Can Linguistic Modeling Be As Accurate As Fuzzy Modeling Without Losing Its Description To A High Degree?

In system modeling with Fuzzy Rule-Based Systems (FRBSs), we may usually find two contradictory requirements, the interpretability and the accuracy of the model obtained. As known, Linguistic Modeling (LM)—where the main requirement is the interpretability—is developed by linguistic FRBSs, while Fuzzy Modeling (FM)—where the main requirement is the accuracy—is developed, among others, by approximate FRBSs. Whilst the fact of making linguistic FRBSs be highly interpretable involves establishing hard restrictions to the rule structure (due to the use of a global semantic) thus losing flexibility, relaxing such restrictions, as approximate FRBSs do (using a local semantic), can make more flexible models to be obtained but losing their interpretability. The main objective of this contribution it to carry out a comparative analysis between LM and FM beyond the classical approach resigned to simply consider LM with a good interpretability but a bad accuracy. Some possibilities to significantly improve the precision of the linguistic models keeping good legibility will be introduced under the assumption that better accuracy could be obtained looking for a good balance between model flexibility and modeling simplicity. The good performance of such improvements opposite to FM will be shown by means of a wide experimental study applying fourteen different learning methods, carefully selected, to four modeling problems with different nature.