欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
核函数选择是支撑向量机（Support Vector Machine,SVM）研究的热点和难点.针对目前SVM核函数的选择没有统一规则的现状,探讨极坐标核在样本分类问题中的应用,提出一种结合样本分布特征进行SVM核选择的方法.首先分析极坐标核的映射原理,采用主成分分析方法（Principle Component Analysis,PCA）对高维数据集合理降维,在得到样本集分布特征的基础上进行SVM核选择,在Matlab环境中,采用四组数据集进行分类实验,验证结合样本分布特征选择SVM核函数的分类效果.实验结果表明,呈类圆形分布的样本集采用极坐标核进行分类,识别率达到100%,训练时间最短,优于采用高斯核SVM的分类效果.该方法提高了SVM的泛化能力,方案具有可行性和有效性.
英文摘要：
In Support Vector Machine study,kernel function selection is hot and difficult.Aiming at the current situation of no unified rules for SVM kernel function,the paper explores polar kernel function application in classification problem,puts forward a new way to select the kernel function based on the characteristics of dataset distribution.First Analysis of the mapping principle of the polar kernel function,then dimension reduction of the high dimensional dataset were processed with Principle Component Analysis method.On the basis of determining dataset distribution,how to select the kernel function was discussed.In the matlab environment,four groups of dataset were adopted to improve the classification experiment.The experimental results illustrate that the classification recognition rate of circle datasets reaches 100% with polar kernel and the training time is the shortest.The classification effect is better than that of using gaussian kernel SVM.The method can improve the generalization ability of SVM and the scheme is practical and feasible.
同期刊论文项目
同项目期刊论文
期刊信息
位置：