基于径向基小波核的多尺度小波支持向量机  

Multi-scale wavelet SVM based on radial wavelet kernel

作　　者：任世锦[1] 吴铁军[1] 

机构地区：[1]浙江大学智能系统与决策研究所工业控制技术国家重点实验室

出　　处：《电路与系统学报》2008年第4期70-76,80,共8页Journal of Circuits and Systems

基　　金：国家863计划资助项目(2002AA412010)

摘　　要：普通支持向量机(SVM)方法用于多尺度回归建模时不能取得满意的精度,而现有的多尺度SVM算法存在只适合均匀分布的样本并可能收敛于局部极值等问题。为解决上述问题,本文提出了一种基于径向基小波核的多尺度小波支持向量机学习算法。文中提出并证明了一种新的径向基小波支持向量核,可提高小波SVM的训练速度和逼近精度。在此基础上,通过解一个二次优化问题可求出多尺度回归建模问题的全局最优解。最终得出的多尺度回归模型能够有效地逼近多尺度信号。仿真结果验证了所提算法的有效性。Original support vector machine （SVM） algorithm can not reach desired precision in multi-scale regression modeling. The existing multi-scale SVM algorithms, on the other hand, are only suitable for uniform sampling and may stick to local minima. To cope with those problems, a multi-scale wavelet SVM algorithm is proposed in this paper by using a new radial base wavelet kernel. This kernel is proposed in this paper and is proved to be able to improve training speed and precision of the wavelet SVM. On this basis, the global optimum of the multi-scale regression modeling problem can be obtained by solving a quadratic programming problem. As a result, the regression model can effectively approximate multi-scale signals. The effectiveness of the proposed algorithm is validated by computer simulation results.

关 键 词：小波 支持向量核 支持向量机 多尺度学习 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]