VISHNU INSTITUTE OF TECHNOLOGY

Department of Artificial Intelligence & Machine Learning

Machine Learning
Question Bank
UNIT-I- 2M

1. Define machine learning.

2. What are the three primary paradigms of machine learning?
3. Give an example of where rote learning is applied in AI.
4. Define learning by induction.
5. What are the main components of a reinforcement learning system?
6. Name the two primary types of data used in machine learning.
7. How is structured data different from unstructured data?
8. Why is model evaluation a crucial stage in machine learning?
9. What is data acquisition in machine learning?
10. Define feature engineering.

UNIT-I- 10M

1. What are the key milestones in the evolution of machine learning?

2. What are the primary paradigms of machine learning, and how do they differ?
3. How does supervised learning compare to unsupervised learning and reinforcement
learning?
4. What is rote learning, and how is it applied in machine learning?
5. What are the challenges of inductive learning in the presence of noisy data?
6. How does reinforcement learning differ from supervised and unsupervised learning?
7. What are the main types of data used in machine learning?
8. What is matching in the context of machine learning?
9. What are the primary stages involved in a machine learning workflow?
10. Why is model evaluation critical before deploying a machine learning system?

UNIT-II-2M

1. Name two commonly used metric distance measures in machine learning.

2. What is the Manhattan distance formula?
3. Provide an example of a non-metric similarity function.
4. What is Hamming distance?
5. How is Jaccard similarity calculated for binary data?
6. What does the k in K-Nearest Neighbor (KNN) represent?
7. List one advantage and one disadvantage of the KNN classifier.
8. Name two metrics used to evaluate the performance of classifiers.
9. What metric is commonly used to evaluate regression algorithms?
10. What is the key difference between the Radius Distance Nearest Neighbor
algorithm and KNN?
UNIT-II- 10M
1. What are proximity measures, and why are they important in nearest neighbor-
based
models?
2. What are the most commonly used distance measures in nearest neighbor
algorithms?
3. How does the Euclidean distance differ from the Manhattan distance, and when is
each more
appropriate?
4. What are non-metric similarity functions, and how do they differ from metric
distance
measures?
5. Why is it important to use specialized measures for binary patterns in machine
learning?
6. Expalin any two classification algorithms that rely on distance measures apart from
KNN?
7. What is the K-Nearest Neighbor (KNN) classifier, and how does it work?
8. What is the Radius Distance Nearest Neighbor algorithm, and how does it differ
from KNN?
9. How is KNN regression different from KNN classification?
10. What metrics are most suitable for assessing the performance of KNN classifiers?

UNIT-III- 2M

1. What is a decision tree? How is it used for classification tasks?

2. What is the Bayes Classifier? Why is it considered optimal?
3. What is Entropy?
4. What is Gini impurity?
5. Define bias.
6. State Bayes Rule
7. Write any two advantages of using Naive Bayes for classification.
8. What is random Forest?
9. Why is the Bayes classifier called the optimal classifier?
10. What is classification? Give an example.

UNIT-III- 10M

1. Explain the bias-variance trade-off in decision trees.

2. Differentiate between classification and regression decision trees.
3. State and explain Bayes’ Rule.
4. How does overfitting and underfitting relate to bias and variance in decision trees?
5. Provide an example to illustrate multi-class classification using the Bayes approach.
6. Discuss the advantages and limitations of using Naive Bayes for classification.
7. Compare and contrast Gini impurity and entropy. When would you use one over the
other?
8. Explain how random forests handle both classification and regression tasks.
9. Discuss a real-world example where random forests would outperform a single
decision tree.
10.Explain the concept of impurity in decision trees.

UNIT-IV -2M

1. What is a linear discriminant? [L1]

2. What is a perceptron classifier? [L1]
3. What is the main objective of SVM for linearly separable data? [L2]
4. What is the kernel trick in SVM? [L2]
5. Name any two kernel functions used in SVM. [L1]
6. What is logistic regression used for? [L1]
7. What is a multi-layer perceptron (MLP)? [L1]
8. Define activation function in the context of MLP. [L2]
9.Compare perceptron and SVM in one sentence. [L4]
10. What is the key difference between SVM and MLP for classification tasks? [L4]

UNIT-IV -10M

1. What are linear discriminants? Explain their role in classification tasks. [L2]
2. Describe the architecture of a perceptron classifier. How does it make decisions?
[L2]
3. Compare and contrast the perceptron classifier with linear discriminants. [L4
4. What is a support vector machine? Explain how it works for linearly separable data.
[L2]
5. Analyze the difference between linearly separable and non-linearly separable cases
in
SVM. [L4]
6. What is the kernel trick in SVM? Provide examples of common kernel functions. [L2]
7. Evaluate the effect of different kernel choices (linear, polynomial, RBF) on SVM
performance. [L5]
8. Explain the difference between linear regression and logistic regression. [L2]
9. What is a multi-layer perceptron (MLP)? Describe its architecture. [L2]
 10. Explain the backpropagation algorithm for training an MLP. Why is it necessary?
[L2]

UNIT-V- 2M

1. Define clustering in the context of machine learning. [L1]

2. List two differences between partitional and hierarchical clustering. [L1]
3. Explain the main idea behind K-means clustering. [L2]
4. Summarize the concept of fuzzy C-means clustering. [L2]
5. Use the K-means algorithm to find the next centroid for the given data points: (2,3),
(3,4), and (5,6). [L3]
6. Identify the key differences between hard partitioning and soft partitioning. [L4]
7. Differentiate between agglomerative and divisive clustering. [L4]
8. Justify why K-means might fail to detect non-spherical clusters. [L5]
9. Critique the limitations of rough clustering compared to fuzzy clustering. [L5]
10.Formulate a step-by-step plan to apply spectral clustering on a small graph. [L6]

UNIT-V- 10M

1. Define clustering and explain its significance in machine learning. Provide examples
of real-world applications. [L1]
2. Explain the differences between divisive and agglomerative clustering, including
their working mechanisms and graphical representations. [L2]
3. Apply the K-means algorithm to cluster the following data points into 2 clusters:
(2,3), (3,4), (5,6), (8,8). Show the centroid calculations for the first iteration. [L3]

4. Demonstrate how the Expectation-Maximization algorithm works, using a simple

example with two Gaussian distributions. [L3]
5. Analyze the impact of different distance metrics (Euclidean, Manhattan, and Cosine
similarity) on K-means clustering results. [L4]
6. Differentiate between hard clustering (K-means) and soft clustering (Fuzzy C-
means). Explain how membership values are calculated in soft clustering. [L4]
7. Critique the limitations of rough K-means clustering and compare it with fuzzy C-
means clustering. Discuss which method works better for overlapping clusters. [L5]
8. Justify the use of Spectral Clustering for non-linearly separable data. Explain why
traditional clustering methods might fail in such cases. [L5]
9. Propose a hybrid clustering approach combining K-means and Expectation-
Maximization for a dataset with complex cluster shapes. [L6]
10. Formulate a step-by-step plan to cluster social network data using Spectral
Clustering.
Highlight how graph-based approaches help uncover hidden patterns. [L6]