Domain Adaptation Using Riemannian Geometry of Spd Matrices.
Gal MamanOr YairDanny EytanRonen TalmonPublished in: ICASSP (2019)
Keyphrases
- domain adaptation
- symmetric positive definite
- riemannian manifolds
- geodesic distance
- lie group
- log euclidean
- geometric structure
- reproducing kernel hilbert space
- euclidean space
- distribution function
- cross domain
- multiple sources
- positive definite
- diffusion tensor
- semi supervised learning
- labeled data
- sentiment classification
- transfer learning
- target domain
- semi supervised
- euclidean distance
- riemannian metric
- document classification
- vector space
- test data
- tensor field
- affine invariant
- manifold learning
- shape analysis
- covariance matrices
- kernel matrix
- data sets
- parameter space
- shape space
- covariance matrix
- kernel function
- unlabeled data
- supervised learning
- reinforcement learning
- feature selection