Acta Metallurgica Sinica(English letters) ›› 2013, Vol. 20 ›› Issue (5): 85-90.doi: 10.1016/S1005-8885(13)60094-1

• Computer Science • 上一篇    下一篇

Differentially private feature selection under MapReduce framework

陈凯1,万文强2,李云2   

  1. 1. College of Computer Science, Nanjing University of Posts and Telecommunications, Nanjing 210023, China 2. Jiangsu High Technology Research Key Laboratory for Wireless Sensor Networks, Nanjing University of Posts and Telecommunications, Nanjing 210023, China 3. Institute of Computer Technology, Nanjing University of Posts and Telecommunications, Nanjing 210023, China
  • 收稿日期:2012-12-17 修回日期:2013-08-16 出版日期:2013-10-30 发布日期:2013-10-29
  • 通讯作者: 李云 E-mail:liyun@njupt.edu.cn
  • 基金资助:
    This work was supported by the National Natural Science Foundation of China (61073114, 61105082), the Scientific Research Foundation of Nanjing University of Posts and Telecommunications (NY212020, NY210010) and the Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD).

Differentially private feature selection under MapReduce framework

  1. 1. College of Computer Science, Nanjing University of Posts and Telecommunications, Nanjing 210023, China 2. Jiangsu High Technology Research Key Laboratory for Wireless Sensor Networks, Nanjing University of Posts and Telecommunications, Nanjing 210023, China 3. Institute of Computer Technology, Nanjing University of Posts and Telecommunications, Nanjing 210023, China
  • Received:2012-12-17 Revised:2013-08-16 Online:2013-10-30 Published:2013-10-29
  • Contact: Yun LI E-mail:liyun@njupt.edu.cn
  • Supported by:
    This work was supported by the National Natural Science Foundation of China (61073114, 61105082), the Scientific Research Foundation of Nanjing University of Posts and Telecommunications (NY212020, NY210010) and the Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD).

摘要: Privacy preserving data mining algorithms are crucial for the personal data analysis, such as medical and financial records. This paper focuses on feature selection and proposes a new privacy preserving distributed algorithm, which can effectively select features based on differential privacy and Gini index under the MapReduce framework. At the same time, the theoretic analysis for privacy guarantee is also presented. Some experiments are conducted on bench-mark datasets, the simulation results indicate that during the selection of important features, the proposed algorithm can preserve privacy information to a certain extent with less time cost than on centralized counterpart.

关键词: data mining, feature selection, differential privacy, MapReduce

Abstract: Privacy preserving data mining algorithms are crucial for the personal data analysis, such as medical and financial records. This paper focuses on feature selection and proposes a new privacy preserving distributed algorithm, which can effectively select features based on differential privacy and Gini index under the MapReduce framework. At the same time, the theoretic analysis for privacy guarantee is also presented. Some experiments are conducted on bench-mark datasets, the simulation results indicate that during the selection of important features, the proposed algorithm can preserve privacy information to a certain extent with less time cost than on centralized counterpart.

Key words: data mining, feature selection, differential privacy, MapReduce