Total No.
of printod pagoA = :1
CSE 1817 OE 21
Roll No. of candidate / I ) � )" )-r / ]
( 0 / 0/ 0 /-? fo] lt�
2021
B.Tech. 7th Semester End-Term Examination
CSE
MACHINE LEARNING
(New Regulation w.e.f 2017-18) &
(New Syllabus w.e.f 2018-19)
B.Tech Papers
Full Marks - 70 Time - Three hours
. .
The figures in the margin indicate full marks for the questions.
Answer Question No� 1 and any four from the rest.
1. Answer the following: (MCQ/Fill in the blanks) (10 X 1 = 10)
(i) Real-Time decisions;- AI game, learning tasks, skill acquisition, and robot
navigation are applications of which of the following:
(a) Superyised Learning: Classification
Qv,..,- Reipforcement Learn'ing
(c) Unsupervised Learning: Clustering
(d) Unsupervised Learning: Regression
(ii) What is perceptron?
A single layer feed-forward neural network with pre-processing
�
{b) Neural network that contains feedback
(c) A double layer auto-associative neural network
(d) An auto-associative neural network
(iii) The Common classes of problems in machine learning is
(a) Classification (b) Clustering
(c) Regression � All of these
[Turn over
(iv) Muchino lonrning algorithms build O model l,ascd on sample data known as
-;fa-)' Training dat{i (b) Transfer data
(c) Validation data (d) Test data
(v) When performing regression or classification, which of the following is the
correct way to preprocess the data
(a) Normalize the data ➔ PCA ➔ training
(b) PCA ➔ normalize PCA output ➔ training
(c) Normalize the data ➔ PCA ➔ normalize PCA output ➔ training
(d) None of the above
(vi) Which of the following is a disadvantage of decision trees?
(a) Factor analysis
(b) Decisiot) trees are robust to outliers
B.Tech Papers
� Decision trees are prone to be overfit
(d) None of the above
(vii) Which of the following methods do we use to best fit the data in Logistic
Regression?
(a) Least Square Error _(b) Maximum Likelihood .
(c) · J�ccard distance . (d)· . Both (a) and (b)
(viii).lf I am_ using all features of my dataset and i achieve 100% accuracy on my
· training set, but 70% on validation set, what should' I look out for? ·
(a) Underfitting � Nothing, the model is perfect
� Overfitting , Non� of these
(d)
(ix) Choose the correct statement about the CART model:
(a) CART follows unsupervised learning techniques
(b) CART follows a supervised learning techniques
(c) CART adopt a greedy approach .
�) both (b) and (c)
(x) · What are the different recommendation engine techniques:
. - Jiii Content based filtering 4!>) Collabora_tive filtering
(c) . Knowledge based system· \}ti) All of these
2. � Define Ma�hine Learning. Briefly explain the types_ of machine learning.
(2+3=5)
CSE 1817 OE 21 2
Tho values of the indopondont vuriublo x nnd dopendont vuriablo y nrc given
bolow.
X y
0
-- - -·-
-.
3 4
4 6
Find the least square regression line y=ax+b. Estimate the value of y when
x= lO. (10)
3. _(a-) How do we build a decision tree? (5)
'(b) Explain the CART Algorithm. (10)
B.Tech Papers
4. .0).. What do you understand by model overfitting? What is bias and variance in
Machine learning? (2+3=5)
� Explain with suitable example collaborative based recommendation
.
system.
(10)
5. (a) Formalize a mathematical model for a biological neuron. ( 7)
(b) A Perceptron can only converge
' on linearly separable. data,
.
support it with
your answer. (8)
6. Write down the answers of the following questions: (3+3+9=15)
(a) What is the role of weights and bias in a neural network?
(b) What do you mean by exploding and vanishing gradients?
(c) Why does a Convolutional Neural Network (CNN) and Recurrent Neural
Network (RNN) work better with iQ1age data and text data respectively?
7. (a) What is the importance of dimensionality reduction in machine learning?
"Linear Discriminant Analysis often ·outperforms in a multi-class
classification task when the class labels are k�own". Illustrate to support
your answer. (2+7=9)
clu tering? Give an examples of k-means clustering in
�- What is K-means _ � _
• real-life Data Appl icatio ns. . (3+3-6)
CSE 1817 OE 21 3
