No Labels?: No Problem! Experiments with active learning strategies for multi-class classification in imbalanced low-resource settings.
Emiel SteeghGiovanni SilenoPublished in: ICAIL (2023)
Keyphrases
- multi class classification
- binary classification
- active learning strategies
- multi label
- class imbalance
- multi class
- cost sensitive
- support vector
- support vector machine
- learning problems
- binary classifiers
- class distribution
- ensemble methods
- graph cuts
- pairwise
- active learning
- prediction accuracy
- class labels
- generalization error
- supervised learning
- cost sensitive learning
- text categorization
- training set
- lower bound
- selection strategies
- learning tasks
- decision trees
- high dimensionality
- text classification
- image classification
- prior knowledge
- training data