摘要
在金融企业中,时间序列是一种重要的数据类型。高效、准确地预测金融时间序列对于企业的运作具有重要意义。提出使用一种具有增量学习能力的模糊神经网络(FNN-IL)应用于金融时间序列的预测。FNN-IL能学习蕴涵在时间序列中的知识,并能跟踪时间序列的运行从而动态调整模糊规则库。对比试验表明FNN-IL的性能优于传统的FNN。
Financial time series is an important data type in financial enterprises. Efficiently and accurately predict financial time series will greatly benefit the operations of financial enterprises. Fuzzy neural network with incremental learning ability （FNN-IL） was proposed to predict financial time series. FNN-IL can automatically learn the knowledge contained in financial time series and track the running procedure of financial time series thus dynamically adjust the rule base. Comparative experimental results demonstrate that the prediction accuracy of FNN-IL is higher than that of the traditional FNN and smaller rule base can be obtained.
出处
《系统仿真学报》
EI
CAS
CSCD
北大核心
2007年第17期4004-4006,共3页
Journal of System Simulation
基金
National Nature Science Foundation of China(60572034)
关键词
时间序列
模糊神经网络
增量学习
预测
time series
Fuzzy Neural Network
incremental learning
prediction
