Scribd
Upload
102 views2 pages

C3W3 - Quiz - Coursera

Uploaded by

Priyanka
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
102 views2 pages

C3W3 - Quiz - Coursera

Uploaded by

Priyanka
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

1.

Classification allows you to identify similarity between two things while siamese networks allow you to categorize 1 / 1 point
things.
False
True

Correct
Correct.

2. Do the two subnetworks in a siamese network share the same parameters? 1 / 1 point
Yes
No

Correct
Correct.

3. When training a siamese network to identify duplicates, which pairs of questions from the following questions do you 1 / 1 point
expect to have the highest cosine similarity ?

Is learning NLP useful for me to get a job? (ANCHOR)


What should I learn to get a job? (POSITIVE)
Where is the job? (NEGATIVE)

Anchor, Positive
Anchor, Negative
Negative, Positive

Correct
Correct.

4. In the triplet loss function below, will decreasing the hyperparameter alpha from 0.5 to 0.2 require more, or less, 1 / 1 point
optimization during training ?
diff = s(A, N ) − s(A, P )
L(A, P , N ) = max(dif f + α, 0)

Less
More.

Correct
Correct. Alpha is the margin, so the smaller it is the less you have to optimize.

5. The orange square below corresponds to the similarity score of question duplicates? 1 / 1 point

True
False

Correct
Correct. They correspond to non question duplicates.

6. What is the closest negative in this set of numbers assuming a duplicate pair similarity of 0.6? 1 / 1 point
[-0.9,-0.4,0.4, 0.8]
-0.9
-0.4
0.4
0.8

Correct
Correct.

7. In one shot learning, is any retraining required when new classes are added? For example, a new bank customer’s 1 / 1 point
signature.
Yes
No

Correct
Correct.

8. During training, you have to update the weights of each of the subnetworks independently. 1 / 1 point
False.
True.

Correct
Correct. You update the same weight.

9. The mean negative is defined as the closest off-diagonal value to the diagonal in each row (excluding the diagonal). 1 / 1 point
True
False

Correct
Correct.

10. In what order are Siamese networks performed in lecture? 1 / 1 point


1. Convert each input into an array of numbers
2. Feed arrays into your model
3. Compare 𝒗1, 𝒗2 using cosine similarity
4. Test against a threshold
1. Convert each input into an array of numbers
2. Feed arrays into your model
3. Run logistic regression classifier
4. Classify by using the probability
1. Convert each input into an array of numbers
2. Feed arrays into your model
3. Run soft-max classifier for all classes
4. Take the arg-max of the probabilities
1. Convert each input into an array of numbers
2. Feed arrays into your model
3. Compare 𝒗1, 𝒗2 using euclidean distance
4. Test against a threshold

Correct

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy