Newton's Method vs Gradient Descent Method in tacking saddle points in Non-Convex Optimization