Artificial Intelligence

advertisement
ENGG1811
© UNSW, CRICOS Provider No: 00098G
Artificial Intelligence slide 1
ENGG1811
© UNSW, CRICOS Provider No: 00098G
Artificial Intelligence slide 2
ENGG1811: Computing for Engineers
Artificial Intelligence
– Philosophical Start
– What Is Artificial Intelligence?
– Knowledge Representation & Reasoning
– Rule-based Reasoning
– Expert Systems/ Knowledge Based System
– Machine Learning
– Decision Trees
– Seeing, Hearing, and Understanding
– AI in Action
– Can Computers be Creative?
– Summary
A * on slide title means it contains
something examinable
ENGG1811
© UNSW, CRICOS Provider No: 00098G
Artificial Intelligence slide 3
Philosophical Start
On May 12th, 1997, the
best chess player in the
world, Gary Kasparov,
lost a six-game chess
match to a computer
named “Deep Blue 2”
The press called this
“humanity’s endgame”
and a “bloody nose for
human[s]”
What was so significant
about this event?
Being able to program a computer to defeat a Grand Master level
chess player had been a long-standing goal of the science of
artificial intelligence - and it was achieved 1.5 decades ago.
ENGG1811
© UNSW, CRICOS Provider No: 00098G
Artificial Intelligence slide 4
What is Artificial Intelligence?
Intelligence is difficult to define and understand, even for
philosophers and psychologists who spend their lives studying it
But this elusive quality is, to many people, the characteristic that
sets humans apart from other species
“What is intelligence, anyway? It is only a word that people use to
name those unknown processes with which our brains solve
problems we call hard. But whenever you learn a skill yourself, you
are less impressed or mystified when other people do the same.
This is why the meaning of “intelligence” seems so elusive: It
describes not some definite thing but only the momentary horizon of
our ignorance about how minds might work.”
- Marvin Minsky, AI researcher
AI is the field of science devoted to making computers perceive,
reason, and act in ways that have, until now, been reserved for
human beings
ENGG1811
© UNSW, CRICOS Provider No: 00098G
Artificial Intelligence slide 5
Why Study AI?
Alan Turing’s work in computer development was inspired by
his vision of building an artificial mind
The Church-Turing thesis claimed that all problems solved by
human thought are reducible to algorithms - that is, that human
intelligence is essentially equivalent to machine intelligence.
Today, scientists around the world are working on the matter
Why would we want artificial intelligence?

to relieve our mental labour, just as machines relieved our
physical labour last century

it should make machines easier to use

it might give some insight into the workings of our own minds

even if we can’t build fully intelligent machines, we might be
able to amplify our own intelligence using AI
ENGG1811
© UNSW, CRICOS Provider No: 00098G
Artificial Intelligence slide 6
Alan Turing (19121954)
Turing, considered by many to be the father
of computer science, was a brilliant
cryptographer and provided much of the
mathematical basis for cracking the German
Enigma codes during WWII.
Despite being largely responsible for saving
thousands of Allied lives he was later
persecuted by the British authorities for his
homosexuality and committed suicide in
1954, aged just 41.
He chose to enact a Snow White death,
eating an apple laced with cyanide (Steve
Jobs denied that the Apple logo is in homage
to Turing, but it makes a great story).
ENGG1811
© UNSW, CRICOS Provider No: 00098G
Source of
images:
Wikipedia
Artificial Intelligence slide 7
* The Computer as a Universal Machine
• Problems can be solved by means of algorithms. An
algorithm is a series of simple, easy steps guaranteed to get
you to a solution
• Algorithms have an amazing property: universality. This
means that if you have a machine that computes any
algorithm, then it can compute all possible algorithms
• A Turing machine can compute
algorithms. Nearly all modern
computers are based on this abstraction
• Nevertheless, some computations can be shown to be
impossible or conclusions undecidable. The halting problem:
given a program and input, does the program complete or run
forever?
ENGG1811
© UNSW, CRICOS Provider No: 00098G
Artificial Intelligence slide 8
* Turing Test
In 1950, Turing published a paper defining intelligence by imagining a
test, called the Turing test
In this test, an interrogator sits alone in room and types questions
into a computer terminal. The questions can be about anythingmaths, science, politics, sports, art, entertainment, emotions anything. The interrogator tries to guess whether those answers
came from a person or a computer on the other end of the wire
If a computer could answer well enough that the interrogator cannot
tell whether there is a person or a machine answering, the machine is
said to be intelligent
?
ENGG1811
© UNSW, CRICOS Provider No: 00098G
Artificial Intelligence slide 9
Major Areas of AI
There are many areas of study that come under the general
heading of AI:
• Expert Systems (also called Knowledge based systems)
• Machine Learning (also known as Data Mining)
• Natural Language Processing
• Voice Recognition and Understanding
• Artificial Vision
• Robotics
ENGG1811
© UNSW, CRICOS Provider No: 00098G
Artificial Intelligence slide 10
Knowledge Representation & Reasoning
Remember how we need ways of encoding, or representing,
numbers, characters, and pictures in bit patterns so that they
could be processed by computer
An important part of artificial intelligence is working out how to
represent still fancier things like objects, events, situations, goals,
and the relationships between them. This is called knowledge
representation
We might try representing knowledge using:
• propositions in mathematical logic
• sets of IF-THEN rules
• trees or networks of linked tokens
• tables of numbers (e.g. probabilities)
• algorithms
• images
ENGG1811
© UNSW, CRICOS Provider No: 00098G
Artificial Intelligence slide 11
* Rule-based Reasoning
The common method of reasoning used by many experts, and is
relatively easy to port to programs, is Rule-Based Reasoning.
Rule 1
IF temperature is less than or equal to zero degrees Celsius,
THEN pure water will freeze
Rule 2
IF water is frozen,
THEN the water has expanded
If we combine Rule 1 and 2, it can be further concluded that if the
temperature is zero degrees C, water will expand. This method is
known as Chaining and is effectively used in expert systems.
ENGG1811
© UNSW, CRICOS Provider No: 00098G
Artificial Intelligence slide 12
* Chaining
There are two types of chaining:
• Forward Chaining - which starts with facts or data and
works towards an outcome by linking several rules.
• Backward Chaining - using the same rules, start with
the outcome, e.g., “water has expanded”, and reasoning
backwards, you can determine cause, e.g., “cause of
water expanding was a temperature less than or equal to
zero degrees C”
ENGG1811
© UNSW, CRICOS Provider No: 00098G
Artificial Intelligence slide 13
* Expert Systems (Knowledge Based Systems)
It can be argued that the basic components used in problem
solving are knowledge and expertise. When someone is
recognised as having some authority, or superior knowledge
and expertise in a particular are, they are commonly called
experts.
Typical characteristics of experts are:
 they tend to be experts in a particular area or domain
 they possess knowledge that is specific to this area
 they possess expertise: the ability to effectively apply
learnt knowledge to solve a particular problem
ENGG1811
© UNSW, CRICOS Provider No: 00098G
Artificial Intelligence slide 14
* Expert Systems (Knowledge Based Systems)
• An Expert System, sometimes known as Knowledge-based
Systems, is software designed to copy the decision-making
process of a human expert within a specific area. Computerbased expert systems began in the 1960’s.
• The main aim in constructing expert systems is to allow
computers to collect and process human expertise.
• The process of building an expert system is known as
Knowledge Engineering. It is the process of adapting expert
human knowledge into a form that can be used by computers.
ENGG1811
© UNSW, CRICOS Provider No: 00098G
Artificial Intelligence slide 15
Expert Systems in Action
Some early expert systems:
•
•
•
•
•
DEC’s XCON configured complex computer systems. It reportedly did
the work of > 300 human experts, with fewer mistakes
American Express uses one to automate checking for fraud and
misuses of its no-limit credit card. This has to be done in few secs
while the customer waits, and the cost of an error can be high
PROSPECTOR is a geological expert system to help find valuable
mineral deposits
PIERS, an expert system used to diagnose blood samples in St.
Vincent Hospital, Sydney
DENDRAL, an expert system that examines the spectroscopic
analysis of an unknown chemical compound and predicts its
molecular structure
ENGG1811
© UNSW, CRICOS Provider No: 00098G
Artificial Intelligence slide 16
* Data, Information, and Knowledge
• Data is a set of discrete facts about events or objects,
often unorganised
• Information is created by aggregating data (i.e. charts,
tables), which often helps us in understanding the data and
in our decision making process
• Information becomes knowledge with questions like “what
implications does this information have for my final
decision?”
• Knowledge is understanding of information based on its
perceived importance
• Knowledge, not information, can lead to a competitive
advantage in business
ENGG1811
© UNSW, CRICOS Provider No: 00098G
Artificial Intelligence slide 18
* Types of Knowledge:
Explicit Knowledge VS Tacit Knowledge
• Explicit knowledge: knowledge codified and digitized in
books, documents, reports, memos, etc.
• Tacit knowledge: knowledge embedded in the human mind
through experience and jobs, often not easy to codify
• Tacit and explicit knowledge have been expressed in terms
of knowing-how and knowing-that, respectively
• We need to use special techniques to codify Tacit
knowledge
ENGG1811
© UNSW, CRICOS Provider No: 00098G
Artificial Intelligence slide 19
Machine Learning
We’ve seen how difficult collecting and maintaining computerised
knowledge is. Acquisition of knowledge will benefit if a machine could
build up its own knowledge from experiences in the world, like a child
learning how to walk. The ability of the machine to discover knowledge
from observations of the world is called machine learning
The definition of learning, as with intelligence, has long fascinated
psychologists, philosophers, scientist and more generally, by parents of
young children.
Psychologists may define the learning process as
The modification of behavior through practice, training, or experience
Teachers might define it as
To acquire knowledge or skill in by study, instruction, or experience
ENGG1811
© UNSW, CRICOS Provider No: 00098G
Artificial Intelligence slide 20
Machine Learning
Several algorithms are currently researched based on certain
areas of learning. Some of these include:
• Decision Tree Learning
• Learning Set of Rules
• Neural Networks
• Reinforcement Learning
• and many more
ENGG1811
© UNSW, CRICOS Provider No: 00098G
Artificial Intelligence slide 22
* Decision Trees
A very simple (and of course funny) example a decision tree
to help a financial institution decide whether a person should
be offered a loan:
ENGG1811
© UNSW, CRICOS Provider No: 00098G
Artificial Intelligence slide 23
Another Decision Tree Example
• Training Data – used to create decision tree
Outlook
sunny
sunny
overcast
rain
rain
rain
overcast
sunny
sunny
rain
sunny
overcast
overcast
rain
ENGG1811
Temperature Humidity Windy
85
85 false
80
90 true
83
78 false
70
96 false
68
80 false
65
70 true
64
65 true
72
95 false
69
70 false
75
80 false
75
70 true
72
90 true
81
75 false
71
80 true
© UNSW, CRICOS Provider No: 00098G
Play
(positive) /
Don't Play
(negative)
Don't Play
Don't Play
Play
Play
Play
Don't Play
Play
Don't Play
Play
Play
Play
Play
Play
Don't Play
Artificial Intelligence slide 24
* … Decision Trees
• Learned Decision Tree: extracts ‘knowledge’ from
training data
ENGG1811
© UNSW, CRICOS Provider No: 00098G
Artificial Intelligence slide 25
Seeing, Hearing and Understanding
An intelligent computer must be able to recognise its surrounding environment
and adapt to changes in it. To do this it must be able to “see” and “hear” what’s
going on
Computer vision is the capability of a computer to mimic the ways that human
brains process and interpret light waves to produce model of reality. Images from
a camera (or other optical sensor) must be used by the machine to automatically
construct and maintain a useful representation of nearby object or events
ENGG1811
© UNSW, CRICOS Provider No: 00098G
Artificial Intelligence slide 26
Hearing, Seeing and Understanding
The ability of a computer to recognise speech as words is called speech
recognition or voice recognition. The machine must analyse vocal sounds
from a microphone, then find representations of the words that correspond to
patterns in those sounds
It’s one thing to hear words, but quite another to understand them!
Natural language processing is the ability of a computer to build knowledge
representations corresponding to the meaning in sentences made up of
recognised words. This is very difficult, because human language is full of
ambiguities, vagueness and depends on a lot of common sense knowledge of
the world
ENGG1811
© UNSW, CRICOS Provider No: 00098G
Artificial Intelligence slide 27
AI in Action: Google’s Driverless Car
• Uses Artificial Intelligence techniques that
– can sense anything near the car, and
– mimic the decisions made by a human driver.
• More reliable!
– no blind spots! It has 360-degree perception,
– could react faster than humans, and importantly,
– does not get distracted, sleepy or intoxicated.
•
The project led by Google engineer Sebastian Thrun, director of
the Stanford Artificial Intelligence Laboratory and co-inventor of
Google Street View.
• …….. Video presentation …….
http://www.cse.unsw.edu.au/~en1811/12s1/AI-examples.html
ENGG1811
© UNSW, CRICOS Provider No: 00098G
Artificial Intelligence slide 28
AI in Action: Kinect
• Kinect
– is a motion sensing input device by Microsoft (for Xbox 360 and
PC).
– uses an infrared projector and camera and a special microchip to
track the movement of objects and individuals in 3D
– can interpret specific gestures, making completely hands-free
control of electronic devices possible
– it enables users to control and interact through a natural user
interface using gestures and spoken commands.
– The machine learning work on human motion capture within
Kinect won the 2011 MacRobert Award for engineering innovation.
ENGG1811
© UNSW, CRICOS Provider No: 00098G
Artificial Intelligence slide 29
Brain Computer Interface (BCI)
• A brain–computer interface (BCI) is a direct communication
pathway between the brain and an external device.
• Computer programs translate signals received from neurons
into actions, that could direct some external activity, such as
control of a cursor or a prosthetic limb.
• BCIs are often directed at assisting, augmenting, or repairing
human cognitive or sensory-motor functions.
• Currently, deliberate conscious thought is required to control
BCI. In future, BCI are likely to work effortlessly.
• ….. Video demonstration …..
http://www.cse.unsw.edu.au/~en1811/12s1/AI-examples.html
ENGG1811
© UNSW, CRICOS Provider No: 00098G
Artificial Intelligence slide 31
Can Computers be Creative?
More examples:
Which riddle was is machine-created?
A. What do you give a hurt lemon? Lemonade.
B. What kind of tree can you wear? A fir coat.
C. What runs around the forest making other animals yawn? A wild boar.
Which “philosophical insight” is the machine-created one?
A. Distance is the soul of reality.
B. Reflections are images of tarnished aspirations.
C. Love is not consolation, it is light.
ENGG1811
© UNSW, CRICOS Provider No: 00098G
Artificial Intelligence slide 34
Summary
AI is the business of making computers perceive, reason, and act
in ways that have, until now, been reserved for human beings.
We want AI because it could relieve our mental labour, and make
machines easier to use. It could also help us understand our own
minds.
An important part of AI is knowledge representation - encoding
objects, events, situations and relationships in computerised form.
(using logic, rules, tree-structures or networks of objects).
The problems of of gathering, maintaining and applying
knowledge have not yet been completely solved.
ENGG1811
© UNSW, CRICOS Provider No: 00098G
Artificial Intelligence slide 44
Download