Prompt as Triggers for Backdoor Attack: Examining the Vulnerability in Language Models.
Shuai ZhaoJinming WenAnh Tuan LuuJunbo ZhaoJie FuPublished in: EMNLP (2023)
Keyphrases
- language model
- attack graphs
- language modeling
- buffer overflow
- penetration testing
- document retrieval
- n gram
- security vulnerabilities
- attack graph
- information retrieval
- probabilistic model
- retrieval model
- speech recognition
- query expansion
- test collection
- language modelling
- statistical language models
- context sensitive
- query terms
- ad hoc information retrieval
- word error rate
- query specific
- smoothing methods
- vector space model
- passage retrieval
- language model for information retrieval
- abstract argumentation
- term dependencies
- dos attacks