| 基于神经网络控制的共轭梯度法 |
| |
| 引用本文： | 沈海荣,刘超,宫宁生. 基于神经网络控制的共轭梯度法[J]. 陶瓷科学与艺术, 2006, 0(6) |
| |
| 作者姓名： | 沈海荣 刘超 宫宁生 |
| |
| 作者单位： | 南京工业大学信息科学与工程学院 江苏南京210009 |
| |
| 摘 要： | 共轭梯度法中搜索步长是通过某种搜索策略得到,许多情况下的收敛速度较慢。为了加快其收敛速度,提出了通过引入具有“先验知识”的神经网络对共轭梯度算法中的搜索步长进行控制。实验结果表明,该模型实现的共轭梯度法对于加快收敛速度有效。
|
| 关 键 词： | 神经网络 共轭梯度法 BP算法 |
Conjugate grade method based on neural network control |
| |
| SHEN Hai-rong,LIU Chao,GONG Ning-sheng. Conjugate grade method based on neural network control[J]. Ceramics Science & Art, 2006, 0(6) |
| |
| Authors: | SHEN Hai-rong LIU Chao GONG Ning-sheng |
| |
| Abstract: | In conjugate grade method,the step searching is obtained by certain strategy,in which the convergent speed is relatively slow in many situations.In order to improve the convergent speed,neural network with "transcendental knowledge"was introduced to control the search step span,then to improve convergent speed of conjugate grade method.The experiment result shows that the conjugate grade method with this model can effectively improve the convergent speed. |
| |
| Keywords: | neural networks conjugate grads method BP arithmetic |
| 本文献已被 CNKI 等数据库收录！ |
|
