Course Code Course Title L T P C
BAITE301 Machine Learning 3 0 2 4
Pre-requisite NIL Syllabus Version
1
Course Objectives
1. To develop a strong grasp of core machine learning concepts.
2. To acquire knowledge and skills in the design, training and fine-tuning of various
machine learning models using supervised learning algorithms and familiar with
principles of neural networks.
3. To learn knowledge and skills for developing machine learning models using
unsupervised learning & reinforcement learning algorithms and understand the
principles of optimization & evaluation metrics.
Course Outcomes
1. Gain a strong knowledge of the fundamental techniques in machine learning.
2. Develop a comprehensive understanding and the ability to implement supervised and
ensemble learning algorithms to solve complex real-world problems.
3. Understand and apply the neural network training process to learn representations for
different types of neural network architectures.
4. Comprehend the basics of unsupervised and reinforcement learning algorithms and
apply them to address practical challenges.
5. Evaluate and optimize machine learning models using appropriate metrics and
techniques.
6. .
Module:1 Introduction 9 hours
Evolution of Artificial Intelligence – Well-Posed Learning Problems - Designing a
Learning Systems - Types and Applications of Machine Learning — Hypothesis spaces
and Complexity - Vapnik-Chervonenkis (VC) dimension, Probably Approximately
Correct (PAC) Learning -Bias - variance trade-off – Data Preprocessing Techniques –
Principal Component Analysis (PCA), Linear Discriminant Analysis (LDA).
Module:2 Supervised Learning 10 hours
Classification - Regression - Decision Tree Classifiers – Pruning – Rules Extraction -
Naïve Bayes - Linear Regression: Single and Multiple Variables - Logistic Regression -
Maximum Margin Classifier – Support Vector Machine – KNN - Random Forests - Model
Combination Schemes - Voting, Ensemble Learning - Bagging, Boosting, Stacking
Module:3 Neural Network Learning 10 hours
Neural Network Representations - Multilayer Perceptron – Back propagation algorithm – Training Procedures –
Improving Convergence – Loss Functions and Optimization – Regularization, Early Stopping – Deep Learning –
Convolution Neural Networks – Recurrent Neural Networks - Generative Neural Networks
Unsupervised Learning and
Module:4 9 hours
Reinforcement Learning
Association Rule Mining - Partitioning Clustering, Gaussian Mixture Models and
Expectation Maximization – Hierarchical Clustering - Basics of Reinforcement Learning
- Reinforcement Learning Framework–Markov Decision Process-Exploration Vs
Exploitation - Polices, Value Functions and Bellman Equations - Solution Methods -Q-
learning
Model Evaluation and
Module:5 5 hours
Optimization
Training, Testing, and Validation Sets -Overfitting – Underfitting – Performance Metrics
for Classifiers - Confusion Matrix - The Receiver Operator Characteristic (ROC) Curve
- Imbalanced Datasets - Measurement Precision - Parameter Tuning - Cross Validation
– Regularization – Performance Metrics for regression and clustering
Module:6 Contemporary Issues 2 hours
Guest Lecture from Industry and R & D Organizations
Total Lecture Hours: 45 hours
Text Book(s)
1. Ethem Alpaydin, "Introduction to Machine Learning", MIT Press, 4th Edition, 2020
2. Mitchell, Tom M., "Machine Learning", McGraw Hill New York Indian Edition, 1st
Edition, 2017
Reference Books
1. Marsland, Stephen , "Machine Learning: An Algorithmic Perspective", Chapman
and Hall CRC, 2nd Edition, 2015
2. Ian Goodfellow, Yoshua Bengio, Aaron Courville, , "Deep Learning", MIT Press, 1st
Edition, 2016
3. Richard S. Sutton and Andrew G. Bart , "Reinforcement Learning An Introduction
", MIT Press, 2nd Edition, 2015
4. Aurelien Geron,, "Hands-On Machine Learning with Scikit-Learn, Keras, and
TensorFlow", O Reilly Media Inc, 2nd Edition, 2019
Indicative Experiments
1.
Implement the data preprocessing techniques for the following
3 hours
a) Handling missing values,
b) normalization
c) standardization
d) one-hot encoding
e) feature selection
2. Implement linear regression using python libraries and evaluate
3 hours
RMSE and R².
3. Demonstrate the support vector machine algorithm for
3 hours
classification
4. Implement KNN for classification on datasets like the Iris
3 hours
dataset.
5. Construct a decision tree classifier for multi-class classification
3 hours
problems in customer churn prediction.
6. Implement K-means clustering algorithm based on their similarity
3 hours
and use customer segmentation and image compression datasets.
7. Implement the fundamental concepts of a perceptron, a basic
building block of neural networks, and implement it for a binary 3 hours
classification task.
8. Implement a simple CNN for image classification using the MNIST
3 hours
dataset.
9. Perform various performance metrics for evaluating machine
3 hours
learning models for classification tasks.
10. Implement optimization parameters used in many machine
learning models, and observe its behavior with different learning
3 hours
rates, convergence and initialization strategies. Explore different loss
functions.
Total Laboratory Hours: 30 hours
Mode of Evaluation :Continuous Assessment Test, Digital Assignment, Quiz, Final
Assessment Test, Lab Continuous Assessment, Lab Final Assessment, Oral
Examination, Seminar, Group Discussion, Presentation
Recommended by Board of Studies : 24-02-2025
Approved by Academic Council : No. 78 12-06-2025