The synthesis of fuzzy systems involves the identification of a struct
ure and its specialization by means of parameter optimization. In doin
g this, symbolic approaches which encode the structure information in
the form of high-level rules allow further manipulation of the system
to minimize its complexity, and possibly its implementation cost, whil
e all-parametric methodologies often achieve better approximation perf
ormance. In this paper, we rely on the concept of a fuzzy set of rules
to tackle the rule induction problem at an intermediate level. An onl
ine adaptive algorithm is developed which almost surely learns the ext
ent to which inclusion of a rule in the rule set significantly contrib
utes to the reproduction of the target behavior, Then, the resulting f
uzzy set of rules can be defuzzified to give a conventional rule set w
ith similar behavior. Comparisons with high-level and low-level method
ologies show that this approach retains the most positive features of
both.