Theoretical Analysis of Self-Training with Deep Networks on Unlabeled Data.
Colin WeiKendrick ShenYining ChenTengyu MaPublished in: CoRR (2020)
Keyphrases
- unlabeled data
- semi supervised learning
- theoretical analysis
- co training
- semi supervised
- labeled data
- semi supervised classification
- training set
- active learning
- supervised learning
- labeled training data
- text classification
- training data
- small set of labeled
- labeled examples
- labeled and unlabeled data
- abundant unlabeled data
- learning algorithm
- machine learning
- transfer learning
- data points
- multi view learning
- text categorization
- number of labeled examples
- unsupervised learning
- class labels
- single view
- training examples
- positive examples
- class distribution
- cost sensitive
- complex networks
- multi view
- pairwise
- information retrieval
- learning tasks
- pairwise constraints
- label propagation
- classification accuracy
- supervised learning algorithms
- transductive learning
- support vector
- feature extraction