ICS 101 Fall 2011 Introduction to Artificial Intelligence Asst. Prof. Lipyeow Lim Information &...
-
Upload
hester-hunt -
Category
Documents
-
view
219 -
download
1
Transcript of ICS 101 Fall 2011 Introduction to Artificial Intelligence Asst. Prof. Lipyeow Lim Information &...
1
ICS 101 Fall 2011
Introduction to Artificial Intelligence
Asst. Prof. Lipyeow LimInformation & Computer Science Department
University of Hawaii at Manoa
10/11/2011 Lipyeow Lim -- University of Hawaii at Manoa
2
What is Artificial Intelligence ?
10/11/2011 Lipyeow Lim -- University of Hawaii at Manoa
3
What is human intelligence ?
10/11/2011 Lipyeow Lim -- University of Hawaii at Manoa
4
What are signs (activities, abilities etc) of human
intelligence ?
Exercise 1: Write down four examples in your worksheet
10/11/2011 Lipyeow Lim -- University of Hawaii at Manoa
5
Approaches to A.I.
10/11/2011 Lipyeow Lim -- University of Hawaii at Manoa
Human-oriented Rationalist
Thinking
Acting
Thinking Humanly Thinking Rationally
Acting Humanly Acting Rationally
6
Definitions of AI (a)• “The exciting new effort to make computer think ... machines
with minds, in the full and literal sense.” (Haugeland, 1985)
• “[The automation of] activities that associate with human thinking, activities such as decision-making, problem solving, learning ...” (Bellman, 1978)
• “The art of creating machines that perform functions that require intelligence when performed by people.” (Kurzweil, 1990)
• “The study of how to make computers do things, at the moment, people are better.” (Rich and Knight, 1991)
10/11/2011 Lipyeow Lim -- University of Hawaii at Manoa
7
Definitions of AI (b)• “The study of mental faculties through the use of
computational models.” (Charniak and McDermott, 1985)
• “The study of the computations that make it possible to perceive, reason, and act.” (Winston, 1992)
• “Computational Intelligence is the study of the design of intelligent agents.”(Poole et al., 1998)
• “AI ... is concerned with intelligent behavior in artifacts.” (Nilsson, 1998)
10/11/2011 Lipyeow Lim -- University of Hawaii at Manoa
8
Acting Humanly: Turing Test (1950)
• Operational test of intelligence• Anticipated all major arguments against AI in following 50 years• Suggested major components of AI: knowledge, reasoning,
language understanding, learning
10/11/2011 Lipyeow Lim -- University of Hawaii at Manoa
Human Interrogator
Human
A.I.
?
9
Thinking Humanly : Cognitive Science
• AI thinks like humans do• How do humans think ?• How can we find out ?
– Introspection– Psychological experiments– Brain imaging
• The goal is to formulate computer programs that mimic how humans think and hence achieve AI!
10/11/2011 Lipyeow Lim -- University of Hawaii at Manoa
10
Thinking Rationally• Aristotle: what are correct arguments/thought
processes?– Syllogism:
• Socrates is a man;• All men are mortal• Therefore Socrates is mortal
– Field of logic• AI programs represent knowledge using
formal logic and solves problems using logical inference/reasoning.
10/11/2011 Lipyeow Lim -- University of Hawaii at Manoa
11
Acting Rationally
• Acting rationally == doing the right thing• What is the “right thing” ?
– Logical / rational– maximize goal achievement, given the available
information• This approach is the focus of many AI efforts!• AI programs are rational agents : programs
that act so as to achieve the best outcome or best expected outcome
10/11/2011 Lipyeow Lim -- University of Hawaii at Manoa
12
AI Today• Robotic Vehicles: Google Self-Drive Car• Speech Recognition: Call routing, Call center• Autonomous planning: Mars Rover• Game Playing: Deep Blue, Watson• Spam Fighting• Logistic Planning: Dynamic Analysis &
Replanning Tool (DART)• Robotics : Roomba• Machine Translation10/11/2011 Lipyeow Lim -- University of Hawaii at Manoa
13
Intelligent Agents
• Perceives its environment through sensors• Acts upon the environment through actuators• Percepts – perceptual input at any given instant• Agent program implements how to map a sequence of
percepts to an action
10/11/2011 Lipyeow Lim -- University of Hawaii at Manoa
Agent
EnvironmentAgent
Program
Sensors
Actuators Actions
Percepts
14
Example: Vacuum Robot
• Vacuum Robot (“agent”) needs to keep two rooms A & B clean. It can sense which room it is in and whether the carpet in that room is dirty. It can either go Right, go Left, or Suck.
10/11/2011 Lipyeow Lim -- University of Hawaii at Manoa
A B
************ ************
15
Example: Vacuum Robot Agent Program
Percept Sequence Action
[A,Clean] Go Right
[A, Dirty] Suck
[B, Clean] Go Left
[B, Dirty] Suck
[A, Clean], [A, Clean] Go Right
10/11/2011 Lipyeow Lim -- University of Hawaii at Manoa
Agent
EnvironmentAgent Program
Sensors
Actuators Actions
PerceptsA B
********* *********
16
Representation & Search• Newell & Simon argue that intelligent activity (human
or machine) is achieved by:
10/11/2011 Lipyeow Lim -- University of Hawaii at Manoa
– Representing significant aspects of a problem using symbol patterns
– Generating potential solutions by applying operations on the representation
– Selecting a solution by searching among these possibilities
17
Example: Tic-Tac-Toe• 2 Player Game: Each gets a symbol 0 or X• Each player tries to get 3 of his/her symbol in a
row/column/diagonal in a 3 by 3 grid.
10/11/2011 Lipyeow Lim -- University of Hawaii at Manoa
Player A
X
0
X
0 X
X
0 X
0 X
0 X
0 X
X
Player B Player A Player B
Player A
Player A Wins !
18
Example: State Space for Tic-Tac-Toe
10/11/2011 Lipyeow Lim -- University of Hawaii at Manoa
x x x
x...x x
0 x 0 x x
0
x
0...
x
0
x
0
... ... ......
19
Exercise • Draw the state space for the vacuum robot
starting from the following initial state for the next two state transitions.
10/11/2011 Lipyeow Lim -- University of Hawaii at Manoa
A B
************ ************
20
Podcast Supplement
Asst. Prof. Lipyeow LimInformation & Computer Science Department
University of Hawaii at Manoa
10/11/2011 Lipyeow Lim -- University of Hawaii at Manoa
21
Approaches to A.I.
10/11/2011 Lipyeow Lim -- University of Hawaii at Manoa
Human-oriented Rationalist
Thinking
Acting
Thinking Humanly Thinking Rationally
Acting Humanly Acting Rationally
22
Acting Humanly: Turing Test (1950)
• Operational test of intelligence• Anticipated all major arguments against AI in following 50 years• Suggested major components of AI: knowledge, reasoning,
language understanding, learning
10/11/2011 Lipyeow Lim -- University of Hawaii at Manoa
Human Interrogator
Human
A.I.
?
23
Thinking Humanly : Cognitive Science
• AI thinks like humans do• How do humans think ?• How can we find out ?
– Introspection– Psychological experiments– Brain imaging
• The goal is to formulate computer programs that mimic how humans think and hence achieve AI!
10/11/2011 Lipyeow Lim -- University of Hawaii at Manoa
24
Thinking Rationally• Aristotle: what are correct arguments/thought
processes?– Syllogism:
• Socrates is a man;• All men are mortal• Therefore Socrates is mortal
– Field of logic• AI programs represent knowledge using
formal logic and solves problems using logical inference/reasoning.
10/11/2011 Lipyeow Lim -- University of Hawaii at Manoa
25
Acting Rationally
• Acting rationally == doing the right thing• What is the “right thing” ?
– Logical / rational– maximize goal achievement, given the available
information• This approach is the focus of many AI efforts!
10/11/2011 Lipyeow Lim -- University of Hawaii at Manoa
AI programs are rational agents : programs that act so as to achieve the best outcome or best expected outcome
26
Intelligent Agents
• Perceives its environment through sensors• Acts upon the environment through actuators• Percepts – perceptual input at any given instant• Agent program implements how to map a sequence
of percepts to an action to achieve some goal
10/11/2011 Lipyeow Lim -- University of Hawaii at Manoa
Agent
EnvironmentAgent
Program
Sensors
Actuators Actions
Percepts
27
Example: Vacuum Robot (“Agent”)
• Goal/outcome: keep two rooms A & B clean. • Percepts: which room it is in and whether the
carpet in that room is dirty. • Actions: go Right, go Left, or Suck.
10/11/2011 Lipyeow Lim -- University of Hawaii at Manoa
Agent
EnvironmentAgent Program
Sensors
Actuators Actions
PerceptsA B
********* *********
28
Example: Vacuum Robot Agent Program
Percept Sequence Action
[A,Clean] Go Right
[A, Dirty] Suck
[B, Clean] Go Left
[B, Dirty] Suck
[A, Clean], [A, Clean] Go Right
10/11/2011 Lipyeow Lim -- University of Hawaii at Manoa
Agent
EnvironmentAgent Program
Sensors
Actuators Actions
PerceptsA B
********* *********
29
Example: Tic-Tac-Toe Agent
• Goal/outcome: Play by the rules and win• Percepts: the tic-tac-toe board and the
human’s move on the board.• Actions: put a “0” on any of the empty cell on
the board
10/11/2011 Lipyeow Lim -- University of Hawaii at Manoa
Human Agent
0
X
30
Example: Car Driving Agent
• Goal/outcome: Navigate and drive to destination safely obeying traffic rules
• Percepts: its own location, speed, direction, and location, speed, direction of nearby objects
• Actions: accelerate, decelerate, hardstop, turn x degrees left, turn x degrees right, maintain speed
10/11/2011 Lipyeow Lim -- University of Hawaii at Manoa
Agent
EnvironmentAgent Program
Sensors
Actuators Actions
Percepts
31
Representation & Search• Newell & Simon argue that intelligent activity
(human or machine) is achieved by:– Representing significant aspects of a problem
using symbol patterns– Generating potential solutions by applying
operations on the representation– Selecting a solution by searching among these
possibilities
10/11/2011 Lipyeow Lim -- University of Hawaii at Manoa
32
Example: Vacuum Robot
• State representation: – Cleanliness status of each room– Location of robot
• State Transitions (generating possibilities)– Triggered by actions of robot (L, R, S)
• Goal states (selecting solutions)– Any state where both rooms are clean
10/11/2011 Lipyeow Lim -- University of Hawaii at Manoa
A B
***** *****Right
A B
*****
A B
*****Suck
A B
Suck
33
Example: Vacuum Robot State Space
10/11/2011 Lipyeow Lim -- University of Hawaii at Manoa
A B
***** *****
RA B
*****A B
*****
S
A B
SR
A B
**********
L
A B
*****
L
L
R
S
RA B
S
A B
*****
LL
S
R
R
L
L RS
S
L
R
34
Example: Tic-Tac-Toe• State representation:
– Tic-tac-toe board– Location of X’s and 0’s
• State Transitions (generating possibilities)– Triggered by players putting
X/0 on empty cells• Goal states (selecting
solutions)– Any state where symbols form
a line10/11/2011 Lipyeow Lim -- University of Hawaii at Manoa
Player A
X
0
X
0 X
X
0 X
0 X
0 X
0 X
X
Player B Player A
Player B Player A
Player A Wins !
35
Example: State Space for Tic-Tac-Toe
10/11/2011 Lipyeow Lim -- University of Hawaii at Manoa
x x x
x...x x
0 x 0 x x
0
x
0...
x
0
x
0
... ... ......
36
Summary• Four different approaches to AI
– Acting vs Thinking– Human-oriented vs Rationalist
• Intelligent Agents– Sensors– Actuators– Agent Program
• Representation and Search– Representing the state of the problem– Generating the state space– Searching the state space for goal states
10/11/2011 Lipyeow Lim -- University of Hawaii at Manoa