mlsanityData to 14 Oct 2026

Machine Unlearning

arXiv 1912.03817 ↗First posted 9 Dec 2019cs.CR cs.AI cs.LGcited by 1,061

Not on the map. Why?

Cited by 756 in LLM safety and security, but placed by its own abstract, which didn’t fit.

Abstract (arXiv)

Once users have shared their data online, it is generally difficult for them to revoke access and ask for the data to be deleted. Machine learning (ML) exacerbates this problem because any model trained with said data may have memorized it, putting users at risk of a successful privacy attack exposing their information. Yet, having models unlearn is notoriously difficult. We introduce SISA training, a framework that expedites the unlearning process by strategically limiting the influence of a data point in the training procedure. While our framework is applicable to any learning algorithm, it is designed to achieve the largest improvements for stateful algorithms like stochastic gradient descent for deep neural networks. SISA training reduces the computational overhead associated with unlearning, even in the worst-case setting where unlearning requests are made uniformly across the training set. In some cases, the service provider may have a prior on the distribution of unlearning requests that will be issued by users. We may take this prior into account to partition and order data accordingly, and further decrease overhead from unlearning. Our evaluation spans several datasets from different domains, with corresponding motivations for unlearning. Under no distributional assumptions, for simple learning tasks, we observe that SISA training improves time to unlearn points from the Purchase dataset by 4.63x, and 2.45x for the SVHN dataset, over retraining from scratch. SISA training also provides a speed-up of 1.36x in retraining for complex learning tasks such as ImageNet classification; aided by transfer learning, this results in a small degradation in accuracy. Our work contributes to practical data governance in machine unlearning.

Key facts. Cited by 1,061 papers in the map (data through 14 Oct 2026). First posted on arXiv 9 Dec 2019.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2021: 26; 2022: 61; 2023: 117; 2024: 263; 2025: 289; 2026: 305.

1,061 papers.

Sort:

By area:

  1. 2023A Comprehensive Survey of Continual Learning: Theory, Method and Applicationcited by 872
  2. 2023Erasing Concepts from Diffusion Modelscited by 539
  3. 2024TOFU: A Task of Fictitious Unlearning for LLMscited by 533
  4. 2024Negative Preference Optimization: From Catastrophic Collapse to Effective Unlearningcited by 504
  5. 2022Knowledge Unlearning for Mitigating Privacy Risks in Language Modelscited by 437
  6. 2025Agentic Context Engineering: Evolving Contexts for Self-Improving Language Modelscited by 398
  7. 2023Detecting Pretraining Data from Large Language Modelscited by 387
  8. 2021Remember What You Want to Forget: Algorithms for Machine Unlearningcited by 371
  9. 2021Membership Inference Attacks on Machine Learning: A Surveycited by 321
  10. 2023Ablating Concepts in Text-to-Image Diffusion Modelscited by 317
  11. 2022A Survey of Machine Unlearningcited by 312
  12. 2023Large Language Model Unlearningcited by 302
38 more
  1. 2024Frontier Models are Capable of In-context Schemingcited by 287
  2. 2021Unrolling SGD: Understanding Factors Influencing Machine Unlearningcited by 279
  3. 2023Towards Unbounded Machine Unlearningcited by 268
  4. 2024MUSE: Machine Unlearning Six-Way Evaluation for Language Modelscited by 261
  5. 2024Rethinking Machine Unlearning for Large Language Modelscited by 257
  6. 2024Continual Learning of Large Language Models: A Comprehensive Surveycited by 234
  7. 2021Machine Unlearning of Features and Labelscited by 229
  8. 2023In-Context Unlearning: Language Models as Few Shot Unlearnerscited by 224
  9. 2022Memorization Without Overfitting: Analyzing the Training Dynamics of Large Language Modelscited by 220
  10. 2021Fast Yet Effective Machine Unlearningcited by 217
  11. 2023Unlearn What You Want to Forget: Efficient Unlearning for LLMscited by 215
  12. 2023Model Sparsity Can Simplify Machine Unlearningcited by 214
  13. 2022Can Bad Teaching Induce Forgetting? Unlearning in Deep Networks using an Incompetent Teachercited by 194
  14. 2023Selective Amnesia: A Continual Learning Approach to Forgetting in Deep Generative Modelscited by 177
  15. 2021Adaptive Machine Unlearningcited by 176
  16. 2024Eight Methods to Evaluate Robust Unlearning in LLMscited by 163
  17. 2024LLM Defenses Are Not Robust to Multi-Turn Human Jailbreaks Yetcited by 151
  18. 2024Towards Safer Large Language Models through Machine Unlearningcited by 139
  19. 2021Graph Unlearningcited by 138
  20. 2023Boundary Unlearningcited by 134
  21. 2022Zero-Shot Machine Unlearningcited by 133
  22. 2024Visual In-Context Learning for Large Vision-Language Modelscited by 128
  23. 2023Explainable Artificial Intelligence (XAI) 2.0: A Manifesto of Open Challenges and Interdisciplinary Research Directionscited by 125
  24. 2024An Adversarial Perspective on Machine Unlearning for AI Safetycited by 123
  25. 2024Machine Unlearning of Pre-trained Large Language Modelscited by 119
  26. 2022The Right to be Forgotten in Federated Learning: An Efficient Realization with Rapid Retrainingcited by 118
  27. 2022Measuring Forgetting of Memorized Training Examplescited by 118
  28. 2023Foundation Models and Fair Usecited by 116
  29. 2021Federated Unlearning via Class-Discriminative Pruningcited by 115
  30. 2024Secret Collusion among AI Agents: Multi-Agent Deception via Steganographycited by 113
  31. 2022Synthetic Text Generation with Differential Privacy: A Simple and Practical Recipecited by 108
  32. 2024SOUL: Unlocking the Power of Second-Order Optimization for LLM Unlearningcited by 104
  33. 2023KGA: A General Machine Unlearning Framework Based on Knowledge Gap Alignmentcited by 99
  34. 2024Guardrail Baselines for Unlearning in LLMscited by 96
  35. 2022Federated Unlearning: How to Efficiently Erase a Client in FL?cited by 95
  36. 2023Machine Unlearning: Solutions and Challengescited by 93
  37. 2024What makes unlearning hard and what to do about itcited by 89
  38. 2024Reversing the Forget-Retain Objectives: An Efficient LLM Unlearning Framework from Logit Differencecited by 88

50 most cited of 1,061 All 1,061 (JSON)