摘要
以医疗数据为应用对象,应用网格搜索和交叉验证的方法选择参数,建立最小二乘支持向量机分类器,进行实际验证,并与使用K近邻分类器(K-NN)和C4.5决策树两种方法的结果进行比较。结果表明,LS-SVM分类器取得较高的准确率,表明最小二乘支持向量机在医疗诊断研究中具有很大的应用潜力。
As the medical diagnose dada an application object, LS-SVM classification hyper-parameters are optimized with grid-search and cross-validation method, carry through to validate the classification performance. Then it is compared to other typical classifications such as K-NN and C4.5 decision tree on the datasets. Computational results indicate that LS-SVM has good performance on the classification recognize, LS-SVM has potential application in medical diagnostics research.
出处
《计算机与数字工程》
2007年第9期21-23,共3页
Computer & Digital Engineering
基金
厦门大学985"海量数据挖掘方法及应用"项目
关键词
最小二乘支持向量机
分类器
医疗诊断
网格搜索
交叉验证
least squares support vector machines (LS-SVM),classification,medical diagnostics,grid-search,cross-validation