论文部分内容阅读
提出了一种新的人工神经元网络结构模型 ,与传统的层次型结构不同的是 ,它具有一种逐步细分型、自相似的结构 .给出了用二层模型模拟任意连续函数的一个可靠方法 ,构造方法是基于函数的局部性质 .
A new artificial neural network structure model is proposed, which is different from the traditional hierarchical structure in that it has a subdivision-type and self-similar structure. A two-layer model is given to simulate an arbitrary continuous function Reliable methods, construction methods are based on the local nature of the function.