Transcript Lecture 1 - MELODI Lab - University of Washington

EE562
ARTIFICIAL INTELLIGENCE
FOR ENGINEERS
University of Washington,
Department of Electrical Engineering
Spring 2005
Instructor: Professor Jeff A. Bilmes
EE562
•
General Introduction to AI for Engineers
–
•
•
•
Lecturer: Prof. Jeff A. Bilmes <[email protected]>
TA: Winyu Chinthammit <[email protected]>
Course home page: http://sssli.ee.washington.edu/courses/ee562
–
–
•
•
•
Can get extra copies of syllabus, problem sets and labs, announcements, copies of the
slides we’ll be using, and other class information.
Bookmark this page for this quarter.
Prerequisites: basic programming, algorithms and data structures, and basic logic
and probability (or permission of instructor, if you are unsure ask me after class).
Textbook: S. Russell and P. Norvig Artificial Intelligence: A Modern Approach Prentice
Hall, 2003, Second Edition
–
•
what does “for Engineers” mean? We will emphasize practical aspects of AI techniques, and
how to use them for real world problems and system building.
excellent text, the standard in the field.
Homework: There will be 3-4 homeworks assigned for the quarter. They will be
combination of standard work problems but will also involve significant programming
assignments. They will be due roughly 2 weeks after assigned (but don’t start late!!)
Exams: There will be both a midterm (May 2nd, 1.5 hours) and a Final (June 8th, 2
hours)
EE562
• Grading: 33% homework, 33% midterm, and 33% final.
• S/NS: Must do all problem sets (need not do
midterm/final).
•
• Class participation is also counted (attendance, asking
and answering questions).
• Last day of class: June 1st, 2005
• Holiday: May 30th, Veterans day.
• Final Exam: Wed, June 8th, 2:30-4:30.
• Reading This Week: AIMA: Chapters 1 and 2.
Course overview
•
•
•
•
•
10 weeks, 19 1.5 hour lectures.
Introduction and Agents (chapters 1,2)
Search, CSP, Games (chapters 3,4,5,6)
Logic (chapters 8,9,10)
Learning (chapters 18,20)
• See (online) syllabus for more detailed
course outline (we may stray from the
outline depending on how things go).
Outline
•
•
•
•
Course overview
What is AI?
A brief history
The state of the art
What is AI?
• But what is intelligence?
– something not entirely well-defined that helps
to distinguish what we call animate objects
from what we call inanimate objects
– But when does an object become animate?
– Does learning play a role? (can an object be
intelligent without learning?)
– Is “living” a necessary condition? Are there
any non-living objects in the world you might
call intelligent?
What is AI?
• What tasks require intelligence?
• The easy (or seemingly mundane)
– Perception (vision, speech)
– Natural Language (understanding, generation,
translation)
– Common sense reasoning
• rational thought, causality, etc.
– Robotics/Motor skills
What is AI?
• What tasks require intelligence?
• The formal
– Games (chess, backgammon, checkers, go)
– Mathematics (geometry, logic, integral
calculus, theorem proving, program
correctness checkers)
What is AI?
• What tasks require intelligence?
• The expert
– Engineering (design, fault finding,
manufacturing planning)
– Scientific analysis and data interpretation,
data mining, problem finding
– Medical diagnosis (doctors)
– Financial analysis (predict the stock market)
– Forensic Science
– Legal Analysis
What is AI?
• What can Humans do? Object recognition:
Object Recognition
• Sometimes it is a continuum.
– Escher, Liberation, 1955
• What is foreground/background?
– Escher, Mosaic, 1957
Object Recognition
• Why we need uncertainty. Is it a face, a vase, or both?
What is AI?
• What can Humans do?
– Speech Recognition:
What is AI?
Views of AI fall into four categories:
Thinking humanly
Acting humanly
Thinking Rationally
Acting Rationally
• Vertical Axis: Thinking  Acting
• Horizontal Axis: Humanly  Rationally
• The textbook advocates "acting rationally“
–
– this is also an engineering perspective. What is
important to get the problem solved. Acting or being
human? To build systems, we care only about acting.
Thinking humanly
Thinking Rationally
Acting humanly
Acting Rationally
Acting humanly: Turing Test
•
•
•
Alan Turing (1950) "Computing machinery and intelligence":
"Can machines think?"  "Can machines behave intelligently?"
Operational test for intelligent behavior: the Imitation Game
•
Needs:
– natural language processing, knowledge representation, automated reasoning,
machine learning, computer vision, speech recognition, robotics
•
•
•
•
Turing predicted that by 2000, a machine might have a 30% chance of
fooling a lay person for 5 minutes
Anticipated all major arguments against AI in following 50 years
Suggested major components of AI: knowledge, reasoning, language
understanding, learning
Thinking humanly
Thinking Rationally
Acting humanly
Acting Rationally
Thinking humanly: cognitive modeling
• 1960s "cognitive revolution": information-processing “psychology”
replaced orthodoxy of behaviorism
– compute as a human would compute
–
• Requires scientific theories of internal activities of the brain
– what level of abstraction? “Knowledge”, “circuits”, only need a “model”
of the process, don’t need to replicate the process (e.g., neuro-)
– How to validate? Requires
• Predicting and testing behavior of human subjects (top-down), or
• Direct identification from neurological data (bottom-up)
•
• Both approaches (roughly, Cognitive Science and Cognitive
Neuroscience) are now considered distinct from AI (which is more
related to computer science)
• Both share with AI:
– existing theories do not yet explain anything close to resembling true
human-level general intelligence. We have a *long* way to go.
• So the various doctrines share a basic principal direction but are
considered different (sub-)fields.
Thinking humanly
Thinking Rationally
Acting humanly
Acting Rationally
Thinking rationally: "laws of thought"
• Irrefutable (prescriptive rather than descriptive) reasoning
processes that must occur (logic)
• Aristotle: what are correct arguments/thought processes?
– Logical forms that rational thinking possesses.
– Ex: “Socrates is a man, all men are mortal, therefore Socrates is
mortal.”
–
• Several Greek schools developed various forms of logic:
notation and rules of derivation for thoughts; may or may
not have proceeded to the idea of mechanization (which is
what we care about in this class)
•
• Direct line through mathematics and philosophy to modern
AI
•
• Problems with this approach:
1. Not all intelligent behavior is mediated by logical deliberation (many
Thinking humanly
Thinking Rationally
Acting humanly
Acting Rationally
Acting rationally: rational agent
• Rational behavior: doing the right thing
– but we don’t care as much how it is happening as
long as it undeniably is happening.
–
• The right thing: that which is expected to
maximize goal achievement, given the available
information
• Doesn't necessarily involve thinking – e.g.,
blinking reflex – but thinking should be in the
service of rational action
• Aristotle (Nicomachean Ethics):
– Every art and every inquiry, and similarly every action
and pursuit, is thought to aim at some good.
–
Rational agents
• An agent is a key idea in this course.
• An agent is an entity that perceives and acts
•
• This course is about designing rational agents
– agents, build to in one way or another, act “rational”
–
• Abstractly, an agent is a function from percept histories
to actions:
•
[f: P*  A]
• Is this real intelligence? Are we deterministic?
• Practically: For any given class of environments and
tasks, we seek the agent (or class of agents) with the
best performance in a given environment at a particular
time.
AI prehistory
• Philosophy
• Mathematics
• Psychology
•
•
•
•
Economics
Linguistics
Neuroscience
Control theory
• Computer
engineering
• Electrical
Engineering
Logic, methods of reasoning, mind as physical
system, foundations of learning, language,
rationality
Formal representation and proof algorithms,
computation, (un)decidability, (in)tractability,
probability, optimization
phenomena of perception and motor control,
experimental techniques, psycho-*
utility, decision theory, game theory
knowledge representation, grammar
physical substrate for mental activity
design systems that maximize an objective
function over time, temporal processes
building fast computing systems
signal processing, acoustics, sound
Abridged history of AI
•
•
•
•
•
1943
1950
1956
1952—69
1950s
• 1966—73
•
•
•
•
•
•
•
1969—79
1980-1986-1987-1988-1995-2003--
McCulloch & Pitts: Boolean circuit model of brain
Turing's "Computing Machinery and Intelligence"
Dartmouth meeting: "Artificial Intelligence" adopted
Look, Ma, no hands!
Early AI programs, including Samuel's checkers
program, Newell & Simon's Logic Theorist,
Gelernter's Geometry Engine
AI discovers computational complexity
Neural network research almost disappears
Early development of knowledge-based systems
AI becomes an industry
Neural networks return to popularity
AI becomes a science
Uncertain reasoning is acknowledged (Pearl)
The emergence of intelligent agents (our text!!)
Human-level AI is back to popularity
State of the art
• Which of the following can be done by computer at the present time?
– Play a decent game of table tennis
– Drive safely along a curving mountain road
– Drive safely along University Avenue
– Buy a week's worth of groceries on the web
– Buy a week's worth of groceries at Whole Foods Market
– Play a decent game of bridge
– Discover and prove a new mathematical theorem
– Design and execute a research program in molecular biology
– Write an intentionally funny story
– Give competent legal advice in a specialized area of law
– Translate spoken English into spoken Swedish in real time
– Converse successfully with another person for an hour
– Perform a complex surgical operation
– Unload any dishwasher and put everything away
– Recognize fluently spoken conversational speech without mistake
State of the art
• Which of the following can be done by computer at the present time?
– Play a decent game of table tennis
– Drive safely along a curving mountain road
– Drive safely along University Avenue
– Buy a week's worth of groceries on the web
– Buy a week's worth of groceries at Whole Foods Market
– Play a decent game of bridge
– Discover and prove a new mathematical theorem
– Design and execute a research program in molecular biology
– Write an intentionally funny story
– Give competent legal advice in a specialized area of law
– Translate spoken English into spoken Swedish in real time
– Converse successfully with another person for an hour
– Perform a complex surgical operation
– Unload any dishwasher and put everything away
– Recognize fluently spoken conversational speech without mistake
State of the art
• Deep Blue defeated the reigning world chess champion Garry
Kasparov in 1997
• Proved a mathematical conjecture (Robbins conjecture) unsolved for
decades (1997), proved in the affirmative
– are all Robbin’s algebras boolean? Algebra that satisfies commutatively,
associatively, and Robbins equation: n(n(x + y) + n(x + n(y))) = x
• “No hands across America” (driving autonomously 98% of the time
from Pittsburgh to San Diego)
• During the 1991 Gulf War, US forces deployed an AI logistics
planning and scheduling program that involved up to 50,000
vehicles, cargo, and people
• NASA's on-board autonomous planning program controlled the
scheduling of operations for a spacecraft
• Proverb solves crossword puzzles better than most humans
(including myself)
• Question: So do these things really require intelligence? How does
the chess program work so well?