摘要
针对现有多分类支持向量机算法所存在的训练时间长、判别速度慢等问题,提出了一种二叉树多类支持向量机算法,该算法能够有效减少支持向量的个数,从而减少训练时间.为了验证算法的有效性,将该算法分别同l-v-r算法和l-v-1算法进行了比较,实验结果表明,提出的算法是有效可行的.
A method of binary tree multi-class support vector machine algorithm has been proposed to solve the problems of the current multi-class support vector machine algorithm,which can effectively reduce the members of support vectors and training time.Compared with l-v-r and l-v-l algorithm,experiments shows that the binary tree multi-class support vector machine algorithm is more effective and feasible.
出处
《新疆大学学报(自然科学版)》
CAS
2011年第1期100-104,共5页
Journal of Xinjiang University(Natural Science Edition)
基金
吉林省科技发展规划项目(20090503)
教育部科技发展中心项目(20090043110010)
关键词
WEB文本分类
二叉树
多分类SVM
Web text classification
binary tree
multi-classification SVM