Copertina

Backdoor in Seconds: Unlocking Vulnerabilities in Large Pre-trained Models via Model Editing

Fuente: arXiv

Salvato in:

Dettagli Bibliografici
Autori principali:	Guo, Dongliang, Hu, Mengxuan, Guan, Zihan, Guo, Junfeng, Hartvigsen, Thomas, Li, Sheng
Natura:	Preprint
Pubblicazione:	2024
Soggetti:	Artificial Intelligence
Accesso online:	Acceder al recurso
Tags:	Aggiungi Tag Nessun Tag, puoi essere il primo ad aggiungerne!!

Documenti analoghi

BalancEdit: Dynamically Balancing the Generality-Locality Trade-off in Multi-modal Model Editing
di: Guo, Dongliang, et al.
Pubblicazione: (2025)

Enhanced Diagnostic Performance via Large-Resolution Inference Optimization for Pathology Foundation Models
di: Hu, Mengxuan, et al.
Pubblicazione: (2026)

MEGen: Generative Backdoor into Large Language Models via Model Editing
di: Qiu, Jiyang, et al.
Pubblicazione: (2024)

Text2Seg: Remote Sensing Image Semantic Segmentation via Text-Guided Visual Foundation Models
di: Zhang, Jielu, et al.
Pubblicazione: (2023)

Efficient Knowledge Editing via Minimal Precomputation
di: Gupta, Akshat, et al.
Pubblicazione: (2025)

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

Large Language Models for Causal Discovery: Current Landscape and Future Directions
di: Wan, Guangya, et al.
Pubblicazione: (2024)

UFID: A Unified Framework for Input-level Backdoor Detection on Diffusion Models
di: Guan, Zihan, et al.
Pubblicazione: (2024)

Are LLMs Ready for Neural-integrated Mechanistic Modeling? A Benchmark and Agentic Framework
di: Guan, Zihan, et al.
Pubblicazione: (2026)

Unlocking Emergent Modularity in Large Language Models
di: Qiu, Zihan, et al.
Pubblicazione: (2023)

Leveraging Pre-trained Large Language Models with Refined Prompting for Online Task and Motion Planning
di: Guo, Huihui, et al.
Pubblicazione: (2025)

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

Backdoor Samples Detection Based on Perturbation Discrepancy Consistency in Pre-trained Language Models
di: Peng, Zuquan, et al.
Pubblicazione: (2025)

ReasonEdit: Editing Vision-Language Models using Human Reasoning
di: Qiu, Jiaxing, et al.
Pubblicazione: (2026)

Img2Loc: Revisiting Image Geolocalization using Multi-modality Foundation Models and Image-based Retrieval-Augmented Generation
di: Zhou, Zhongliang, et al.
Pubblicazione: (2024)

Revisiting Pre-trained Language Models for Vulnerability Detection
di: Li, Youpeng, et al.
Pubblicazione: (2025)

Exploiting the Vulnerability of Large Language Models via Defense-Aware Architectural Backdoor
di: Miah, Abdullah Arafat, et al.
Pubblicazione: (2024)

Alignment-Weighted DPO: A principled reasoning approach to improve safety alignment
di: Hu, Mengxuan, et al.
Pubblicazione: (2026)

FGBERT: Function-Driven Pre-trained Gene Language Model for Metagenomics
di: Duan, ChenRui, et al.
Pubblicazione: (2024)

Data Darwinism Part I: Unlocking the Value of Scientific Data for Pre-training
di: Qin, Yiwei, et al.
Pubblicazione: (2026)

Pre-train and Fine-tune: Recommenders as Large Models
di: Jiang, Zhenhao, et al.
Pubblicazione: (2025)

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

Selection-Based Vulnerabilities: Clean-Label Backdoor Attacks in Active Learning
di: Zhi, Yuhan, et al.
Pubblicazione: (2025)

How to Set the Learning Rate for Large-Scale Pre-training?
di: Zhou, Yunhua, et al.
Pubblicazione: (2026)

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

Protecting Copyright of Medical Pre-trained Language Models: Training-Free Backdoor Model Watermarking
di: Kong, Cong, et al.
Pubblicazione: (2024)

CBV: Clean-label Backdoor Attacks on Vision Language Models via Diffusion Models
di: Guo, Ji, et al.
Pubblicazione: (2026)

Your Compiler is Backdooring Your Model: Understanding and Exploiting Compilation Inconsistency Vulnerabilities in Deep Learning Compilers
di: Chen, Simin, et al.
Pubblicazione: (2025)

Robust Knowledge Distillation Based on Feature Variance Against Backdoored Teacher Model
di: Chen, Jinyin, et al.
Pubblicazione: (2024)

Synthesize-on-Graph: Knowledgeable Synthetic Data Generation for Continue Pre-training of Large Language Models
di: Ma, Shengjie, et al.
Pubblicazione: (2025)

How to Set the Batch Size for Large-Scale Pre-training?
di: Zhou, Yunhua, et al.
Pubblicazione: (2026)

PersGuard: Preventing Malicious Personalization via Backdoor Attacks on Pre-trained Text-to-Image Diffusion Models
di: Liu, Xinwei, et al.
Pubblicazione: (2025)

A Unified Understanding of Adversarial Vulnerability Regarding Unimodal Models and Vision-Language Pre-training Models
di: Zheng, Haonan, et al.
Pubblicazione: (2024)

Unlocking the Power of Large Language Models for Entity Alignment
di: Jiang, Xuhui, et al.
Pubblicazione: (2024)

Effective Backdoor Mitigation in Vision-Language Models Depends on the Pre-training Objective
di: Verma, Sahil, et al.
Pubblicazione: (2023)

A Vision-Language Pre-training Model-Guided Approach for Mitigating Backdoor Attacks in Federated Learning
di: Gai, Keke, et al.
Pubblicazione: (2025)

Mutual Information Guided Backdoor Mitigation for Pre-trained Encoders
di: Han, Tingxu, et al.
Pubblicazione: (2024)

Prompt as Triggers for Backdoor Attack: Examining the Vulnerability in Language Models
di: Zhao, Shuai, et al.
Pubblicazione: (2023)

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

Unlocking the Power of Spatial and Temporal Information in Medical Multimodal Pre-training
di: Yang, Jinxia, et al.
Pubblicazione: (2024)