0ICPC404

Machine Learning
Dr. Tahseen A. Mulla (Ph. D - CSE)
Associate Professor
Department of Computer Science and Engineering
[Internet of Things and Cyber Security including Blockchain
Technology]
Unit 1

Machine Learning
Overview
• Machine Learning Definition
• Terminology
• Types of Learning
• Machine Learning problem categories
• Machine Learning architecture
• Process Life Cycle
• Performance Measures
• Tools and Frameworks
• Data Visualization
Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 3
Machine Learning
Machine Learning – (Definition)
Machine learning is a field of computer
science that aims to teach computers how
to learn and act without being explicitly
programmed

Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 5
Machine Learning – (Definition)
According to Tom Mitchell –

A computer program is said to learn from experience ’E’ with

respect to some task ’T’ and some performance
measure ’P’,
and
if its performance on T,
as measured by P,
improves with experience E

Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 6
 Lets see some
Machine Learning videos

Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 7
Applications of Machine
Learning
Applications of Machine Learning

• Natural Language Processing

• Insurance Claim Analysis

• Bioinformatics and Medical Diagnosis

• Image Processing and Pattern Recognition

• Search Engines

• Financial Market Analysis

Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 9
History of Machine Learning
History of Machine Learning
The “nearest
Arthur Samuel wrote the neighbor” algorithm
Frank Rosenbelt
first computer learning was written, allowing
designed the first
Alan Turing creates program “The game of computers to begin
neural network for
the “Turing Test” to checkers” and the IBM using very basic
computers which
determine if a computer improved the pattern recognition
simulates the thought
computer has real game more it played
processes of the
intelligence studying which moves made
human brain
To pass this test, a up winning strategies and
computer must be incorporating those moves
able to fool a human into its program
into believing it is also
human

1967
1952 1957
1950
Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 11
History of Machine Learning Work on Machine
Learning shifts from
knowledge-driven
approach to a data-
Gerald Dejong introduces Terry Sejnowski driven approach.
the concept of Explanation invents NetTalk, Now machines used
Based Learning (EBL), in which learns to to learn from the
which a computer analyses pronounce words the results
Students at Stanford training data and creates a same way a baby
University invent the general rule it can follow by does
“Stanford Cart” discarding the unimportant
which can navigate data
obstacles in a room
on its own

1990
1981 1985
1979
Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 12
History of Machine Learning
The Microsoft Google Brain is
“Kinect” can track 20 developed, and its
human features at a deep neural network
Geoffrey Hinton coins rate of 30 times per can learn to discover
the term “Deep second, allowing and categorize
Learning” to explain people to interact objects
new algorithms that let with the computer
computers “see” and via movements and
distinguish objects and gestures
text in images and
IBM’s Deep Blue videos
beats the world
champion at chess

2011
2006 2010
1997
Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 13
 Google’s Artificial

History of Machine Learning Intelligence algorithm

beats a professional
player at Chinese
board game “Go”,
which is considered the
Facebook develops world’s most complex
DeepFace, a software board game and is
Google’s X lab
algorithm that is able to many times harder than
developers develops
recognize or verify Amazon launches its chess
a machine learning
algorithm that is able individuals on photos own Machine
to autonomously to the same level as Learning platform
browse YouTube humans can
videos to identify the
videos that contains
cats

2016
2014 2015
2012
Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 14
Types of Machine Learning
Types of Machine Learning
Machine
Learning

Supervised Unsupervised Reinforcement

Learning Learning Learning

Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 16
Supervised Learning
• Supervised learning is where
you have input variables (x)
and an output variable (Y) and
you use an algorithm to learn
the mapping function from the
input to the output
Y = f(X)
• The goal is to approximate the
mapping function so well that
when you have new input data
(x) that you can predict the
output variables (Y) for that
data
Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 17
Unsupervised Learning
• Unsupervised learning is where you only have input data (X)
and no corresponding output variables

• The goal for unsupervised learning is to model the underlying

structure or distribution in the data in order to learn more
about the data

Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 18
Reinforcement Learning
• It is about taking suitable action
to maximize reward in a
particular situation

• It is employed by various
software and machines to find
the best possible behavior or
path it should take in a specific
situation

Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 19
Mind Map of Machine Learning
Machine Learning

Supervised Unsupervised Reinforcement

Learning Learning Learning

Continuous Categorical NO LABELED DEPENDS ON THE

variable variable DATA AGENT
Eg:
Classificati
Regression Clustering Association Optimized
on
Marketing
Eg:
Eg: Housing Eg: Market Eg:
Eg: Medical Customer
Price Basket Driverless
Imaging Segmentatio
prediction Analysis cars
n
Eg:
Gaming
Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 20
Hypothesis Space
Hypothesis Space
• Hypothesis space is the set from which the machine learning
algorithm would determine the best possible which would
best describe the target function or the outputs
Unknown Target
function
t: X  Y
Learning Final
Training Examples
Algorith Hypothe
(x1, y1), …, (xn, yn)
m (A) sis

Hypothesis Space

Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 22
Hypothesis Space

• Lets consider this co-ordinate

that shows the distribution of
some data

• Suppose we have the test data

for which we have to determine
the outputs

Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 23
Hypothesis Space
The test data will look like
this

Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 24
Hypothesis Space
• Based on the data, algorithm and constraints this coordinate
plane can be divided in the following ways -

Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 25
Hypothesis Space
• Hypothesis space (H) is the
composition of all legal best
possible ways to divide the
coordinate plane so that it
best maps input to proper
output

• Each individual best possible

way is called a hypothesis (h)

Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 26
Machine Learning Architecture
Machine Learning Architecture
Gatheri
ng Data

Data
Deploy
Prepara
ment
tion

Data
Model
Wrangli
Testing
ng

Data
Model
Analyzin
Training
g

Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 28
Gathering Data
• It is the first step of the Machine Learning life cycle. The goal
of this is to identify and obtain all data-related problems

• We need to identify the different data collect from various

sources such as files, database, internet and other devices

• The quantity and quality of the collected data will determine

the efficiency of the output

Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 29
Data Preparation
• After collecting the data, we need to prepare it for further
steps

• Data preparation is where we put our data into a suitable

place and prepare it to use in our Machine Learning training

• Data preparation can be divided into two processes –

• Data Exploration – To understand the nature of data that we have to
work with. The characteristics of data such as it properties, format
and quality of the data
• Data Pre-processing – Used for analysis
Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 30
Data Wrangling
• It is the process of cleaning and converting raw data into a useable format

• It is the process of cleaning the data, selecting the variable to use and
transforming the data in a proper format to make it more suitable for
analysis

• Cleaning of data is required to address the quality issues

• In real-world applications, collected data nay have various issues such as –

• Missing values
• Duplicate data
• Invalid data
• Noise
Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 31
Data Analysis
• The cleaned and prepared data is passed on to the analysis step

• This step involves –

• Selection of analytical techniques
• Building the models
• Reviewing the result

• We can use this for determining the type of problems, to select

the machine learning techniques such as regression,
classification, cluster analysis and association to build the
model using prepared data and evaluate the model
Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 32
Model Training
• We train our model to improve its performance for better
outcome of the problem

• Various datasets are used to train the model using various

machine learning algorithms

• Training a model is required so that it can understand the

various patterns, rules and features

Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 33
Model Testing
• The machine learning model has been trained on a given
dataset, then we test the model

• Here we check for the accuracy of our model by providing a

test dataset to it

• Testing the model determines the percentage accuracy of the

model as per the requirement of the problem statement

Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 34
Deployment
• If the model is preparing an accurate result as per our
requirement, then the model is deployed in the real system

• Before deploying the project we check whether it is

performing using available data or not

Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 35
Performance Evaluation
Performance Evaluation
• Performance is very essential for a model in Machine
Learning

• We can train our supervised machine learning models all day

long, but unless you evaluate its performance, you can never
know if your model is useful or not

Why evaluation in necessary?

Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 37
Performance Measures
• Confusion Matrix
• Accuracy
• Precision and Recall
• ROC & AUC
• Log/Logarithmic Loss
• F1 score
• Mean Absolute Error
• Mean Squared Error
• Coefficient of Determination
Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 38
Tools and Framework
Tools

Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 40
Framework

Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 41
Data Visualization
Data Visualization
• Line Plot

Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 43
Data Visualization
• Line Plot
• Bar Chart

Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 44
Data Visualization
• Line Plot
• Bar Chart
• Pie and Donut Chart

Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 45
Data Visualization
• Line Plot
• Bar Chart
• Pie and Donut Chart
• Histogram Plot

Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 46
Data Visualization
• Line Plot
• Bar Chart
• Pie and Donut Chart
• Histogram Plot
• Scatter Plot

Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 47
Data Visualization
• Line Plot
• Bar Chart
• Pie and Donut Chart
• Histogram Plot
• Scatter Plot
• Kernel Density Estimation

Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 48
Data Visualization
• Line Plot
• Bar Chart
• Pie and Donut Chart
• Histogram Plot
• Scatter Plot
• Kernel Density Estimation
• Box and Whisker Plot

Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 49
Data Visualization
• Line Plot
• Bar Chart
• Pie and Donut Chart
• Histogram Plot
• Scatter Plot
• Kernel Density Estimation
• Box and Whisker Plot
• Word Cloud and Network Diagrams

Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 50
Data Visualization
• Line Plot
• Bar Chart
• Pie and Donut Chart
• Histogram Plot
• Scatter Plot
• Kernel Density Estimation
• Box and Whisker Plot
• Word Cloud and Network Diagrams
• Correlation Matrices
Wednesday,
Dr.January
Tahseen2A. Mulla Unit 1 - Machine Learning 51
Let’s summarize
 Introduction to Probability and Statistics
 Machine Learning definition
 Machine Learning terminology
 Types of Machine Learning
 Machine Learning problem categories End of Module -
 Machine Learning Architecture 1
 Machine Learning Process
 Machine Learning Lifecycle
 Performance measures
 Tools and framework
 Data Visualization