English

应用科学学报 >

2023 , Vol. 41 >Issue 3: 488 - 499

DOI: https://doi.org/10.3969/j.issn.0255-8297.2023.03.010

计算机科学与应用

一种可信执行环境下的联邦逻辑回归评分卡系统

展开
  • 1. 扬州大学 广陵学院, 江苏 扬州 225000;
    2. 国汽智图(北京) 科技有限公司, 北京 100176;
    3. 中国电子系统技术有限公司, 北京 100141

收稿日期: 2022-06-25

  网络出版日期: 2023-06-16

基金资助

2022年教育部产学合作协同育人项目(No. 220604309011432)资助

收起

Federated Logistic Regression Scorecard System under Trusted Execution Environment

Expand
  • 1. Guangling College, Yangzhou University, Yangzhou 225000, Jiangsu, China;
    2. China Intelligent and Connected Vehicles Data(Beijing) Co., Ltd., Beijing 100176, China;
    3. China Electronic System Technology Co., Ltd., Beijing 100141, China

Received date: 2022-06-25

  Online published: 2023-06-16

Fold

摘要

为了构造评分卡模型并保证数据的隐私性，提出一种可信执行环境下的联邦逻辑回归系统。该系统利用可信执行环境的强安全性来抵御参数交互过程中的推演攻击，通过联合归一化和改进的联邦平均方法分别解决局部数据尺度的不一致性和类别不均衡分布下的评分卡模型有效性问题。在一个公开信用卡违约数据集上的测试结果表明： 所提出的改进联邦平均方法与典型联邦平均方法相比，能更有效地应对类别不均衡分布问题；与同态加密联邦学习系统相比，能大大提高执行效率。

关键词： 逻辑回归; 联邦学习; 可信执行环境; 类别不均衡

本文引用格式

史汶泽, 陆林, 秦文杰, 于涛 . 一种可信执行环境下的联邦逻辑回归评分卡系统[J]. 应用科学学报, 2023 , 41(3) : 488 -499 . DOI: 10.3969/j.issn.0255-8297.2023.03.010

Abstract

A federated logistic regression system under trusted execution environment is proposed to build a scorecard model while ensuring data privacy. This system uses the strong security of trusted execution environment to resist inference attacks in the process of parameter exchanging. Then, a joint normalization method and an improved federated average method are utilized to solve the problem of inconsistency of local data scale and improve the effectiveness of scorecard model under class imbalance condition, respectively. Test results on a public credit-overdue data set show that the improved federated average is more effective than typical federated average method in handling the problem of imbalanced class distribution. Compared with homomorphic encryption-based federated learning systems, the proposed federated logistic regression system performs a greatly improved execution efficiency.

Key words： logistic regression; federated learning; trusted execution environment; class imbalance

参考文献

[1] 刘荣珍. 基于逻辑回归和机器学习的个人信用风险研究[D]. 兰州: 兰州大学, 2021.
[2] 刘荣弟. 基于Logistic回归的信用评分模型研究[D]. 大连: 大连理工大学, 2018.
[3] Gang D, Lai K K, Yen J. Credit scorecard based on logistic regression with random coefficients [J]. Procedia Computer Science, 2010, 1(1): 2463-2468.
[4] Bian Y, Yang C, Zhao J L, et al. Good drivers pay less: a study of usage-based vehicle insurance models [J]. Transportation Research Part A: Policy and Practice, 2018, 107: 20-34.
[5] Huang Y, Meng S. Automobile insurance classification ratemaking based on telematics driving data [J]. Decision Support Systems, 2019, 127(12): 1-11.
[6] 边俊源. 基于驾驶行为数据分析的UBI车险定价策略研究[D]. 南京: 南京邮电大学, 2020.
[7] Mcmahan B, Moore E, Ramage D, et al. Communication-efficient learning of deep networks from decentralized data [C]//Artificial Intelligence and Statistics. PMLR, 2017: 1273-1282.
[8] 杨强. 联邦学习: 人工智能的最后一公里[J]. 智能系统学报, 2020, 15(1): 183-186. Yang Q. Federated learning: the last kilometer of artificial intelligence [J]. CAAI Transactions on Intelligent Systems, 2020, 15(1): 183-186. (in Chinese)
[9] 谭作文, 张连福. 机器学习的隐私保护研究综述[J]. 软件学报, 2020, 31(7): 2127-2156. Tan Z W, Zhang L F. Survey on privacy preserving techniques for machine learning [J]. Journal of Software, 2020, 31(7): 2127-2156. (in Chinese)
[10] 刘艺璇, 陈红, 刘宇涵, 等. 联邦学习中的隐私保护技术[J]. 软件学报, 2022, 33(3): 1057-1092. Liu Y X, Chen H, Liu Y H, et al. Privacy-preserving techniques in federated learning [J]. Journal of Software, 2022, 33(3): 1057-1092. (in Chinese)
[11] Le T P, Aono Y, Hayashi T, et al. Privacy-preserving deep learning via additively homomorphic encryption [J]. IEEE Transactions on Information Forensics and Security, 2018, 13(5): 1333-1345.
[12] Liu Y, Kang Y, Xing C, et al. A secure federated transfer learning framework [J]. IEEE Intelligent Systems, 2020, 35(4): 70-82.
[13] Bonawitz K, Ivanov V, Kreuter B, et al. Practical secure aggregation for privacy-preserving machine learning [C]//Proceedings of 2017 ACM SIGSAC Conference on Computer and Communications Security, 2017: 1175-1191.
[14] Wei K, Li J, Ding M, et al. Federated learning with differential privacy: algorithms and performance analysis [J]. IEEE Transactions on Information Forensics and Security, 2020, 15: 3454-3469.
[15] Naehrig M, Lauter K E, Vaikuntanathan V. Can homomorphic encryption be practical? [C]//Proceedings of the 3rd ACM Cloud Computing Security Workshop. ACM, 2011: 113-124.
[16] Melis L, Song C, De Cristofaro E D, et al. Exploiting unintended feature leakage in collaborative learning [C]//2019 IEEE Symposium on Security and Privacy (SP). IEEE, 2019: 691-706.
[17] Bagdasaryan E, Poursaeed O, Shmatikov V. Differential privacy has disparate impact on model accuracy [J]. Advances in Neural Information Processing Systems, 2019, 32: 1-10.
[18] Mckeen F, Alexandrovich I, Anati I, et al. Intel software guard extensions (Intel SGX) support for dynamic memory management inside an enclave [C]//Hardware & Architectural Support for Security & Privacy. ACM, 2016: 1-9.
[19] 郑显义, 李文, 孟丹. TrustZone技术的分析与研究[J]. 计算机学报, 2016, 39(9): 1912-1928. Zheng X Y, Li W, Meng D. Analysis and research on TrustZone technology [J]. Chinese Journal of Computers, 2016, 39(9): 1912-1928. (in Chinese)
[20] 董春涛, 沈晴霓, 罗武, 等. SGX应用支持技术研究进展[J]. 软件学报, 2021, 32(1): 137-166. Dong C T, Shen Q N, Luo W, et al. Research progress of SGX application supporting techniques [J]. Journal of Software, 2021, 32(1): 137-166. (in Chinese)
[21] 舒俊宜. 一种基于可信执行环境的机密计算框架设计与实现[D]. 北京: 北京大学, 2021.
[22] 张英骏, 冯登国, 秦宇, 等. 基于TrustZone的强安全需求环境下可信代码执行方案[J]. 计算机研究与发展, 2015, 52(10): 2224-2238. Zhang Y J, Feng D G, Qin Y, et al. A TrustZone-based trusted code execution with strong security requirements [J]. Journal of Computer Research and Development, 2015, 52(10): 2224- 2238. (in Chinese)
[23] 张英骏, 冯登国, 秦宇, 等. 基于TrustZone的开放环境中敏感应用防护方案[J]. 计算机研究与发展, 2017, 54(10): 2268-2283. Zhang Y J, Feng D G, Qin Y, et al. A TrustZone based application protection scheme in highly open scenarios [J]. Journal of Computer Research and Development, 2017, 54(10): 2268-2283. (in Chinese)
[24] Yu D, Ran D, Long L, et al. POSTER: rust SGX SDK: towards memory safety in intel SGX enclave [C]//2017 ACM SIGSAC Conference. ACM, 2017: 2491-2493.
[25] Chen Y, Luo F, Li T, et al. A training-integrity privacy-preserving federated learning scheme with trusted execution environment [J]. Information Sciences, 2020, 522: 69-79.
[26] Zhang X, Li F, Zhang Z, et al. Enabling execution assurance of federated learning at untrusted participants [C]//IEEE INFOCOM 2020- IEEE Conference on Computer Communications. IEEE, 2020: 1877-1886.
[27] Mo F, Haddadi H, Katevas K, et al. PPFL: privacy-preserving federated learning with trusted execution environments [C]//Proceedings of the 19th Annual International Conference on Mobile Systems, Applications, and Services, 2021: 94-108.
[28] 姜建林. 基于可信执行环境的联邦学习模型安全聚合技术研究[D]. 武汉: 武汉大学, 2021.
[29] 宋蕾, 马春光, 段广晗, 等. 基于数据纵向分布的隐私保护逻辑回归[J]. 计算机研究与发展, 2019, 56(10): 2243-2249. Song L, Ma C G, Duan G H, et al. Privacy-preserving logistic regression on vertically partitioned data [J]. Journal of Computer Research and Development, 2019, 56(10): 2243-2249. (in Chinese)
[30] Krawczyk B. Learning from imbalanced data: open challenges and future directions [J]. Progress in Artificial Intelligence, 2016, 5(4): 221-232.
Options
文章导航

/

〈 〉