CSE 523 : Machine Learning (ML)

Lecture - 1
Introduction to Machine Learning Course

Dhaval Patel, Ph.D

Assistant Professor,
School of Engineering and Applied Science,
Ahmedabad University, Gujarat, India

January 07, 2020

 Outline
• Introduction to Machine Learning
- Traditional Programming vs Machine Learning (ML)
- When do we use ML?
- ML Applications and recent developments
- Artificial Intelligence vs Machine Learning vs Deep Learning
- What is Machine Learning?

• About CSE 523 – A Course on Machine Learning

- Course Philosophy, Objectives and Outcomes
- Evaluation Components
- Project Discussions

• A Brief on Project Areas

- Natural Language Processing (NLP)
- Intelligent Transportation System (ITS)
- 5G/6G Wireless Networks
- Environment and Healthcare
- Biology/Bioinformatics 2
 Outline
• Introduction to Machine Learning
- Traditional Programming vs Machine Learning (ML)
- When do we use ML?
- ML Applications and recent developments
- Artificial Intelligence vs Machine Learning vs Deep Learning
- What is Machine Learning?

• About CSE 523 – A Course on Machine Learning

- Course Philosophy, Objectives and Outcomes
- Evaluation Components
- Project Discussions

• A Brief on Project Areas

- Natural Language Processing (NLP)
- Intelligent Transportation System (ITS)
- 5G/6G Wireless Networks
- Environment and Healthcare
- Biology/Bioinformatics 3
Introduction to Machine Learning
- Traditional Programming vs Machine Learning (1/5)

4
Introduction to Machine Learning
- When do we use ML (2/5)

Human expertise does not exist - Humans can’t explain their expertise:
Navigating on Mars Speech recognition

Models must be customized: Models are based on huge amounts of data

5 :
Personalized medicine Genomics
Introduction to Machine Learning
- When do we use ML (2/5)

Human expertise does not exist -

Navigating on Mars

6
Introduction to Machine Learning
- When do we use ML (2/5)

Human expertise does not exist - Humans can’t explain their expertise:
Navigating on Mars Speech recognition

7
Introduction to Machine Learning
- When do we use ML (2/5)

Human expertise does not exist - Humans can’t explain their expertise:
Navigating on Mars Speech recognition

Models must be customized: 8

Personalized medicine
Introduction to Machine Learning
- When do we use ML (2/5)

Human expertise does not exist - Humans can’t explain their expertise:
Navigating on Mars Speech recognition

Models must be customized: Models are based on huge amounts of data

9 :
Personalized medicine Genomics
Introduction to Machine Learning
- When do we use ML (2/5)

A classic example of a task that requires machine learning:

It is very hard to say what makes a 2

10
Introduction to Machine Learning
- When do we use ML (2/5)

Some more examples of tasks that are best solved by using a ML algorithm:

Recognizing patterns Web search:

– Facial identities or facial expressions
– Handwritten or spoken words • Computational biology
– Medical images
• Finance
Generating patterns • E-commerce
– Generating images or motion sequences
• Space exploration
Recognizing anomalies
– Unusual credit card transactions • Robotics
– Unusual patterns of sensor readings in a
nuclear power plant • Information extraction

Prediction: • Social networks

– Future stock prices or currency exchange
rates • Debugging software 11
Introduction to Machine Learning
- When do we use ML (2/5)

Some more examples of tasks that are best solved by using a ML algorithm:

Recognizing patterns
– Facial identities or facial expressions
– Handwritten or spoken words
– Medical images

12
Introduction to Machine Learning
- When do we use ML (2/5)

Some more examples of tasks that are best solved by using a ML algorithm:

Recognizing patterns
– Facial identities or facial expressions
– Handwritten or spoken words
– Medical images

Generating patterns
– Generating images or motion sequences

13
Introduction to Machine Learning
- When do we use ML (2/5)

Some more examples of tasks that are best solved by using a ML algorithm:

Recognizing patterns
– Facial identities or facial expressions
– Handwritten or spoken words
– Medical images

Generating patterns
– Generating images or motion sequences

Recognizing anomalies
– Unusual credit card transactions
– Unusual patterns of sensor readings in a
nuclear power plant

14
Introduction to Machine Learning
- When do we use ML (2/5)

Some more examples of tasks that are best solved by using a ML algorithm:

Recognizing patterns
– Facial identities or facial expressions
– Handwritten or spoken words
– Medical images

Generating patterns
– Generating images or motion sequences

Recognizing anomalies
– Unusual credit card transactions
– Unusual patterns of sensor readings in a
nuclear power plant

Prediction:
– Future stock prices or currency exchange
15
rates
Introduction to Machine Learning
- When do we use ML (2/5)

Some more examples of tasks that are best solved by using a ML algorithm:

Recognizing patterns Web search:

– Facial identities or facial expressions
– Handwritten or spoken words • Computational biology
– Medical images
• Finance
Generating patterns • E-commerce
– Generating images or motion sequences
• Space exploration
Recognizing anomalies
– Unusual credit card transactions • Robotics
– Unusual patterns of sensor readings in a
nuclear power plant • Information extraction

Prediction: • Social networks

– Future stock prices or currency exchange
rates • Debugging software 16
Introduction to Machine Learning
- ML Applications (3/5)

17
Introduction to Machine Learning
- ML Applications (3/5)

18
Source: https://www.apple.com/siri/
Introduction to Machine Learning
- ML Applications (3/5)

19
Introduction to Machine Learning
- ML Applications (3/5)

20
Introduction to Machine Learning
- ML Applications (3/5)

Cancer Detection

21
Introduction to Machine Learning
- ML Applications (3/5)

22
Introduction to Machine Learning
- ML Applications (3/5)

23
Introduction to Machine Learning
- ML Applications (3/5)

24
Introduction to Machine Learning
- Artificial Intelligence vs Machine Learning vs Deep Learning (4/5)

The capacity of a
computer to learn from
experience, i.e. to
modify its processing on
the basis of newly
acquired information.

25
Introduction to Machine Learning
- Artificial Intelligence vs Machine Learning vs Deep Learning (4/5)

The capacity of a
computer to learn from
experience, i.e. to
modify its processing on
the basis of newly
acquired information.

26
Introduction to Machine Learning
- Artificial Intelligence vs Machine Learning vs Deep Learning (4/5)

The capacity of a
computer to learn from
experience, i.e. to
modify its processing on
the basis of newly
acquired information.

Source:
https://en.oxforddictionaries.com/definiti
on/machine_learning 27
Introduction to Machine Learning
- What is ML (5/5)
“Learning is any process by which a system improves performance from
experience.”

28
Introduction to Machine Learning
- What is ML (5/5)
“Learning is any process by which a system improves performance from
experience.”

29
Introduction to Machine Learning
- What is ML (5/5)
“Learning is any process by which a system improves performance from
experience.”

Herbert A. Simon
The Sveriges
Riksbank Prize in
Economic Sciences
in Memory of Alfred
Nobel, 1978

30
Introduction to Machine Learning
- What is ML (5/5)
“Learning is any process by which a system improves performance from
experience.”

Herbert A. Simon
The Sveriges
Riksbank Prize in
Economic Sciences
in Memory of Alfred
Nobel, 1978

✓ Herbert Simon was one of the founding fathers of artificial intelligence.

✓ No other scientist better understood of computers. the future of machines and the ultimate
importance

✓ By 1965, Simon was certain that “machines will be capable of doing any work a man can do”

✓ His visionary perspective on decision making processes, climate change and flaws in
31
economic theories prove to be even more relevant and crucial in the 21st century.
Introduction to Machine Learning
- What is ML (5/5)

“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 tasks in T, as measured by P,
improves with experience E”

Tom M. Mitchell, Example: Image Classification using ML

Professor of Machine Learning
Department, School of Computer E: Past data with images having labels or
Science, Carnegie Mellon University
assigned classes

T: The task of assigning class to new, unlabelled

images

P: The performance measure indicated by the

percentage of images correctly classified.
32
 Outline
• Introduction to Machine Learning
- Traditional Programming vs Machine Learning (ML)
- When do we use ML?
- ML Applications and recent developments
- Artificial Intelligence vs Machine Learning vs Deep Learning
- What is Machine Learning?

• About CSE 523 – A Course on Machine Learning

- Course Philosophy, Objectives and Outcomes
- Evaluation Components
- Project Discussions

• A Brief on Project Areas

- Natural Language Processing (NLP)
- Intelligent Transportation System (ITS)
- 5G/6G Wireless Networks
- Environment and Healthcare
- Biology/Bioinformatics 33
About CSE 523 – A Course on Machine Learning
- Course Philosophy, Objectives and Outcomes

34
About CSE 523 – A Course on Machine Learning
- Course Philosophy, Objectives and Outcomes

35
About CSE 523 – A Course on Machine Learning
- Course Philosophy, Objectives and Outcomes

36
About CSE 523 – A Course on Machine Learning
- Course Philosophy, Objectives and Outcomes

37
About CSE 523 – A Course on Machine Learning
- Course Philosophy, Objectives and Outcomes

The objectives of this course are:

1. To understand the four pillars of ML like Regression, Dimensionality reduction,

Density estimation and Classification and to learn about various machine learning
algorithms.

2. To understand mathematics as a working principle of machine learning models.

3. To implement various ML algorithms in TensorFlow and to solve various real-

world problems in the areas like Natural language processing, Intelligent
Transportation System, 5G/6G wireless networks, Environment and climate
change and Bioinformatics.

4. Understanding of Artificial Neural Network

5. Focus on supervised and unsupervised form of learning through case studies

38
About CSE 523 – A Course on Machine Learning
- Course Philosophy, Objectives and Outcomes

The following are the expected learning outcomes from this course:

1. To apply machine learning as a tool to address real-life problems/challenges.

2. Ability to develop a new machine-learning algorithm to challenge and address

the state of the art machine learning algorithms.

3. To create a strong foundation for related courses like Deep Learning, Computer
Vision, Reinforcement Learning, Natural language processing and other similar
courses that are commonly found in.

4. Learn a course to match the ABET standard.

5. Select an appropriate feature for classification and derive insight into

fundamental building blocks of machine learning.

6. Apply specific machine learning tool based on application context and

understand the mathematical and statistical principles underlying the tool. 39
About CSE 523 – A Course on Machine Learning
- Evaluation Components

End semester exam - 40%

Mid semester exam – 25%

Project - 35%

40
About CSE 523 – A Course on Machine Learning
- Project Guidelines

Can I carry out project in group ?

The B.Tech students can carry their project in a group (max. four students in a group).
Students have autonomy to choose the partner. However, M.Tech and PhD students
have to carry their project independently only (no group)

How would the project submission be done?

There would be total FIVE project submissions. Each project submission is highly correlated
to its previous submission, covering different aspects of the course.

1. Project Abstract Submission (in week-3, will be launched soon) (5% weightage)

2. Project Module #1 : Regression (5% weightage)

3. Project Module #2 : Dimensionality reduction (5% weightage)

4. Project Module #3: Density Estimation and Classification (5% weightage)

41
5. Innovation and final submission (15%)
About CSE 523 – A Course on Machine Learning
- Project Guidelines

What is the project all about?

As a part of the project component comprising of 35% weightage, the students

would be required to do a project. There are five project areas that have been
identified. They are listed as below:

1. Natural Language Processing (NLP)

2. Intelligent Transportation System (ITS)

3. 5G/6G Wireless Networks

4. Environment and Climate change

5. Biology /Bioinformatics
42
About CSE 523 – A Course on Machine Learning
- Tools / Software / Platform

43
About CSE 523 – A Course on Machine Learning
- Tools / Software / Platform

44
About CSE 523 – A Course on Machine Learning
- Schedule up to Mid Semester Examination

Week # Execution
1. L1 and L2
Week 1 2. Project Group Formation
3. Tutorial #1 – ML lab hands-on, Tensorflow Introduction, Example
1. L3 and L4
Week 2 2. Project execution and implementation, Research articles in group
3. Demonstration of Project areas-one implementation

1. L5 and L6
Week 3
2. Project viva and Abstract Submission
1. L7 to L13
Week 4-5-6
2. Math Quiz + L1-L6
Midsem Exam
Post Midsem Feedback form-I

45
Thank you !!
dhaval.patel@ahduni.edu.in

46