摘要
针对传统对支持向量机多类分类算法(Multi-TWSVM)中出现的模糊性问题,提出了一种基于遗传算法的决策树对支持向量机(GA-DTTSVM)多类分类算法。GA-DTTSVM用遗传算法对特征数据建立决策树,通过构建决策树可以分离样本的模糊区域,提高模糊区域样本的识别率。在决策树的每个节点上用对支持向量机(TWSVM)训练分类器,最后用训练的分类器进行分类和预测。实验结果表明,与决策树对支持向量机(DTTSVM)多类分类算法以及Multi-TWSVM相比,GA-DTTSVM多类分类算法具有较高的分类精度和较快的训练速度。
Aiming at the fuzzy problem in Multi-class Twin Support Vector Machine(Multi-TWSVM), a new method of Decision Tree Twin Support Vector Machine based on Genetic Algorithm(GA-DTTSVM)is proposed. GA-DTTSVM builds the decision tree with the feature data by genetic algorithm to separate the fuzzy region of samples, so that the sample recognition rate can be improved. For each node of the decision tree this paper uses the Twin Support Vector Machine(TWSVM)to train a classifier, and finally it uses the trained classifier for classification and prediction. The experiments show that GA-DTTSVM algorithm can get higher classification accuracy and faster training speed compared with Decision Tree Twin Support Vector Machine algorithm(DTTSVM)and Multi-TWSVM.
作者
邹丽
蒋芸
陈娜
沈健
胡学伟
李志磊
ZOU Li;JIANG Yun;CHEN Na;SHEN Jian;HU Xuewei;LI Zhilei(College of Computer Science and Engineering, Northwest Normal University, Lanzhou 730070, China)
出处
《计算机工程与应用》
CSCD
北大核心
2016年第21期76-80,共5页
Computer Engineering and Applications
基金
国家自然科学基金(No.61163036
No.61163039)
2012年度甘肃省高校基本科研业务费专项资金项目
甘肃省高校研究生导师项目(No.1201-16)
西北师范大学第三期知识与创新工程科研骨干项目(No.nwnu-kjcxgc-03-67)
关键词
遗传算法
对支持向量机
分类和预测
genetic algorithm
twin support vector machine
classification and prediction
