Department of Computer Science and Engineering

Academic Year 2024 – 2025

Question Bank
Course Code / Name: CS3491 ARTIFICIAL INTELLIGENCE AND MACHINE
LEARNING

Year / Sem. / Sec: II/IV /A, B & C

UNIT –1 PROBLEM SOLVING

PART – A

D.L
Q.No Questions CO BT
(E/M/D)

1. Define Artificial Intelligence. CO1 R E

2 State the role of heuristic functions in problem-solving. CO1 U E

3 List the four components of a problem in AI. CO1 R E

4 Recall the steps involved in a simple problem-solving technique. CO1 R M

5 Name the characteristics of AI. CO1 R E

Summarize the steps of the Depth-First Search algorithm to

6 CO1 U M
evaluate its performance.

7 What is meant by a heuristic function? CO1 R E

8 Identify the main problem faced by Hill-Climbing search. CO1 R M

9 Differentiate between Blind Search and Heuristic Search. CO1 U M

10 List the types of algorithms in adversarial search CO1 R E

Identify what a local maximum is in search techniques and its

11 CO1 U M
impact on problem-solving.

12 Define the concept of Constraint Satisfaction Problems (CSP). CO1 R E

13 What are the properties of the Minimax Algorithm? CO1 R

14 Define Adversarial Search and its importance in AI. CO1 R E

Interpret the full joint probability distribution in the context of

15 CO1 U M
probabilistic reasoning

16 Classify different types of search algorithms and provide examples CO1 U E

 Find the path from S to G using greedy search. The heuristic
values h of each node below the name of the node

17 CO1 AP M

18 Define A* tree search. CO1 R E

Illustrate how Constraint Satisfaction Problems (CSP) are applied U M

19 CO1
in real-world scenarios.

How alpha-beta pruning enhances the efficiency of the mini max U M

20 CO1
algorithm.

Which solution would BFS find to move from node S to node G if

run on the graph below?
21 CO1 AP M

PART B

D.L *
Q.No Questions M CO* BT*
(E/M/D)

Outline the uninformed search strategies like breadth-first search and 16

1 CO1 U E
depth-first search with examples.

Discuss five real-world applications of artificial intelligence, highlighting 16 CO1

2 their significance and impact on various industries. Provide examples to AP M
support your discussion.

3 Find the most cost-effective path to reach from start state A to final state 16 CO1 AP M
J using A* Algorithm.
 i. Explain the concept of problem-solving agents in detail. 8 CO1 U E

ii. Solve the 8-puzzle problem given below. 8 AP

4

5 Outline the heuristic search strategies with suitable examples. 16 CO1 U E

Analyze the Classical "Water Jug Problem." Break down the state 16 CO1
6 AN M
space representation for this problem and evaluate the solution.

Solve the given problem. Describe the operators involved in it. 16 CO1 AN M

Consider a Water Jug Problem : You are given two jugs, a 4-gallon one
and a 3-gallon one. Neither has any measuring markers on it. There is a
7 pump that can be used to fill the jugs with water. How can you get exactly
2 gallons of water into the 4-gallon jug ? Explicit Assumptions: A jug can
be filled from the pump, water can be poured out of a jug onto the ground,
water can be poured from one jug to another and that there are no other
measuring devices available.

Discuss in detail about hill climbing algorithm by using 8-queens 16 CO1

8 U E
problem.

Discuss about constraint satisfaction problem with algorithm for solving 16 CO1
9 U E
a crypt arithmetic problem.

Solve the following Crypt arithmetic problem using constraints

satisfaction search procedure.

CROSS
16 CO1
10 +ROADS AP M
------------
DANGER
----------------
 UNIT –2 PROBABILISTIC REASONING
PART – A

D.L
Q.No Questions CO BT
(E/M/D)

1. How is uncertainty defined? CO2 U E

2 State Bayes’ Rule state. CO2 R E

3 Why are Bayesian Networks used? CO2 U E

4 State the principle of Maximum Expected Utility (MEU). CO2 R M

5 How would you describe causal networks? CO2 U M

6 What does the full joint probability distribution specify? CO2 U E

7 Given that P(A)=0.3,P(A|B)=0.4 and P(B)=0.5, Compute P(B|A). CO2 R M

8 Differentiate uncertainty with ignorance. CO2 R M

9 How Does the Bayes Theorem Work? CO2 U M

10 Define hybrid Bayesian network. CO2 R E

11 Define Exact Inference. CO2 U M

12 List the purpose of the do-operator in causal networks. CO2 R E

13 Mention the needs of probabilistic reasoning in AI. CO2 R E

14 Define decision theory. CO2 R E

15 Clarify the concept of Conditional probability with an example. CO2 U M

16 What an atomic event is in the context of probability theory? CO2 R E

17 Define default reasoning. CO2 R M

18 What is approximate inference? CO2 U M

19 Define Structural Causal Models. CO2 R E

20 How reasoning by default works in uncertain environments? CO2 U M

 PART – B

D.L *
Q.No Questions M CO* BT*
(E/M/D)

i.Elobrate unconditional probability and conditional probability with an 8

example.
1 CO2 U E
ii.Explain the steps followed to construct a Bayesian network with an
example. 8

What do you mean by inference in Bayesian networks? Outline 16 E

2 CO2 U
inference by enumeration with an example.

3 Summarize in detail about Naive Bayes Models. 16 CO2 U M

Describe naive Bayes models and their applications in classification 16

4 tasks. Provide a step-by-step explanation of how a naive Bayes CO2 AP M
classifier works, including the calculation of posterior probabilities.

Apply the methods of exact inference to a Bayesian network to 16

5 CO2 AP M
compute the probability of specific variables given observed evidence.

Explain the methods used to obtain approximate inference from a 16 E

6 Bayesian network and describe when and why these methods are CO2 U
preferred over exact inference.

Define uncertain knowledge, prior probability and conditional 16 M

7 probability. State the Bayes’ theorem. How it is useful for decision CO2 U
making under uncertainty? Explain belief networks briefly?

Explore exact inference techniques in Bayesian networks, such as 16

8 variable elimination and junction tree algorithms. Compare these CO2 U M
techniques in terms of computational efficiency and accuracy.

Consider the following set of propositions 16

• Patient has spots

• Patient has measles
• Patient has high fever
9 • Patient has Rocky mountain spotted fever. CO2 AP M
• Patient has previously been inoculated against
measles.
• Patient was recently bitten by a tick
• Patient has an allergy.
A) Create a network that defines the casual connections
among these nodes.
 B) Make it a Bayesian network by constructing the
necessary conditional probability matrix

Explain Causal Network or Causal Bayesian Network in Machine 16

10 CO2 U M
Learning.

UNIT – III– Supervised Learning

PART A

D.L
Q.No Questions CO BT
(E/M/D)

1 Define Machine Learning. CO3 R E

2 Define Supervised learning. CO3 R E

3 State bayes rule. CO3 R E

4 List the advantages and disadvantages of Bayesian Regression. CO3 R E

5 Define Unsupervised learning. CO3 R E

6 Why is random forest better than SVM? CO3 U M

7 Why overfitting happens? How can you avoid overfitting? CO3 U E

List out the different types of algorithm techniques in Machine CO3 R E

8
Learning.

9 What is the principle of least squares? CO3 U E

Do you think 50 small decision trees are better than a large one? CO3 U M
10
Why?

Outline the difference between supervised learning and CO3 U M

11
unsupervised learning.

12 List out the advantages of Naive Bayes. CO3 R E

13 Differentiate ‘Training dataset’ and ‘Test dataset’. CO3 U E

Mention the difference between artificial learning and machine CO3 U M

14
learning?

What are the three stages to build the hypotheses or model in CO3 R E
15
machine learning?

16 How does Naive Bayes classify data using probabilities? CO3 U M

 17 When would you use random forests vs SVM and why? CO3 U M

18 What is Decision tree? CO3 R E

19 What is Random forest? CO3 R E

20 How did the classification differ from the regression? CO3 U M

PART-B

D.L *
Q.No Questions M CO* BT*
(E/M/D)

Elaborate on logistic regression with an example. Explain the process 16

1 CO3 U E
of computing coefficient.

What is a classification tree? Explain the steps to construct a 16 CO3 M

2 U
classification tree. List and explain about different procedures used.

3 Describe the general procedure of random forest algorithm. 16 CO3 U E

4 With a suitable example explain knowledge extraction in detail. 16 CO3 U E

Explain the principle of the gradient descent algorithm. Accompany 16 CO3 M

5 U
your explanation with a neat diagram.

6 Explain Naive Bayes Classifier with an example. 16 CO3 U E

7 Explain SVM Algorithm in Detail. 16 CO3 U E

8 Explain the following a) Linear regression b) Logistic Regression 16 CO3 U E

9 Explain Decision Tree Classification with suitable example. 16 CO3 U E

Compare and contrast the situations in which using a Random Forest 16 CO3
10 algorithm would be more suitable than a Support Vector Machine AN M
(SVM), and explain the reasons behind the preference.

UNIT –IV Ensemble Techniques and Unsupervised Learning

PART – A

D.L
Q.No Questions CO BT
(E/M/D)

1. Define unsupervised learning. CO4 R E

2 Define voting and list its types. CO4 R E

3 Define ensemble learning. CO4 R E

4 List the three types of ensemble learning. CO4 R E

5 List the properties of the K-means algorithm. CO4 R M

6 Define bootstrapping. CO4 R E

7 What types of classifiers are used in the weighted voting method? CO4 R E

8 What is unsupervised learning? CO4 R E

9 Define voting in machine learning and mention its types. CO4 R E

10 Write a short note ensemble method and its importance. CO4 U M

11 What is stacking in ensemble learning? CO4 R E

12 Summarize clustering and its importance in machine learning. CO4 U M

13 What is K-Nearest Neighbour (KNN) algorithm? CO4 U E

14 State the principle of maximum likelihood and its significance. CO4 U M

15 What is K-means clustering in unsupervised learning? CO4 U E

What is Gaussian Mixture Model (GMM) and the role of CO4 U M

16
Expectation-Maximization (EM) in it.

17 State the logic behind Gaussian processes. CO4 U M

18 Why are ensemble methods used in machine learning? CO4 R M

19 Compare and contrast K-means and KNN. CO4 U M

20 What factors influence the performance of the KNN algorithm? CO4 U E

How do Gaussian Mixture Models (GMMs) differ from K-means CO4 U

21 M
clustering?

22 Differentiate between bagging and boosting. CO4 U M

23 List the types of Boosting. CO4 R E

24 How to select the optimal K value? CO4 R M

25 Define Blending. CO4 R E

 PART B

D.L *
Q.No Questions M CO* BT*
(E/M/D)

Explain the steps involved in the AdaBoost algorithm with a simple 16

1 CO4 U M
example.

Outline the steps of the Expectation-Maximization (EM) algorithm and

2 describe its significance in clustering problems. U M
16 CO4

Describe the various ensemble learning techniques with suitable

3 examples. M
16 CO4 U

Explain the different learning techniques in unsupervised learning,

4 highlighting their applications. U M
16 CO4

List the applications of clustering algorithms and analyze their

5 advantages and disadvantages in solving real-world problems. M
16 CO4 U

Discuss in detail about the strengths and limitations of combining 16 CO4

6 U M
multiple classifiers using voting techniques.

Explain the K-means clustering algorithm and apply it to your own 16 CO4
7 AP M
dataset in detail about each step.

Elaborate on the K-Nearest Neighbor (KNN) algorithm, explain how it 16 CO4

8 AP E
can be applied to classification tasks.

Apply the Expectation-Maximization (EM) algorithm to a practical 16 CO4

9 AP D
problem involving Gaussian distributions.

Using ensemble learning techniques such as bagging, boosting, and

10 stacking, apply them to solve a classification problem and evaluate the AP D
results. 16 CO4
 UNIT-V NEURAL NETWORKS

PART A

D.L
Q.No Questions CO BT
(E/M/D)

1. Define a perceptron. CO5 R E

2 State any two limitation of the perceptron model CO5 U M

3 Define Multilayer Perceptron (MLP). CO5 R E

4 How does a perceptron make decisions? CO5 U E

Why is a Multilayer Perceptron preferred over a perceptron for CO5

5 U M
complex tasks?

6 What is the purpose of an activation function in a neural network? CO5 R E

7 Name two commonly used activation functions in deep learning. CO5 R E

8 Why is gradient descent used in neural network training? CO5 U E

9 What is backpropagation in neural network training? CO5 U E

10 Mention one advantage of the ReLU activation function. CO5 U E

11 What is gradient descent in neural network training? CO5 R E

12 Define stochastic gradient descent (SGD). CO5 R E

What is the role of the error backpropagation algorithm in training CO5

13 U E
neural networks?

Mention one difference between batch gradient descent and CO5

14 U M
stochastic gradient descent.

15 Why is optimization important in neural network training? CO5 U E

16 What is the vanishing gradient problem? CO5 U E

Why is the ReLU activation function preferred over Sigmoid or CO5

17 U M
Tanh in deep networks?

18 Define the difference between shallow and deep neural networks. CO5 U M

19 What causes unit saturation in neural networks? CO5 U E

20 Mention one advantage of deep networks over shallow networks. CO5 U E

 21 What is hyperparameter tuning in neural networks? CO5 U E

22 Define batch normalization and its purpose in deep learning. CO5 U E

23 What is the role of regularization in machine learning models? CO5 U E

24 How does dropout help prevent overfitting in neural networks? CO5 U E

25 Mention one common method for hyperparameter tuning. CO5 R E

PART B

D.L *
Q.No Questions M CO* BT*
(E/M/D)

Explain the architecture of a single layer perceptron (SLP) and explain 16

1 CO5 U E
its operation. Mention its advantages and disadvantages.

Draw the architecture of a Multilayer perceptron (MLP) and explain its 16 CO5 E
2 U
operation. Mention its advantages and disadvantages.

3 Explain the stochastic optimization methods for weight determination. 16 CO5 U E

4 Describe back propagation and features of back propagation. 16 CO5 U E

Develop a Back propagation algorithm for Multilayer Feed forward 16 CO5

5 neural network consisting of one input layer, one hidden layer and AP M
output layer from first principles.

6 Explain the role of activation functions in neural networks. 16 CO5 U E

Explain the factors that affect the performance of multilayer feed- 16 CO5 E
7 U
forward neural network.

8 Differentiate between Shallow Net & Deep Learning Net. 16 CO5 U E

How do you tune hyperparameters for better neural network 16 CO5

9 U E
performance? Explain in detail.

Elaborate the process of training hidden layers by ReLU in deep 16 CO5

10 U E
networks.

Faculty HoD Dean-Academics Principal