Anh Viet Doan BDS Sep 21 S4 SYD Machine Learning MTE

BDS Sep 21 S4 SYD Machine Learning MTE
Anh Viet Doan | 02 Mar 2023
Mercer-Mettl | Anh Viet Doan Page 1 / 15

Test Name::
BDS Sep 21 S4 SYD Machine Learning MTE
Overall Status: Completed Detailed Status: Test-taker Completed Test Finish Time: March 02, 2023 12:40:31 PM AEDT
Anh Viet Doan

AV
anh.bs21don043@spjain.org
Last Name: Doan
Date of birth: Feb 3, 2003
Contact No: 0452189276
SP Jain student ID/Campus ID: BS21DON043
Country: Vietnam
Test-Taker ID: - 94288934

Question-Wise Details
Section 1 20 55m 17s 7/10

Section #1 question(s) Time taken Marks Scored
Q.
Question 1 Time taken: 3m 48s Marks Scored: 0.5/0.5
1
Machine learning definition is, “A computer program is said to learn from experience E with respect to some task T and some performance
measure P, if its performance on T, as measured by P, improves with experience E”.
Based on this definition, what is E for a model used by driverless car.

Choose the most appropriate option
Response:
OPTIONS RESPONSE ANSWER
Dataset containing images have boxes with labels around objects
Dataset containing images but not having boxes with labels around objects
Accuracy of identifying objects in images
Precision of identifying objects in images
Recall of identifying objects in images
Both D & E
Q.
Question 2 Time taken: 15s Marks Scored: 0.5/0.5
2
Which of the following is most suitable for classifying a sample into categories: good, bad and neutral?
Response:
Logistic regression
SoftMax regression
Linear regression
Any of the above

Q.
3
Which of the following can you use if you know that your loss function is convex?
Response:
Batch gradient descent
Stochastic gradient descent
Mini-batch gradient descent
All of the above
None of the above
Q.
Question 4 Time taken: 1m 51s Marks Scored: 0/0.5
4
Which of the following is not a supervised learning algorithm?
Response:
Support Vector Machine
Linear Regression
DBSCAN
K-Means
All of the above
Only C and D

Q.
5
Reinforcement Learning System consists of which of the following components
Response:
Agent
Policy
Both A and B
Penalty
Q.
6
Suppose you are using gradient descent the which of the following would you use to find an optional learning rate
Response:
Grid CV search
Randomized CV search
Hyperopt CV search
All of the above

Q.
7
Suppose your model is:
In this model, we vary till we find a model that best fits our data.
This is an example of
Response:
Parametric Learning
Non-parametric learning
Reinforcement learning
Semi-supervised learning
Q.
8
Which of the following models will require the least storage after the model has been learned?
Response:
Decision tree
Instance Based Model
Deep learning Model
Linear regression model

Q.
9
Which of the following are ways to solve linear regression?
Response:
Normal Equation
Mini - Batch Gradient Descent
Stochastic Gradient Descent
All of the above
Q.
10
Which of the following will do go through a group (of samples ) containing less samples than the total dataset before taking the next step (or
before doing the next update)?
Response:
Stochastic gradient descent
Mini-Batch gradient descent
Batch Gradient Ascent

Q.
11
Which of the following are limitations of gradient descent?
Response:
Gradient decent might converge to a local minimum.
Gradient descent might take a long time to cross a plateau
With a fixed number of iterations, gradient decent may never find the global
minimum
All of the above
Q.
12
You are building a linear regression model which cost function would you pick
Response:
Either A or B can be used. It will lead to the same minima

Q.
13
Which of the following use Manhattan norm to perform regularization?
Response:
Lasso regression
Ridge regression
Both A & B
None of the above
Q.
Question 14 Time taken: 51s Marks Scored: 0/0.5
14
In the code sample below, how many rounds of training will happen?
Response:
30
45

Q.
Question 15 Time taken: 1m Marks Scored: 0.5/0.5
15
The vector form of gradient vector is given by:
Where m = number of samples, n = number of features
What will be the shape of gradient vector ∇ if we have 30 features?
Response:
30 x 1
30 x m
m x 30
mx1
1x1

Q.
16
Suppose the following linear regression model
is written as
Then what is the value of
Response:
Q.
17
What library is used to perform vectorized computations in Machine learning?
Response:
NumPy
Pandas
Sklearn
Tensorflow

Q.
18
If you have a very large training set, then will mini - Batch Gradient Descent train faster than Stochastic Gradient Descent?
Response:
Yes
No
Cannot say based on information provided
Q.
19
Suppose your linear regression model is as follows:
If you use gradient descent, then which of the following equations are used to calculate the gradient for the 1st feature
Response:

Q.
Question 20 Time taken: 2m 25s Marks Scored: 0.5/0.5
20
Let’s say for logistic regression:
After applying the sigmoid function which of the following samples, will the model not be confident about predicting
Response:
(7,0)
(0,7)
(7,7)
(0,0)
All of the above
Only A and B
A, B, C but not D

Question-Wise Details
Section 2 1 9m 36s 7/10

Section #2 question(s) Time taken Marks Scored
Q.
Question 1 Time taken: 9m 36s Marks Scored: 7/10
1
Section B : There are 3 Coding Questions ( 10 Marks)
Write your code in the Jupyter notebook named “BDS22 Feb 23 Mid-term Section B.ipynb”
Save the notebook as <Your_name>.ipynb
Upload the notebook to submission point as instructed.
End of Examination
Response:
Anh Viet Doan.ipynb

31.04 KB Click to Download
File Upload Settings for this question:
File Types Allowed All File Types Allowed by Mettl
Max marks: 10
EVALUATOR’S NAME MARKS EVALUATION COMMENTS LAST MODIFIED ON
Aditya Narvekar 7.0 q2 - did not write the function 01:55 PM Mar 2, 2023
using markdown (-1) q3 - should
have written the equation based
on fitting a linear regression
model (-2)

About the Report
This Report is generated electronically on the basis of the inputs received from the assessment takers. This Report including the AI flags that are generated in
case of availing of proctoring services, should not be solely used/relied on for making any business, selection, entrance, or employment-related decisions. Mettl
accepts no liability from the use of or any action taken or refrained from or for any and all business decisions taken as a result of or reliance upon anything,
including, without limitation, information, advice, or AI flags contained in this Report or sources of information used or referred to in this Report.

Anh Viet Doan BDS Sep 21 S4 SYD Machine Learning MTE

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Anh Viet Doan BDS Sep 21 S4 SYD Machine Learning MTE

Uploaded by

Copyright:

Available Formats

BDS Sep 21 S4 SYD Machine Learning MTE

Anh Viet Doan | 02 Mar 2023

Mercer-Mettl | Anh Viet Doan Page 1 / 15

BDS Sep 21 S4 SYD Machine Learning MTE

Anh Viet Doan

Last Name: Doan

Date of birth: Feb 3, 2003

Contact No: 0452189276

SP Jain student ID/Campus ID: BS21DON043

Mercer-Mettl | Anh Viet Doan Page 2 / 15

Section 1 20 55m 17s 7/10

Based on this definition, what is E for a model used by driverless car.

OPTIONS RESPONSE ANSWER

Dataset containing images have boxes with labels around objects

Accuracy of identifying objects in images

Precision of identifying objects in images

Recall of identifying objects in images

OPTIONS RESPONSE ANSWER

Any of the above

Mercer-Mettl | Anh Viet Doan Page 3 / 15

OPTIONS RESPONSE ANSWER

Batch gradient descent

Stochastic gradient descent

Mini-batch gradient descent

All of the above

None of the above

Which of the following is not a supervised learning algorithm?

OPTIONS RESPONSE ANSWER

Support Vector Machine

All of the above

Mercer-Mettl | Anh Viet Doan Page 4 / 15

Reinforcement Learning System consists of which of the following components

OPTIONS RESPONSE ANSWER

OPTIONS RESPONSE ANSWER

All of the above

Mercer-Mettl | Anh Viet Doan Page 5 / 15

Suppose your model is:

OPTIONS RESPONSE ANSWER

OPTIONS RESPONSE ANSWER

Instance Based Model

Deep learning Model

Linear regression model

Mercer-Mettl | Anh Viet Doan Page 6 / 15

Which of the following are ways to solve linear regression?

OPTIONS RESPONSE ANSWER

Mini - Batch Gradient Descent

Stochastic Gradient Descent

Batch gradient descent

All of the above

OPTIONS RESPONSE ANSWER

Batch gradient descent

Stochastic gradient descent

Mini-Batch gradient descent

Batch Gradient Ascent

Mercer-Mettl | Anh Viet Doan Page 7 / 15

Which of the following are limitations of gradient descent?

OPTIONS RESPONSE ANSWER

Gradient decent might converge to a local minimum.

Gradient descent might take a long time to cross a plateau

All of the above

OPTIONS RESPONSE ANSWER

Either A or B can be used. It will lead to the same minima

Mercer-Mettl | Anh Viet Doan Page 8 / 15