Latent space decomposition into task-specific and domain-specific subspaces for domain adaptation.
Takaya UedaIkuko NishikawaPublished in: IJCNN (2020)
Keyphrases
- domain adaptation
- latent space
- domain specific
- target domain
- low dimensional
- lower dimensional
- transfer learning
- parameter space
- high dimensional data
- high dimensional
- dimensionality reduction
- feature space
- general purpose
- cross domain
- latent variables
- labeled data
- semi supervised
- principal component analysis
- data points
- gaussian process
- multiple sources
- test data
- manifold learning
- generative model
- semi supervised learning
- unlabeled data
- data mining
- reinforcement learning
- training data
- original data
- learning tasks
- learning algorithm
- distance metric
- active learning
- feature vectors
- sentiment classification
- matrix factorization
- multi task
- face recognition
- machine learning