mlsanityData to 7 Oct 2026

Deep Leakage from Gradients

arXiv 1906.08935 ↗First posted 21 Jun 2019cs.LG cs.CR stat.MLcited by 1,004

Not on the map. Why?

Cited by 50 in LLM safety and security, but placed by its own abstract, which didn’t fit.

Abstract (arXiv)

Exchanging gradients is a widely used method in modern multi-node machine learning system (e.g., distributed training, collaborative learning). For a long time, people believed that gradients are safe to share: i.e., the training data will not be leaked by gradient exchange. However, we show that it is possible to obtain the private training data from the publicly shared gradients. We name this leakage as Deep Leakage from Gradient and empirically validate the effectiveness on both computer vision and natural language processing tasks. Experimental results show that our attack is much stronger than previous approaches: the recovery is pixel-wise accurate for images and token-wise matching for texts. We want to raise people's awareness to rethink the gradient's safety. Finally, we discuss several possible strategies to prevent such deep leakage. The most effective defense method is gradient pruning.

Key facts. Cited by 1,004 papers in the map (data through 7 Oct 2026). First posted on arXiv 21 Jun 2019.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 140; 2022: 185; 2023: 197; 2024: 191; 2025: 158; 2026: 133.

1,004 papers.

Sort:

By area:

  1. 2021FedProto: Federated Prototype Learning across Heterogeneous Clientscited by 349
  2. 2021Federated Learning on Non-IID Data: A Surveycited by 345
  3. 2021See through Gradients: Image Batch Recovery via GradInversioncited by 320
  4. 2021A Field Guide to Federated Optimizationcited by 312
  5. 2021Trustworthy AI: From Principles to Practicescited by 206
  6. 2021Evaluating Gradient Inversion Attacks and Defenses in Federated Learningcited by 196
  7. 2023Text Embeddings Reveal (Almost) As Much As Textcited by 189
  8. 2021Federated Learning with Buffered Asynchronous Aggregationcited by 188
  9. 2021FedKD: Communication Efficient Federated Learning via Knowledge Distillationcited by 185
  10. 2023Towards Building the Federated GPT: Federated Instruction Tuningcited by 184
  11. 2021Federated Learning for Smart Healthcare: A Surveycited by 182
  12. 2024Security and Privacy Challenges of Large Language Models: A Surveycited by 179
38 more
  1. 2021Graph Condensation for Graph Neural Networkscited by 150
  2. 2022Reconstructing Training Data with Informed Adversariescited by 143
  3. 2022Federated Learning from Pre-Trained Models: A Contrastive Learning Approachcited by 141
  4. 2022PromptFL: Let Federated Participants Cooperatively Learn Prompts Instead of Models -- Federated Learning in Age of Foundation Modelcited by 133
  5. 2022Reconstructing Training Data from Trained Neural Networkscited by 129
  6. 2023ProPILE: Probing Privacy Leakage in Large Language Modelscited by 127
  7. 2021When the Curious Abandon Honesty: Federated Learning Is Not Privatecited by 126
  8. 2021Distributed Learning in Wireless Networks: Recent Progress and Future Challengescited by 125
  9. 2022FedALA: Adaptive Local Aggregation for Personalized Federated Learningcited by 121
  10. 2023Machine Learning for Synthetic Data Generation: A Reviewcited by 118
  11. 2022The Right to be Forgotten in Federated Learning: An Efficient Realization with Rapid Retrainingcited by 109
  12. 2021Robbing the Fed: Directly Obtaining Private Data in Federated Learning with Modified Modelscited by 106
  13. 2022Vertical Federated Learning: Concepts, Advances and Challengescited by 104
  14. 2021Gradient Inversion with Generative Image Priorcited by 101
  15. 2021PPFL: Privacy-preserving Federated Learning with Trusted Execution Environmentscited by 93
  16. 2021Trustworthy AI: A Computational Perspectivecited by 93
  17. 2022Trustworthy Graph Neural Networks: Aspects, Methods and Trendscited by 88
  18. 2021Label Leakage and Protection in Two-party Split Learningcited by 87
  19. 2021Large Scale Private Learning via Low-rank Reparametrizationcited by 87
  20. 2021Do Not Let Privacy Overbill Utility: Gradient Embedding Perturbation for Private Learningcited by 86
  21. 2022Position: Considerations for Differentially Private Learning with Large-Scale Public Pretrainingcited by 86
  22. 2023Multimodal Federated Learning via Contrastive Representation Ensemblecited by 85
  23. 2021FedNLP: Benchmarking Federated Learning Methods for Natural Language Processing Taskscited by 84
  24. 2022Survey on Federated Learning Threats: concepts, taxonomy on attacks and defences, experimental study and challengescited by 81
  25. 2023Sentence Embedding Leaks More Information than You Expect: Generative Embedding Inversion Attack to Recover the Whole Sentencecited by 81
  26. 2021FedProc: Prototypical Contrastive Federated Learning on Non-IID datacited by 80
  27. 2022Untargeted Backdoor Watermark: Towards Harmless and Stealthy Dataset Copyright Protectioncited by 80
  28. 2021Federated Reinforcement Learning: Techniques, Applications, and Open Challengescited by 79
  29. 2022Source-Free Unsupervised Domain Adaptation: A Surveycited by 79
  30. 2021Federated Reconstruction: Partially Local Federated Learningcited by 78
  31. 2023Federated Learning for Medical Image Analysis: A Surveycited by 77
  32. 2021CAFE: Catastrophic Data Leakage in Vertical Federated Learningcited by 76
  33. 2022When Federated Learning Meets Pre-trained Language Models' Parameter-Efficient Tuning Methodscited by 74
  34. 2022Fishing for User Data in Large-Batch Federated Learning via Gradient Magnificationcited by 72
  35. 2022Auditing Privacy Defenses in Federated Learning via Generative Gradient Leakagecited by 72
  36. 2022Recovering Private Text in Federated Learning of Language Modelscited by 67
  37. 2021Understanding Clipping for Federated Learning: Convergence and Client-Level Differential Privacycited by 66
  38. 2021Eluding Secure Aggregation in Federated Learning via Model Inconsistencycited by 66

50 most cited of 1,004 All 1,004 (JSON)