一种模糊小脑模型神经网络

A Kind of Fuzzy CMAC Neural Network

为了简化网络、提高网络的学习能力、便于对一些系统的建模,本文提出了一种小脑模型神经网络,将模糊逻辑的推理过程用小脑模型神经网络表示出来,其输入层采用模糊化的感受野,能有效地减少输入层的容量,提高逼近能力。由于采用系统的模糊信息,可以按实际问题的性质初始化网络的结构与参数,有利于提高学习的收敛速度。

In this paper a kind of fuzzy CMAC neural network is presented to simplify network,enhance the learning ability,and to model some systems easily. It expresses the inference of fuzzy logic with CMAC neural network. Its input layer adopts the fuzzy receptive field,which can decrease the capacitance of input layer effectively and enhance the ability of approximation. Because of application of the fuzzy information of the system,the networks structure and parameters can be initialized by the nature of the real world problem,thus improving its learning convergence speed and making the learning result describe the problem nature more realistically.