NEURO-FUZZY SYSTEMS: Learning Models

The goal of this research is the analysis of learning models by using of arithmetic operations applied in a neuro-fuzzy system (NFS). The research integrates the concepts between artificial neural network (ANN) and the fuzzy sets theory (FST). In order to assess the validity of the proposal, an FNS is proposed to diagnose paroxysmal events involving epileptic events (EE) and non-epileptic events(NEE). This article describes the learning models showing some results obtained through the use of min/max arithmetic operation by using the NEFCLASS model (neuro fuzzy classification) with the result of Einstein´s product and sum arithmetic operations through the use of a backpropagation neural network (BPNN). After the simulations had been performed, one has verified that through the use of different arithmetic operations in the fuzzy rules base, the end results may be different, resulting in a bigger or smaller rate of hits of the NFS