弹性核k-NN分类算法及其在药物构效关系中的应用
|
摘要
核方法利用核函数可以有效地解决非线性问题,在药物构效关系领域得到了广泛的应用.本文提出了一种新的弹性核k-最近邻算法(EKk-NN).首先,基于加权多项式核和径向基函数核构造了一种信息丰富的弹性核,所构造的弹性核能有效地利用局部核和全局核的优点,同时也为构造核函数提供了一种可行的方法;然后,在核方法的框架下,将弹性核耦合到k-最近邻算法.实际数据集的实验和分析表明,EKk-NN能明显提高分类性能.
The kernel approaches have been gaining popularity in the field of drug Structure-Activity Relationship, which could effectively solve nonlinear problems by using the kernel function. In the present study, a novel Elastic Kernel k-Nearest Neighbor algorithm(EKk-NN) has been proposed. First, an informative novel elastic kernel is constructed based on polynomial kernel and radial basis function kernel.The constructed elastic kernel can effectively integrate the advantages of local kernel and global kernel, which can provide a feasible way for building the kernel function. Then, under the framework of kernel methods,this elastic kernel is extended to the k-Nearest Neighbor algorithm. Compared with the traditional kernel k-NN,experiments and analysis on the real data sets have shown that EKk-NN can significantly improve the performance of classification, which is really an attractive alternative technique.
The kernel approaches have been gaining popularity in the field of drug Structure-Activity Relationship, which could effectively solve nonlinear problems by using the kernel function. In the present study, a novel Elastic Kernel k-Nearest Neighbor algorithm(EKk-NN) has been proposed. First, an informative novel elastic kernel is constructed based on polynomial kernel and radial basis function kernel.The constructed elastic kernel can effectively integrate the advantages of local kernel and global kernel, which can provide a feasible way for building the kernel function. Then, under the framework of kernel methods,this elastic kernel is extended to the k-Nearest Neighbor algorithm. Compared with the traditional kernel k-NN,experiments and analysis on the real data sets have shown that EKk-NN can significantly improve the performance of classification, which is really an attractive alternative technique.
引文
[1]COURANT R,HILBERT D.Methods of mathematical physics:vol.1[M].New York:Interscience Publication,1953.
[2]HART P E.The condensed nearest neighbor rule[J].IEEETransactions on Information Theory,1968,14(3):515-516.
[3]WILSON D L.Asymptotic properties of nearest neighbor rules using edited data[J].IEEE Transactions on Systems,Man and Cybernetics,1972,SMC-2(3):408-421.
[4]ZHENG W F,TROPSHA A.Novel variable selection quantitative Structure-Property Relationship approach based on the k-Nearest-Neighbor principle[J].Journal of Chemical Information and Computer Sciences,2000,40(1):185-194.
[5]AJMANI S,JADHAV K,KULKARNI S A.Threedimensional QSAR using the k-Nearest Neighbor method and its interpretation[J].Journal of Chemical Information and Modeling,2006,46(1):24-31.
[6]刘松华,张军英,许进,等.Kernel-k NN:基于信息能度量的核k-最近邻算法[J].自动化学报,2010,36(12):1681-1688.
[7]YU K,JI L,ZHANG X G.Kernel Nearest-Neighbor algorithm[J].Neural Processing Letters,2002,15(2):147-156.
[8]HASTIE T,TIBSHIRANI R,FRIEDMAN J.The elements of statistical learning:data mining,inference and prediction:second edition[M].Berlin:Springer,2009.
[9]TAYLOR J S,CRISTIANINI N.Kernel methods for pattern analysis[M].London:Cambridge University Press,2004.
[10]SCH?LKOPF B,SMOLA A J.Learning with kernels[M].Cambridge:MIT Press,2002.
[11]XUE Y,LI Z R,YAP C W,et al.Effect of molecular descriptor feature selection in support vector machine classification of pharmacokinetic and toxicological properties of chemical agents[J].Journal of Chemical Information&Computer Sciences,2004,44(5):1630-1638.
[12]FAWCETT T.An introduction to ROC analysis[J].Pattern Recognition Letters,2006,27(8):861-874.
[2]HART P E.The condensed nearest neighbor rule[J].IEEETransactions on Information Theory,1968,14(3):515-516.
[3]WILSON D L.Asymptotic properties of nearest neighbor rules using edited data[J].IEEE Transactions on Systems,Man and Cybernetics,1972,SMC-2(3):408-421.
[4]ZHENG W F,TROPSHA A.Novel variable selection quantitative Structure-Property Relationship approach based on the k-Nearest-Neighbor principle[J].Journal of Chemical Information and Computer Sciences,2000,40(1):185-194.
[5]AJMANI S,JADHAV K,KULKARNI S A.Threedimensional QSAR using the k-Nearest Neighbor method and its interpretation[J].Journal of Chemical Information and Modeling,2006,46(1):24-31.
[6]刘松华,张军英,许进,等.Kernel-k NN:基于信息能度量的核k-最近邻算法[J].自动化学报,2010,36(12):1681-1688.
[7]YU K,JI L,ZHANG X G.Kernel Nearest-Neighbor algorithm[J].Neural Processing Letters,2002,15(2):147-156.
[8]HASTIE T,TIBSHIRANI R,FRIEDMAN J.The elements of statistical learning:data mining,inference and prediction:second edition[M].Berlin:Springer,2009.
[9]TAYLOR J S,CRISTIANINI N.Kernel methods for pattern analysis[M].London:Cambridge University Press,2004.
[10]SCH?LKOPF B,SMOLA A J.Learning with kernels[M].Cambridge:MIT Press,2002.
[11]XUE Y,LI Z R,YAP C W,et al.Effect of molecular descriptor feature selection in support vector machine classification of pharmacokinetic and toxicological properties of chemical agents[J].Journal of Chemical Information&Computer Sciences,2004,44(5):1630-1638.
[12]FAWCETT T.An introduction to ROC analysis[J].Pattern Recognition Letters,2006,27(8):861-874.