On Noisy Negative Curvature Descent: Competing with Gradient Descent for Faster Non-convex Optimization

Open in new window