TrojFSP: Trojan Insertion in Few-shot Prompt Tuning

arXiv:2312.10467v3 Announce Type: replace
Abstract: Prompt tuning is one of the most effective solutions to adapting a fixed pre-trained language model (PLM) for various downstream tasks, especially with only a few input samples. However, the security issues, e.g., Trojan attacks, of prompt tuning on a few data samples are not well-studied. Transferring established data poisoning attacks directly to few-shot prompt tuning presents multiple challenges. One significant issue is the \textit{poisoned imbalance issue}, where non-target class samples are added to the …

abstract arxiv attacks cs.lg data few-shot however language language model prompt prompt tuning samples security solutions tasks type