|
|||||
|
|
| 基于决策树构造神经网络的方法及应用 | |
| 引用本文： | 李宁,谢振华,商琳,陈世福. 基于决策树构造神经网络的方法及应用[J]. 复旦学报(自然科学版), 2004, 43(5): 696-699 |
| 作者姓名： | 李宁 谢振华 商琳 陈世福 |
| 作者单位： | 南京大学,计算机软件新技术国家重点实验室,南京,210093;南京大学,计算机软件新技术国家重点实验室,南京,210093;南京大学,计算机软件新技术国家重点实验室,南京,210093;南京大学,计算机软件新技术国家重点实验室,南京,210093 |
| 基金项目： | 江苏省自然科学基金资助项目(BK2003067) |
| 摘 要： | 基于知识的神经网络KBNN(Knowledge-Based Neural Network)在提高神经计算性能方面效果显著，其构造方法基于已得的规则.利用决策树对数值数据的分割能力和神经网络准确的逼近收敛能力，提出了一种利用C4.5决策树从数据中自动提取规则生成神经网络的构造方法.该方法易于构造、收敛速度快、精度较高，在高速公路路面破损智能识别系统中使用效果良好. |
| 关 键 词： | 神经网络 基于知识的神经网络 决策树 |
| 文章编号： | 0427-7104(2004)05-0696-04 |
Design Neural Network Based on Decision Tree Rules and Its Application Research |
|
| LI Ning,XIE Zhen-hua,SHANG Lin,CHEN Shi-fu ty,Nanjing,China). Design Neural Network Based on Decision Tree Rules and Its Application Research[J]. Journal of Fudan University(Natural Science), 2004, 43(5): 696-699 | |
| Authors: | LI Ning XIE Zhen-hua SHANG Lin CHEN Shi-fu ty Nanjing China) |
| Affiliation: | LI Ning,XIE Zhen-hua,SHANG Lin,CHEN Shi-fu ty,Nanjing210093,China) |
| Abstract: | Knowledge-Based Neural Network can obviously improve the efficiency of neural computing, which is constructed on acquired rules. An approach to construct a neural network based on rules generate by C 4.5 is projected, combining the capability of cutting numeric data of decision tree and the capability of accurate approximation of neural network, which is fast convergent and highly precise. The method is realized easily and shows better performance in the System of Recognition of Freeway's Surface Diseases. |
| Keywords: | neural network KBNN decision tree |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ | |