基于差分隐私的直方图发布方法  

Histogram Publishing Method Based on Differential Privacy

作　　者：汤文兵[1] 张伟媛 Tang Wenbing;Zhang Weiyuan(School of Computer Science and Engineering,Anhui University of Science and Technology,Huainan 232001)

机构地区：[1]安徽理工大学计算机科学与工程学院,淮南232001

出　　处：《现代计算机》2022年第14期75-78,共4页Modern Computer

摘　　要：差分隐私是目前应用较为广泛的工具,由于它忽略攻击者的任意背景知识,提供了强大的理论隐私保障,能够在数据发布期间保护数据的隐私。直方图发布作为直观的数据发布方法之一,也有着隐私披漏的风险。针对该问题,本文提出了一种基于差分隐私保护的直方图发布方法,实验结果证明,将这种方法应用到统计上可大大提高隐私的安全性和数据发布的可靠性。Differential privacy is currently a widely used tool.Because it ignores any background knowledge of the attacker,it provides a strong theoretical privacy guarantee and can protect data privacy during data release.Histogram publishing,as one of the intuitive data publishing methods,also has the risk of privacy leaks.In response to this problem,this paper proposes a histogram publishing method based on differential privacy protection,and the experimental results prove that applying this method to statis⁃tics can greatly improve the security of privacy and the reliability of data publishing.

关 键 词：差分隐私 数据发布 直方图 隐私保护 

分 类 号：TP309[自动化与计算机技术—计算机系统结构]