Introduction to Artificial Intelligence 37213502 Fall 2002

1
Introduction to Artificial Intelligence(372-1-350
2)Fall 2002
2
Course Overview

• Practical aspects
• 13 lessons
• Lisp
• Specific topics, systems and programs
• General issues and ideas
• 6 exercises (20)
• Final exam (80)
• Course site
• Lecture notes
• Slides
• Assignments
• Exercises turn-in

3
Course Overview

• Practical aspects (cont)
• Instractor
• Tsvika
• Email tsvikak_at_bgumail
• Office room 11 bldg 93
• Tel [phone]
• Office hours Wed. 0900-1100
• Teaching assistant
• Diana
• Email diana_at_bgumail
• Office room 11 bldg 93
• Tel [phone]
• Office hours Wed. 0900-1100

4
Course Overview

• Practical aspects (cont)
• Support software, Exercises and programming
  Diana later today.

5
Course Overview

• Topics
• Introduction of AI
• What is AI and a brief history of AI
• Historical systems, ideas and implementation
• AI basics
• Search
• Knowledge representation
• Applications and advanced topics
• Introduction to paradigms of AI programming

6
Course Overview

• Books
• Artificial Intelligence a Modern Approach
  Russell Norvig Intro and bakground
• Introduction to paradigms of AI programming
  Norvig Specific issues, examples and code

7
What is AI?

• Some definitions of intelligence
• Ability to learn or understand or to deal with
  new or trying situations the skilled use of
  knowledge Webster
• Ability to apply knowledge to manipulate one's
  environment or to think abstractly as measured by
  objective criteria (as tests) Webster
• Ability to acquire, analyze, understand and
  creatively apply the knowledge
• Ability to reason (think) and intelligently
  handle (behave) information.
• AI therefore draws from many other fields
  linguistics cognitive science psychology
  philosophy mathematics physicology neurology
  prosthetics engineering

8
What is AI?

• We can't really define AI well, because
• We lack a good definition of intelligence
• What's considered AI research is very broad and
  continues to change

9
What is AI?

• AI strives to build intelligent entities as well
  as understand them
• Formally initiated in 1956
• AI encompasses a huge variety of subfields
• Knowledge representation
• Reasoning
• Machine learning
• More

10
What is AI?
The automation of activities that we
associate with human thinking, activities such as
decisionmaking, problem solving, learning
'' (Bellman, 1978)
The study of mental faculties through the use
of computational models'' (CharniakMcDermott,
1985)
The study of how to make computers do things at
which, at the moment, people are better''
(RichKnight, 1991)
The branch of computer science that is
concerned with the automation of intelligent
behavior'' (LugerStubblefield, 1993)
11
What is AI?
Views of AI fall into four categories
Thinking humanly
Thinking rationally
Acting humanly
Acting rationally
12
What is AI?

• Building systems that think like humans
• Machines with minds
• Automate human thinking
• How do we do this?
• Develop a precise theory of mind, through
  experimentation and introspection, then write a
  computer program that implements it.

13
What is AI?

• How do we know when we've got it right?
• Present a problem to both humans and computer
• Trace the steps both follow to obtain their
  answers
• Compare the results
• Notes
• This approach is more concerned with capturing
  the process than with that process' results.
• Researchers using this approach might prefer that
  a program obtain the wrong'' answer than the
  right'' one if that is what a human would do.

14
What is AI?

• Building systems that act like humans
• Doing things that (we think) require intelligence
  
• Doing things that humans presently do better than
  computers

15
What is AI?

• How do we do this?
• Implement all cognitive tasks
• Cognitive tasks include
• Natural language processing (for communication)
• Knowledge representation (to store information)
• Automated reasoning (to answer questions)
• Machine learning (to adapt)
• Computer vision (for perception)
• Robotics (to manipulate things)

16
What is AI?

• How do we know when we've got it right?
• Write programs that perform all' cognitive tasks
  
• See if the computer can fool an interrogator
• The Turing Test (circa 1950)

17
What is AI?

• The Turing Test
• Turing said that we could tell when
  intelligence'' had been achieved when a
  computer could fool an interrogator into thinking
  that the computer is a human.
• Turing suggested using a teletype for interaction
  between interrogator and subject later
  suggestions include use of video signal to test
  perception/manipulation as well.

18
What is AI?

• Notes
• The techniques the computer uses to act as a
  human need not be anything like the techniques a
  person would use.
• Researchers here don't care (much) about the
  process used to achieve effects, but rather the
  achievement of the effects themselves.
• Researchers here again would prefer a wrong''
  answer to a right'' answer if that is what a
  human would do.
• Some feel that once a computer does something as
  well as or better than a human, it is not
  properly artificial intelligence at work!

19
What is AI?

• Building systems that think rationally
• Capture correct'' reasoning processes
• Strategies for complex problem solving A loose
  definition of rational thinking Irrefutable
  reasoning process
• How do we do this?
• Develop a formal model of reasoning (formal
  logic) that always'' leads to the right''
  answer
• Implement this model.
• How do we know when we've got it right?
• When we can prove that the results of the
  programmed reasoning
• are correct

20
What is AI?

• Notes
• It is really hard to represent some information
  in a formal way, particularly when data is
  contradictory or incomplete
• Most algorithms for formal reasoning are very
  expensive
• Researchers here would be unhappy with a system
  that came up with
• the wrong answers from the right'' data, even
  if that's what a human would do.
• Researchers usually don't care if the reasoning
  process is anything like that of a human
• Researchers want the process to be correct''

21
What is AI?

• Building systems that act rationally
• Emulate intelligent behavior
• Act so that desired goals are achieved How do we
  do this?
• Figure out how to make correct decisions, which
  sometimes means thinking rationally and other
  times means having rational reflexes
• Implement Turing test cognitive skills'' to
  perceive and act

22
What is AI?

• How do we know when we've got it right?
• When goals are achieved.
• Notes
• We will concentrate on this approach, also known
  as the rational agent approach
• Researchers want the system to do the right thing
  
• Researchers don't usually care if the process is
  that of a human
• Researchers want to be sure that the system will
  act properly

23
What is AI?

• Subfields of Artificial Intelligence
• Problem solving
• Lots of early success here
• Solving puzzles
• Playing chess
• Mathematics (integration)
• Uses techniques like search and problem reduction

24
What is AI?

• Logical reasoning
• Prove things by manipulating database of facts

25
What is AI?

• Language understanding and semantic modeling
• One of the earliest problems
• Some success within limited domains
• How can we understand' written/spoken language?
• Includes answering questions, translating between
  languages, learning from written text, and speech
  recognition

26
What is AI?

• Some aspects of language understanding
• Associating spoken words with actual' word
• Understanding language forms, such as
  prefixes/suffixes/roots
• Syntax how to form grammatically correct
  sentences
• Semantics understanding meaning of words,
  phrases, sentences
• Context
• Conversation

27
What is AI?

• Automatic Programming
• Writing computer programs given some sort of
  description
• Some success with semi-automated methods
• Some error detection systems

28
What is AI?

• Pattern Recognition
• Computer-aided identification of
  objects/shaps/sounds
• Needed for speech and picture understanding
• Requires signal acquisition, feature extraction,
  ...

29
What is AI?

• Expert Systems and Expertise
• Somewhat more recent.
• Designers often called knowledge engineers
• Translate things that an expert knows and rules
  that an expert uses to make decisions into a
  computer program
• Problems include
• Knowledge acquisition (or how do we get the
  information)
• Explanation (of the answers)
• Knowledge models (what do we do with info)
• Handling uncertainty

30
What is AI?

• Planning, Robotics and Vision
• Planning how to perform actions
• Manipulating devices
• Recognizing objects in pictures
• Machine Learning and Neural Nets
• Can we remember' solutions, rather than
  recalculating them?
• Can we deduce additional facts from present data?
  
• Can we model the physical aspects of the brain?

31
What is AI?

• Languages and Environments
• AI investigation has led to the development of
  new languages and environments, partially because
  of the size and scope of the interesting
  problems.
• Examples
• LISP, Prolog, CLIPS
• Object oriented techniques (according to some)
• Automated programming

32
What is AI?

• Techniques important in AI
• Data acquisition and representation.
  Intelligence/intelligent behavior requires
  knowledge, which is
• Voluminous
• Hard to characterize
• Constantly changing

33
What is AI?

• How can one capture formally (i.e., computerize)
  everything needed for intelligent behavior? Some
  questions...
• How do you store all of that data in a useful
  way?
• Can you get rid of some?
• How can you store decision making steps?

34
What is AI?

• Characteristics of good data representation
  techniques
• Captures general situation rather than being
  overly specific
• Understandable by the people who provide it
• Easily modified to handle errors, changes in
  data, and changes in perception
• Of general use
• Self-refining''

35
What is AI?

• Search methods.
• How can we move between steps in a decision
  making process?
• How can you find the info you need in a large
  data set?
• Given a choice of possible decision sequences,
  how do you pick a good one?

36
What is AI?

• Planning.
• Given a goal, how do you figure out what to do?
• Programming languages
• Is there a programming language that can help?

37
What is AI?

• Primary AI Languages (Reference Luger, 1993)
• Prolog
• First Prolog program France, 1970
• Developed as part of a natural languageunderstandi
  ng project.
• Based on theorem proving research
• Major development at University of Edinburgh,
  1975-79
• Adopted by the Japanese Fifth Generation
  Computing Project circa 1980

38
What is AI?

• Logic programming language
• Programs composed of facts and rules
• Executes by applying first-order predicate
  calculus/unification to programs
• Interactive interpreter, compiler
• Tell the computer what is true and what needs to
  be done, rather than how to do it.

39
What is AI?

• Example
• likes(deb, horses) . likes(deb, dogs) . ?-
  likes(deb, horses) . yes ?- likes(deb, X) .
  Xhorses Xdogs We can also add rules, such as
  the following
• likes(deb, Y) - horse(Y) . horse(robin) . ?-
  likes(deb, robin) . yes

40
What is AI?

• LISP
• Proposed by McCarthy, late 1950s contemporary of
  COBOL, FORTRAN.
• Functional programming language based on lambda
  calculus/recursive function theory
• Intended as a language for symbolic rather than
  numeric computation
• Originally very simple, but has since been
  extended
• Interactive interpreter, compiler
• Uses atoms, lists, functions.

41
What is AI?

• Example
• (defun hypotenuse (x y) (sqrt ( (square x)
  (square y)))) gt (hypotenuse 4 3) 5

42
What is AI?

• To summarize
• Provide solution to problems where there is no
  practical algorithmic solution, while human
  solves these problems with reasonable succsess

Thinking humanly
Thinking rationally
Acting humanly
Acting rationally
43
AI brief history

• Early years 1950-1970
• Early systems and tools GPS, Eliza, Lisp,
  Neural Networks, Genetic Algorithms
• 1970- 1980
• Knowledge based systems
• 1980- 1990
• AI becomes an Industry
• 1990
• Machine learning ANNs, GA, Belief Networks
• Intelligent Agents
• Vision, Speech