Identifying and attacking the saddle point problem in high-dimensional non-convex optimization

Open in new window