MLQP神经元网络的学习算法及其典型应用

Learning algorithm of multilayer quadratic perceptron and its typical apllications

论述了一种多层二阶神经元MLQP(MultilayerQuadraticPerceptron)网络模型的结构和学习算法。这种模型综合了一般多层神经元网络和高阶神经元网络的特点,其结构简单,可调整权数量适中,学习速度快。文中以典型的模式分类和函数逼近问题为例,比较了这种网络和传统的一阶网络以及两种其他类型的二阶网络的学习速度,验证了MLQP快速收敛性。

This paper presents the structure and learning algortithm of multilayer Quadratic perceptron (MLQP) that combines advantages of multi layer perceptrons and high order neural networks.The features of MLQP are in its simple structure,powerful mapping ability,practical number of adjustable connection weights and fast learning speed.Its learning speeds are compared with the multi layer percepton and other two kinds of the second order neural networks on pattern classification and function approximation problems,and its fast convergence property is proved.