摘要
提出一种自适应权值的支持向量机集成学习方法。该方法以Bagging方法为基础,结合部分AdaBoost算法权值更新的思想,给各个子分类器赋予权值,同时结合支持向量机本身的特性,对训练数据进行选择,加大训练样本的差异性。相比较传统的Bagging方法,结合SVM的特性来更有针对性的训练错分样本。文中使用4个UCI数据集进行对比实验,结果表明本文算法相比较传统的Bagging算法可以在一定程度上提高分类器的泛化能力。
This paper presents the SVMensemble learning method based on adaptive weight.The method isbased on the Bagging method,and combined with the idea that AdaBoost algorithm updates the weights of sample.In order to increase the diversity of training samples,the algorithm gives each sub -classifier weight,and combinesthe characteristics of SVMto select the training data.Compared with the traditional Bagging method,the algorithmcan pay more attention to the wrong samples,by combining the characteristics of the SVM.Four UCI data sets areused to do the experiment by us.Compared with the Bagging algorithm,the result shows that the algorithm canimprove the generalization ability of the classifier to some extent.
出处
《山东师范大学学报(自然科学版)》
CAS
2015年第1期20-23,共4页
Journal of Shandong Normal University(Natural Science)
基金
教育部博士点基金资助项目(20113704110001);山东省自然科学基金资助项目(ZR2010FM021);山东省科技攻关计划及泰山学者基金资助项目(2013GGX10125).