The balancing principle for parameter choice in distance-regularized domain adaptation.
Werner ZellingerNatalia ShepelevaMarius-Constantin DinuHamid Eghbal-zadehHoan Duc NguyenBernhard NesslerSergei V. PereverzyevBernhard Alois MoserPublished in: NeurIPS (2021)
Keyphrases
- domain adaptation
- labeled data
- cross domain
- multiple sources
- semi supervised
- transfer learning
- sentiment classification
- semi supervised learning
- manifold alignment
- document classification
- test data
- target domain
- unlabeled data
- training data
- text classification
- supervised learning
- co training
- text mining
- support vector machine
- active learning
- prior knowledge
- high dimensional
- data sets