Liberating Seen Classes: Boosting Few-Shot and Zero-Shot Text Classification via Anchor Generation and Classification Reframing.
Han LiuSiyang ZhaoXiaotong ZhangFeng ZhangWei WangFenglong MaHongyang ChenHong YuXianchao ZhangPublished in: AAAI (2024)
Keyphrases
- text classification
- feature selection
- document classification
- machine learning
- class labels
- text categorization
- n gram
- labeled data
- classification accuracy
- text mining
- text classifiers
- decision trees
- naive bayes
- knn
- positive and unlabeled examples
- supervised learning
- weak learners
- data cleaning
- learning algorithm
- base classifiers
- classification algorithm
- feature set
- feature space
- unlabeled data
- ensemble learning
- text classification tasks
- shot detection
- knowledge discovery
- binary classification problems
- feature vectors
- weak classifiers
- classification systems
- support vector
- feature extraction
- ensemble methods
- cost sensitive
- key frames
- data mining
- video sequences
- action recognition