mlsanityData to 7 Oct 2026

First Provable Guarantees for Practical Private FL: Beyond Restrictive Assumptions

arXiv 2512.21521 ↗First posted 25 Dec 2025cs.LG math.OC stat.MLcited by 0

Not on the map. Why?

Abstract (arXiv)

Federated Learning (FL) enables collaborative training on decentralized data. Differential privacy (DP) is crucial for FL, but current private methods often rely on unrealistic assumptions (e.g., bounded gradients or heterogeneity), hindering practical application. Existing works that relax these assumptions typically neglect practical FL features, including multiple local updates and partial client participation. We introduce Fed-α-NormEC, the first differentially private FL framework providing provable convergence and DP guarantees under standard assumptions while fully supporting these practical features. Fed-α-NormE integrates local updates (full and incremental gradient steps), separate server and client stepsizes, and, crucially, partial client participation, which is essential for real-world deployment and vital for privacy amplification. Our theoretical guarantees are corroborated by experiments on private deep learning tasks.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 59 of its 70 references that are in the map. First posted on arXiv 25 Dec 2025.

Builds on

59 of 70 references.

References · 59

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  2. 2016Communication-Efficient Learning of Deep Networks from Decentralized Datacited by 7,592
  3. 2019Advances and Open Problems in Federated Learningcited by 3,021
  4. 2016Deep Learning with Differential Privacycited by 2,682
  5. 2016Federated Learning: Strategies for Improving Communication Efficiencycited by 1,461
  6. 2020Adaptive Federated Optimizationcited by 1,036
  7. 2018Exploiting Unintended Feature Leakage in Collaborative Learningcited by 509
  8. 2017Differentially Private Federated Learning: A Client Level Perspectivecited by 451
  9. 2020A Unified Theory of Decentralized SGD with Changing Topology and Local Updatescited by 361
  10. 2021A Field Guide to Federated Optimizationcited by 312
  11. 2019Tighter Theory for Local SGD on Identical and Heterogeneous Datacited by 266
  12. 2018Expanding the Reach of Federated Learning by Reducing Client Resource Requirementscited by 162
38 more
  1. 2020Federated Learning with Compression: Unified Analysis and Sharp Guaranteescited by 159
  2. 2020Understanding Gradient Clipping in Private SGD: A Geometric Perspectivecited by 140
  3. 2022ProxSkip: Yes! Local Gradient Steps Provably Lead to Communication Acceleration! Finally!cited by 134
  4. 2021EF21: A New, Simpler, Theoretically Better, and Practically Faster Error Feedbackcited by 131
  5. 2021When the Curious Abandon Honesty: Federated Learning Is Not Privatecited by 126
  6. 2020Optimal Client Sampling for Federated Learningcited by 123
  7. 2020Random Reshuffling: Simple Analysis with Vast Improvementscited by 112
  8. 2023Stochastic Controlled Averaging for Federated Learning with Communication Compressioncited by 100
  9. 2020Local SGD: Unified Theory and New Efficient Methodscited by 88
  10. 2022Automatic Clipping: Differentially Private Deep Learning Made Easier and Strongercited by 87
  11. 2023Revisiting Gradient Clipping: Stochastic bias and tight convergence guaranteescited by 83
  12. 2020From Local SGD to Local Fixed-Point Methods for Federated Learningcited by 74
  13. 2021Differentially Private Federated Learning on Heterogeneous Datacited by 74
  14. 2018Random Shuffling Beats SGD after Finite Epochscited by 66
  15. 2021Understanding Clipping for Federated Learning: Convergence and Client-Level Differential Privacycited by 66
  16. 2019How Good is SGD with Random Shuffling?cited by 63
  17. 2020Linearly Converging Error Compensated SGDcited by 61
  18. 2019Federated Learning with Bayesian Differential Privacycited by 60
  19. 2015Incremental Gradient, Subgradient, and Proximal Methods for Convex Optimization: A Surveycited by 44
  20. 2020On the Outsized Importance of Learning Rates in Local Update Methodscited by 34
  21. 2022SoteriaFL: A Unified Framework for Private Federated Learning with Communication Compressioncited by 34
  22. 2021Proximal and Federated Random Reshufflingcited by 33
  23. 2023Momentum Provably Improves Error Feedback!cited by 31
  24. 2022EF21-P and Friends: Improved Theoretical Communication Complexity for Distributed Optimization with Bidirectional Compressioncited by 28
  25. 2019Efficient Privacy-Preserving Stochastic Nonconvex Optimizationcited by 27
  26. 2022Private Non-Convex Federated Learning Without a Trusted Servercited by 26
  27. 2022Server-Side Stepsizes and Sampling Without Replacement Provably Help in Federated Optimizationcited by 23
  28. 2022A Communication-Efficient Distributed Gradient Clipping Algorithm for Training Deep Neural Networkscited by 20
  29. 2020Private and Communication-Efficient Edge Learning: A Sparse Differential Gaussian-Masking Distributed SGD Approachcited by 18
  30. 2022Federated Optimization Algorithms with Random Reshuffling and Gradient Compressioncited by 18
  31. 2021Random Shuffling Beats SGD Only After Many Epochs on Ill-Conditioned Problemscited by 16
  32. 2023EControl: Fast Distributed Optimization with Compression and Error Controlcited by 14
  33. 2023Federated Learning with Regularized Client Participationcited by 13
  34. 2024The Limits and Potentials of Local SGD for Distributed Heterogeneous Learning with Intermittent Communicationcited by 12
  35. 2019Gradient Descent with Compressed Iteratescited by 11
  36. 2019Distributed Fixed Point Methods with Compressed Iteratescited by 11
  37. 2022Federated Random Reshuffling with Compression and Variance Reductioncited by 11
  38. 2023Clip21: Error Feedback for Gradient Clippingcited by 10

50 most cited of 59 All 59 (JSON)

Built on by

Not cited yet.