SEMESTER –V

MACHINE LEARNING II

Semester: 5 CIE Marks 50

Course Code 22ADS53 SEE Marks 50
Hours/Week (L: T: P) 3:2:0 Duration of SEE (hours): 03
Type of Course PC Credits 04
Prerequisites (if any): Machine Learning I
Course Learning Objectives:
Sl. No Course Learning Objectives (CLO)
Understanding of the fundamental classification algorithms and challenges of supervised
1
algorithms
2 Become conversant with types of multiclass classification algorithms, and their applicability
3 Familiarize with the unsupervised machine learning algorithms
4 To improve the performance of the algorithms using hyperparameter tuning techniques.
5 To understand and apply reinforcement learning concepts to the real data.

No. of RBT
Module 1
Hours Level
Supervised Learning (Classification):
Support Vector Machine (SVC and SVR), Kernel Methods, Random Forest, Ensemble 10 L3
classification methods (Bagging and Boosting Techniques).
Module 2
Multiclass Classification: Multiclass classification problem, Generalization bounds,
Uncombined multi-class algorithms, aggregated multi-class algorithms, Performance
Metrics. 10 L3
Explainable AI: Introduction to XAI, LIME, SHAP.
Module 3
Unsupervised Learning:
Introduction to Unsupervised Learning, Clustering, k-means Clustering, Bisecting k-
means, K-Means as special case of Expectation Maximization, Agglomerative 10 L3
Clustering and Divisive Clustering, DBSCAN, Comparing and Evaluating Clustering
Algorithms, Semi-Supervised Learning models.
Module 4
Hyperparameter Tuning: Overview, Manual Search, Grid Search, Random Search,
Random Search with Hyperopt, Bayesian Optimization, Multi-fidelity Optimization,
Optuna. 10 L3
 Module 5
Reinforcement Learning(RL):
Key elements of RL, the RL Algorithm, how RL differs from other ML paradigms,
The Markov Decision Process, Action space, Episodic and Continuous tasks, Return 10 L3
and discount factor, The Value function, Q-function, model-based and model-free
learning, types of environments, Applications.

Course Outcomes:
Upon successful completion of this course, the students will be able to

CO1 Apply supervised learning techniques to real data using classification algorithms
CO2 Apply multiclass classification learning techniques to real data
CO3 Apply unsupervised learning algorithms for prediction.

CO4 Apply hyperparameter tuning techniques to improve the performance of the model
CO5 Apply reinforcement learning algorithms to solve real-world problems.

PO PO PO PO PO PO PO PO PO PO- PO- PO- PSO PSO

-1 -2 -3 -4 -5 -6 -7 -8 -9 10 11 12 -1 -2
CO1 2 2 2 3 3 2 3 2 3
CO2 2 2 2 3 3 2 3 2 3
CO3 3 2 3 3 3 2 3 3 2 2
CO4 3 2 3 3 3 2 3 3 2 3
CO5 3 3 3 3 3 3 3 3 2 2
Average 3 2 3 3 3 2 3 3 2 3

High-3: Medium-2: Low-1

Text Books:

1. Introduction to Machine Learning with Python‖ Sarah Guido, Andreas C. Müller, O‘ Reilly, 2017.
2. Deep Reinforcement Learning with Python – Sudharsan Ravichandiran, by Packt Publishing – 2nd
edition 2020.
3. Principles of Data Science by Sinan Ozdemir, Sunil Kakade, Packt Publishing Limited,
2nd Edition, 2018.
Reference Books:

1. Principles of Soft Computing by S N Sivanandam and S N Deepa, 3rd Edition, Wiley.

2. Introduction to Machine Learning‖, by Ethem Alpaydin, PHI Learning, 2nd Edition, 2019.

E-Books / Web References:

 1. https://towardsdatascience.com/hyperparameter-tuning-for-machine-learning-models-
1b80d783b946.
2. https://smartlabai.medium.com/reinforcement-learning-algorithms-an-intuitive-overview-
904e2dff5bbc.

MOOCs:

1. Udemy.
2. Coursera
3. NPTEL

Scheme of Examination:
Semester End Examination (SEE):
SEE Question paper is to be set for 100 marks and the marks scored will be proportionately
reduced to 50. There will be two full questions (with a maximum of four sub-questions) from each
module carrying 20 marks each. Students are required to answer any five full questions choosing at least
one full question from each module.

Continuous Internal Evaluation (CIE):

Three Tests are to be conducted for 40 marks each. Average marks scored is added to test component.
CIE is executed by way of two quizzes / Alternate Assessment Tools (AATs), and Three tests. Two
quizzes are to be conducted and each quiz is evaluated for 5 marks adding up to 10 marks.

All quizzes are conducted online. Faculty may adopt innovative methods for conducting quizzes
effectively.

Some possible AATs: seminar/ assignments/term paper/ open ended experiments/ mini-projects/ concept
videos/ partial reproduction of research work/ oral presentation of research work/ group activity/
developing a generic toolbox for problem solving/ report based on participation in create-a-thon/ make-a-
thon/ code-a-thon/ hack-a-thon conducted by reputed organizations/ any other.

Typical Evaluation pattern for regular courses is shown in Table.

Table: Distribution of weightage for CIE & SEE of Regular courses

Component Marks Total Marks

CIE Test-1 30
CIE Test-2 30
CIE CIE Test-3 30 50
Lab 20
SEE Semester End Examination 50 50
Grand Total 100
 MACHINE LEARNING II LAB

Semester: 5 CIE Marks 50

Course Code 22ADSL54(Lab) SEE Marks 50
Hours/Week (L:T:P) 0:0:2 Duration of SEE (hours): 03
Type of Course PC Credits 01

Prerequisites: Basics of Probability

Course Learning Objectives: The course will enable students to:

CLO1 Implement the advanced programming skills of Python.

CLO2 To develop skills to analyze Unsupervised, Semi-supervised machine learning algorithms.

CLO3 To strengthen the ability to the students to identify and apply the suitable algorithm for the
given real-world problem.
CLO4 Enables to gain knowledge in practical applications of machine learning.

Sl.N O Experiments

1 Write a program to demonstrate Support Vector Machine using different Kernel

functions and compare the performance with the other ML algorithms.

2 Write a program to implement Bagging and Boosting classifiers and compare the
results with the single classifiers.
3 Write a program to demonstrate pipeline in Machine Learning.

4 Write a program to classify the data using Multiclass classification algorithm 1.

5 Write a program to classify the data using Multiclass classification algorithm 2.

6 Write a program to cluster the data using K-Means clustering algorithm.

7 Write a program to implement Label Propagation algorithm ( Semi – Supervised

Learning
)
8 Write a program to demonstrate Random Forest algorithm and improve the
performance using different Hyper Parameter Tuning Techniques (Randomized and
Grid search CV).
9 Write a program to demonstrate on classification algorithm and improve the
performance using different Hyper Parameter Tuning Techniques (Bayesian and
Optuna).
10 Write a program to implement Q-Learning.
Course Outcomes: Upon successful completion of this course, students will be able to

CO1 Understand, appreciate and effectively explain the underlying concepts of Machine Learning
algorithms.
CO2 Code, debug and demonstrate the working nature of different types of Ensemble algorithms.

CO3 Choose the appropriate Unsupervised learning algorithm on any given dataset.

CO4 Implement Semi-supervised learning algorithms on real-world problems.

CO5 Implement Q-Learning Reinforcement Learning algorithm on any gaming dataset.

Conduct of Practical Examination:

1. All laboratory experiments are to be included for practical examination.

2. Breakup of marks and the instructions printed on the cover page of the answer script to
be strictly adhered by the examiners.
3. Students can pick one experiment from the questions lot prepared by the examiners.
4. Change of experiment is allowed only once and 15% Marks allotted to the
procedure part to be made zero.

Scheme of Examination (CIE):

In order to encourage innovative methods while delivering a course, the faculty members have
been encouraged to use the Alternative Assessment Tool (AAT). The AAT enhances the
autonomy (freedom and flexibility) of individual faculty and enables them to create
innovative pedagogical practices.

Possible AATs are - seminar/ assignments/term paper/ open ended experiments/ mini-projects/
concept videos/ partial reproduction of research work/ oral presentation.