Badedit: Backdooring large language models by model editing

Y Li, T Li, K Chen, J Zhang, S Liu, W Wang… - arXiv preprint arXiv …, 2024 - arxiv.org
Mainstream backdoor attack methods typically demand substantial tuning data for
poisoning, limiting their practicality and potentially degrading the overall performance when …

Personalization as a shortcut for few-shot backdoor attack against text-to-image diffusion models

Y Huang, F Juefei-Xu, Q Guo, J Zhang, Y Wu… - Proceedings of the …, 2024 - ojs.aaai.org
Although recent personalization methods have democratized high-resolution image
synthesis by enabling swift concept acquisition with minimal examples and lightweight …

SAME: Sample Reconstruction against Model Extraction Attacks

Y Xie, J Zhang, S Zhao, T Zhang, X Chen - Proceedings of the AAAI …, 2024 - ojs.aaai.org
While deep learning models have shown significant performance across various domains,
their deployment needs extensive resources and advanced computing infrastructure. As a …

Parameter Disparities Dissection for Backdoor Defense in Heterogeneous Federated Learning

W Huang, M Ye, Z Shi, G Wan, H Li… - The Thirty-eighth Annual …, 2024 - openreview.net
Backdoor attacks pose a serious threat to federated systems, where malicious clients
optimize on the triggered distribution to mislead the global model towards a predefined …

Semantic-guided Prompt Organization for Universal Goal Hijacking against LLMs

Y Huang, C Wang, X Jia, Q Guo, F Juefei-Xu… - arXiv preprint arXiv …, 2024 - arxiv.org
With the rising popularity of Large Language Models (LLMs), assessing their trustworthiness
through security tasks has gained critical importance. Regarding the new task of universal …

FedQP: Towards Accurate Federated Learning using Quadratic Programming Guided Mutation

J Weng, Z Xia, R Li, M Hu, M Chen - arXiv preprint arXiv:2411.15847, 2024 - arxiv.org
Due to the advantages of privacy-preserving, Federated Learning (FL) is widely used in
distributed machine learning systems. However, existing FL methods suffer from low …

SampDetox: Black-box Backdoor Defense via Perturbation-based Sample Detoxification

Y Yang, C Jia, DK Yan, M Hu, T Li, X Xie, X Wei… - The Thirty-eighth Annual … - openreview.net
The advancement of Machine Learning has enabled the widespread deployment of
Machine Learning as a Service (MLaaS) applications. However, the untrustworthy nature of …