Riemannian stochastic optimization methods avoid strict saddle points | Read Paper on Bytez