When can unlabeled data improve the learning rate?
Christina GöpfertShai Ben-DavidOlivier BousquetSylvain GellyIlya O. TolstikhinRuth UrnerPublished in: COLT (2019)
Keyphrases
- unlabeled data
- learning rate
- labeled data
- learning algorithm
- semi supervised learning
- semi supervised
- active learning
- co training
- improve the classification accuracy
- training data
- supervised learning
- semi supervised classification
- text classification
- unlabeled training data
- labeled and unlabeled data
- labeled examples
- text categorization
- data points
- training set
- domain adaptation
- unsupervised learning
- training examples
- convergence rate
- prior knowledge
- class labels
- labeled training data
- machine learning
- number of labeled examples
- information retrieval
- genetic algorithm
- small number
- linear combination
- multi view