Welcome to

WOA 7015
Advanced Machine Learning

Saw Shier Nee

[email protected]
Dept. of AI, Fac. of Computer Science and IT, University of
Malaya
 Innovative Way to Learn Machine Learning

• We all learn together - sharing sessions

• Students are expected to learn most techniques online by
themselves.
• What is the use of this course?
• This course provides concepts and understandings
that cannot be easily self-learned.
• This course provides the real life experience for doing
deep learning.
• This course teaches you how to think in different ways
Four types of people

No skills & Have skills

Beggars cannot but cannot Good
explain explain servants
ideas ideas

No skills Have skills

but can and can
Bosses explain explain Masters
ideas ideas
Who will succeed in the Future?

No skills & Have skills

Beggars cannot but cannot Good
explain explain servants
ideas ideas

No skills Have skills

but can and can
Bosses explain explain Masters
ideas ideas
 In this course,
• We aim to be a master
What is AI? ML? DL?
 History of AI
“Can machines
think?”
Alan Turing https://www.bbc.com/news/technology-35785875

introduce a Alpha Go
method for defeated Lee
machine https://editions.covecollective.org/
chronologies/alan-turing-
First AI Second AI Sedol, world
intelligence %E2%80%9Ccomputing-
machinery-and-intelligence
Winter Winter champion of Go
%E2%80%9D-1950

1950 1966 1975 1980 1987 2016 Now

First Chatbot: Expert System AI in

ELIZA everywhere

https://www.computerhope.com/jargon/e/
eliza.htm
Impact of AI
Transporta
tion

Manufactur
Agriculture
ing

AI
Retail Healthcare

Media Education

8
Agriculture

9
 Manufacturing
Inside Tesla's Crazy AI Manufacturing Revolution

https://www.youtube.com/watch?
v=uobJbNyGVjo&ab_channel=CasgainsCapital
10
 Google Skin Condition Prediction
Healthcare

https://australiannewsreview.com/google-unveils-ai-tool-that-can-
identify-288-skin-conditions/
Healthcare – 2021’s papers
 Retails

https://usmsystems.com/top-use-cases-of-artificial-
intelligence-in-retail-industry/ https://medium.com/@humansforai/retail-d4ca38b4de3f
 Transportation
Tokyo – Autonomous Taxi Waymo company - self-
driving minivans and trucks

https://asia.nikkei.com/
Transportation - Traffic

Drone Taxi
How AI affect employment?

16
Current AI limitation

17
AI Robot - Sophia

18
 How human brain works?

86 billions of neurons

https://www.livescience.com/29365-human-brain.html
19
How information is transmitted via neurons?

20
 Deep learning Intuition

21
https://www.researchgate.net/publication/339446790_Using_a_Data_Driven_Approach_to_Predict_Waves_Generated_by_Gravity_Driven_Mass_Flows
Deep learning

22
Deep learning
DenseNet

InceptionNet

23
 Deep learning - Accuracy vs Operation

Model
complexity
Computing
Resources

24
https://towardsdatascience.com/neural-network-architectures-156e5bad51ba
How we learn?

Input Output

Unclear Unclear
input output

Precise Precise
Input output
25
How we teach machine to learn?

Input Output

Unclear Unclear
input output

Precise Precise
Input output
26
Different levels

Can
fundamentally
Can combine change deep
methods to learning in
create new research
things in Deep
Can understand Learning
what they doing
and can explain

Can get
results but
cannot judge
if the output is
reasonable 27
Design AI is not difficult

Prerequisite: Coding

28
Design AI is not difficult
Libraries
Integrated Development Environment (IDE)

29
What is your coding level? Write your real
name.

ⓘ Start presenting to display the poll results on this slide.

Code-less Platform – Drag and Drop

31
Everyone is learning AI

32
What set us apart?

Can
fundamentally
Can combine change deep
methods to learning in
create new research
things in Deep
Can understand Learning
what they doing
and can explain

Can get
results but
cannot judge
if the output is
reasonable 33
Go into the maths

Can
fundamentally
Can combine change deep
methods to learning in
create new research
things in Deep
Can understand Learning
what they doing
and can explain

Can get
results but
cannot judge
if the output is
reasonable 34
Let’s play a game. Refresh your maths

35
 Can you guess the simplification of this
equation?

Material from C.S.Chan 36

Material from C.S.Chan 37
 What to expect?

• Basic fundamental of probability

• Using probability theory to develop machines that can learn

from data.

• Heavy maths is involved.

 Assessment
• Assignment – 50% (week 2 – 6 – Spectrum )
• Group Projects – 50% (Week 6 Announcement)
• Pre-submission report – 5%, (week 10 - Spectrum)
• Presentation – 5%, (week 14)
• Final Report + Code Submission – 40%, (week 14 - Spectrum)

39