A generic framework for enhancing the interpretability Of granular computing-based information

One of the main advantages of Granular Computing and Fuzzy Logic is the transparency and interpretability features that are available to the user. In this paper we present a systematic data granulation algorithm for the elicitation of Fuzzy rules and show how the granular data and relational information extracted during the data mining process can be translated into Fuzzy Logic statements with enhanced interpretability. Notions of granular cardinality, distribution and distance are used to apply linguistic hedges to two-sided Gaussian Fuzzy membership functions. The proposed methodology is applied to a biomedical dataset relating to Electrical Impedance Tomography (EIT) measurements of lung ventilation showing good agreement and interpretability between the captured knowledge and the theoretical and physiological expectations.