chapter 6

Question 1

what is the meaning of regression model?

Answer 1

the model that predicts a real-valued number for something.

Question 2

Best combination between loss function and activation units.

Answer 2

1- logistic unit + cross-entropy&raquo_space;»» Binary classification.
2-softmax unit + categorical cross-entropy&raquo_space;»» Multiclassification.
3-linear unit + MSE&raquo_space;»»Regression.

Question 3

The softmax unit

Answer 3

is a generalization of the logistic sigmoid function but extended to multiple outputs.

Question 4

what is the meaning of classification model?

Answer 4

The model that measures the probability of a particular class being present and distinguishing it from other classes.

Question 5

The linear output unit

Answer 5

used to predict a value not limited to the range 0 to 1, and uses the identity function as an activation function.

Question 6

Regularization techniques

Answer 6

is a technique that aims at reducing the gap between training error and test error.

Question 7

Weight decay

Answer 7

one of the regularization techniques that implemented by adding a penalty term to the loss function.

Question 8

Dropout

Answer 8

one of the regularization techniques that implemented by randomly removing a subset of the neurons from the network during training.