Ridge and LASSO regression are meant to improve classical linear regression for large number of predictors by adding a penalty term. But are they per definition a linear model such as polynomial regression and of course classical linear regression?