面向隐私保护的非聚合式数据共享综述

doi:10.11959/j.issn.1000-436x.2021120

通信学报 ›› 2021, Vol. 42 ›› Issue (6): 195-212.doi: 10.11959/j.issn.1000-436x.2021120

所属专题：联邦学习

面向隐私保护的非聚合式数据共享综述

李尤慧子¹, 殷昱煜¹, 高洪皓²^,³, 金一⁴, 王新珩⁵

¹ 杭州电子科技大学计算机学院，浙江杭州 310018
² 上海大学计算机工程与科学学院，上海 200444
³ 韩国嘉泉大学计算机工程系，城南市 461701
⁴ 北京交通大学计算机与信息技术学院，北京 100044
⁵ 西交利物浦大学电气与电子工程系，江苏苏州 215123

修回日期:2021-03-10 出版日期:2021-06-25 发布日期:2021-06-01
作者简介:李尤慧子（1989− ），女，河南新蔡人，博士，杭州电子科技大学副教授，主要研究方向为边缘计算、隐私安全、移动互联网计算、高能效系统
殷昱煜（1980− ），男，重庆人，博士，杭州电子科技大学教授，主要研究方向为边缘计算、服务计算、大数据分析、软件形式化方法等
高洪皓（1985− ），男，浙江临海人，博士，上海大学副教授、韩国嘉泉大学教授，主要研究方向为软件形式化验证、服务协同计算、无线网络和工业物联网、智能医学影像处理等
金一（1982− ），女，河北石家庄人，博士，北京交通大学教授、博士生导师，主要研究方向为机器学习与认知计算、人工智能及应用、图像感知与识别
王新珩（1968− ），男，山东平度人，博士，西交利物浦智能工程学院教授、博士生导师，主要研究方向为物联网、室内定位、智能化服务和智慧城市
基金资助:
国家重点研发计划基金资助项目(2020YFB2103805);国家自然科学基金资助项目(61802093);国家自然科学基金资助项目(61972358)

Survey on privacy protection in non-aggregated data sharing

Youhuizi LI¹, Yuyu YIN¹, Honghao GAO²^,³, Yi JIN⁴, Xinheng WANG⁵

¹ School of Computer Science, Hangzhou Dianzi University, Hangzhou 310018, China
² School of Computer Engineering and Science, Shanghai University, Shanghai 200444, China
³ Department of Computer Engineering, Gachon University, Seongnam 461701, South Korea
⁴ School of Computer and Information Technology, Beijing Jiaotong University, Beijing 100044, China
⁵ Department of Electrical and Electronic Engineering, Xi'an Jiaotong-Liverpool University, Suzhou 215123, China

Revised:2021-03-10 Online:2021-06-25 Published:2021-06-01
Supported by:
The National Key Research and Development Program of China(2020YFB2103805);The National Natural Science Foundation of China(61802093);The National Natural Science Foundation of China(61972358)

摘要/Abstract

摘要：

海量数据价值虽高但与用户隐私关联也十分密切，以高效安全地共享多方数据且避免隐私泄露为目标，介绍了非聚合式数据共享领域的研究发展。首先，简述安全多方计算及其相关技术，包括同态加密、不经意传输、秘密共享等；其次，分析联邦学习架构，从源数据节点和通信传输优化方面探讨现有研究；最后，整理对比面向隐私保护的非聚合式数据共享框架，为后续研究方案构建和运行提供支撑。此外，总结提出非聚合式数据共享领域的挑战和潜在的研究方向，如复杂多参与方场景、优化开销平衡、相关安全隐患等。

关键词: 隐私保护, 数据共享, 联邦学习, 安全多方计算

Abstract:

Although there is a great value hidden in the massive data, it can also easily expose user privacy.Aiming at efficiently and securely sharing data from multiple parties and avoiding leakage of user private information, the development of related research and technologies on the non-aggregated data sharing field was introduced.Firstly, secure multi-party computing and its technologies were briefly described, including homomorphic encryption, oblivious transfer, secret sharing, etc.Secondly, the federated learning architecture was analyzed from the aspects of source data nodes and transmission optimization.Finally, the existing non-aggregated data sharing frameworks were listed and compared.In addition, the challenges and future potential research directions were summarized, such as complex multi-party scenarios, the balance between optimization and cost, as well as related security risks.

Key words: privacy protection, data sharing, federated learning, secure multi-party computation

中图分类号:

TN92

李尤慧子, 殷昱煜, 高洪皓, 金一, 王新珩. 面向隐私保护的非聚合式数据共享综述[J]. 通信学报, 2021, 42(6): 195-212.

Youhuizi LI, Yuyu YIN, Honghao GAO, Yi JIN, Xinheng WANG. Survey on privacy protection in non-aggregated data sharing[J]. Journal on Communications, 2021, 42(6): 195-212.

图/表 17

图1

图2

表1

表2

图3

表3

表4

图4

表5

表6

图5

图6

图7

图8

图9

图10

图11

参考文献 86

[87]	ALEDHARI M , RAZZAK R , PARIZI R M ,et al. Federated learning:a survey on enabling technologies,protocols,and applications[J]. IEEE Access, 2020,8: 699-725.
[88]	LE H Q , NGUYEN M N , HONGF C S . A federated learning for+mage classification with heterogeneous data[J]. Journal of Korea Information Science Society, 2020: 352-354.
[89]	WANG H Y , SREENIVASAN K , RAJPUT S ,et al. Attack of the tails:yes,you really can backdoor federated learning[J]. arXiv Preprint,arXiv:2007.05084, 2020.
[90]	YU B , QIU W Y , CHEN C ,et al. SubMito-XGBoost:predicting protein submitochondrial localization by fusing multiple feature information and eXtreme gradient boosting[J]. Bioinformatics, 2020,36(4): 1074-1081.
[91]	MCMAHAN H B , ANDREW G . A general approach to adding differential privacy to iterative training procedures[J]. arXiv Preprint,arXiv:1812.06210, 2018.
[92]	AUGENSTEIN S , MCMAHAN H B , RAMAGE D ,et al. Generative models for effective ML on private,decentralized datasets[J]. arXiv Preprint,arXiv:1911.06679, 2019.
[93]	GONZALEZ E , LEDIJU BELL M A . A GPU approach to real-time coherence-based photoacoustic imaging and its application to photoacoustic visual servoing[J]. Photons Plus Ultrasound:Imaging and Sensing,, 2020,1124:1124054.
[94]	LI W Q , MILLETARì F , XU D G ,et al. Privacy-preserving federated brain tumour segmentation[C]// Machine Learning in Medical Imaging. Berlin:Springer, 2019: 133-141.
[95]	RANGAN R , TURAKHIA N , JOLY A . Corrections to “countering load-to-use stalls in the NVIDIA Turing GPU”[J]. IEEE Micro, 2021,41(1): 83.
[96]	WANG Z S , BANAWAN K , ULUKUS S . Multi-party private set intersection:an information-theoretic approach[J]. IEEE Journal on Selected Areas in Information Theory, 2021,2(1): 366-379.
[97]	KOLESNIKOV V , MATANIA N , PINKAS B ,et al. Practical multi-party private set intersection from symmetric-key techniques[C]// The 2017 ACM SIGSAC Conference on Computer and Communications Security. New York:ACM Press, 2017: 257-272.
[98]	ALI M , MOHAJERI J , SADEGHI M R ,et al. Attribute-based fine-grained access control for outscored private set intersection computation[J]. Information Sciences, 2020,536: 222-243.
[99]	ZHANG Y , DUCHI J C , WAINWRIGHT M J . Communication- efficient algorithms for statistical optimization[J]. The Journal of Machine Learning Research, 2013,14(1): 3321-3363.
[100]	GOODFELLOW I J , POUGET-ABADIE J , MIRZA M ,et al. Generative adversarial nets[C]// International Conference on Neural Information Processing Systems. Berlin:Springer, 2014: 2672-2680.
[101]	LIU Y , PENG J L , YU J J Q ,et al. PPGAN:privacy-preserving generative adversarial network[C]// 2019 IEEE 25th International Conference on Parallel and Distributed Systems. Piscataway:IEEE Prss, 2019: 985-989.
[102]	XIE L Y , LIN K X , WANG S ,et al. Differentially private generative adversarial network[J]. arXiv Preprint,arXiv:1802.06739, 2018.
[103]	DUCHI J C , JORDAN M I , WAINWRIGHT M J . Local privacy,data processing inequalities,and statistical minimax rates[J]. arXiv Preprint,arXiv:1302.3203, 2013.
[104]	QIN Z , YU T , YANG Y ,et al. Generating synthetic decentralized social graphs with local differential privacy[C]// The 2017 ACM SIGSAC Conference on Computer and Communications Security. New York:ACM Press, 2017: 425-438.
[1]	于文平 . 中国物联网市场发展现状分析及未来五年发展趋势预测[J]. 物联网技术, 2018,8(3): 9-11.
	YU W P . Analysis of the development status of China's Internet of things market and forecast of its development trend in the next five years[J]. Internet of Things Technologies, 2018,8(3): 9-11.
[105]	张佳程, 彭佳, 王雷 . 大数据环境下的本地差分隐私图信息收集方法[J]. 信息网络安全, 2020,20(6): 44-56.
	ZHANG J C , PENG J , WANG L . A graph information collection method based on local differential privacy in big data environment[J]. Netinfo Security, 2020,20(6): 44-56.
[2]	瑞星 . 2020 年中国网络安全报告[J]. 信息安全研究, 2021,7(2): 102-109.
	RISING . China’s cybersecurity report in 2020[J]. Journal of Informa-tion Security Research, 2021,7(2): 102-109.
[3]	REGULATION G D P . Regulation (EU) 2016/679 of the European Parlia ment and of the Council of 27 April 2016 on the protection of natural persons with regard to the processing of personal data and on the free movement of such data,and repealing Directive 95/46[J]. Official Journal of the European Union (OJ). 2016,59(1-88): 294.
[4]	YAO A C , . Protocols for secure computation[C]// 23rd IEEE Annual Symposium on Foundations of Computer Science. Piscataway:IEEE Press, 1982: 160-164.
[5]	WINTERNITZ R S , . A secure one-way hash function built from DES[C]// 1984 IEEE Symposium on Security and Privacy. Piscataway:IEEE Press, 1984:88.
[6]	SHI W S , CAO J , ZHANG Q ,et al. Edge computing:vision and challenges[J]. IEEE Internet of Things Journal, 2016,3(5): 637-646.
[7]	ION M , KREUTER B , NERGIZ E ,et al. Private intersection-sum protocol with applications to attributing aggregate AD conversions[J]. IACR Cryptology ePrint Archive, 2017,2017:738
[8]	RIVEST R L , ADLEMAN L , DERTOUZOS M L . On data banks and privacy homomorphisms[J]. Foundations of Secure Computation, 1978,4(11): 169-180.
[9]	RIVEST R L , SHAMIR A , ADLEMAN L . A method for obtaining digital signatures and public-key cryptosystems[J]. Communications of the ACM, 1978,21(2): 120-126.
[10]	PAILLIER P , . Public-key cryptosystems based on composite degree residuosity classes[C]// International Conference on the Theory and Application of Cryptographic Techniques. Berlin:Springer, 1999: 223-238.
[11]	ABADI A , TERZIS S , METERE R ,et al. Efficient delegated private set intersection on outsourced private datasets[J]. IEEE Transactions on Dependable and Secure Computing, 2019,16(4): 608-624.
[12]	RUAN O , WANG Z H , MI J ,et al. New approach to set representation and practical private set-intersection protocols[J]. IEEE Access, 2019,7: 64897-64906.
[13]	KAVOUSI A , MOHAJERI J , SALMASIZADEH M . Improved secure efficient delegated private set intersection[C]// 2020 28th Iranian Conference on Electrical Engineering. Piscataway:IEEE Press, 2020: 1-6.
[14]	RESENDE A C D , ARANHA D F . Faster unbalanced private set intersection[M]. Berlin: Springer, 2018.
[15]	BALDI P , BARONIO R , DE CRISTOFARO E ,et al. Countering GATTACA:efficient and secure testing of fully-sequenced human genomes[C]// The 18th ACM Conference on Computer and communications Security. New York:ACM Press, 2011: 691-702.
[16]	CHEN H , LAINE K , RINDAL P . Fast private set intersection from homomorphic encryption[C]// The 2017 ACM SIGSAC Conference on Computer and Communications Security. New York:ACM Press, 2017:299.
[17]	FREEDMAN M J , NISSIM K , PINKAS B . Efficient private matching and set intersection[C]// Advances in Cryptology-EUROCRYPT. Berlin:Springer, 2004: 1-9.
[18]	INBAR R , OMRI E , PINKAS B . Efficient scalable multiparty private set-intersection via garbled bloom filters[C]// Security and Cryptography for Networks. Berlin:Springer, 2018: 235-252.
[19]	HUANG Y , CHAPMAN P , EVANS D . Privacy-preserving applications on Smartphones[C]// 6th USENIX Workshop on Hot Topics in Security. Berkeley:USENIX Association, 2011: 1-6.
[20]	PINKAS B , SCHNEIDER T , TKACHENKO O ,et al. Efficient circuit-based PSI with linear communication[C]// Advances in Cryptology - EUROCRYPT 2019. Berlin:Springer, 2019: 122-153.
[21]	PINKAS B , SCHNEIDER T , WEINERT C ,et al. Efficient circuit-based PSI via Cuckoo Hashing[C]// EUROCRYPT 2018. Berlin:Springer, 2018: 125-157.
[22]	HUANG H F , CHANG C C . A new design for efficent t-out-n oblivious transfer scheme[C]// 19th International Conference on Advanced Information Networking and Applications. Los Alamitos:IEEE Computer Society, 2005: 28-30.
[23]	徐彦蛟, 李顺东, 王道顺 ,等. 基于椭圆曲线公钥系统的不经意传输协议[J]. 计算机科学, 2013,40(12): 186-191.
	XU Y J , LI S D , WANG D S ,et al. Oblivious transfer based on elliptic curve public key cryptosystems[J]. Computer Science, 2013,40(12): 186-191.
[24]	ISHAI Y , KILIAN J , NISSIM K ,et al. Extending oblivious transfers efficiently[C]// Advances in Cryptology - CRYPTO 2003. Berlin:Springer, 2003: 145-161.
[25]	ASHAROV G , LINDELL Y , SCHNEIDER T ,et al. More efficient oblivious transfer and extensions for faster secure computation[C]// The 2013 ACM SIGSAC Conference on Computer ＆ communications Security. New York:ACM Press, 2013: 673-701.
[26]	CHASE M , MIAO P H . Private set intersection in the Internet setting from lightweight oblivious PRF[C]// Advances in Cryptology CRYPTO 2020. Berlin:Springer, 2020: 34-63.
[27]	CIAMPI M , ORLANDI C . Combining private set-intersection with secure two-party computation[C]// Security and Cryptography for Networks. Berlin:Springer, 2018:105.
[28]	KARAKO? F , KüP?ü A . Linear complexity private set intersection for secure two-party protocols[C]// Cryptology and Network Security. Cham:Springer International Publishing, 2020: 409-429.
[29]	PINKAS B , ROSULEK M , TRIEU N ,et al. SpOT-light:lightweight private set intersection from sparse OT extension[C]// Advances in Cryptology - CRYPTO 2019. Berlin:Springer, 2019: 401-431.
[30]	PINKAS B , SCHNEIDER T , ZOHNER M . Scalable private set intersection based on OT extension[J]. ACM Transactions on Privacy and Security, 2018,21(2): 1-35.
[31]	RINDAL P , ROSULEK M . Improved private set intersection against malicious adversaries[C]// Advances in Cryptology - EUROCRYPT 2017. Berlin:Springer, 2017: 235-259.
[32]	RINDAL P , ROSULEK M . Malicious-secure private set intersection via dual execution[C]// The 2017 ACM SIGSAC Conference on Computer and Communications Security. New York:ACM Press, 2017: 1229-1242.
[33]	GHOSH S , NILGES T . An algebraic approach to maliciously secure private set intersection[C]// Advances in Cryptology - EUROCRYPT 2019. Berlin:Springer, 2019:1064.
[34]	KOLESNIKOV V , KUMARESAN R , ROSULEK M ,et al. Efficient batched oblivious PRF with applications to private set intersection[C]// The 2016 ACM SIGSAC Conference on Computer and Communications Security. New York:ACM Press, 2016: 818-829.
[35]	DONG C Y , CHEN L Q , WEN Z K . When private set intersection meets big data:an efficient and scalable protocol[C]// The 2013 ACM SIGSAC Conference On Computer ＆ Communications Security. New York:ACM Press, 2013: 789-800.
[36]	GROCE A , RINDAL P , ROSULEK M . Cheaper private set intersection via differentially private leakage[J]. Proceedings on Privacy Enhancing Technologies, 2019,2019(3): 6-25.
[37]	SHAMIR A . How to share a secret[J]. Communications of the ACM, 1979,22(11): 612-613.
[38]	CHENG K W , FAN T , JIN Y L ,et al. SecureBoost:a lossless federated learning framework[J]. IEEE Intelligent Systems, 2561,pp(99): 1.
[39]	CRAMER R , DAMG?RD I , MAURER U . General secure multi-party computation from any linear secret-sharing scheme[C]// Advances in Cryptology-EUROCRYPT 2000. Berlin:Springer, 2000: 316-334.
[40]	RABIN M O , . Randomized Byzantine generals[C]// 24th Annual Symposium on Foundations of Computer Science. Piscataway:IEEE Press, 1983: 403-409.
[41]	LYU S , YE J H , YIN S J ,et al. Unbalanced private set intersection cardinality protocol with low communication cost[J]. Future Generation Computer Systems, 2020,102: 1054-1061.
[42]	MIAO P H , PATEL S , RAYKOVA M ,et al. Two-sided malicious security for private intersection-sum with cardinality[C]// Advances in Cryptology - CRYPTO 2020. Berlin:Springer, 2020: 3-33.
[43]	SHI R H , ZHANG M W . A feasible quantum protocol for private set intersection cardinality[J]. IEEE Access, 2019,7: 105-112.
[44]	DONG C Y , LOUKIDES G . Approximating private set union/intersection cardinality with logarithmic complexity[J]. IEEE Transactions on Information Forensics and Security, 2017,12(11): 2792-2806.
[45]	CERULLI A , CRISTOFARO E , SORIENTE C . Nothing refreshes like a RePSI:reactive private set intersection[C]// Applied Cryptography and Network Security. Berlin:Springer, 2018: 280-300
[46]	ATENIESE G , CRISTOFARO E , TSUDIK G . (if) size matters:size-hiding private set intersection[C]// Public Key Cryptography. Berlin:Springer, 2011: 156-173.
[47]	YANG Q , LIU Y , CHEN T J ,et al. Federated machine learning[J]. ACM Transactions on Intelligent Systems and Technology, 2019,10(2): 1-19.
[48]	ZHAO Y , LI M , LAI L ,et al. Federated learning with non-IID data[J]. arXiv Preprint,arXiv:1806.00582, 2018.
[49]	GOETZ J , MALIK K , BUI D ,et al. Active federated learning[J]. arXiv Preprint,arXiv:1909.12641, 2019.
[50]	KANG J W , XIONG Z H , NIYATO D ,et al. Incentive mechanism for reliable federated learning:a joint optimization approach to combining reputation and contract theory[J]. IEEE Internet of Things Journal, 2019,6(6): 10700-10714.
[51]	YOSHIDA N , NISHIO T , MORIKURA M ,et al. Hybrid-FL for wireless networks:cooperative learning mechanism using non-IID data[C]// IEEE International Conference on Communications. Piscataway:IEEE Press, 2020: 1-7.
[52]	SARIKAYA Y , ERCETIN O . Motivating workers in federated learning:a stackelberg game perspective[J]. IEEE Networking Letters, 2020,2(1): 23-27.
[53]	WANG S Q , TUOR T , SALONIDIS T ,et al. Adaptive federated learning in resource constrained edge computing systems[J]. IEEE Journal on Selected Areas in Communications, 2019,37(6): 1205-1221.
[54]	LIU L M , ZHANG J , SONG S H ,et al. Client-edge-cloud hierarchical federated learning[C]// IEEE International Conference on Communications. Piscataway:IEEE Press, 2020: 1-6.
[55]	SHOKRI R , SHMATIKOV V . Privacy-preserving deep learning[C]// The 22nd ACM SIGSAC Conference on Computer and Communications Security. New York:ACM Press, 2015: 1310-1321.
[56]	KONE?NY JB , MCMAHAN B , RAMAGE D . Federated optimization:Distributed optimization beyond the datacenter[J]. arXiv Preprint,arXiv:1511.03575, 2015.
[57]	LI T , SAHU A K , ZAHEER M ,et al. Federated optimization in heterogeneous networks[J]. arXiv Preprint,arXiv:1812.06127, 2018.
[58]	CHEN X , CHEN T , SUN H ,et al. Distributed training with heterogeneous data:bridging median-and mean-based algorithms[J]. arXiv Preprint,arXiv:1906.01736, 2019.
[59]	KARIMIREDDY S P , KALE S , MOHRI M ,et al. Scaffold:Stochastic controlled averaging for federated learning[J]. arXiv Preprint,arXiv:1910.06378, 2019.
[60]	XIE C , KOYEJO S , GUPTA I . Asynchronous federated optimization[J]. arXiv Preprint,arXiv:1903.03934, 2019.
[61]	LI D , WANG J . FedMD:Heterogenous federated learning via model distillation[J]. arXiv Preprint,arXiv:1910.03581, 2019.
[62]	LIU L M , ZHANG J , SONG S H ,et al. Edge-assisted hierarchical federated learning with non-IID data[J]. arXiv Preprint,arXiv:1905.06641, 2019.
[63]	RAMANAN P , NAKAYAMA K . BAFFLE:blockchain based aggregator free federated learning[C]// 2020 IEEE International Conference on Blockchain. Piscataway:IEEE Press, 2020: 72-81.
[64]	ROY A G , SIDDIQUI S , P?LSTERL S ,et al. BrainTorrent:a peer-to-peer environment for decentralized federated learning[J]. arXiv Preprint,arXiv:1905.06731, 2019.
[65]	LIU Y , KANG Y , ZHANG X W ,et al. A communication efficient collaborative learning framework for distributed features[J]. arXiv Preprint,arXiv:1912.11187, 2019.
[66]	YAO X , HUANG C F , SUN L F . Two-stream federated learning:reduce the communication costs[C]// 2018 IEEE Visual Communica tions and Image Processing. Piscataway:IEEE Press, 2018: 1-4.
[67]	STICH S U . Local SGD converges fast and communicates little[J]. arXiv Preprint,arXiv:1805.09767, 2018.
[68]	WANG J Y , JOSHI G . Adaptive communication strategies to achieve the best error-runtime trade-off in local-update SGD[J]. arXiv Preprint,arXiv:1810.08313, 2018.
[69]	MOHRI M , SIVEK G , SURESH A T . Agnostic federated learning[J]. arXiv Preprint,arXiv:1902.00146, 2019.
[70]	ALISTARH D , GRUBIC D , LI J ,et al. QSGD:communication-efficient SGD via gradient quantization and encoding[C]// Advances in Neural Information Processing Systems. Massachusetts:MIT Press, 2017: 1709-1720.
[71]	KHALED A , RICHTáRIK P . Gradient descent with compressed iterates[J]. arXiv Preprint,arXiv:1909.04716, 2019.
[72]	VOGELS T , KARIMIREDDY S P , JAGGI M . PowerSGD:practical low-rank gradient compression for distributed optimization[C]// Advances in Neural Information Processing Systems. Massachusetts:MIT Press, 2019: 14259-14268.
[73]	BERNSTEIN J , ZHAO J , AZIZZADENESHELI K ,et al. SignSGD with majority vote is communication efficient and fault tolerant[J]. arXiv Preprint,arXiv:1810.05291, 2018.
[74]	JIANG Y H , KONE?NY J , RUSH K ,et al. Improving federated learning personalization via model agnostic meta learning[J]. arXiv Preprint,arXiv:1909.12488, 2019.
[75]	SHEN T , ZHANG J , JIA X K ,et al. Federated mutual learning[J]. arXiv Preprint,arXiv:2006.16765, 2020.
[76]	SAVAZZI S , NICOLI M , RAMPA V . Federated learning with cooperating devices:a consensus approach for massive IoT networks[J]. IEEE Internet of Things Journal, 2020,7(5): 4641-4654.
[77]	SATTLER F , WIEDEMANN S,MüLLER K R ,et al. Robust and communication-efficient federated learning from non-I.I.D.data[J]. IEEE Transactions on Neural Networks and Learning Systems, 2019,31(9): 3400-3413.
[78]	KANG J W , XIONG Z H , NIYATO D ,et al. Incentive design for efficient federated learning in mobile networks:a contract theory approach[C]// 2019 IEEE VTS Asia Pacific Wireless Communications Symposium. Piscataway:IEEE Press, 2019: 1-5.
[79]	MCMAHAN B , MOORE E , RAMAGE D ,et al. Communication-efficient learning of deep networks from decentralized data[C]// Artificial Intelligence and Statistics. New York:ACM Press, 2017: 1273-1282.
[80]	KISS á , LIU J , SCHNEIDER T ,et al. Private set intersection for unequal set sizes with mobile applications[J]. Proceedings on Privacy Enhancing Technologies, 2017,2017(4): 177-197.
[81]	NASR M , SHOKRI R , HOUMANSADR A . Improving deep learning with differential privacy using gradient encoding and denoising[J]. arXiv Preprint,arXiv:2007.11524, 2020.
[82]	MOHASSEL P , RINDAL P . ABY3:a mixed protocol framework for machine learning[C]// The 2018 ACM SIGSAC Conference on Computer and Communications Security. New York:ACM Press, 2018: 35-52.
[83]	何雯, 白翰茹, 李超 . 基于联邦学习的企业数据共享探讨[J]. 信息与电脑(理论版), 2020,32(8): 173-176.
	HE W , BAI H R , LI C . Research on enterprise data sharing based on federated learning[J]. China Computer ＆ Communication, 2020,32(8): 173-176.
[84]	ZHANG C , LI S , XIA J ,et al. Batchcrypt:efficient homomorphic encryption for cross-silo federated learning[C]// USENIX Annual Technical Conference,Berkeley:USENIX Association, 2020: 493-506.
[85]	FRIEDMAN J , HASTIE T , TIBSHIRANI R . Additive logistic regression:a statistical view of boosting (with discussion and a rejoinder by the authors)[J]. The Annals of Statistics, 2000,28(2): 337-407.
[86]	GANG L , CHAWATHE S S . Privacy-preserving inter-database operations[C]// 2nd Symposium on Intelligence and Security Informatics. Berlin:Springer, 2004: 66-82.

类型	文献及协议		应用场景	安全	支持多方	输出	运行时间/s （输入元素个数为2¹⁶）	输入上限k （输入元素个数为2^k）	备注
	文献[11]	O-PSI	云外包	半诚实	√	交集	—	20	—
同态加密		EO-PSI					415.8
	文献[12]	PSI	云外包	半诚实	×	交集大小	1 670	20	—
	文献[18]	PSI	本地	半诚实	√	交集	32.78	19	—
	文献[20]	Basic	本地	半诚实	×	交集	13.767	20	LAN
逻辑电路		Advanced					9.763
逻辑电路	文献[21]	Separate	本地	半诚实	×	交集	11.251	20	LAN
		Combined					9.076		1 Thread
	文献[26]	PSI	本地	半诚实	×	交集	0.63	24	LAN
				恶意
	文献[29]	Spot-low	本地	半诚实	×	交集	13.7	24	100 Mbit/s
不经意传输		Spot-fast		恶意			2.91		1 Thread
不经意传输	文献[30]	PSI	本地	半诚实	×	交集	0.702	24	LAN
									1 Thread
	文献[31]	PSI	本地	恶意	×	交集	9.7	20	LAN
	文献[32]	EC-ROM	本地	恶意	×	交集	0.94	24	LAN
		DE-ROM					1.3		1 Thread
注：O-PSI在输入为2¹⁵时，运行时间达到57 357 s，时间过大，故未标出。

模型个性化程度	文献	对个性化模型的保护	全局模型直接可用	接入设备动态改变	局限性	模型个性化方法
完全自主	文献[62]	支持	不支持	自适应	对持有更多样本数据的参与者不公平	知识蒸馏
模型体系结构由中央服务器控制	文献[63]	无	支持	自适应	模型个性化程度不高	模型不可知元学习
完全自主	文献[64]	无	不支持	不支持	需要公共数据集	知识蒸馏

解决方法	文献	核心方法/特点	局限性
	文献[49]	根据值函数选择一个优化（有用程度较高）的数据子集	客户不断收集（并可能获取）数据，并且在许多情况下，分发可能是不稳定的
参与者选择	文献[61]	主动管理客户端，FedCS为客户端设置了在FL（federated learning）协议中下载、更新和上载ML模型的最后期限	在动态的场景中工作平均资源量以及更新和上载所需的时间会动态波动
	文献[76]	只有极少数客户端（例如，少于1％）允许其数据上传到FL服务器。通过考虑每个客户端的数据分布和通道条件来计划数据上载客户端和模型上载客户端	考虑客户的能源消耗
	文献[77]	采用deepQ学习算法，该算法允许服务器学习和找到最佳决策	未考虑Non-IID数据
	文献[62]	契约理论	确保本地模型更新的可靠性
奖励机制	文献[63]	将声誉与合同理论相结合的有效激励机制	只考虑了一个联邦服务器
	文献[64]	斯塔克伯格博弈的平衡解	服务器获利较低，只考虑了一个联邦服务器

名称	有损压缩	特点	下行压缩	对Non-IID数据有稳健性	压缩的倍数/通信时间缩小倍数（选取最好情况）	测试数据集
深度梯度压缩^[70]	×	动量修正、局部梯度剪裁、动量因子映射	×	√	压缩了270～600倍	CIFAR10、ImageNet、Penn Treebank、AN4、Librispeech
量化SGD^[71]	√	量化、黑盒压缩	×	×	训练减少了2.5～6.8倍	ImageNet、CIFAR10、MNIST、CMU AN4、AlexNet、VGG、ResNet
PowerSGD^[72]	√	功率迭代、快速	×	×	压缩了24～128倍	CIFAR10
SignSGD^[73]	√	top-k梯度稀疏化，实现下游压缩以及权重更新的分层和最佳Golomb编码	√	√	压缩了1 050倍	CIFARA、KWSA、F-MNIST
稀疏三进制压缩^[75]	√	多数投票的方式	√	×	速度提高25%	ImageNet、RESNET50

算法—框架名称	Baseline	特点	通信情况	隐私保护	适用场景
FedAvg^[48]	FeedSGD	迭代平均	与同步随机梯度下降相比，通信效率提高10~100倍	DP、MPC	非平衡数据non-IID数据
DSVGR^[57]	SVRG DANE	稀疏数据(结构)	—	—	不平衡数据可用节点少
Parallel Restarted SGD^[65]	Parallel mini-batch SGD parallel SGD	并行、重启	与parallel mini-batch SGD相比减少了O(T^1/4)的通信交流次数	—	非凸优化
PASGD^[69]	Fully synchronous SGD	周期平均	VGG-16中，ADACOMIS比全同步SGD快3.5倍；ResNet-50中，ADACOMM比完全同步SGD快2倍	—	—
AFL^[49]	Active Learning	主动、选择性抽样迭代	减少20%～70%的迭代	有	—
AFL ^[61]	FedAvg	分布、公平	—	—	集中式模型
Fmore^[54]	—	激励机制，迁移学习	减少近51.3%的训练轮数，真实系统，时间减少了38.4%	—	移动边缘计算
FedBCD^[66]	—	基于块坐标梯度下降	可以显著降低通信成本	有	共享样本
MMD^[67]	单流	双流模型代替单一模型	非IID数据分布所需的通信轮数减少了20%以上	—	non-IID数据
Mini-batch SGD^[68]	SGDZ	小型批量	通信回合的数量最多可以减少T/2 （其中T表示总步数）	—	高网络时延带宽受限

面向隐私保护的非聚合式数据共享综述

Survey on privacy protection in non-aggregated data sharing

在线阅读

PDF下载

可视化

摘要/Abstract

引用本文

使用本文

图/表 17

参考文献 86

相关文章 15

Metrics

推荐阅读 0

框架	支持类型	安全协议	主要支持场景	软件	Kubernetes	开发机构
JUGO	安全两方计算	同态加密	商业应用	不开源	不支持	矩阵元
		混淆电路
Tasty	安全多方计算	PSI	学术研究	开源	不支持	—
		同态加密
		混淆电路
PaddleFL	横向联邦学习	PSI	学术研究	开源	仅Data Parallel支持	百度
	纵向联邦学习	秘密分享
		同态加密
FATE	横向联邦学习	PSI	商用数据共享	开源	支持	微众银行
	纵向联邦学习	秘密分享
	联邦迁移学习	同态加密
TTF	横向联邦学习	DP	学术研究	开源	不支持	谷歌
		GAN
Clara	横向联邦学习	DP	商业应用	不开源	支持	英伟达

[1]	马鑫迪, 李清华, 姜奇, 马卓, 高胜, 田有亮, 马建峰. 面向Non-IID数据的拜占庭鲁棒联邦学习[J]. 通信学报, 2023, 44(6): 138-153.
[2]	金彪, 李逸康, 姚志强, 陈瑜霖, 熊金波. GenFedRL：面向深度强化学习智能体的通用联邦强化学习框架[J]. 通信学报, 2023, 44(6): 183-197.
[3]	田有亮, 吴柿红, 李沓, 王林冬, 周骅. 基于激励机制的联邦学习优化算法[J]. 通信学报, 2023, 44(5): 169-180.
[4]	张佳乐, 朱诚诚, 孙小兵, 陈兵. 基于GAN的联邦学习成员推理攻击与防御方法[J]. 通信学报, 2023, 44(5): 193-205.
[5]	冯涛, 陈李秋, 方君丽, 石建明. 基于本地化差分隐私和属性基可搜索加密的区块链数据共享方案[J]. 通信学报, 2023, 44(5): 224-233.
[6]	李开菊, 许强, 王豪. 冗余数据去除的联邦学习高效通信方法[J]. 通信学报, 2023, 44(5): 79-93.
[7]	余晟兴, 陈泽凯, 陈钟, 刘西蒙. DAGUARD：联邦学习下的分布式后门攻击防御方案[J]. 通信学报, 2023, 44(5): 110-122.
[8]	姜慧, 何天流, 刘敏, 孙胜, 王煜炜. 面向异构流式数据的高性能联邦持续学习算法[J]. 通信学报, 2023, 44(5): 123-136.
[9]	夏莹杰, 朱思雨, 刘雪娇. 区块链架构下具有条件隐私的车辆编队跨信任域高效群组认证研究[J]. 通信学报, 2023, 44(4): 111-123.
[10]	胡柏吉, 张晓娟, 李元诚, 赖荣鑫. 支持多功能的V2G网络隐私保护数据聚合方案[J]. 通信学报, 2023, 44(4): 187-200.
[11]	徐明, 张保俊, 伍益明, 应晨铎, 郑宁. 面向网络攻击和隐私保护的多智能体系统分布式共识算法[J]. 通信学报, 2023, 44(3): 117-127.
[12]	刘雪娇, 曹天聪, 夏莹杰. 区块链架构下高效的车联网跨域数据安全共享研究[J]. 通信学报, 2023, 44(3): 186-197.
[13]	余晟兴, 陈钟. 基于同态加密的高效安全联邦学习聚合框架[J]. 通信学报, 2023, 44(1): 14-28.
[14]	汤凌韬, 王迪, 刘盛云. 面向非独立同分布数据的联邦学习数据增强方案[J]. 通信学报, 2023, 44(1): 164-176.
[15]	范绍帅, 吴剑波, 田辉. 面向能量受限工业物联网设备的联邦学习资源管理[J]. 通信学报, 2022, 43(8): 65-77.