应用遗传算法优化子空间的SVM分类算法

摘要/Abstract

摘要： 提出了一种应用遗传算法优化子空间的SVM分类算法GS-SVM。该算法首先改进样本选择策略,采用基于置信度和凸包的样本选择方法,考虑类间距离和样本分布等因素,选择典型代表样本作为SVM的新训练集；然后采用矩阵式混合编码方式,利用遗传算法一并优化代表样本的特征子空间和SVM分类参数,并根据特征优化后的代表样本,构建SVM分类模型。在UCI的11个数据集上进行的仿真实验结果表明,该算法在大部分数据集上均可获得较小的样本规模和特征维数,以及较高的分类精度。

关键词: 子空间分类,遗传算法,支持向量机,样本选择,凸包

Abstract: This paper presented a new GA based Subspace classification algorithm for SVM(GS-SVM)．A modified sample selection method is adopted to select a subset of training data based on both the confidence and the convex hull．Then the representative samples are selected to train the SVM models by considering the distances between classes and the sample distribution．The algorithm adopts the matrix-form mixed encoding．Genetic algorithm is used to optimize the feature subspace of representative samples and the classification parameters of SVM simultaneously．The SVM classification model is produced based on the representative samples with the optimized feature subspace．Experimental results on eleven UCI datasets illustrate that the proposed algorithm is able to select both smaller sample subset and feature size,and achieve higher classification accuracy than the traditional classification algorithms.

Key words: Subspace classification,Genetic algorithm,Support vector machine,Sample selection,Convex hull

蒋华荣,郁雪. 应用遗传算法优化子空间的SVM分类算法[J]. 计算机科学, 2013, 40(11): 255-260. https://doi.org/

JIANG Hua-rong and YU Xue. GA-based Subspace Classification Algorithm for Support Vector Machines[J]. Computer Science, 2013, 40(11): 255-260. https://doi.org/

参考文献

[1] He Qiang,Xie Zong-xia,Hu Qing-hua,et al．Neighborhoodbased sample and feature selection for SVM classification learning[J]．Neurocomputing,2011,74(10):1585-1594
[2] Huang C L,Wang C J．A GA-based feature selection and parameters optimization for support vector machines[J]．Expert Systems with Applications,2006,35(4):231-240
[3] 王世卿,曹彦．基于遗传算法和支持向量机的特征选择研究[J]．计算机工程与设计,2010,31(18):4088-4092
[4] Tian Jin,Li Min-qiang,Chen Fu-zan．Dual-population based coevolutionary algorithm for designing RBFNN with feature selection[J]．Expert Systems with Applications,2010,37(10):6904-6918
[5] Oja E．Subspace methods of pattern recognition[M]．New York:Research Studies Press,1983
[6] Qiu Guo-ping,Fang Jian-zhong．Classification in an informative sample subspace[J]．Pattern Recognition,2008,41(3):949-960
[7] Cevikalp H,Neamtu M,Barkana A．Kernel common vector me-thod:A Novel nonlinear subspace classifier for pattern recognition[J]．IEEE Transactions On Systems,Man and Cybernetics,2007,37(4):937-951
[8] Wen Ying．An improved discriminative common vectors and support vector machine based face recognition approach[J]．Expert Systems with Applications,2012,39(4):4628-4632
[9] Sakano H,Mukawa N,Nakamura T．Kernel mutual subspacemethod and its application for object recognition[J]．Electronics and Communications in Japan(Part II:Electronics),2005,88(6):45-53
[10] Kazuhiro F,Osamu Y．The kernel orthogonal mutual subspace method and its application to 3D object recognition[C]∥8th Asian Conference on Computer Vision．Tokyo,2007:467-476
[11] Zhu Mei-hong,Li Ai-hua．Random subspace method for improving performance of credit cardholder classification[C]∥Mode-ling Risk Management for Resources and Environment in China．Berlin,2011:257-264
[12] Watanabe S,Lambert P F,Kulikowski C A,et al．Evaluation and selection of variables in pattern recognition[C]∥Computer and Information Sciences II．New York,1967
[13] Zhang Peng,Peng Jing,Domeniconi C．Kernel pooled local subspaces for classification[J]．IEEE Transactions on Systems,Man,and Cybernetics,Part B:Cybernetics,2005,5(3):489-502
[14] Kitamura T,Takeuchi S,Abe S,et al．Subspace-based support vector machines for pattern classification[J]．Neural Networks,2009,22(5/6):558-567
[15] Wang Li-po,Zhou N,Chu Feng．A General Wrapper Approach to Selection of Class-Dependent Features[J]．IEEE Transactions on Neural Networks,2008,19(7):1267-1278
[16] Vapnik V N．The nature of statistical learning theory[M].Springer,1999
[17] Shin H,Cho S．Invariance of neighborhood relation under input space to feature space mapping[J]．Pattern Recognition Letters,2005,26(6):707-718
[18] 翟俊海,李胜杰,王熙照．基于粗糙集技术的压缩近邻规则[J]．计算机科学,2012,39(2):236-239
[19] Wang Ji-gang,Neskovic P,Cooper L N．Training data selection for support vector machines[C]∥Advances in Natural Computation．Changsha,2005:554-564
[20] 周晓飞,姜文瀚,杨静宇．基于子空间样本选择的最近凸包分类器[J]．计算机工程,2008,34(12):167-168
[21] 姜文瀚,周晓飞,杨静宇．核子类凸包样本选择方法及其SVM应用[J]．计算机工程,2008,4(16):212-214
[22] Zhou Xiao-fei,Jiang Wen-han,Tian Ying-jie,et al．Kernel subclass convex hull sample selection method for svm on face recognition[J].Neurocomputing,2010,10-12(73):2234-2246

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed