Points: 1 Consider a simple setting in which we are predicting the height of a person in centimeters based on their weight. Suppose we include the weight measured in kilograms (kg) and milligrams (mg) as two separate features and we tune the coefficient of the L1 regularization to include only one feature. Without normalizing the data before training, which feature would be selected after the model is trained? Weight in mg Weight in kg Saved Q9.2 Points: 1 Alice is training a model and her training error is rapidly decreasing but her validation error is increasing. What should she do? O Increase regularization. O Decrease regularization. No additional regularization changes are needed. Saved Q9.3 Points: 1 Suppose Alice finds that her model is overfitting and she decides to add L2 regularization with regularization coefficient > to her model. As she increases the regularization coefficient 1, which of the following are true? Bias increases O Variance increases Bias decreases Variance decreases Saved SAMSUNG