CODE: 2066PC03 R20 SET - 2

MALLA REDDY ENGINEERING COLLEGE FOR WOMEN

(Autonomous Institution – UGC, Govt. of India)
(Affiliated to JNTU, Hyderabad, Approved by AICTE - - ISO 9001:2015 Certified)
Accredited by NBA & NAAC – ‘A’ Grade
NIRF India Ranking, Accepted by MHRD, Govt. of India

B.TECH III YEAR II SEMESTER REGULAR END EXAMINATIONS, MAY-2023

MACHINE LEARNING
(CSE – AI&ML)
[Time: 3 Hours] [Max. Marks: 70]
PART – A
(5 x 2 = 10M)
Note: 1. This Part consists of 8 QUESTIONS
2. Answer any 5 questions. Each question carries 2 Marks

1 A What are bias and variance in machine learning? 2M BTL1

B What is a logic model in machine learning? Define. 2M BTL1
C What are the consequences of over-fitting on a machine learning 2M BTL2
model's performance?
D What is the main idea behind bagging in ensemble learning? 2M BTL1
E What are the key components of an association rule, such as 2M BTL2
antecedent, consequent, support, or confidence?
F How does passive reinforcement learning differ from active 2M BTL2
reinforcement learning?
G What is the trade-off between exploration and exploitation in 2M BTL1
reinforcement learning?
H How do meta learning algorithms learn to adapt to new tasks or 2M BTL2
environments?

PART – B
(5 x 12 = 60M)
Note: 1. This Part consists of 10 QUESTIONS
2. Answer any 1 question from each Section. Each question carries 12Marks.
3. Illustrate your answers with NEAT sketches wherever necessary.
SECTION - I
2.A What is a geometric model in machine learning, and how is it different 6M BTL2
from a probabilistic model?
2.B What is reinforcement learning, and how does it differ from supervised 6M BTL3
and unsupervised learning?
(OR)
3.A What is the approximation generalization tradeoff in machine learning, 6M BTL2
and how does it affect the performance of a model?
3.B What are some techniques for handling noisy data in machine learning, 6M BTL1
such as data cleaning, data augmentation, and outlier detection?

SECTION - II
4.A What is the feasibility of learning, and how can we determine whether a 6M BTL1
problem is learnable?
4.B How can you tune the hyper-parameters of a soft margin SVM using 6M BTL3
CODE: 2066PC03 R20 SET - 2
cross-validation?

(OR)
5.A What is multivariate linear regression and how is it different from 8 M BTL1
univariate linear regression? How do you handle categorical variables in
multivariate linear regression?
5.B What are the advantages and disadvantages of using k-fold cross-validation 4 M BTL2
over holdout validation?

SECTION - III
6.A What are the differences between the Euclidean distance and Manhattan 6 M BTL2
distance metrics in the nearest neighbor algorithm? How can you handle
categorical variables in the nearest neighbor algorithm? Explain with
examples.
6.B What is hierarchical clustering, and how does it differ from other clustering 6M BTL3
techniques, such as K-means and density-based clustering?

(OR)
7.A What is K-means clustering, and how does it work? Describe the main steps of
6 M BTL3
the algorithm, including initialization, assignment, and update.
7.B What is Locality Sensitive Hashing (LSH), and how does it work? Describe 6 M BTL3
the main idea behind LSH, including the use of random projections and
hash functions to efficiently search for similar data points in
high-dimensional spaces.

SECTION – IV
8.A What is descriptive rule learning, and how does it differ from predictive 6 M BTL1
modeling and classification?
8.B Explain the concept of bootstrapping in bagging, and how it can be used to 6 M BTL2
generate multiple training sets from a single dataset.

(OR)
9.A Explain the concept of ranking trees, and how they can be used to learn a 6 M BTL2
ranking function from pairwise preferences.
9.B What are ordered rule lists, and how do they differ from unordered rule 6 M BTL1
sets or decision trees?

SECTION – V
10.A Explain the difference between model-based and model-free reinforcement 6 M BTL2
learning, and how it relates to the concept of generalization.
10.B What are some applications of machine learning in robot control, such as 6 M BTL1
in autonomous navigation, grasping, and manipulation?
(OR)
11.A What is adaptive dynamic programming (ADP), and how does it differ 6M BTL3
from other machine learning techniques, such as supervised learning and
reinforcement learning?
11.B How does temporal-difference (TD) learning work in reinforcement 6M BTL4
learning, and what are some advantages and disadvantages of this approach?
----***----