摘要
针对云计算环境中机器学习方案存在的安全和隐私泄露问题,提出了一种新的基于云计算的隐私保护支持向量机分类方案.用户端使用二维滞后复Logistic映射和Paillier加密方案将明文数据集加密后上传至模型服务器后,模型服务器通过协议与云服务器协同计算分类结果并返回至用户端.实验结果表明该方案能够在半诚实安全模型下安全计算的同时保护各参与者隐私,且精度较高.
Aiming at the security and privacy leakage problems of current machine learning schemes in cloud computing environment,a classification scheme of privacy preservation support vector machine based on cloud computing is proposed.The client terminal uploads the plaintext dataset to the model server with two-dimension lag complex Logistic mapping and Paillier encryption scheme.Then,the model server collaborates with the cloud server through the protocol to compute classification results and return them to the client terminal.The experiments show that this scheme can protect the privacy of each participant in the semi-honest security model while the security computation is performed,the accuracy is comparatively high.
作者
马凤英
吴黎明
张芳芳
MA Fengying;WU Liming;ZHANG Fangfang(School of Information and Automation Engineering,Qilu University of Technology(Shandong Academy of Sciences),Jinan Shandong 250353,China)
出处
《指挥与控制学报》
CSCD
2023年第3期346-354,共9页
Journal of Command and Control
基金
山东省科技重大创新工程(2020CXGC010901)
山东省高等学校“青创科技支持计划”项目(2021KJ025)
齐鲁工业大学(山东省科学院)产学研协同创新基金(2021CXY-13,2021CXY-14)资助。
关键词
云计算
机器学习
支持向量机
隐私保护
同态加密
混沌加密
cloud computing
machine learning
support vector machine
privacy preservation
homomorphic encryption
chaotic encryption
