previous    up   next

Multiple Propositions Rule


Assume the following multiple propositions rule using the premises operator AND:

IF ( tex2html_wrap_inline1518 is tex2html_wrap_inline1520 ) and ...and ( tex2html_wrap_inline1522 is tex2html_wrap_inline1524 ) THEN Y is B.

The generalized modus ponens can be expressed by:

equation286

 

However, the computation of the generalized modus ponens with formula 5 needs the evaluation of several Cartesian products and projections.

So, to reduce computing time and memory space, each n-multiple proposition rule is cut in n simple rules, and the generalized modus ponens can be applied on each simple rule to provide n fuzzy sets tex2html_wrap_inline1532 which are combined in a unique fuzzy set tex2html_wrap_inline1366 with an appropriate combination (depending on the premises operator AND, OR, AVG) as following :

  • AND : tex2html_wrap_inline1536
  • OR : tex2html_wrap_inline1538
  • AVG : tex2html_wrap_inline1540

Effectively, with the 2 following Brouwer-Gödel implication properties:

tex2html_wrap_inline1542

          and

    tex2html_wrap_inline1544


it is easy to show the equivalence between formula 5 and the rule cutting method for the operator AND and OR. However there is no equivalence with the operator AVG ( tex2html_wrap_inline1546 ), but it can be approximated.

Basically, combination operators can be separated in three families [1]:

AVG is a compromise operator between conjunctive and disjunctive operators like tex2html_wrap_inline1548 . Experimental result shows hat the behaviour of the AVG operator is constant with such a computation.


      previous    up   next     
  
 IRIT-UPS