mlsanityData to 7 Oct 2026

Machine Unlearning: Solutions and Challenges

arXiv 2308.07061 ↗First posted 14 Aug 2023cs.LG cs.AIcited by 85

On the map: LLM safety and security → Machine unlearning

Abstract (arXiv)

Machine learning models may inadvertently memorize sensitive, unauthorized, or malicious data, posing risks of privacy breaches, security vulnerabilities, and performance degradation. To address these issues, machine unlearning has emerged as a critical technique to selectively remove specific training data points' influence on trained models. This paper provides a comprehensive taxonomy and analysis of the solutions in machine unlearning. We categorize existing solutions into exact unlearning approaches that remove data influence thoroughly and approximate unlearning approaches that efficiently minimize data influence. By comprehensively reviewing solutions, we identify and discuss their strengths and limitations. Furthermore, we propose future directions to advance machine unlearning and establish it as an essential capability for trustworthy and adaptive machine learning models. This paper provides researchers with a roadmap of open problems, encouraging impactful contributions to address real-world needs for selective data removal.

Key facts. Cited by 85 papers in the map (data through 7 Oct 2026). Placed in Machine unlearning, one of 12 directions in LLM safety and security. Builds on 74 of its 139 references that are in the map. First posted on arXiv 14 Aug 2023.

Builds on

74 of 139 references.

Closest ancestors · 17

Same direction, newest first

  1. 2023Boundary Unlearning21 Mar
  2. 2022Proof of Unlearning: Definitions and Instantiation20 Oct
  3. 2022Verifiable and Provably Secure Machine Unlearning17 Oct
  4. 2022Algorithms that Approximate Data Removal: New Results and Limitations25 Sep
  5. 2022A Survey of Machine Unlearning6 Sep
  6. 2022An Introduction to Machine Unlearning2 Sep
  7. 2022Quark: Controllable Text Generation with Reinforced Unlearning26 May
  8. 2022Deep Unlearning via Randomized Conditionally Independent Hessians15 Apr
  9. 2022Knowledge Removal in Sampling-based Bayesian Inference24 Mar
  10. 2022Markov Chain Monte Carlo-Based Machine Unlearning: Unlearning What Needs to be Forgotten28 Feb
  11. 2022Zero-Shot Machine Unlearning14 Jan
  12. 2021Fast Yet Effective Machine Unlearning17 Nov
5 more
  1. 2021On the Necessity of Auditable Algorithmic Definitions for Machine Unlearning22 Oct
  2. 2021Unrolling SGD: Understanding Factors Influencing Machine Unlearning27 Sep
  3. 2021Machine Unlearning of Features and Labels26 Aug
  4. 2021Adaptive Machine Unlearning8 Jun
  5. 2021Remember What You Want to Forget: Algorithms for Machine Unlearning4 Mar

Other references · 57

  1. 2016Communication-Efficient Learning of Deep Networks from Decentralized Datacited by 7,592
  2. 2015Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shiftcited by 6,979
  3. 2017A Unified Approach to Interpreting Model Predictionscited by 6,488
  4. 2016Pointer Sentinel Mixture Modelscited by 2,992
  5. 2016Membership Inference Attacks against Machine Learning Modelscited by 2,221
  6. 2017Understanding Black-box Predictions via Influence Functionscited by 1,714
  7. 2019Explainable Artificial Intelligence (XAI): Concepts, Taxonomies, Opportunities and Challenges toward Responsible AIcited by 1,543
  8. 2017Towards A Rigorous Science of Interpretable Machine Learningcited by 1,400
  9. 2019Machine Unlearningcited by 1,020
  10. 2018Overcoming catastrophic forgetting with hard attention to the taskcited by 672
  11. 2013Speech Recognition with Deep Recurrent Neural Networkscited by 666
  12. 2019Eternal Sunshine of the Spotless Net: Selective Forgetting in Deep Networkscited by 599
38 more
  1. 2019Certified Data Removal from Machine Learning Modelscited by 512
  2. 2020Graph Neural Networks in Recommender Systems: A Surveycited by 493
  3. 2019Making AI Forget You: Data Deletion in Machine Learningcited by 447
  4. 2014New insights and perspectives on the natural gradient methodcited by 435
  5. 2020Exploring Self-attention for Image Recognitioncited by 313
  6. 2021Membership Inference Attacks on Machine Learning: A Surveycited by 306
  7. 2020Descent-to-Delete: Gradient-Based Methods for Machine Unlearningcited by 303
  8. 2017Certified Defenses for Data Poisoning Attackscited by 288
  9. 2023Multimodal Deep Learningcited by 257
  10. 2020Approximate Data Deletion from Machine Learning Modelscited by 250
  11. 2020Auditing Differentially Private Machine Learning: How Private is Private SGD?cited by 213
  12. 2020When Machine Unlearning Jeopardizes Privacycited by 181
  13. 2020Forgetting Outside the Box: Scrubbing Deep Networks of Information Accessible from Input-Output Observationscited by 171
  14. 2020Mixed-Privacy Forgetting in Deep Networkscited by 158
  15. 2020DeltaGrad: Rapid retraining of machine learning modelscited by 143
  16. 2018Backdoor Embedding in Convolutional Neural Network Models via Invisible Perturbationcited by 131
  17. 2021Graph Unlearningcited by 129
  18. 2022The Privacy Onion Effect: Memorization is Relativecited by 128
  19. 2020Machine Unlearning for Random Forestscited by 125
  20. 2020Variational Bayesian Unlearningcited by 114
  21. 2022The Right to be Forgotten in Federated Learning: An Efficient Realization with Rapid Retrainingcited by 109
  22. 2021Federated Unlearning via Class-Discriminative Pruningcited by 108
  23. 2020On the Privacy Risks of Algorithmic Fairnesscited by 82
  24. 2021Proof-of-Learning: Definitions and Practicecited by 81
  25. 2019Measure Contribution of Participants in Federated Learningcited by 78
  26. 2021Practical Blind Membership Inference Attack via Differential Comparisonscited by 73
  27. 2021Manipulating SGD with Data Ordering Attackscited by 73
  28. 2022Recommendation Unlearningcited by 71
  29. 2022Backdoor Defense with Machine Unlearningcited by 61
  30. 2023GIF: A General Graph Unlearning Strategy via Influence Functioncited by 59
  31. 2020Learn to Forget: Machine Unlearning via Neuron Maskingcited by 56
  32. 2022PUMA: Performance Unchanged Model Augmentation for Training Data Removalcited by 56
  33. 2022FedRecover: Recovering from Poisoning Attacks in Federated Learning using Historical Informationcited by 53
  34. 2020Formalizing Data Deletion in the Context of the Right to be Forgottencited by 46
  35. 2021Survey: Leakage and Privacy at Inference Timecited by 44
  36. 2023GNNDelete: A General Strategy for Unlearning in Graph Neural Networkscited by 43
  37. 2022VeriFi: Towards Verifiable Federated Unlearningcited by 41
  38. 2023Heterogeneous Federated Knowledge Graph Embedding Learning and Unlearningcited by 39

50 most cited of 57 All 57 (JSON)

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 2; 2024: 22; 2025: 36; 2026: 25.

85 papers.

Sort:

By area:

  1. 2024Catastrophic Failure of LLM Unlearning via Quantizationcited by 76
  2. 2024Unified Gradient-Based Machine Unlearning with Remain Geometry Enhancementcited by 43
  3. 2024Threats, Attacks, and Defenses in Machine Unlearning: A Surveycited by 34
  4. 2024Continual Forgetting for Pre-trained Vision Modelscited by 29
  5. 2025Protecting Users From Themselves: Safeguarding Contextual Privacy in Interactions with Conversational Agentscited by 26
  6. 2024Towards Scalable Exact Machine Unlearning Using Parameter-Efficient Fine-Tuningcited by 25
  7. 2025SAEs Can Improve Unlearning: Dynamic Sparse Autoencoder Guardrails for Precision Unlearning in LLMscited by 21
  8. 2024Learning to Unlearn for Robust Machine Unlearningcited by 19
  9. 2024Is Retain Set All You Need in Machine Unlearning? Restoring Performance of Unlearned Models with Out-Of-Distribution Imagescited by 18
  10. 2025LLM Unlearning with LLM Beliefscited by 17
  11. 2025NoT: Federated Unlearning via Weight Negationcited by 15
  12. 2023DUCK: Distance-based Unlearning via Centroid Kinematicscited by 13
38 more
  1. 2024Remaining-data-free Machine Unlearning by Suppressing Sample Contributioncited by 13
  2. 2024Towards Natural Machine Unlearningcited by 11
  3. 2024Ferrari: Federated Feature Unlearning via Optimizing Feature Sensitivitycited by 11
  4. 2025ForgetMe: Evaluating Selective Forgetting in Generative Modelscited by 11
  5. 2025Agents Are All You Need for LLM Unlearningcited by 9
  6. 2023DeepClean: Machine Unlearning on the Cheap by Resetting Privacy Sensitive Weights using the Fisher Diagonalcited by 7
  7. 2025Existing Large Language Model Unlearning Evaluations Are Inconclusivecited by 7
  8. 2025DualOptim: Enhancing Efficacy and Stability in Machine Unlearning with Dual Optimizerscited by 6
  9. 2025A Unified Gradient-based Framework for Task-agnostic Continual Learning-Unlearningcited by 6
  10. 2025Do Not Mimic My Voice: Speaker Identity Unlearning for Zero-Shot Text-to-Speechcited by 6
  11. 2025Approximate Domain Unlearning for Vision-Language Modelscited by 6
  12. 2024Verifying Machine Unlearning with Explainable AIcited by 5
  13. 2025Position: Bridge the Gaps between Machine Unlearning and AI Regulationcited by 5
  14. 2025"Alexa, can you forget me?" Machine Unlearning Benchmark in Spoken Language Understandingcited by 5
  15. 2026FG-OrIU: Towards Better Forgetting via Feature-Gradient Orthogonality for Incremental Unlearningcited by 5
  16. 2024Red-Teaming Segment Anything Modelcited by 4
  17. 2024Moderating the Generalization of Score-based Generative Modelcited by 4
  18. 2025Practical Continual Forgetting for Pre-trained Vision Modelscited by 4
  19. 2025Machine Learners Should Acknowledge the Legal Implications of Large Language Models as Personal Datacited by 4
  20. 2025Deep Contrastive Unlearning for Language Modelscited by 4
  21. 2025Unlearning Comparator: A Visual Analytics System for Comparative Evaluation of Machine Unlearning Methodscited by 4
  22. 2025Auditing Approximate Machine Unlearning for Differentially Private Modelscited by 4
  23. 2025CoUn: Empowering Machine Unlearning via Contrastive Learningcited by 4
  24. 2026Forget Many, Forget Right: Scalable and Precise Concept Unlearning in Diffusion Modelscited by 4
  25. 2026FaLW: A Forgetting-aware Loss Reweighting for Long-tailed Unlearningcited by 3
  26. 2024Eternal Sunshine of the Mechanical Mind: The Irreconcilability of Machine Learning and the Right to be Forgottencited by 2
  27. 2024LMEraser: Large Model Unlearning through Adaptive Prompt Tuningcited by 2
  28. 2024The Approximate Fisher Influence Function: Faster Estimation of Data Influence in Statistical Modelscited by 2
  29. 2024Uncertainty Quantification of Data Shapley via Statistical Inferencecited by 2
  30. 2024Neural Machine Unrankingcited by 2
  31. 2025Exploring Incremental Unlearning: Techniques, Challenges, and Future Directionscited by 2
  32. 2025Unlearning as Ablation: Toward a Falsifiable Benchmark for Generative Scientific Discoverycited by 2
  33. 2025Selective, Controlled and Domain-Agnostic Unlearning in Pretrained CLIP: A Training- and Data-Free Approachcited by 2
  34. 2026Distinguishable Deletion: Unifying Knowledge Erasure and Refusal for Large Language Model Unlearningcited by 2
  35. 2024GraphMU: Repairing Robustness of Graph Neural Networks via Machine Unlearningcited by 1
  36. 2024Instance-Level Difficulty: A Missing Perspective in Machine Unlearningcited by 1
  37. 2024Neural Corrective Machine Unrankingcited by 1
  38. 2025A Framework for Cryptographic Verifiability of End-to-End AI Pipelinescited by 1

50 most cited of 85 All 85 (JSON)