An Experimental Study on the Interpretability of Fuzzy Systems

Interpretability is one of the most significant properties of Fuzzy Systems which are widely acknowledged as gray boxes against other Soft Computing techniques such as Neural Networks usually regarded as black boxes. It is essential for applications with high human interaction (decision support systems in medicine, economics, etc). The use of accuracy indices to guide the fuzzy modeling process is broadly extended. In turn, although there have been a few attempts to define Interpretability indices, we are still far away from having a universal index. With the aim of evaluating the most used indices an experimental analysis (in the form of a web poll) was carried out yielding some useful clues to keep in mind regarding Interpretability assessment. Results extracted from the poll show the inherent subjectivity of the measure because we collected a huge diversity of answers. Nevertheless, comparing carefully all the answers, it was possible to find out some interesting user profiles. Keywords— Fuzzy modeling, Interpretability assessment.