arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2601.06385cs.CR
收藏

鱼鳞隐私下的噪声削减:一种实用的噪声校准方法

Noise Reduction for Pufferfish Privacy: A Practical Noise Calibration Method

Beijing Institute of Technology(北京理工大学);University of Auckland(奥克兰大学);Qi An Xin Technology Group Inc.(奇安信科技集团股份有限公司);Zhongguancun Laboratory(中关村实验室)

Wenjin Yang, Ni Ding, Zijian Zhang, Jing Sun, Zhen Li, Yan Wu, Jiahang Sun, Haotian Lin, Yong Liu, Jincheng An, Liehuang Zhu

AI总结:

本文提出了一种实用的噪声校准方法,用于在鱼鳞隐私下提高数据效用,实验表明数据效用提升了47%至87%。

摘要:

本文介绍了一种放松的噪声校准方法,以在实现鱼鳞隐私的同时提高数据效用。这项工作基于已有的1-瓦尔德斯坦(Kantorovich)机制,通过缓解现有的过于严格的条件,从而导致过多噪声,提出了一个实用的机制设计算法作为通用解决方案。我们证明,与1-瓦尔德斯坦机制相比,我们的方法在所有隐私预算ε和先验信念下,总存在严格噪声削减,且噪声削减(也代表数据效用的提升)在低隐私预算情况下显著增加——这在实际应用中很常见。我们还分析了不同先验分布下噪声削减的变化和最优性。此外,当加性噪声最大时,我们引入的最坏情况1-瓦尔德斯坦机制仍保持噪声削减的所有性质。我们进一步表明,最坏情况1-瓦尔德斯坦机制等同于ℓ₁-灵敏度方法。在三个真实世界数据集上的实验结果表明,数据效用提高了47%至87%。

英文摘要:

This paper introduces a relaxed noise calibration method to enhance data utility while attaining pufferfish privacy. This work builds on the existing $1$-Wasserstein (Kantorovich) mechanism by alleviating the existing overly strict condition that leads to excessive noise, and proposes a practical mechanism design algorithm as a general solution. We prove that a strict noise reduction by our approach always exists compared to $1$-Wasserstein mechanism for all privacy budgets $ε$ and prior beliefs, and the noise reduction (also represents improvement on data utility) gains increase significantly for low privacy budget situations--which are commonly seen in real-world deployments. We also analyze the variation and optimality of the noise reduction with different prior distributions. Moreover, all the properties of the noise reduction still exist in the worst-case $1$-Wasserstein mechanism we introduced, when the additive noise is largest. We further show that the worst-case $1$-Wasserstein mechanism is equivalent to the $\ell_1$-sensitivity method. Experimental results on three real-world datasets demonstrate $47\%$ to $87\%$ improvement in data utility.

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。

↑