CReST: A Class-Rebalancing Self-Training Framework for Imbalanced Semi-Supervised Learning.
Chen WeiKihyuk SohnClayton MellinaAlan L. YuilleFan YangPublished in: CVPR (2021)
Keyphrases
- semi supervised learning
- semi supervised
- manifold regularization
- semi supervised classification
- unlabeled data
- regularization framework
- co training
- unsupervised learning
- supervised learning
- labeled data
- discriminative models
- multi view learning
- training data
- boosting framework
- semi supervised learning algorithms
- machine learning
- information retrieval
- multiple instance learning
- learning models
- metric learning
- label information
- learning problems
- multi class
- graph based semi supervised learning
- class distribution
- class labels
- training examples
- similarity measure
- feature selection
- opinion detection
- abundant unlabeled data