基于茫然传输协议的FATE联邦迁移学习方案

doi:10.19678/j.issn.1000-3428.0064452

计算机工程 ›› 2023, Vol. 49 ›› Issue (2): 24-30. doi: 10.19678/j.issn.1000-3428.0064452

基于茫然传输协议的FATE联邦迁移学习方案

郑云涛¹, 叶家炜²

1. 复旦大学计算机科学技术学院上海市智能信息处理重点实验室, 上海 200433;
2. 复旦大学计算机科学技术学院教育部网络信息安全审计与监控工程研究中心, 上海 200433

收稿日期:2022-04-13 修回日期:2022-05-16 发布日期:2022-06-21
作者简介:郑云涛(1997-),男,硕士研究生,主研方向为多方安全计算、区块链、密码学;叶家炜,工程师、硕士。
基金资助:
上海市基础研究重点项目（21JC1400600）。

FATE Federated Transfer Learning Scheme Based on Oblivious Transfer Protocol

ZHENG Yuntao¹, YE Jiawei²

1. Shanghai Key Laboratory of Intelligent Information Processing, School of Computer Science, Fudan University, Shanghai 200433, China;
2. Engineering Research Center of Cyber Security Auditing and Monitoring, Ministry of Education, School of Computer Science, Fudan University, Shanghai 200433, China

Received:2022-04-13 Revised:2022-05-16 Published:2022-06-21

摘要/Abstract

摘要： 利用不同来源的数据参与机器学习模型的训练，能够使得训练出的模型的预测结果更加准确，然而大量的数据收集则会产生隐私方面的相关问题。FATE联邦迁移学习是一种基于同态加密的联邦学习框架，但FATE联邦迁移学习中同态加密计算复杂，收敛速度相对较慢，导致模型训练效率低。提出一种基于茫然传输协议的安全矩阵计算方案。通过实现矩阵加法和乘法及数乘的安全计算，完成参与两方交互下具有数据隐私保护特性机器学习模型的损失函数计算与梯度更新，并以此构造更高效的FATE联邦迁移学习算法方案。在此基础上，通过茫然传输扩展协议和通信批量处理，减少需要调用的茫然传输协议的数量，缩减通信轮数，从而降低茫然传输协议带来的通信消耗。性能分析结果表明，该方案的安全模型满足安全性和隐私保护性，并且具有一定的可扩展性，在局域网环境下，相比基于同态加密的方案，模型收敛的平均时间缩短约25%，并且随着数据样本特征维度的增加，该方案仍能保持稳定的收敛速度。

关键词: 联邦迁移学习, 安全多方计算, 秘密共享, 茫然传输协议, 同态加密

Abstract: Machine learning models can be trained on a large amount of data collected from different sources, thereby improving the prediction accuracy.However, data collection between different organizations causes privacy issues. The Federated Artificial intelligence Technology Enabler (FATE) Federated Transfer Learning (FTL) is a secure machine learning framework based on homomorphic encryption.However, the convergence speed is relatively low owing to the computational efficiency of homomorphic encryption.In this study, a secure multiparty computation scheme for matrix computation based on Oblivious Transfer(OT) protocol is proposed to design a two-party machine-learning scheme to construct a more efficient FATE FTL.In addition, the communication consumption caused by the OT protocol is reduced via OT extension and batch processing.The performance analysis shows that the scheme ensures security, privacy preservation, and scalability in practical applications.On the other hand, the convergence time of the proposed scheme is approximately 25% better than that FATE FTL framework based on a homomorphic encryption scheme in a local-area network environment. As the feature dimension of the data samples increases, the advantage of the convergence speed of this scheme can remain stable, proving that this scheme has a practical application significance.

Key words: Federated Transfer Learning(FTL), secure multiparty computation, secret sharing, Oblivious Transfer(OT) protocol, homomorphic encryption

中图分类号:

TP391

郑云涛, 叶家炜. 基于茫然传输协议的FATE联邦迁移学习方案[J]. 计算机工程, 2023, 49(2): 24-30.

ZHENG Yuntao, YE Jiawei. FATE Federated Transfer Learning Scheme Based on Oblivious Transfer Protocol[J]. Computer Engineering, 2023, 49(2): 24-30.

http://www.ecice06.com/CN/Y2023/V49/I2/24

图/表 4

参考文献

[1] GENTRY C, HALEVI S.Implementing gentry's fully-homomorphic encryption scheme[C]//Proceedings of Annual International Conference on the Theory and Applications of Cryptographic Techniques.Berlin, Germany:Springer, 2011:129-148.
[2] SHAMIR A.How to share a secret[J].Communications of the ACM, 1979, 22(11):612-613.
[3] RABIN M O.How to exchange secrets by oblivious transfer:TR-81[R].Aiken Computation Laboratory, Harvard University, 1981.
[4] DAMGÅRD I, PASTRO V, SMART N, et al.Multiparty computation from somewhat homomorphic encryption[C]//Proceedings of CRYPTOʼ12.Berlin, Germany:Springer, 2012:643-662.
[5] DEMMLER D, SCHNEIDER T, ZOHNER M.ABY-A framework for efficient mixed-protocol secure two-party computation[C]//Proceedings of 2015 Network and Distributed System Security Symposium.Washington D.C., USA:IEEE Press, 2015:257-268.
[6] MCMAHAN B, MOORE E, RAMAGE D, et al.Communication-efficient learning of deep networks from decentralized data[C]//Proceedings of IEEE PMLRʼ17.Washington D.C., USA:IEEE Press, 2017:1273-1282.
[7] PAPERNOT N, ABADI M, ERLINGSSON Ú, et al.Semi-supervised knowledge transfer for deep learning from private training data[EB/OL].[2022-03-10].https://arxiv.org/abs/1610.05755.
[8] YUROCHKIN M, AGARWAL M, GHOSH S, et al.Bayesian nonparametric federated learning of neural networks[EB/OL].[2022-03-10].https://arxiv.org/abs/1905.12022.
[9] YANG T, ANDREW G, EICHNER H, et al.Applied federated learning:improving google keyboard query suggestions[EB/OL].[2022-03-10].https://arxiv.org/abs/1812.02903.
[10] HARDY S, HENECKA W, IVEY-LAW H, et al.Private federated learning on vertically partitioned data via entity resolution and additively homomorphic encryption[EB/OL].[2022-03-10].https://arxiv.org/abs/1711.10677.
[11] SHARMA S, XING C P, LIU Y, et al.Secure and efficient federated transfer learning[C]//Proceedings of 2019 IEEE International Conference on Big Data.Washington D.C., USA:IEEE Press, 2019:2569-2576.
[12] GEYER R C, KLEIN T, NABI M.Differentially private federated learning:a client level perspective[EB/OL].[2022-03-10].https://arxiv.org/abs/1712.07557.
[13] MOHASSEL P, ZHANG Y P.SecureML:a system for scalable privacy-preserving machine learning[C]//Proceedings of IEEE Symposium on Security and Privacy.Washington D.C., USA:IEEE Press, 2017:19-38.
[14] LIU Y, KANG Y, XING C P, et al.A secure federated transfer learning framework[J].IEEE Intelligent Systems, 2020, 35(4):70-82.
[15] PAILLIER P.Public-key cryptosystems based on composite degree residuosity classes[C]//Proceedings of International Conference on the Theory and Applications of Cryptographic Techniques.Berlin, Germany:Springer, 1999:223-238.
[16] ZHANG C, LI S, XIA J, et al.BatchCrypt:efficient homomorphic encryption for cross-silo federated learning[C]//Proceedings of 2020 USENIX Annual Technical Conference.Washington D.C., USA:IEEE Press, 2020:493-506.
[17] KONEČNÝ J, MCMAHAN H B, RAMAGE D, et al.Federated optimization:distributed machine learning for on-device intelligence[EB/OL].[2022-03-10].https://arxiv.org/abs/1610.02527.
[18] NAOR M, PINKAS B.Efficient oblivious transfer protocols[C]//Proceedings of the 12th Annual ACM-SIAM Symposium on Discrete Algorithms.New York, USA:ACM Press, 2001:448-457.
[19] ISHAI Y, KILIAN J, NISSIM K, et al.Extending oblivious transfers efficiently[C]//Proceedings of CRYPTO'03.Berlin, Germany:Springer, 2003:145-161.
[20] AONO Y, HAYASHI T, PHONG L T, et al.Scalable and secure logistic regression via homomorphic encryption[C]//Proceedings of the 6th ACM Conference on Data and Application Security and Privacy.New York, USA:ACM Press, 2016:142-144.
[21] 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.
[22] KIM A, SONG Y, KIM M, et al.Logistic regression model training based on the approximate homomorphic encryption[J].BMC Medical Genomics, 2018, 11(4):83.
[23] CHEN H, KIM M, RAZENSHTEYN I, et al.Maliciously secure matrix multiplication with applications to private deep learning[C]//Proceedings of ASIACRYPT'20.Berlin, Germany:Springer, 2020:31-59.
[24] BOYLE E, COUTEAU G, GILBOA N, et al.Efficient pseudorandom correlation generators:silent OT extension and more[C]//Proceedings of CRYPTO'19.Berlin, Germany:Springer, 2019:489-518.
[25] BONAWITZ K, EICHNER H, GRIESKAMP W, et al.Towards federated learning at scale:system design[EB/OL].[2022-03-10].https://arxiv.org/abs/1902.01046.

选择文件类型/文献管理软件名称

选择包含的内容

基于茫然传输协议的FATE联邦迁移学习方案

FATE Federated Transfer Learning Scheme Based on Oblivious Transfer Protocol

RichHTML

PDF

可视化

被引次数

摘要/Abstract

引用本文

使用本文

图/表 4

参考文献

相关文章 15

编辑推荐

Metrics

本文评价

[1]	杨越佳, 华蓓, 钟志威, 高咪. 基于同态加密的隐私保护逻辑回归协同计算[J]. 计算机工程, 2023, 49(4): 23-31.
[2]	唐敏, 张宇浩, 邓国强. 一种高效的非交互式隐私保护逻辑回归模型[J]. 计算机工程, 2023, 49(4): 32-42,51.
[3]	徐鑫, 温蜜. 基于数据库驱动认知无线电网络的位置隐私保护方案[J]. 计算机工程, 2022, 48(2): 164-172.
[4]	王元庆, 刘百祥. 一种去中心化的隐私保护匿名问卷方案[J]. 计算机工程, 2021, 47(6): 123-131.
[5]	张晓均, 张经伟, 黄超, 唐伟. 可验证的云存储医疗加密数据统计分析方案[J]. 计算机工程, 2021, 47(6): 32-37,43.
[6]	李秋贤, 周全兴, 王振龙, 丁红发, 潘齐欣. 基于隐私保护的可证明安全委托计算协议[J]. 计算机工程, 2021, 47(5): 131-137.
[7]	张静, 何铮, 葛炳辉, 汤永利, 叶青. 一种高效的百万富翁问题协议及其应用[J]. 计算机工程, 2021, 47(2): 168-175.
[8]	夏高, 何成万. 一种基于异或运算的(k,n)门限秘密共享算法[J]. 计算机工程, 2021, 47(10): 111-115,124.
[9]	陈丽燕, 芮廷先, 吕光金. 基于区块链智能合约的个人征信隐私保护方案[J]. 计算机工程, 2020, 46(7): 30-35.
[10]	周能, 张敏情, 林文兵. 基于秘密共享的可分离密文域可逆信息隐藏算法[J]. 计算机工程, 2020, 46(10): 112-119.
[11]	刘文超, 潘峰, 杨晓元, 周潭平, 涂广升. 基于cuFHE的同态比较运算器[J]. 计算机工程, 2019, 45(9): 143-146,152.
[12]	解扬,苗付友,白建峰. 基于整数规划的一般访问结构秘密共享方案[J]. 计算机工程, 2019, 45(6): 165-170.
[13]	王彩芬,赵冰,刘超,成玉丹,许钦百. 基于整数多项式环的多对一全同态加密算法[J]. 计算机工程, 2019, 45(4): 130-135.
[14]	李志佳, 夏玮. 基于差值直方图平移的密文域可逆信息隐藏算法[J]. 计算机工程, 2019, 45(11): 152-158.
[15]	沈鑫娣,翟东君,张得天,刘安. 基于LSH的隐私保护POI推荐算法[J]. 计算机工程, 2019, 45(1): 96-102.

模态框（Modal）标题

选择文件类型/文献管理软件名称

选择包含的内容

基于茫然传输协议的FATE联邦迁移学习方案

FATE Federated Transfer Learning Scheme Based on Oblivious Transfer Protocol

RichHTML

PDF

可视化

被引次数

摘要/Abstract

引用本文

使用本文

图/表 4

参考文献

相关文章 15

编辑推荐

Metrics

本文评价