| Uncertain Relation Suited to Overfitting of BP Neural Network |
| |
| 作者姓名： | 任继平 李祚泳 江春华 |
| |
| 作者单位： | School of Computer science and engineering，UESTC Chengdu 610054 China，Chengdu University of Information Technology Chengdu 610041 China，School of Computer science and engineering，UESTC Chengdu 610054 China |
| |
| 基金项目： | Supported by the the Nation Natural Science Foundation of China (No.40271024) |
| |
| 摘 要： | Researches have been done on the overfitting of BP neural network1~7]. The uncertain relation between learning and generation suited to overfitting of BP neural network modeling is given in Ref.8]. By numerical test, the range of overfitting parameter in uncertain relation is determined as 10?6~10?3. Further, it is pointed out that parameter selection relates to the complexity of certain problems (approximate functions), which, can be described by multicorrelative coefficients. But for mul…
|
Uncertain Relation Suited to Overfitting of BP Neural Network |
| |
| Ren Jiping,Li Zuoyong,Jiang chunhua,YANG Haomiao.Uncertain Relation Suited to Overfitting of BP Neural Network[J].Journal of Electronic Science Technology of China,2004,2(1). |
| |
| Authors: | Ren Jiping Li Zuoyong Jiang chunhua YANG Haomiao |
| |
| Affiliation: | 1. School of Cornputer science and engineering,UESTC Chengdu 610054 China
2. Chengdu University of Information Technology Chengdu 610041 China |
| |
| Abstract: | A general uncertainty relation between the change of weighted value which represents learning ability and the discrimination error of unlearning sample sets which represents generalization ability is revealed in the modeling of back propagation (BP) neural network.Tests of numerical simulation for multitype of complicated functions are carried out to determine the value distribution (l×10-5～5×10-4) of overfitting parameter in the uncertainty relation.Based on the uncertainty relation,the overfitting in the training process of given sample sets using BP neural network can be judged. |
| |
| Keywords: | back progagation neural network overfitting learning ability generalization ability uncertainty relation |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
