Quiz 5

Milestones

Rubric

All questions on the quizzes in this course are out of 3 points. When assessing these quizzes, score answers as follows:

• 3 is reserved for excellent responses that that fully explain the answer to the question, show a true understanding of the question being asked, and don’t contain any errors. As a general guideline, we expect no more than 25-30% of answers are likely to receive a 3 on any particular question.
• 2 is awarded for a good response that’s getting at the right idea, but might not explain it fully or might have a few minor errors. 2, indeed, is what the course expects most students are likely to earn. As a general guideline, we expect approximately 50-60% of answers are likely to receive a 2 on any particular question.
• 1 is awarded for an inadequate response that misses the mark, is wholly incorrect, or otherwise fails to answer what the question is asking.
• 0 is given to answers that are exceedingly brief, or blank, and as such are effectively non-responses.

The above rubric should be applied for all questions on the quiz.

If you give a score lower than full credit, please include a brief comment in the “Provide comments specific to this submission” section of the form. You can also re-use comments in the “Apply Previously Used Comments” to save time so that you don’t find yourself typing the same thing over and over!

Grading Responsibilities

Answer Key

Question 1

1. 0, as the calculated result is less than activation.
2. 0, as the calculated result is less than activation.
3. 1, since the value is greater than the activation point, the output is 1.
4. 11, since the value is greater than the activation point, we preserve the value.

Question 2

1. Dropout helps to avoid overfitting. By randomly dropping out nodes from the network while training the network, the network learns not to rely too heavy on any one connection.
2. Hidden layers allow the network to learn nonlinear decision boundaries. A single-layer network is only capable of learning a linear separation between classes, which isn’t a good fit (it may “underfit”) for many data sets.
3. Backpropagation allows us to determine how to adjust the weights in hidden layers when training a neural network in such a way that it minimizes error/loss in the network. This makes it possible to learn parameters to maximize the accuracy of a multi-layer neural network.

Question 3

1. 15. Each of the 4 inputs is connected to each of the 3 outputs, for a total of 12 weights. Each of the 3 outputs has a bias, for a total of 15 weights.
2. 44. Each of the 3 inputs is connected to each of the 5 hidden units, for a total of 15 weights — if we include the 5 biases, that becomes 20 weights. That is added to each of the 5 hidden units connected to each of the 4 output units, for a total of 20 weights, plus an additional 4 biases for the output units.

Question 4

1.

2.