The second hidden layer provides the aggregation of membership levels calculated in the first hidden layer and consists of h multiplication blocks. The third hidden layer contains adjustable synaptic weights w w w h12 , ,..., determined during supervised learning. The fourth hidden layer consists of two adders and computes the sums of output signals of the third and second hidden layers. And, at last, the fifth (output) hidden layer performs defuzzification and, as a result, the NFS output signal () yk is computed