Fuzzy cognitive maps based models for pattern classification: Advances and challenges

Abstract

Fuzzy Cognitive Maps (FCMs) have proven to be a suitable methodology for the design of knowledge-based systems. By combining both uncertainty depiction and cognitive mapping, this technique represents the knowledge of systems that are characterized by ambiguity and complexity. In short, FCMs can be defined as recurrent neural networks that include elements of fuzzy logic during the knowledge engineering phase. While the literature contains many studies claiming how this Soft Computing technique is able to model complex and dynamical systems, we explore another promising research field: the use of FCMs in solving pattern classification problems. This is motivated by the transparency of the decision model attached to these cognitive, neural networks. In this chapter, we revise some prominent advances in the area of FCM-based classifiers and open challenges to be confronted.