[论文] Privacy-Aligned Personalized Federated Learning with Compact Adaptatio...

论文概要

研究领域: ML 作者: Yilin Xu, Chun Hei Michael Shiu, Chih Wei Ling, Linqi Song 发布时间: 2026-09-14 arXiv: 2609.15950

中文摘要

记录级差分隐私在个性化联邦学习中暴露了一个结构性错位：当客户端特定变化是低维的，而训练反复释放高维更新时。本文通过仅释放一次私有客户端上下文并将重复适应限制在固定系数空间来解决这一错位。除了降维之外，分解生成器还诱导了一种自适应优化几何，重塑了含噪更新，受控消融表明其私有训练增益的大部分由径向演化保留。为了进一步降低通信成本，我们通过变长量化直接实现系数更新的高斯机制，期望码长有限，使量化误差本身充当所需的隐私扰动而非额外失真。在 MNIST 和 CIFAR-10 上，我们的设计在各种隐私预算和客户端异质性下匹配或优于全模型私有适应，同时在 CIFAR-10 上 ε=16 时将受保护上行链路减少了2.67倍，且未来客户端准确率相当。

原文摘要

Record-level differential privacy exposes a structural misalignment in personalized federated learning when client-specific variation is low-dimensional while training repeatedly releases high-dimensional updates. In this paper, we address this misalignment by releasing a private client context once and confining repeated adaptation to a fixed coefficient space. Beyond dimensionality reduction, the factorized generator induces an adaptive optimization geometry that reshapes noisy updates, and controlled ablations show that most of its private-training gain is retained by radial evolution. To further reduce the communication cost, we realize the Gaussian mechanism for coefficient updates directly through variable-length quantization with finite expected code length, so that the quantization...

#论文 #arXiv #ML #小凯