850 likes | 1.04k Views
Lecture 10A Overview of Expert Systems. MINE 432 Industrial Automation and Robotics. There is nothing more difficult to carry out, nor more doubtful of success, nor more dangerous to handle than to initiate a n ew order of things. . Niccolò di Bernardo dei Machiavelli ( 1469-1527 )
E N D
Lecture 10AOverview of Expert Systems MINE 432 Industrial Automation and Robotics
There is nothing more difficult to carry out, nor more doubtful of success, nor more dangerous to handle than to initiate a new order of things. Niccolò di Bernardo dei Machiavelli (1469-1527) in Il Principe (The Prince) 1513 (1532)
Elements of AI or Soft Computing Tools and Methods • Expert Systems • Fuzzy Logic • Artificial Neural Networks • Genetic Algorithms Applications • Natural Language Programming • Problem Diagnosis / Troubleshooting • Process Control and Analysis • Robotics and Thinking Machines (Machine Intelligence)
Syntax • Object • Attributes (an Attribute can become an Object) • Values (a Value can be a word, a string, a number) • Conjunctions (IF – AND – OR) • Predicate Functions is is_not is more_or_less is_definitely is not_definitely might_be is_almost is_very is_not_very
Knowledge Elements • Facts and Classes (Degree of Belief, Age) • Rules (inferencing) • Procedures (functions) • Meta-Knowledge • Explanations (What does that mean?); • Rule descriptions (Why are you asking?); • Justifications (How did you make that conclusion?)
Rule Examples • Rule Porridge-popping IF froth.conditions.porridge-popping AND collector.addition_rate.minimum is FALSE THEN collector.addition_rate_change.Negative_Big is TRUE ELSE MACRO ("Water_addition_increase") • Rule Water_addition_increase IF froth.conditions.porridge-popping AND collector.addition_rate.minimum AND feed.Pb_grade.high is FALSE THEN feed.water_addition.Increase is TRUE ELSE feed.tonnage_rate.Decrease is TRUE
Types of Objects <O><A> are known as Keyword Doublets <O><A><V> are known as Keyword Triplets • Logical KWT hat.colour.green • String KWT hat.colour.@string @string assigned "green" • Numerical KWT hat.size.@f assigned 7.5 hat.size.@i assigned 7 hat.size.@d assigned 7.5 hat.size.@l assigned 7
Object Associations • Logical KWT - mutually exclusive sets - multi-choice sets - fuzzy sets • Logical KWT hat.colour.green • String KWT hat.colour.@string @string assigned "green" • Numerical KWT hat.size.@f assigned 7.5 hat.size.@i assigned 7 hat.size.@d assigned 7.5 hat.size.@l assigned 7
Rule Structure Degree of Truth ( IF "A" AND "B" OR "C" AND "D" THEN "G" ELSE "H" 88 >> 88 ) 94 ( 76 >> 65 ) 65 v v 88 If the Confidence Level ≤ 88
Rule Structure Degree of Truth IF "A" AND "B" OR "C" AND "D" THEN "G" ELSE "H" >> 88 88 ( 94 >> 94 ) 76 >> 65 65 v v 65 If the Confidence Level ≤ 65
Rule Structure Degree of Truth IF "A" AND "B" AND "C" OR "D" THEN "G" ELSE "H" >> 88 88 ( 94 >> 76 ) 76 >> 65 65 v v 76 If the Confidence Level ≤ 76
Search Strategies • Forward-chaining (data-driven) • From conclusion to premise • Backward-chaining (goal-driven) • From premise to conclusion • Depth-first searching (likely faster) • Breadth-first searching (more accurate) • Combinations work best • Knowledge states of facts • Unknown (at start) • Instantiated (during the search) • Not known (after searching)
Procedural Search Strategies • FIND ("<O><A><V>") • From conclusion to conclusion • APPLYRULE ("<O><A><V>") • From conclusion to premise • MACRO("Rule_Name") • GOTO ("RULE_Name")
Functions • Inference control functions • String functions • Presentation functions • Mathematical functions • Statistical functions • Data/time functions • Dynamic variable functions • Knowledge base management functions • Miscellaneous functions
Uncertainty in Expert Systems Degree of Belief in a fact false to true 0-100 hat.colour.green 95 Degree of Truth in a statement hat colour is green 95 hat colour might_be green 100 hat colour is_definitely green 0
Uncertainty in Expert Systems Degree of Truth in a premise IF hat colour is green 95 AND shirt colour is_not purple 70 The AND conjunction takes the MINIMUM DoT So this premise has a DoT of MIN(95,70) = 70
Uncertainty in Expert Systems Degree of Truth in a premise IF hat colour is green 95 AND shirt colour is_not purple 70 OR hat colour is green 95 AND shirt colour is_not red 80 The OR conjunction takes the MAXIMUM DoT Premise has a DoT of MAX(MIN(95,70),MIN(95,80)) = 80
Uncertainty in Expert Systems Net Degree of Truth in a conclusion IF hat colour is green 95 AND shirt colour is_not purple 70 OR hat colour is green 95 AND shirt colour is_not red 80 THEN clothing attire is fashionable CF = 90 CF = Certainty Factor (hence the rule is "fuzzy") >> 70 >> 80 >> 80 v v 72
User Interface • Customized questions and explanations • Hypertext documents • Forms for input and output • Animations and pictures • Smart reports • Knowledge units within the interface