CSC 480: Artificial Intelligence by mWrNSE

VIEWS: 11 PAGES: 20

									                               16 September 2012
     ARTIFICIAL INTELLIGENCE




    Introduction

1
 EXAMPLES OF DEFINITIONS OF AI




                                                                     16 September 2012
 Cognitive     approaches
     emphasis on the way systems work or “think”


 Behavioral      approaches
     only activities observed from the outside are taken into
      account


 Human-like        systems
     try to emulate human intelligence


 Rational     systems
   systems that do the “right thing”
   idealized concept of intelligence                            2
SYSTEMS THAT THINK LIKE HUMANS




                                                             16 September 2012
   “[The automation of] activities that we associate
    with human thinking, activities such as decision-
    making, problem solving, learning …”
    [Bellman, 1978]


   “The art of creating machines that perform
    functions that require intelligence when performed
    by people”
    [Kurzweil, 1990]




                                                         3
SYSTEMS THAT THINK RATIONALLY




                                                  16 September 2012
 “Thestudy of mental faculties through the
 use of computational models”
 [Charniak and McDermott, 1985]

 “The study of the computations that make
 it possible to perceive, reason, and act”
 [Winston, 1992]




                                              4
SYSTEMS THAT ACT RATIONALLY




                                                   16 September 2012
 “Afield of study that seeks to explain and
 emulate intelligent behavior in terms of
 computational processes”
 [Schalkhoff, 1990]

 “The branch of computer science that is
 concerned with the automation of
 intelligent behavior”
 [Luger and Stubblefield, 1993]


                                               5
COGNITIVE MODELING




                                                              16 September 2012
     to construct theories of how the
 Tries
 human mind works

 Usescomputer models from AI and
 experimental techniques from psychology

 MostAI approaches are not directly based
 on cognitive models
   often difficult to translate into computer programs
   performance problems



                                                          6
RATIONAL THINKING




                                                    16 September 2012
 Based    on abstract “laws of thought”
     usually with mathematical logic as tool


 Problems and knowledge must be
 translated into formal descriptions

    system uses an abstract reasoning
 The
 mechanism to derive a solution

        real-world problems may be
 Serious
 substantially different from their abstract
 counterparts                                   7
RATIONAL AGENTS




                                                               16 September 2012
 An   agent that does “the right thing”
  it achieves its goals according to what it knows
  perceives information from the environment
  may utilize knowledge and reasoning to select actions




                                                           8
BEHAVIORAL AGENTS




                                                       16 September 2012
 Anagent that exhibits some behavior
 required to perform a certain task
    may simply map inputs onto actions
    simple behaviors may be assembled into more
     complex ones




                                                   9
FOUNDATIONS OF ARTIFICIAL INTELLIGENCE




                                                                  16 September 2012
 Philosophy
     theories of language, reasoning, learning, the mind


 Mathematics
     formalization of tasks and problems (logic, computation,
      probability)


 Linguistics
   understanding and analysis of language
   knowledge representation



 Psychology
                                                                 10
FOUNDATIONS OF ARTIFICIAL INTELLIGENCE
CONT.




                                            16 September 2012
 Computer      science
    provides tools for testing theories
    programmability
    speed
    storage




                                           11
CONCEPTION (LATE 40S, EARLY 50S)




                                                16 September 2012
 Artificial   neurons (McCulloch and Pitts,
 1943)

 Learning     in neurons (Hebb, 1949)

 Chess   programs (Shannon, 1950; Turing,
 1953)

 Neural   computer (Minsky and Edmonds,
 1951)
                                               12
BABY STEPS (LATE 1950S)




                                                       16 September 2012
 Demonstration of programs solving simple
 problems that require some intelligence

 Development     of some basic concepts and
 methods
   Lisp (McCarthy, 1958)
   formal methods for knowledge representation and
    reasoning




                                                      13
(EARLY 1960S)




                                            16 September 2012
 GeneralProblem Solver (Newell and
 Simon, 1961)

 Shakey    the robot (SRI)

 Algebraic   problems (Bobrow, 1967)

 Neuralnetworks (Widrow and Hoff, 1960;
 Rosenblatt, 1962; Winograd and Cowan,
 1963)


                                           14
(LATE 60S, EARLY 70S)




                                             16 September 2012
       networks can learn, but not very
 Neural
 much (Minsky and Papert, 1969)

       systems are used in some real-life
 Expert
 domains

 Knowledgerepresentation schemes
 become useful



                                            15
AI GETS A JOB (EARLY 80S)




                                                            16 September 2012
 Commercial       applications of AI systems
     R1 expert system for configuration of DEC computer
      systems (1981)


 Expert    system shells

 AI   machines and tools




                                                           16
(LATE 80S)




                                             16 September 2012
      all, neural networks can learn more
 After
 in multiple layers (Rumelhart and
 McClelland, 1986)

 HiddenMarkov models help with speech
 problems




                                            17
(90S)




                                                    16 September 2012
 Handwriting    and speech recognition work

 AI   is in the driver’s seat (Pomerleau, 1993)




                                                   18
INTELLIGENT AGENTS APPEAR (MID-90S)




                                                       16 September 2012
 Distinctionbetween hardware (robots) and
 software (softbots)

 Agent    architectures

 Situated    agents
     embedded in real environments with continuous
      inputs


 Web-based      agents

                                                      19
CHAPTER SUMMARY




                                               16 September 2012
 Introduction    to important concepts and
 terms

 Relevance    of Artificial Intelligence

 Influence    from other fields

 Historical development of the field of
 Artificial Intelligence


                                              20

								
To top