摘要
标准的支持向量回归机对于参数的选取有很强的依赖性.当选取的参数不恰当,或当数据受到噪声的污染时,回归的效果将受到较大的影响.笔者将训练点被正确划分的程度引入到支持向量回归机模型中,通过理论推导,提出了一种新的支持向量回归机TSVR,并给出了TSVR算法收敛的相关证明.同时,通过大量的数值实验,证明了TSVR具有较好的回归效果,其回归结果对参数的选取较不敏感,具有比标准的支持向量回归机更好的性质.
The standard support vector machine largely relies on the parameter selection.The inappropriate parameters will compromise the results.Meanwhile,the results will be affected if the data are contaminated by noises.Hence,it is necessary to improve the standard support vector machine.In this paper,the distances being classified correctly is introduced into the model of the support vector machine.A new type of support vector machine for regression TSVR is proposed and some relevant theorems are brought forth.The results of numeric experiments show that the TSVR method performs better than standard support vector machine and is not sensitive to the selections of regression parameters.
出处
《宁波大学学报(理工版)》
CAS
2010年第3期52-55,共4页
Journal of Ningbo University:Natural Science and Engineering Edition