摘要
利用GMDH(GroupMethodofDataHanding)神经网络进行数据预测。与一般的前馈神经网络不同 ,GMDH网络的结构确定于训练过程之中 ,因而可大大提高神经网络性能 ,从而特别适用于数据预测。文中提供的两个实例表明 ,利用GMDH网络进行数据预测的结果与实际值符合得很好 ,特别在金融预测方面有很好的应用前景。
Prediction using GMDH(Group Method of Data Handling) neural network is studied. Unlike the normal feedforward neural network, the structure of GMDH network is configured through training. To do this improves the performance of GMDH networks largely so that it is particularly suitable for prediction. Two examples of prediction using GMDH networks are given and the results show that the predicted data agree with the real data very well so that the GMDH could be used in prediction, particularly in financial prediction.
出处
《华东船舶工业学院学报》
2000年第4期72-76,共5页
Journal of East China Shipbuilding Institute(Natural Science Edition)
关键词
神经网络
数据预测
GMDH网络
neural networks data
prediction
group method of data handling networks
