Stability of Fixed-Point Values in Reduced Fuzzy Cognitive Map Models

Abstract

The authors have already presented their method for reducing oversized FCM models, and also have analyzed the prediction error of the reduced models. These investigations assumed that models have a single fixed-point attractor. The novelty of this paper is that it deals with the stability behavior of the fixed-point attractor value of original-reduced model pairs and compares the number of fixed-point attractors found, the asymptotic values of the concepts, and also checks if any limit cycles or chaotic behavior occur. The method of comparison and also the first results made with two real-life and one synthetic model are presented and some conclusions are taken.