Transcript File

Fundamentals of AI
Introduction
Overview
Syllabus
Grading
 Topics

What is AI?

Four competing views
Agents
Course Goals
Summary
COSC 159 - Fundamentals of AI
2
Syllabus
Instructor information
Prerequisites

Programming Languages
Textbook

Russell and Norvig, AIMA, 2nd Edition
Attendance
COSC 159 - Fundamentals of AI
3
Grading
Qualities of good work
Communication
 Correctness
 Validation
 Comparison
 Efficiency

Your work will be graded on all aspects
COSC 159 - Fundamentals of AI
4
Topics Covered
Definitions of AI
Agents
Problem representation and solving

Searching, heuristics, optimization
Knowledge representation and reasoning

Logic
Planning problems
Uncertainty
Learning
More topics if we have time
COSC 159 - Fundamentals of AI
5
What is AI?
Understand and build intelligent entities

Artificial refers to building entities
What is intelligence?
Understand and build an entity emulating a
human?
 Understand and build an entity that is
rational?

COSC 159 - Fundamentals of AI
6
Rationality
An ideal concept of intelligence
Doing the right thing given available
information
 How do we define the right thing?

Suppose put your hand down on a hot
stove. What is the rational response?
Rationality does not always mean doing
the best possible thing
COSC 159 - Fundamentals of AI
7
COSC 159 - Fundamentals of AI
8
Rationality
Given the situation, was the boss’ action
irrational?
What would make the boss’ action
irrational?
COSC 159 - Fundamentals of AI
9
Competing Views of AI
Many definitions that can be classified
as follows (Russell and Norvig, 2003)
Like Humans
Rationally
Thinking
Acting
COSC 159 - Fundamentals of AI
10
Acting Humanly
Turing test (1950)
COSC 159 - Fundamentals of AI
11
Acting Humanly
Goal: Make computers/entities act like
humans




Natural language processing
Knowledge representation
Automated reasoning
Machine learning
It is not important how the actions are
chosen, as long as results in behavior
indistinguishable from a human
COSC 159 - Fundamentals of AI
12
Thinking Humanly
Understand cognition

Defined as the mental process of knowing,
including aspects such as awareness,
perception, reasoning, and judgment.
Simulate cognition on computers
Cognitive science

Experimental investigation of humans and
animals
The how is important
COSC 159 - Fundamentals of AI
13
Thinking Rationally
Attempt to codify “right thinking”

Aristotle’s syllogisms (reasonings or patterns of
argument)
Logical approach


Formal methods of representing knowledge
Formal methods of reasoning
Again, how a conclusion is reached is
important
COSC 159 - Fundamentals of AI
14
Acting Rationally
Entities that do the right thing
The how isn’t necessarily important
Couple rational thinking with other methods
 What if there is no provably correct action?

Consider the hot stove again

Did the action require rational thought?
Are reflex actions intelligent?
COSC 159 - Fundamentals of AI
15
Applications
Autonomous planning and scheduling

NASA’s Remote Agent
Game playing

IBM’s Deep Blue
Autonomous control

ALVINN, drove 98% of the time across the country
Diagnosis

Medical diagnosis
Pattern recognition

Data mining and bioinformatics
COSC 159 - Fundamentals of AI
16
Characteristics of AI Problems
Frequently hard

NP-hard, which implies there is no known efficient
general solution
Frequently complex

Messy data, such as images, pressure, locations,
natural language, etc.
Frequently imprecise

Uncertain situations
Autonomy

Cannot require human intervention, must adapt
COSC 159 - Fundamentals of AI
17
Agents
An agent is something that acts
In this class, we will build software agents

Agents that act rationally
How are agents different from other
programs?





Autonomous
Perceptive
Persistent
Adaptable
Assume the goals of other agents
COSC 159 - Fundamentals of AI
18
Agents
Agent
Percepts
Sensors
Environment
?
Actuators
Actions
COSC 159 - Fundamentals of AI
19
Definitions
Percept sequence

History of everything agent has perceive
Agent function

Map from percept sequence to action
Agent program

Implementation of agent function
COSC 159 - Fundamentals of AI
20
Example
Consider a world that has a starving monkey
and a banana. Whenever the monkey is in
the same location as the banana, the monkey
will eat it. After eating the banana, the
monkey falls asleep.
We would like to build a simulation for the
environment with a software agent
representing the monkey.
Consider a world with two locations.
COSC 159 - Fundamentals of AI
21
Example
U
D
COSC 159 - Fundamentals of AI
22
Example
Assumptions
Monkey can see the bananas and knows
its location
 Defines percepts: (Location, Contents)

Actions

Up, down, eat, sleep
COSC 159 - Fundamentals of AI
23
Example
Agent function should move monkey to
the bananas, eat the bananas, then
sleep
One possible agent program is to create
a table mapping a percept sequence to
appropriate action

Table-driven agent
COSC 159 - Fundamentals of AI
24
Table
Percept Sequence
Action
(U, Empty)
Down
(U, Bananas)
Eat
(D, Empty)
Up
…
(U, Empty),(D, Bananas)
Eat
(U, Bananas),(U,Empty)
Sleep
…
(D,Empty),(U,Bananas),(U,Empty)
COSC 159 - Fundamentals of AI
???
25
Questions to ponder
Is a table driven agent a good way to
implement rational behavior?
Are all sequences of percepts possible
in the environment?
What if the monkey didn’t know its
location, could you still devise a solution
to the problem? How would the
percepts change?
COSC 159 - Fundamentals of AI
26
Measuring Rational Behavior
What does it mean for an agent to do
the right thing?

The right action is the one causing the
agent to be most successful.
A performance measure embodies the
criterion for an agent’s success.
COSC 159 - Fundamentals of AI
27
Performance Measures
Simple performance measure for
monkey and bananas

The monkey has eaten and fallen asleep.
Suppose you have two monkeys, one
that sleeps right after eating and one
that wanders around and then falls
asleep. Which one is better? Why?
COSC 159 - Fundamentals of AI
28
Performance Measures
Consider more complex environments




What performance measure is appropriate for the
economy?
What about for stocks?
How about medical diagnoses?
What about driving a car?
Performance measures are not easy to
determine, but you must design one for each
environment
COSC 159 - Fundamentals of AI
29
Rationality
Rational behavior at any given time
depends on four things
Performance measure
 Agent’s prior knowledge
 Actions agent can perform
 Agent’s percept sequence

COSC 159 - Fundamentals of AI
30
Course Goals
Understand and build intelligent entities

Rational agents
Formulate search problems

Solve using uninformed and informed algorithms
Represent and reason about knowledge

Logic
Formulate and solve planning problems

STRIPS, partial order planners
COSC 159 - Fundamentals of AI
31
Course Goals (cont.)
Reason in uncertain situations

Probability, Bayesian networks
Introduce machine learning

Inductive learning, decision trees
COSC 159 - Fundamentals of AI
32
For Next Time
Read through chapters 1 and 2.
Think about how you would implement a
simulation for the two location monkey
and banana world.
COSC 159 - Fundamentals of AI
33
Summary
AI is the study and implementation of
intelligent entities
Several perspectives on AI

We will take the rational action perspective
The agent framework provides a
unifying approach to AI
Applications of AI are widespread and
complex
COSC 159 - Fundamentals of AI
34