Problem Solving - Human vs. Machine Intelligence

Aug 16, 2014 ISS Open Day 2014
Knowledge Engineering Group
Problem Solving 
Human vs. Machine Intelligence

2
 Watson scores landslide victory over the best players on
the TV game show Jeopardy, Feb 2011
 Watson earned $77,147, versus
$24,000 for Ken Jennings, and
$21,600 for Brad Rutter
Human Intelligence Machine Intelligence
Watson took 25 IBM scientists 4 years, and around $30 million to create.
Man vs. Machine: IBM Watson
?Do you
also know
?

3
What experts say
 Can Watson decide to create Watson? … While computers can
calculate and construct, they cannot decide to create. We are
far from there. … Our ability to create is what allows us to
discover and create new knowledge and technology.
 Pradeep Khosla, Carnegie Mellon University
What Watson’s creators say
 I see human intelligence consuming machine intelligence,
not the other way around.
 David Ferrucci, IBM's lead researcher on Watson
Human Machine Intelligence

4
 Human intelligence has been taken as the gold standard of
machine intelligence.
 Turing Test [Alan Turing, 1950]
The Gold Standard & AI Dream
The Turing test of artificial intelligence proposes a simple
game where a hidden computer A and a person B converse
with another person C.
If C is unable to distinguish which he is conversing with
then the computer can be said to be able to “think”.
Modern computer still
cannot pass the test !
image: www.rutherfordjournal.org

5
 There are many aspects to describe human intelligence.
 An important one is the capability of problem solving using
knowledge
 Using existing knowledge to
solve new problem
 Learning of new knowledge
from new experience
 Discovery of knowledge
 Machine intelligence is expected to do the same.
Human Intelligence
image: jeeda-lovenpassion.blogspot.com

6
Problem Solving
 Typical (but COMPLEX) problem solving tasks:
 Planning
 Decision
 Classification / categorization
 Prediction / forecasting
image: undsci.berkely.edu

7
Problem Solving: examples
 Planning
 new housing development
 MRT stations / lines
 a course timetable for school
 … …
 Decision / optimization
an optimal investment decision
a good university to attend
… …
image: techchai.com
image: www.do2learn.com

8
 Classification / categorization
 Classifying customer groups for adopting
suitable campaign / promotion strategies
 Categorizing customer complaints for effective responses /
follow-up actions
 … …
image: www.marketing.savant.com
image: cacm.acm.org

9
 Prediction / forecast
 weather forecasting
financial market trends
demand of electricity
… …
image: www.paloalto.com

10
√ Faster
√ Cheaper
? Better
 Computer / software systems for problem solving and
decision making, with the support of
 Advance IT technologies
 Internet, Mobile, Cloud, Data storage, …
 Data / information
 Past records, relevant reference, …
Intelligent Systems
 Why?
 Humans performing the same tasks would be considered as being
“intelligent”
Automated Problem Solving

11
 An intelligent system (IS) is an embodiment
of machine intelligence. IS can typically solve problems …
 in well defined domain
 for well specified tasks
 with well described performance targets
 Intelligent Systems make use of …
 intelligent algorithms and domain knowledge
and/or
 machine learning techniques and sources of data
Machine Intelligence
Provided
by human
Developed
by human
Created by
human and/or
machine
Built by
human

12
 A big picture of intelligent system (at data / information level)
Intelligent System
Intelligent
algorithms
Data
Domain
knowledge
human or
machine
intelligence
User
Developer
Intelligent
system

13
 A complex task that requires the developer with …
 competencies in knowledge and skills of exploiting intelligent
systems techniques, and
 experience and knowledge of problem domains
 Knowledge engineering (KE), is an important discipline of
Artificial Intelligence (AI). KE comprises:
Methodologies,
Techniques, and
Practical approaches
for the successful development of intelligent systems
Developing Intelligent Systems
image: www.123rf.com

14
 Knowledge Based System for Marketing (Local retailer)
 Indoor Location Prediction on Mobile Devices (SAP)
 Intelligent Cab Management System
 Length of stay prediction (Local Hospital)
 Prediction of Colorectal cancer recurrence (TTSH)
 Customer Complaints Handling (Local Bank)
 Insurance Quote Comparison (Local Insurer)
 Career Management system (Armed Forces)
 JWT World Wide IT Intelligent System (JWT)
 Regulatory Knowledge based system (Sennheiser)
 Storm water drainage design (Engineering)
 Weather forecasting system (NEA)
 Intelligent System for evaluating musical chord progression (NIE)
Many more in different areas …
Systems Developed by KE Students
Business
Service /
finance
Healthcare
/ medicine
Urban develop
/ environment
Enterprise /
organization
Arts /
Music
image: www.projectinsight.net

15
 Intelligent system for a local leading department store
[KE24 batch, Jan 2014]
 Using data mining techniques and
other intelligent techniques
 Business analytics
 Perform spending analysis on
their card member base
 Build models using past promotion
 Generate a list of card members who are likely respond to
advertising for promotion

16
 Indoor Location Intelligence [KE24 batch, Jan 2014]
 Realizing next place prediction in a shopping mall
 Studying the historical data to
work out patterns of the paths and
recognizing the behavior of the
shoppers using the data extracted
from WiFi signal of the shoppers
 An extended application that
relates indoor geographic or location
contexts to business data as part of
the decision making process.

17
 In November 2013, IBM announced Watson as an
application development platform in the Cloud.
 IBM also recently announced
its intention to open Watson
to corporate developers, to
advance a new generation of
apps infused with Watson‘s
cognitive computing intelligence.
 Nearly 2,000 individuals and organizations have contacted IBM to
share ideas for building cognitive applications that redefine how
businesses and consumers make decisions.
IBM Watson: recent news
Users can also be contributors!

18
 Learn and interact naturally with people to extend what
either humans or machines could do on their own.
 Help human experts make better decisions by penetrating
the complexity of Big Data.
 Human and machines working together.
Cognitive Computing Systems

19
 Humans work together with
machines that are more intelligent
 Understanding natural languages
 Computing with words
 Perception-based reasoning
 Competitive & collaborative
 Learning & adapting to changing environment
 … …
Towards the Future
The gold standard is also evolving !
yesterday today tomorrow
image: darleneglasgow.wordpress.com
image: www.doyouknow.inimage: www.genetic-programming.org
image: informatics.indiana.edu

Thank you!

Problem Solving - Human vs. Machine Intelligence

More Related Content

What's hot (20)

Viewers also liked (18)

Similar to Problem Solving - Human vs. Machine Intelligence (20)

More from NUS-ISS (20)

Recently uploaded (20)

Problem Solving - Human vs. Machine Intelligence