Revisit Few-shot Intent Classification with PLMs: Direct Fine-tuning vs. Continual Pre-training.
Haode ZhangHaowen LiangLi-Ming ZhanXiao-Ming WuAlbert Y. S. LamPublished in: ACL (Findings) (2023)
Keyphrases
- fine tuning
- training samples
- training set
- fine tuned
- pattern recognition
- classification accuracy
- training phase
- supervised learning
- decision trees
- viable alternative
- automatic classification
- training process
- pattern classification
- classification algorithm
- support vector
- avoid overfitting
- feature extraction
- classification scheme
- text classification
- classification models
- classification systems
- image classification
- labelled data
- classification method
- preprocessing
- multi class
- svm classifier
- benchmark datasets
- training dataset
- feature space
- video sequences
- training data
- machine learning algorithms
- training patterns
- feature selection
- classification rules
- classification performances
- machine learning methods