Riemannian stochastic optimization methods avoid strict saddle points

Open in new window