| 基于曲线曲面图形神经网络学习方法的研究 |
| |
| 引用本文： | 包健,周华英.基于曲线曲面图形神经网络学习方法的研究[J].实验室研究与探索,2005,24(4):19-21,27. |
| |
| 作者姓名： | 包健 周华英 |
| |
| 作者单位： | 杭州电子科技大学,计算机学院智能控制实验室,浙江,杭州,310018 |
| |
| 基金项目： | 浙江省自然科学基金(M603060) |
| |
| 摘 要： | 提出一种基于曲线曲面图形神经网络学习方法，即由点集构成的曲线曲面图形作为神经网络的学习样本，利用人工神经网络直接对曲线、曲面进行学习和训练，并以曲线、曲面图形的直观显示来表示网络的训练效果。同时还提出了基于曲线曲面图形神经网络结构的构建方法和训练过程的改进方案。实验结果表明，该构建方法和改进方案在提高训练精度、加快收敛速度方面收到了较好的效果。
|
| 关 键 词： | BP神经网络 网络结构 算法 |
| 文章编号： | 1006-7167(2005)04-0019-04 |
Study on Learning Method about Curve Graph and Curved Surface Graph of Neural Network |
| |
| BAO Jian,ZHOU Hua-ying.Study on Learning Method about Curve Graph and Curved Surface Graph of Neural Network[J].Laboratory Research and Exploration,2005,24(4):19-21,27. |
| |
| Authors: | BAO Jian ZHOU Hua-ying |
| |
| Abstract: | A new network study method based on curve graph and curved surface graph was brought forward. The training samples are the curve graph and the curved surface graph, and the neural network trains these graphs directly and shows the intuitive training results with the corresponding curve graph and curved surface graph. How to construct the network structure and how to improve the training process were also presented. The lab results indicate that these methods are simple and convenient, which converge fast and improve the precision. |
| |
| Keywords: | BP neural network network structure algorithm |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
