论文部分内容阅读
泛化是前馈神经网络最重要的问题之一。本文从提高泛化能力方面对神经网络的拓扑结构问题进行了分析和研究，给出了基于提高泛化能力的网络拓扑结构构造算法。
Generalization is one of the most important problems in feedforward neural networks. In this paper, the topological structure of neural networks is analyzed and studied in terms of improving the generalization ability, and the algorithm of network topology construction based on generalization ability is given.