| 鲁棒区间回归分析的神经网络学习算法 |
| |
| 引用本文： | 黄磊,张百灵,黄茜,徐秉铮.鲁棒区间回归分析的神经网络学习算法[J].华南理工大学学报(自然科学版),1996(4). |
| |
| 作者姓名： | 黄磊 张百灵 黄茜 徐秉铮 |
| |
| 作者单位： | 华南理工大学无线电与自动控制研究所 |
| |
| 基金项目： | 国家八六三高技术基金,汕头大学人工智能与模式识别研究所基金 |
| |
| 摘 要： | 在许多工程实践问题中，区间回归分析是处理区间数据的一个重要手段。文献中用神经网络实现的区间回归分析都假定给定的训练数据是无噪的。当训练数据被污染时，这些方法的性能将急剧下降。本文针对一种区间回归的神经网络实现方法，提出了两种新的神经网络学习算法，能够有效地从有噪训练数据里建立一个鲁棒式的非线性区间回归模型。所提出的学习算法充分利用了对训练数据质量的估计知识。仿真结果表明了这种方法的有效性。
|
| 关 键 词： | 区间回归 鲁棒性 前向神经网络 BP算法 模糊回归 |
LEARNING ALGORITHMS OF NEURAL NETWORKS FOR ROBUST INTERVAL REGRESSION ANALYSIS |
| |
| Huang Lei,Zhang Bailing,Huang Qian,Xu Bingzheng.LEARNING ALGORITHMS OF NEURAL NETWORKS FOR ROBUST INTERVAL REGRESSION ANALYSIS[J].Journal of South China University of Technology(Natural Science Edition),1996(4). |
| |
| Authors: | Huang Lei Zhang Bailing Huang Qian Xu Bingzheng |
| |
| Abstract: | A method by means of neural networks for interval regression analysis hasbeen proposed on the assumption that the given training data are totally "good data".However, when noisy training data are present, its performance will decline sharply. Inthis paper, we propose two new learning algorithms, which can determine a robustnonlinear interval regression model from the noisy training data. The proposed algorithms make full use of the estimated knowledge about the quality of the training data.Simulation results show that the proposed algorithms are very effective. |
| |
| Keywords: | s:interval regression robust feedforward neural networks BP algorithm fuzzy regression |
| 本文献已被 CNKI 维普 等数据库收录！ |
