RATT: Leveraging Unlabeled Data to Guarantee Generalization.
Saurabh GargSivaraman BalakrishnanJ. Zico KolterZachary LiptonPublished in: ICML (2021)
Keyphrases
- unlabeled data
- labeled data
- semi supervised learning
- semi supervised
- active learning
- labeled and unlabeled data
- supervised learning
- co training
- training data
- semi supervised classification
- labeled examples
- labeled training data
- data points
- text classification
- training set
- text categorization
- supervised learning algorithms
- learning algorithm
- small set of labeled
- class labels
- positive examples
- prior knowledge
- number of labeled examples
- training examples
- domain adaptation
- unsupervised learning
- transfer learning
- multi view
- label propagation
- natural language processing
- object recognition
- unlabeled instances
- sample selection bias
- semisupervised learning
- feature selection