Adaptation of Fuzzy Cognitive Maps - a Comparison Study

Abstract

This paper deals with the experimental study and comparison of various adaptation methods for setting-up parameters of fuzzy cognitive maps (FCMs). A survey is given of the best known methods, which are mostly based on unsupervised learning. The authors show better performance using supervised learning, namely least mean square approaches. Experiments were done on a simulation example of autonomous vehicle navigation. The paper is concluded by comparing their efficacy and properties.