《计算机应用》唯一官方网站 ›› 2023, Vol. 43 ›› Issue (4): 1160-1168.DOI: 10.11772/j.issn.1001-9081.2022030337
所属专题: 网络空间安全
收稿日期:
2022-03-21
修回日期:
2022-05-13
接受日期:
2022-05-25
发布日期:
2023-04-11
出版日期:
2023-04-10
通讯作者:
尹春勇
作者简介:
屈锐(1999—),男,江苏宿迁人,硕士研究生,主要研究方向:差分隐私、联邦学习。
Received:
2022-03-21
Revised:
2022-05-13
Accepted:
2022-05-25
Online:
2023-04-11
Published:
2023-04-10
Contact:
Chunyong YIN
About author:
QU Rui, born in 1999, M. S. candidate. His research interests include differential privacy, federated learning.
摘要:
联邦学习(FL)可以有效保护用户的个人数据不被攻击者获得,而差分隐私(DP)则可以实现FL的隐私增强,解决模型训练参数导致的隐私泄露问题。然而,现有的基于DP的FL方法只关注统一的隐私保护预算,而忽略了用户的个性化隐私需求。针对此问题,提出了一种两阶段的基于个性化差分隐私的联邦学习(PDP-FL)算法。在第一阶段,依据用户的隐私偏好对用户隐私进行分级,并添加满足用户隐私偏好的噪声,以实现个性化隐私保护,同时上传隐私偏好对应的隐私等级给中央聚合服务器;在第二阶段,为实现对全局数据的充分保护,采取本地和中心同时保护的策略,并根据用户上传的隐私等级,添加符合全局DP阈值的噪声,以量化全局的隐私保护水平。实验结果表明,在MNIST和CIFAR-10数据集上,PDP-FL算法的分类准确度分别为93.8%~94.5%和43.4%~45.2%,优于基于本地化差分隐私的联邦学习(LDP-Fed)和基于全局差分隐私的联邦学习(GDP-FL),同时满足了个性化隐私保护的需求。
中图分类号:
尹春勇, 屈锐. 基于个性化差分隐私的联邦学习算法[J]. 计算机应用, 2023, 43(4): 1160-1168.
Chunyong YIN, Rui QU. Federated learning algorithm based on personalized differential privacy[J]. Journal of Computer Applications, 2023, 43(4): 1160-1168.
方法 | 优点 | 缺点 | 保护对象 | 保护场景 |
---|---|---|---|---|
同态加密[ | 准确性高,隐私保护严格 | 计算开销高,通信开销高,不保护发布模型 | 本地梯度 | 统一预算场景 |
多方安全计算[ | 准确性较高,隐私保护严格 | 通信开销高,协议复杂脆弱 | 本地梯度 | 统一预算场景 |
中心化差分隐私[ | 准确性较高,通信开销低,计算量较低 | 需要中心服务器可信 | 中心梯度 | 统一预算场景 |
本地化差分隐私[ | 通信开销低,扰动方案灵活 | 准确性低 | 本地梯度 | 统一预算场景 |
安全混洗[ | 准确性较高,隐私保护严格 | 计算开销高 | 本地梯度 | 统一预算场景 |
PDP-FL | 可以提供个性化隐私保证,准确性较高 | 需要合理的隐私分级 | 本地和中心梯度 | 个性化预算场景 |
表1 基于联邦学习的隐私保护方法比较
Tab. 1 Comparison of privacy protection methods based on federated learning
方法 | 优点 | 缺点 | 保护对象 | 保护场景 |
---|---|---|---|---|
同态加密[ | 准确性高,隐私保护严格 | 计算开销高,通信开销高,不保护发布模型 | 本地梯度 | 统一预算场景 |
多方安全计算[ | 准确性较高,隐私保护严格 | 通信开销高,协议复杂脆弱 | 本地梯度 | 统一预算场景 |
中心化差分隐私[ | 准确性较高,通信开销低,计算量较低 | 需要中心服务器可信 | 中心梯度 | 统一预算场景 |
本地化差分隐私[ | 通信开销低,扰动方案灵活 | 准确性低 | 本地梯度 | 统一预算场景 |
安全混洗[ | 准确性较高,隐私保护严格 | 计算开销高 | 本地梯度 | 统一预算场景 |
PDP-FL | 可以提供个性化隐私保证,准确性较高 | 需要合理的隐私分级 | 本地和中心梯度 | 个性化预算场景 |
图3 CIFAR-10数据集上多场景下不同隐私预算的分类准确度比较
Fig. 3 Comparison of classification accuracies of different privacy budgets in multiple scenarios on CIFAR-10 dataset
数据集 | 隐私保护 | 算法 | 场景 | 准确度/% | 损失 |
---|---|---|---|---|---|
MNIST | 无 | FedAvg | 无 | 95.10 | 1.2 |
GDP | GDP-FL | 无 | 94.20 | 1.5 | |
LDP | LDP-Fed | 无 | 93.70 | 1.4 | |
PDP | PDP-FL | 场景1 | 93.90 | 1.3 | |
场景2 | 1.2 | ||||
场景3 | 93.80 | 1.3 | |||
场景4 | 94.00 | 1.3 | |||
CIFAR-10 | 无 | FedAvg | 无 | 46.01 | 1.2 |
GDP | GDP-FL | 无 | 44.22 | 1.5 | |
LDP | LDP-Fed | 无 | 43.60 | 1.4 | |
PDP | PDP-FL | 场景1 | 43.55 | 1.4 | |
场景2 | 1.2 | ||||
场景3 | 43.40 | 1.4 | |||
场景4 | 44.10 | 1.3 |
表2 MNIST和CIFAR-10数据集上FedAvg、GDP-FL、LDP-Fed和PDP-FL的对比结果
Tab. 2 Comparison between FedAvg GDP-FL,LDP-Fed and PDP-FL on MNIST CIFAR-10 datasets
数据集 | 隐私保护 | 算法 | 场景 | 准确度/% | 损失 |
---|---|---|---|---|---|
MNIST | 无 | FedAvg | 无 | 95.10 | 1.2 |
GDP | GDP-FL | 无 | 94.20 | 1.5 | |
LDP | LDP-Fed | 无 | 93.70 | 1.4 | |
PDP | PDP-FL | 场景1 | 93.90 | 1.3 | |
场景2 | 1.2 | ||||
场景3 | 93.80 | 1.3 | |||
场景4 | 94.00 | 1.3 | |||
CIFAR-10 | 无 | FedAvg | 无 | 46.01 | 1.2 |
GDP | GDP-FL | 无 | 44.22 | 1.5 | |
LDP | LDP-Fed | 无 | 43.60 | 1.4 | |
PDP | PDP-FL | 场景1 | 43.55 | 1.4 | |
场景2 | 1.2 | ||||
场景3 | 43.40 | 1.4 | |||
场景4 | 44.10 | 1.3 |
1 | LÓPEZ K L, GAGNÉ C, GARDNER M A. Demand-side management using deep learning for smart charging of electric vehicles[J]. IEEE Transactions on Smart Grid, 2019, 10(3): 2683-2691. 10.1109/tsg.2018.2808247 |
2 | LIN W Y, HU Y H, TSAI C F. Machine learning in financial crisis prediction: a survey[J]. IEEE Transactions on Systems, Man, and Cybernetics, Part C (Applications and Reviews), 2012, 42(4): 421-436. 10.1109/tsmcc.2011.2170420 |
3 | CHEN S Z, YU D X, ZOU Y F, et al. Decentralized wireless federated learning with differential privacy[J]. IEEE Transactions on Industrial Informatics, 2022, 18(9):6273-6282. 10.1109/tii.2022.3145010 |
4 | BARRENO M, NELSON B, JOSEPH A D, et al. The security of machine learning[J]. Machine Learning, 2010, 81(2): 121-148. 10.1007/s10994-010-5188-5 |
5 | 张梅舒,徐雅斌. 多维数值型敏感属性数据的个性化隐私保护方法 [J]. 计算机应用, 2020, 40(2):491-496. |
ZHANG M S, XU Y B. Personalized privacy protection method for data with multiple numerical sensitive attributes[J]. Journal of Computer Applications, 2020, 40(2): 491-496. | |
6 | 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 federal learning[J]. Journal of Software, 2022, 33(3): 1057-1092. | |
7 | WU N, FAROKHI F, SMITH D, et al. The value of collaboration in convex machine learning with differential privacy[C]// Proceedings of the 2020 IEEE Symposium on Security and Privacy. Piscataway: IEEE, 2020: 304-317. 10.1109/sp40000.2020.00025 |
8 | WANG B L, YAO Y S, SHAN S, et al. Neural cleanse: identifying and mitigating backdoor attacks in neural networks[C] // Proceedings of the 2019 IEEE Symposium on Security and Privacy. Piscataway: IEEE, 2019: 707-723. 10.1109/sp.2019.00031 |
9 | YUAN X Y, HE P, ZHU Q L, et al. Adversarial examples: attacks and defenses for deep learning[J]. IEEE Transactions on Neural Networks and Learning Systems, 2019, 30(9): 2805-2824. 10.1109/tnnls.2018.2886017 |
10 | YUAN J W, YU S C. Privacy preserving back-propagation neural network learning made practical with cloud computing[J]. IEEE Transactions on Parallel and Distributed Systems, 2014, 25(1): 212-221. 10.1109/tpds.2013.18 |
11 | BONAWITZ K, IVANOV V, KREUTER B, et al. Practical secure aggregation for privacy-preserving machine learning[C]// Proceedings of the 2017 ACM SIGSAC Conference on Computer and Communications Security. New York: ACM, 2017: 1175-1191. 10.1145/3133956.3133982 |
12 | 邱鑫源,叶泽聪,崔翛龙,等. 联邦学习通信开销研究综述[J]. 计算机应用, 2022, 42(2): 333-342. 10.11772/j.issn.1001-9081.2021020232 |
QIU X Y, YE Z C, CUI X L, et al. Survey of communication overhead in federated learning[J]. Journal of Computer Applications, 2022, 42(2): 333-342. 10.11772/j.issn.1001-9081.2021020232 | |
13 | DWORK C, ROTH A. The algorithmic foundations of differential privacy[J]. Foundations and Trends® in Theoretical Computer Science, 2013, 9(3/4): 211-407. 10.1561/0400000042 |
14 | ZHANG J L, ZHAO Y C, WANG J Y, et al. FedMEC: improving efficiency of differentially private federated learning via mobile edge computing[J]. Mobile Networks and Applications, 2020, 25(6): 2421-2433. 10.1007/s11036-020-01586-4 |
15 | TRUEX S, LIU L, CHOW K H, et al. LDP-Fed: federated learning with local differential privacy[C]// Proceedings of the 3rd ACM International Workshop on Edge Systems, Analytics and Networking. New York: ACM, 2020: 61-66. 10.1145/3378679.3394533 |
16 | ZHAO Y, ZHAO J, YANG M M, et al. Local differential privacy-based federated learning for Internet of Things[J]. IEEE Internet of Things Journal, 2021, 8(11): 8836-8853. 10.1109/jiot.2020.3037194 |
17 | WU X, ZHANG Y T, SHI M Y, et al. An adaptive federated learning scheme with differential privacy preserving[J]. Future Generation Computer Systems, 2022, 127: 362-372. 10.1016/j.future.2021.09.015 |
18 | PHONG L T, 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. 10.1109/tifs.2017.2787987 |
19 | SONG J C, WANG W Z, GADEKALLU T R, et al. EPPDA: an efficient privacy-preserving data aggregation federated learning scheme[J]. IEEE Transactions on Network Science and Engineering, 2022(Early Access): 1-1. 10.1109/tnse.2022.3153519 |
20 | GONG M G, FENG J L, XIE Y. Privacy-enhanced multi-party deep learning[J]. Neural Networks, 2020, 121: 484-496. 10.1016/j.neunet.2019.10.001 |
21 | McMAHAN B, MOORE E, RAMAGE D, et al. Communication-efficient learning of deep networks from decentralized data[C]// Proceedings of the 20th International Conference on Artificial Intelligence and Statistics. New York: JMLR.org, 2017: 1273-1282. |
22 | ABADI M, CHU A, GOODFELLOW I, et al. Deep learning with differential privacy[C]// Proceedings of the 2016 ACM SIGSAC Conference on Computer and Communications Security. New York: ACM, 2016: 308-318. 10.1145/2976749.2978318 |
23 | WANG S W, HUANG L S, NIE Y W, et al. Local differential private data aggregation for discrete distribution estimation[J]. IEEE Transactions on Parallel and Distributed Systems, 2019, 30(9): 2046-2059. 10.1109/tpds.2019.2899097 |
24 | ARACHCHIGE P C M, BERTOK P, KHALIL I, et al. Local differential privacy for deep learning[J]. IEEE Internet of Things Journal, 2020, 7(7): 5827-5842. 10.1109/jiot.2019.2952146 |
25 | WANG N, XIAO X K, YANG Y, et al. Collecting and analyzing multidimensional data with local differential privacy[C]// Proceedings of the IEEE 35th International Conference on Data Engineering. Piscataway: IEEE, 2019: 638-649. 10.1109/icde.2019.00063 |
26 | GEYER R C, KLEIN T, NABI M. Differentially private federated learning: a client level perspective[EB/OL]. (2018-03-01) [2022-03-05].. |
27 | 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. 10.1109/tifs.2020.2988575 |
28 | WU S H, YU M Q, AHMED M A M, et al. FL-MAC-RDP: federated learning over multiple access channels with Rényi differential privacy[J]. International Journal of Theoretical Physics, 2021, 60(7): 2668-2682. 10.1007/s10773-021-04867-0 |
29 | LIU R, CAO Y, YOSHIKAWA M, et al. FedSel: federated SGD under local differential privacy with top-k dimension selection[C]// Proceedings of the 2020 International Conference on Database Systems for Advanced Applications. Cham: Springer, 2020: 485-501. 10.1007/978-3-030-59410-7_33 |
30 | 莫慧凌,郑海峰,高敏,等. 基于联邦学习的多源异构数据融合算法[J]. 计算机研究与发展, 2022, 59(2):478-487. 10.7544/issn1000-1239.20200668 |
MO H L, ZHEN H F, GAO M, et al. Multi-source heterogeneous data fusion based on federated learning[J]. Journal of Computer Research and Development, 2022, 59(2): 478-487. 10.7544/issn1000-1239.20200668 | |
31 | ZHAO L C, WANG Q, ZOU Q, et al. Privacy-preserving collaborative deep learning with unreliable participants[J]. IEEE Transactions on Information Forensics and Security, 2020, 15: 1486-1500. 10.1109/tifs.2019.2939713 |
32 | GIRGIS A, DATA D, DIGGAVI S, et al. Shuffled model of differential privacy in federated learning[C]// Proceedings of the 24th International Conference on Artificial Intelligence and Statistics. New York: JMLR.org, 2021: 2521-2529. 10.1109/jsait.2021.3056102 |
33 | ZHANG L F, ZHU T Q, XIONG P, et al. A robust game-theoretical federated learning framework with joint differential privacy[J]. IEEE Transactions on Knowledge and Data Engineering, 2022(Early Access): 1-1. 10.1109/tkde.2021.3140131 |
34 | AVENT B, KOROLOVA A, ZEBER D, et al. BLENDER: enabling local search with a hybrid differential privacy model[C]// Proceedings of the 26th USENIX Security Symposium. Berkeley: USENIX Association, 2017: 747-764. 10.29012/jpc.680 |
35 | HU R, GUO Y X, LI H N, et al. Personalized federated learning with differential privacy[J]. IEEE Internet of Things Journal, 2020, 7(10): 9530-9539. 10.1109/jiot.2020.2991416 |
36 | YANG G, WANG S W, WANG H J. Federated learning with personalized local differential privacy[C]// Proceedings of the IEEE 6th International Conference on Computer and Communication Systems. Piscataway: IEEE, 2021: 484-489. 10.1109/icccs52626.2021.9449232 |
37 | LI H R, XIONG L, JI Z L, et al. Partitioning-based mechanisms under personalized differential privacy[C]// Proceedings of the 2017 Pacific-Asia Conference on Knowledge Discovery and Data Mining, LNCS 10234. Cham: Springer, 2017: 615-627. |
38 | INAN A, GURSOY M E, SAYGIN Y. Sensitivity analysis for non-interactive differential privacy: bounds and efficient algorithms[J]. IEEE Transactions on Dependable and Secure Computing, 2020, 17(1): 194-207. 10.1109/tdsc.2017.2734664 |
39 | JORGENSEN Z, YU T, CORMODE G. Conservative or liberal? personalized differential privacy[C]// Proceedings of the IEEE 31st International Conference on Data Engineering. Piscataway: IEEE, 2015: 1023-1034. 10.1109/icde.2015.7113353 |
40 | ZHANG Y P, QU Y Y, GAO L X, et al. GPDP: game-enhanced personalized differentially private smart community[C]// Proceedings of the 2021 IEEE International Conferences on Internet of Things and IEEE Green Computing and Communications and IEEE Cyber, Physical and Social Computing and IEEE Smart Data and IEEE Congress on Cybermatics. Piscataway: IEEE, 2021: 238-243. 10.1109/ithings-greencom-cpscom-smartdata-cybermatics53846.2021.00048 |
41 | LeCUN Y, BOTTOU L, BENGIO Y, et al. Gradient-based learning applied to document recognition[J]. Proceedings of the IEEE, 1998, 86(11): 2278-2324. 10.1109/5.726791 |
42 | KRIZHEVSKY A. Learning multiple layers of features from tiny images[R/OL]. (2009-04-08) [2022-03-05].. 10.1016/j.tics.2007.09.004 |
[1] | 张治政, 张啸剑, 王俊清, 冯光辉. 结合差分隐私与安全聚集的联邦空间数据发布方法[J]. 《计算机应用》唯一官方网站, 2024, 44(9): 2777-2784. |
[2] | 陈廷伟, 张嘉诚, 王俊陆. 面向联邦学习的随机验证区块链构建[J]. 《计算机应用》唯一官方网站, 2024, 44(9): 2770-2776. |
[3] | 沈哲远, 杨珂珂, 李京. 基于双流神经网络的个性化联邦学习方法[J]. 《计算机应用》唯一官方网站, 2024, 44(8): 2319-2325. |
[4] | 陈学斌, 任志强, 张宏扬. 联邦学习中的安全威胁与防御措施综述[J]. 《计算机应用》唯一官方网站, 2024, 44(6): 1663-1672. |
[5] | 罗玮, 刘金全, 张铮. 融合秘密分享技术的双重纵向联邦学习框架[J]. 《计算机应用》唯一官方网站, 2024, 44(6): 1872-1879. |
[6] | 余孙婕, 曾辉, 熊诗雨, 史红周. 基于生成式对抗网络的联邦学习激励机制[J]. 《计算机应用》唯一官方网站, 2024, 44(2): 344-352. |
[7] | 彭鹏, 倪志伟, 朱旭辉, 陈千. 改进萤火虫群算法协同差分隐私的干扰轨迹发布[J]. 《计算机应用》唯一官方网站, 2024, 44(2): 496-503. |
[8] | 张祖篡, 陈学斌, 高瑞, 邹元怀. 基于标签分类的联邦学习客户端选择方法[J]. 《计算机应用》唯一官方网站, 2024, 44(12): 3759-3765. |
[9] | 高瑞, 陈学斌, 张祖篡. 面向部分图更新的动态社交网络隐私发布方法[J]. 《计算机应用》唯一官方网站, 2024, 44(12): 3831-3838. |
[10] | 巫婕, 钱雪忠, 宋威. 基于相似度聚类和正则化的个性化联邦学习[J]. 《计算机应用》唯一官方网站, 2024, 44(11): 3345-3353. |
[11] | 陈学斌, 屈昌盛. 面向联邦学习的后门攻击与防御综述[J]. 《计算机应用》唯一官方网站, 2024, 44(11): 3459-3469. |
[12] | 张帅华, 张淑芬, 周明川, 徐超, 陈学斌. 基于半监督联邦学习的恶意流量检测模型[J]. 《计算机应用》唯一官方网站, 2024, 44(11): 3487-3494. |
[13] | 陈学斌, 单丽洋, 郭如敏. 基于差分隐私的直方图发布方法综述[J]. 《计算机应用》唯一官方网站, 2024, 44(10): 3114-3121. |
[14] | 尹春勇, 周永成. 双端聚类的自动调整聚类联邦学习[J]. 《计算机应用》唯一官方网站, 2024, 44(10): 3011-3020. |
[15] | 徐雪冉, 杨庚, 黄喻先. 横向联邦学习中差分隐私聚类算法[J]. 《计算机应用》唯一官方网站, 2024, 44(1): 217-222. |
阅读次数 | ||||||
全文 |
|
|||||
摘要 |
|
|||||