Docstoc
EXCLUSIVE OFFER FOR DOCSTOC USERS
Try the all-new QuickBooks Online for FREE.  No credit card required.

Materialy06Lecture6- ICM Expert Systems.ppt

Document Sample
Materialy06Lecture6- ICM Expert Systems.ppt Powered By Docstoc
					Slovak University of Technology
Faculty of Material Science and Technology in Trnava




                Intelligent Control
                Methods
        Lecture 5: Expert Systems
Briefly from the history (recapitulation):

n   Initially: Universal problem solvers, without
    orientation to problem
    ¨ inefficient

n   70th years: Universal problem solvers for
    concrete problems
n   Since 80th years: Special knowledge for
    concrete tasks solution



                                                    2
 Briefly from the history (recapitulation):
AI-system
quality                                        Specialized
                                             knowledge about
                                             concrete problem
                                                   area
                            Universal solution
                           methods for concrete
                                problems
             Universal solution
            methods for universal
                 problems


    1950       1960      1970       1980     1990     2000


                                                                3
n   AI-system is able to solve intelectual tasks, if it
    has knowledge about problem area.

n   Result of this principle: Expert systems
n   ES: Program systems, which use expert
    knowledge for problem solution in narrow
    problem area



                                                      4
Place of ES in AI-systems:

                           AI-systems

                Knowledge systems

                            ES




                                        5
Technology of ES-design:
        Knowledge engineering

        Questions, tasks


    Expert
                        Knowledge      ES
   (problem
                       engineer (IT)
     area)
        Answers, solutions




                                            6
ES-attributes:
  n   Core of ES: Knowledge-base
      ¨ module  separated from inference
      ¨ knowledge formulated explicitly
      ¨ sizeable (2000 – 3000 units)
  n   Solutions on expert level
      ¨ creative,   exact, fast, effective
  n   Explanatory competence
  n   Prognostic possibilities
  n   Institutional memory
  n   Possibilities for experts teaching

                                             7
 ES Architecture:
                     SHELL
                    Explanatory
                      module

DB (facts about                      KB (knowledge
                  Inference module
   problem                           about problem area)

                  Communication
                     module




                       User


                                                           8
ES advantages (compared with
people experts):
n   ES-knowledge are permanent
    ¨ People   forget, are tired and moody
n   ES-knowledge can be easy transferred
    ¨ Copying,   not study
n   ES-knowledge can be easy documented
    ¨ Incontrast to people knowledge they are
      explicitly represented by data structures
n   ES-knowledge can be used in many
    places in the same time                       9
People experts advantages
(compared with ES):
n   People are creative
    ¨ They reorganize information, master
      unexpected situations
n People learn
n People receive information by sense
    ¨ ES:   only texts, sometimes by sensors
n   People reason in context
    ¨ InKB of ES is only knowledge from problem
      area
                                                  10
Implementation of ES is proper, when:


n The task is solvable by phone,
n Experts are disposable and
n The solution takes time 3 min. – 3 ours.




                                             11
Implementation of ES is proper, when:
n   Experts are available and
n   the experts are able to describe their knowledge
    and
n   solution of various experts are not opposite and
n    the solved problem needs only specific, not
    general knowledge and
n   the problem needs only intellectual, not physical
    givennesses and
n   the task is not new and
n   the task is not easy or extremely difficult and
                                                    12
Implementation of ES is proper, when:

n The solution brings profit or
n experts are not available or expensive or
n they are needed in more places in the
  same time or
n knowledge are needed after expert
  outgoing or
n the system should work in dangerous
  conditions.
                                              13
Main application areas of ES:
n   Interpretation (data analysis with the goal to
    determine the state)
    ¨ Datacan be uncertain, incorrect, incomplete...)
    ¨ Used in chemistry, geology, health, ...

n   Prognosis (determination of the current state
    development)
    ¨ Incomplete   information
    ¨ Different alternatives conditioned by next events
    ¨ Prices, harvest, ...



                                                          14
Main application areas of ES:
n   Diagnostics (technical or medical)
    ¨ Determination   of malfunction (illness) by
      symptoms
    ¨ In a broad the same like classification
n   Design (documentation creating needed
    for object implementation)
    ¨ Electronic circuits (computers VAX)
    ¨ Different alternatives by concrete conditions
    ¨ Design of complex units as a set of simple
      ones (modularization)
                                                      15
Main application areas of ES:
n   Planning (activities design)
    ¨ ES  TATR – planning of flying attack
    ¨ Activities depends on conditions, which will be
      known later
    ¨ Many possibilities
n   Monitoring (observation of the actual
    state and its comparison with expected
    one)
    ¨ Real time,
    ¨ Nuclear reactors.
                                                   16
Main application areas of ES
(principally):

n Diagnostic ES: solution alternatives exist
  ahead, ES only select the best ones
  (interpretation, prognosis, diagnostics,
  monitoring)
n Generative ES: solution originates during
  the solution process (planning, design)

                                           17

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:0
posted:7/25/2013
language:English
pages:17