Large Language Models Are Better Adversaries: Exploring Generative Clean-Label Backdoor Attacks Against Text Classifiers.
Wencong YouZayd HammoudehDaniel LowdPublished in: EMNLP (Findings) (2023)
Keyphrases
- language model
- text classifiers
- query generation
- text categorization
- text classification
- multi label
- language modeling
- naive bayes
- probabilistic model
- n gram
- retrieval model
- document classification
- generative model
- test collection
- query expansion
- text data
- language modeling framework
- information retrieval
- unsupervised learning
- naive bayes classification
- feature selection
- classification algorithm
- class labels
- semantic features
- knowledge discovery
- learning algorithm
- smoothing methods
- vector space model
- text mining
- bag of words
- knn
- active learning
- naive bayes classifier
- relevance model
- training data
- low dimensional
- visual features