arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2104.01808cs.CRcs.DScs.LG
收藏

多方差分隐私下的频率估计:单次和流式

Frequency Estimation Under Multiparty Differential Privacy: One-shot and Streaming

Hong Kong University of Science and Technology(香港科技大学);University of Warwick(华威大学)

Ziyue Huang, Yuan Qiu, Ke Yi, Graham Cormode

摘要:

我们研究了隐私和通信约束下的频率估计的基本问题,其中数据分布在$k$各方之间。我们考虑两种应用场景:(1)单次,其中数据是静态的,聚合器进行一次性计算;(2)流式传输,其中每一方随着时间的推移接收项目流,聚合器不断监控频率。我们采用多方差分隐私(MDP)模型,它比局部差分隐私(LDP)和(集中式)差分隐私更具有一般性。我们的协议实现最优性(对数因子)允许的两个约束条件中更严格的。特别是,当专用于$\varepsilon$ -LDP模型时,我们的协议使用$O(k\max\{ \varepsilon, \frac{1}{\varepsilon} \})$通信位和$O(k \log u)$公共随机位实现了$\sqrt{k}/(e^{Θ(\varepsilon)}-1)$的错误,其中$u$是域的大小.

英文摘要:

We study the fundamental problem of frequency estimation under both privacy and communication constraints, where the data is distributed among $k$ parties. We consider two application scenarios: (1) one-shot, where the data is static and the aggregator conducts a one-time computation; and (2) streaming, where each party receives a stream of items over time and the aggregator continuously monitors the frequencies. We adopt the model of multiparty differential privacy (MDP), which is more general than local differential privacy (LDP) and (centralized) differential privacy. Our protocols achieve optimality (up to logarithmic factors) permissible by the more stringent of the two constraints. In particular, when specialized to the $\varepsilon$-LDP model, our protocol achieves an error of $\sqrt{k}/(e^{Θ(\varepsilon)}-1)$ using $O(k\max\{ \varepsilon, \frac{1}{\varepsilon} \})$ bits of communication and $O(k \log u)$ bits of public randomness, where $u$ is the size of the domain.

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。

↑