scientific article; zbMATH DE number 1420883
A fuzzy if-then rule ``if \(A(x)\) then \(B(y)\) is usually interpreted as a fuzzy relation \(\Theta_{AB}(x,y)\) whose truth value, for each \(x\) and \(y\), is equal to \(\theta(A(x),B(y))\) for some logical connective \(\theta\). If we have two rules ``if \(A(x)\) then \(B(y)\) and ``if \(B(y)\) then \(C(z)\), then, intuitively, we can conclude that ``if \(A(x)\) then \(C(z)\); so, we have a fuzzy relation \(\Theta_{AC}(x,z)=\theta(A(x),C(z))\). On the other hand, since the relation between \(A\) and \(C\) is a composition of the relations between \(A\) and \(B\) and between \(B\) and \(C\), we can use the extension principle and compute this relation as \(\widetilde \Theta_{AC}(x,z)=\sup_y (\Theta_{AB}(x,y) \& \Theta_{BC}(y,z))\) for an appropriately chosen t-norm \(\&\). It is desirable to require that these two natural approaches lead to the same relation \(\Theta_{AC}=\widetilde \Theta_{AC}\), i.e., that the \textit{chaining} of fuzzy rules is well-defined. It turns out that for Mamdani-style \(\theta=\&\), the chaining property is true under reasonable conditions that fuzzy sets are normal (attain value 1), and the extension principle uses the same t-norm \(\theta\). Similarly, chaining is well-defined if we use R-implication \(a\to b:=\sup\{c\mid a\& c\leq b\}\) and if both \(A(x)\) and \(B(y)\) attain all the values from the interval \([0,1]\). For S-implication \(a\to b:=b\vee\neg a\), the chaining property is often false, and it is not clear under what reasonable conditions this property holds true. For the entire collection see [Zbl 0927.00008].
- The min-max composition rule and its superiority over the usual max-min composition rule
- When \((S,N)\)-implications are \((T,T _{1})\)-conditional functions?
- Investigations of fuzzy implications satisfying generalized hypothetical syllogism
- scientific article; zbMATH DE number 1206953 (Why is no real title available?)
- scientific article; zbMATH DE number 1303436 (Why is no real title available?)
- Correct models of fuzzy IF-THEN rules are continuous
This page was built for publication:
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4944952)
