基于代数神经网络的分圆多项式判定及学习算法  

Cyclotomic Base on Algebra Neural Networks of Tests Polynomials and Learning Algorithms

作　　者：周永权[1] 马武瑜[1] 

机构地区：[1]广西民族学院数学与计算机科学系,南宁530006

出　　处：《系统工程与电子技术》2000年第5期79-82,共4页Systems Engineering and Electronics

摘　　要：首先对MP神经元进行推广 ,给出了一元多项式代数神经元、一元多项式代数神经网络 ,设计出一类用于一元多项式二元运算的神经元模型 ,它是单输入单输出的 3层多项式神经网络 ,给出了两种用于判定一多项式为分圆多项式的神经网络模型及学习算法 ,把判定一多项式为分圆多项式转化成线性方程组求解问题 ,通过算例说明了该方法的有效性。Firstly the MP neuron is popularized, and the concepts of polynomials algebra neurons and polynomials algebra neural networks are proposed. A kind of polynomias neurons model of applying to polynomials operations is designed, that is the single input and single output three layers polynomials neural networks. Two kinds of testing cyclotomic polynomials neural networks model and learning algoirthms are designed, and the tests for cyclotomic polynomials are changed into solving the line equation group problem. The example and the results show that the algorithm is efficiency.

关 键 词：代数神经网络 分圆多项式判定 学习算法 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] O174.14[自动化与计算机技术—控制科学与工程]