| 摘要: |
| 机器学习被广泛应用于自动推理、自然语言处理、模式识别、计算机视觉、智能机器人等人工智能领域，成为许多领域研究与技术应用中必不可少的一个工具。然而，机器学习本身存在隐私安全问题，已经引起了越来越多的关注。本文专门针对机器学习中的隐私问题进行了分类和较为详细的介绍，提出了基于攻击对象的隐私威胁分类方式，并清晰地展示了防御技术的研究思路，最后给出了亟待解决的问题和发展方向。 |
| 关键词: 机器学习 隐私威胁 隐私保护 |
| DOI：10.19363/J.cnki.cn10-1380/tn.2019.09.01 |
| 投稿时间：2019-06-15修订日期：2019-08-13 |
| 基金项目:本课题得到国家自然科学基金（No.U1836105）资助。 |
|
| A Survey of Privacy Preserving in Machine Learning |
| ZHAO Zhendong,CHANG Xiaolin,WANG Yixiang |
| Beijing Key Laboratory of Security and Privacy in Intelligent Transportation, Beijing 100044, China;School of Computer and Information Technology, Beijing Jiaotong University, Beijing 100044, China |
| Abstract: |
| Machine learning is widely used in artificial intelligence, such as automatic reasoning, natural language processing, pattern recognition, computer vision and intelligent robots. It has become an indispensable tool in many fields of research and applications. However, there exist privacy issues in machine learning have attracted more and more attention. This paper specifically classifies and introduces the privacy issues in machine learning, proposes a new classification method of privacy threat according to the attack object, and clearly shows the research ideas of the defense technology. Finally, the problem that needs to be solved and the direction of development is given. |
| Key words: machine learning privacy threat privacy protection |
