mlsanityData to 7 Oct 2026

DMM: Distributed Matrix Mechanism for Differentially-Private Federated Learning Based on Constant-Overhead Linear Secret Resharing

arXiv 2410.16161 ↗First posted 21 Oct 2024cs.CR cs.LGcited by 3

Not on the map. Why?

Abstract (arXiv)

Federated Learning (FL) solutions with central Differential Privacy (DP) have seen large improvements in their utility in recent years arising from the matrix mechanism, while FL solutions with distributed (more private) DP have lagged behind. In this work, we introduce the distributed matrix mechanism to achieve the best-of-both-worlds; better privacy of distributed DP and better utility from the matrix mechanism. We accomplish this using a novel cryptographic protocol that securely transfers sensitive values across client committees of different training iterations with constant communication overhead. This protocol accommodates the dynamic participation of users required by FL, including those that may drop out from the computation. We provide experiments which show that our mechanism indeed significantly improves the utility of FL models compared to previous distributed DP mechanisms, with little added overhead.

Key facts. Cited by 3 papers in the map (data through 7 Oct 2026). Builds on 27 of its 54 references that are in the map. First posted on arXiv 21 Oct 2024.

Builds on

27 of 54 references.

References · 27

  1. 2016Communication-Efficient Learning of Deep Networks from Decentralized Datacited by 7,592
  2. 2018Federated Optimization in Heterogeneous Networkscited by 2,971
  3. 2016Deep Learning with Differential Privacycited by 2,682
  4. 2020Adaptive Federated Optimizationcited by 1,036
  5. 2018LEAF: A Benchmark for Federated Settingscited by 780
  6. 2019Local Model Poisoning Attacks to Byzantine-Robust Federated Learningcited by 479
  7. 2016Concentrated Differential Privacy: Simplifications, Extensions, and Lower Boundscited by 339
  8. 2020Data Poisoning Attacks Against Federated Learning Systemscited by 266
  9. 2018Subsampled R'enyi Differential Privacy and Analytical Moments Accountantcited by 234
  10. 2021Practical and Private (Deep) Learning without Sampling or Shufflingcited by 169
  11. 2018Amplification by Shuffling: From Local to Central Differential Privacy via Anonymitycited by 163
  12. 2016Concentrated Differential Privacycited by 139
15 more
  1. 2020The Discrete Gaussian for Differential Privacycited by 134
  2. 2021The Distributed Discrete Gaussian Mechanism for Federated Learning with Secure Aggregationcited by 127
  3. 2020Hiding Among the Clones: A Simple and Nearly Optimal Analysis of Privacy Amplification by Shufflingcited by 96
  4. 2019Hypothesis Testing Interpretations and Renyi Differential Privacycited by 84
  5. 2021The Skellam Mechanism for Differentially Private Federated Learningcited by 66
  6. 2022Improved Differential Privacy for SGD via Optimal Private Linear Operators on Adaptive Streamscited by 59
  7. 2023(Amplified) Banded Matrix Factorization: A unified approach to private trainingcited by 49
  8. 2022Multi-Epoch Matrix Factorization Mechanisms for Private Machine Learningcited by 48
  9. 2020A Better Bound Gives a Hundred Rounds: Enhanced Privacy Guarantees via f-Divergencescited by 21
  10. 2022DReS-FL: Dropout-Resilient Secure Federated Learning for Non-IID Clients via Secret Data Sharingcited by 17
  11. 2024Efficient and Near-Optimal Noise Generation for Streaming Differential Privacycited by 17
  12. 2023Flamingo: Multi-Round Single-Server Secure Aggregation with Applications to Private Federated Learningcited by 16
  13. 2022SecureFedYJ: a safe feature Gaussianization protocol for Federated Learningcited by 9
  14. 2021SASH: Efficient Secure Aggregation Based on SHPRG For Federated Learningcited by 4
  15. 2024Secure Stateful Aggregation: A Practical Protocol with Applications in Differentially-Private Federated Learningcited by 2

Built on by

Cited by year

Citing papers per year: 2025: 3.

3 papers.

  1. 2025Back to Square Roots: An Optimal Bound on the Matrix Factorization Error for Multi-Epoch Differentially Private SGDcited by 7
  2. 2025Learning Rate Scheduling with Matrix Factorization for Private Trainingcited by 2
  3. 2025FedVideoMAE: Efficient Federated Video Moderation with Differential Privacy and Secure Aggregationcited by 0