Linguistic Hedges for Ant-generated Rules

FRANTIC, a system inspired by insect behaviour for inducing fuzzy IF-THEN rules, is enhanced to produce rules with linguistic hedges. FRANTIC is evaluated against an earlier version of itself and against several other fuzzy rule induction algorithms, and the results are highly encouraging. Rule comprehensibility is maintained while an improvement in the accuracy of the rulebases induced is observed. Equally important, the increase in computation expense due to the improved richness in the hypothesis language is acknowledged, and several ways of resolving this are discussed.