Riemannian stochastic optimization methods avoid strict saddle points.
Ya-Ping HsiehMohammad Reza Karimi JagharghAndreas KrausePanayotis MertikopoulosPublished in: NeurIPS (2023)
Keyphrases
- optimization methods
- saddle points
- stochastic methods
- stage stochastic programs
- scale space
- optimization method
- critical points
- optimization problems
- simulated annealing
- saddle point
- shape analysis
- model selection
- multiscale
- optimization algorithm
- genetic programming
- graphical models
- cost function
- multi objective
- evolutionary algorithm
- search space