| 一种基于遗传算法的BP网络改进方法 |
| |
| 引用本文： | 蒋蓉蓉.一种基于遗传算法的BP网络改进方法[J].微计算机信息,2007,23(31):234-236. |
| |
| 作者姓名： | 蒋蓉蓉 |
| |
| 作者单位： | 北京科技大学信息工程学院计算机系,北京,100083 |
| |
| 基金项目： | 国家自然科学基金;北京科技大学校科研和教改项目 |
| |
| 摘 要： | 为克服和改进传统的BP算法的不足，发挥神经网络和遗传算法各自的优势，提出一种新的基于遗传算法的改进的BP网络训练方法。在美国手写体数字标准数据集MNIST库的实验结果表明，该方法提高了识别率，增加了网络的泛化能力，并且极大地节省了存储空间，缩短了学习时间。
|
| 关 键 词： | BP网络 遗传算法 手写体数字 识别 |
| 文章编号： | 1008-0570（2007）11-1-0234-03 |
| 修稿时间： | 2007-08-13 |
A New BP Network Improved Method Based on Genetic Algorithm |
| |
| JIANG RONGRONG.A New BP Network Improved Method Based on Genetic Algorithm[J].Control & Automation,2007,23(31):234-236. |
| |
| Authors: | JIANG RONGRONG |
| |
| Affiliation: | School of Information Engineering, University of Science and Technology Beijing, Beijing, 100083 |
| |
| Abstract: | To overcome and improve on the lack of the traditional BP algorithm, and bring into play the advantage of artificial neural network and genetic algorithm, the paper proposes a kind of training method based on genetic algorithm. Experimental results on MNIST database show that the presented method obtains higher recognition accuracy, improves the generalization ability, and reduces the training time and storage space. |
| |
| Keywords: | BP network genetic algorithm handwritten numeral recognition |
| 本文献已被 维普 万方数据 等数据库收录！ |
