i) Solve how the CART (Classification

i) Describe
and Regression
the process Tree)
by which the CART algorithm constructs
algorithm constructs a decision tree. decision trees.
ii) Construct the role of Gini Impurityii)and
Compare
Entropyand
as explain
splittinghow Gini Impurity and Entropy are used as
criteria in decision trees. measures to select the best feature split.
Show the key idea behind Gradient Boosting?
(OR) (OR)
How a decision tree make predictions?
Define
i) a medoid
Develop in clustering.
and justify a machine learning solutionausing
i) Construct ensemble
machine learning model for a noisy classification
What is the difference between PCA and Factor Analysis?
methods such as Bagging, AdaBoost, problemGradient Boosting,
utilizing ensembleand algorithms like Bagging, AdaBoost,
List the applications of Independent Component Analysis.
XGBoost for main
Discover the a classification problem
objective of with noisy
Gradient
Principal [Link].
Boosting,
Component and XGBoost, and provide a rationale for your
Identify
ii) why exploration
Organize ensembleis important
techniques in active
design learning?
choices.
(Bagging, AdaBoost,
Compare Boosting,
Gradient active reinforcement
XGBoost) learning ii) and
to enhance active ensemble
model
Apply reinforcement
accuracy andlearning methods—such as Bagging,
learning.
robustness in the presence
Name the importance of noise. AdaBoost,
of generalization Gradientlearning
in reinforcement Boosting, and XGBoost—to improve model
Outline the performance and resilience when dealing with noisy data.
Analyze anddisadvantages
explain variousof partitional
direct utility estimation.
clustering algorithms such
as K-Means and Clustering Around Medoids
Examine(PAM).
and Discuss
describetheir
different partitional clustering methods,
working principles, mathematical formulations,
including advantages, and Partitioning Around Medoids (PAM).
K-Means and
limitations. Illustrate with suitable examples
Explainandtheir
diagrams.
underlying mechanisms, mathematical models,
(OR) strengths, and weaknesses, and support your discussion with
Examine the concept and working appropriate
of Hierarchical Clustering
examples in
and diagrams.
detail. Discuss the difference between Agglomerative and Divisive
approaches with suitable diagrams. Analyze the various linkage
methods used, illustrate the of a dendrogram for a sample dataset. (OR)
Explain the concept and working of Hierarchical Clustering.
Classify Principal Component Analysis (PCA) and Linear
Differentiate between Agglomerative and Divisive methods, and
Discriminant Analysis (LDA) on a multivariate dataset. Justify the
illustrate linkage types and dendrogram construction with example.
choice of technique based on objectives like variance
Distinguish between Principal Component Analysis (PCA) and
preservation or class separability.
Linear Discriminant Analysis (LDA) for multivariate datasets, and
(OR)
justify the choice of method based on objectives such as variance
Contrast the dimensionality reduction pipeline using ICA and
preservation or class separability.
Factor Analysis for a signal processing problem. Discuss how
(OR)
feature independence and latent variables are captured.
Examine a dimensionality reduction framework using Independent
Examine the approaches of direct utility estimation, adaptive
Component Analysis (ICA) and Factor Analysis for a signal
dynamic programming, and temporal-difference (TD) learning in
processing application, explaining how the methods extract
passive reinforcement learning. Evaluate how each method works,
independent features and uncover underlying latent factors.
along with its strengths, limitations, and suitable application
Analyze the methods of direct utility estimation, adaptive dynamic
scenarios
programming, and temporal-difference (TD) learning in passive
(OR)
reinforcement learning, discussing their working principles,
Compare and contrast passive and active reinforcement
advantages, drawbacks, and appropriate use cases.
learning. Analyze how exploration influences the learning of an
(OR)
action-utility function, and justify the trade-off between
Differentiate between passive and active reinforcement learning,
exploration and exploitation in real-world environments.
explaining the role of exploration in developing the action-utility
Apply the role of policy search methods in reinforcement learning.
function. Discuss the balance between exploration and exploitation
Discuss when they are preferred over value-based methods, and