CSI3026 MACHINE LEARNING

Assessment Methods

Assessment Max. Weightage

type Marks

Quiz 1 10 10

Quiz 2 10 10

Assignment 10 10
CAT – I 30 15
CAT – II 15
30
FAT 40
100
 Module 1
Module:

Introduction to Machine Learning

Introduction

Google Search Engine

Amazon E-

mail
Human can learn from past experience
and make decision of its own

5
What is this object?

6
 What is this object?

CAR

CAR

BIKE
It is a CAR
BIKE

7
Let us ask the same question to him
What is this object?

8
Let us ask the same question to him
What is this object?
?

9
[ But, he is a human being. He can observe and
learn ]
et us make him learn

show him

1
1
et us make him learn

CAR

show him

CAR

BIKE

BIKE

1
2
Let us ask the same question now
What is this object?

CAR

CAR

BIKE
BIKE
Past experience 10
Let us ask the same question now
CAR What is this object?

CAR

CAR

BIKE
BIKE

14
What about a Machine ?

Machines follow instructions

[ It can not take decision of its own

15
What about a Machine ?

We can ask a machine

• To perform an arithmetic operations such as

• Addition
• Multiplication
• Division

Machines follow instructions

16
What about a Machine ?

• Comparison

• Print

• Plotting a chart

Machines follow instructions

17
What is Machine Learning?

[ We want a machine to act like a human]

18
What is Machine Learning?

[ to identify this object.]

19
What is Machine Learning?

Price in 2025?

[ predict the price in future]

20
 What is Machine Learning?

I made met him yesterday

[ Natural Language understand, and correct grammar ]

21
 What is Machine Learning?

recognize face

[ Recognize Faces ]
22
What is Machine Learning?

[ What do we do?

Just like, what we did to human,

we need to provide experience

to the machine.

23
What is Machine Learning?

[
This what we called as Data
or Training dataset

+ So, we first need to provide

training dataset to the
machine
]

Dataset

24
What is Machine Learning?

+ +
[ Then, devise algorithms and execute programs on the
data

With respect to the underlying target tasks ]

Dataset

25
What is Machine Learning?

+ + +

Dataset [ Then, using the programs, Identify

required rules ]
26
What is Machine Learning?

+ + +

Dataset [extract required patterns ]

27
What is Machine Learning?

+ + +

Dataset [ Identify relations ]

28
What is Machine Learning?

+ + + =

Dataset [ So that machine can derive inferences

from the data ]
29
In summary, what is machine learning?

Given a machine learning problem

• Identify and create the appropriate dataset

• Perform computation to learn

• Required rules, pattern and relations

• Output the decision

30
In summary, what is machine learning?

Machine learning is a subset of artificial intelligence that gives

systems the ability to learn and optimize processes without
having to be consistently programmed.
programmed Simply put, machine
learning uses data, statistics and trial and error to “learn” a
specific task without ever having to be specifically coded for
the task.
What is Machine Learning?

Machine Learning
Learn from past experiences
Improve the performances of intelligent programs

Definition (Mitchell 1997)

A computer program is said to learn from experience E with respect to
some class of tasks T and performance measure P, if its performance at
the tasks improves with the experiences
The concept of learning in a ML system

• Learning = Improving performance with experience

at some task
– Improve over task T,
– With respect to performance measure, P
– Based on experience, E.

7
 Motivating Example Learning to Filter Spam

ample:
ample Spam Filtering
am - is all email the user does not want
receive and has not asked to receive
T: Identify Spam Emails
P:
% of spam emails that were filtered
% of ham/ (non-spam) emails that were
incorrectly filtered-out
E: a database of emails that were
labelled by users
What is Machine Learning?
Traditional Programming

Data
Output
Program Computer

Machine Learning

Data
Computer Program
Output
 Real Time Applications
ogle’s GNMT(Google
GNMT Neural Machine Translation)

35%
35% of Amazon’s revenue is generated by P
Face Book Recommendations.

Gmail
Paypal Google

Maps Uber eCommerce Losses to Online Payment

Fraud to Exceed $48 Billion Globall
in every year
 Machine Learning Paradigms
•Supervised

•Unsupervised Learning

•Reinforcement learning

We as human being solve various types of problem in our day-to-day life, <pause> Various decisions
need to be taken.
Depending on the nature of the problem, machine learning tasks can be broadly divided in ]
40
What is Supervised Learning?

• A category of machine learning that

uses labeled datasets to train
algorithms to predict outcomes and
recognize patterns.
patterns
• These algorithms are given labeled training
data to learn the relationship between the
input and the outputs.
outputs
41
What is Supervised Learning?
CAR

CAR

+ BIKE
= Training Dataset
BIKE

Samples Labels

[In supervised learning, we need some thing called a Labelled Training Dataset ]
42
What is Supervised Learning?
CAR

CAR

+ BIKE
= Training Dataset =
BIKE

Samples Labels

[ Given a labelled dataset, the task is to devise a function which takes the dataset, and a new sample, and
produces an output value.]
43
What is Supervised Learning?
CAR

CAR

+ BIKE
= Training Dataset =
BIKE

Samples Labels

[ Given a labelled dataset, the task is to devise a function which takes the dataset, and a new sample, and
produces an output value.]
44
What is Supervised Learning?
CAR

CAR

+ BIKE
= Training Dataset = CAR
BIKE

Samples Labels

[ Given a labelled dataset, the task is to devise a function which takes the dataset, and a new sample, and
produces an output value.]
45
What is Supervised Learning?
CAR
Classification
CAR

+ BIKE
= Training Dataset = CAR
BIKE

Samples Labels

[ If the possible output values of the function are predefined and discrete/categorical, it is called
Classification
33
What is Supervised Learning?
CAR
Classification
CAR

+ BIKE
= Training Dataset = CAR
BIKE

Samples Labels

[ Predefined classes means, it will produce output only from the labels defined in the dataset. For example,
even if we input a bus, it will produce either CAR or BIKE ]
47
Classifier
Elephant
Elephant

Classifier

Tiger Identify the Animal ?

Dataset
48
 Classification Applications
Pattern recognition
Face recognition: Pose, lighting, occlusion (glasses, beard), make-
make
up, hair style
Character recognition: Different handwriting styles. Speech
recognition: Temporal dependency.
Use of a dictionary or the syntax of the language.
Sensor fusion: Combine multiple modalities; eg, visual (lip image)
and acoustic for speech
Medical diagnosis: From symptoms to illnesses
Web Advertizing: Predict if a user clicks on an ad on
the Internet.
Supervised Learning : Applications

Prediction of future cases: Use the rule to predict the output for future inputs

Knowledge extraction: Learning a rule from data Compression: The rule is simpler

than the data it explains

Outlier detection: Exceptions that are not covered by the rule, e.g., fraud Novelty Detection :

Previously unseen but valid case

Regression
Regression

= 20500.50

Dataset

[ If the possible output values of the function are continuous real values, then it is called Regression
51
[
The classification and Regression problems are supervised, because the decision depends on the
characteristics of the ground truth labels or values present in the dataset, which we define as experience
]

52
What is Unsupervised Learning?
Learning

• A type of machine learning that

learns from data without human
supervision.
• Unsupervised machine learning models are
given unlabeled data and allowed to
discover patterns and insights without any
explicit guidance or instruction.
instruction
53
What is Unsupervised Learning?
Learning
CAR

CAR

BIKE

BIKE

Dataset

[ In the unsupervised learning, we do not need to know the labels or Ground truth values ]
54
What is Unsupervised Learning?
Learning

Clustering
Dataset

[ The task is to identify the patterns like group the similar objects together ]
39
What is Unsupervised Learning?
Learning

Association Rules Mining

Dataset

[ Association rules like ]

56
More Example Unsupervised Learning

Dataset
57
More Example Unsupervised Learning

58
Unsupervised Learning : Applications

Document grouping Custering gene of

Individual
Organizing Computing Clusters Social
Network
Market Segment
What is Reinforcement Learning?
• It is a machine learning model that is similar
to supervised learning, but the algorithm isn’t
trained using sample data.
data
• This model learns as it goes by using trial and
error.
• A sequence of successful outcomes will be
reinforced to develop the best recommendation
or policy for a given problem.
problem
[ It is also known as learning from trials and errors ]
60
What is Reinforcement Learning?

61
What is Reinforcement Learning?

62
What is Reinforcement Learning?

63
Another Example

Agent Task Environment

64
Reinforcement Learning

Punishment

65
Reinforcement Learning

Reward

66
Reinforcement Learning

Reward
Baby Learn from the Trials and Errors

Reinforcement Learning 67
Another example: Computer playing chess
ML basic concepts
There are many different types of machine learning algorithms, with hundreds
published each day, and they’re typically grouped by either learning style (i.e.
supervised learning, unsupervised learning, semi-supervised learning) or by similarity
in form or function (i.e. classification, regression, decision tree, clustering, deep
learning, etc.).

Regardless of learning style or function, all combinations of machine learning

algorithms consist of the following:

• Representation
• Evaluation
• Optimization
ML basic concepts
• Representation:
• It is basically the space of allowed models (the hypothesis space)
• A set of classifiers or the language that a computer understands.
• This implies how to represent knowledge.
• Evaluation:
• objective/scoring function.
• This is the way to evaluate candidate programs (hypotheses).
• How do we differentiate good models from bad ones.
• Optimization:
• search method; often the highest-scoring classifier, for example; there are both off-the-shelf and
custom optimization methods used
• what is our process for finding the good models among all the possible models
ML basic concepts
The fundamental goal of machine learning
algorithms is to generalize beyond the training
samples i.e. successfully interpret data that it has
never ‘seen’ before.
Important Terminologies and Definitions
• Examples: Items or instances of data used for learning or evaluation. In our spam
problem, these examples correspond to the collection of email messages we will
use for learning and testing.

• Features: The set of attributes, often represented as a vector, associated to an

example. In the case of email messages, some relevant features may include the
length of the message, the name of the sender, various characteristics of the
header, the presence of certain keywords in the body of the message, and so on.

• Labels: Values or categories assigned to examples. In classification problems,

examples are assigned specific categories, for instance, the spam and non-spam
categories in our binary classification problem.
problem In regression, items are assigned
real-valued labels.
Important Terminologies and Definitions
• Training sample: Examples used to train a learning algorithm.

• Validation sample: Examples used to tune the parameters of a learning algorithm

when working with labeled data.

• Test sample: Examples used to evaluate the performance of a learning algorithm.

• Loss function: A function that measures the difference, or loss, between a

predicted label and a true label.

• Hypothesis set: A set of functions mapping features (feature vectors) to the set of
• labels Y.
Summary

what is machine learning

what are the machine learning paradigms

[ In this lesion, we have learnt ]