Averaging on the Bures-Wasserstein manifold: dimension-free convergence of gradient descent.
Jason M. AltschulerSinho ChewiPatrik GerberAustin J. StrommePublished in: CoRR (2021)
Keyphrases
- update rule
- lower dimension
- high dimension
- objective function
- cost function
- intrinsic dimension
- convergence rate
- euclidean space
- pointwise
- diffusion maps
- iterative algorithms
- loss function
- manifold learning
- convergence speed
- neural network
- low dimensional
- tangent space
- conjugate gradient
- temporal dimension
- particle swarm optimization
- principal component analysis
- high dimensional
- feature space
- learning algorithm