Practical Secure Aggregation for Privacy-Preserving Machine Learning
The DOI matches its Crossref record; 56 of the 56 references it cites were found in a registry or an index.
Abstract
We design a novel, communication-efficient, failure-robust protocol for secure aggregation of high-dimensional data. Our protocol allows a server to compute the sum of large, user-held data vectors from mobile devices in a secure manner (i.e. without learning each user's individual contribution), and can be used, for example, in a federated learning setting, to aggregate user-provided model updates for a deep neural network. We prove the security of our protocol in the honest-but-curious and active adversary settings, and show that security is maintained even if an arbitrarily chosen subset of users drop out at any time. We evaluate the efficiency of our protocol and show, by complexity analysis and a concrete implementation, that its runtime and communication overhead remain low even on large data sets and client pools. For 16-bit input values, our protocol offers $1.73 x communication expansion for 210 users and 220-dimensional vectors, and 1.98 x expansion for 214 users and 224-dimensional vectors over sending data in the clear.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“Although SMC guarantees that none of the parties shares anything with each other or with any third party, it can not prevent an adversary from learning some individual information”
From Federated Learning for Healthcare Informatics (2021) · cites this paper as [8]
In the citing paper
“By combining local inference with trusted aggregation, TDAI can enhance both privacy and robustness of anomaly detection pipelines [239].”
From Anomaly detection and root-cause identification in microservices: a survey (2026) · cites this paper as [239]
In the citing paper
“FL is still in its infancy and is still vulnerable to various challenges [148,149].”
From Smart Wearables for the Detection of Cardiovascular Diseases: A Systematic Literature Review (2023) · cites this paper as [148]
In the citing paper
“Therefore, many FL frameworks use secure aggregation [81], [82], [83], [84] algorithms in the aggregation process, at which point most of the poisoning attack defense methods mentioned above will be invalid.”
From Poisoning attacks in federated learning: A survey (2023) · cites this paper as [81]
In the citing paper
“Bonawitz et al. [62] introduce a secure aggregation protocol for protecting individual model updates.”
From Handling privacy-sensitive medical data with federated learning: Challenges and future directions (2023) · cites this paper as [62]
6 more claims about this paper
Create a free account to see every claim other papers make about this one.
Citations
Cites
New directions in cryptography
Diffie, W., & Hellman, M. (1976) · IEEE Transactions on Information Theory · doi:10.1109/tit.1976.1055638 · Cited by 14,713
Communication-Efficient Learning of Deep Networks from Decentralized Data
McMahan, H. B., Moore, E., Ramage, D., Hampson, S., & Arcas, B. A. Y. (2016) · arXiv · doi:10.48550/arxiv.1602.05629 · Cited by 14,215 · cited 1×
“maintains a private database of her text messages securely on her own mobile device, and a shared global model is trained under the coordination of a central server based upon highly processed, minimally scoped, ephemeral updates from users [20], [10].”
BackgroundSelf-citationShamir, A. (1979) · Communications of the ACM · doi:10.1145/359168.359176 · Cited by 13,756 · cited 1×
“We rely on Shamir’s t -out-of- n Secret Sharing [38], which allows a user to split a secret s into n shares, such that any t shares can be used to reconstruct s , but any set of at most t − 1 shares gives no information about s .”
Method(2016) · Cited by 11,679
Deep Learning with Differential Privacy
Abadi, M., Chu, A., Goodfellow, I., McMahan, H. B., Mironov, I., Talwar, K., & Zhang, L. (2016) · Proceedings of the 2016 ACM SIGSAC Conference on Computer and Communications Security · doi:10.1145/2976749.2978318 · Cited by 6,578 · cited 3×
“This is particular advantageous in the local privacy setting [13], which offers provable guarantees for the protection of individual training examples [14], [15] even when the data aggregator is not assumed to be trusted [16], [17].”
BackgroundMethodSelf-citationThe Algorithmic Foundations of Differential Privacy
Dwork, C., & Roth, A. (2014) · Foundations and Trends® in Theoretical Computer Science · doi:10.1561/0400000042 · Cited by 5,557
Dwork, C. (2006) · Lecture Notes in Computer Science · doi:10.1007/11787006_1 · Cited by 5,291 · cited 1×
“In such cases, secure aggregation composes well with differential privacy [12].”
BackgroundMembership Inference Attacks Against Machine Learning Models
Shokri, R., Stronati, M., Song, C., & Shmatikov, V. (2017) · 2017 IEEE Symposium on Security and Privacy (SP) · doi:10.1109/sp.2017.41 · Cited by 4,754
The Algorithmic Foundations of Differential Privacy
Dwork, C., & Roth, A. (2013) · doi:10.1561/9781601988195 · Cited by 3,856
Federated Learning: Strategies for Improving Communication Efficiency
Konečný, J., McMahan, H. B., Yu, F. X., Richtárik, P., Suresh, A. T., & Bacon, D. (2016) · arXiv · doi:10.48550/arxiv.1610.05492 · Cited by 3,766
Self-citation
Research works printed but not linked: 3
- Stephen Advokat. 1987. Publication Of Bork's Video Rentals Raises Privacy Issue. Chicago Tribune (1987).
- Michael Barbaro, Tom Zeller, and Saul Hansell. 2006. A face is exposed for AOL searcher no. 4417749. New York Times 9, 2008 (2006).
- Elaine Shi, HTH Chan, Eleanor Rieffel, Richard Chow, and Dawn Song. 2011. Privacy-preserving aggregation of time-series data. In Annual Network & Distributed System Security Symposium (NDSS). Internet Society.
Cited by
Items 801–900 of 1,035
Verifiable and privacy preserving federated learning without fully trusted centers
Han, G., Zhang, T., Zhang, Y., Xu, G., Sun, J., & Cao, J. P. (2021) · Journal of Ambient Intelligence and Humanized Computing · doi:10.1007/s12652-020-02664-x · Cited by 51
“Bonawitz et al. (2017) designed a high-dimensional data security aggregation protocol for federated learning.”
BackgroundFLRA: A Reference Architecture for Federated Learning Systems
Lo, S. K., Lu, Q., Paik, H.-Y., & Zhu, L. (2021) · Lecture notes in computer science · doi:10.1007/978-3-030-86044-8_6 · Cited by 53
Your Labels are Selling You Out: Relation Leaks in Vertical Federated Learning
Qiu, P., Zhang, X., Ji, S., Du, T., Pu, Y., Zhou, J. W., & Wang, T. (2022) · IEEE Transactions on Dependable and Secure Computing · doi:10.1109/tdsc.2022.3208630 · Cited by 51
Variance Reduced ProxSkip: Algorithm, Theory and Application to Federated Learning
Malinovsky, G., Yi, K., & Richtárik, P. (2022) · neural information processing systems · doi:10.52202/068431-1104 · Cited by 51
FedDual: Pair-Wise Gossip Helps Federated Learning in Large Decentralized Networks
Chen, Q., Wang, Z., Wang, H., & Lin, X. (2022) · IEEE Transactions on Information Forensics and Security · doi:10.1109/tifs.2022.3222935 · Cited by 50
“In the second degree, we consider a widely adopted honest-but-curious (or semi-honest) participants and active adversaries security model [10], [11], [22], [45].”
BackgroundThe FeatureCloud Platform for Federated Learning in Biomedicine: Unified Approach
Matschinske, J., Späth, J., Bakhtiari, M., Probul, N., Majdabadi, M. M. K., Nasirigerdeh, R., Torkzadehmahani, R., Hartebrodt, A., Orbán, B.-A., Fejér, S.-J., et al. (2023) · Journal of Medical Internet Research · doi:10.2196/42621 · Cited by 50
“Therefore, to improve data privacy, FL can be combined with privacy-enhancing technologies (PETs) [10], such as secure aggregation [11] or differential privacy (DP) [12,13].”
BackgroundMultifactor Incentive Mechanism for Federated Learning in IoT: A Stackelberg Game Approach
Chen, Y., Hui, Z., Li, T., Li, J., & Zhou, H. (2023) · IEEE Internet of Things Journal · doi:10.1109/jiot.2023.3306778 · Cited by 50
“Homomorphic encryption [10], [11] and secure multiparty computation [12] are often used to secure IoT data privacy, but complex cryptographic operations lead to poor efficiency and significant expense.”
BackgroundRPIFL: Reliable and Privacy-Preserving Federated Learning for the Internet of Things
Wang, R., Lai, J., Li, X., He, D., & Khan, M. K. (2023) · Journal of Network and Computer Applications · doi:10.1016/j.jnca.2023.103768 · Cited by 50
Prototype-guided Knowledge Transfer for Federated Unsupervised Cross-modal Hashing
Li, J., Li, F., Zhu, L., Cui, H. J., & Li, J. (2023) · ACM International Conference on Multimedia (ACM MM) · doi:10.1145/3581783.3613837 · Cited by 50
“In addition, in practical applications, federated learning-based methods usually add additional privacy protection techniques to further improve system reliability [2], such as Homo-geneous Encryption (HE) [53].”
BackgroundMethodPriVeriFL: Privacy-Preserving and Aggregation-Verifiable Federated Learning
Wang, L., Polato, M., Brighente, A., Conti, M., Zhang, L., & Xu, L. (2024) · IEEE Transactions on Services Computing · doi:10.1109/tsc.2024.3451183 · Cited by 50
“However, the above two schemes, like [18], require four rounds of interactions to aggregate model updates from one iteration of training, and participants need to synchronize secret keys and masks.”
InfluentialThe Next Breakthroughs of Artificial Intelligence: The Interdisciplinary Nature of AI
Zhuang, Y., Cai, M., Li, X., Luo, X., Yang, Q., & Wu, F. Q. (2020) · Engineering · doi:10.1016/j.eng.2020.01.009 · Cited by 50
“The model can be transferred (in part) from one party to another under a security scheme [15,16], so that any party cannot reverse-engineer the data of any other parties.”
BackgroundCitadel: Protecting Data Privacy and Model Confidentiality for Collaborative Learning
Zhang, C., Xia, J., Yang, B., Puyang, H., Wang, W., Chen, R., Akkuş, İ. E., Aditya, P., & Yan, F. (2021) · ACM Symposium on Cloud Computing (SoCC) · doi:10.1145/3472883.3486998 · Cited by 52
“Our centralized zero-sum masking approach protects the model confidentiality, while guaranteeing the same level of privacy for data owners as in secure aggregation [13], but without the time-consuming synchronous distributed mask generation.”
BackgroundInfluentialMoshpit SGD: Communication-Efficient Decentralized Training on Heterogeneous Unreliable DevicesPreprint
Ryabinin, M., Gorbunov, E., Plokhotnyuk, V., & Pekhimenko, G. G. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2103.03239 · Cited by 51
“This issue is most prevalent in federated learning [11, 29, 30].”
BackgroundZero-Knowledge Proof Frameworks: A Systematic SurveyPreprint
Sheybani, N., Ahmed, A., Kinsy, M. A., & Koushanfar, F. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2502.07063 · Cited by 50
“The introduction of secure aggregation [47], which was devised to secure individual user updates, has made it much easier for malicious users to perform Byzantine attacks.”
Federated Learning: A Survey on Privacy-Preserving Collaborative IntelligencePreprint
Rahman, R. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2504.17703 · Cited by 50
“To mitigate these risks, techniques such as differential privacy [15], secure multiparty computation [14], and robust aggregation mechanisms [33] have been integrated into FL protocols.”
InfluentialMitigating the Backdoor Attack by Federated Filters for Industrial IoT Applications
Hou, B., Gao, J., Guo, X., Baker, T., Zhang, Y., Wen, Y., & Liu, Z. (2022) · IEEE Transactions on Industrial Informatics · doi:10.1109/tii.2021.3112100 · Cited by 49
ADFL: A Poisoning Attack Defense Framework for Horizontal Federated Learning
Guo, J., Li, H., Huang, F., Liu, Z., Peng, Y., Li, X., Ma, J., Menon, V. G., & Igorevich, K. K. (2022) · IEEE Transactions on Industrial Informatics · doi:10.1109/tii.2022.3156645 · Cited by 49
RevFRF: Enabling Cross-Domain Random Forest Training With Revocable Federated Learning
Liu, Y., Ma, Z., Yang, Y., Liu, X., Ma, J., & Ren, K. (2022) · IEEE Transactions on Dependable and Secure Computing · doi:10.1109/tdsc.2021.3104842 · Cited by 49
“Moreover, since the data reconstruction attack towards the sub-updates is discovered [4], [5], a consistent exploration to enhance the privacy-preserving strategy of FL has never been stopped [6], [7].”
BackgroundEfficient Federated Learning for AIoT Applications Using Knowledge Distillation
Liu, T., Xia, J., Ling, Z., Fu, X., Yu, S., & Chen, M. (2022) · IEEE Internet of Things Journal · doi:10.1109/jiot.2022.3229374 · Cited by 49
“, secure aggregation [22] and variational Bayes-based method [23]) for FL can be easily integrated into our approach to avoid the risk of privacy exposure.”
MethodDistributed, Private, Sparse Histograms in the Two-Server Model
Bell, J., Gascón, A., Ghazi, B., Kumar, R., Manurangsi, P., Raykova, M., & Schoppmann, P. (2022) · ACM SIGSAC Conference on Computer and Communications Security (CCS) · doi:10.1145/3548606.3559383 · Cited by 49
“Applications can be found in many different contexts including product analysis and browser telemetry [10, 15, 26, 27], understanding the spread of viruses [4, 58], and detecting distributed attacks and fraud behavior [13, 63].”
BackgroundMalik, H., Anees, T., Naeem, A., Naqvi, R. A., & Loh, W.-K. (2023) · Bioengineering · doi:10.3390/bioengineering10020203 · Cited by 52
“The study [20] developed a method that safeguards users’ privacy while allowing for the secure aggregation of gradients through the utilization of the federated learning global model.”
MethodA Systematic Survey for Differential Privacy Techniques in Federated Learning
Zhang, Y., Lu, Y., & Liu, F. (2023) · Journal of Information Security · doi:10.4236/jis.2023.142008 · Cited by 50
Verifiable Federated Learning With Privacy-Preserving Data Aggregation for Consumer Electronics
Xie, H., Wang, Y., Ding, Y., Yang, C., Zheng, H., & Qin, B. (2023) · IEEE Transactions on Consumer Electronics · doi:10.1109/tce.2023.3323206 · Cited by 49
“…result by colluding to upload incorrect gradient values, the scheme can still guarantee that no party can learn more information than the aggregated result of the gradient from a large subset of honest participants, or deviate widely from the direction of correct aggregation results [41].”
BackgroundMao, Q., Wang, L., Yu, L., Han, L., Wang, Z., & Chen, K. (2023) · Knowledge-Based Systems · doi:10.1016/j.knosys.2023.111338 · Cited by 50
Privacy-Preserving and Secure Industrial Big Data Analytics: A Survey and the Research Framework
Liu, L., Li, J., Lv, J., Wang, J., Zhao, S., & Lu, Q. (2024) · IEEE Internet of Things Journal · doi:10.1109/jiot.2024.3353727 · Cited by 49
“For example, Bonawitz et al. [78] proposed a secure aggregation model that combines techniques like secret sharing to prevent the server from decrypting individual client gradients, thus hiding information from malicious servers.”
MethodZheng, J., Chen, Y., & Lai, Q. (2024) · Future Generation Computer Systems · doi:10.1016/j.future.2024.03.020 · Cited by 49
Privacy-preserving federated learning based on partial low-quality data
Wang, H., Wang, Q., Ding, Y., Tang, S., & Wang, Y. (2024) · Journal of Cloud Computing Advances Systems and Applications · doi:10.1186/s13677-024-00618-8 · Cited by 49
“To provide a clear visualization of the experimental results, we compared them with the PPML [32] approach.”
Kang, Y., Gu, H., Tang, X., He, Y., Zhang, Y., He, J., Han, Y., Fan, L., Chen, K., & Yang, Q. (2024) · ACM Transactions on Intelligent Systems and Technology · doi:10.1145/3701039 · Cited by 49
“We measure the training cost 𝜖 𝑐 of BC as follows: In this setting, a solution contains the following variables to optimize: learning rate, the number of neurons in each layer for the two hidden-layer MLP and LetNet, and BC quantization bit width chosen in the range of [ 2 , 4 , 8 , 16 ] .”
Challenges of Privacy-Preserving Machine Learning in IoT
Zheng, M., Xu, D., Jiang, L., Gu, C., Tan, R., & Cheng, P. (2019) · Proceedings of the First International Workshop on Challenges in Artificial Intelligence and Machine Learning for Internet of Things · doi:10.1145/3363347.3363357 · Cited by 36
“Besides, secure data aggregation is applied to aggregate the updates from individual participants [6].”
MethodFedBC: Blockchain-based Decentralized Federated Learning
Wu, X. G., Wang, Z., Zhao, J., Zhang, Y., & Wu, Y. (2020) · IEEE International Conference on Artificial Intelligence and Computer Applications (ICAICA) · doi:10.1109/icaica50127.2020.9182705 · Cited by 49
“[10][11] established a secret sharing and symmetric encryption mechanism to prevent curious participants from causing privacy leakage.”
BackgroundFederated Learning with Gaussian Differential Privacy
Chuanxin, Z., Yi, S., & Degang, W. (2020) · Proceedings of the 2020 2nd International Conference on Robotics, Intelligent Control and Artificial Intelligence · doi:10.1145/3438872.3439097 · Cited by 33
“Although secure multi-party computing and homomorphic encryption can provide good user-level privacy protection, the excessive computational overhead makes it impossible to perform large-scale, low-power federated learning tasks [16][18].”
BackgroundExtraction of Complex DNN Models: Real Threat or Boogeyman?
Atli, B. G., Szyller, S., Juuti, M., Marchal, S., & Asokan, N. (2020) · Communications in computer and information science · doi:10.1007/978-3-030-62144-5_4 · Cited by 48
An Efficient Approach for Cross-Silo Federated Learning to Rank
Wang, Y., Tong, Y., Shi, D., & Xu, K. (2021) · IEEE International Conference on Data Engineering (ICDE) · doi:10.1109/icde51399.2021.00102 · Cited by 49
“In both settings, privacy protection often becomes the core issue and encryption-based methods like secret sharing [18] and noisybased methods like differential privacy [19]–[21] have been applied.”
MethodFederated recommenders: methods, challenges and future
Alamgir, Z., Khan, F. K., & Karim, S. (2022) · Cluster Computing · doi:10.1007/s10586-022-03644-w · Cited by 50
“A reasonable amount of work has been done to improve security in FR [10, 11], and make it more personalized [12, 13].”
BackgroundA Federated Multi-View Deep Learning Framework for Privacy-Preserving RecommendationsPreprint
Huang, M., Li, H., Bai, B., Wang, C., Bai, K., & Fei, W. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2008.10808 · Cited by 49
“…can be 2, and the aggregated gradient will be sent to each view for next round of training, solutions based on Secure Multi-Party Computation (MPC) [5] can not be used to implement local_secure_aggregate() since the aggregation result is accurate and it will expose another view’s gradients by…”
BackgroundMethodInfluentialStevens, T., Skalka, C., Vincent, C., Ring, J. H., Clark, S., & Near, J. P. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2112.06872 · Cited by 49
“Combining secure aggregation [8, 12] with differential privacy [21,29] ensures end-to-end privacy in federated learning systems.”
InfluentialOptimal Algorithms for Mean Estimation under Local Differential PrivacyPreprint
Asi, H., Feldman, V., & Talwar, K. (2022) · arXiv (Cornell University) · doi:10.48550/arxiv.2205.02466 · Cited by 49
Randomized Quantization is All You Need for Differential Privacy in Federated LearningPreprint
Youn, Y., Hu, Z., Ziani, J., & Abernethy, J. (2023) · arXiv (Cornell University) · doi:10.48550/arxiv.2306.11913 · Cited by 50
“One is secure multi-party computation, which protects the communication between local devices and the learner, preventing an attacker from intercepting messages sent between them (Bell et al., 2020; Bonawitz et al., 2017).”
BackgroundMethodRomoa: Robust Model Aggregation for the Resistance of Federated Learning to Model Poisoning Attacks
Mao, Y., Yuan, X., Zhao, X., & Zhong, S. (2021) · Lecture Notes in Computer Science · doi:10.1007/978-3-030-88418-5_23 · Cited by 48
Securing Distributed SGD Against Gradient Leakage Threats
Wei, W., Liu, L., Zhou, J., Chow, K.-H., & Wu, Y. (2023) · IEEE Transactions on Parallel and Distributed Systems · doi:10.1109/tpds.2023.3273490 · Cited by 48
“between a client and the federated server [41].”
BackgroundShen, G., Fu, Z., Gui, Y., Susilo, W., & Zhang, M. (2023) · Information Sciences · doi:10.1016/j.ins.2023.119261 · Cited by 49
Towards Efficient Communication and Secure Federated Recommendation System via Low-rank Training
Nguyen, N.-H., Nguyen, T.-A., Nguyen, T. A., Hoang, V. T., Le, D. D., & Wong, K. (2024) · International World Wide Web Conference (WWW) · doi:10.1145/3589334.3645702 · Cited by 48
Learning from Differentially Private Neural Activations with Edge Computing
Mao, Y., Yi, S., Li, Q., Feng, J., Xu, F., & Zhong, S. (2018) · 2018 IEEE/ACM Symposium on Edge Computing (SEC) · doi:10.1109/sec.2018.00014 · Cited by 40
“But user’s private data will be violated seriously if the central server is untrusted [1].”
BackgroundMethodInfluentialRiazi, M. S., Rouhani, B. D., & Koushanfar, F. (2019) · IEEE Security & Privacy · doi:10.1109/msec.2019.2935666 · Cited by 48
Federated Learning of Unsegmented Chinese Text Recognition Model
Zhu, X., Wang, J., Hong, Z., Xia, T., & Xiao, J. (2019) · 2019 IEEE 31st International Conference on Tools with Artificial Intelligence (ICTAI) · doi:10.1109/ictai.2019.00186 · Cited by 39
“In such a system, participating clients are typically assumed to be honest, while the server is considered honest-but-curious [19].”
BackgroundFast-convergent federated learning with class-weighted aggregation
Ma, Z., Zhao, M., Cai, X., & Jia, Z. (2021) · Journal of Systems Architecture · doi:10.1016/j.sysarc.2021.102125 · Cited by 49
NN-EMD: Efficiently Training Neural Networks Using Encrypted Multi-Sourced Datasets
Xu, R., Joshi, J. B. D., & Li, C. (2021) · IEEE Transactions on Dependable and Secure Computing · doi:10.1109/tdsc.2021.3074439 · Cited by 48
From Server-Based to Client-Based Machine Learning: A Comprehensive Survey
Gu, R., Niu, C., Wu, F., Chen, G., Hu, C., Lyu, C., & Wu, Z. (2021) · ACM Computing Surveys · doi:10.1145/3424660 · Cited by 63
“After the development of FedAvg [70], to better ensure the security of the aggregation process in federated learning, Google proposed a practical secure aggregation method [13] for privacy-preserving machine learning.”
BackgroundMethodDecentralized Learning of Generative Adversarial Networks from Non-iid DataPreprint
Yonetani, R., Takahashi, T., Hashimoto, A., & Ushiku, Y. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1905.09684 · Cited by 55
“More recent work has tried to make federated learning more communication efficient [36, 37, 38], secure [19, 39], and applicable to a practical wireless setting [40, 41], but under the setting of standard supervised learning.”
BackgroundMethodPractical Privacy Attacks on Vertical Federated LearningPreprint
Weng, H., Zhang, J., Ma, X., Xue, F., Wei, T., Ji, S., & Zong, Z. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2011.09290 · Cited by 67
“Based on the characteristics of data distribution, the general federated learning can be categorized into horizontal federated learning [2], [3] and vertical federated learning [47], [19].”
BackgroundLocal Differential Privacy for Federated Learning
Chamikara, M. A. P., Liu, D., Camtepe, S. A., Nepal, ., Grobler, M. M., Bertök, P., & Khalil, I. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-17140-6_10 · Cited by 47
Yang, Y., Yang, R., Peng, H. M., Li, Y., Li, T., Liao, Y., & Zhou, P. (2023) · International World Wide Web Conference (WWW) · doi:10.1145/3543507.3583500 · Cited by 47
“TextCNN [21] used in � has four convolution kernels of size [2, 3, 4, 5].”
BackgroundMethodAgrEvader: Poisoning Membership Inference against Byzantine-robust Federated Learning
Zhang, Y., Bai, G., Chamikara, M. A. P., Ma, M., Shen, L., Wang, J., Nepal, ., Xue, M., Wang, L., & Liu, J. K. (2023) · International World Wide Web Conference (WWW) · doi:10.1145/3543507.3583542 · Cited by 49
“Secure aggregation employs cryptographic mechanisms such as homomorphic encryption (HE) [13, 35, 48] and secure multiparty computation (MPC) [10, 18, 33, 39, 62] to securely update the global model.”
BackgroundLi, Q., Liu, Z., Li, Q. L., & Xu, K. (2023) · ACM SIGSAC Conference on Computer and Communications Security (CCS) · doi:10.1145/3576915.3623134 · Cited by 48
“As a result, the aggregation algorithm in FPPDL [61] is unable to handle malicious DPs; and Omnilytics [54] only supports four DPs using the simple Secure-Aggregation algorithm [13] with the multi-Krum [12] algorithm to remove outliers, while incurring significant gas cost (at least 1000 times more thanmartFL, as show…”
MethodFotohi, R., Aliee, F. S., & Farahani, B. (2024) · Applied Soft Computing · doi:10.1016/j.asoc.2024.111764 · Cited by 47
“An approach based on HE to guarantee the confidentiality of gradients is proposed by Bonawitz et al. (Bonawitz, Ivanov, Kreuter, Marcedone, McMahan, Patel, Ramage, Segal, and Seth, 2017).”
Zheng, Y., Chang, C.-H., Huang, S.-H., Chen, P., & Picek, S. (2024) · IEEE Journal on Emerging and Selected Topics in Circuits and Systems · doi:10.1109/jetcas.2024.3477348 · Cited by 49
“To enhance the robustness against client dropout problem [192], Shamir t -out-of- n secret sharing is adopted in [176] by incorporating a dual-mask privacy-preserving aggregation protocol based on ( t , n ) threshold secret sharing and key agreement, where n is the number of clients.”
Recommender Systems Meet Large Language Model Agents: A Survey
Zhu, X., Wang, Y., Gao, H., Xu, W., Wang, C., Liu, Z., Wang, K., Jin, M., Pang, L. X., Weng, Q., et al. (2025) · Foundations and Trends® in Privacy and Security · doi:10.1561/3300000050 · Cited by 47
PrivColl: Practical Privacy-Preserving Collaborative Machine Learning
Zhang, Y., Bai, G., Li, X., Curtis, C., Chen, C., & Ko, R. K. L. (2020) · Lecture notes in computer science · doi:10.1007/978-3-030-58951-6_20 · Cited by 46
Mutual Information Driven Federated Learning
Uddin, M. P., Xiang, Y., Lu, X., Yearwood, J. L., & Gao, L. (2020) · IEEE Transactions on Parallel and Distributed Systems · doi:10.1109/tpds.2020.3040981 · Cited by 47
RPPTD: Robust Privacy-Preserving Truth Discovery Scheme
Chen, J., Liu, Y., Xiang, Y., & Sood, K. (2021) · IEEE Systems Journal · doi:10.1109/jsyst.2021.3099103 · Cited by 47
“Then fault-tolerance is achieved in [7], [11], [15] under the assumption that there is no collusion between servers, and nobody guarantees that the two servers never collude in the real environment [16], [17], whether or not the two servers are from the same operator.”
BackgroundLightweight Federated Learning for Large-Scale IoT Devices With Privacy Guarantee
Wei, Z., Pei, Q., Zhang, N., Liu, X., Wu, C., & Taherkordi, A. (2021) · IEEE Internet of Things Journal · doi:10.1109/jiot.2021.3127886 · Cited by 47
“Theapproaches based on secret sharing [24] are relatively computationally efficient, and can be made robust against participants dropouts.”
BackgroundMethodInfluentialSecure Aggregation in Federated Learning via Multiparty Homomorphic Encryption
Hosseini, E., & Khisti, A. J. (2021) · IEEE Globecom Workshops (GC Wkshps) · doi:10.1109/gcwkshps52748.2021.9682053 · Cited by 47
Federated Neural Architecture SearchPreprint
Yuan, J., Xu, M., Zhao, Y., Bian, K., Huang, G., Liu, X., & Wang, S. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2002.06352 · Cited by 51
“, improving the privacy guarantees by differential privacy [30] and secure aggregation [6], reducing the communication cost among cloud and clients through weights compression [21, 29], are complementary and orthogonal to DecNAS.”
ResultSparsified Secure Aggregation for Privacy-Preserving Federated LearningPreprint
Ergun, I., Sami, H. U., & Güler, B. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2112.12872 · Cited by 47
“In addition to the pairwise seeds, user i also creates a private random seed s ( t ) i , which protects the privacy of the user if a user delayed instead of being dropped, which was shown in [11].”
InfluentialEnhancing anomaly detection in distributed power systems using autoencoder-based federated learning
Kea, K., Han, Y., & Kim, T.-K. (2023) · PLoS ONE · doi:10.1371/journal.pone.0290337 · Cited by 46
“This iterative approach enables the global model to be updated while preserving the privacy and security of the clients’ data [13].”
BackgroundAli, H., Chen, D., Harrington, M., Salazar, N., Ameedi, M. A., Khan, A. F., Butt, A. R., & Cho, J.-H. (2023) · IEEE Access · doi:10.1109/access.2023.3326410 · Cited by 48
“SMC is highly useful in adversarial environments as FL protocols have successfully adopted SMC to prevent observation of client data from aggregating servers or necessary edge computing environments [16], [104], [105].”
MethodMaintaining Privacy in Face Recognition Using Federated Learning Method
Woubie, A., Solomon, E., & Attieh, J. (2024) · IEEE Access · doi:10.1109/access.2024.3373691 · Cited by 50
“The adoption of secure multiparty computation serves to safe-guard individual model updates, ensuring their privacy and confidentiality [44].”
BackgroundA Federated Learning Approach to Routing in Challenged SDN-Enabled Edge Networks
Sacco, A., Esposito, F., & Marchetto, G. (2020) · IEEE International Conference on Network Softwarization (NetSoft) · doi:10.1109/netsoft48620.2020.9165506 · Cited by 46
“However, recent improvements have been focusing on overcoming the analytical challenges [22], [23], or to further improve the security aspects [24], [25].”
BackgroundVerifiable, Reliable, and Privacy-Preserving Data Aggregation in Fog-Assisted Mobile Crowdsensing
Yan, X., Ng, W. W. Y., Zeng, B., Lin, C., Liu, Y., Lu, L., & Gao, Y. (2021) · IEEE Internet of Things Journal · doi:10.1109/jiot.2021.3068490 · Cited by 46
PIVODL: Privacy-Preserving Vertical Federated Learning Over Distributed Labels
Zhu, H., Wang, R., Jin, Y., & Liang, K. (2021) · IEEE Transactions on Artificial Intelligence · doi:10.1109/tai.2021.3139055 · Cited by 46
“Secure multi-party computation [12], [13], homomorphic encryption (HE) [14] and differential privacy (DP) [15] are three most common privacy preserving mechanisms that are theoretically and empirically proved to be effective for HFL.”
BackgroundMethodEfficient Privacy Preserving Edge Intelligent Computing Framework for Image Classification in IoT
Fagbohungbe, O. I., Reza, S. R., Dong, X., & Qian, L. (2021) · IEEE Transactions on Emerging Topics in Computational Intelligence · doi:10.1109/tetci.2021.3111636 · Cited by 46
Towards Efficient Synchronous Federated Training: A Survey on System Optimization Strategies
Jiang, Z., Wang, W., Li, B., & Yang, Q. (2022) · IEEE Transactions on Big Data · doi:10.1109/tbdata.2022.3177222 · Cited by 46
“One of the most impactful works is Google’s secure aggregation (SecAgg) protocol, where pseudorandom masks are used for data confidentiality and Shamir’s secret sharing scheme [125] is used to accommodate client dropout [78].”
MethodInfluentialFedCKE: Cross-Domain Knowledge Graph Embedding in Federated Learning
Huang, W. X., Liu, J., Li, T., Ji, S., Wang, D., & Huang, T. (2022) · IEEE Transactions on Big Data · doi:10.1109/tbdata.2022.3205705 · Cited by 46
“FedCKE can also be combined with the encryption mechanisms [47] used in the federated learning foundational framework to prevent the leakage of private information from the client to the server.”
BackgroundSenate: A Maliciously-Secure MPC Platform for Collaborative AnalyticsPreprint
Poddar, R., Kalra, S., Yanai, A., Deng, R., Popa, R. A., & Hellerstein, J. M. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2010.13752 · Cited by 46
FedVS: Straggler-Resilient and Privacy-Preserving Vertical Federated Learning for Split ModelsPreprint
Li, S., Yao, D., & Liu, J. (2023) · arXiv (Cornell University) · doi:10.48550/arxiv.2304.13407 · Cited by 46
“Recently in (Shi et al., 2022), it is proposed to use secure aggregation (Bonawitz et al., 2017) for privacy protection in asynchronous training of linear and logistic regression models over vertically partitioned data, which is nevertheless faced with slow convergence from asynchronous model…”
BackgroundBeyond class-level privacy leakage: breaking record-level privacy in federated learning
Yuan, X., Ma, X., Zhang, L., Fang, Y., & Wu, D. (2022) · IEEE Internet of Things Journal · doi:10.1109/jiot.2021.3089713 · Cited by 45
FedNLP: A Research Platform for Federated Learning in Natural Language Processing.
Lin, B., He, C., Zeng, Z., Wang, H., Huang, Y., Soltanolkotabi, M., Ren, X., & Avestimehr, S. (2021) · Cited by 45
Liu, G., Tian, Z., Chen, J., Wang, C., & Liu, J. (2023) · IEEE Transactions on Information Forensics and Security · doi:10.1109/tifs.2023.3303718 · Cited by 46
Differentially Private Federated Clustering Over Non-IID Data
Li, Y., Wang, S., Chi, C., & Quek, T. Q. S. (2023) · IEEE Internet of Things Journal · doi:10.1109/jiot.2023.3312852 · Cited by 45
“However, such a strategy requires complicated encryption protocols and substantial extra communication and computation cost [37], thus not applicable to large-scale FL models.”
BackgroundFL-ODP: An Optimized Differential Privacy Enabled Privacy Preserving Federated Learning
Iqbal, M., Tariq, A., Adnan, M., Din, I. U., & Qayyum, T. (2023) · IEEE Access · doi:10.1109/access.2023.3325396 · Cited by 45
“Extensive research efforts have been directed towards developing privacy-preserving techniques for machine learning [17], [18].”
BackgroundFRAMU: Attention-Based Machine Unlearning Using Federated Reinforcement Learning
Shaik, T., Tao, X., Li, L., Xie, H., Cai, T., Zhu, X., & Li, Q. (2024) · IEEE Transactions on Knowledge and Data Engineering · doi:10.1109/tkde.2024.3382726 · Cited by 45
“While privacy has been partially addressed through Differential Privacy [30] and Secure Multiparty Computation [31], these techniques often compromise on model efficiency.”
BackgroundHan, B., Li, B., Jurdak, R., Zhang, P., Zhang, H. Y., Feng, P., & Yuen, C. (2025) · IEEE Internet of Things Journal · doi:10.1109/jiot.2024.3524632 · Cited by 46
“Bonawitz et al. [31] proposed the first SA-based PPFL scheme, which ingeniously combined double masking and secret sharing techniques to protect privacy.”
Beyond the Cloud: Federated Learning and Edge AI for the Next Decade
Thomas, S. G., & Myakala, P. K. (2025) · Journal of Computer and Communications · doi:10.4236/jcc.2025.132004 · Cited by 45
Optimizing Federated Learning With Aggregation Strategies: A Comprehensive Survey
Khan, N., Nisar, S., Khan, M. A., Rehman, Y. A. U., Noor, F., & Barb, G. (2025) · IEEE Open Journal of the Computer Society · doi:10.1109/ojcs.2025.3590102 · Cited by 47
“To achieve these, advanced message-passing techniques that optimize both communication and the model across clients are used to ensure FL systems can accommodate millions of clients [18], [19], [20], [21], [22], [23].”
Privacy in Neural Network Learning: Threats and Countermeasures
Chang, S., & Li, C. (2018) · IEEE Network · doi:10.1109/mnet.2018.1700447 · Cited by 36
“[6] considered local gradients from one participant in each update as a private vector, and designed a protocol to compute the sum of gradient vectors from different participants in a secure fashion.”
BackgroundJatain, D., Sıngh, V., & Dahiya, N. (2021) · Journal of King Saud University - Computer and Information Sciences · doi:10.1016/j.jksuci.2021.05.016 · Cited by 45
“In their federated learning framework, some researchers have ensured the privacy of aggregated user updates using a secure aggregation scheme (Bonawitz et al., 2017).”
InfluentialTowards a robust and trustworthy machine learning system development: An engineering perspective
Xiong, P., Buffett, S., Iqbal, S., Lamontagne, P., Mamun, M., & Molyneaux, H. (2022) · Journal of Information Security and Applications · doi:10.1016/j.jisa.2022.103121 · Cited by 46
“Privacy-Preserving Training/Prediction FHE & polynomial approximation Training Phase Prediction Phase [13, 75] MPC Prediction phase [12] Distributed training [26] Secret sharing & additive HE [82, 91, 115, 145] FHE + MPC Prediction phase [76, 111] Functional encryption”
BackgroundMethodAMBLE: Adjusting mini-batch and local epoch for federated learning with heterogeneous devices
Park, J., Yoon, D., Yeo, S., & Oh, S. (2022) · Journal of Parallel and Distributed Computing · doi:10.1016/j.jpdc.2022.07.009 · Cited by 45
Lin, L., & Zhang, X. (2022) · IEEE Internet of Things Journal · doi:10.1109/jiot.2022.3233024 · Cited by 45
“Therehave been several researches on privacy-preserving federated learning [15], [18], [22], [23], [24], [25], [26], [27], [28], [29], [30], [31], [32], [33], [34].”
BackgroundOne-shot Empirical Privacy Estimation for Federated LearningPreprint
Andrew, G., Kairouz, P., Oh, S., Oprea, A., McMahan, H. B., & Suriyakumar, V. M. (2023) · arXiv (Cornell University) · doi:10.48550/arxiv.2302.03098 · Cited by 45
“Technologies such as secure aggregation [Bonawitz et al., 2017, Bell et al., 2020] can be applied to ensure that even the central server cannot inspect individual updates, but only their aggregate.”
Self-citationFederated Learning without Full Labels: A SurveyPreprint
Jin, Y., Liu, Y., Chen, K., & Yang, Q. (2023) · arXiv (Cornell University) · doi:10.48550/arxiv.2303.14453 · Cited by 45
“Up to now, a plethora of FL techniques has been proposed, focusing primarily on addressing the issues of data heterogeneity [13, 15], system heterogeneity [14, 17], data privacy and security [16, 18], and communication efficiency [12, 19].”
BackgroundPreserving Privacy in Large Language Models: A Survey on Current Threats and Solutions
Miranda, M., Ruzzetti, E. S., Santilli, A., Zanzotto, F. M., Bratières, S., & Rodola, E. (2025) · Transactions on machine learning research · doi:10.48550/arxiv.2408.05212 · Cited by 48
A Generic Federated Recommendation Framework via Fake Marks and Secret Sharing
Lin, Z., Pan, W., Yang, Q., & Ming, Z. (2023) · ACM Transactions on Information Systems · doi:10.1145/3548456 · Cited by 44
“Fortunately, there has been some works on these problems that give feasible solutions [4, 29].”
BackgroundInfluentialNot All Features Are Equal: Discovering Essential Features for Preserving Prediction Privacy
Mireshghallah, F., Taram, M., Jalali, A., Elthakeb, A. T., Tullsen, D., & Esmaeilzadeh, H. (2020) · arXiv · doi:10.48550/arxiv.2003.12154 · Cited by 44
LightVeriFL: A Lightweight and Verifiable Secure Aggregation for Federated Learning
Buyukates, B., So, J., Mahdavifar, H., & Avestimehr, S. (2024) · IEEE Journal on Selected Areas in Information Theory · doi:10.1109/jsait.2024.3391849 · Cited by 44
“Proof: The use of a secure aggregation scheme such as SecAgg [6] or LightSecAgg [8] takes care of the input privacy in the aggregation protocol.”
BackgroundMethodInfluentialOn the Information Theoretic Secure Aggregation With Uncoded Groupwise Keys
Wan, K., Yao, X., Sun, H., Ji, M., & Caire, G. (2024) · IEEE Transactions on Information Theory · doi:10.1109/tit.2024.3422087 · Cited by 44
“…the authors in [9] proposed an information theoretic formulation of the secure aggregation problem with user dropouts originally considered in [6], whose objective is to characterize the fundamental limits of the communication cost while preserving the information theoretic security of the…”
BackgroundMethodResultInfluentialFedNIDS: A Federated Learning Framework for Packet-Based Network Intrusion Detection System
Nguyen, Q. H., Hore, S., Shah, A., Le, T. Q., & Bastian, N. D. (2024) · Digital Threats Research and Practice · doi:10.1145/3696012 · Cited by 44
“Subsequently, the clients share the model updates in a secure and aggregated manner [13].”
InfluentialFLeet: Online Federated Learning via Staleness Awareness and Performance Prediction
Damaskinos, G., Guerraoui, R., Kermarrec, A.-M., Nitu, V., Patra, R., & Taïani, F. (2020) · International Middleware Conference · doi:10.1145/3423211.3425685 · Cited by 44
“Differential privacy [26] is a popular technique for privacy-preserving FL with formal guarantees [9].”
BackgroundSecure Collaborative Deep Learning Against GAN Attacks in the Internet of Things
Chen, Z., Fu, A., Zhang, Y., Liu, Z., Zeng, F., & Deng, R. H. (2020) · IEEE Internet of Things Journal · doi:10.1109/jiot.2020.3033171 · Cited by 44
“Collaborative learning is followed by some improvenments that focus on user dynamics [35], integrity verification [36] and security issues [13], [14].”
BackgroundDeTrust-FL: Privacy-Preserving Federated Learning in Decentralized Trust Setting
Xu, R., Baracaldo, N., Zhou, Y., Anwar, A., Kadhe, S., & Ludwig, H. (2022) · IEEE ... International Conference on Cloud Computing · doi:10.1109/cloud55607.2022.00065 · Cited by 44
“Pairwise masking [9], [22] is a lightweight computation approach where each input is masked with one-time pads while the aggregated output cancels out those pads.”
BackgroundMethodInfluential
Metadata
- Authors
- Bonawitz, K., Ivanov, V., Kreuter, B., Marcedone, A., McMahan, H. B., Patel, S., Ramage, D., Segal, A., & Seth, K.
- Affiliations
- Google (United States) (US)
- Cornell Tech (US)
- Venue
- Proceedings of the 2017 ACM SIGSAC Conference on Computer and Communications Security
- Publisher
- ACM
- Published
- 2017-10-30
- Volume
- 2017
- Pages
- 1175-1191
- ISSN
- 1543-7221 (ISSN-L)
- OpenAlex
- W2767079719
- Cited by
- 3,956
- References
- 55
- Citations by year
- 2026: 701 · 2025: 712 · 2024: 643 · 2023: 564 · 2022: 400 · 2021: 436 · 2020: 299 · 2019: 152 · 2018: 42 · 2017: 7
- FWCI
- 62.37
- Type
- conference paper
- Language
- en
- Open access
- closed
- Locations
- ACM SIGSAC Conference on Computer and Communications Security (CCS) (publishedVersion)
- IACR Cryptology ePrint Archive (submittedVersion)
- Topics
- Physical Sciences › Computer Science › Computer Networks and Communications › Cryptography and Data Security
- Physical Sciences › Computer Science › Artificial Intelligence › Privacy-Preserving Technologies in Data
- Physical Sciences › Computer Science › Information Systems › Cloud Data Security Solutions
- Keywords
- secure aggregation, privacy-preserving machine learning, federated learning, high-dimensional data, communication efficiency, user dropout, honest-but-curious adversary, active adversaries, secure multi-party computation
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.