摘要
针对实际应用中经常出现的异类数据源,采用多核学习的支持向量机受到关注.然而随着核函数数量的增多,计算量也随之大大增加.为了解决这一问题,该提出了一种利用协同聚类对多核支持向量机的训练数据进行简化的方法,可以减少支持向量机的数目,从而减少计算量.实验结果显示,提出的方法可以提高多核支持向量机的效率,同时还不会影响分类精度.
Support vector machine based on multiple kernel learning is proposed due to the learning problems involve multiple and heterogeneous data sources, however, the increase of kernels will increase the computation of multiple kernel learning inevitably. To solve this problem, a new cluster method is presented, which is called cooperative clustering. Applying the cooperative clustering to multiple kernel SV M, the number of support vectors will be reduced, the time complexity of computation is also reduced. The experimental results reveal that the time consumptions of training and testing are decreased and the classification efficiency is maintained at the same level as the origin after applying our method.
出处
《北京交通大学学报》
EI
CAS
CSCD
北大核心
2008年第2期10-13,共4页
JOURNAL OF BEIJING JIAOTONG UNIVERSITY
基金
国家自然科学基金资助项目(60503017)
北京交通大学科技基金资助项目(2006XM007)
关键词
协同聚类
多核学习
核函数
支持向量机
cooperative clustering
multiple kernel learning
kernel function
support vector machine