面向实际应用的声明引导文本后门攻击
Claim-Guided Textual Backdoor Attack for Practical Applications
AI总结：
针对现有后门攻击需在模型分发后操作输入的局限性，提出声明引导后门攻击（CGBA），利用固有文本声明作为触发器，通过声明提取、聚类和定向训练诱导模型在特定声明上异常，同时保持干净数据性能，显著提升实际攻击的可行性与隐蔽性。
摘要：
自然语言处理的最新进展以及大型语言模型的日益广泛使用暴露了新的安全漏洞，例如后门攻击。以往的后门攻击要求在模型分发后进行输入操作以激活后门，这在实际应用中存在局限性。针对这一空白，我们提出了一种新颖的声明引导后门攻击（CGBA），它通过利用固有的文本声明作为触发器，消除了此类操作的需要。CGBA利用声明提取、聚类和定向训练，诱使模型在目标声明上行为异常，同时不影响其在干净数据上的性能。CGBA在多个数据集和模型上展现了其有效性和隐蔽性，显著提升了实际后门攻击的可行性。我们的代码和数据将发布于https://github.com/PaperCGBA/CGBA。
英文摘要：
Recent advances in natural language processing and the increased use of large language models have exposed new security vulnerabilities, such as backdoor attacks. Previous backdoor attacks require input manipulation after model distribution to activate the backdoor, posing limitations in real-world applicability. Addressing this gap, we introduce a novel Claim-Guided Backdoor Attack (CGBA), which eliminates the need for such manipulations by utilizing inherent textual claims as triggers. CGBA leverages claim extraction, clustering, and targeted training to trick models to misbehave on targeted claims without affecting their performance on clean data. CGBA demonstrates its effectiveness and stealthiness across various datasets and models, significantly enhancing the feasibility of practical backdoor attacks. Our code and data will be available at https://github.com/PaperCGBA/CGBA.
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。
