Leveraging Domain-Unlabeled Data in Offline Reinforcement Learning across Two Domains.
Soichiro NishimoriXin-Qiang CaiJohannes AckermannMasashi SugiyamaPublished in: CoRR (2024)
Keyphrases
- maximum likelihood
- unlabeled data
- transfer learning
- target domain
- labeled data
- semi supervised learning
- reinforcement learning
- domain adaptation
- cross domain
- semi supervised
- active learning
- labeled examples
- supervised learning
- learning algorithm
- training data
- knowledge transfer
- co training
- semi supervised classification
- labeled training data
- text classification
- learning tasks
- data points
- text categorization
- domain specific
- training examples
- labeled and unlabeled data
- training set
- prior knowledge
- background knowledge
- training samples
- small set of labeled
- machine learning
- data sets
- data analysis
- test data
- pairwise
- learning problems
- learning process
- class labels
- number of labeled examples
- select relevant features
- label propagation
- multiple sources
- machine learning algorithms
- data mining
- real world