Escaping from saddle points on Riemannian manifolds.
Yue SunNicolas FlammarionMaryam FazelPublished in: CoRR (2019)
Keyphrases
- saddle points
- riemannian manifolds
- euclidean space
- scale space
- critical points
- vector space
- multi class classification
- parameter space
- feature space
- manifold learning
- geometric structure
- geodesic distance
- reproducing kernel hilbert space
- shape analysis
- covariance matrix
- image features
- data points
- high dimensional
- training set