METHODS AND EXPERIMENTS REGARDING LEARNING FUZZY COGNITIVE MAPS (FCM) WEIGHTS USED IN FUPOL

In order to develop a suitable method for computing Fuzzy Cognitive Maps (FCM) weights in FUPOL project, we compare two methods for computing FCM weights using as historical data the data obtained by simulating a simple FCM model. First method Least Square Error (LSE) is based on optimization formalism, the second method Adaptive Neuro Fuzzy System (ANFIS) combines Fuzzy Inference Systems with the optimization approach. Based on this comparison, we can decide if ANFIS method can be used alone or it is necessary to combine it with other methods.