Incorporating Method of Time Correlation and SVM Based on Time Geodesic Distance

摘要：

Support vector machine (SVM) has been used in many fields as a new learning method developed in recent years.When dealing with time series forecasting problem one encounters time correlation prior knowledge of time series data. If prior knowledge at hand can be incorporated into Support Vector learning machines, the generalization performance of SVM may be improved efficiently.In order to incorporate time correlation into SVM, this paper presents time geodesic distance for structural feature of learning data, and this presented new metric can be made use of by classification methods based on distance in training of learning machine. Comparing with the traditional SVM based on air quality database, the presented approach can greatly improve the generalization performance of SVM.

关键词： Support Vector Machine Time Series Forecasting Time Geodesic Distance

作者: Ping WANG Guisheng ZHANG

作者单位: Shanxi University,Taiyuan,030006,P. R. China Shanxi Police Academy,Taiyuan,030021,P. R. China Shanxi University,Taiyuan,030006,P. R. China

会议类型: 国际会议

会议名称: International Conference on Advances in Engineering 2011(2011年工程研究进展国际学术会议 ICAE2011)

会议地点: 南京

会议语种:英文

页码: 823-827

在线出版日期: 2011-12-17（万方平台首次上网日期，不代表论文的发表时间）