arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2104.14808cs.CRcs.LG
收藏

改进的矩阵高斯差分隐私机制

Improved Matrix Gaussian Mechanism for Differential Privacy

Shanghai Jiao Tong University(上海交通大学)

Jungang Yang, Liyao Xiang, Weiting Li, Wei Liu, Xinbing Wang

摘要:

近年来,机器学习的广泛应用对大规模、高维数据产生了巨大的需求,而数据的隐私问题也引起了人们的严重关注。差分隐私(DP)机制传统上是针对标量值而不是针对像矩阵这样的结构数据开发的。基于$ (\varepsilon,δ) $ -差分隐私的充要条件,提出了一种改进的矩阵高斯机制(IMGM)。IMGM只对噪声协方差矩阵的奇异值施加约束,这为设计留下了空间。在矩阵值DP的合法噪声分布中,我们发现最佳分布是i.i.d.。高斯噪声,并且DP约束成为每个元素上的噪声下限。我们进一步推导出IMGM的紧合成方法。除了理论分析外,在各种模型和数据集上的实验也验证了IMGM在相同的隐私保证下比现有的机制产生更高的效用。

英文摘要:

The wide deployment of machine learning in recent years gives rise to a great demand for large-scale and high-dimensional data, for which the privacy raises serious concern. Differential privacy (DP) mechanisms are conventionally developed for scalar values, not for structural data like matrices. Our work proposes Improved Matrix Gaussian Mechanism (IMGM) for matrix-valued DP, based on the necessary and sufficient condition of $ (\varepsilon,δ) $-differential privacy. IMGM only imposes constraints on the singular values of the covariance matrices of the noise, which leaves room for design. Among the legitimate noise distributions for matrix-valued DP, we find the optimal one turns out to be i.i.d. Gaussian noise, and the DP constraint becomes a noise lower bound on each element. We further derive a tight composition method for IMGM. Apart from the theoretical analysis, experiments on a variety of models and datasets also verify that IMGM yields much higher utility than the state-of-the-art mechanisms at the same privacy guarantee.

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。

↑