arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2103.01953cs.ITcs.CRcs.LGmath.IT
收藏

基于用户采样和无线聚合的联邦学习隐私放大

Privacy Amplification for Federated Learning via User Sampling and Wireless Aggregation

The University of Arizona(亚利桑那大学)

Mohamed Seif, Wei-Ting Chang, Ravi Tandon

摘要:

在本文中,我们研究了联邦学习的问题,在一个无线信道与用户采样,高斯多址信道建模,受到中央和本地差分隐私(DP/LDP)的约束。已经表明,无线信道的叠加性质提供了带宽高效梯度聚合的双重益处,以及对用户的强DP保证。具体地,中央DP隐私泄漏已经被示出为按比例缩放为$\mathcal{O}(1/K^{1/2})$,其中$K$是用户的数量。它也被证明,用户采样耦合正交传输可以增强中央DP隐私泄漏具有相同的缩放行为。在这项工作中,我们表明,通过加入无线聚合和用户采样,可以获得更强的隐私保证。我们提出了一个私人无线梯度聚合方案,它依赖于独立随机参与决策的每个用户。我们提出的方案的中心DP泄漏的尺度为$\mathcal{O}(1/K^{3/4})$。此外,我们发现,LDP也提高了用户抽样。我们还提出了分析所提出的方案的收敛速度和研究的无线资源,收敛和隐私之间的权衡理论和经验的两种情况下,采样的参与者的数量是$(a)$已知的,或$(b)$未知的参数服务器。

英文摘要:

In this paper, we study the problem of federated learning over a wireless channel with user sampling, modeled by a Gaussian multiple access channel, subject to central and local differential privacy (DP/LDP) constraints. It has been shown that the superposition nature of the wireless channel provides a dual benefit of bandwidth efficient gradient aggregation, in conjunction with strong DP guarantees for the users. Specifically, the central DP privacy leakage has been shown to scale as $\mathcal{O}(1/K^{1/2})$, where $K$ is the number of users. It has also been shown that user sampling coupled with orthogonal transmission can enhance the central DP privacy leakage with the same scaling behavior. In this work, we show that, by join incorporating both wireless aggregation and user sampling, one can obtain even stronger privacy guarantees. We propose a private wireless gradient aggregation scheme, which relies on independently randomized participation decisions by each user. The central DP leakage of our proposed scheme scales as $\mathcal{O}(1/K^{3/4})$. In addition, we show that LDP is also boosted by user sampling. We also present analysis for the convergence rate of the proposed scheme and study the tradeoffs between wireless resources, convergence, and privacy theoretically and empirically for two scenarios when the number of sampled participants are $(a)$ known, or $(b)$ unknown at the parameter server.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑