They are Not Completely Useless: Towards Recycling Transferable Unlabeled Data for Class-Mismatched Semi-Supervised Learning.
Zhuo HuangJian YangChen GongPublished in: IEEE Trans. Multim. (2023)
Keyphrases
- semi supervised learning
- unlabeled data
- labeled data
- class labels
- semi supervised
- co training
- label information
- semi supervised classification
- labeled examples
- supervised learning
- training data
- labeled and unlabeled data
- labeled samples
- label propagation
- positive examples
- labeled training data
- active learning
- semi supervised learning methods
- training examples
- unsupervised learning
- manifold regularization
- learning problems
- transfer learning
- unlabeled examples
- number of labeled examples
- improve the classification accuracy
- machine learning
- learning algorithm
- text classification
- semi supervised learning algorithms
- unlabeled training data
- supervised learning algorithms
- regularization framework
- data points
- multi view learning
- decision trees
- metric learning
- learning models
- unlabeled samples
- pairwise
- sample selection bias
- data mining
- abundant unlabeled data