{"total":13,"items":[{"citing_arxiv_id":"2606.06320","ref_index":45,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Learning What to Forget: Improving LLM Unlearning via Learned Token-Level Importance","primary_cat":"cs.LG","submitted_at":"2026-06-04T15:56:32+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":7.0,"formal_verification":"none","one_line_summary":"ATWU jointly optimizes model parameters and token weights via a linear scorer on hidden states, recovering oracle forget-specific tokens under a separation condition and achieving SOTA forget-retain trade-offs on TOFU and RWKU.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.05946","ref_index":7,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Short paper: Models in the dark -- Rectification and erasure under GDPR in ML supply chains","primary_cat":"cs.LG","submitted_at":"2026-06-04T09:46:32+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":5.0,"formal_verification":"none","one_line_summary":"Survey identifying technical and supply-chain barriers to GDPR data subject rights in ML, with new framing of 'models in the dark' for downstream opacity.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.03291","ref_index":3,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Multilingual Unlearning in LLMs: Transfer, Dynamics, and Reversibility","primary_cat":"cs.CL","submitted_at":"2026-06-02T07:55:52+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"Unlearning in multilingual LLMs suppresses rather than erases knowledge in later layers, with transfer varying by language similarity and reversible via inference-time steering.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2605.20915","ref_index":42,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Calibration vs Decision Making: Revisiting the Reliability Paradox in Unlearned Language Models","primary_cat":"cs.CL","submitted_at":"2026-05-20T08:59:23+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":5.0,"formal_verification":"none","one_line_summary":"Unlearned language models retain low calibration error but show increased shortcut reliance on the TOFU benchmark, extending the reliability paradox to machine unlearning.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2605.12122","ref_index":4,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Disentangled Sparse Representations for Concept-Separated Diffusion Unlearning","primary_cat":"cs.LG","submitted_at":"2026-05-12T13:39:07+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":7.0,"formal_verification":"none","one_line_summary":"SAEParate disentangles sparse representations in diffusion models via contrastive clustering and nonlinear encoding to enable more precise concept unlearning with reduced side effects.","context_count":1,"top_context_role":"background","top_context_polarity":"background","context_text":"Diffusion-based text-to-image (T2I) models have recently achieved remarkable progress in image synthesis, producing highly realistic and diverse visual content [31, 33]. However, these models can also generate undesirable content, including violent or sexually explicit imagery [ 32], as well as content that raises copyright or privacy concerns. This has motivated growing interest in machine unlearning [4], which aims to remove the influence of specific data or concepts from trained models. T2I diffusion unlearning methods can be broadly categorized into two approaches: modifying model parameters [11, 13, 27, 36] or suppressing the generation of target concepts without directly editing the model [9, 25]. Among them, SAeUron [9] employs sparse autoencoders (SAEs) [7] as auxiliary"},{"citing_arxiv_id":"2605.11170","ref_index":174,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Unlearning with Asymmetric Sources: Improved Unlearning-Utility Trade-off with Public Data","primary_cat":"cs.LG","submitted_at":"2026-05-11T19:28:33+00:00","verdict":null,"verdict_confidence":null,"novelty_score":null,"formal_verification":null,"one_line_summary":null,"context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2605.11161","ref_index":19,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Interpretability Can Be Actionable","primary_cat":"cs.LG","submitted_at":"2026-05-11T19:08:21+00:00","verdict":"CONDITIONAL","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"Interpretability research should be judged by actionability—the degree to which its insights support concrete decisions and interventions—rather than explanatory power alone.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2605.07242","ref_index":2,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"MEMOREPAIR: Barrier-First Cascade Repair in Agentic Memory","primary_cat":"cs.AI","submitted_at":"2026-05-08T04:57:29+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":7.0,"formal_verification":"none","one_line_summary":"MemoRepair formalizes the cascade update problem in agentic memory and solves it via a min-cut reduction that eliminates invalidated memory exposure to 0% while recovering 91-94% of valid successors at 57-76% of baseline repair cost.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2605.04713","ref_index":11,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Not Every Subject Should Stay: Machine Unlearning for Noisy Engagement Recognition","primary_cat":"cs.CV","submitted_at":"2026-05-06T10:03:06+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":5.0,"formal_verification":"none","one_line_summary":"Approximate subject-level unlearning recovers 89.3% and 92.5% of oracle performance gains on EngageNet and DAiSEE at roughly one-quarter the retraining cost in K=3 forget-set regimes.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2605.01047","ref_index":6,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"LLM Ghostbusters: Surgical Hallucination Suppression via Adaptive Unlearning","primary_cat":"cs.CR","submitted_at":"2026-05-01T19:20:31+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"Adaptive Unlearning suppresses package hallucinations in code-generating LLMs by 81% while preserving benchmark performance, using model-generated data and no human labels.","context_count":1,"top_context_role":"background","top_context_polarity":"background","context_text":"Models on Their Own Data. InInternational Conference on Learning Representa- tions (ICLR). [5] Lucas Bourtoule, Varun Chandrasekaran, Christopher A. Choquette-Choo, Hen- grui Jia, Adelin Travers, Baiwu Zhang, David Lie, and Nicolas Papernot. 2021. Machine Unlearning. InProceedings of the 42nd IEEE Symposium on Security and Privacy (SP). IEEE, 141-159. [6] Yinzhi Cao and Junfeng Yang. 2015. Towards Making Systems Forget with Machine Unlearning. In2015 IEEE Symposium on Security and Privacy. IEEE, 463-480. doi:10.1109/SP.2015.35 [7] Mark Chen, Jerry Tworek, Heewoo Jun, Qiming Yuan, Henrique Ponde de Oliveira Pinto, Jared Kaplan, Harri Edwards, Yuri Burda, Nicholas Joseph, Greg Brockman, Alex Ray, Raul Puri, Gretchen Krueger, Michael Petrov, Heidy Khlaaf,"},{"citing_arxiv_id":"2604.09544","ref_index":5,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Large Language Models Generate Harmful Content Using a Distinct, Unified Mechanism","primary_cat":"cs.CL","submitted_at":"2026-04-10T17:58:31+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"Harmful generation in LLMs relies on a compact, unified set of weights that alignment compresses and that are distinct from benign capabilities, explaining emergent misalignment.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2509.22483","ref_index":3,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"OFMU: Optimization-Driven Framework for Machine Unlearning","primary_cat":"cs.LG","submitted_at":"2025-09-26T15:31:32+00:00","verdict":null,"verdict_confidence":null,"novelty_score":null,"formal_verification":null,"one_line_summary":null,"context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2501.19202","ref_index":3,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Improving LLM Unlearning Robustness via Random Perturbations","primary_cat":"cs.CL","submitted_at":"2025-01-31T15:12:20+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":7.0,"formal_verification":"none","one_line_summary":"LLM unlearning is reframed as inadvertently installing backdoor triggers on forget-tokens; Random Noise Augmentation is introduced as a defense that improves robustness with theoretical guarantees.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null}],"limit":50,"offset":0}