摘要： 将泛函神经元结构变形，建立Sigma-Pi泛函网络模型，给出Sigma-Pi泛函网络学习算法。采用数值分析的方法，将Sigma-Pi泛函网络应用于异或问题，结果表明，该网络对于某些问题具有很强的分类能力。该方法的优点在于利用一元函数作为基函数来实现高维函数的逼近，在函数逼近技术上，有着重要的应用价值。

关键词: 泛函神经元, 泛函网络, Sigma-Pi泛函网络, 基函数簇, 异或问题

Abstract: This paper converts a structure of a functional neuron, presents a Sigma-Pi functional network(SPFN) structure, and proposes the Sigma-Pi functional networks learning algorithm. Using numerical analysis method, the Sigma-Pi functional network is applied to XOR problem. The results demonstrate that the functional network has powerful classification capability. The method has the advantages of using a variable function for multi-dimensional function approximation and important practical significance.

Key words: Functional neuron, Functional networks, Sigma-Pi functional networks, Base functions, XOR problem