Version 1.

3
CS4811: Artificial Intelligence
Lab Manual

Capital University of Science &

Technology

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY

Department of Computer Science
1
 Lab Course Development Team

Supervision and Dr. M. Abdul Qadir

Coordination Professor/ Dean Faculty of
Computing

Lab Designer Ms. Syeda Amna Rizwan

Associate Lecturer Faculty of

Computing

Ms. Snober Naseer

Associate Lecturer Faculty of

Computing
Lab Reviewer Mr. Salman Ahmed

Senior Lecturer Faculty of

Computing

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 2

 Course Outline

Weekly Plan Topics Pg. no

Week 1: Lab 1 Python for Beginners 4

Week 2 : Lab 2 Python Libraries and Practices 17

Week 3 : Lab 3 Uninformed Search: Breadth First Search and Depth First 23
Search
Week 4 : Lab 4 A* Search 30

Week 5 : Lab 5 Minimax Algorithm 39

Week 6 : Lab 6 Alpha Beta Pruning 46

Week 7 : Lab 7 Hill Climbing 57

Week 8: Lab 8 Constraint Satisfaction Problem 65

Week 9 Mid Term

Week 10 : Lab 9 First Order Logic (FOL) Inference by using Python Libraries 74

Week 11 : Lab 10 Deep Learning Inference 83

Week 12 : Lab 11 Basic Machine Learning & Naïve Bayes 91

Week 13 : Lab 12 Supervised Learning and Unsupervised Learning 99

Week 14 : Lab 13 Performance Metrics in Machine Learning 108

Week 15 : Lab 14 Natural Language Processing 115

Week 16 Final Term

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 3

 Lab Manual for Artificial Intelligence

Lab-1: Python for Beginners

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 4

 1. Introduction to Python

Python is a widely used general-purpose, high level programming language. It was

created by Guido van Rossum in 1991 and further developed by the Python Software
Foundation. It was designed with an emphasis on code readability, and its syntax allows
programmers to express their concepts in fewer lines of code. Python is a programming
language that lets you work quickly and integrate systems more efficiently.

Python is:

 Interpreted: it can execute at runtime, and changes in a program are instantly

perceptible. To be very technical, Python has a compiler. The difference when compared
to Java or C++ is how transparent and automatic it is. With Python, we don't have to
worry about the compilation step as it's done in real-time. The tradeoff is that interpreted
languages are usually slower than compiled ones.

 Semantically Dynamic: you don't have to specify types for variables and there is
nothing that makes you do it.

 Object-Oriented: everything in Python is an object. But you can choose to write code
in an object-oriented, procedural, or even functional way.

 High level: you don't have to deal with low-level machine details.
This lab is a course on the introduction of Python language. The lab allows you to learn
the core of the language in a matter of hours instead of weeks.
1.1. Use of Python

Python has been growing a lot recently partly because of its many uses in the following
areas:

 System Scripting: it's a great tool to automate everyday repetitive tasks.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 5

  Data Analysis: it is a great language to experiment with and has tons of libraries and
tools to handle data, create models, visualize results and even deploy solutions. This is
used in areas like Finance, E-commerce, and Research.
 Web Development: frameworks like Django and Flask allow the development of web
applications, API's, and websites.
 Machine Learning: Tensorflow and Pytorch are some of the libraries that allow
scientists and the industry to develop and deploy Artificial Intelligence solutions in
Image Recognition, Health, Self-driving cars, and many other fields.

1.2. Syntax
1.2.1. Semicolons

Python doesn't use semicolons to finish lines. A new line is enough to tell the interpreter
that a new command is beginning.

The print () method will display something.

In this example, we have two commands that will display the messages inside the single
quotes.

But the following is wrong due to the semicolons in the end:

1.2.2. Indentation

Many languages use curly-brackets to define scope.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 6

 Python's interpreter uses only indentation to define when a scope ends and another one
starts.

This means you have to be aware of white spaces at the beginning of each line -- they
have meaning and might break your code if misplaced.

This definition of a function works:

This doesn't work because the indentation of the second line is missing and will throw
an error:

1.2.3. Case Sensitivity and Variables

This Python is case sensitive. So the variables name and Name are not the same thing
and store different values.

As you can see, variables are easily created by just assigning values to them using the =
symbol.

This means name stores 'Renan' and Name stores 'Moura'.

1.2.4. Comments
Finally, to comment something in your code, use the hash mark #.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 7

 The commented part does not influence the program flow.

1.2.5. Types
To store data in Python you need to use a variable. And every variable has its type
depending on the value of the data stored. Python has dynamic typing, which means you
don't have to explicitly declare the type of your variable -- but if you want to, you can.
Lists, Tuples, Sets, and Dictionaries are all data types and have dedicated sections later
on with more details, but we'll look at them briefly here.

1.2.6. User Input

If you need to interact with a user when running your program in the command line (for
example, to ask for a piece of information), you can use the input () function.

1.2.7. Operators

In a programming language, operators are special symbols that you can apply to your
variables and values in order to perform operations such as arithmetic/mathematical and
comparison. Python has lots of operators that you can apply to your variables and I will
demonstrate the most used ones.

1.2.7.1. Arithmetic Operators

Arithmetic operators are the most common type of operators and also the most
recognizable ones. They allow you to perform simple mathematical operations. They are:

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 8

  +: Addition
 -: Subtraction
 *: Multiplication
 /: Division
 **: Exponentiation
 //: Floor Division, rounds down the result of a division
 %: Modulus, gives you the remainder of a division

Let's see a program that shows how each of them is used.

1.2.7.2. Concatenation

Concatenation is when you have two or more strings and you want to join them into one.
This is useful when you have information in multiple variables and want to combine
them.

For instance, in this next example I combine two variables that contain my first name
and my last name respectively to have my full name.

The + operator is used to concatenate.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 9

 1.2.7.3. Comparison Operators

Use comparison operators to compare two values. These operators return either True or
False. They are:

 ==: Equal
 !=: Not equal
 >: Greater than
 <: Less than
 >=: Greater than or equal to
 <=: Less than or equal to

Let's see a program that shows how each of them is used.

1.2.7.4. Logical Operators

Logical operators are used to combine statements applying Boolean algebra. They are:

 and: True only when both statements are true

 or: False only when both x and y are false
 not: The not operator simply inverts the input, True becomes False and vice versa.

Let's see a program that shows how each one is used.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 10

 1.2.8. Conditionals

Conditionals are one of the cornerstones of any programming language. They allow you
to control the program flow according to specific conditions you can check.

1.2.8.1. if statement

The way you implement a conditional is through the if statement. The general form of
an if statement is:

The expression contains some logic that returns a Boolean, and the statement is
executed only if the return is True. A simple example:

1.2.8.2. if else or elif statement

In our last example, the program only does something if the condition returns True.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 11

 But we also want it to do something if it returns False or even check a second or third
condition if the first one wasn't met.

In this example, we swapped Bob's and Sarah's age. The first condition will return
False since Sarah is older now, and then the program will print the phrase after the
else instead.

Now, consider the example below with the elif.

1.3. Objective of the Experiment

After completing this lab, the students should be able to:

 Understand Python and its syntax in detail.

1.4. Practice Task
1. Go to the URL
https://www.jetbrains.com/pycharm/download/?section=windows#section=windows . To
download Pycharm python IDE and install it.

2. For Python installation, go to the URL

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 12

 https://www.python.org/downloads/release/python-373/ and install Windows x86-64
executable installer.

3. Open Pycharm and create new project named Example1 by clicking on file in menu bar
and then on new project.

4. After creating new project right click on project name and create a python file name first
program as shown in figures below.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 13

 5. When file created write following code

6. Run file by clicking on Run in menu bar

7. After that you can see output in output box. Function returned 2.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 14

 1.5. Evaluation Tasks

1.5.1. Task 1 (Marks 10)

Write a python program to print the multiplication table for a number and range given
by a user?
For example: the number is 2 and the range is 11 to 13 and the output will be:
2 x 11 = 22
2 x 12 = 24
2 x 13 = 26

1.5.2. Task 2 (Marks 10)

Write a python program using function to check whether the positive integer is prime or
not. The number will be given by the user that lies between 0 to 1000. For example, for
input 5 the output will be displayed as “prime number”. For input 6, the output will be
“not a prime number”.

1.5.3. Task 3 (Marks 10)

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 15

 Write a Python program that accepts a positive integer (n) and computes the sum of
squares from 1 to n. For example: n is 3 so the sum will be 14 and for n is 5 the sum will
be 55.

1.6. Evaluation using Rubrics

The evaluation criteria for this lab will be based on the completion of the following tasks.
Each task is assigned the marks percentage which will be evaluated by the instructor in
the lab whether the student has finished the complete/partial task(s).
Criteria Level 0 Level 1 Level 2 Level 3 Level 4 Total
(0%) (1-50)% (51-65)% (66-80)% (81-100)% Score
Procedural Awareness
Practical Implementation
Program Correctness
Use of Software Tool
Sub Total Marks

1.7. Out comes

After completing this lab, students will be able to learn Python syntax and implement
different Python concepts in detail.

1.8. Further Readings

1.8.1. Books
Artificial Intelligence: A modern approach 3rd Edition by Stuart Russell & Peter Norvig

1.8.2. Links
https://pynative.com/python-basic-exercise-for-beginners/

https://www.w3resource.com/python-exercises/

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 16

 Lab Manual for Artificial Intelligence

Lab-2: Python Libraries & Practices

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 17

 2. Introduction to Python Libraries

A library is a collection of pre-written code that you can use to perform specific tasks.
Libraries are often used to reduce the amount of code a programmer needs to write by
providing reusable functions or classes that can be called upon as needed.

All libraries are divided into two classifications. All libraries are either Static or
Dynamic.

 Static: These are compiled into a program’s executable code, making it self-
contained and independent of external changes. Static libraries are linked with a
program at compile time, resulting in a single executable file that includes all of the
code from the library. As a result, static libraries are often used to improve the
performance of a program by avoiding the overhead of loading a library at runtime.

 Dynamic: Dynamic libraries are loaded at runtime rather than linked with a program
at compile time. This allows programs to use the library without having to be re-
compiled if the library is updated. In addition, dynamic libraries are often used to
provide platform-specific functionality or to allow multiple programs to share a single
copy of the library in memory.

You can write libraries in most programming languages, and they can be used by
programs written in that same language or a different language. For example, a library
written in C++ could be used by a program written in Python.

Libraries can perform various tasks, such as file input and output, data manipulation,
network communication, and more. They can also provide access to system resources,
such as the file system or hardware devices.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 18

 2.1. Python Libraries in AI

Python libraries play a significant role in making AI the technology they are today. Here
are some of the well-known Python libraries that are leveraged in the development of
Artificial Intelligence applications:

 Numpy: It is a popular Python library used to handle multi-dimensional data and

complex mathematical functions being used on the data. The NumPy library powers the
speed of computation of mathematical expressions and execution of complex functions
working on arrays. Discretion of Fourier transformations, Statistical operations, and
linear algebra, Support for n-dimensional arrays, Data cleaning and manipulation, and
Random simulations.
 Pandas: Pandas is a prominent Python library generally used for Machine Learning
concepts. It is basically a data analysis library that analyses and manipulates the data.
Pandas make it easier for the developers to work with structured multidimensional data
and time series concepts and produce efficient results, Data alignment and handling of
missing data, merging and joining of datasets, Dataset reshaping and pivoting, data
filtration, data manipulation and analysis, indexing of the data.
 Matplotlib: It is a data visualization library used for designing plots and graphs. The
library itself is an extension of SciPy and handles complex data models of Pandas as
well as NumPy data structures. Matplotlib offer features such as Basemap, GTK tools,
Cartopy, Mplot4d, etc., that help in generating image plots, 3D plots, contour plots, and
more. High-quality diagrams, plots, histograms, graphs, etc,Intuitive and easy to
use,GUI toolkit support,Map projections,Recognition of data patterns.
 SciPy: It is a Python library that originates from NumPy. SciPy is leveraged by Python
development services to perform technical and scientific computing on large sets of
data. The library has built-in array optimization and linear algebra modules that help in

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 19

 scientific analysis and engineering. Array manipulation subroutines, User-friendliness,
Data visualization and manipulation, Scientific and technical analysis.
 Scikit-learn: It is a powerful Python library that was originally generated to serve the
purpose of data modelling and building machine learning algorithms. It has a simple,
engaging, and consistent interface that is exceptionally user-friendly, making it easy to
use and share data. Data modelling, End-to-end Machine Learning algorithms, Model
selection, Classification of data, Dimensionality reduction, and Pre-processing of the
data.
 TensorFlow: It is an open-source ML library used for reaching data and production
purposes. The library is offered by Google and can be used to make ML model building
easier. TensorFlow offers a flexible framework and architecture that enables it to run
on various computational platforms. However, it has its own tensor processing unit
(TPU) through which the best results can be obtained. Creating deep learning models,
Natural Language Processing, Abstraction capabilities, Managing deep neural
networks, Image, text, and speech recognition.
 Keras: It is an open-source Python neural network library that is offered as an extension
to the TensorFlow library. Keras is designed for building and evaluating neural
networks within machine learning and deep learning models. The library is flexible,
modular, and extensible and can be integrated with objectives, optimizers, layers, and
activation functions. Activation and cost functions, Data pooling, developing neural
layers, Batch normalization, and Building deep learning models.
 PyTorch: It is a highly popular production-ready Python library generally used in
Machine Learning concepts. The library supports GPU acceleration and provides
performance optimization to deep neural networks. PyTorch is mostly used for boosting
the performance of deep learning frameworks and is supported by a vast Python
community. Development of deep learning models, Matrix-vector multiplication,

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 20

 Statistical distribution and operations, Greater control over datasets, and Python
libraries are highly beneficial when working with AI technologies.

2.2. Objective of the Experiment

After completing this lab, the students should be able to:

 Understand different Python libraries, its usage and practical implementation.

2.3. Evaluation Tasks

2.3.1. Task 1 (Marks 10)

Write a code to create a basic line plot using matplotlib.

2.3.2. Task 2 (Marks 10)

Write a code to read a CSV file into a pandas data Frame?
Note: You can take any CSV file of your choice from the given link
https://www.kaggle.com/datasets?fileType=csv
2.4. Evaluation using Rubrics

The evaluation criteria for this lab will be based on the completion of the following tasks.
Each task is assigned the marks percentage which will be evaluated by the instructor in
the lab whether the student has finished the complete/partial task(s).
Criteria Level 0 Level 1 Level 2 Level 3 Level 4 Total
(0%) (1-50)% (51-65)% (66-80)% (81-100)% Score
Procedural Awareness
Practical Implementation
Program Correctness
Use of Software Tool
Sub Total Marks

2.5. Out comes

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 21
 After completing this lab, students will be able to learn Python syntax and implement
different Python concepts in detail.

2.6. Further Readings

2.6.1. Books
Artificial Intelligence: A modern approach 3rd Edition by Stuart Russell & Peter Norvig

2.6.2. Links
https://www.edureka.co/blog/python-libraries/

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 22

 Lab Manual for Artificial Intelligence

Lab-3: Uninformed Search: Breadth First

Search, Depth First Search

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 23

 3. Introduction

The objective of this lab is to introduce the concept of uninformed search algorithms and
their usage. Uninformed search is a class of general-purpose search algorithms which
operates in brute force-way. Uninformed search algorithms do not have additional
information about state or search space other than how to traverse the tree, so it is also
called blind search. Following are some of the types of uninformed search algorithms:

 Breadth-first Search

 Depth-first Search
3.1. Breadth First Search (BFS)

Breadth-first search (BFS) is an algorithm for traversing or searching tree or graph

data structures. It starts at the tree root and explores all of the neighbor nodes at the present
depth prior to moving on to the nodes at the next depth level. It uses a queue (First in First
Out) instead of a stack and it checks whether a vertex has been discovered before enqueueing
the vertex rather than delaying this check until the vertex is dequeued from the queue.

3.2. Depth First Search (DFS)

Depth-first search is an algorithm for traversing or searching tree or graph data structures.
The algorithm starts at the root node (selecting some arbitrary node as the root node in the
case of a graph) and explores as far as possible along each branch before backtracking. So the
basic idea is to start from the root or any arbitrary node and mark the node and move to the
adjacent unmarked node and continue this loop until there is no unmarked adjacent node.
Then backtrack and check for other unmarked nodes and traverse them. Finally print the
nodes in the path.
3.3. Relevant Lecture Readings
Revise Lecture Uninformed search

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 24

 3.4. Objectives of the experiment

 To get knowledge of implementing Breadth First Search, Depth First Search and
Uniform Cost Search

 To know the difference between the above mentioned uninformed search algorithms

3.5. Concept Map

This concept map will help students to understand the main concepts of topics covered
in the lab. In this lab, students will be provided with a detailed content to learn that how
built in libraries can be used to implement concept of BFS, DFS and Uniform Cost Search
in python.

3.6. Breadth First Search (BFS) Pseudo Code

BFS (G, s) //Where G is the graph and s is the source node

let Q be queue.

Q.enqueue(s) //Inserting s in queue until all its neighbor vertices are marked.
mark s as visited.
while (Q is not empty)
//Removing that vertex from queue, whose neighbor will be visited
now
v = Q.dequeue( )//processing all the neighbours of v for all neighbours w of v in
Graph G if w is not visited
Q.enqueue( w )//Stores w in Q to further visit its neighbor mark w as visited.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 25

 3.7. Depth First Search (DFS) Implementation

DFS-iterative (G, s)//Where G is graph and s is source vertex let S be

stack
S.push(s )//Inserting s in stack mark s as visited.
while (S is not empty):
//Pop a vertex from stack to visit next
v= S.top( )
S.pop( )
//Push all the neighbours of v in stack that are not visited for
all neighbours w of v in Graph G:
if w is not visited :
S.push( w ) //mark w as visited
DFS-recursive(G, s) //mark s as visited
for all neighbours w of s in Graph G: if w is not visited:
DFS-recursive(G, w)

3.8. Homework before Lab

Revise the concepts of BFS and DFS.

3.9. Tools

Pycharm

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 26

 3.10. Practice Task
Implementation of BFS in Python.

3.11. Evaluation Tasks

This section will provide more practice exercises which you need to finish during the
lab. You need to finish the tasks in the required time.

3.11.1. Task 1 (Marks 10)

Write a Python function to perform topological sorting of a directed acyclic graph (DAG)

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 27

 using BFS. You are required to write a Python function to count the number of connected
components in an undirected graph using BFS.

Test Case 1

3.11.2. Task 2 (Marks 10)

You are given the following tree whose starting node is A and Goal node is F.

You are required to implement the Depth First Search on the given graph in Python.

3.11.3. Task 3 (Marks 10)

Implement Water Jug Problem using the DFS in Python, initial stage (0,0) and final stage
(2,Y). A Water Jug Problem: You are given two jugs, a 4-gallon one and a 3- gallon one,
a pump which has unlimited water which you can use to fill the jug, and the ground on
which water may be poured. Neither jug has any measuring markings on it. How can you
get exactly 2 gallons of water in the 4-gallon jug?

3.12. Evaluation using Rubrics

The evaluation criteria for this lab will be based on the completion of the following tasks.
Each task is assigned the marks percentage which will be evaluated by the instructor in

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 28

 the lab whether the student has finished the complete/partial task(s).
Criteria Level 0 Level 1 Level 2 Level 3 Level 4 Total
(0%) (1-50)% (51-65)% (66-80)% (81-100)% Score
Procedural Awareness
Practical Implementation
Program Correctness
Use of Software Tool
Sub Total Marks

3.12.1. Out comes

After completing this lab, students will know implementation of uninformed search i-e.,
BFS and DFS in Python.

3.13. Further Readings

3.13.1. Books
Artificial Intelligence: A modern approach 3rd Edition by Stuart Russell & Peter Norvig

3.13.2. Links
https://www.geeksforgeeks.org/breadth-first-search-or-bfs-for-a-graph/
https://www.geeksforgeeks.org/depth-first-search-or-dfs-for-a-graph/
https://pythoninwonderland.wordpress.com/2017/03/18/how-to-implement-breadth-first-search-in-
python/
https://en.wikipedia.org/wiki/Depth-first_search
https://www.hackerearth.com/practice/algorithms/graphs/depth-first-search/tutorial/

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 29

 Lab Manual for Artificial Intelligence

Lab-04: A* Search

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 30

 4. Introduction

In this lab you will learn about the concepts of A* Search. This algorithm is one of the best
and popular technique used in path-finding and graph traversals. The idea is avoid expanding
nodes which are expansive.

4.1. Relevant Lecture Readings

Revise Lectures A* Search Algorithm, Admissibility and Monotonicity conditions

4.2. Objective of the Experiment

After completing this lab, the student should be able to:

 Clearly understand difference between uninformed searches and informed searches.

 Better understanding of heuristic function.

 Understand search mechanism where cost is involved.

4.3. Concept Map

This concept map will help students to understand the main concepts of topics covered in the
lab. In this lab, students will be provided with a detailed content to learn that how built in
libraries can be used to implement concept of A* search algorithm in python.

4.4. A* Search

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 31

 1. Numbers written on edges represent the distance between nodes. Numbers written on
nodes represent the heuristic value.

2. Given the graph, find the cost-effective path from A to G. That is A is the source node
and G is the goal node.

3. Now from A, we can go to point B or E, so we compute f(x) for each of them,

4. A → B = g(B) + h(B) = 2 + 6 = 8

5. A → E = g(E) + h(E) = 3 + 7 = 10

6. Since the cost for A → B is less, we move forward with this path and compute the f(x)
for the children nodes of B.

7. Now from B, we can go to point C or G, so we compute f(x) for each of them,

8. A → B → C = (2 + 1) + 99= 102

9. A → B → G = (2 + 9 ) + 0 = 11

10. Here the path A → B → G has the least cost but it is still more than the cost of A → E,
thus we explore this path further.

11. Now from E, we can go to point D, so we compute f(x),

12. A → E → D = (3 + 6) + 1 = 10

13. Comparing the cost of A → E → D with all the paths we got so far and as this cost is
least of all we move forward with this path.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 32

 14. Now compute the f(x) for the children of D

15. A → E → D → G = (3 + 6 + 1) +0 = 10

16. Now comparing all the paths that lead us to the goal, we conclude that A → E → D →
G is the most cost-effective path to get from A to

4.5. Homework before Lab

Attempt the following questions. You are required to bring this task with you and submit
to the lab instructor.

4.5.1. Problem description

Design the solution of the following 8 puzzle problem given in figure 4.2. Your solution
must contain all steps until goal state not find as described the solution design in section
4.1.

Figure 4.2: 8 Puzzle Problem

4.6. Procedure & Tools

4.6.1. Tools

Pycharm

4.7. Practice Task

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 33

 This is a direct implementation of A* on a graph structure given below in Figure 4.3. The
heuristic function is defined as 1 for all nodes for the sake of simplicity and brevity. The graph
is represented with an adjacency list, where the keys represent graph nodes, and the values
contain a list of edges with the corresponding neighboring nodes.

Figure 4.3

Write the given code in .py file in Pycharm and run it.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 34

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 35
CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 36
 OUTPUT

4.8. Evaluation Tasks

This section will provide more practice exercises which you need to finish during the
lab. You need to finish the tasks in the required time.

4.8.1. Task 1 (Marks 10)

Implement the A* code in python on the following test cases (for help see the
walkthrough task). In test case 1, the starting node is 8 and the last node is 6 whereas in
test case 2 the starting node is 0 and the ending node is 4.

4.8.2. Task 2 (Marks 10)

Implement the 8 Puzzle Problem using A* search in Python.

The puzzle consists of an area divided into a grid, 3 by 3 for the 8-puzzle. On each grid

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 37

 square is a tile, expect for one square which remains empty. Thus, there are eight tiles in the
8-puzzle. A tile that is next to the empty grid square can be moved into the empty space,
leaving its previous position empty in turn. Tiles are numbered, 1 thru 8 for the 8-puzzle, so
that each tile can be uniquely identified. Heuristic for 8-puzzle problem is Number of
Misplaced Tiles. You can design your own heuristic.

4.8.3. Out comes

After completing this lab, students will to know implementation of A* search algorithm.

4.9. Evaluation Criteria

The evaluation criteria for this lab will be based on the completion of the following tasks.
Each task is assigned the marks percentage which will be evaluated by the instructor in
the lab whether the student has finished the complete/partial task(s).
Criteria Level 0 Level 1 Level 2 Level 3 Level 4 Total
(0%) (1-50)% (51-65)% (66-80)% (81-100)% Score
Procedural Awareness
Practical Implementation
Program Correctness
Use of Software Tool
Sub Total Marks

4.10. Further Readings

4.10.1. Books
Artificial Intelligence: A modern approach 3rd Edition by Stuart Russell & Peter Norvig

4.10.2. Links
https://www.geeksforgeeks.org/a-search-algorithm/
https://www.annytab.com/a-star-search-algorithm-in-
python/#:~:text=The%20A*%20search%20algorithm%20uses,heuristic%20to%20gui
de%20the%20search.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 38

 Lab Manual for Artificial Intelligence

Lab-05: Min-max Algorithm

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 39

 5. Introduction
Games have always been an important application area for heuristic algorithms. In
playing games whose state space may be exhaustively delineated, the primary difficulty
is in accounting for the actions of the opponent. This can be handled easily by assuming
that the opponent uses the same knowledge of the state space as us and applies that
knowledge in a consistent effort to win the game. Minmax implements game search
under referred to as MIN and MAX.

5.1. Relevant Lecture Readings

Revise Lectures Game Playing and Mini-max algorithm

5.2. Objective of the Experiment

After completing this lab, the student should be able to:

 Clearly understand the concept of Min-Max algorithm and able to use an optimization
technique for min-max algorithm (Alpha-Beta).
5.3. Concept map
In this lab, we’re going to discuss Minimax algorithm and its applications in Artificial
Intelligence. As it’s a game theory algorithm, we’ll implement a simple game using it.

5.4. Min-Max Algorithm

Min-max is a decision-making algorithm, typically used in a turn-based, two player games.
The goal of the algorithm is to find the optimal next move. In the algorithm, one player is
called the maximizer, and the other player is a minimizer. If we assign an evaluation score
to the game board, one player tries to choose a game state with the maximum score, while
the other chooses a state with the minimum score.

In other words, the maximizer works to get the highest score, while the minimizer tries
get the lowest score by trying to counter moves. Examples of such games are chess,
poker, checkers, tic-tac-toe.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 40

 Our goal is to find the best move for the player. To do so, we can just choose the node
with best evaluation score. To make the process smarter, we can also look ahead and
evaluate potential opponent’s moves. For each move, we can look ahead as many moves
as our computing power allows. The algorithm assumes that the opponent is playing
optimally. Technically, we start with the root node and choose the best possible node.
We evaluate nodes based on their evaluation scores. In our case, evaluation function can
assign scores to only result nodes (leaves). Therefore, we recursively reach leaves with
scores and back propagate the scores. Consider the below game tree:

Maximizer starts with the root node and chooses the move with the maximum score.
Unfortunately, only leaves have evaluation scores with them, and hence the algorithm
has to reach leaf nodes recursively. In the given game tree, currently it’s the minimizer’s
turn to choose a move from the leaf nodes, so the nodes with minimum scores (here,
node 3 and 4) will get selected. It keeps picking the best nodes similarly, till it reaches
the root node.

5.5. Home Task

Read and understand the concept of minmax algorithm. Try to design the solution of any
example on paper and bring it in lab.

5.6. Procedure & Tools

Pycharm
5.7. Practice Task
Consider a game which has 4 final states and paths to reach final state are from root to 4

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 41

 leaves of a perfect binary tree as shown below. Assume you are the maximizing player
and you get the first chance to move, i.e., you are at the root and your opponent at next
level.

Since this is a backtracking based algorithm, it tries all possible moves, then backtracks
and makes a decision.

 Maximizer goes LEFT: It is now the minimizers turn. The minimizer now has a choice
between 3 and 5. Being the minimizer it will definitely choose the least among both, that
is 3
 Maximizer goes RIGHT: It is now the minimizers turn. The minimizer now has a choice
between 2 and 9. He will choose 2 as it is the least among the two values.
Being the maximizer you would choose the larger value that is 3. Hence the optimal
move for the maximizer is to go LEFT and the optimal value is 3.

Now the game tree looks like the above figure, and the above tree shows two possible
scores when maximizer makes left and right moves.

Note: Even though there is a value of 9 on the right sub tree, the minimizer will never
pick that. We must always assume that our opponent plays optimally. Below is the
implementation for the same.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 42

 Output:

5.8. Evaluation Tasks

This section will provide more practice exercises which you need to finish during the
lab. You need to finish the tasks in the required time.

5.8.1. Task 1 (Marks 10)

Apply the following algorithms on the tree below (use python).

 Min-Max

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 43

 5.9.1. Task 2 (Marks 10)
Implement tic-tac-toe game in python using min-max algorithm.
5.9.2. Out comes
After completing this lab, students will to know about Minimax algorithm and its
applications in Artificial Intelligence.

5.10. Evaluation Criteria

The evaluation criteria for this lab will be based on the completion of the following
tasks. Each task is assigned the marks percentage which will be evaluated by the
instructor in the lab whether the student has finished the complete/partial task(s).
Criteria Level 0 Level 1 Level 2 Level 3 Level 4 Total
(0%) (1-50)% (51-65)% (66-80)% (81-100)% Score
Procedural Awareness
Practical Implementation
Program Correctness
Use of Software Tool
Sub Total Marks

5.11. Further Readings

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 44

 5.11.1. Books
Artificial Intelligence: A modern approach 3rd Edition by Stuart Russell & Peter Norvig

5.11.2. Links
https://www.geeksforgeeks.org/minimax-algorithm-in-game-theory-set-1-introduction/

http://homepage.ufp.pt/jtorres/ensino/ia/alfabeta.html

https://stackabuse.com/minimax-and-alpha-beta-pruning-in-python/

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 45

 Lab Manual for Artificial Intelligence

Lab-06: Alpha-beta Pruning

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 46

 6. Introduction
Alpha-beta pruning is a modified version of the minimax algorithm. It is an optimization
technique for the minimax algorithm. The basic idea behind this modification to the min-
max search algorithm is the following. During the process of searching for the next move,
not every move (i.e. every node in the search tree) needs to consider in order to reaching
a correct decision. Alpha-beta pruning can be applied at any depth of a tree, and
sometimes it not only prune the tree leaves but also entire sub-tree.
The two-parameter can be defined as:
 Alpha: The best (highest-value) choice we have found so far at any point along
the path of Maximizer. The initial value of alpha is -∞.

 Beta: The best (lowest-value) choice we have found so far at any point along the
path of Minimizer. The initial value of beta is +∞.

The main condition which required for alpha-beta pruning is: α>=β

6.1. Relevant Lecture Readings

Revise Lecture alpha-beta pruning

6.2. Objective of the Experiment

After completing this lab, the student should be able to:
 Clearly understand the concept of Min-Max algorithm and able to use an
optimization technique for min-max algorithm (Alpha-Beta).
6.3. Concept map
In this lab, we’re going to the advantages of using the algorithm and see how it can be
improved it using Alpha-Beta Pruning. As it’s a game theory algorithm, we’ll implement
a simple game using it. We’ll also discuss.

6.4. Alpha Beta Pseudo Code

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 47

 .

6.5. Working of Alpha-Beta Pruning

Let's take an example of two-player search tree to understand the working of Alpha-beta
pruning.

Step 1: At the first step the Max player will start first move from node A where α= -∞
and β= +∞, these value of alpha and beta passed down to node B where again α= -∞ and
β= +∞, and Node B passes the same value to its child D.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 48

 Step 2: At Node D, the value of α will be calculated as its turn for Max. The value of α
is compared with firstly 2 and then 3, and the max (2, 3) = 3 will be the value of α at
node D and node value will also 3.

Step 3: Now algorithm backtrack to node B, where the value of β will change as this is
a turn of Min, Now β= +∞, will compare with the available subsequent nodes value, i.e.
min (∞, 3) = 3, hence at node B now α= -∞, and β= 3.

In the next step, algorithm traverse the next successor of Node B which is node E, and
CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 49
 the values of α= -∞, and β= 3 will also be passed.

Step 4: At node E, Max will take its turn, and the value of alpha will change. The current
value of alpha will be compared with 5, so max (-∞, 5) = 5, hence at node E α= 5 and β=
3, where α>=β, so the right successor of E will be pruned, and algorithm will not traverse
it, and the value at node E will be 5.

Step 5: At next step, algorithm again backtrack the tree, from node B to node A. At node
A, the value of alpha will be changed the maximum available value is 3 as max (-∞, 3)=
3, and β= +∞, these two values now passes to right successor of A which is Node C.

At node C, α=3 and β= +∞, and the same values will be passed on to node F.

Step 6: At node F, again the value of α will be compared with left child which is 0, and
max(3,0)= 3, and then compared with right child which is 1, and max(3,1)= 3 still α
remains 3, but the node value of F will become 1.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 50

 Step 7: Node F returns the node value 1 to node C, at C α= 3 and β= +∞, here the value
of beta will be changed, it will compare with 1 so min (∞, 1) = 1. Now at C, α=3 and β=
1, and again it satisfies the condition α>=β, so the next child of C which is G will be
pruned, and the algorithm will not compute the entire sub-tree G.

Step 8: C now returns the value of 1 to A here the best value for A is max (3, 1) = 3.
Following is the final game tree which is the showing the nodes which are computed and
nodes which has never computed. Hence the optimal value for the maximizer is 3 for
this example.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 51

 6.6. Home Task

Read and understand the concept of alpha beta pruning. Try to design the solution of any
example on paper and bring it in lab.

6.7. Procedure & Tools

Pycharm
6.8. Practice Task
Here is the implementation of alpha beta pruning.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 52

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 53
 Output:

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 54

 6.9. Evaluation Tasks
This section will provide more practice exercises which you need to finish during the
lab. You need to finish the tasks in the required time.
6.9.1. Task 1 (Marks 10)
Apply the following algorithm on the tree below (use python).

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 55

 6.9.2. Task 2 (Marks 10)
How can you represent a game tree in Python for a game like Connect Four or Checkers?
Implement the alpha-beta pruning algorithm for one of these games.
6.9.3. Outcome
After completing this lab, students will to know about Minimax algorithm and its
applications in Artificial Intelligence

6.10. Evaluation Criteria

The evaluation criteria for this lab will be based on the completion of the following
tasks. Each task is assigned the marks percentage which will be evaluated by the
instructor in the lab whether the student has finished the complete/partial task(s).
Criteria Level 0 Level 1 Level 2 Level 3 Level 4 Total
(0%) (1-50)% (51-65)% (66-80)% (81-100)% Score
Procedural Awareness
Practical Implementation
Program Correctness
Use of Software Tool
Sub Total Marks

6.11. Further Readings

6.11.1. Books
Artificial Intelligence: A modern approach 3rd Edition by Stuart Russell & Peter Norvig

6.11.2. Links
https://www.geeksforgeeks.org/minimax-algorithm-in-game-theory-set-1-introduction/

http://homepage.ufp.pt/jtorres/ensino/ia/alfabeta.html
https://stackabuse.com/minimax-and-alpha-beta-pruning-in-python/

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 56

 Lab Manual for Artificial Intelligence

LAB-07: Hill Climbing

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 57

 7. Introduction

In the field of AI, many complex algorithms have been used. It is also important to find out
an optimal solution. Hill climbing algorithm is one such optimization algorithm used in the
field of Artificial Intelligence. It is a mathematical method which optimizes only the
neighboring points and is considered to be heuristic. A heuristic method is one of those
methods which does not guarantee the best optimal solution. This algorithm belongs to the
local search family.

Local search algorithms are used on complex optimization problems where it tries to find
out a solution that maximizes the criteria among candidate solutions. A candidate
solution is considered to be the set of all possible solutions in the entire functional region
of a problem.

7.1. Relevant Lecture Readings

Revise Lectures Hill Climbing Algorithm (Simple hill Climbing, Steepest-Ascent hill
climbing)

7.2. Objective of the Experiment

After completing this lab, the student should be able to:

 Clearly understand Hill climbing problems in artificial intelligence.

7.3. Concept Map

This concept map will help students to understand the main concepts of topics covered
in the lab. In this lab, students will be provided with a detailed content to learn that what
the concept of hill climbing is and how to implement it in python.

7.4. Overview

Hill Climbing is a heuristic search used for mathematical optimization problems in the

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 58

 field of Artificial Intelligence.
Given a large set of inputs and a good heuristic function, it tries to find a sufficiently
good solution to the problem. This solution may not be the global optimal maximum.

7.5. Working Process

This algorithm works on the following steps in order to find an optimal solution.

 It tries to define the current state as the state of starting or the initial state.

 It generalizes the solution to the current state and tries to find an optimal solution. The
solution obtained may not be the best.

 It compares the solution which is generated to the final state also known as the goal state.

 It will check whether the final state is achieved or not. If not achieved, it will try to find
another solution.

7.6. Types of Hill Climbing

There are various types of Hill Climbing which are-

 Simple Hill climbing

 Steepest-Ascent Hill climbing

 Stochastic Hill climbing

7.7. Applications of Hill climbing technique

7.7.1. Robotics

Hill climbing is mostly used in robotics which helps their system to work as a team and
maintain coordination.

7.7.2. Marketing

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 59

 The algorithm can be helpful in team management in various marketing domains where
hill climbing can be used to find an optimal solution. The travelling time taken by a sale
member or the place he visited per day can be optimized using this algorithm.

7.8. Homework before Lab

Attempt the following questions. You are required to bring this task with you and submit
to the lab instructor.

7.8.1. Problem description

Read and under the detailed concepts of hill climbing algorithm and their usage in AI
field.

7.9. Procedure & Tools

Pycharm

7.10. Practice Task

We will perform a simple study in Hill Climbing on a greeting “Hello World!”. We will
see how the hill climbing algorithm works on this. We will generate random solutions
and evaluate our solution. If the solution is the best one, our algorithm stops; else it will
move forward to the next step. Write the given code in .py file and run it.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 60

 Output:

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 61

7.11. Evaluation Tasks
This section will provide more practice exercises which you need to finish during the
lab. You need to finish the tasks in the required time.

7.11.1. Task 1 (Marks 10)

Implement the following goal state finding problem using hill climbing algorithm.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 62

 Hint: Key point while solving any hill-climbing problem is to choose an appropriate
heuristic function.

7.11.2. Task 2 (Marks 10)

Consider a 2D grid with a starting point (0, 0) and a target point (5, 5). You are allowed
to move either up, down, left, or right by one unit at a time. Implement the Hill Climbing
algorithm in Python to find the shortest path from the starting point to the target point.

Remember to consider edge cases, handle obstacles if any, and optimize your
implementation for efficiency.

7.11.3. Out comes

After completing this lab, students will to know how to deal with Hill climbing problem.

7.12. Evaluation Criteria

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 63

 The evaluation criteria for this lab will be based on the completion of the following tasks.
Each task is assigned the marks percentage which will be evaluated by the instructor in
the lab whether the student has finished the complete/partial task(s).
Criteria Level 0 Level 1 Level 2 Level 3 Level 4 Total
(0%) (1-50)% (51-65)% (66-80)% (81-100)% Score
Procedural Awareness
Practical Implementation
Program Correctness
Use of Software Tool
Sub Total Marks

7.13. Further Readings

7.13.1. Books

Artificial Intelligence: A modern approach 3rd Edition by Stuart Russell & Peter Norvig

7.13.2. Links
https://www.mygreatlearning.com/blog/an-introduction-to-hill-climbing-algorithm/

https://www.geeksforgeeks.org/introduction-hill-climbing-artificial-intelligence/
https://www.baeldung.com/java-hill-climbing-algorithm

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 64

 77

Lab Manual for Artificial Intelligence

Lab-08: Constraint Satisfaction Problems

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 65

 8. Introduction

A constraint satisfaction problem (CSP) consists of

 A set of variables,
 A domain for each variable, and
 A set of constraints.
The aim is to choose a value for each variable so that the resulting possible world satisfies
the constraints; we want a model of the constraints.

8.1. Relevant Lecture Readings

Revise Lecture Constraint Satisfaction Problem

8.2. Objective of the Experiment

After completing this lab, the student should be able to:

 Clearly understand constraint satisfaction problems in artificial intelligence.

8.3. Concept Map

This concept map will help students to understand the main concepts of topics covered
in the lab. In this lab, students will be provided with a detailed content to learn that how
built in libraries can be used to implement concept of CSP in python.

8.4. Overview
Formally speaking, a constraint satisfaction problem (or CSP) is defined by a set of
variables, {X1; X2; : : : ; Xn}, and a set of constraints, {C1; C2; : : : ; Cm}. Each variable
Xi has a nonempty domain Di of possible values. Each constraint Ci involves some
subset of the variables and specifies the allowable combinations of values for that subset.
A state of the problem is defined by an assignment of values to some or all of the
variables,{Xi=vi; Xj=vj; : : :}. An assignment that does not violate any constraints is
called a consistent or legal assignment. A complete assignment is one in which every

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 66

 variable is mentioned, and a solution to a CSP is a complete assignment that satisfies all
the constraints.

8.5. Varieties of constraints

8.5.1. Unary Constraints
This is the simplest type of constraint, e.g. unary constraint can be. South Australia can’t
be assigned green color.
SA ≠ green or WA = red
8.5.2. Binary Constraints
Binary constraint relates two variables, e.g. SA ≠ NSW is a binary constraint that
involves two pairs of variables.

8.5.3. Higher-order Constraints

Higher-order constraints involves two or more variables e.g. SA ≠ NSW ≠ Q. We will
see detailed examples of this constraint in following map coloring problem.

We are looking at the map of Australia showing each of its states and territories as shown
in above figure (a), we are given the task of coloring each region either red, green or blue
in such a way that no neighboring regions have the same color.

To formulate this CSP, we define the variables to the regions: WA, NT, Q, NSW, V,
SA, T.

The domain of each variable is the set {red, green, blue}.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 67

 The constraints require neighboring regions to have distinct colors, e.g. if WA is
given red color then NT and SA can’t have red color, their domain will be limited to
{green, blue}.

It is helpful to visualize a CSP as a constraint graph as shown in figure (b). The nodes of
the graph correspond to variables of the problem.

We can give incremental formulation as a standard search problem as follows:

 Initial State: the empty assignment {}, in which all variables are unassigned.
 Successor Function: a value can be assigned to any unassigned variable, provided
that it doesn’t conflict with previously assigned variables.
 Goal Test: the current assignment is complete.
 Path Cost: a constraint cost e.g 1 for every step.
To solve the above problem we use backtracking. Backtracking search is used for a depth
first search that chooses values for one variable at a time and backtracks when a variable
has no legal values left to assign.

Assign domain (colors) to one variable at a time. Continue…

…

But backtracking is not efficient yet. We can improve efficiency of backtracking using
some General-Purpose methods. Most Constraint Variable

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 68

 1. Most Constraining Variable
2. Least Constraint Value
8.5.4. Most Constraint Variable
Pick the variable with Minimum Remaining Values (MRV). E.g. WA can have values
{red, green, blue}, and NT can have values {red, green}, so you have to choose NT first.

8.5.5. Most Constraining Variable

Choose the variable with the most constraints on remaining variables (most edges in
graph)

8.5.6. Least Constraint Value

LSV guides the choice of which value to assign next. Choose the least constraining value:

Now let’s perform the general rules on the above problem.

WA NT SA Q NSW V T

Initial R,G,B R,G,B R,G,B R,G,B R,G,B R,G,B R,G,B

Domain
SA = Red G,B G,B R G,B G,B G,B R,G,B

NT = Blue G B R G B G R,G,B

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 69

 T = Red G B R G B G R

8.6. Homework before Lab

Attempt the following questions. You are required to bring this task with you and submit
to the lab instructor.

8.6.1. Problem description

Design the solution of the following CSP also make table to show the process of coloring
on paper given in figure 5.1. Your solution must contain all steps until the whole map is
not colored as described in the solution design in section 4.2.3.

Figure 10.1: CSF Problem

8.7. Procedure & Tools
Pycharm

8.8. Practice Task

Write the following code in .py file and run it.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 70

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 71
 Output:

8.9. Evaluation Tasks

This section will provide more practice exercises which you need to finish during the
lab. You need to finish the tasks in the required time.

8.9.1. Task 1 (Marks 10)

Implement the home task problem in python.

8.9.2. Task 2 (Marks 10)

The eight queen’s puzzle is the problem of placing eight chess queens on an 8×8
chessboard so that no two queens threaten each other; thus, a solution requires no two queens
share the same row, column, or diagonal. Formulate this problem as a CSP problem.

8.9.3. Out comes

After completing this lab, students will to know how to deal with CSP.

8.10. Evaluation Criteria

The evaluation criteria for this lab will be based on the completion of the following tasks.
Each task is assigned the marks percentage which will be evaluated by the instructor in
the lab whether the student has finished the complete/partial task(s).
Criteria Level 0 Level 1 Level 2 Level 3 Level 4 Total
(0%) (1-50)% (51-65)% (66-80)% (81-100)% Score
Procedural Awareness
Practical Implementation
Program Correctness

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 72

 Use of Software Tool
Sub Total Marks

8.11. Further Readings

8.11.1.Books

Artificial Intelligence: A modern approach 3rd Edition by Stuart Russell & Peter Norvig

8.11.2. Links
https://www.geeksforgeeks.org/m-coloring-problem-backtracking-5/
https://freecontent.manning.com/constraint-satisfaction-problems-in-python/

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 73

 Lab Manual for Artificial Intelligence
Lab-09: First Order Logic (FOL) inference
by using Python Libraries

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 74

 9. Introduction
First-order logic, often abbreviated as FOL or FOL, is a formal system used in
mathematics, philosophy, computer science, and artificial intelligence to represent and
reason about statements and relationships between objects in a structured and precise
manner. It's also known as first-order predicate logic, first-order predicate calculus, or
simply predicate logic.
First-order logic is a fundamental logic system that extends propositional logic (which
deals with simple true/false statements) to handle more complex statements involving
objects, relations, and quantifiers. Here are some key components and concepts of first-
order logic:
1. Syntax: In first-order logic, you work with symbols to represent various elements. These
symbols include variables, constants, predicates, functions, and logical connectives.
2. Variables: Symbols that can represent any object, like 'x', 'y', 'z'.
3. Constants: Symbols representing specific objects, like 'John', 'Mary'.
4. Predicates: Symbols representing relationships or properties, like 'IsRed', 'IsParentOf'.
5. Functions: Symbols representing operations that return objects, like 'Add', 'Multiply'.
6. Logical Connectives: Symbols for logical operations, including 'and' (∧), 'or' (∨), 'not'
(¬), 'implies' (→), and 'if and only if' (↔).
7. Quantifiers: Symbols like ∀ (for all) and ∃ (there exists) used to make statements about
entire sets of objects.
8. Semantics: First-order logic provides a way to interpret these symbols and assign
meaning to statements through a formal semantics. It defines how predicates, functions,
and quantifiers behave in terms of truth values.
9. Formulas: Sentences or statements in first-order logic are called formulas. These can be
simple, like "IsRed(Apple)", or complex, involving quantifiers, conjunctions,
disjunctions, and implications.
10.Quantifiers: Quantifiers are a fundamental aspect of first-order logic. They allow you
to make statements about sets of objects.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 75

 o The universal quantifier (∀) represents "for all" and asserts that a statement holds true
for all objects in a given set.
o The existential quantifier (∃) represents "there exists" and asserts that there is at least
one object in a given set for which a statement holds true.
11.Axioms and Inference Rules: First-order logic is typically accompanied by a set of
axioms and inference rules that govern how to derive new statements from existing ones.
Common inference rules include modus ponens and universal generalization.
12.Soundness and Completeness: First-order logic is both sound and complete. Soundness
means that if a conclusion is derived using valid inference rules, it is necessarily true.
Completeness means that any statement that is logically valid within the language of
first-order logic can be proven using the rules of the system.
First-order logic is a powerful tool for formal reasoning and forms the basis for various
AI systems, knowledge representation languages, and automated theorem provers. It
provides a rigorous framework for expressing and reasoning about a wide range of real-
world problems and concepts.

9.1. Relevant Lecture Readings

Revise Lectures First Order Logic and its components, Inference in FOL, Reasoning in
AI using FOL.

9.2. Objective of the Experiment

The lab aims to equip students with the knowledge and skills to work effectively with
First-Order Logic and use it as a tool for formal reasoning and problem-solving.

9.3. Concept Map

This concept map will help students to understand the main concepts of topics covered
in the lab. In this lab, students will be provided with a detailed content to learn that how
built in libraries can be used for First Order Logic implementation.

9.4. Libraries

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 76

 If you want to work with First-Order Logic (FOL) in Python, you can use various
libraries and tools to help you represent, manipulate, and reason with FOL statements.
Here are some popular libraries and tools for working with FOL in Python:
1. SymPy:
o SymPy is a powerful Python library for symbolic mathematics.
o It can be used to create and manipulate symbolic expressions, including those in FOL.
o SymPy provides support for logic operations, quantifiers, and proof techniques.
Example: You can use SymPy to create FOL formulas and perform logical reasoning.

2. Pyke:
o Pyke is a knowledge-based inference engine (expert system) that uses FOL for
knowledge representation and reasoning.
o It's designed for building expert systems and rule-based applications.
o Pyke can be used to define knowledge bases and rules using FOL.
3. Prover9/Mace4:
o Prover9 and Mace4 are theorem proving tools that can handle FOL.
o While they are standalone programs, you can use Python to interact with them and
automate theorem proving tasks.
4. NetworkX is a Python library for creating and analyzing complex networks.
o It can be used to model and work with graphs and networks, which can be a useful
representation for certain FOL applications.
5. Pyke Knowledge Engine:
o Pyke Knowledge Engine is a Python library that provides an integrated environment for

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 77

 rule-based programming and knowledge representation using FOL.
6. Natural Language Processing Libraries:
o Libraries like spaCy, NLTK, and Stanford NLP can be used to extract FOL-like
representations from natural language text, which can be a useful starting point for FOL
reasoning.
When choosing a library or tool for your FOL coding needs in Python, consider your
specific requirements and whether you need basic FOL manipulation, complex theorem
proving, expert system development, or natural language understanding capabilities.
Your choice may also depend on the complexity of the FOL tasks you need to
accomplish.

9.5. Tools
Prolog is a popular programming language and logic programming environment used for
implementing and working with logic-based systems. There are several Prolog
implementations (also called Prolog compilers or Prolog systems) available, each with
its own features and advantages. Here are some commonly used Prolog implementations:
1. SWI-Prolog:
o SWI-Prolog is a widely used and actively maintained Prolog system.
o It is open-source and available on multiple platforms, including Windows, macOS, and
Linux.
o SWI-Prolog has a rich set of libraries and extensions, making it suitable for various
applications.
o It includes a built-in development environment with a graphical debugger and a web-
based IDE called SWISH.
2. GNU Prolog:
o GNU Prolog is an open-source Prolog compiler designed to be compatible with the ISO
standard for Prolog.
o It is available on various platforms, including Unix-like systems, Windows, and macOS.
o GNU Prolog focuses on being a lightweight and efficient Prolog system.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 78

 3. YAP (Yet another Prolog):
o YAP is a high-performance Prolog compiler that emphasizes efficiency.
o It is available on Unix-like systems, Windows, and macOS.
o YAP offers various extensions and libraries for advanced Prolog programming.
4. SICStus Prolog:
o SICStus Prolog is a commercial Prolog system known for its high-performance and
advanced features.
o It offers a wide range of libraries and tools for applications like constraint logic
programming and natural language processing.
o SICStus Prolog is available for Windows, Linux, and macOS.
5. ECLiPSe:
o ECLiPSe is a versatile constraint logic programming system that extends Prolog.
o It is often used for constraint satisfaction problems and optimization tasks.
o ECLiPSe supports several programming paradigms, including constraint logic
programming (CLP).
The choice of Prolog implementation depends on factors such as your specific project
requirements, platform compatibility, performance considerations, and whether you
prefer open-source or commercial solutions. SWI-Prolog and GNU Prolog are often
recommended for beginners due to their accessibility and robust community support.

9.6. Practice Task

In this example, we will use the SymPy library, which provides excellent support for
symbolic mathematics and logic. Here's how to implement FOL using SymPy:
1. Step 1: Install SymPy
If you haven't already installed SymPy, you can do so using pip:

2. Step 2: Define Symbols

In SymPy, you first need to define symbols to represent variables and constants in your

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 79

 FOL statements:

3. Step 3: Define Predicates

Predicates represent relationships or properties. You can define predicates as functions
of symbols:

4. Step 4: Create FOL Formulas

You can create FOL formulas using these predicates, logical connectives, and
quantifiers:

5. Step 5: Perform Logical Operations

You can use SymPy's logical operators to perform operations on FOL formulas:

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 80

 6. Step 6: Check Validity
SymPy can be used to check the validity of FOL formulas:

7. Step 7: Substitution and Evaluation

You can substitute values into FOL formulas and evaluate them:

9.7. Evaluation Tasks

This section will provide more practice exercises which you need to finish during the
lab. You need to finish the tasks in the required time.

9.7.1. Task 1 (Marks 10)

Translate the following English sentence into FOL and implement it in Python using
SymPy:

1. "Every cat is an animal."

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 81

 2. "If it is raining, then the ground is wet."

3. "For all x, if x is a person, then there exists a y such that y is the parent of x."

9.7.2. Task 2 (Marks 10)

To write the negation of the FOL statement "Some birds can fly" in Python using the
SymPy library, you can use the Not operator to negate the entire statement.

9.7.3. Out comes

After completing this lab, students will know the practical implementation of FOL using
Python libraries.

9.8. Evaluation Criteria

The evaluation criteria for this lab will be based on the completion of the given tasks.
Each task is assigned the marks percentage which will be evaluated by the instructor in
the lab whether the student has finished the complete/partial task(s).
Criteria Level 0 Level 1 Level 2 Level 3 Level 4 Total
(0%) (1-50)% (51-65)% (66-80)% (81-100)% Score
Procedural Awareness
Practical Implementation
Program Correctness
Use of Software Tool
Sub Total Marks

9.9. Further Readings

9.9.1. Books
Artificial Intelligence: A modern approach 3rd Edition by Stuart Russell & Peter
Norvig

9.9.2. Links
https://avigad.github.io/lamr/implementing_first_order_logic.html

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 82

 Lab Manual for Artificial Intelligence
Lab-10: Deep Learning Inference

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 83

 10. Introduction

Deep learning inference refers to the process of using a trained deep learning model to
make predictions or decisions based on new, unseen data. In deep learning, models are
typically composed of multiple layers of artificial neurons (like neural networks) that
have been trained on large datasets to learn complex patterns and relationships within
the data.

Here's a step-by-step explanation of deep learning inference:

1. Training: Initially, a deep learning model is trained on a labeled dataset. During training,
the model learns to recognize patterns and make predictions based on the input data and
associated labels. This involves adjusting the model's internal parameters (weights and
biases) through a process called backpropagation and optimization techniques such as
gradient descent.

2. Model Deployment: Once the model is trained and its parameters are optimized, it can be
deployed for inference. Deployment can occur on various platforms, including cloud
servers, edge devices (like smartphones or IoT devices), or even embedded systems.

3. Inference: During inference, the trained model is presented with new, unseen data (input).
The model processes this input data through its layers of neurons, and the output is generated
based on what the model has learned during training. This output can take different forms,
depending on the nature of the task. For example, in image classification, the output might
be a label identifying the object in the image.

4. Decision-Making: The output generated by the model during inference is used for decision-
making or prediction. In applications like image recognition, natural language processing,
and autonomous driving, deep learning models are used to make decisions based on the
input data. For instance, a deep learning model in a self-driving car might use inference to
detect obstacles and make decisions about steering and braking.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 84

 5. Feedback Loop: In some cases, the output of the model during inference can be used to
improve the model itself. This can be done by collecting feedback data, adjusting model
parameters, and retraining the model periodically to keep it up-to-date and accurate.

Deep learning inference is a critical component of many modern applications, including

image and speech recognition, natural language understanding, recommendation systems,
and autonomous systems, among others. It enables these systems to make real-time
decisions and predictions based on the patterns and knowledge they have learned from
training data.

10.1. Relevant Lecture Readings

Revise Lecture Probabilistic Reasoning in AI and Bayes Theorem in AI.

10.2. Objective of the Experiment

After completing this lab, the student should be able to:

 Clearly understand deep learning models in artificial intelligence.

10.3. Concept Map

This concept map will help students to understand the main concepts of topics covered
in the lab. In this lab, students will be provided with a detailed content to learn that what
the concept of deep learning is and how to implement these models in python.

10.4. Types of Deep Learning Models

There are several types of deep learning models, each designed for specific tasks and
data types. Here are some common types:

1. Feedforward Neural Networks (FNN): Also known as multi-layer perceptron (MLPs),

these are the foundational deep learning models composed of input, hidden, and output
layers. They're used for various tasks like classification and regression.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 85

 2. Convolutional Neural Networks (CNN): Primarily designed for image recognition
tasks, CNNs utilize specialized layers like convolutional and pooling layers to
automatically learn and recognize patterns in images.

3. Recurrent Neural Networks (RNN): Suited for sequence data, RNNs have connections
that loop back on themselves, allowing them to process sequences in a time-dependent
manner. Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) are
popular RNN variants.

4. Transformer Models: Introduced by the "Attention Is All You Need" paper,

transformers have revolutionized natural language processing tasks. They use attention
mechanisms to process sequences of data in parallel, enabling efficient processing of
long-range dependencies. Models like BERT, GPT, and T5 are based on transformers.

5. Auto encoders: Used for unsupervised learning, auto encoders learn to encode data into
a compressed representation and then decode it back to the original data. They're used
for tasks like data denoising, dimensionality reduction, and anomaly detection.

6. Generative Adversarial Networks (GAN): GANs consist of two networks—the

generator and the discriminator—competing in a game-like setting. They're used for
generating new data instances that resemble a given dataset, leading to applications in
image and text generation.

7. Variational Auto encoders (VAE): VAEs combine auto encoders with probabilistic
modeling, enabling the generation of new data instances while exploring the underlying
data distribution.

8. Siamese Networks: These networks are used for tasks like similarity and distance
measurement. They have two or more identical subnetworks, sharing weights, and are
often used for tasks like face recognition and signature verification.

9. Graph Neural Networks (GNN): Designed to work with graph-structured data, GNNs
can capture information from nodes and their connections in graphs, making them useful

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 86

 for applications like social network analysis and recommendation systems.

10.Attention-Based Models: Beyond transformers, attention mechanisms are used in

various contexts to weigh the importance of different elements in a sequence, such as in
image captioning and visual question answering.

These are just a few examples of the diverse range of deep learning models. Each type
is designed to address specific types of data and tasks, and researchers continue to
develop new architectures to tackle increasingly complex problems.

10.5. Procedure & Tools

Pycharm

10.6. Practice Task

We will perform a simple practice task to build and train a deep learning model in Python
using TensorFlow and Keras. In this example, we'll create a feedforward neural network
for image classification using the famous MNIST dataset of handwritten digits.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 87

 In this code:

1. We import TensorFlow and Keras.

2. We load the MNIST dataset of handwritten digits, which consists of 28x28 pixel images.

3. We preprocess the data by scaling pixel values to the range [0, 1].

4. We build a feedforward neural network model using Keras. This model has an input layer
that flattens the 28x28 images into 1D arrays, a hidden layer with ReLU activation, a
dropout layer to reduce overfitting, and an output layer with 10 neurons (one for each
digit) using softmax activation.

5. We compile the model, specifying the optimizer ('adam'), loss function

('sparse_categorical_crossentropy' for multiclass classification), and evaluation metric
('accuracy').

6. We train the model on the training data for 10 epochs with a validation split of 20%.

7. Finally, we evaluate the model's accuracy on the test data and print the result.

The output will show the test accuracy of the trained model, indicating how well it
performs on classifying handwritten digits in the MNIST dataset.

10.7. Evaluation Tasks

This section will provide more practice exercises which you need to finish during the

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 88

 lab. You need to finish the tasks in the required time.

10.7.1. Task 1 (Marks 10)

Load a dataset of your choice (e.g., MNIST, CIFAR-10, Oxford Flowers 102) using
Python and a CNN deep learning framework. Preprocess the data, including
normalization, splitting into training and testing sets, and data augmentation if
applicable.

10.7.2. Task 2 (Marks 10)

Train the neural network on the training data. Monitor training progress by visualizing
training and validation loss and accuracy. Set an appropriate number of epochs and batch
size. For example: You are required to train a CNN model for image classification. The
dataset you are using is Oxford Flowers 102, epoch’s ranges from 10 to 100 and batch
size ranges from 32 to 256.

10.7.3. Out comes

After completing this lab, students will to know how to deal with different deep learning
models.

10.8. Evaluation Criteria

The evaluation criteria for this lab will be based on the completion of the following tasks.
Each task is assigned the marks percentage which will be evaluated by the instructor in
the lab whether the student has finished the complete/partial task(s).
Criteria Level 0 Level 1 Level 2 Level 3 Level 4 Total
(0%) (1-50)% (51-65)% (66-80)% (81-100)% Score
Procedural Awareness
Practical Implementation
Program Correctness
Use of Software Tool
Sub Total Marks

10.9. Further Readings

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 89

 10.9.1. Books

Artificial Intelligence: A modern approach 3rd Edition by Stuart Russell & Peter
Norvig

10.9.2. Links
https://colab.research.google.com/drive/19YwLLREPmE0ZFoneEKQpSkXRIIF9pPP

https://www.geeksforgeeks.org/introduction-hill-climbing-artificial-intelligence/

https://www.mathworks.com/discovery/deep-learning.html

https://www.projectpro.io/article/deep-learning-projects-for-beginners/441

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 90

 Lab Manual for Artificial Intelligence
Lab-11: Basic ML and Naïve Bayes

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 91

 11. Introduction

Machine learning is a subfield of artificial intelligence (AI) that focuses on developing

algorithms and models that enable computers to learn from and make predictions or
decisions based on data, without being explicitly programmed to perform a specific task. In
other words, it's about creating systems that can improve their performance on a given task
through experience (data) rather than through explicit programming.

Here are some key aspects of machine learning:

 Data-Driven Learning: Machine learning algorithms learn from data. You provide a
machine learning model with a dataset that includes examples (input data) and their
corresponding outcomes (labels or target values), and the model learns patterns and
relationships within the data.

 Generalization: A primary goal of machine learning is to create models that can generalize
well to new, unseen data. This means that the model should be able to make accurate
predictions or decisions on data it hasn't encountered during training.

 Automated Learning: Machine learning models automatically adapt and improve their
performance as they receive more data. This adaptability is a key feature, as it allows models
to handle changing and evolving environments.

 Applications: Machine learning is used in a wide range of applications, including image

and speech recognition, natural language processing, recommendation systems, fraud
detection, autonomous vehicles, medical diagnosis, and much more.

11.1. Use of Machine Learning

 Complex Tasks: Machine learning excels at tasks that are difficult for humans to program
explicitly due to their complexity or the need to process vast amounts of data. For example,
it's challenging to write precise rules for recognizing objects in images, but machine learning
models can learn to do this from labeled image data.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 92

  Pattern Recognition: Machine learning is excellent at recognizing patterns and making
predictions based on those patterns. This ability is valuable in fields like finance for
predicting stock prices, in healthcare for diagnosing diseases, and in marketing for
predicting customer behavior.

 Automation: Machine learning can automate repetitive tasks, allowing humans to focus on
more creative and complex aspects of their work. For example, in customer support,
chatbots powered by machine learning can handle routine inquiries, freeing up human
agents for more complex customer interactions.

 Personalization: Machine learning is used to create personalized experiences, such as

personalized recommendations on streaming platforms, personalized marketing messages,
and even personalized medicine.

In summary, machine learning is a powerful tool for solving complex problems and making
predictions or decisions based on data. Its ability to learn from data and generalize to new
situations makes it a valuable technology in a wide range of fields and applications.

11.2. Naïve Bayes

The Bayesian Classification represents a supervised learning method as well as a

statistical method for classification. Assumes an underlying probabilistic model and it
allows us to capture uncertainty about the model in a principled way by determining
probabilities of the outcomes. It can solve diagnostic and predictive problems. This
Classification is named after Thomas Bayes (1702-1761), who proposed the Bayes
Theorem. Bayesian classification provides practical learning algorithms and prior
knowledge and observed data can be combined.

Bayesian Classification provides a useful perspective for understanding and evaluating

many learning algorithms. It calculates explicit probabilities for hypothesis and it is
robust to noise in input data.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 93

 Naive Bayes classifiers are a collection of classification algorithms based on Bayes’
Theorem. It is not a single algorithm but a family of algorithms where all of them share
a common principle, i.e. every pair of features being classified is independent of each
other.
Bayes theorem can be written as:

11.2.1. Types of Naive Bayes Classifier

 Multinomial Naive Bayes

This is mostly used for document classification problem, i.e. whether a document
belongs to the category of sports, politics, technology etc. The features/predictors used
by the classifier are the frequency of the words present in the document.

 Bernoulli Naive Bayes

This is similar to the multinomial Naive Bayes but the predictors are Boolean variables.
The parameters that we use to predict the class variable take up only values yes or no, for
example if a word occurs in the text or not.
 Gaussian Naive Bayes

When the predictors take up a continuous value and are not discrete, we assume that
these values are sampled from a Gaussian distribution. Naive Bayes algorithms are
mostly used in sentiment analysis, spam filtering, and recommendation systems etc.
They are fast and easy to implement.

11.3. Relevant Lecture Readings

Revise Lecture No. Bayes Theorem in AI and Introduction to ML

11.4. Objective of the Experiment

After completing this lab, the student should be able to:

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 94

  Different concepts of machine learning and implementation of Naïve Bayes theorem.

11.5. Concept Map

This concept map will help students to understand the main concepts of topics covered
in the lab. In this lab, students will be provided with a detailed content to learn that what
the concept of machine learning is and how to implement different ML algorithms in
python.

11.6. Procedure & Tools

Pycharm

11.7. Homework before Lab

Attempt the following questions. You are required to bring this task with you and submit
to the lab instructor.

11.7.1. Problem description

Apply the given formula on the given dataset Figure 1 to classify whether the day is
suitable for playing golf or not. The columns represent these features and the rows
represent individual entries. If we take the first row of the dataset, we can observe that is
not suitable for playing golf if the outlook is rainy, temperature is hot, humidity is high
and it is not windy. We make two assumptions here, one as stated above we consider that
these predictors are independent. That is, if the temperature is hot, it does not necessarily

mean that the humidity is high. Another assumption made here is that all the predictors
have an equal effect on the outcome. That is, the day being windy does not have more
importance in deciding to play golf or not. According to this example, Bayes theorem
can be rewritten as:
The variable y is the class variable (play golf), which represents if it is suitable to play
golf or not given the conditions. Variable X represent the parameters/features.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 95

 Figure 2: Dataset

11.8. Practice Task

Implementing a Naive Bayes classifier in Python is relatively straightforward using

popular libraries like Scikit-Learn (sklearn). Scikit-Learn provides a simple and efficient
implementation of various Naive Bayes algorithms, including Gaussian Naive Bayes,
Multinomial Naive Bayes, and Bernoulli Naive Bayes, which are suitable for different
types of data.

Here's a basic example of how to implement Naive Bayes classification using Scikit-
Learn:

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 96

 Output:

Accuracy: 95.67

11.9. Evaluation Tasks

This section will provide more practice exercises which you need to finish during the
lab. You need to finish the tasks in the required time.

11.9.1. Task 1 (Marks 10)

You can use a dataset of emails labeled as spam or not spam (ham).
Build a Naive Bayes classifier to classify emails as spam or not spam based on their
content (e.g., subject, body). You may need to preprocess the text data, extract relevant

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 97

 features, and represent the data appropriately.

11.9.2. Task 2 (Marks 10)

Create your own dataset for a classification task of your choice. It could be related to a
topic or domain you are interested in, such as sports, movies, or food.
Design the dataset, gather data, and define classes or labels. Then, implement a Naive
Bayes classifier to perform the classification task.
11.9.3. Out comes

After completing this lab, students will to know how to deal with Naïve Bayes classifier.

11.10. Evaluation Criteria

The evaluation criteria for this lab will be based on the completion of the following tasks.
Each task is assigned the marks percentage which will be evaluated by the instructor in
the lab whether the student has finished the complete/partial task(s)

Criteria Level 0 Level 1 Level 2 Level 3 Level 4 Total

(0%) (1-50)% (51-65)% (66-80)% (81-100)% Score
Procedural Awareness
Practical Implementation
Program Correctness
Use of Software Tool
Sub Total Marks

11.11. Further Readings

11.11.1. Books
Artificial Intelligence: A modern approach 3rd Edition by Stuart Russell & Peter Norvig

11.11.2. Links
https://www.geeksforgeeks.org/naive-bayes-classifiers/
https://towardsdatascience.com/naive-bayes-classifier-81d512f50a7c

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 98

 Lab Manual for Artificial Intelligence
Lab-12: Supervised & Unsupervised Learning

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 99

 12. Introduction

Supervised learning and unsupervised learning are two fundamental approaches in machine
learning, and they differ in terms of their objectives and how they use labeled or unlabeled
data. Here's a comparison of the two with examples:

1. Supervised Learning:

 Objective: In supervised learning, the goal is to learn a mapping from input data to known
output labels. It involves training a model to make predictions or classifications based on
labeled training data.

 Labeled Data: Supervised learning requires a dataset with both input features and
corresponding output labels.

Examples:

Classification: Classifying emails as spam or not spam. Given a dataset of emails, each
labeled as spam or not spam, a supervised learning algorithm learns to classify new,
unlabeled emails.

Regression: Predicting house prices based on features like square footage, number of
bedrooms, and location. The model learns to predict the price (a continuous value) from
the input features.

Image Classification: Identifying objects in images. Given a dataset of images with

labeled objects (e.g., cats, dogs, cars), the model learns to classify new images.

2. Unsupervised Learning:

 Objective: Unsupervised learning aims to discover patterns or structure within unlabeled

data. It does not have access to explicit output labels and often involves grouping or
clustering similar data points.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 100

  Unlabeled Data: Unsupervised learning uses data without any predefined output labels.

Examples:

Clustering: Grouping similar data points together. For instance, in customer segmentation,
you might have data on customer behavior but not their segment. Unsupervised learning can
cluster customers based on their behavior to identify different market segments.

Dimensionality Reduction: Reducing the number of features while retaining important

information. Principal Component Analysis (PCA) is an example used to reduce the
dimensionality of data for visualization or compression.

Anomaly Detection: Identifying unusual or rare data points in a dataset. This can be used
in fraud detection, where unusual financial transactions need to be flagged.

Example Comparison:

Imagine you have a dataset containing customer purchase history, and you want to analyze
it:

Supervised Learning: If you have labeled data with information about which customers
made a purchase (output label) and other customer attributes (input features), you can use
supervised learning to build a model to predict which new customers are likely to make a
purchase. This would be a classification task.

Unsupervised Learning: If you only have the purchase history without labels indicating
whether each customer made a purchase or not, you can use unsupervised learning to cluster
customers based on their purchase behavior. This might reveal distinct customer segments,
even though you don't know what those segments represent in advance.

In summary, supervised learning is used when you have labeled data and want to predict or
classify based on known outcomes, while unsupervised learning is used when you want to
uncover patterns or structure within unlabeled data. Each approach is valuable in different
scenarios and serves distinct machine learning purposes.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 101

 12.1. Supervised and Unsupervised Learning Algorithms

12.1.1. Supervised Learning Algorithms

Here are some common supervised learning algorithms:

1. Linear Regression: Linear regression is used for predicting a continuous target variable
(a real number) based on one or more input features. It fits a linear equation to the data
and is particularly useful for regression tasks.

2. Logistic Regression: Unlike linear regression, logistic regression is used for binary
classification problems, where the goal is to predict one of two classes. It models the
probability that a given input belongs to a particular class.

3. Decision Trees: Decision trees are versatile algorithms that can be used for both
classification and regression tasks. They partition the data into subsets based on the
values of input features and make decisions at each node to classify or predict the target
variable.

4. Random Forest: Random Forest is an ensemble learning method that combines multiple
decision trees to improve predictive performance and reduce overfitting. It is commonly
used for both classification and regression tasks.

5. Support Vector Machines (SVM): SVM is a classification algorithm that finds the
optimal hyperplane to separate data into different classes. It works well for linear and
non-linear classification tasks.

6. K-Nearest Neighbors (KNN): KNN is a simple yet effective algorithm for classification
and regression. It makes predictions based on the majority class or average value of its k
nearest neighbors in the training data.

7. Naive Bayes: Naive Bayes is a probabilistic algorithm based on Bayes' theorem. It is

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 102

 often used for text classification and spam filtering but can be applied to other
classification tasks as well.

8. Neural Networks: Neural networks, particularly deep learning models, have gained
popularity for their ability to handle complex data and tasks. They consist of
interconnected layers of neurons and are used for a wide range of tasks, including image
and speech recognition, natural language processing, and more.

12.1.2. Unsupervised Learning Algorithms

Here are some common unsupervised learning algorithms:

1. Clustering Algorithms:

 K-Means Clustering: This algorithm partitions the data into 'K' clusters based on
similarity. It assigns data points to clusters by minimizing the distance between data
points and the cluster centroids.

 Hierarchical Clustering: Hierarchical clustering builds a tree-like structure of clusters,

where data points are grouped together based on their similarity.

 Gaussian Mixture Models (GMM): GMM models data as a mixture of multiple

Gaussian distributions and can capture complex patterns in the data.

2. Dimensionality Reduction Algorithms:

 Principal Component Analysis (PCA): PCA reduces the dimensionality of data by

projecting it onto a lower-dimensional subspace while preserving the most important
information.

 Autoencoders: Autoencoders are neural network-based models that learn to encode and
decode data, effectively reducing its dimensionality. They are used for tasks like feature

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 103

 extraction and denoising.

3. Generative Models:

 Generative Adversarial Networks (GANs): GANs consist of two neural networks, a

generator, and a discriminator, which are trained adversarial. GANs can generate new
data instances that resemble the training data.

12.2. Relevant Lecture Readings

Revise Lectures Supervised and Unsupervised Learning.

12.3. Objective of the Experiment

After completing this lab, the student should be able to:

 Different concepts of supervised and unsupervised learning.

12.4. Concept Map

This concept map will help students to understand the main concepts of topics covered
in the lab. In this lab, students will be provided with a detailed content to learn that what
the concept of supervised and unsupervised learning is and how to implement different
algorithms in python.

12.5. Procedure & Tools

Pycharm

12.6. Homework before Lab

Attempt the following questions. You are required to bring this task with you and submit
to the lab instructor.

12.6.1. Problem description

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 104

  Suppose you have a dataset with historical temperature data and energy consumption for
a building. How can you apply supervised learning to predict energy consumption based
on temperature?
 Given a dataset of customer purchase histories, how can you use unsupervised learning
to segment customers into distinct groups based on their buying behavior? Provide an
example of an algorithm you might use.

12.7. Practice Task

Here's an example of implementing the k-Nearest Neighbors (KNN) classification

algorithm in Python using scikit-learn, and then plotting the accuracy results with a
graph.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 105

 Output:

12.8. Evaluation Tasks

This section will provide more practice exercises which you need to finish during the
lab. You need to finish the tasks in the required time.

12.8.1. Task 1 (Marks 10)

Implement linear regression in Python using scikit-learn to predict housing prices. Load
a dataset, split it into training and testing sets, fit the model, and evaluate its performance
using mean squared error.
Datasets links: https://www.kaggle.com/datasets/yasserh/housing-prices-dataset

12.8.2. Task 2 (Marks 10)

Apply Principle Component Analysis (PCA) to reduce the dimensionality of a high-
dimensional dataset of stock market features. Show the cumulative explained variance
and visualize the reduced dataset.
12.8.3. Out comes

After completing this lab, students will to know how to deal with different supervised
and unsupervised learning algorithms.

12.9. Evaluation Criteria

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 106

 The evaluation criteria for this lab will be based on the completion of the following tasks.
Each task is assigned the marks percentage which will be evaluated by the instructor in
the lab whether the student has finished the complete/partial task(s)

Criteria Level 0 Level 1 Level 2 Level 3 Level 4 Total

(0%) (1-50)% (51-65)% (66-80)% (81-100)% Score
Procedural Awareness
Practical Implementation
Program Correctness
Use of Software Tool
Sub Total Marks

12.10. Further Readings

12.10.1. Books
Artificial Intelligence: A modern approach 3rd Edition by Stuart Russell & Peter Norvig

12.10.2. Links
https://colab.research.google.com/github/RPI-DATA/course-intro-ml-
app/blob/master/content/notebooks/08-intro-modeling/04-knn.ipynb#scrollTo=nhXd2F5ewXPo

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 107

 Lab Manual for Artificial Intelligence
LAB-13: Performance Metrics in Machine Learning

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 108

 13. Introduction
Performance metrics in machine learning are used to evaluate the quality and
effectiveness of a machine learning model's predictions. These metrics help you
understand how well your model is performing on a given task, such as
classification, regression, or clustering. The choice of performance metrics depends
on the specific problem you are trying to solve and the nature of your data. Here we
only discuss some performance metrics for classification and regressions metrics of
machine learning tasks:

Classification Metrics:
1. Accuracy: This is one of the most basic metrics and measures the ratio of
correctly predicted instances to the total instances in the dataset. However,
it may not be suitable for imbalanced datasets.
Example: If a binary classification model correctly predicts 800 out of 1000
instances, the accuracy is 80%.
2. Precision: Precision measures the proportion of true positive predictions out
of all positive predictions. It is useful when false positives are costly.
Example: Out of 100 positive predictions, 85 were true positives, so the
precision is 85%.
𝑇𝑃
𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛 =
𝑇𝑃 + 𝐹𝑃
3. Recall (Sensitivity or True Positive Rate): Recall calculates the proportion
of true positive predictions out of all actual positives. It is important when
false negatives are costly.
Example: There were 100 actual positive instances, and the model correctly
predicted 85 of them, so the recall is 85%.
𝑇𝑃
𝑟𝑒𝑐𝑎𝑙𝑙 =
𝑇𝑃 + 𝐹𝑁

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 109

 4. F1-Score: The F1-Score is the harmonic mean of precision and recall,
providing a balanced measure between the two.
Example: If precision is 0.85 and recall is 0.80, the F1-Score is 0.825.
𝑝𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛 × 𝑟𝑒𝑐𝑎𝑙𝑙
𝑓1𝑠𝑐𝑜𝑟𝑒 =
𝑝𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛 + 𝑟𝑒𝑐𝑎𝑙𝑙
5. ROC AUC (Receiver Operating Characteristic Area under Curve):
ROC AUC quantifies the model's ability to distinguish between classes
across different thresholds. It's especially useful when you need to control
the trade-off between true positive and false positive rates.
Example: A ROC AUC score of 0.90 indicates good discrimination between
classes.
6. Confusion Matrix: Not a single metric but a table that summarizes the
performance of a classification algorithm, showing true positives, true
negatives, false positives, and false negatives.
Regression Metrics:
1. Mean Absolute Error (MAE): MAE measures the average absolute
difference between predicted and actual values. It is robust to outliers but not
differentiable.
Example: If the MAE is 5, it means, on average, predictions are off by 5 units
from actual values.
2. Mean Squared Error (MSE): MSE measures the average squared
difference between predicted and actual values. It punishes larger errors more
than MAE.
Example: If the MSE is 25, it means, on average, predictions are off by 25
squared units from actual values.
3. Root Mean Squared Error (RMSE): RMSE is the square root of MSE and
provides an interpretable metric in the same units as the target variable.
Example: If RMSE is 5, it means, on average, predictions are off by 5 units

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 110

 from actual values.
4. R-squared (Coefficient of Determination): R-squared measures the
proportion of the variance in the target variable explained by the model. It
ranges from 0 to 1, with higher values indicating better fit.
Example: An R-squared of 0.75 means 75% of the variance in the target
variable is explained by the model.

13.1. Relevant Lecture Readings

Revise Lecture Performance Metrics
13.2. Objective of the Experiment
After completing this lab, the student should be able to:
 Understand the concept of different performance metrics such as precision, recall,
F1 score etc. 

13.3. Concept map

This concept map will help students to understand the main concepts of topics
covered in the lab. In this lab, students will be provided with a detailed content to
learn about different performance metrics and their implementation in classification
and regression metrics.

13.4. Homework before Lab

Attempt the following questions. You are required to bring this task with you and
submit to the lab instructor.

13.4.1. Problem description

1. You have a classifier that predicts whether an email is spam or not. Out of 1000
emails, it correctly identifies 400 spam emails and incorrectly classifies 50 non-
spam emails as spam. Calculate precision, recall, and F1-Score for this classifier.

2. You have a regression model for predicting house prices. The model has an
RMSE of 10. What does this RMSE value signify in the context of your model's

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 111

 predictions?

13.5. Procedure & Tools

Pycharm
Google Colab

13.6. Practice Task

Certainly! Here's a practice task in Python that involves using performance
metrics for a classification problem. In this task, you'll evaluate the performance
of a binary classification model using common metrics such as accuracy,
precision, recall, and the ROC curve. You can use libraries like scikit-learn to
simplify the process.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 112

 Output:

13.7. Evaluation Tasks

This section will provide more evaluation exercises which you need to finish
during the lab. You need to finish the tasks in the required time.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 113

 13.7.1. Task 1
Write Python code to calculate and print the accuracy, precision, recall, and F1-Score of a
binary classification model using scikit-learn. Assume you have the true labels (y_true) and
predicted labels (y_pred).

13.7.2. Task 2
Implement a Python function to compute and print the Mean Absolute Error (MAE), Mean
Squared Error (MSE), Root Mean Squared Error (RMSE), and R-squared (R^2) for a
regression model using scikit-learn. Assume you have the true target values (y_true) and
predicted values (y_pred).

13.7.3. Out comes

After completing this lab, students will be able to compute different performance metrics.

13.8. Evaluation Criteria

The evaluation criteria for this lab will be based on the completion of the following
tasks. Each task is assigned the marks percentage which will be evaluated by the
instructor in the lab whether the student has finished the complete/partial task(s)

Table 3: Evaluation of the Lab

Criteria Level 0 Level 1 Level 2 Level 3 Level 4 Total
(0%) (1-50)% (51-65)% (66-80)% (81-100)% Score
Procedural Awareness
Practical Implementation
Program Correctness
Use of Software Tool
Sub Total Marks

13.9. Further Readings

13.9.1. Books
Artificial Intelligence: A modern approach 3rd Edition by Stuart Russell
& Peter Norvig

13.9.2. Links
https://www.javatpoint.com/performance-metrics-in-machine-learning

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 114

 Lab Manual for Artificial Intelligence
LAB-14: Natural Language Processing (NLP)

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 115

 14. Introduction
Natural Language Processing (NLP) is the development of applications or services
that understand human language. Here are some practical examples of natural
language processing (NLP), such as speech recognition, speech translation,
understanding complete sentences, understanding synonyms of matching words, and
generating grammatically correct complete sentences and paragraphs.

14.1. Practical Examples of NLP

1. Search Engine: such as Google, Yahoo, etc. Google search engine knows you
are a technician, so it shows technology-related results.
2. Social Network Push: Facebook News Feed. If the News Feed algorithm knows
that your interest is natural language processing, relevant ads and posts will be
displayed.
3. Speech Engine: such as Apple’s Siri.
14.2. Libraries of NLP

Here are some open-source natural language processing libraries (NLPs):

 Natural language toolkit (NLTK): The Natural Language Toolkit (NLTK) is the
most popular Natural Language Processing Library (NLP), written in Python, and
has very strong community support behind it. NLTK is also very easy to use and , it
is the simplest natural language processing (NLP) library.For installation NLTK
Windows/Linux/Mac, you can install NLTK using the pip command.To install
NLTK for the first time, you need to install the NLTK extension package. This will
bring up the NLTK download window to choose which packages need to be
installed:

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 116

  SpaCy: It is a modern and efficient NLP library that offers fast and accurate
tokenization, part-of-speech tagging, named entity recognition, and dependency
parsing. It is known for its speed and ease of use.

 Scikit-learn: It is a popular machine learning library that includes various NLP

functionalities. It provides tools for text preprocessing, feature extraction, and
classification algorithms that can be applied to NLP tasks.

 TensorFlow and Keras: TensorFlow is a powerful deep learning library, and Keras
is a high-level neural networks API that runs on top of TensorFlow. They are
commonly used for building and training deep learning models for NLP tasks like
text classification, sentiment analysis, and machine translation.
 PyTorch: PyTorch is another popular deep learning library that provides dynamic
computational graphs and a flexible framework for building neural networks. It is
widely used for NLP tasks, including language modeling, text generation, and
machine translation.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 117

 These libraries offer a wide range of functionalities and tools to handle various NLP
tasks efficiently. They are constantly evolving and provide extensive documentation and
community support, making them valuable resources for NLP practitioners and
researchers

14.3. Relevant Lecture Readings

Revise Lecture Natural Language Processing and its parameters
14.4. Objective of the Experiment
After completing this lab, the student should be able to:
 Understand the concept of different parameters of NLP and its practical
implementation. 

14.5. Concept map

This concept map will help students to understand the main concepts of topics
covered in the lab. In this lab, students will be provided with a detailed content to
learn about different Natural language processing parameters and their
implementation using neural networks.

14.6. Homework before Lab

Attempt the following questions. You are required to bring this task with you and
submit to the lab instructor.

14.6.1. Problem description

1. Write Python code to tokenize a text document into words or tokens.

2. How can you use the CountVectorizer from the scikit-learn library to create a
BoW representation of a collection of text documents? Provide an example.

14.7. Procedure & Tools

Pycharm
Google Colab

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 118

 14.8. Practice Task
Implementing GloVe (Global Vectors for Word Representation) from scratch
in Python is a complex task, and the full code can be quite lengthy. However, I
can provide you with an outline of the steps involved and some key code
snippets to get you started. To implement GloVe, you'll need a large text corpus
and the following main steps:

Data Preparation:

 Tokenize the corpus into words or tokens.

 Create a word co-occurrence matrix that captures how often words co-
occur in the context of each other.

Initialize Word Vectors:

 Initialize word vectors for each word in the vocabulary.

 Initialize bias terms for each word.

Define the Objective Function:

 Define the loss function based on the word co-occurrence matrix and
word vectors.

Optimization:

 Use an optimization algorithm (e.g., stochastic gradient descent) to

minimize the loss function.

 Update word vectors and bias terms iteratively.

Training:

 Train the model on the corpus for a specified number of epochs.

Obtain Word Embedding:

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 119

  After training, the word vectors can be used as word embedding.

Here's a simplified code snippet to give you an idea of how you might start
implementing GloVe using Python:

Output:

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 120

 14.9. Evaluation Tasks
This section will provide more evaluation exercises which you need to finish
during the lab. You need to finish the tasks in the required time.

14.9.1. Task 1 (Marks 10)

Implement text preprocessing steps such as lowercasing, removing punctuation, and stop
word removal.

14.9.2. Practice Task 2 (Marks 10)

Load pre-trained word embedding (e.g., Word2Vec or GloVe) using python libraries and
use them to find similar words for a given word?

14.9.3. Out comes

After completing this lab, students will be able to understand and implement Natural
Language Processing.

14.10. Evaluation Criteria

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 121

 The evaluation criteria for this lab will be based on the completion of the following
tasks. Each task is assigned the marks percentage which will be evaluated by the
instructor in the lab whether the student has finished the complete/partial task(s)

Table 3: Evaluation of the Lab

Criteria Level 0 Level 1 Level 2 Level 3 Level 4 Total
(0%) (1-50)% (51-65)% (66-80)% (81-100)% Score
Procedural Awareness
Practical Implementation
Program Correctness
Use of Software Tool
Sub Total Marks

14.11. Further Readings

14.11.1. Books
Artificial Intelligence: A modern approach 3rd Edition by Stuart Russell
& Peter Norvig
14.11.2. Links
http://www.dataversity.net/natural-language-processing/

https://www.wired.com/insights/2014/02/growing-importance-natural-language-processing/

https://radimrehurek.com/gensim/models/keyedvectors.html

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 122

Glossary

1. Artificial Intelligence (AI): The field of computer science that focuses on creating systems or algorithms
capable of performing tasks that typically require human intelligence, such as understanding natural
language, recognizing patterns, and making decisions.
2. Machine Learning (ML): A subfield of AI that involves the development of algorithms that enable
computers to learn and improve their performance on a specific task through experience and data.
3. Deep Learning: A subset of machine learning that uses neural networks with multiple layers (deep neural
networks) to automatically learn patterns and representations from data.
4. Neural Network: A computational model inspired by the structure and function of the human brain, used in
machine learning and deep learning for tasks such as image recognition and natural language processing.
5. Supervised Learning: A type of machine learning in which an algorithm learns from labeled training data
to make predictions or classify new, unseen data.
6. Unsupervised Learning: A type of machine learning in which an algorithm learns patterns and structures in
data without explicit labels or supervision.
7. Reinforcement Learning: A machine learning paradigm where an agent learns to make decisions by
interacting with an environment and receiving rewards or penalties based on its actions.
8. Natural Language Processing (NLP): The branch of AI that focuses on the interaction between computers
and human language, enabling machines to understand, interpret, and generate human language.
9. Feature Engineering: The process of selecting, transforming, or creating relevant features from raw data to
improve the performance of machine learning models.
10. Data Preprocessing: The tasks involved in cleaning, formatting, and transforming raw data into a suitable
format for training machine learning models.
11. Overfitting: A common issue in machine learning where a model performs well on the training data but
poorly on unseen data, indicating that it has learned noise rather than meaningful patterns.
12. Bias-Variance Trade-off: A fundamental concept in machine learning, where a model's ability to generalize
is balanced between underfitting (high bias) and overfitting (high variance).
13. TensorFlow: An open-source machine learning framework developed by Google that is widely used for
building and training deep learning models.
14. PyTorch: An open-source deep learning framework developed by Facebook's AI Research lab (FAIR) that
is known for its flexibility and dynamic computation graph.
15. Hyperparameter: A configuration setting that is not learned from data but is set prior to training a machine
learning model, such as learning rate or the number of hidden layers in a neural network.
16. Algorithm: A set of step-by-step instructions for solving a specific problem or task, often used
interchangeably with "model" in the context of machine learning.
17. Bias: Systematic errors or inaccuracies in a model's predictions or decisions, often resulting from flaws in
the data or the model's design.
18. Variance: The variability of a model's predictions when trained on different subsets of the data, indicating
how sensitive the model is to changes in the training data.
19. Cross-Validation: A technique used to assess a model's performance by splitting the data into multiple
subsets for training and testing to obtain a more robust evaluation.

CAPITAL UNIVERSITY OF SCIENCE & TECHNOLOGY 123