JISE


  [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13]


Journal of Information Science and Engineering, Vol. 38 No. 1, pp. 57-82


Game-based Theory Rational Delegation Learning Scheme


KANG XIANG1,2, YOU-LIANG TIAN1,2,+, SHENG GAO3,
CHANG-GEN PENG1,2 AND WEI-JIE TAN1
1State Key Laboratory of Public Big Data
College of Computer Science and Technology

2Institute of Cryptography and Data Security
Guizhou University
Guiyang, 550025 P.R. China

3School of Information
Central University of Finance and Economics
Beijing, 100081 P.R. China
E-mail: xiangkang5258@gmail.com; youliangtian@163.com
+


Many enterprises or smart mobile devices collecting user data (e.g. smart-watches and wearable healthcare devices, etc.) are limited by their own computing power, so can't mine useful information from the data. To address this problem, this paper proposes a rational delegation machine learning pattern. In the proposed pattern, we firstly construct a reliable game model, and then build a formal model of delegation machine learning by delegation computation thought. Finally, we design a rational delegation learning scheme (RDLS) for decision tree model. The feasibility and reliability of the scheme are guaranteed by the incentive and constraint mechanism of game model. Moreover, we analyze the security and performance of the proposed scheme, the results show that the scheme reduces the client’s computing costs and can not disclose any useful information. Last but not least, the experimental result demonstrates that the scheme can obtain a decision tree model with high accuracy in the case of ensuring the security of data.


Keywords: rational delegation learning, game theory, decision tree, machine learning, privacy protection

  Retrieve PDF document (JISE_202201_04.pdf)