Transcript (Knowledge-Based Systems). - Industrial Engineering Department

Artificial Intelligence and Its
Relevance to Industrial
Engineers
Assoc.Prof.Dr. Hasan H. ÖNDER
In this presentation we will review why industrial
engineers should be interested in expert
systems (AI-knowledge-based )systems.
Brief history of AI
• AI research originated in the 1950’s,
• IEEE Computer Society in December,
1984.
AI research deals
•
•
•
•
using computers to emulate the reasoning,
problem-solving,
creativity,
planning behaviors of human beings so
that they can solve problems that are too
large or too complex to be solved with
traditional techniques.
Branches of AI
•
•
•
•
Fuzzy Sets,
Neuron Networks,
Genetic Algorithms,
Expert Systems (Knowledge-Based
Systems).
Expert Systems or KnowledgeBased Systems.
• expert system is considered to be the area
of AI with the most promise for commercial
applications.
• computer programs that solve difficult
problems that are traditionally by human
experts.
Knowledge-Based Systems.
• Knowledge-based systems are concerned
with applying knowledge to solve that
ordinarily require human intelligence
and/or expertise.
• Expert systems are the knowledge-based
systems that consist of computer
programs design to represent and apply
factual knowledge of specific areas of
expertise to solve problems.
• They emphasize domain-specific problem
solving strategies and employ selfknowledge to reason about their own
interference process and provide
explanation or justifications for conclusions
reached
Expert systems typically consist of
three components:
•
•
•
A knowledge base: consists of all facts,
rules, relations ets. Used by the experts.
inference systems: contains the lines of
reasoning followed by the expert to
organize and control the steps taken to
solve problems in a given domain.
A dialog system: user interface which
communicates with the knowledge-base
through the inference system.
Components of Expert systems
Knowledge
acquition
Knowledge
base
Inference
system
User
interfac
e
Expert systems are needed to supplement,
complement, or replace certain human expert
functions because:
•
•
•
•
•
Human expertise is a scarce resource
whose supply is never guaranteed.
Human get tired, forget, or simply
becomes indolent.
Humans are inconsistent in their day to
day decisions for the same set of data.
Human gets quit, have bad days, harbor
bias, or insubordinate.
Human lie, die, and hide.
The system can;
•
•
•
•
•
•
•
•
•
•
•
diagnose,
monitor,
analyze,
interpret,
consult,
plan,
design,
instruct,
clarify,
learn,
and conceptualize specific topics in industrial engineering.
Industrial Engineer:
• associated with the ‘human aspects’ of job
functions
• concerned about the natural, physical and
emotional limitations of man
• decision-making and management
• Success or failure of may undertaking is
directly dependent on the quality of
decisions. The need for expert systems
arises because there are certain inherent
human characteristics that tend to impair
the optimality of decisions.
• . Computers have been used extensively
for traditional decision support systems
(DSS). Now expert systems are extending
the frontiers of computer usage for
decision-making. This is in direct response
to a need created by the limitations of man
in decision environments.
Application areas of AI and
computers:
•
•
•
•
•
•
•
Manufacturing:
Robotics/vision
Factory Planning and design
CAD
Human factors/ organizational implications
Decision-aids
Human-machine communications and intelligent
interfaces
• Production planning
• Repair and fault –diagnosis
• Process control
Potential IE applications
• Computer aided manufacturing: production
management, robotics, quality control, flexible
manufacturing systems
• Banking, financing, economics,
• Business administration, accounting, human resources
management.
• Law, legislation, regulation, enforcement, contract
management, taxation systems.
• Insurance.
• Office automations,
• Computer aided design: electronics, architecture,
engineering, and construction, structural design etc.
Potential IE applications
•
•
•
•
Technical diagnosis and maintanence.
Logistics.
Computer-aided education
Computer engineering: configuration management,
reliability, safety, auditing, security, capacity and network
management.
• Software engineering: specification, design, verification
and validation, project management, quality control,
maintenance.
• Humanities and social science.
• Decision support systems: command control,
communications, and intelligence, data fusion.
Potential IE applications
• Complex systems control, simulation, simulation
and training.
• Mathematics, statistics and data analysis,
numerical analysis, risk assessment.
• Signal processing and pattern recognition;
vision, speech processing.
• Agriculture and food industry,
• Information retrieval, data base management.
• Project management; planning, scheduling,
monitoring, controlling, heuristics formulation.
Potential IE applications
• With the proliferation of personal
computers and the increased movement
towards micro-computer-based expert
systems, industrial engineers should take
the lead in developing systems that can
provide assistance for real-time
operational decisions.
Potential IE applications
• A question that may be come up at the
point of time is, why should Industrial
engineers be concerned with AI, which is a
domain of computer scientists?
Potential IE applications
• In addition to the need to give more
emphasis to emerging technologies,
(preparing us for a better role in the more
automated systems of the future), AI
technologies offer significant opportunities
to improve performance and also
alternative approaches to deal with some
of our current problems like integration of
CAD/CAM or CIM systems.
Time has come for industrial
engineers
• Industrial engineers by virtue of their
decision-oriented responsibilities are in a
unique position to utilize the emerging
technology of expert systems. Industrial
engineers are will known for their
involvements in decision processes for job
functions ranging from facility location
planning, process control, and economic
analysis to automation of manufacturing
systems.
• Artificial intelligence is being hailed as a
technology that will dictate how business and
industry will operate in the future. Laying the
groundwork now for the marriage of expert
systems and IE job functions will help assure the
realization of the full potentials of this young
branch of computer science. Industrial engineers
can, thus, take the lead in preparing work
environments for the impending ‘invasion’ of
intelligent computers and software.
I.E
Analysis/sysnthesis
DOMAIN EXPERT
Knowledge/experties
KNOWLEDGE ENGINEER
System development
WORKİNG MEMORY
KNOWLEDGE BASE
INFERENCE ENGINE
Recommendation
Consultation
CLIENT
IE and expert systems interfaces
Concluding remarks
• An industrial engineer should consider the knowledgebased as one more tool in his/her bag of tools.
• Computers have changed the industrial Engineer’s way
of thinking and become a part of our systems approach
to problem-solving. It is the IE who integrates skills of
engineering with the tools of mathematics and computer
science to formulate and build models for design,
analysis, evaluation and prediction purposes. Future
computers are predicted to introduce staggering
changes in our abilities to use computers; and AI is part
of this future making it important for some of the IE’s to
get involved in AI.
• The developments in knowledge-based systems
have reached such a stage that they are now
ready to be taken out of laboratories and into the
real world. However, even where they are
suitable for applications, they are not yet enough
to displace the human link completely and it is
not expected that they will replace these experts.
This is because the expert systems knowledge
comes from the knowledge base and the
heuristic rules developed from those solve all
possible decision situations.
• Thus an expert system, once developed to
the prototype stage and found satisfactory,
needs to go through continuous learning
process whenever new situations are
encountered. This learning comes from
adding new knowledge in the knowledge
base and/or adding/modifying heuristic
rules.
• Another imported point is that AI products
often are not stand-alone type of products
and have to be integrated with other
systems in a large context. In AI breadth of
technical understanding will be more
important for successful research. Need
for greater understanding substantiates
the concept that IE’s should take more
interest in AI concepts and applications.