Fuzzy inference system for learning inference rules.

摘要

A proposition truth value storing means (100) stores a truth value mu A of a first proposition part (A) and a truth value mu B of a second proposition part (B). A rule truth value learning means 125 learns a rule truth value with mu A and mu B. (A) rule validity learning means (150) learns a rule validity degree fr. A rule truth value storing means (120) stores the rule truth value. A rule validity degree storing means (130) stores the rule validity degree fr. A bounded difference calculating means (160) calculates an inference validity degree (fi). A quantizer (165) determines whether or not (fi) is equal to or larger than a predetermined threshold value. An inference result storing means (170) outputs a rule truth value mu R as a truth value of the other of the propositions (A) and (B). A minimum value calculating circuit (180) selects the rule validity degree fr or the inference validity degree fi, whichever smaller. A result validity storing means (190) outputs (fr) or (fi), whichever smaller, as an inference result validity degree.