【作者】 刘力丰； 米红； 耿代；

【Author】 LIU Li-feng, MI Hong, GENG Dai (Xiamen University, Laboratory of Pattern Recognition and Intelligence System, Xiamen 365001, China)

【机构】 厦门大学模式识别与智能系统研究所；

【摘要】 讨论了在文献[4]中提出的非线性协整存在性理论,指出非线性协整关系反映了向量时间序列分量之间的非线性均衡关系;同时阐述了回归型支持向量机(SVR)的基本结构及训练方法;并在此基础上利用回归支持向量机的非线性回归能力,分析了估计非线性协整函数的统计学习方法,最后通过对预设的一组向量时间序列进行仿真研究, 并估计其间的非线性协整函数,来说明SVR方法在非线性协整建模应用中的有效性．更多还原

【Abstract】 This paper discussed the concept of nonlinear co-integration relationship which is presented in referenced paper[4], and indicate that nonlinear co-integration could reflect some nonlinear equilibriums among variables in vector time series. At the same time we expound the essential framework of support vector machine for regression (SVR) and its training algorithm. And then we analyze the statistic learning algorithm for estimating nonlinear co-integration function based on SVR regression. At last we present a emulation to a specific vector time series, in order to illuminate validity of SVR used in application of nonlinear co-integration.更多还原