基于差分隐私的直方图发布方法

Histogram Publishing Method Based on Differential Privacy

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.