Newton's method

Newton's method is an alternative to Gradient Descent. In contrast to GD, which uses the first derivative to approach the optimal model, Newton's method adds the second derivative to converge faster.

Newton's method has the drawback of being more computationally expensive due to the need to find the second derivative.

$w_{j}=w_{j}-a{\frac {\frac {\partial l}{\partial w_{j}}}{\frac {\partial ^{2}l}{\partial w_{j}^{2}}}}$