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