On the Usefulness of Fuzzy Rule Based Systems Based on Hierarchical Linguistic Fuzzy Partitions

In the recent years, a high number of fuzzy rule learning algorithms have been developed with the aim of building the Knowledge Base of Linguistic Fuzzy Rule Based Systems. In this context, it emerges the necessity of managing a flexible structure of the Knowledge Base with the aim of modeling the problems with a higher precision. In this work, we present a short overview on the Hierarchical Fuzzy Rule Based Systems, which consists in a hierarchical extension of the Knowledge Base, preserving its structure and descriptive power and reinforcing the modeling of those problem subspaces with more difficulties by means of a hierarchical treatment (higher granularity) of the rules generated in these areas. Finally, this methodology includes a summarisation step by means of a genetic rule selection process in order to obtain a compact and accurate model. We will show the goodness of this methodology by means of a case of study in the framework of imbalanced data-sets in which we compare this learning scheme with some basic Fuzzy Rule Based Classification Systems and with the well-known C4.5 decision tree, using the proper statistical analysis as suggested in the specialised literature. Finally, we will develop a discussion on the usefulness of this methodology, analysing its advantages and proposing some new trends for future work on the topic in order to extract the highest potential of this technique for Fuzzy Rule Based Systems.