COURSE DATA SHEET

PROGRAM: AI&DS DEGREE: B.TECH

COURSE: MACHINE LEARNING SEMESTER: IV CREDITS: 3
COURSE CODE: AL3451 REGULATION: 2021 COURSE TYPE: CORE /ELECTIVE / BREADTH/ S&H
COURSE AREA/DOMAIN: Software Domain CONTACT HOURS: 3
CORRESPONDING LAB COURSE CODE (IF LAB COURSE NAME (IF ANY):Machine Learning
ANY):AD3461 Laboratory

SYLLABUS:
UNIT DETAILS HOURS
UNIT I INTRODUCTION TO MACHINE LEARNING
Review of Linear Algebra for machine learning; Introduction and motivation for machine
I learning; Examples of machine learning applications, Vapnik-Chervonenkis (VC) dimension, 8
Probably Approximately Correct (PAC) learning, Hypothesis spaces, Inductive bias,
Generalization, Bias variance trade-off.
UNIT II SUPERVISED LEARNING
Linear Regression Models: Least squares, single & multiple variables, Bayesian linear
regression, gradient descent, Linear Classification Models: Discriminant function – Perceptron
II 11
algorithm, Probabilistic discriminative model - Logistic regression, Probabilistic generative
model – Naïve Bayes, Maximum margin classifier – Support vector machine, Decision Tree,
Random Forests
UNIT III ENSEMBLE TECHNIQUES AND UNSUPERVISED LEARNING
Combining multiple learners: Model combination schemes, Voting, Ensemble Learning -
III 9
bagging, boosting, stacking, Unsupervised learning: K-means, Instance Based Learning: KNN,
Gaussian mixture models and Expectation maximization.
UNIT IV NEURAL NETWORKS
Multilayer perceptron, activation functions, network training – gradient descent optimization –
IV stochastic gradient descent, error backpropagation, from shallow networks to deep networks – 9
Unit saturation (aka the vanishing gradient problem) – ReLU, hyperparameter tuning, batch
normalization, regularization, dropout.
UNIT V DESIGN AND ANALYSIS OF MACHINE LEARNING EXPERIMENTS
Guidelines for machine learning experiments, Cross Validation (CV) and resampling – K-fold
V CV, bootstrapping, measuring classifier performance, assessing a single classification 8
algorithm and comparing two classification algorithms – t test, McNemar’s test, K-fold CV
paired t test
TOTAL HOURS 45

TEXT/REFERENCE BOOKS:
T/R BOOK TITLE/AUTHORS/PUBLICATION

T Ethem Alpaydin, “Introduction to Machine Learning”, MIT Press, Fourth Edition, 2020.
T Stephen Marsland, “Machine Learning: An Algorithmic Perspective, “Second Edition”, CRC Press, 2014.
R Christopher M. Bishop, “Pattern Recognition and Machine Learning”, Springer, 2006.
R Tom Mitchell, “Machine Learning”, McGraw Hill, 3rd Edition, 1997

R Mehryar Mohri, Afshin Rostamizadeh, Ameet Talwalkar, “Foundations of Machine Learning”, Second Edition,
MIT Press, 2012, 2018.
R Ian Goodfellow, Yoshua Bengio, Aaron Courville, “Deep Learning”, MIT Press, 2016
 R Sebastain Raschka, Vahid Mirjalili , “Python Machine Learning”, Packt publishing, 3rd Edition, 2019.

COURSE PRE-REQUISITES:
C.CODE COURSE NAME DESCRIPTION SEM
- - - -

COURSE OBJECTIVES:
1 To understand the basic concepts of machine learning.

2 To understand and build supervised learning models.

3 To understand and build unsupervised learning models.
4 To evaluate the algorithms based on corresponding metrics identified
5 To design and analyse machine learning experiments, employing Cross Validation and resampling methods

COURSE OUTCOMES:
PO(1..12) &
S.NO DESCRIPTION PSO(1..2)
MAPPING
PO1,PO2,PO3,PO4,P
Students will be able to understand the fundamental concepts of machine learning
C212.1 O9,PO10,P011,PO12
and their applications in real-world problems.
& PSO1,PSO2
PO1,PO2,PO3,PO4,P
Students will be to evaluate supervised learning techniques, applying them adeptly
C212.2 SO5,PO9,PO10,P011
to address both regression and classification tasks with efficacy and precision.
,PO12 & PSO1,PSO2
Students will be able to utilize diverse model combination schemes to enhance PO1,PO2,PO3,PO4,P
C212.3 unsupervised learning techniques for effective pattern recognition, clustering, and SO5,PO9,PO10,P011
predictive modeling tasks. ,PO12 & PSO1,PSO2

Students will able to demonstrate multilayer perceptrons, including activation PO1,PO2,PO3,PO4,P

C212.4 functions, and excel in optimizing network performance through hyper parameter SO5,PO9,PO10,P011
tuning techniques. ,PO12 & PSO1,PSO2

Students will able to design and analyze machine learning experiments, employing PO1,PO2,PO3,PO4,P
C212.5 Cross Validation and resampling methods to ensure robustness and reliability in SO5,PO9,PO10,P011
model evaluation and performance estimation. ,PO12 & PSO1,PSO2
 COURSE OUTCOMES VS POS MAPPING (DETAILED; HIGH:3; MEDIUM:2; LOW:1):
SNO COURSE OUTCOMES PO1 PO2 PO3 PO4 PO5 PO6 PO7 PO8 PO9 PO10 PO11 PO1 PSO PSO PSO
2 1 2 3
Students will be able to
understand the fundamental
concepts of machine 2 1 2 1 - - - - 3 3 2 2 2 2 2
C212.1
learning and their
applications in real-world
problems.
Students will be to evaluate
supervised learning
techniques, applying them
C212.2 adeptly to address both 2 3 3 1 2 - - - 2 2 2 1 3 1
regression and classification 1
tasks with efficacy and
precision.
Students will be able to
utilize diverse model
combination schemes to
enhance unsupervised
C212.3 learning techniques for 2 1 3 3 2 - - - 1 1 1 1 1 2 2
effective pattern
recognition, clustering, and
predictive modeling tasks.

Students will able to

demonstrate multilayer
perceptrons, including
activation functions, and 2 3 3 2 1 - - - 3 2 3 2 1 2
C212.4 2
excel in optimizing network
performance through hyper
parameter tuning
techniques.
Students will able to design
and analyze machine
learning experiments,
employing Cross Validation
C212.5 and resampling methods to 2 2 3 1 2 - - - 3 1 1 1 2 2 1
ensure robustness and
reliability in model
evaluation and performance
estimation.

Avg 2 2 3 2 1 - - - 2 2 2 1 2 2 2

* For Entire Course, PO /PSO Mapping; 1 (Low); 2(Medium); 3(High) Contribution to PO/PSO

PO1 Engineering Knowledge PO7 Environment and Sustainability PSO1 Modern

Technology
PO2 Problem Analysis PO8 Ethics PSO2 Sustainable
Development
PO3 Design and Development PO9 Individual and Team Work PSO3 Leadership

PO4 Investigations PO10 Communication Skills

PO5 Modern Tools PO11 Project Management and Finance

PO6 Engineer and Society PO12 Life Long Learning

JUSTIFICATION FOR MAPPING

SNO PO/PSO MAPPED JUSTIFICATION
Knowledge of various Machine Learning techniques solving complex
PO1
Engineering problems.
Principles of mathematics and engineering sciences are used in various aspects of Problem
PO2
analysis in Machine Learning approaches.
Heuristic search techniques can be used to design solutions to complex engineering
PO3
problems.
PO4 Dissertation on Feature Engineering and Selection in Complex Problems
PO9 Individual and team work allows practitioners to delve deeply into specific areas of interest,
gaining expertise and specialization of machine learning techniques
PO10 Effective communication facilitates the transfer of knowledge and machine learning concepts
C212.1 to grasp difficult concepts.
PO11 Machine learning provides powerful tools for decision making and optimization in project
management and finance.
PO12 Lifelong learning cultivates critical thinking skills and creativity, essential for problem-
solving and innovation in machine learning.
PSO1 Various learning approaches to acquire skills to design, analyse and develop algorithms and
implement those using high-level programming language
PSO2 Theoretical underpinnings of machine learning, which is essential for building more
advanced skills and knowledge in the field.
PSO3 Leadership with a solid understanding of machine learning concepts can facilitate cross-
functional collaboration, bridge communication gaps to ensure project success.
Engineering knowledge provides the mathematical framework necessary to understand the
PO1 core concepts of machine learning, such as linear algebra, calculus, probability theory, and
statistics.
PO2 Conducting EDA to uncover patterns, trends, and relationships in the data.
Supervised learning models can be trained on datasets containing of designs on Predictive
PO3
Modelling for Aesthetic Preferences
Investigate regression algorithms such as linear regression, polynomial regression, ridge
PO4 regression, and lasso regression. Understand their underlying principles, advantages,
limitations, and suitable applications.
PO5 Supports supervised learning tasks with distributed computing capabilities and integration
with popular libraries like Tensor Flow and Sickie-learn.
PO9 Training the selected model on the training data and evaluate its performance using team
C212.2 work for appropriate metrics
PO10 Communication skills are crucial for explaining the insights and implications derived from
supervised learning models.
PO11 Finance principles facilitate budgeting and financial planning for supervised learning
projects.
PO12 Lifelong learning encourages the integration of domain knowledge into supervised learning
models
PSO1 Understanding of the core principles and theoretical underpinnings of machine learning,
which is essential for building more advanced skills and knowledge in the field.
PSO2 Machine learning models using relevant evaluation metrics and techniques, which is crucial
for making informed decisions and improving model efficacy.
PSO3 The ability to evaluate supervised learning techniques and apply them adeptly to regression
and classification tasks with efficacy and precision is essential for leadership development.
C212.3 Identify common patterns or features in engineering data without the need for labelled
PO1
examples.
By leveraging unsupervised learning algorithms, make data-driven decisions based on the
PO2
inherent structure and relationships present in their data.
By incorporating unsupervised learning into system design can gain valuable insights into
PO3
the underlying characteristics of the data
Investigating unsupervised learning methods can help improve the accuracy and efficiency
PO4
of anomaly detection systems.
Modern tools based on their superior performance can lead to quicker analysis and more
PO5
timely insights from large datasets.
With unsupervised learning individuals have the freedom to experiment with different
PO9 algorithms, techniques, and approaches to uncover hidden patterns and insights within the
data.
Articulating the value proposition of unsupervised learning in terms of its potential to
PO10
improve decision-making, optimize processes, reduce costs, and drive innovation.
PO11 Unsupervised learning can help project managers identify patterns and anomalies in project
data
 Lifelong learners can leverage unsupervised learning to identify areas of interest, assess their
PO12
strengths and weaknesses, and design customized learning.
Theoretical foundations of Decision tree and Artificial Neural network classifier, acquire
PSO1 skills to design, analyse and develop algorithms and implement them using high-level
programming language
Mastering the implementation of deep learning models using popular frameworks in the use
PSO2
of modern tools and technologies relevant to the field of machine learning.
Unsupervised learning projects entail assessing and mitigating risks associated with model
PSO3
performance, scalability, and interpretability.
Engineering knowledge enables the selection and customization of performance metrics that
PO1
are meaningful and relevant to the problem domain.
Problem analysis helps identify the assumptions and limitations of different modelling
PO2
approaches.
Evaluators can design comprehensive validation experiments, cross-validation strategies, and
PO3 testing procedures to ensure that the models are rigorously evaluated under diverse
conditions and scenarios.
Investigations provide empirical evidence to support model evaluation and comparison,
PO4
enabling evaluators to make evidence-based decisions
Modern tools offer automated workflows, algorithms, and pipelines that streamline the
PO5
model evaluation process
PO9 Team work promotes knowledge sharing and continuous learning
PO10 Communication skills with best practices in model evaluation focuses broader community.
PO11 Finance principles facilitate budgeting and financial planning for model evaluation projects
Lifelong learners can assess the strengths and weaknesses of different models more
PO12
effectively and make informed decisions.
C212.4 Study of classifier model working acquire skills to design, analyse and develop algorithms
PSO1
and implement them using high-level programming language
Machine learning techniques to solve real-world problems, aligning with the program of
PSO2
being able to address practical challenges using their knowledge and skills.
The application of neural networks must adhere to ethical principles to ensure fairness,
PSO3
transparency, and accountability.
Equipping students with the ability to design and analyse machine learning experiments
PO1 using cross-validation and resampling methods not only strengthens their engineering
knowledge but also prepares them to address complex real-world problems
Problem analysis enables students to choose the most suitable algorithms, pre-processing
PO2 methods, and evaluation metrics based on the nature of the problem, the available data, and
the desired outcomes.
Enable students to tackle complex problems, optimize solutions for performance and
PO3
scalability, and apply machine learning techniques effectively in real-world applications.
Conducting investigations of complex systems is integral to machine learning education as it
PO4 enables students to develop a deeper understanding of the underlying data and system
dynamics.
Modern tools often incorporate the latest advancements and techniques in machine learning
PO5
research.
Integrating individual and teamwork skills into machine learning education prepares students
C212.5 PO9
for the collaborative nature of real-world machine learning projects
Communication skills in machine learning education, students develop the ability to convey
PO10
complex ideas clearly.
Project management and finance concepts into machine learning education equips students
PO11 with the skills and knowledge necessary to effectively plan, execute, and evaluate machine
learning projects.
Lifelong learning fosters adaptability and flexibility, enabling students to navigate diverse
PO12
career paths and tackle new challenges in the dynamic field of machine learning.
Modern technology facilitates hands-on learning experiences, allowing students to apply
PSO1
theoretical concepts to practical, real-world scenarios.
By considering sustainable development principles, students learn to design machine
PSO2
learning experiments that minimize data storage requirements.
Leadership helps to develop problem-solving skills, resilience and adaptability, enabling
PSO3
them to overcome challenges and achieve project goals.
GAPS IN THE SYLLABUS - TO MEET INDUSTRY/PROFESSION REQUIREMENTS, POs:
SNO DESCRIPTION PROPOSED
ACTIONS
1 Graph Neural Networks Introduction to Graph Neural
Networks will be given as
Content beyond Syllabus

PROPOSED ACTIONS: TOPICS BEYOND SYLLABUS/ASSIGNMENT/INDUSTRY VISIT/GUEST LECTURER/NPTEL ETC

TOPICS BEYOND SYLLABUS/ADVANCED TOPICS/DESIGN:

1 Transformer Models
2 Graph Neural Networks

WEB SOURCE REFERENCES:

1 https://onlinecourses.nptel.ac.in/noc24_cs60/preview
2 https://www.geeksforgeeks.org/machine-learning/

DELIVERY/INSTRUCTIONAL METHODOLOGIES:
☐ CHALK & TALK ☐ STUD. ASSIGNMENT ☐ WEB RESOURCES ☐ NPTEL/OTHERS
☐ LCD/SMART BOARDS ☐ STUD. SEMINARS ☐ ADD-ON COURSES ☐ WEBNIARS

ASSESSMENT METHODOLOGIES-DIRECT
☐ ASSIGNMENTS ☐ STUD. SEMINARS ☐ TESTS/MODEL EXAMS ☐ UNIV. EXAMINATION
☐ STUD. LAB PRACTICES ☐ STUD. VIVA ☐ MINI/MAJOR PROJECTS ☐ CERTIFICATIONS
☐ ADD-ON COURSES ☐ OTHERS

ASSESSMENT METHODOLOGIES-INDIRECT
☐ ASSESSMENT OF COURSE OUTCOMES (BY FEEDBACK, ☐ STUDENT FEEDBACK ON FACULTY (TWICE)
ONCE)
☐ ASSESSMENT OF MINI/MAJOR PROJECTS BY EXT. ☐ OTHERS
EXPERTS

INNOVATIONS IN TEACHING/LEARNING/EVALUATION PROCESSES:

1. Student Seminar
2. Animation Videos
3. Practice code

Prepared by Approved by
(Faculty) (HOD)