Imagen de Portada

Secure Transfer Learning: Training Clean Models Against Backdoor in (Both) Pre-trained Encoders and Downstream Datasets

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autores principales:	Zhang, Yechao, Zhou, Yuxuan, Li, Tianyu, Li, Minghui, Hu, Shengshan, Luo, Wei, Zhang, Leo Yu
Formato:	Preprint
Publicado:	2025
Materias:	Machine Learning Cryptography and Security
Acceso en línea:	Acceder al recurso
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

Ejemplares similares

MARS: A Malignity-Aware Backdoor Defense in Federated Learning
por: Wan, Wei, et al.
Publicado: (2025)

DarkFed: A Data-Free Backdoor Attack in Federated Learning
por: Li, Minghui, et al.
Publicado: (2024)

Why Does Little Robustness Help? A Further Step Towards Understanding Adversarial Transferability
por: Zhang, Yechao, et al.
Publicado: (2023)

Spa-VLM: Stealthy Poisoning Attacks on RAG-based VLM
por: Yu, Lei, et al.
Publicado: (2025)

ECLIPSE: Expunging Clean-label Indiscriminate Poisons via Sparse Diffusion Purification
por: Wang, Xianlong, et al.
Publicado: (2024)

Pre-trained Encoder Inference: Revealing Upstream Encoders In Downstream Machine Learning Services
por: Fu, Shaopeng, et al.
Publicado: (2024)

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

Transferable Watermarking to Self-supervised Pre-trained Graph Encoders by Trigger Embeddings
por: Zhao, Xiangyu, et al.
Publicado: (2024)

Low Rank Comes with Low Security: Gradient Assembly Poisoning Attacks against Distributed LoRA-based LLM Systems
por: Dong, Yueyan, et al.
Publicado: (2026)

TED-LaST: Towards Robust Backdoor Defense Against Adaptive Attacks
por: Mo, Xiaoxing, et al.
Publicado: (2025)

Patronus: Identifying and Mitigating Transferable Backdoors in Pre-trained Language Models
por: Zhao, Tianhang, et al.
Publicado: (2025)

Watermarking LLM-Generated Datasets in Downstream Tasks
por: Liu, Yugeng, et al.
Publicado: (2025)

Not All Prompts Are Secure: A Switchable Backdoor Attack Against Pre-trained Vision Transformers
por: Yang, Sheng, et al.
Publicado: (2024)

Mellivora Capensis: A Backdoor-Free Training Framework on the Poisoned Dataset without Auxiliary Data
por: Pu, Yuwen, et al.
Publicado: (2024)

Securely Fine-tuning Pre-trained Encoders Against Adversarial Examples
por: Zhou, Ziqi, et al.
Publicado: (2024)

Fine-tuning is Not Fine: Mitigating Backdoor Attacks in GNNs with Limited Clean Data
por: Zhang, Jiale, et al.
Publicado: (2025)

SSCL-BW: Sample-Specific Clean-Label Backdoor Watermarking for Dataset Ownership Verification
por: Wang, Yingjia, et al.
Publicado: (2025)

Probe-Me-Not: Protecting Pre-trained Encoders from Malicious Probing
por: Ding, Ruyi, et al.
Publicado: (2024)

Privacy Backdoors: Enhancing Membership Inference through Poisoning Pre-trained Models
por: Wen, Yuxin, et al.
Publicado: (2024)

DPDSyn: Improving Differentially Private Dataset Synthesis for Model Training by Downstream Task Guidance
por: Jia, Mingxuan, et al.
Publicado: (2026)

BadTemplate: A Training-Free Backdoor Attack via Chat Template Against Large Language Models
por: Wang, Zihan, et al.
Publicado: (2026)

"MCP Does Not Stand for Misuse Cryptography Protocol": Uncovering Cryptographic Misuse in Model Context Protocol at Scale
por: Yan, Biwei, et al.
Publicado: (2025)

Real is not True: Backdoor Attacks Against Deepfake Detection
por: Sun, Hong, et al.
Publicado: (2024)

UltraClean: A Simple Framework to Train Robust Neural Networks against Backdoor Attacks
por: Zhao, Bingyin, et al.
Publicado: (2023)

Activation Gradient based Poisoned Sample Detection Against Backdoor Attacks
por: Yuan, Danni, et al.
Publicado: (2023)

A Systematic Study of Code Obfuscation Against LLM-based Vulnerability Detection
por: Li, Xiao, et al.
Publicado: (2025)

Instruction Backdoor Attacks Against Customized LLMs
por: Zhang, Rui, et al.
Publicado: (2024)

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

BadMerging: Backdoor Attacks Against Model Merging
por: Zhang, Jinghuai, et al.
Publicado: (2024)

Towards Model Extraction Attacks in GAN-Based Image Translation via Domain Shift Mitigation
por: Mi, Di, et al.
Publicado: (2024)

UnlearnShield: Shielding Forgotten Privacy against Unlearning Inversion
por: Xue, Lulu, et al.
Publicado: (2026)

Poisoning the Pixels: Revisiting Backdoor Attacks on Semantic Segmentation
por: Zhang, Guangsheng, et al.
Publicado: (2026)

When Forgetting Triggers Backdoors: A Clean Unlearning Attack
por: Arazzi, Marco, et al.
Publicado: (2025)

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

VertMark: A Unified Training-Free Robust Watermarking Framework for Vertical Domain Pre-trained Language Models
por: Kong, Cong, et al.
Publicado: (2026)

On the Robustness of Graph Reduction Against GNN Backdoor
por: Zhu, Yuxuan, et al.
Publicado: (2024)

Backdoor Attacks and Defenses in Computer Vision Domain: A Survey
por: Abbasi, Bilal Hussain, et al.
Publicado: (2025)

One-to-Multiple Clean-Label Image Camouflage (OmClic) based Backdoor Attack on Deep Learning
por: Wang, Guohong, et al.
Publicado: (2023)

Transferable Backdoor Attacks for Code Models via Sharpness-Aware Adversarial Perturbation
por: Chang, Shuyu, et al.
Publicado: (2026)

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