Introduction to AI Outline Artificial Intelligence What is an AI?

Outline
What is an AI?
Introduction to AI
ECE457 Applied Artificial Intelligence
Fall 2007
Lecture #1
Russell & Norvig, chapter 1
Agents
Environments
Russell & Norvig, chapter 2
ECE457 Applied Artificial Intelligence
Artificial Intelligence
Assembly-line robots,
auto-pilot, Mars
exploration robots,
RoboCup, etc.
Medical diagnostics,
business advice,
technical help, etc.
Natural language
Spam filtering,
translation, document
summarization, etc.
ECE457 Applied Artificial Intelligence
Systems that…
Humanly
Neural
Think
Act
Expert systems
Computer players in
video games
Robotics
R. Khoury (2007)
Page 2
What is an AI?
Artificial intelligence is all around us
R. Khoury (2007)
Page 3
Rationally
networks
Theorem
proving
ELIZA
Deep Blue
Rationality vs. Humans: emotions, instincts,
etc.
Thinking vs. acting: Turing test vs. Searle’s
Chinese room
Engineers (and this course) focus mostly on
rational systems
ECE457 Applied Artificial Intelligence
R. Khoury (2007)
Page 4
1
Act Rationally
Perceive the environment, and act so as to
achieve one’s goal
Not necessary to do the best action
Think Rationally
There’s not always an absolutely best action
There’s not always time to find the best action
An action that’s good enough can be acceptable
R. Khoury (2007)
Page 5
Think Rationally
2.
3.
X = Y/Z ⇔ XZ = Y
X=Y⇔
X+Z=Y+Z
X*Y+X*Z⇔
X * (Y + Z)
a.
b.
c.
d.
e.
4.
5.
6.
Example: Game playing
Sample approach: Tree-searching strategies
Problem: Choosing what to do given the
constraints
ECE457 Applied Artificial Intelligence
1.
Informal knowledge
Uncertainty
Search space
ECE457 Applied Artificial Intelligence
R. Khoury (2007)
Page 6
Act Humanly
b² = AH * c
a² = BH * c
a² + b² =
BH * c + AH * c
a² + b² =
c * (AH + BH)
a² + b² = c²
“Turing-test” AI
Improve human-machine interactions
up to human-human level
Drawbacks:
b/c = AH/b
a/c = BH/a
AH + BH = c
ECE457 Applied Artificial Intelligence
Uses logic to reach a decision or goal
via logical inferences
Example: Theorem proving
Sample approach: First-order logic
Problems:
R. Khoury (2007)
Page 7
In some cases, requires dumbing down the
AI
Lots of man-made devices work well
because they don’t imitate nature
ECE457 Applied Artificial Intelligence
R. Khoury (2007)
Page 8
2
Think Humanly
Computer vision
Natural language processing
ECE457 Applied Artificial Intelligence
R. Khoury (2007)
Page 9
Types of Agents
Model-based agent
Goal-based agent
Keeps track of perception history
ECE457 Applied Artificial Intelligence
Sensors
R. Khoury (2007)
Percepts
Actions
Actuators
Current
State
Considers what will happen given its actions
Selected
Action
Adds the ability to choose between
conflicting/uncertain goals
If-then
Rules
Adds the ability to learn from its experiences
ECE457 Applied Artificial Intelligence
Page 10
Environment
Learning agent
Actuators
Agent
Program
Sensors
Utility-based agent
A rational agent has
an agent program
that allows it to do
the right action given
its precepts
Simple Reflex Agent
Selects action based only on current perception of
the environment
Environment
Sensors to perceive
its environment
Actuators to act upon
its environment
Simple reflex agent
An agent has
Actions
Cognitive science
Neural networks
Helps in other fields
Percepts
Rational Agents
R. Khoury (2007)
Page 11
ECE457 Applied Artificial Intelligence
R. Khoury (2007)
Page 12
3
Simple Reflex Agent
Model-Based Agent
Dune II (1992) units were
simple reflex agents
Harvester rules:
IF at refinery AND not empty
THEN empty
IF at refinery AND empty
THEN go harvest
IF harvesting AND not full
THEN continue harvesting
IF harvesting AND full
THEN go to refinery
IF under attack by infantry
THEN squash them
ECE457 Applied Artificial Intelligence
R. Khoury (2007)
Page 13
Goal-Based Agent
Environment
Percepts
Current
State
Previous
perceptions
Current
State
Previous
perceptions
World changes
R. Khoury (2007)
Impact of actions
ECE457 Applied Artificial Intelligence
R. Khoury (2007)
Percepts
Page 14
Current
State
Goal
Previous
perceptions
Page 15
Actions
Actuators
Sensors
Selected
Action
Impact of actions
ECE457 Applied Artificial Intelligence
If-then
Rules
World changes
Environment
Actuators
State if I do
action X
Selected
Action
Utility-Based Agent
Actions
Sensors
Actuators
Sensors
Environment
Percepts
Actions
State if I do
action X
Happiness in
that state
World changes
Selected
Action
Utility
Impact of actions
ECE457 Applied Artificial Intelligence
R. Khoury (2007)
Page 16
4
Learning Agent
Properties of the Environment
Environment
Percepts
Actions
Actuators
Sensors
Performance Element
Knowledge
Critic
Feedback
Learning
Problem
Element Learning Goals Generator
R. Khoury (2007)
Page 17
Properties of the Environment
Static vs. dynamic vs. semi-dynamic
Alone vs. team-mates vs. opponents
Sudoku vs. sport team vs. chess
ECE457 Applied Artificial Intelligence
R. Khoury (2007)
Fully observable, deterministic, sequential, static,
discrete, single-agent
Monopoly
Fully observable, stochastic, sequential, static,
discrete, competitive multi-agent
Driving a car
Partially observable, stochastic, sequential,
dynamic, continuous, cooperative multi-agent
Assembly-line inspection robot
Page 19
Page 18
Crossword Puzzle
Single agent vs. cooperative vs. competitive
R. Khoury (2007)
Properties of the Environment
Finite distinct states vs. uninterrupted sequence
Chess vs. driving
Independent episodes vs. series of events
Face recognition vs. chess
ECE457 Applied Artificial Intelligence
World waits for agent vs. world goes on without
agent vs. world waits but agent timed
Translation vs. driving vs. chess with timer
Controlled by agent vs. randomness vs.
multiagents
Sudoku vs. Yahtzee vs. chess
Episodic vs. sequential
Discrete vs. continuous
See everything vs. hidden information
Chess vs. Stratego
Deterministic vs. stochastic vs. strategic
Changes
ECE457 Applied Artificial Intelligence
Performance
standard
Fully observable vs. partially observable
Fully observable, deterministic, episodic, dynamic,
continuous, single-agent
ECE457 Applied Artificial Intelligence
R. Khoury (2007)
Page 20
5