Something for (almost) nothing: Improving deep ensemble calibration using unlabeled data.
Konstantinos PitasJulyan ArbelPublished in: CoRR (2023)
Keyphrases
- unlabeled data
- labeled data
- ensemble learning
- semi supervised learning
- semi supervised
- training data
- training set
- learning algorithm
- active learning
- co training
- semi supervised classification
- supervised learning
- labeled and unlabeled data
- text classification
- labeled examples
- labeled training data
- data points
- text categorization
- supervised learning algorithms
- training examples
- class distribution
- class labels
- transfer learning
- number of labeled examples
- domain adaptation
- positive examples
- small set of labeled
- unsupervised learning
- decision trees
- machine learning
- prior knowledge
- neural network
- base classifiers
- ensemble methods
- naive bayes
- label propagation
- pairwise
- supervised and semi supervised