What is
Artificial
Intelligence?
• The theory and development of
computer systems able to perform tasks
normally requiring human intelligence,
such as visual perception, speech
recognition, decision making, and
translation between languages.” –
Google
AI is the new electricity
“About 100 years ago,
electricity transformed every
major industry. AI has
advanced to the point where
it has the power to
transform” every major
sector in coming years.–
Andrew Ng
Ref: AI Singapore
What drives AI?
Cutting Edge
Results in a
Variety of Fields
Bigger
datasets
Faster
Computers
Neural Nets
Faster hardware is one of the key areas driving the modern era of AI.
Bigger Datasets
In 2020, it is expected that:
• The average internet user will generate ~1.5 GB of traffic per day.
• A smart hospital will generate 3,000 GB/day.
• Self-driving cars are each generating over 4,000 GB/day.
• Connected planes will generate 40,000 gigabytes per day.
• A connected factory will generate 1 million gigabytes per day.
The MIC represents a continuum from simple, scripted automation to
superhuman intelligence and highlights the functional capabilities of different
levels of machine intelligence.
https://www.topbots.com/topbots-ai-machine-intelligence-continuum/?utm_medium=article&utm_source=medium&utm_campaign=AI-Continuum
Artificial Narrow Intelligence vs
Artificial General Intelligence
Artificial General Intelligence
Understand Abstract Concepts
Explain Why
Be Creative Like Children
Tell Right From Wrong
Have Emotions
Beat Go World Champions
Read Facial Expressions
Write Music
Diagnose Mental Disorders
Comfort Earthquake Survivors
Artificial Narrow Intelligence
Use cases in our daily life
• Smart Phone – Voice Assistants,
image enhancement, App store
Recommendations, Face unlock
• Transportation – Dynamic Pricing in
Travel, hospitality, logistic…
• Web Services – Email Filtering,
search, translation, Facebook and
LinkedIn recommendations.
• Sales – Netflix, Spotify, Amazon
Recommendation Engines, Customer
Support Queries (and Chatbots)
• Security – Video Surveillance, Cyber
Security
• Financial – Catching Fraud
Definitions
Artificial
Intelligence
Machine
Learning
Deep
Learning Deep Learning
A subset of machine learning in which multilayered
neural networks learn from vast amount of data.
Machine Learning
Subset of Al techniques which use statistical methods
to enable machines to improve with experiences.
Artificial Intelligence
Any technique which enables computers to
sense, reason, act and adapt
Is this weird?
Anomaly
detection
Is this pressure
gauge reading
normal?
Is this message
from the internet
typical?
How many?
How Much?
Regression
What will the
temperature be
next Tuesday?
What will my
fourth quarter
sales be?
How is this
organized?
Clustering
Which viewers
like the same
types of
movies?
Which printer
models fail the
same way?
What should I
do?
Reinforce
Learning
If I'm a self-
driving car: At a
yellow light,
brake or
accelerate?
For a robot
vacuum: Keep
vacuuming, or
go back to the
charging
station?
Is this A or B?
(Classification)
Will this tire fail
in the next 1,000
miles: Yes or
no?
Which brings in
more
customers: a $5
coupon or a
25% discount?
5 questions data science answers
Machine Learning Limitations
• Suppose you wanted to
determine if an image is of a cat
or a dog.
• What features would you use?
• This is where Deep Learning
can come in.
Dog and cat recognition
What is deep learning?
Deep Learning
“Machine learning that involves
using very complicated models
called “deep neural networks”."
(Intel)
Models determine best
representation of original data; in
classic machine learning, humans
must do this.
Deep Learning Example
Classic
Machine
Learning
Step 1: Determine
features.
Step 2: Feed them
through model.
“Arjun"
Neural Network
“Arjun"
Deep
Learning
Steps 1 and 2
are combined
into 1 step.
Machine
Learning
Classifier
Algorithm
Feature
Detection
AI Services
• Google’s AI Services for Companies
• https://experiments.withgoogle.com/collection/ai
• Google’s cloud-based AI Tools
• https://ai.google
• Microsoft AI Language Translator
• Amazon’s AI Platform
• Alibaba Machine Learning Platform for AI
AI changes Job Market
• Jobs that Don’t Involve
Large Quantities of
Data
• Jobs Based on Human
Interaction
• Jobs that Have Minimal
Repetition or Routine
• Jobs that are Difficult to
Learn Through Simple
Observation
https://www.futuristspeaker.com/business/20-common-jobs-in-2040/
Can a robot do your job?
http://bit.ly/2C3igDx
http://bit.ly/2C3igDx
Current Manpower and Projected Demand
for ICT Professional 2018 - 2020
87800
33600
15500
100800
35900
20200
IT Development Network and Infrastructure Critical Emerging Tech
2017 2020
13,100 IT Development and 4700 Critical Emerging Tech Projected Demand
IMDA 2018
Survey
Critical Emerging Tech Specialists
a. Includes Data analysts/Data scientists, Machine
Learning/Artificial Intelligence Engineer, IT Security
specialists, IT Security Operations Analysts/Engineers,
Infocomm R&D, Internet of Things (IoT) Engineer, Embedded
Systems/Fireware Developers, IoT Solution Architect
b. Projected demand to grow by another 4,700 headcounts in the
next three years (2018 – 2020)
Estimate of Technology Displaced Jobs
by Country
8.1%
9.5 m
7.4%
1.2 m
10.1%
4.5 m
20.6%
0.5 m
11.9%
4.9 m
13.8%
7.5 m
Indonesia Malaysia Philippines Singapore Thailand Vietnam
Displacement Percentage
https://www.oxfordeconomics.com/recent-releases/dd577680-7297-4677-aa8f-450da197e132
Technology and the Future of Asean Jobs : The Impact of AI on Workers in ASEAN’s 6 Largest Economies
Projected Proportion of Technology
Displaced Jobs in Singapore
Technology and the Future of Asean Jobs : The Impact of AI on Workers in ASEAN’s 6
Largest Economies
https://www.oxfordeconomics.com/recent-releases/dd577680-7297-4677-aa8f-450da197e132
Job Roles in AI
Role AI Roles Composition Summary of Key Functions
Role 1 AI Researcher / AI
Scientist / Data
Scientist (often PhD /
Master)
~10 - 20% Research & develop algorithms;
Develop data strategy, models
and insights
Role 2 AI Engineers / AI
Developers
~25 - 35% Develop AI software and
products; Experiment, configure
and test algorithms
Role 3 Data Engineer ~25 - 35% Design and implement data
systems and solutions
Role 4 AI Application
Developers | System
Integrators | Infra
Engineers
~25 - 35% Develop / integrate AI
applications;
Deploy infra for AI services
DO YOU WANT TO LEARN AI?
Get it I
still want
you to learn
something.
Y N
Use pre-
trained ML
APIs in
your apps
Take AI/ML
short/
online
courses
DO YOU DEVELOP
APP?
DO YOU WANT TO
CODE YOUR OWN
ML?
ARE YOU ALREADY
AN SOFTWARE
ENGINEER?
DO YOU WANT TO BE
AN AI ENGINEER?
ARE YOU FOCUSED ON
VERTICAL INDUSTRY?
(e.g. healthcare, finance, retail)
CAN YOU STUDY AI
FULL-TIME?
ARE YOU LEARNING FOR
PROFESSIONAL GROWTH
OR PERSONAL CURIOSITY?
Read an
entire AI
report
Watch
videos
from AI
experts
Use code-
free ML
tools
Enroll in
part-time
courses
e.g. SDAAI
Enroll in
immersive
bootcamp
e.g. TIPP AI
Set daily
Google
alerts
Y N
N Y
N Y Y N
NY
Y N
Pro Per
Est 12 wksEst 1 yr Est 20 days Est 5 days Est 2 days
Inspiration from Allie K. Miller
Specialist Diploma in Applied
Artificial Intelligence
Total Hours = 270 hours
Time to Complete = 12 months
PDC in Fundamentals
of Artificial Intelligence
(120 hours)
PDC in Applications of
Artificial Intelligence
150 hours)
Math for Machine
Learning (30 hours)
Pattern Recognition
and Anomaly Detection
(30 hours)
Introduction to
Programming (30
hours)
Recommender
Systems (30 hours)
Introduction to Data
Management for
Machine Learning (30
hours)
Virtual Assistants (30
hours)
Machine Learning
Fundamentals (30
hours)
Capstone Project* (60
hours)
Create Azure account
Create Custom Vision resources
Setup
Allow the classifier to know what constitutes a given
class.
Train the classifier
Use unseen data to test your classifier.
Test your model
Gather training and validation images from internet
or other sources
Prepare images
Check for accuracy, recall and probability threshold.
Evaluate the classifier
Use various techniques to improve your classifier.
Improve your classifier
01
02
03
04
05
06
Hotdog/Not-Hotdog
- HBO’s Silicon Valley
Training an image classifier
Hands-on
Training an image classifier
Hands-on NLP
• Use Google Cloud Platform, we will:
• Classify Text
• Named Entity Recognition
• Sentiment Analysis
• Syntax Analysis
https://cloud.google.com/natural-language/docs/apis
Hands on – Object Recognition
• Workflow
1. Load Data.
2. Define Model.
3. Compile Model.
4. Fit/Train Model.
5. Evaluate Model.
6. Predict with new data.