模糊反向传播神经网络FBP的函数逼近能力研究

Research on Function Approximation Capabilities of Fuzzy Backpropagation Neural Networks

研究了模糊反向传播神经网络FBP(FuzzyBackpropagation)的函数逼近能力。给出了单调连续函数的FBP按序单调特性、连续映射定理以及非函数一致逼近定理 ,并且说明了FBP虽然能保持连续性映射 ,但不如原神经网络具有函数逼近能力。

Studies the function approximation capabilities of FBP. The results here show the ordered monotony property, continuous mapping theorem and its non approximation theorem to function. It is proved that although FBP can keep continuous mapping property, their capabilities to approximate function is worse than that of traditional neural networks.