积进行近似。在 CIFAR-100 和 FEMNIST 图像上，PI-SME 在多项反转指标和匹配损失上比最强代理基线更忠实地重建了客户端私有输入。该工作投稿至 IEEE ICASSP 2027。
正文
Abstract:Federated learning lets many clients train a shared model together without ever sending their private data to a central server. Each client shares only a model update, and this update should reveal far less about the client than its raw training examples would. This premise is what protects the privacy of the clients. Gradient inversion attacks challenge it directly by trying to reconstruct a client's private input images from the single update it shared. Under FedAvg, a client's update accumulates several local training steps, so the server sees only the two endpoints of a hidden weight trajectory. Recent gradient inversion attacks fit a surrogate model along the path between these two endpoints but they still read its gradient at a single point. We propose the Path-Integral Surrogate Model Extension (PI-SME) which treats the accumulated update as a path integral of the gradient field and approximates it by Gauss--Legendre quadrature over several nodes along a learnable Bézier path. On CIFAR-100 and FEMNIST images across a range of trajectory lengths and class-restricted batches PI-SME reconstructs the private inputs more faithfully than the strongest surrogate baseline on several inversion metrics and the matching loss.
| Comments: | 5 pages, 2 figures, 3 tables. Submitted to IEEE ICASSP 2027 |
| Subjects: | Machine Learning (cs.LG); Distributed, Parallel, and Cluster Computing (cs.DC) |
| Cite as: | arXiv:2610.03597 [cs.LG] |
| (or arXiv:2610.03597v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.03597 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Agnivo Ghosh Mr [view email]
[v1]
Fri, 2 Oct 2026 17:03:12 UTC (52 KB)
来源：arXiv：cs.LG · arxiv.org