Imagen de Portada

Backdoor Token Unlearning: Exposing and Defending Backdoors in Pretrained Language Models

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autores principales:	Jiang, Peihai, Lyu, Xixiang, Li, Yige, Ma, Jing
Formato:	Preprint
Publicado:	2025
Materias:	Cryptography and Security Artificial Intelligence Computation and Language
Acceso en línea:	Acceder al recurso
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

Ejemplares similares

DUP: Detection-guided Unlearning for Backdoor Purification in Language Models
por: Hu, Man, et al.
Publicado: (2025)

Defending Against Weight-Poisoning Backdoor Attacks for Parameter-Efficient Fine-Tuning
por: Zhao, Shuai, et al.
Publicado: (2024)

Unlearning Backdoor Attacks for LLMs with Weak-to-Strong Knowledge Distillation
por: Zhao, Shuai, et al.
Publicado: (2024)

Instructions as Backdoors: Backdoor Vulnerabilities of Instruction Tuning for Large Language Models
por: Xu, Jiashu, et al.
Publicado: (2023)

UOR: Universal Backdoor Attacks on Pre-trained Language Models
por: Du, Wei, et al.
Publicado: (2023)

Exploring Backdoor Vulnerabilities of Chat Models
por: Hao, Yunzhuo, et al.
Publicado: (2024)

AutoBackdoor: Automating Backdoor Attacks via LLM Agents
por: Li, Yige, et al.
Publicado: (2025)

Backdoor4Good: Benchmarking Beneficial Uses of Backdoors in LLMs
por: Li, Yige, et al.
Publicado: (2026)

Universal Vulnerabilities in Large Language Models: Backdoor Attacks for In-context Learning
por: Zhao, Shuai, et al.
Publicado: (2024)

Stealthy and Persistent Unalignment on Large Language Models via Backdoor Injections
por: Cao, Yuanpu, et al.
Publicado: (2023)

A Survey of Recent Backdoor Attacks and Defenses in Large Language Models
por: Zhao, Shuai, et al.
Publicado: (2024)

Gracefully Filtering Backdoor Samples for Generative Large Language Models without Retraining
por: Wu, Zongru, et al.
Publicado: (2024)

Acquiring Clean Language Models from Backdoor Poisoned Datasets by Downscaling Frequency Space
por: Wu, Zongru, et al.
Publicado: (2024)

AttackVLA: Benchmarking Adversarial and Backdoor Attacks on Vision-Language-Action Models
por: Li, Jiayu, et al.
Publicado: (2025)

Direct Token Optimization: A Self-contained Approach to Large Language Model Unlearning
por: Lee, Hong kyu, et al.
Publicado: (2025)

Concept-Guided Backdoor Attack on Vision Language Models
por: Shen, Haoyu, et al.
Publicado: (2025)

bi-GRPO: Bidirectional Optimization for Jailbreak Backdoor Injection on LLMs
por: Ji, Wence, et al.
Publicado: (2025)

LoRATK: LoRA Once, Backdoor Everywhere in the Share-and-Play Ecosystem
por: Liu, Hongyi, et al.
Publicado: (2024)

BadJudge: Backdoor Vulnerabilities of LLM-as-a-Judge
por: Tong, Terry, et al.
Publicado: (2025)

Claim-Guided Textual Backdoor Attack for Practical Applications
por: Song, Minkyoo, et al.
Publicado: (2024)

A Study of Backdoors in Instruction Fine-tuned Language Models
por: Raghuram, Jayaram, et al.
Publicado: (2024)

Unleashing the Unseen: Harnessing Benign Datasets for Jailbreaking Large Language Models
por: Zhao, Wei, et al.
Publicado: (2024)

Stealthy Backdoor Attacks against LLMs Based on Natural Style Triggers
por: Wei, Jiali, et al.
Publicado: (2026)

Backdoor Attribution: Elucidating and Controlling Backdoor in Language Models
por: Yu, Miao, et al.
Publicado: (2025)

RobustKV: Defending Large Language Models against Jailbreak Attacks via KV Eviction
por: Jiang, Tanqiu, et al.
Publicado: (2024)

Securing Multi-turn Conversational Language Models From Distributed Backdoor Triggers
por: Tong, Terry, et al.
Publicado: (2024)

Watch Out for Your Agents! Investigating Backdoor Threats to LLM-Based Agents
por: Yang, Wenkai, et al.
Publicado: (2024)

Token by Token, Compromised: Backdoor Vulnerabilities in Unified Autoregressive Models
por: Braun, Tobias, et al.
Publicado: (2026)

Unlearn to Relearn Backdoors: Deferred Backdoor Functionality Attacks on Deep Learning Models
por: Shin, Jeongjin, et al.
Publicado: (2024)

Internal Safety Collapse in Frontier Large Language Models
por: Wu, Yutao, et al.
Publicado: (2026)

P2P: A Poison-to-Poison Remedy for Reliable Backdoor Defense in LLMs
por: Zhao, Shuai, et al.
Publicado: (2025)

SynGhost: Invisible and Universal Task-agnostic Backdoor Attack via Syntactic Transfer
por: Cheng, Pengzhou, et al.
Publicado: (2024)

Breaking PEFT Limitations: Leveraging Weak-to-Strong Knowledge Transfer for Backdoor Attacks in LLMs
por: Zhao, Shuai, et al.
Publicado: (2024)

Uncovering and Aligning Anomalous Attention Heads to Defend Against NLP Backdoor Attacks
por: Jin, Haotian, et al.
Publicado: (2025)

Buffer is All You Need: Defending Federated Learning against Backdoor Attacks under Non-iids via Buffering
por: Lyu, Xingyu, et al.
Publicado: (2025)

Backdoors in RLVR: Jailbreak Backdoors in LLMs From Verifiable Reward
por: Guo, Weiyang, et al.
Publicado: (2026)

Mitigating Backdoor Threats to Large Language Models: Advancement and Challenges
por: Liu, Qin, et al.
Publicado: (2024)

Defending against Jailbreak through Early Exit Generation of Large Language Models
por: Zhao, Chongwen, et al.
Publicado: (2024)

Backdoor Defense in Diffusion Models via Spatial Attention Unlearning
por: Jha, Abha, et al.
Publicado: (2025)

Backdoor Attack with Invisible Triggers Based on Model Architecture Modification
por: Ma, Yuan, et al.
Publicado: (2024)