Regularization is used to address overfitting problems as it penalizes the loss function by adding a multiple of an L1 (LASSO) or an L2 (Ridge) norm of your weights vector w.