计算机科学 ›› 2025, Vol. 52 ›› Issue (11A): 241200115-9.doi: 10.11896/jsjkx.241200115
李进成, 李英娜, 付国庆
LI Jincheng, LI Yingna, FU Guoqing
摘要： 在信息时代,数据成为一种宝贵资源。数据共享在驱动人工智能领域发展的同时,也带来了隐私泄露的风险。全同态加密(Fully Homomorphic Encryption,FHE)技术为各种机器学习算法的实现提供了一条安全路径,它允许在密文数据上直接进行运算。然而,在密文数据上进行运算会产生很高的计算开销,因此需要以“FHE友好”的方式重新设计算法。对此,基于CKKS全同态加密算法,采用低次近似的阶跃函数和轻量级的交互协议取代复杂的非线性运算,提出了一种新的隐私保护决策树方案,实现了密文下决策树的训练与推理。最后,在4个UCI数据集上进行了对比实验,实验结果显示,提出的方案在平均AUC和平均F1-Score指标上分别达到0.92与0.77,优于PrivaTree方案与SecDT方案,同时展现出更强的稳定性。
中图分类号:
| [1]OLINDER N,FEDYAKIN K,KORNEEVA E.Personal dataprotection in the internet of things [C]//Proceedings of the 1st International Scientific Conference Legal Regulation of the Digital Economy and Digital Relations:Problems and Prospects of Development.New York,USA:Atlantis Press,2021:227-232. [2]ZHANG J,CHEN B,ZHAO Y,et al.Data security and privacy-preserving in edge computing paradigm:Survey and open issues [J].IEEE Access,2018,6:18209-18237. [3]SACHDEV A,BHANSALI M.Enhancing cloud computing security using AES algorithm [J].International Journal of Computer Applications,2013,67(9):19-23. [4]BRAKERSKI Z.Fully homomorphic encryption without modulus switching from classical GapSVP [C]//Annual Cryptology Conference.Heidelberg,Germany:Springer,2012:868-886. [5]CHEON J H,KIM A,KIM M,et al.Homomorphic encryption for arithmetic of approximate numbers [C]//Advances in Cryptology－ASIACRYPT 2017.Heidelberg,Germany:Springer International Publishing,2017:409-437. [6]MASAHIRO Y.Fully Homomorphic encryption without bootstrapping [M].Saarbrücken,Germany:LAP LAMBERT Academic Publishing,2015. [7]FAN J,VERCAUTEREN F.Somewhat practical fully homo-morphic encryption [EB/OL].[2012-03-22].https://ia.cr/2012/144. [8]ALLOGHANI M,ALANI M M,AL-JUMEILY D,et al.A systematic review on the status and progress of homomorphic encryption technologies [J].Journal of Information Security and Applications,2019,48:102362. [9]CHEN H,GILAD-BACHRACH R,HAN K,et al.Logistic regression over encrypted data from fully homomorphic encryption [J].BMC Medical Genomics,2018,11(4):3-12. [10]CROCKETT E.A low-depth homomorphic circuit for logisticregression model training [EB/OL].[2024-05-01].https://ia.cr/2020/1483. [11]GILAD-BACHRACH R,DOWLIN N,LAINE K,et al.Cryp-tonets:Applying neural networks to encrypted data with high throughput and accuracy [C]//International Conference on Machine Learning.New York,USA:PMLR,2016:201-210. [12]BOST R,POPA R A,TU S,et al.Machine learning classification over encrypted data [EB/OL].[2015-01-12].https://ia.cr/2014/331. [13]COCK M D,DOWSLEY R,HORST C,et al.Efficient and pri-vate scoring of decision trees,support vector machines and logistic regression models based on pre-computation[J].IEEE Transactions on Dependable and Secure Computing,2019,16(2):217-230. [14]ZHAO J,ZHU H,WANG F,et al.Efficient and privacy-preserving tree-based inference via additive homomorphic encryption [J].Information Sciences,2023,650:119480. [15]TUENO A,BOEV Y,KERSCHBAUM F.Non-interactive private decision tree evaluation [C]//34th Annual IFIP WG 11.3 Conference.Berlin,Germany:Springer International Publishing,2020:174-194. [16]XU K,TAN B H M,WANG L P,et al.Privacy-preserving outsourcing decision tree evaluation from homomorphic encryption [J].Journal of Information Security and Applications,2023,77:103582. [17]TUENO A,KERSCHBAUM F,KATZENBEISSER S.Private evaluation of decision trees using sublinear cost [J].Proceedings on Privacy Enhancing Technologies,2019(1):266-286. [18]LU W J,ZHOU J J,SAKUMA J.Non-interactive and Output Expressive Private Comparison from Homomorphic Encryption [C]//Proceedings of the 2018 on Asia Conference on Computer and Communications Security.New York,USA:ACM,2018:67-74. [19]HAN Z,GE C,WU B,et al.Privet:A privacy-preserving federated incremental decision trees [J].IEEE Transactions on Ser-vices Computing,2023,16(3):1964-1975. [20]ZHENG Y,XU S,WANG S,et al.Privet:A privacy-preserving vertical federated learning service for gradient boosted decision tables [J].IEEE Transactions on Services Computing,2023,16(5):3604-3620. [21]YAMAMOT F,OZAWA S,WANG L.eFL-Boost:Efficientfederated learning for gradient boosting decision trees [J].IEEE Access,2022,10:43954-43963. [22]ZHAO J,ZHU H,XU W,et al.SGBoost:An efficient and priva-cy-preserving vertical federated tree boosting framework [J].IEEE Transactions on Information Forensics and Security,2023,18:1022-1036. [23]BLATT M,GUSEV A,POLYAKOV Y,et al.Optimized homomorphic encryption solution for secure genome-wide association studies [J].BMC Medical Genomics,2020,13(7):83. [24]BOURA C,GAMA N,GEORGIEVA M,et al.Chimera:Combining ring-lwe-based fully homomorphic encryption schemes [J].Journal of Mathematical Cryptology,2020,14(1):316-338. [25]EL ZEIN Y,LEMAY M,HUGUENIN K.PrivaTree:Collabora-tive privacy-preserving training of decision trees on biomedical data [J].IEEE/ACM Transactions on Computational Biology and Bioinformatics,2023,21(1):1-13. [26]CHEN Y C,CHANG C C,HUNG C C,et al.SecDT:privacy-preserving outsourced decision tree classification without polynomial forms in edge-cloud computing [J].IEEE Transactions on Signal and Information Processing over Networks,2022,8:1037-1048. |
|
||
首页