| 一种改进的BP算法 |
| |
| 引用本文： | 万里青.一种改进的BP算法[J].火力与指挥控制,1995(2). |
| |
| 作者姓名： | 万里青 |
| |
| 作者单位： | 洛阳航空工业总公司电光设备研究所 |
| |
| 摘 要： | ＢＰ算法是目前应用极为广泛的神经网络算法，但它也存在一些不足。文中提出了采用共轭梯度法及黄金分割相结合的改进ＢＰ算法（ＭＢＰ），自适应调整学习率，提高了运算速度，减少了迭代次数。最后将标准ＢＰ算法和ＭＢＰ算法进行了比较，仿其结果表明：ＭＢＰ算法的学习次数和收敛速度得到极大改善。
|
| 关 键 词： | 神经网络，BP算法，自适应学习率 |
A Modified BP Algorithm |
| |
| Wan Liqing.A Modified BP Algorithm[J].Fire Control & Command Control,1995(2). |
| |
| Authors: | Wan Liqing |
| |
| Abstract: | Back-propagation algorithm is one of the most popular aigorithm inneural network. In this paper, an improved BP algorithm with conjugate gradient methodand adaptive learning rate is present.A much fast convergence is obtained and the learningtime is reduced. Finally, the MBP algorithm is compared with the standard BP algorithm. The results shown that the learning speed of MBP algorithm is increased greatly. |
| |
| Keywords: | neural network BP algorithm adaptive learning rate |
| 本文献已被 CNKI 等数据库收录！ |
