 Solution

Writing the log-likelihood as $\ell (\beta )=\log L(\beta )$, the Fisher information matrix is the expected negative Hessian matrixUnder the usual regularity conditions it is equivalently the expected outer product of the score function.