Quiz's

Question 1

Which of the following is NOT an activation function?

Tanh

Softmax

Cross-entropy

Sigmoid

Answer 1

Cross-entopy

Question 2

Which statement regarding forward and backward propagation is correct?

Question 3

What is the difference between forward and back propogation?

Answer 3

Forward propagation’s purpose is to obtain the model errors, while backward propagation’s is to update the weights.

Explanation:
Forward propagation is when we run the model with some input data. After that, we obtain the output from which, along with the targets, we can determine the errors of our network and the loss function. Meanwhile, back-propagation is when we compute each weight’s contribution to these errors in order to know how to update it.

Question 4

What is one way we can deal with overfitting?

Answer 4

Early stopping

Question 5

What is the best approach to the initialization of weights?

Answer 5

The Xavier initialization