mlsanityData to 7 Oct 2026

Unrolling SGD: Understanding Factors Influencing Machine Unlearning

arXiv 2109.13398 ↗First posted 27 Sep 2021cs.LG cs.CRcited by 274

On the map: LLM safety and security → Machine unlearning

Abstract (arXiv)

Machine unlearning is the process through which a deployed machine learning model is made to forget about some of its training data points. While naively retraining the model from scratch is an option, it is almost always associated with large computational overheads for deep learning models. Thus, several approaches to approximately unlearn have been proposed along with corresponding metrics that formalize what it means for a model to forget about a data point. In this work, we first taxonomize approaches and metrics of approximate unlearning. As a result, we identify verification error, i.e., the L2 difference between the weights of an approximately unlearned and a naively retrained model, as an approximate unlearning metric that should be optimized for as it subsumes a large class of other metrics. We theoretically analyze the canonical training algorithm, stochastic gradient descent (SGD), to surface the variables which are relevant to reducing the verification error of approximate unlearning for SGD. From this analysis, we first derive an easy-to-compute proxy for verification error (termed unlearning error). The analysis also informs the design of a new training objective penalty that limits the overall change in weights during SGD and as a result facilitates approximate unlearning with lower verification error. We validate our theoretical work through an empirical evaluation on learning with CIFAR-10, CIFAR-100, and IMDB sentiment analysis.

Key facts. Cited by 274 papers in the map (data through 7 Oct 2026). Placed in Machine unlearning, one of 12 directions in LLM safety and security. Builds on 20 of its 44 references that are in the map. First posted on arXiv 27 Sep 2021.

Builds on

20 of 44 references.

Closest ancestors · 2

Same direction, newest first

  1. 2021Adaptive Machine Unlearning8 Jun
  2. 2021Remember What You Want to Forget: Algorithms for Machine Unlearning4 Mar

Other references · 18

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  2. 2020Language Models are Few-Shot Learnerscited by 32,199
  3. 2014Very Deep Convolutional Networks for Large-Scale Image Recognitioncited by 15,267
  4. 2019DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lightercited by 4,243
  5. 2016Deep Learning with Differential Privacycited by 2,682
  6. 2016Membership Inference Attacks against Machine Learning Modelscited by 2,221
  7. 2019Machine Unlearningcited by 1,020
  8. 2018The Secret Sharer: Evaluating and Testing Unintended Memorization in Neural Networkscited by 847
  9. 2019Eternal Sunshine of the Spotless Net: Selective Forgetting in Deep Networkscited by 599
  10. 2019Certified Data Removal from Machine Learning Modelscited by 512
  11. 2020Amnesiac Machine Learningcited by 305
  12. 2020Systematic Evaluation of Privacy Risks of Machine Learning Modelscited by 273
6 more
  1. 2020When Machine Unlearning Jeopardizes Privacycited by 181
  2. 2020Forgetting Outside the Box: Scrubbing Deep Networks of Information Accessible from Input-Output Observationscited by 171
  3. 2020DeltaGrad: Rapid retraining of machine learning modelscited by 143
  4. 2019Analyzing Information Leakage of Updates to Natural Language Modelscited by 89
  5. 2020Machine Unlearning: Linear Filtration for Logit-based Classifierscited by 89
  6. 2021Proof-of-Learning: Definitions and Practicecited by 81

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 1; 2022: 8; 2023: 20; 2024: 79; 2025: 97; 2026: 69.

274 papers.

Sort:

By area:

  1. 2024Negative Preference Optimization: From Catastrophic Collapse to Effective Unlearningcited by 484
  2. 2023SalUn: Empowering Machine Unlearning via Gradient-based Weight Saliency in Both Image Classification and Generationcited by 376
  3. 2022A Survey of Machine Unlearningcited by 298
  4. 2023Large Language Model Unlearningcited by 288
  5. 2023Towards Unbounded Machine Unlearningcited by 261
  6. 2024MUSE: Machine Unlearning Six-Way Evaluation for Language Modelscited by 251
  7. 2024Rethinking Machine Unlearning for Large Language Modelscited by 247
  8. 2024Model Merging in LLMs, MLLMs, and Beyond: Methods, Theories, Applications and Opportunitiescited by 247
  9. 2023Model Sparsity Can Simplify Machine Unlearningcited by 206
  10. 2023To Generate or Not? Safety-Driven Unlearned Diffusion Models Are Still Easy To Generate Unsafe Images ... For Nowcited by 204
  11. 2021On the Necessity of Auditable Algorithmic Definitions for Machine Unlearningcited by 164
  12. 2024Defensive Unlearning with Adversarial Training for Robust Concept Erasure in Diffusion Modelscited by 164
38 more
  1. 2024Towards Safer Large Language Models through Machine Unlearningcited by 134
  2. 2022The Privacy Onion Effect: Memorization is Relativecited by 128
  3. 2024SOUL: Unlocking the Power of Second-Order Optimization for LLM Unlearningcited by 104
  4. 2022Federated Unlearning: How to Efficiently Erase a Client in FL?cited by 90
  5. 2024Guardrail Baselines for Unlearning in LLMscited by 90
  6. 2024What makes unlearning hard and what to do about itcited by 89
  7. 2023Machine Unlearning: Solutions and Challengescited by 85
  8. 2024Reversing the Forget-Retain Objectives: An Efficient LLM Unlearning Framework from Logit Differencecited by 85
  9. 2022Towards Adversarial Evaluations for Inexact Machine Unlearningcited by 82
  10. 2024Inexact Unlearning Needs More Careful Evaluations to Avoid a False Sense of Privacycited by 78
  11. 2024Catastrophic Failure of LLM Unlearning via Quantizationcited by 76
  12. 2024LLM Unlearning via Loss Adjustment with Only Forget Datacited by 71
  13. 2024Protecting Privacy in Multimodal Large Language Models with MLLMU-Benchcited by 61
  14. 2025Rethinking LLM Unlearning Objectives: A Gradient Perspective and Go Beyondcited by 59
  15. 2024Position: LLM Unlearning Benchmarks are Weak Measures of Progresscited by 58
  16. 2024Scissorhands: Scrub Data Influence via Connection Sensitivity in Networkscited by 50
  17. 2024Machine Unlearning in Generative AI: A Surveycited by 49
  18. 2024Are we making progress in unlearning? Findings from the first NeurIPS unlearning competitioncited by 48
  19. 2024UnlearnCanvas: Stylized Image Dataset for Enhanced Machine Unlearning Evaluation in Diffusion Modelscited by 47
  20. 2023Exploring the Landscape of Machine Unlearning: A Comprehensive Survey and Taxonomycited by 44
  21. 2024Challenging Forgets: Unveiling the Worst-Case Forget Sets in Machine Unlearningcited by 44
  22. 2023GNNDelete: A General Strategy for Unlearning in Graph Neural Networkscited by 43
  23. 2024Unified Gradient-Based Machine Unlearning with Remain Geometry Enhancementcited by 43
  24. 2024To Each (Textual Sequence) Its Own: Improving Memorized-Data Unlearning in Large Language Modelscited by 42
  25. 2024Evaluating Copyright Takedown Methods for Language Modelscited by 42
  26. 2025Exploring Criteria of Loss Reweighting to Enhance LLM Unlearningcited by 39
  27. 2024Dissecting Language Models: Machine Unlearning via Selective Pruningcited by 38
  28. 2022Hidden Poison: Machine Unlearning Enables Camouflaged Poisoning Attackscited by 37
  29. 2024Fast Exact Unlearning for In-Context Learning Data for LLMscited by 33
  30. 2024Machine Unlearning: A Comprehensive Surveycited by 33
  31. 2024Revisiting Who's Harry Potter: Towards Targeted Unlearning from a Causal Intervention Perspectivecited by 33
  32. 2023Inductive Graph Unlearningcited by 30
  33. 2024A Review on Machine Unlearningcited by 30
  34. 2024Erasing Concepts from Text-to-Image Diffusion Models with Few-shot Unlearningcited by 29
  35. 2025MMUnlearner: Reformulating Multimodal Machine Unlearning in the Era of Multimodal Large Language Modelscited by 29
  36. 2023SAFE: Machine Unlearning With Shard Graphscited by 28
  37. 2025Decoupled Distillation to Erase: A General Unlearning Method for Any Class-centric Taskscited by 28
  38. 2023Towards Machine Unlearning Benchmarks: Forgetting the Personal Identities in Facial Recognition Systemscited by 27

50 most cited of 274 All 274 (JSON)