Discrepancy-Based Active Learning for Domain Adaptation.
Antoine de MathelinFrançois DeheegerMathilde MougeotNicolas VayatisPublished in: ICLR (2022)
Keyphrases
- domain adaptation
- active learning
- labeled data
- semi supervised
- transfer learning
- unlabeled data
- semi supervised learning
- target domain
- cross domain
- multiple sources
- co training
- manifold alignment
- supervised learning
- learning algorithm
- machine learning
- training examples
- training set
- covariate shift
- document classification
- sentiment classification
- text categorization
- unsupervised learning
- knowledge discovery
- learning process
- reinforcement learning
- training data
- class labels
- learning tasks
- concept space
- databases
- data sets