Regulation: R23

Subjectcode:R2322421
TIRUMALAENGINEERINGCOLLEGE SET-2
(Autonomous)
II [Link] II Semester Regular Examinations APRIL-2025
Machine Learning
(CSE(AIML))
Duration: 3 hours Maximum Marks: 70
Note:
1. This question paper contains two parts A and B.
2. Part A is compulsory which carries 20 marks. Answer all questions in PartA.
3. Part B consists of 5 Units. Answer any question from each unit.
4. Each question carries 10marks and may have a,b,c,d as sub questions.
Part-A
All the following questions carry equal marks (10x2M=20Marks)
Blooms
[Link] Marks CO
Questions Taxonomy
1. attainment
Level
Explain the importance of data representation in Machine
a) 2M CO1
Learning.
How is model prediction different from model
b) 2M CO1
evaluation?
c) What is the role of distance weighting in KNN? 2M CO2
Why is choosing the correct value of K important in
d) 2M CO2
KNN?
e) How does Random Forest overcome over fitting? 2M CO3
How does pruning help in improving Decision Tree
f) 2M CO3
performance?
g) What is the Kernel Trick in SVM? 2M CO4
What is the main difference between Logistic Regression
h) 2M CO4
And Linear Regression?
i) What is Rough Clustering? 2M CO5
j) Define partitioning clustering with an example. 2M CO5
Part-B
Answer All the following questions. (5x10M=50Marks)
CO Blooms
Marks
[Link] Questions attainm Taxonomy
ent Level
UNIT-I
Compare and contrast Learning by Rote, Learning by
2) Induction, and Reinforcement Learning with suitable 10M CO1
examples.
OR
A binary classification model produces the following
confusion matrix:
Actual\Predicted Positive (P) Negative(N)
3) Positive (p) 70 30 10M CO1
Negative(N) 20 80

Compute Precision, Recall, and F1-score for the Positive class.

Page1of2
 UNIT-II
Explain different proximity and distance measures used in
4) 10M CO2
Machine Learning with suitable examples.
OR
A dataset contains 5 data points in a 2D space:
Point X Y
A 2 3
B 4 7
C 6 5
D 8 9
E 3 6
5) A query point Q=(5,5)is given. 10M CO2

(a) Compute the Euclidean distance between Q and

UNIT-III
Explain the working of Decision Trees for classification with
6) 10M CO3
A suitable example.
OR
A Random Forest model uses 5 Decision Trees to classify a
new sample. The predictions of the trees are:
Decision Tree Predicted Class
Tree1 A
Tree2 B
Tree3 A
Tree4 A
7) Tree5 B 10 M CO3

(a) Compute the final class label using majority

UNIT-IV
Explain how Support Vector Machines(SVM)works for
8) classification. Derive the mathematical formulation for finding 10M CO4
the optimal hyper plane.
OR
Explain how a Multi-Layer Perceptron (MLP)works.
9) Derive the Back propagation Algorithm for training an 10M CO4
MLP.
UNIT-V
Compare soft clustering and hard clustering with real-
10) World examples Discuss the advantages of Fuzzy C- 10M CO5
Means clustering over K-Means.
OR
Explain the Expectation-Maximization(EM) Algorithm for
11) clustering. Solve a numerical example showing how EM 10M CO5
Updates cluster parameters.

Page2of2
Page3of2