BriefGPT.xyz
Dec, 2024
揭示大型语言模型的脆弱性:对抗性诈骗检测与性能分析
Exposing LLM Vulnerabilities: Adversarial Scam Detection and Performance
HTML
PDF
Chen-Wei Chang, Shailik Sarkar, Shutonu Mitra, Qi Zhang, Hossein Salemi...
TL;DR
本研究解决了大型语言模型(LLMs)在诈骗检测任务中对于对抗性诈骗信息的脆弱性问题。通过建立一个包含原始和对抗性诈骗信息的综合数据集,扩展了传统的诈骗检测二元分类为更细化的诈骗类型。研究发现,LLMs在对抗性例子面前表现出高误分类率,并提出了增强模型鲁棒性的策略。
Abstract
Can we trust
Large Language Models
(LLMs) to accurately predict scam? This paper investigates the vulnerabilities of LLMs when facing adversarial scam messages for the task of
Scam Detection
. We addressed this is
→