What is regularization in machine learning and what do L1 and L2 regularization penalize?

Question

Data Science with Python — Medium

What is regularization in machine learning and what do L1 and L2 regularization penalize?

Accepted Answer

Regularization in machine learning adds a penalty term to prevent overfitting. L1 (Lasso) penalizes absolute coefficients to encourage sparsity, while L2 (Ridge) penalizes squared coefficients to shrink all coefficients.