Negative curvature obstructs acceleration for strongly geodesically convex optimization, even with exact first-order oracles.
Christopher CriscitielloNicolas BoumalPublished in: COLT (2022)
Keyphrases
- convex optimization
- interior point methods
- convex relaxation
- higher order
- convex optimization problems
- first order logic
- primal dual
- total variation
- multiscale
- convex constraints
- norm minimization
- low rank
- scale space
- active learning
- convex formulation
- semidefinite program
- multiresolution
- video sequences
- optimal solution
- image segmentation