Flowing on Riemannian Manifold: Domain Adaptation by Shifting Covariance.
Zhen CuiWen LiDong XuShiguang ShanXilin ChenXuelong LiPublished in: IEEE Trans. Cybern. (2014)
Keyphrases
- domain adaptation
- riemannian manifolds
- reproducing kernel hilbert space
- covariance matrices
- covariance matrix
- euclidean space
- multiple sources
- vector space
- labeled data
- sentiment classification
- target domain
- transfer learning
- cross domain
- semi supervised learning
- semi supervised
- maximum likelihood
- manifold learning
- test data
- principal component analysis
- document classification
- geometric structure
- parameter space
- data sets
- distance measure
- pairwise
- feature space
- unlabeled data
- gaussian distribution
- euclidean distance
- co training
- geodesic distance
- domain specific