arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2409.18169cs.CRcs.AIcs.LG
收藏

大型语言模型有害微调攻击与防御:综述

Harmful Fine-tuning Attacks and Defenses for Large Language Models: A Survey

School of Computer Science, Georgia Institute of Technology(佐治亚理工学院计算机科学系)

Tiansheng Huang, Sihao Hu, Fatih Ilhan, Selim Furkan Tekin, Ling Liu

AI总结:

本文综述了大型语言模型中有害微调攻击的威胁模型、防御设计及评估方法,分析了现有攻击与防御机制,并提出了未来研究方向。

摘要:

近期研究表明,新兴的微调即服务商业模式暴露了严重的安全问题:用户上传少量有害数据进行微调即可破坏模型的安全对齐。有害微调攻击已引起学术界和工业界广泛关注。本文首先系统地制定了有害微调的威胁模型和基本假设,然后从攻击设置、防御设计和评估方法三个基本视角全面回顾了有害微调。首先,我们提出了该问题的威胁模型,介绍了有害微调攻击及其变种。接着,我们系统地回顾了现有文献中的代表性攻击、防御方法及不良影响的机械分析。最后,我们介绍了评估方法并概述了未来研究方向,这些可以作为该领域未来发展的指南和关键视角。我们还维护了一个精选的论文列表,可在https://github.com/git-disl/awesome_LLM-harmful-fine-tuning-papers获取。

英文摘要:

Recent research demonstrates that the nascent fine-tuning-as-a-service business model exposes serious safety concerns: fine-tuning with a few harmful data uploaded from the users can compromise the safety alignment of the model. The attack, known as harmful fine-tuning attack, has generated broad research interests in both academia and industry. In this paper, we first systematically formulate the threat model and basic assumptions of harmful fine-tuning. Then, we provide a comprehensive review of harmful fine-tuning from three fundamental perspectives: attack setting, defense design, and evaluation methodology. First, we present the threat model of the problem and introduce the harmful fine-tuning attack and its variants. Next, we systematically survey representative attacks, defense methods, and mechanical analysis of adverse effects in the existing literature. Finally, we introduce the evaluation methodology and outline future research directions, which can serve as guidelines and crucial perspectives for the future development of the subject. We also maintain a curated list of relevant papers, which are made accessible at https://github.com/git-disl/awesome_LLM-harmful-fine-tuning-papers

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑