arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2602.03277cs.LG
收藏

BlockRR:一种用于标签差分隐私的RR型算法统一框架

BlockRR: A Unified Framework of RR-type Algorithms for Label Differential Privacy

School of Mathematics and Statistics & Institute of Interdisciplinary Research for Mathematics and Applied Science & Hubei Key Laboratory of Engineering Modeling and Scientific Computing(数学与统计学学院及数学与应用科学交叉研究机构及湖北工程建模与科学计算重点实验室)

Haixia Liu, Yi Ding

AI总结:

BlockRR是一种统一的RR型算法框架,用于提升标签差分隐私下的模型性能,在高隐私和中等隐私范围内表现更优。

摘要:

在本文中,我们介绍了BlockRR,一种新颖的随机响应机制,用于标签差分隐私。该框架将已有的RR型机制作为特殊案例进行泛化,通过特定的参数设置,消除了对单独案例分析的需要。理论上,我们证明了BlockRR满足ε-标签DP。我们还设计了一种基于从标签先验信息导出的权重矩阵的BlockRR分区方法;平行组合原理确保了两种此类机制的组合仍保持ε-标签DP。经验上,我们在两种具有不同类别不平衡程度的CIFAR-10变体上评估了BlockRR。结果表明,在高隐私和中等隐私范围(ε≤3.0)内,我们的方法在测试准确率和每类准确率的平均值之间取得了更好的平衡。在低隐私范围(ε≥4.0)内,所有方法将BlockRR还原为标准RR,而无额外性能损失。

英文摘要:

In this paper, we introduce BlockRR, a novel and unified randomized-response mechanism for label differential privacy. This framework generalizes existed RR-type mechanisms as special cases under specific parameter settings, which eliminates the need for separate, case-by-case analysis. Theoretically, we prove that BlockRR satisfies $ε$-label DP. We also design a partition method for BlockRR based on a weight matrix derived from label prior information; the parallel composition principle ensures that the composition of two such mechanisms remains $ε$-label DP. Empirically, we evaluate BlockRR on two variants of CIFAR-10 with varying degrees of class imbalance. Results show that in the high-privacy and moderate-privacy regimes ($ε\leq 3.0$), our propsed method gets a better balance between test accuaracy and the average of per-class accuracy. In the low-privacy regime ($ε\geq 4.0$), all methods reduce BlockRR to standard RR without additional performance loss.

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。

↑