Hello, Here is the formula of the kernelized ridge regression, shouldn't there be the lambda term acting as penalizer?
I think that the formula is correct. If you expand the left term you find 1/2 lambda a.T @ a which is the regularization term
I am sorry I didn't understand your answer, is there another explanation because I checked and the formula always mentions an extra lambda.
The regularization term \( \lambda \Vert \alpha\Vert^2\) is hidden in (and is exactly) the term \(\alpha^\top \lambda I_N \alpha\). However the matrix \(X\) needs to be full rank for this problem to be equivalent to ridge regression.
Could I ask how to get this formula? I am sorry that I cannot deduce this.
Possible error in pb11 exam2018
Hello,
Here is the formula of the kernelized ridge regression, shouldn't there be the lambda term acting as penalizer?
I think that the formula is correct. If you expand the left term you find 1/2 lambda a.T @ a which is the regularization term
1
I am sorry I didn't understand your answer, is there another explanation because I checked and the formula always mentions an extra lambda.
The regularization term \( \lambda \Vert \alpha\Vert^2\) is hidden in (and is exactly) the term \(\alpha^\top \lambda I_N \alpha\). However the matrix \(X\) needs to be full rank for this problem to be equivalent to ridge regression.
Could I ask how to get this formula? I am sorry that I cannot deduce this.
Add comment