Extracting compact fuzzy rules based on adaptive data approximation using B-splines

We first discuss the importance of making a controller interpretable and give an overview of the existing models and structures for that purpose. We then summarise our approach to designing fuzzy controllers based on the B-spline model by learning. Too large number of rules will not only result in the over-fitting problem, but also the lost of interpretability of the model. By using an optimal partition algorithm and using linguistic modificators like "between", "at most", "at least" etc., the rule base can be reduced to the minimum. We tested this approach in different benchmark problems and achieved a rule compression ratio till 71%. In this way, the readability of a rule base is significantly improved.