mlsanityData to 14 Oct 2026

LoTUS: Large-Scale Machine Unlearning with a Taste of Uncertainty

arXiv 2503.18314 ↗First posted 24 Mar 2025cs.LG cs.AI cs.CVcited by 13

On the map: LLM safety and security → Machine unlearning

Abstract (arXiv)

We present LoTUS, a novel Machine Unlearning (MU) method that eliminates the influence of training samples from pre-trained models, avoiding retraining from scratch. LoTUS smooths the prediction probabilities of the model up to an information-theoretic bound, mitigating its over-confidence stemming from data memorization. We evaluate LoTUS on Transformer and ResNet18 models against eight baselines across five public datasets. Beyond established MU benchmarks, we evaluate unlearning on ImageNet1k, a large-scale dataset, where retraining is impractical, simulating real-world conditions. Moreover, we introduce the novel Retrain-Free Jensen-Shannon Divergence (RF-JSD) metric to enable evaluation under real-world conditions. The experimental results show that LoTUS outperforms state-of-the-art methods in terms of both efficiency and effectiveness. Code: https://github.com/cspartalis/LoTUS.

Key facts. Cited by 13 papers in the map (data through 14 Oct 2026). Placed in Machine unlearning, one of 12 directions in LLM safety and security. Builds on 23 of its 39 references that are in the map. First posted on arXiv 24 Mar 2025.

Builds on

23 of 39 references.

Closest ancestors · 9

Same direction, newest first

  1. 2024Are we making progress in unlearning? Findings from the first NeurIPS unlearning competition13 Jun
  2. 2023Towards Machine Unlearning Benchmarks: Forgetting the Personal Identities in Facial Recognition Systems3 Nov
  3. 2023Fast Model Debias with Machine Unlearning19 Oct
  4. 2023Fast Machine Unlearning Without Retraining Through Selective Synaptic Dampening15 Aug
  5. 2023Boundary Unlearning21 Mar
  6. 2023Towards Unbounded Machine Unlearning20 Feb
  7. 2023Learning to Unlearn: Instance-wise Unlearning for Pre-trained Classifiers27 Jan
  8. 2022Can Bad Teaching Induce Forgetting? Unlearning in Deep Networks using an Incompetent Teacher17 May
  9. 2021Fast Yet Effective Machine Unlearning17 Nov

Other references · 14

  1. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  2. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 30,186
  3. 2014ImageNet Large Scale Visual Recognition Challengecited by 8,498
  4. 2016Membership Inference Attacks against Machine Learning Modelscited by 2,281
  5. 2019Machine Unlearningcited by 1,061
  6. 2018MINE: Mutual Information Neural Estimationcited by 683
  7. 2019Eternal Sunshine of the Spotless Net: Selective Forgetting in Deep Networkscited by 618
  8. 2019Certified Data Removal from Machine Learning Modelscited by 535
  9. 2023Selective Amnesia: A Continual Learning Approach to Forgetting in Deep Generative Modelscited by 177
  10. 2023CIFAKE: Image Classification and Explainable Identification of AI-Generated Synthetic Imagescited by 84
  11. 2024CPR: Retrieval Augmented Generation for Copyright Protectioncited by 24
  12. 2023Leave-one-out Distinguishability in Machine Learningcited by 20
2 more
  1. 2024Generative Unlearning for Any Identitycited by 19
  2. 2024Potion: Towards Poison Unlearningcited by 14

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 5; 2026: 8.

13 papers.

Sort:

By area:

  1. 2026VLA-Forget: Vision-Language-Action Unlearning for Embodied Foundation Modelscited by 10
  2. 2026RAZOR: Ratio-Aware Layer Editing for Targeted Unlearning in Vision Transformers and Diffusion Modelscited by 8
  3. 2025The Right to be Forgotten in Pruning: Unveil Machine Unlearning on Sparse Modelscited by 7
  4. 2025Evaluating the Defense Potential of Machine Unlearning against Membership Inference Attackscited by 2
  5. 2025A Unified Framework for Diffusion Model Unlearning with f-Divergencecited by 2
  6. 2025IMU: Influence-guided Machine Unlearningcited by 1
  7. 2026Erased, but Not Gone: Output Forgetting Is Not True Forgettingcited by 1
  8. 2025Unleashing Uncertainty: Efficient Machine Unlearning for Generative AIcited by 0
  9. 2026Stake the Points: Structure-Faithful Instance Unlearningcited by 0
  10. 2026BARRIER: Bounded Activation Regions for Robust Information Erasurecited by 0
  11. 2026Interference-Aware Multi-Task Unlearningcited by 0
  12. 2026SPACE: Source-free Proxy Anchor Concept Erasure for MLLMscited by 0
1 more
  1. 2026IREU: Identity-Related Encoder-Only Unlearning for Customized Portrait Generationcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Forgetting Similar Samples: Can Machine Unlearning Do it Better?
  2. 2025When to Forget? Complexity Trade-offs in Machine Unlearning
  3. 2024Deep Unlearn: Benchmarking Machine Unlearning for Image Classification
  4. 2023Model Sparsity Can Simplify Machine Unlearning
  5. 2024nabla τ: Gradient-based and Task-Agnostic machine Unlearning
  6. 2025Efficient Machine Unlearning by Model Splitting and Core Sample Selection
  7. 2026Efficient Unlearning through Maximizing Relearning Convergence Delay
  8. 2024Machine Unlearning for Image-to-Image Generative Models