Toward fast reasoning for fuzzy logic inference

This paper presents a method for seducing computations in fuzzy logic inference. Reduction of computations is achieved via three points: (1) avoid match-degree computations over the whole universe; (2) remove redundant match-degree calculation in common IF-clauses; and (3) omit calculations of the same fuzzy value in the ´THEN-part´ of different rules. By using these techniques, we show several times improvements over previous methods by truck backer-upper control systems. We also note that no information of linguistic concept is lost in proposed method.