A Unified Formulation for the Bures-Wasserstein and Log-Euclidean/Log-Hilbert-Schmidt Distances between Positive Definite Operators.
Hà Quang MinhPublished in: GSI (2019)
Keyphrases
- positive definite
- diffusion tensor
- covariance matrix
- geometric structure
- reproducing kernel hilbert space
- distance measure
- kernel function
- euclidean space
- anisotropic diffusion
- riemannian manifolds
- euclidean distance
- magnetic resonance images
- white matter
- distance function
- learning theory
- kernel methods
- similarity measure
- image processing
- human brain
- data points
- domain adaptation
- special case
- lower bound
- support vector
- kernel machines
- semi parametric