基于支持向量机优化参数的集成多核典型相关分析
|
摘要
提出一种可有效处理多个数据集合之间变量关系的集成多核典型相关分析方法,构造一个特殊的核函数,使其更好地将原始样本数据映射到高维空间;基于支持向量机,在选择一个优化参数的基础上最大化多组数据集变量间的关系,以寻求整体相关性最大。在多特征手写体数字库上的实验证明,相比传统的典型相关分析与核典型相关分析方法,基于优化参数的集成多核典型相关分析方法具有更优的性能。
This paper presents an integrated multi-kernel canonical correlation analysis(IMKCCA) method,that can efficiently deal with the relationship among variables in multiple data sets. In this method, a special kernel function is constructed, which helped to project the original sample data onto a high-dimensional space; in addition, optimization parameters of SVM are selected to maximize the relationships among multiple groups of data set. Tests on Multiple Feature Data Set illustrated that, compared with the traditional canonical correlation analysis(CCA) and kernel canonical correlation analysis(KCCA), IMKCCA based on the optimum parameters of SVM has better performance.
This paper presents an integrated multi-kernel canonical correlation analysis(IMKCCA) method,that can efficiently deal with the relationship among variables in multiple data sets. In this method, a special kernel function is constructed, which helped to project the original sample data onto a high-dimensional space; in addition, optimization parameters of SVM are selected to maximize the relationships among multiple groups of data set. Tests on Multiple Feature Data Set illustrated that, compared with the traditional canonical correlation analysis(CCA) and kernel canonical correlation analysis(KCCA), IMKCCA based on the optimum parameters of SVM has better performance.
引文
[1]杨静,李文平,张健沛.大数据典型相关分析的云模型方法[J].通信学报,2013,34(10):121-134.
[2]孙平,徐宗本,申建中.基于核化原理的非线性典型相关判别分析[J].计算机学报,2004,27(6):789-795.
[3]邱爱昆,朱嘉钢.基于集成学习的多重集典型相关分析方法[J].计算机工程与应用,2017,53(6):162-168,173.
[4]许洁,梁久祯,吴秦,等.核典型相关分析特征融合方法及应用[J].计算机科学,2016,43(1):35-39.
[5]YUAN Y H,SHEN X B,XIAO Z Y,et al.Multiview correlation feature learning with multiple kernels[C]//HE X F,GAO X B,ZHANG Y N,et al.Intelligence science and big data engineering.Big data and machine learning techniques:5th International Conference,ISc IDE2015,Suzhou,China,June 14-16,2015.Revised selected papers partⅡ.1st ed.Cham:Springer,2015:518-528.
[6]徐健.基于多种测度的术语相似度集成计算研究[J].情报学报,2013,32(6):618-628.
[7]梁吉业,冯晨娇,宋鹏.大数据相关分析综述[J].计算机学报,2016,39(1):1-18.
[8]DUAN K,KEERTHI S S,POO A N.Evaluation of simple performance measures for tuning SVM hyperparameters[J].Neurocomputing,2003,51(4):41-49.
[9]郭美丽,覃锡忠,贾振红,等.基于改进的网格搜索SVR的话务预测模型[J].计算机工程与科学,2014,36(4):707-712.
[10]杨思春,高超,戴新宇,等.基于SVM的中文查询分类[J].情报学报,2011,30(9):946-950.
[11]YOSHIDA K,YOSHIMOTO J,DOYA K.Sparse kernel canonical correlation analysis for discovery of nonlinear interactions in highdimensional data[J].Bmc Bioinformatics,2017,18(1):108-118.
[12]郑小霞,钱锋.高斯核支持向量机分类和模型参数选择研究[J].计算机工程与应用,2006,1(1):77-79.
[13]汪廷华,陈峻婷.核函数的选择研究综述[J].计算机工程与设计,2012,33(3):1181-1186.
[14]罗小燕,陈慧明,卢小江,等.基于网格搜索与交叉验证的SVM磨机负荷预测[J].中国测试,2017,43(1):132-135,144.
[15]WANG J F,ZHANG L,CHEN G X.A parameter optimization method for an SVM based on improved grid search algorithm[J].Applied Science and Technology,2012,39(3):28-31.
[16]周航星,陈松灿.有序判别典型相关分析[J].软件学报,2014,25(9):2018-2025.
[17]YANG X H,LIU W F,TAO D P,et al.Canonical correlation analysis networks for two-view image recognition[J].Information Sciences,2017,385(5):338-352.
[2]孙平,徐宗本,申建中.基于核化原理的非线性典型相关判别分析[J].计算机学报,2004,27(6):789-795.
[3]邱爱昆,朱嘉钢.基于集成学习的多重集典型相关分析方法[J].计算机工程与应用,2017,53(6):162-168,173.
[4]许洁,梁久祯,吴秦,等.核典型相关分析特征融合方法及应用[J].计算机科学,2016,43(1):35-39.
[5]YUAN Y H,SHEN X B,XIAO Z Y,et al.Multiview correlation feature learning with multiple kernels[C]//HE X F,GAO X B,ZHANG Y N,et al.Intelligence science and big data engineering.Big data and machine learning techniques:5th International Conference,ISc IDE2015,Suzhou,China,June 14-16,2015.Revised selected papers partⅡ.1st ed.Cham:Springer,2015:518-528.
[6]徐健.基于多种测度的术语相似度集成计算研究[J].情报学报,2013,32(6):618-628.
[7]梁吉业,冯晨娇,宋鹏.大数据相关分析综述[J].计算机学报,2016,39(1):1-18.
[8]DUAN K,KEERTHI S S,POO A N.Evaluation of simple performance measures for tuning SVM hyperparameters[J].Neurocomputing,2003,51(4):41-49.
[9]郭美丽,覃锡忠,贾振红,等.基于改进的网格搜索SVR的话务预测模型[J].计算机工程与科学,2014,36(4):707-712.
[10]杨思春,高超,戴新宇,等.基于SVM的中文查询分类[J].情报学报,2011,30(9):946-950.
[11]YOSHIDA K,YOSHIMOTO J,DOYA K.Sparse kernel canonical correlation analysis for discovery of nonlinear interactions in highdimensional data[J].Bmc Bioinformatics,2017,18(1):108-118.
[12]郑小霞,钱锋.高斯核支持向量机分类和模型参数选择研究[J].计算机工程与应用,2006,1(1):77-79.
[13]汪廷华,陈峻婷.核函数的选择研究综述[J].计算机工程与设计,2012,33(3):1181-1186.
[14]罗小燕,陈慧明,卢小江,等.基于网格搜索与交叉验证的SVM磨机负荷预测[J].中国测试,2017,43(1):132-135,144.
[15]WANG J F,ZHANG L,CHEN G X.A parameter optimization method for an SVM based on improved grid search algorithm[J].Applied Science and Technology,2012,39(3):28-31.
[16]周航星,陈松灿.有序判别典型相关分析[J].软件学报,2014,25(9):2018-2025.
[17]YANG X H,LIU W F,TAO D P,et al.Canonical correlation analysis networks for two-view image recognition[J].Information Sciences,2017,385(5):338-352.