几种局部连接神经网络结构及性能的分析与比较

Analysis and Comparison of the Structure and Performance of Several Local Neural Networks

对人工神经网络中的局部连接网络进行了综合研究，比较和分析了CMAC、B样条和RBF这3种用于函数逼近的局部网络的性能。还分析了以ART-2为代表的局部网络中用于模式识别的网络结构及功能。最后在对以上几种网络进行模糊化的基础上，讨论了相应模糊神经网络的性能，为选择和应用局部神经网络提供了一定的依据。

The paper synthesizes the local neural networks, analyzes and compares the structures and functions of three local networks CMAC, B-spline and RBF, which are often used to approach functions. Then it discusses the structure of ART-2, depicts their fuzzy structures and performances based on the fuzzy model of these local networks. The study and analysis in the paper are useful for selecting and designing local neural networks.