Please tell me if my understanding is right or not! I am stuck on this for long!
First, Let's consider Linear regression. Now, For this we define a cost function without regularization. After obtaining the optimal parameters, we plot the learning curve.
If the learning curve tells us there is a high bias, We increase the features or the degree. And then we again plot the learning curve... If the learning curve is good, Then, we select it
If the learning curve says there is high variance, Then we introduce a regularization term and find the best value of lambda for this model using the lands curve. We choose this model finally. .
Is this right? If not, Please correct me!