Learning consistent, complete and compact sets of fuzzy rules in conjunctive normal form for regression problems

Abstract

When a flexible fuzzy rule structure such as those with antecedent in conjunctive normal form is used, the interpretability of the obtained fuzzy model is significantly improved. However, some important problems appear related to the interaction among this set of rules. Indeed, it is relatively easy to get inconsistencies, lack of completeness, redundancies, etc. Generally, these properties are ignored or mildly faced. This paper, however, focuses on the design of a multiobjective genetic algorithm that properly considers all these properties thus ensuring an effective search space exploration and generation of highly legible and accurate fuzzy models.

Memetic Type-2 Fuzzy System Learning for Load Forecasting

“…This random selection of examples maximises the diversity of the initial population whilst also ensures that each example fires at least one fuzzy rule. This mechanism ensures a proper covering over all the training sample [20] and also specifies a variable number of rules in the initial individuals. Hence, each fuzzy rule is initialised with an unordered set of fuzzy antecedents, one per input (i.e.…”

MOCA-I: Discovering Rules and Guiding Decision Maker in the Context of Partial Classification in Large and Imbalanced Datasets

“…This is only an overview of problems that may happen. Casillas et al identified more possible inconsistencies risen by Pittsburgh modeling [15]. We propose to use a Pittsburgh representation where each solution is a ruleset.…”

Reducing the Complexity of Genetic Fuzzy Classifiers in Highly-Dimensional Classification Problems

“…In the previous years, various genetic learning approaches have been considered for creating GFRBCSs 22,[27][28][29][30][31] , each exhibiting different benefits and drawbacks. Here we concentrate on the so-called iterative rule learning (IRL) approach [29][30][31] , which is the methodology followed by the proposal of this paper.…”