- most common type of regularization - for logistic [[Logistic Regression]]: - $w \in \mathbb{R}^{n_x}, b \in \mathbb{R}$ - $J(w, b) = \frac{1}{m}\sum_{i=1}^{m} L(\hat{y}^{(i)}, y^{(i)}) + \frac{\lambda}{2m} ||w||^2_2$ - $||w||^2_2 = \sum_{j=1}^{n_x} = w^Tw$