Difference between gradient and Jacobian in gradient descent

1.5k Views Asked by At

What is the difference between the computation of gradient (the partial derivative of error w.r.t. weight) in gradient descent and the computation of the Jacobian in Levenberg-Marquardt algorithm?

1

There are 1 best solutions below

6
On

The "Jacobian" in this context is the Jacobian matrix of the gradient, which is the Hessian of the underlying objective function. It is therefore quite a bit more expensive to compute than the gradient (because you need $O(n^2)$ numerical second derivatives instead of $O(n)$ numerical first derivatives).