Out-Of-Domain Unlabeled Data Improves Generalization.
Seyed Amir Hossein SaberiAmir NajafiAlireza HeidariMohammad Hosein MovasaghiniaAbolfazl S. MotahariBabak H. KhalajPublished in: ICLR (2024)
Keyphrases
- unlabeled data
- labeled data
- semi supervised learning
- semi supervised
- domain adaptation
- active learning
- co training
- training data
- supervised learning
- transfer learning
- labeled examples
- semi supervised classification
- text classification
- data points
- text categorization
- target domain
- labeled and unlabeled data
- learning algorithm
- training set
- labeled training data
- cross domain
- number of labeled examples
- class labels
- unsupervised learning
- semisupervised learning
- small set of labeled
- supervised learning algorithms
- label propagation
- labeled data for training
- decision boundary
- positive examples
- domain specific
- learning problems
- domain experts
- multi view learning
- feature selection
- supervised and semi supervised
- information retrieval
- machine learning