In Ridge Regression, the cost function is J(w) = MSE + λΣ w ᵢ ². If the regularization strength λ is increased substantially, what is the most likely effect on the trained model?

The L2 penalty term λΣ w ᵢ ² discourages large weight magnitudes; as λ increases, the optimizer is pushed to shrink weights closer to zero to keep the penalty term small, at the cost of fitting the training data slightly less closely. This trades reduced variance (less sensitivity to training-data noise, less overfitting) for increased bias (a simpler, more constrained model). Unlike L1 (Lasso) regularization, L2 (Ridge) shrinks weights smoothly toward zero but essentially never sets them to exactly zero, so it does not perform automatic feature selection.