发表机构
City University of Hong Kong; University of British Columbia; Hebrew University of Jerusalem（香港城市大学; 不列颠哥伦比亚大学; 耶路撒冷希伯来大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
针对个性化联邦学习中低维客户端变化与高维更新释放的结构错位，提出一次性释放私有上下文并限制适配于系数空间，结合变长量化实现高斯机制，在MNIST和CIFAR-10上匹配或超越全模型私有适配，并大幅降低通信成本。
AI 中文摘要
记录级差分隐私在个性化联邦学习中暴露了一种结构性错位：当客户端特定变化是低维的，而训练反复释放高维更新时，这种错位尤为明显。本文通过一次性释放私有客户端上下文，并将重复适配限制在固定的系数空间内，来解决这一错位问题。除了降维之外，因子化生成器还引入了一种自适应优化几何，重塑了噪声更新，受控消融实验表明，其私有训练收益的大部分由径向演化保留。为进一步降低通信成本，我们通过有限期望码长的变长量化直接实现系数更新的高斯机制，使量化误差本身充当所需的隐私扰动，而非额外的失真。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超越了全模型私有适配，同时在CIFAR-10上，在ε=16时，将受保护的上行链路减少了2.67倍，且未来客户端精度相当。
英文摘要
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 error itself serves as the required privacy perturbation rather than extra distortion. Across MNIST and CIFAR-10, our design matches or outperforms full-model private adaptation across privacy budgets and client heterogeneity, while reducing protected uplink by a factor of 2.67 at \(\varepsilon=16\) on CIFAR-10 with comparable future-client accuracy.
Comments8 pages, 1 figure
