Universal Approximation Theorem for Bipolar Fuzzy Madaline Neural Network

Multiple Adaptive Linear Neuron network is a three layer neural network. Weights of this network are considered to be bipolar fuzzy sets. In Bipolar fuzzy MADALINE neural network, binary and bipolar sigmoid activation functions are used to determine the output value. The output function of bipolar fuzzy MADALINE neural network acts as a function approximator. This paper deals with Universal approximation theorem for bipolar fuzzy MADALINE neural network which is proved using Stone-Weierstrass theorem. Further, Convergence theorem for selecting different learning rates for obtaining the accurate output is developed. The Identification of bipolar fuzzy MADALINE neural network with three layers is determined.