摘要
以线性无关的基函数作为隐层神经元的激励函数,构建了一类基函数神经网络,且推导出该类神经网络的学习算法;在此基础上,设计了一种基于指数增长和折半删减的快速最小隐神经元数目确定算法.仿真实验表明,此算法能自适应地、快速有效地确定网络最小隐层神经元数目.
In this paper, a basis function neural network is constructed, of which the hidden-layor neurons are activated with linear independence basis function. Accordingly, the learning algorithm for the constructed neural network is derived and a fast algorithm based on exponential-growth and binary-delete search strategy is proposed to determinate the optimal number of hidden-layor neurons. The simulation results substantiate that our algorithm can adaptively, quickly and efficiently determine number of hidden neurons in the neural network.
出处
《微电子学与计算机》
CSCD
北大核心
2010年第1期57-60,共4页
Microelectronics & Computer
基金
国家自然科学基金项目(60643004
60775050)
浙江大学CAD/CG国家重点实验室开放课题(A0908)
关键词
广义多项式
神经网络
结构自适应确定
指数增长
折半删减
general polynomial
neural network
structure-adaptive-determination
exponential growth
binary ,search
