DCT Lexicon · Neuro-fuzzy
ANFIS (adaptive neuro-fuzzy inference)
Adaptive neuro-fuzzy inference system: a Sugeno rule base, with graded premises and affine conclusions, whose proposals are adopted only after an exact check.
Vol. II, Glossary of Key Terms, p. 393, shortened in the book's words
In depth
In the book
Adaptive neuro-fuzzy inference system: a Sugeno rule base, with graded premises and affine conclusions, whose conclusions are fitted by least squares and whose terms by gradient steps; it can approximate any continuous map, and its proposals are adopted only after an exact check (Theorem 10.3).
Vol. II, Glossary of Key Terms, p. 393, verbatim
Formal statement
In a first-order Sugeno system, rule fires at the strength , the product of its terms’ memberships, and concludes the affine value ; the output is (i) Reading. The output is a weighted average of the rules’ conclusions; where one rule fires far more strongly than the rest, the output follows it.
Vol. II, Theorem 10.3 (Neuro-Fuzzy Rules), verbatim
Common misreading
A common mistake is to adopt a learned proposal without the exact check.
Vol. II, LOS 10.4, verbatim
How AXIOM uses it
AXIOM reads the readiness answers a second time as graded declarations, through a first-order Takagi–Sugeno system in ANFIS form, and shows beside the certified step how firmly the answers support it. It certifies nothing: the attainable step is set by the book's rule, the weakest critical capability, and the reading never changes it. The survey use is AXIOM's proposed construct, stated in every response.
Drafted for the author's approval, from AXIOM's method passport "How firm the step is (ANFIS)" (D-093)
In the book: Vol. II, Ch. 10: Neuro-Fuzzy Robust Enterprise Optimization
Related: Membership function (graded declaration) · Readiness · Attainable step
V2.Thm.10.3 · V2.Def.10.1


