| 摘要: |
| 差分隐私是2006年由DWORK提出的一种新型的隐私保护机制，它主要针对隐私保护中，如何在分享数据时定义隐私，以及如何在保证可用性的数据发布时，提供隐私保护的问题，这两个问题提出了一个隐私保护的数学模型。由于差分隐私对于隐私的定义不依赖于攻击者的背景知识，所以被作为一种新型的隐私保护模型广泛地应用于数据挖掘，机器学习等各个领域。本文介绍了差分隐私的基础理论和目前的研究进展，以及一些已有的差分隐私保护理论和技术，最后对未来的工作和研究热点进行了展望。 |
| 关键词: 差分隐私 隐私保护 数据发布 数据挖掘 机器学习 |
| DOI：10.19363/J.cnki.cn10-1380/tn.2018.09.08 |
| 投稿时间：2017-09-11修订日期：2018-02-22 |
| 基金项目:本课题得到国家自然科学基金——广东省联合基金（No.U1401251）资助。 |
|
| A Survey on Differential Privacy |
| LI Xiaoguang,LI Hui,LI Fenghua,ZHU Hui |
| School of Cyber Engineering, Xidian University, Xi'an 710071, China |
| Abstract: |
| Differential privacy is a privacy preserving mathematical model which was proposed by Dwork at 2006, it aims to solve two mainly problems which are how to define privacy while sharing data and how to publish data to satisfy privacy while provide utility. As the definition of differential privacy doesn't depend on background knowledge of adversaries, it is regarded as a new privacy preserving mechanism to apply in many fields such as data mining and machine learning. In this article, we introduce basic theories and present research progress, besides, we talk about existing theories and technologies of differential privacy. At the end, we discuss directions and underlying research hotspots in the future. |
| Key words: differential privacy privacy preserving data publishing data mining machine learning |
