| 一种确定神经网络最优结构的新算法 |
| |
| 引用本文： | 吕晓丽,张瑞,王博,崔新月. 一种确定神经网络最优结构的新算法[J]. 西安文理学院学报, 2014, 0(4): 4-8 |
| |
| 作者姓名： | 吕晓丽 张瑞 王博 崔新月 |
| |
| 作者单位： | 西北大学数学学院; |
| |
| 基金项目： | 陕西省自然科学基金(2014JM1016) |
| |
| 摘 要： | 从神经网络结构设计问题出发,提出一种确定神经网络最优隐节点个数的新方法.该算法首先按照等差数列增加隐节点,确定最优隐节点个数的范围;然后利用折半删减法确定最优隐节点个数.数值实验表明该算法在保持良好泛化能力的同时能自适应地、快速有效地确定网络最小隐神经元数目.
|
| 关 键 词： | ELM算法 神经网络 等差增长 折半删减 |
A New Algorithm of Determining Optimal Structure of Neural Network |
| |
| LV Xiao-li,ZHANG Rui,WANG Bo,CUI Xin-yue. A New Algorithm of Determining Optimal Structure of Neural Network[J]. Journal of Xi‘an University of Arts & Science:Natural Science Edition Edi Edition, 2014, 0(4): 4-8 |
| |
| Authors: | LV Xiao-li ZHANG Rui WANG Bo CUI Xin-yue |
| |
| Affiliation: | (School of Mathematics, Northwest University, Xi＇an 710127, China) |
| |
| Abstract: | This study addresses the structure of neural network and proposes a new method for determining the number of the optimal hidden nodes. Based on the algorithm, the hidden nodes are increased according to the arithmetic progression and the scope of the optimal hidden nodes is determined. The method of 50% deletion is then applied in determining the number of opti- mal hidden nodes. Simulation result proves that the proposed algorithm is adaptive, effective, and rapid in determining the minimal number of hidden neurons of the network while maintai- ning good generalization capability. |
| |
| Keywords: | ELM algorithm neural network arithmetic growth 50% deletion |
| 本文献已被 CNKI 维普 等数据库收录！ |
|
