Apply your knowledge of Applied Machine Learning to answer the following:

$[0\mathrm{.}5$ mark each$]$

$ln,$ values are shifted and rescaled so that they end up ranging from $0$ to

$$ scaling.

a$.$ Min$-$max scaling.

b$.$ Standardization.

c$.$ Variance shifting.

d$.$ Mean$-$min scaling.

$2.$ What is not true about standardization

$2.$ Standardization is much more affected by outliers, than min$-$max scaling.

b$.$ Standardization does not bound values to a specific range between $0$ and $1.$

c$.$ Standardization subtracts the mean value from the input datapoint, and then it divides by the standard deviation.

d$.$ All of the above.

$3.$ If you are making a prediction model for people age using their photos, you should:

a$.$ randomly split the data into train and test.

use the future to predict the past.

C$.$ split the data into train and test set based on people rather than individual instance.

d$.$ None of the above.

$4.$ Selecting your model to get optimal test performance is risky because

a$.$ What works best for the particular test set might not actually be the best ir general.

b$.$ Overfitting to the test data.

c$.$ a and b$.$

d$.$ None of the above.

$5.$ A neural network can be used as:

a$.$ Regression model.

b$.$ Binary classifier.

c$.$ Multi$-$class classification.

$($d$)$ All of the above.

$6.$ The process of computing the gradient and updating neural network weights is

a$.$ Forward$-$propagation.

b$.$ Backpropagation.

c$.$ Feed$-$forwarding.

d$.$ Weights training.