They are Not Completely Useless: Towards Recycling Transferable Unlabeled Data for Class-Mismatched Semi-Supervised Learning.
Zhuo HuangYing TaiChengjie WangJian YangChen GongPublished in: CoRR (2020)
Keyphrases
- semi supervised learning
- unlabeled data
- labeled data
- class labels
- semi supervised
- co training
- supervised learning
- semi supervised classification
- label information
- positive examples
- labeled and unlabeled data
- active learning
- manifold regularization
- labeled examples
- machine learning
- label propagation
- training data
- unsupervised learning
- labeled samples
- domain adaptation
- semi supervised learning algorithms
- labeled training data
- learning problems
- learning models
- number of labeled examples
- semi supervised learning methods
- text classification
- unlabeled samples
- improve the classification accuracy
- decision trees
- multi instance
- training set
- unlabeled examples
- supervised learning algorithms
- knowledge discovery
- multi view learning
- training examples
- data mining
- metric learning
- abundant unlabeled data