摘要
粗集和神经网络相结合反映的是人类智能的定性和定量、清晰和隐含、串行和并行相互交叉混合的常规思维机理。本文建立这样一种混合模型用于指数趋势的预测 ,通过粗集对数据的二维约简预处理消除了样本中的噪声和冗余 ,提高神经网络预测精度的同时还降低了学习负担 ;利用遗传算法进行属性离散化和网络权重的优化获得了较优的预测精度。对上证综指的实证分析表明 ,这种混合模型的性能明显优于 BP和
Rough set integrated neural network method reflects the human's normal thinking mechanism which mixes the method of quantitative and qualitative, clear and uncertain, serial and parallel. This paper builds such a model which using rough set's 2 dimension reductive ability to reduce the noise and redundancy in the samples. So it improves the neural network's forecasting accuracy as well as reducing its' burden of learning. GA also is used in this paper to the attribute's discretion and neural network learning to find the optimized forecasting accuracy. Case study shows the hybrid model is more competitive than the similar neural network model.
出处
《系统工程理论方法应用》
2002年第2期157-162,共6页
Systems Engineering Theory·Methodology·Applications
基金
国家自然科学基金资助项目 (697740 3 6)
