ORDisCo: Effective and Efficient Usage of Incremental Unlabeled Data for Semi-supervised Continual Learning.
Liyuan WangKuo YangChongxuan LiLanqing HongZhenguo LiJun ZhuPublished in: CoRR (2021)
Keyphrases
- semi supervised
- unlabeled data
- labeled and unlabeled data
- semi supervised learning
- labeled data
- supervised learning
- active learning
- co training
- learning algorithm
- incremental learning
- unsupervised learning
- labeled examples
- semi supervised classification
- positive examples
- transductive learning
- supervised learning algorithms
- domain adaptation
- partially labeled data
- learning problems
- training examples
- manifold regularization
- learning frameworks
- multi view learning
- learning tasks
- class labels
- learning process
- sample selection bias
- small set of labeled
- supervised and semi supervised
- prior knowledge
- pairwise
- label propagation
- labeled training data
- multi view
- data points
- pairwise constraints
- negative examples
- inductive inference
- text classification
- labeling effort
- object recognition
- training data