Scribd
Upload
325 views

Module 2 Quiz: Email

The document is a quiz on machine learning concepts with 10 multiple choice questions. It covers topics like generative adversarial networks, advantages and disadvantages of deep learning, machine learning loss functions like accuracy and precision, metrics like false negatives and positives, crowdsourcing to generate training data, overfitting, the role of test data, and the importance of feature engineering. The quiz results show the student scored full marks on all 10 questions.

Uploaded by

Nguyễn Quốc Mạnh
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
325 views

Module 2 Quiz: Email

The document is a quiz on machine learning concepts with 10 multiple choice questions. It covers topics like generative adversarial networks, advantages and disadvantages of deep learning, machine learning loss functions like accuracy and precision, metrics like false negatives and positives, crowdsourcing to generate training data, overfitting, the role of test data, and the importance of feature engineering. The quiz results show the student scored full marks on all 10 questions.

Uploaded by

Nguyễn Quốc Mạnh
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 3

Module 2 Quiz

Tổng điểm19/19
Total points 10
Graded Quiz • 30 min

Email *
nguyenquocmanh46k29.1@gmail.com

 
1.Generative adversarial networks:? *
1/1

Are designed to effectively respond to cybersecurity attacks


Use two different networks in tandem to create realistic but synthetic content (e.g. artificial
images)
 
Help to deal with explainability concerns in neural networks
Require less training data than other deep learning networks

 
2.Which of the following is NOT an advantage of deep learning networks:? *
1/1

They eliminate some of the “feature engineering” tasks required in a typical machine learning
workflow
All choices are advantages of deep learning networks
With increasing amounts of data, they become increasingly accurate
They run much faster than non-deep learning machine learning approaches
 
 
3. Which of the following is not a “loss function” used to evaluate the performance of
machine learning algorithms? *
1/1

Recall
Accuracy
Precision
Validity
 
 
4.Which of the following machine learning loss functions refers to the fraction of
labels that the algorithm correctly predicts: *
1/1

Validity
Precision
Accuracy
 
Recall

 
5.Which of the following machine learning loss functions measures how many
instances identified as “positive” by the algorithm are correctly identified? *
1/1

Validity
Precision
 
Recall
Accuracy

 
6.Consider a model that predicts whether a patient has a condition and should receive
a particular treatment (corresponding to a positive recommendation by the algorithm).
Which of the following performance metrics would you want to minimize for a context
such as this assuming there are high costs for a sick patient missing the treatment
but fairly low costs of a well patient receiving the treatment? *
1/1

False negatives
 
False positives
Accuracy
No choices are correct

 
7. Sending a data set without the correct predicted answers (unlabeled data) to a
crowdsourcing platform for gig workers to add the correct answers (labels) is an
example of: *
1/1

A way to generate training data that a machine learning algorithm can use to learn
 
A way to cross-validate algorithmic performance
A way to extract relevant “features” from a data set that a machine learning algorithm can use to
learn
A way to evaluate machine learning performance

 
8. Which of the following best describes “overfitting”? *
10/10

No choices are correct


A model that fits too closely with the training data, but fails to perform well on predicting
observations that are not from the training data
 
Applying a prediction model to a set of data which has features that do not match that of the
training data
Training a neural network past the point at which the cost function is minimized

 
9.What is the role of test data in a machine learning workflow? Select all that apply  *
1/1

None of the choices are correct


It provides a data set that a machine learning algorithm can use to learn algorithm parameters
Both choices are correct
It provides a data set that the model did not use to learn, on which to evaluate model
performance
 
 
10. Which of the following is true about “feature engineering”, the process used to
create measures from the raw data that a machine learning model uses to make
predictions?
1/1

It is the final step in a machine learning workflow


It is less important when using text-based data
It is often difficult to create features effectively without having some expertise in the relevant
domain from which the data is drawn
 
It is more important for deep learning than for simpler techniques

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