How can I approximate the Hessian matrix in convex optimization?

632 Views Asked by At

I can manually find the exact second order partial derivatives but the computation takes a big too long.

I know that I can find linear approximation to the gradient via $ f(x) \approx f(x_{0}) + \nabla f(x_{0})*(x-x_{0}) $

but how can I approximate the hessian matrix? I feel like I might have to use taylor approximation but I'm not sure