Feed-Forward Source-Free Latent Domain Adaptation via Cross-Attention.
Ondrej BohdalDa LiShell Xu HuTimothy M. HospedalesPublished in: CoRR (2022)
Keyphrases
- feed forward
- domain adaptation
- multiple sources
- target domain
- source domain
- back propagation
- concept space
- cross domain
- artificial neural networks
- transfer learning
- neural nets
- neural network
- labeled data
- training data
- sentiment classification
- source and target domains
- semi supervised learning
- test data
- unlabeled data
- semi supervised
- document classification
- latent variables
- training samples
- covariate shift
- training examples
- text categorization
- domain specific
- co training
- information extraction
- data sources
- feature space
- data mining