Privacy-Preserving Deep Learning on Big Data in Cloud  

作　　者：Yongkai Fan Wanyu Zhang Jianrong Bai Xia Lei Kuanching Li 

机构地区：[1]State Key Laboratory of Media Convergence and Communication,Communication University of China,Beijing 100024,China [2]Industrial and Commercial Bank of China Zhuhai Branch,Zhuhai 519000,China [3]Dept.of Computer Science and Information Engineering,Providence University,Taichung 43301,China

出　　处：《China Communications》2023年第11期176-186,共11页中国通信（英文版）

基　　金：This work was partially supported by the Natural Science Foundation of Beijing Municipality(No.4222038);by Open Research Project of the State Key Laboratory of Media Convergence and Communication(Communication University of China),the National Key R&D Program of China(No.2021YFF0307600);Fundamental Research Funds for the Central Universities.

摘　　要：In the analysis of big data,deep learn-ing is a crucial technique.Big data analysis tasks are typically carried out on the cloud since it offers strong computer capabilities and storage areas.Nev-ertheless,there is a contradiction between the open nature of the cloud and the demand that data own-ers maintain their privacy.To use cloud resources for privacy-preserving data training,a viable method must be found.A privacy-preserving deep learning model(PPDLM)is suggested in this research to ad-dress this preserving issue.To preserve data privacy,we first encrypted the data using homomorphic en-cryption(HE)approach.Moreover,the deep learn-ing algorithm’s activation function—the sigmoid func-tion—uses the least-squares method to process non-addition and non-multiplication operations that are not allowed by homomorphic.Finally,experimental re-sults show that PPDLM has a significant effect on the protection of data privacy information.Compared with Non-Privacy Preserving Deep Learning Model(NPPDLM),PPDLM has higher computational effi-ciency.

关 键 词：big data cloud computing deep learning homomorphic encryption PRIVACY-PRESERVING 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] TP309[自动化与计算机技术—控制科学与工程]