Robust Assignment of Labels for Active Learning with Sparse and Noisy Annotations.
Daniel KaluzaAndrzej JanuszDominik SlezakPublished in: CoRR (2023)
Keyphrases
- active learning
- label noise
- metadata
- ground truth labels
- noisy environments
- high dimensional
- pairwise
- annotation effort
- training examples
- labeled data
- random sampling
- learning algorithm
- active learning framework
- clustering algorithm
- training data
- sparse data
- class labels
- learning strategies
- missing data
- computationally efficient
- unlabeled data
- semi supervised