Docstoc

Introduction

Document Sample
Introduction Powered By Docstoc
					  Artificial
Intelligence

  Introduction
      2010
                 AI
 You are a caveman (or woman)
 I travel back in time and bring you a
  LapTop PC and show you some of the
  things it is capable of doing.
 Question : Would you, as a caveman,
  consider the computer to be
  intelligent?
 Hands up if you think the computer is
  intelligent?
         Big questions

   Can machines think?
   If so, how?
   If not, why not?
   What does this say about humans?
   What does this say about the mind?
     AI Long Term Goals

Produce intelligent behaviour in machines

   Why use computers at all?
    – They can do things better than us

    – Big calculations quickly and reliably

   We do intelligent things
    – So get computers to do intelligent things
      Aims of the Course
   Define what we mean by AI (or at least
    give us a working definition for this
    course)

   Know how to write “AI” programs that
    – Explore search spaces using both blind
      and heuristic search techniques
    – Implement Neural Networks (Machine
      Learning)
                       Why AI?
   Engineering: To get machines to do a wider variety
    of useful things
     – e.g., understand spoken natural language,
       recognize individual people in visual scenes, find
       the best travel plan for your vacation, etc.
   Cognitive Science: As a way to understand how
    natural minds and mental phenomena work
     – e.g., visual perception, memory, learning,
       language, etc.
   Philosophy: As a way to explore some basic and
    interesting (and important) philosophical questions
     – e.g., the mind body problem, what is
       consciousness, etc.
What’s easy and what’s hard for AI?
   It’s been easier to mechanize many of the high-level tasks
    we usually associate with “intelligence” in people
      – e.g., symbolic integration, proving theorems, playing
         chess, medical diagnosis
   It’s been very hard to mechanize tasks that lots of animals
    can do
      – walking around without running into things
      – catching prey and avoiding predators
      – interpreting complex sensory information (e.g., visual,
         aural, …)
      – modeling the internal states of other animals from their
         behavior
      – working as a team (e.g., with pack animals)
   Is there a fundamental difference between the two
    categories?
What can AI systems do?
Here are some example applications
 Computer vision: face recognition from a large set

 Robotics: autonomous (mostly) automobile

 Natural language processing: simple machine
  translation
 Expert systems: medical diagnosis in a narrow
  domain
 Spoken language systems: ~1000 word continuous
  speech
 Planning and scheduling: Hubble Telescope
  experiments
 Learning: text categorization into ~1000 topics

 User modeling: Bayesian reasoning in Windows help
  (the infamous paper clip…)
 Games: Grand Master level in chess (world
  champion), checkers, etc.
    IBM’s Deep Blue versus Kasparov

   On May 11, 1997, Deep
    Blue was the first computer
    program to beat reigning
    chess champion Kasparov
    in a 6 game match (2 : 1
    wins, with 3 draws)
   Massively parallel
                                       Searched the game tree •
    computation (259th most
                                   from 6-12 ply usually, up to
    powerful supercomputer in
                                     40 ply in some situations.
    1997)
                                   One ply corresponds to –
   Evaluation function criteria
                                          one turn of play.
    learned by analyzing
    thousands of master
    games
      Shakey (1966-1972)   Robotics
                                             Cog (90s)




                            Robocup Soccer
Kismet (late 90s, 2000s)       (2000s)             Boss (2007)
                Autonomous Vehicles

   Think of “robot wars” but with a computer doing the
    controls – not a human on the radio control
   Suppose you are going to take a winning robot from
    robot wars,
    – and make it autonomous,
    – and still win.
   What software components do you think you will
    need?
   Partial List:
    – Vision processing
    – Path planning
           don’t want to fall into holes
    – Action planning: “if I move to the right of my opponent then I can
      use my hammer to attack the weak spot on their left flank”
DARPA grand challenge
    Stanley’s Technology
                                                                             Path
                                                                         Planning


    Laser Terrain Mapping


Learning from Human Drivers                  Adaptive Vision
         Sebastian




                     Stanley



                               Images and movies taken from Sebastian Thrun’s multimedia website.
       Europa Hydrobot
 http://www.resa.net/nasa/images/gem/HYDR
 OBOT.JPG
                Semantic Web

Current web pages are HTML
 Text written with a markup language of
  “tags” to tell the browser what to do
 Hypertext links
    – “click here to go to a different page”
   Structural Markup Only
    – E.g.
        <h2>This  is a level 2 header</h2>
        <p>This is a paragraph with <b>this</b> in
         bold</p>
            “Semantic Web”

Want to move to “semantic markup”
 Text written with a markup language of
  “tags” to tell the reader the meaning of
  the entries
 In contrast, “structural markup” only
  says how to format them on the page
 Example: XML, e.g.
     <person>
        <last-name>Berners-Lee</last-name>
        <first-name>Tim</first-name>
     </person>
                  “Semantic Web”

There is a lot of work on “Ontologies”:
Ontology = Sets of rules for the meanings of
  the entries in a semantic markup language

Want to use the ontologies in order to
 verify that the information is consistent with
  the ontology
    – e.g. for housing might have tags for house type and
      numbers of floors
    – and could check that a “bungalow” has only one floor
   deduce new information
    – e.g. that the bungalow only has one floor even if we are
      not told so explicitly
         Bioinformatics
   MERL: constraint-based           Genetic Motif Discovery
    approach to protein folding       and Mapping
       Why is AI hard?
Two usual ingredients (for standard AI)
 Representation

  – need to represent our knowledge in
    computer readable form
 Reasoning

  – need to be able to manipulate knowledge
    and derive new knowledge
  – many possible ways to do this, but most
    give rubbish
  – finding the successful way usually
    involves search
Both of these are hard.
    The Travelling Salesman
         Problem (TSP)
   A salesperson has to visit a number of cities
   (S)He can start at any city and must finish at that
    same city
   The salesperson must visit each city only once
   For example, with 5 cities a possible tour is:
                 A
                                    C
                            D

                     B
                              E
    Combinatorial Explosion
A 50 City TSP has 1.52 * 1064 possible solutions
Age of the universe is 15 billion (1.5 * 1010) years
There are 30 million seconds in a year
Age of universe is about 45 * 1016 seconds
A 10GHz computer might do 109 tours per second
Running since start of universe, it would still only have
done 1026 tours
Not even close to evaluating all tours!
Need to be clever about how to solve such search
problems!
             Scaling properties

   Big Oh notation: O(f(n))
    – n is a measure of the size of the problem
    – function f(n) is an upper bound on the
      asymptotic (large n) behaviour (of the
      runtime)
    – ignores constant factors
    – e.g. ( 5 n2 + 4 n ) is        O(n2)

   “Polynomial” means O(n), O(n2), etc.
    – That is, O(nk) for some fixed k
   “Exponential” means O(2an) for some
    fixed a.
                                    n/20
             Turing Test
   Three rooms contain a person, a
    computer, and an interrogator.
   The interrogator can communicate with
    the other two by teleprinter.
   The interrogator tries to determine which
    is the person and which is the machine.
   The machine tries to fool the interrogator
    into believing that it is the person.
   If the machine succeeds, then we
    conclude that the machine can think.
    The Loebner contest
   A modern version of the Turing Test, held annually,
    with a $100,000 cash prize.
   Named after Hugh Loebner
   http://www.loebner.net/Prizef/loebner-prize.html
   Restricted topic (removed in 1995) and limited time.
   Participants include a set of humans and a set of
    computers and a set of judges.
   Scoring
     – Rank from least human to most human.
     – Highest median rank wins $2000.
     – If better than a human, win $100,000. (Nobody
       yet…)
                       Who does AI?
   Academic researchers (perhaps the most Ph.D.-generating
    area of computer science in recent years)
     – Some of the top AI schools: CMU, Stanford, Berkeley,
       MIT, UIUC, UMd, U Alberta, UT Austin, ... (and, of course,
       Swarthmore!)
   Government and private research labs
     – NASA, NRL, NIST, IBM, AT&T, SRI, ISI, MERL, ...
   Lots of companies!
     – Google, Microsoft, Honeywell, Teknowledge, SAIC,
       MITRE, Fujitsu, Global InfoTek, BodyMedia, ...
       The course topics
   introduction to AI

   AI application areas

   Knowledge representation

   Search space

   Machine learning
              What is AI?
  There are no crisp definitions
Q. What is artificial intelligence?
A. It is the science and engineering of making
   intelligent machines, especially intelligent computer
   programs. It is related to the similar task of using
   computers to understand human intelligence, but AI
   does not have to confine itself to methods that are
   biologically observable.
Q. Yes, but what is intelligence?
A. Intelligence is the computational part of the ability
   to achieve goals in the world. Varying kinds and
   degrees of intelligence occur in people, many
   animals and some machines.
         AI definition


AI is a branch of computer science and it

 concerned with intelligent behavior.
      AI fundamental

The two fundamental concerns of AI are

 knowledge representation and search.
Knowledge Representation


   It is the problem of capturing in a
    formal language suitable for computer
    manipulation.

   We will study predicate calculus as a
    language for AI
      Search Problem

Search is a problem-solving technique to

 explores successive stages in

 problem-solving process.
         AI applications
1.   Game playing

2.   Theorem proving

3.   Expert System

4.   Natural Language understanding

5.   Planning and Robotics
Knowledge representation


AI is concerned with qualitative rather

 than quantitative problem solving and

 with reasoning rather than calculation.
         Representation
           Language

   An AI representation language must :

    – Handle qualitative knowledge

    – Allow new knowledge to be inferred from

      set of facts and rules
         Search Problem
   We need to define a space to search
    in to find a problem solution

   To successfully design and implement
    search algorithm, we must be able to
    analyze and predict its behavior.
   State Space Search

One tool to analyze the search space is

 to represent it as space graph, so by

 use graph theory we analyze the

 problem and solution of it.
          Graph Theory
 A graph consists of a set of nodes and a
   set of arcs or links connecting pairs of
   nodes.
                    River2




Island1                                 Island2




                    River1
       Graph structure
 Nodes = {a, b, c, d, e}
 Arcs  = {(a,b), (a,d), (b,c),….}
                           d
                 b
         c



             e
                       a
                  Tree
 A tree is a graph in which two nodes
  have at most one path between them.
 The tree has a root.
                   a

              b         c      d

          e   f        g h i       j
 Space representation
In the space representation of a
 problem, the nodes of a graph
 correspond to partial problem solution
 states and arcs correspond to steps in
 a problem-solving process
              Example
   Let the game of Tic-Tac-toe



                1   2   3
                8       4
                7   6   5
                            1       2       3
                            8               4
                            7       6       5




1   2   3   1       4       3           1       4   3       1   4   3
7   4   6   8       7       6           7       8   6       7   6   4
5   8   2   5       8       2           5       6   2       5   8   2




1   1   3       1       3       3           1       4   3       1   4   3
7   4   6       7       4       6           1       7   6       5   7   6
5   8   2       5       8       2           5       8   2       7   8   2
    Strategies for search

   The strategies for state space search

    are: Data-driven and goal-driven

    search
      Data-Driven search
   It is called forward chaining

   The problem solver begins with the
    given facts and a set of legal moves or
    rules for changing state to arrive to the
    goal.
      Goal-Driven Search
   Take the goal that we want to solve
    and see what rules or legal moves
    could be used to generate this goal.

   So we move backward.
Search Implementation
 In both types of moving search, we
  must find the path from start state to a
  goal.
 We use goal-driven search if
    – The goal is given in the problem
    – There exist a large number of rules
    – Problem data are not given
Search Implementation
   The data-driven search is used if
    – All or most data are given
    – There are a large number of potential
      goals
    – It is difficult to form a goal
             Project

We want to form an algorithm to data-

 driven and goal driven search

				
DOCUMENT INFO
Shared By:
Categories:
Stats:
views:9
posted:3/11/2012
language:
pages:49
Description: Artificial intelligence Academic lecture