Fuzzy Reasoning

Origin. Lotfi Zadeh's fuzzy set theory (1965) was extensively adopted and applied in Japan starting in the 1980s. Japanese engineers applied fuzzy logic to consumer products (cameras, washing machines, rice cookers) and industrial control systems, demonstrating practical value that Western AI research initially dismissed.

Mechanism. Classical logic requires precise categories; fuzzy logic permits graded membership. A temperature is not simply "hot" or "cold" but belongs to each category to a degree. Rules operate on fuzzy sets: "if temperature is somewhat high and trend is increasing, then increase cooling moderately." The outputs of multiple rules are aggregated and defuzzified into a crisp action. This captures expert heuristics that resist precise formalization.

Procedure. Define the input variables and their fuzzy sets (linguistic terms with membership functions). Define the output variables similarly. Encode expert knowledge as fuzzy rules: if [fuzzy condition] then [fuzzy action]. For each input, compute its membership in each relevant fuzzy set. Fire all applicable rules; aggregate their outputs. Defuzzify to produce a crisp control action (typically by centroid).

Applies to. Control systems where precise models are unavailable, consumer products, decision support systems, any domain where experts reason in qualitative terms.

Limitations. Membership functions and rules are designed by the engineer; the system cannot learn them (though neuro-fuzzy hybrids address this). Fuzzy systems are difficult to verify and can produce unexpected behaviors at boundary conditions. The success of Japanese fuzzy applications was often in narrow domains; broader applicability was oversold.