all InfoSec news
Defending Against Weight-Poisoning Backdoor Attacks for Parameter-Efficient Fine-Tuning
Feb. 20, 2024, 5:11 a.m. | Shuai Zhao, Leilei Gan, Luu Anh Tuan, Jie Fu, Lingjuan Lyu, Meihuizi Jia, Jinming Wen
cs.CR updates on arXiv.org arxiv.org
Abstract: Recently, various parameter-efficient fine-tuning (PEFT) strategies for application to language models have been proposed and successfully implemented. However, this raises the question of whether PEFT, which only updates a limited set of model parameters, constitutes security vulnerabilities when confronted with weight-poisoning backdoor attacks. In this study, we show that PEFT is more susceptible to weight-poisoning backdoor attacks compared to the full-parameter fine-tuning method, with pre-defined triggers remaining exploitable and pre-defined targets maintaining high confidence, even …
application arxiv attacks backdoor backdoor attacks cs.ai cs.cr defending fine-tuning language language models parameter poisoning question security strategies study updates vulnerabilities
More from arxiv.org / cs.CR updates on arXiv.org
Jobs in InfoSec / Cybersecurity
Intern, Cyber Security Vulnerability Management
@ Grab | Petaling Jaya, Malaysia
Compliance - Global Privacy Office - Associate - Bengaluru
@ Goldman Sachs | Bengaluru, Karnataka, India
Cyber Security Engineer (m/w/d) Operational Technology
@ MAN Energy Solutions | Oberhausen, DE, 46145
Armed Security Officer - Hospital
@ Allied Universal | Sun Valley, CA, United States
Governance, Risk and Compliance Officer (Africa)
@ dLocal | Lagos (Remote)
Junior Cloud DevSecOps Network Engineer
@ Accenture Federal Services | Arlington, VA