Imagen de Portada

Robust LLM Unlearning Against Relearning Attacks: The Minor Components in Representations Matter

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autores principales:	Xiao, Zeguan, Xu, Xuanzhe, Chen, Yun, Wang, Yong, Yang, Jian, Hu, Yanqing, Chen, Guanhua
Formato:	Preprint
Publicado:	2026
Materias:	Computation and Language
Acceso en línea:	Acceder al recurso
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

Ejemplares similares

Representation-Guided Parameter-Efficient LLM Unlearning
por: Xiao, Zeguan, et al.
Publicado: (2026)

Modeling LLM Unlearning as an Asymmetric Two-Task Learning Problem
por: Xiao, Zeguan, et al.
Publicado: (2026)

Distract Large Language Models for Automatic Jailbreak Attack
por: Xiao, Zeguan, et al.
Publicado: (2024)

Towards Bridging the Reward-Generation Gap in Direct Alignment Algorithms
por: Xiao, Zeguan, et al.
Publicado: (2025)

MiLoRA: Harnessing Minor Singular Components for Parameter-Efficient LLM Finetuning
por: Wang, Hanqing, et al.
Publicado: (2024)

Enhancing Uncertainty Estimation in LLMs with Expectation of Aggregated Internal Belief
por: Xiao, Zeguan, et al.
Publicado: (2025)

Towards LLM Unlearning Resilient to Relearning Attacks: A Sharpness-Aware Minimization Perspective and Beyond
por: Fan, Chongyu, et al.
Publicado: (2025)

Toward Automated Robustness Evaluation of Mathematical Reasoning
por: Hou, Yutao, et al.
Publicado: (2025)

Meta-Unlearning on Diffusion Models: Preventing Relearning Unlearned Concepts
por: Gao, Hongcheng, et al.
Publicado: (2024)

SeqAR: Jailbreak LLMs with Sequential Auto-Generated Characters
por: Yang, Yan, et al.
Publicado: (2024)

BiasScope: Towards Automated Detection of Bias in LLM-as-a-Judge Evaluation
por: Lai, Peng, et al.
Publicado: (2026)

Beyond the Surface: Enhancing LLM-as-a-Judge Alignment with Human via Internal Representations
por: Lai, Peng, et al.
Publicado: (2025)

Evaluating the Robustness of Multimodal Agents Against Active Environmental Injection Attacks
por: Chen, Yurun, et al.
Publicado: (2025)

Temperature Matters: Enhancing Watermark Robustness Against Paraphrasing Attacks
por: Idrissi, Badr Youbi, et al.
Publicado: (2025)

FinSafetyBench: Evaluating LLM Safety in Real-World Financial Scenarios
por: Hou, Yutao, et al.
Publicado: (2026)

Pi-SQL: Enhancing Text-to-SQL with Fine-Grained Guidance from Pivot Programming Languages
por: chi, Yongdong, et al.
Publicado: (2025)

Defending LLM Watermarking Against Spoofing Attacks with Contrastive Representation Learning
por: An, Li, et al.
Publicado: (2025)

Collapse of Irrelevant Representations (CIR) Ensures Robust and Non-Disruptive LLM Unlearning
por: Sondej, Filip, et al.
Publicado: (2025)

Defending Against Alignment-Breaking Attacks via Robustly Aligned LLM
por: Cao, Bochuan, et al.
Publicado: (2023)

Which Retain Set Matters for LLM Unlearning? A Case Study on Entity Unlearning
por: Chang, Hwan, et al.
Publicado: (2025)

Maximizing Local Entropy Where It Matters: Prefix-Aware Localized LLM Unlearning
por: Zhai, Naixin, et al.
Publicado: (2026)

Robust LLM Unlearning with MUDMAN: Meta-Unlearning with Disruption Masking And Normalization
por: Sondej, Filip, et al.
Publicado: (2025)

InstructDiff: Domain-Adaptive Data Selection via Differential Entropy for Efficient LLM Fine-Tuning
por: Su, Junyou, et al.
Publicado: (2026)

CLUE: Conflict-guided Localization for LLM Unlearning Framework
por: Chen, Hang, et al.
Publicado: (2025)

From Theft to Bomb-Making: The Ripple Effect of Unlearning in Defending Against Jailbreak Attacks
por: Zhang, Zhexin, et al.
Publicado: (2024)

Robustness of Large Language Models Against Adversarial Attacks
por: Tao, Yiyi, et al.
Publicado: (2024)

Improving LLM Unlearning Robustness via Random Perturbations
por: Huu-Tien, Dang, et al.
Publicado: (2025)

PACIT: Unlocking the Power of Examples for Better In-Context Instruction Tuning
por: Xue, Tianci, et al.
Publicado: (2023)

Soft Token Attacks Cannot Reliably Audit Unlearning in Large Language Models
por: Chen, Haokun, et al.
Publicado: (2025)

Fake News in Sheep's Clothing: Robust Fake News Detection Against LLM-Empowered Style Attacks
por: Wu, Jiaying, et al.
Publicado: (2023)

LLM-Agnostic Semantic Representation Attack
por: Lian, Jiawei, et al.
Publicado: (2026)

Unveiling Over-Memorization in Finetuning LLMs for Reasoning Tasks
por: Ruan, Zhiwen, et al.
Publicado: (2025)

LLM Unlearning with LLM Beliefs
por: Li, Kemou, et al.
Publicado: (2025)

FlexLLM: Exploring LLM Customization for Moving Target Defense on Black-Box LLMs Against Jailbreak Attacks
por: Chen, Bocheng, et al.
Publicado: (2024)

Dual-Space Smoothness for Robust and Balanced LLM Unlearning
por: Yan, Han, et al.
Publicado: (2025)

ImPart: Importance-Aware Delta-Sparsification for Improved Model Compression and Merging in LLMs
por: Yang, Yan, et al.
Publicado: (2025)

Unlearning's Blind Spots: Over-Unlearning and Prototypical Relearning Attack
por: Ha, SeungBum, et al.
Publicado: (2025)

SeTAR: Out-of-Distribution Detection with Selective Low-Rank Approximation
por: Li, Yixia, et al.
Publicado: (2024)

Emoji Attack: Enhancing Jailbreak Attacks Against Judge LLM Detection
por: Wei, Zhipeng, et al.
Publicado: (2024)

Rotation Control Unlearning: Quantifying and Controlling Continuous Unlearning for LLM with The Cognitive Rotation Space
por: Zhang, Xiang, et al.
Publicado: (2025)