UNIVERSITY OF WESTMINSTER

TRENDS IN COMPUTER SCIENCE

4COSC008C.2

MACHINE LEARNING

2.(b) Overview of Machine Learning.

Describe and compare two different machine learning techniques.

NAME: MADURAPPERUMA ARACHCHILLAGE LISITHA KAWODYA

MADURAPPERUMA.

UOW NUMBER: w1998857

IIT NUMBER: 20220560

GROUP E5:

Name IIT ID UOW Number Topic

C.R.S.A. Hewavickrama 20222237 w1985556 2(a)
M.A.L.K.Madurapperum 20220560 w1998857 2(b)
a
K.S.N.Wickramanayaka 20221974 w1985670 2(c)
T.M.H.M Thennakoon 20222162 w1985547 2(d)
1 Table of Content
1 Introduction.................................................................................................................................3

1.1 Overview of Machine Learning..........................................................................................4

2 Supervised Learning....................................................................................................................5

2.1 Regression............................................................................................................................5

2.2 Classification........................................................................................................................6

3 Unsupervised Learning...............................................................................................................7

4 Comparison between Supervised Learning and Unsupervised Learning...............................8

4.1 Advantages of Supervised Learning and Unsupervised Learning...................................8

4.2 Disadvantages of Supervise Learning and Unsupervised Learning.................................9

5 Conclusion..................................................................................................................................10

6 Critical Evaluation....................................................................................................................10

References..........................................................................................................................................11

1|Page
List of Figures
Figure 1.1 Comparison between traditional programming-based system and machine learning based
system....................................................................................................................................................3
Figure 2.21 Regression model and Classification model.......................................................................6

List of Tables
Table 4.11 Advantages of Supervised learning and Unsupervised learning...........................................8
Table 4.22 Disadvantages of Supervised learning and Unsupervised learning......................................9

2|Page
1 Introduction

Machine learning is one aspect of artificial intelligence. It allows machines or computers

to make decisions and learn on their own. (Kapoor,Sachdeva,2022). In traditional
programming, the output was represented after the required data inputs, process and
outputs were given, but in machine learning, after the required inputs, process and outputs
are given to the relevant model or computer, a figure or an object matching the previously
given inputs is used to get the desired output. Then the model or computer will analyse
the object or figure and provide the desired output to the user.(Shyan,Singh,2021).

Figure 1.1 Comparison between traditional

programming-based system and machine
learning based system.

Machine learning techniques divided into main three categories

1) Supervise Learning
2) Unsupervised Learning
3) Reinforcement Learning

This report provides a clear description of supervised learning and unsupervised

learning and a comparison of the two techniques.

3|Page
1.1 Overview of Machine Learning.

The concept of machine learning first appeared in 1959 because of the experiments
conducted by Arthur Samuel who studied the field. The method used here is to enable the
computer to perform tasks automatically without being pre-programmed. This can be
considered as the origin of Machine learning. Machine learning is a subset of artificial
intelligence. The next subset is deep learning. At present it can be considered as a field which
is updated day by day. Because Machine learning and related theories are used a lot in many
fields nowadays. Examples of techniques that use the concept of machine learning.

1) Finance Social Networks

2) E- commerce
3) Web search
4) Healthcare
5) Robotics
6) Debugging software
7) Information extraction
8) Computationally biology
9) Space exploration
10) Product Recommendation etc
(Shyam,Singh,2021)

Similarly, many business organizations are using this concept for their operations and
their future development. Examples of such companies are Google Brain (2011), Alex
Net (2012), Deep Farce (2014) and Open AI (2015).
(Shyam,Singh,2021)

4|Page
2 Supervised Learning.

In supervised learning, the machine is trained by inputting and outputting data that
matches the machine or model. Data input to the machine while training it must be
labeled data.

There are several types of algorithms that can be used to create a form of machine
learning called supervised learning
1) Linear Regression
2) Logistic Regression
3) Support Vector Machine (SVM)
4) K-Nearest Neighbors (KNN)
5) Nalive Bias
6) Decision Tree
7) Random Forest

Supervised Learning divided into two categories.

2.1 Regression
The process that takes place here is that the input data to the algorithm cannot be
classified. In simpler terms, the output or the value given here is always a constant
value. (Nasteski,2017)

For example, when the algorithm is given the value of tomorrow's ambient
temperature, it can return a value from negative infinity to positive infinity because it
is constantly changing. As such, this always returns only constant values
(Nasteski ,2017)

5|Page
2.2 Classification
This is the process of separating similar elements into one group. If the input labeled
data is separated into two groups it is called binary classification and if more than two
groups it is called multiclass classification. (Shyam,Singh,2021)

For example, after an test using a group of cells taken from a patient's body, when the
algorithm is instructed to find cancer cells among the group of cells obtained, it will
perform those tasks, that is, the user must have taught the model how to identify
cancer cells, then the model will Different cancer cells are isolated and imaged

Figure 2.21 Regression model and Classification

model

6|Page
3 Unsupervised Learning

Unsupervised learning mainly trains on unlabeled data. This allows the machine to self-study
and categorize the data. Unsupervised learning techniques are especially used in multimedia
content such as clustering or partitioning.(Shyam,Singh,2021)

Several forms of clustering are used in unsupervised learning

1) K-Means Clustering
2) Hierarchical Clustering
3) Principal Component Analysis (PCA)
4) Singular Value Decomposition
5) Independent Component Analysis

7|Page
4 Comparison between Supervised Learning and Unsupervised
Learning

This chapter aims to discuss a comparison between supervised learning and unsupervised
learning. Here the advantages and disadvantages of supervised learning and unsupervised
learning are discussed.

4.1 Advantages of Supervised Learning and Unsupervised Learning

Supervise learning Unsupervised learning

Supervised learning allows data to be Labeled data is expensive when used
collected from previous experiences to for algorithms. But unsupervised
create output data learning solves clustering problems
without learning and labeling given
examples.
Criteria-based experiences help Being able to replace labels with data
improve performance after classification
Supervised learning helps to solve Compared to other learning
different type of real-world problems techniques, it helps a lot in finding
patterns in data

Table 4.11 Advantages of Supervised learning and Unsupervised learning

8|Page
4.2 Disadvantages of Supervise Learning and Unsupervised Learning

Supervised Learning Unsupervised Learning

Exceeding the decision-making threshold Outputs may be less accurate due to less
and not delivering the desired output when input data
training.
Not being able to get good examples from Training the model on new data without
every class while training for classification prior knowledge
Categorizing data into large categories is a An algorithm takes more time in the
tough challenge learning process

Table 4.22 Disadvantages of Supervised learning and Unsupervised learning

(Shyam,Singh,2021)

9|Page
5 Conclusion

Machine learning is a subset of Artificial Intelligence. The purpose of this grant is to create
autonomous programs based on artificial intelligence. This report discusses two popular
machine learning techniques, Supervised learning and Unsupervised learning techniques, and
this report contains a comparison of the two methods.

6 Critical Evaluation

The facts that emerged from researching these two topics are that Machine learning technique
2 has more or less advantages as well as disadvantages. This means that the Unsupervised
learning method is very suitable compared to Supervised learning. Because the cost to be
applied for unsupervised learning is minimal, I believe that it is more appropriate if it is a
method that can be used by people while minimizing its shortcomings.

10 | P a g e
References

JavaTpoint.Regression vs Classification in Machine learning[image]. Available from

Regression vs Classification in Machine Learning - Javatpoint [Accessed 15th February 2023]

Kapoor,R.andSachdeva,D.(2022).Machine Learning Techniques.India:AGPG Bokks.

Available
fromhttps://www.researchgate.net/publication/367161404_Machine_Learning_Techniques
[Accessed 10th February 2023]

Martin,M.E,Cazorla,M.and Escolano,S.O.(2020).Machine learning Techniques Assistive

Robotics.Switzerland:MDPL Available from

https://library-collections-search.westminster.ac.uk/discovery/fulldisplay?
docid=alma997314574803711&context=L&vid=44WST_INST:WST_VUA&lang=en&searc
h_scope=MyInst_and_CI&adapto [Accessed 10th February 2023]

Nasteski,V.(2017).An overview of the supervised machine learning methods.1-8. Available

from

https://www.researchgate.net/profile/Vladimir-Nasteski/publication/
328146111_An_overview_of_the_supervised_machine_learning_methods/links/
5c1025194585157ac1bba147/A

n-overview-of-the-s [Accessed 11th February 2023]

Shyam,E. and Singh,R.(2021).A Taxonomy of Machine Learning Techniques. Advancements

in Robotics,8(3,2021),18-22.Available from

11 | P a g e
https://www.researchgate.net/publication/
358089496_A_Taxonomy_of_Machine_Learning_Techniques [Accessed 10th February 2023]

Stefanus,R.(2019).Conventional computing Vs Machine Learning. Available from

https://rstefanus16.medium.com/conventional-programming-vs-machine-learning-
a3b7b3425531 [Accessed 16th February 2023]

12 | P a g e