Model Parameter Selection of the Universum SVM Based on Particle Swarm Optimization
-
摘要: 分类器的模型参数对分类结果有直接影响.针对引入无关样本的Universum SVM算法中模型参数选择问题,采用粒子群优化(particle swarm optimization,PSO)算法对其进行优化.该方法概念简单、计算效率高且受问题维数变化的影响较小,可实现对多个参数同时优选.此外,在PSO中粒子适应度函数的选择是一个关键问题.考虑k遍交叉验证法的估计无偏性,利用交叉验证误差作为评价粒子优劣的适应值.通过舌象样本数据实验,对参数优选前后测试样本识别正确率进行比较,实验结果验证了该算法的有效性.
-
关键词:
- UniversumSVM /
- 参数选择 /
- 粒子群算法
Abstract: The model parameters of a classifier directly affect the classification results.According to the traits of additional irrelevant samples in the learning process of Universum SVM,this paper optimizes parameters with particle swarm optimization(PSO) due to its simple concept,high computational efficiency,and less impact by the changes of the problem dimension;therefore,several parameters can be simultaneously optimized.Besides,selection for fitness function is a key factor in PSO algorithm.According to its unbiased estimation,k-fold cross validation error is considered as the fitness value,by which an evaluation on the particle can be obtained.Finally,through experiment on tongue samples,the recognition accuracy rates on test samples before and after optimizing the parameters are compared.Result verifies the effectiveness of the proposed algorithm.-
Keywords:
- Universum SVM /
- parameter selection /
- particle swarm optimization(PSO)
-
-
[1] 邓乃扬,田英杰.数据挖掘中的新方法——支持向量机[M].北京:科学出版社,2004. [2] 张春华.支持向量机中最优化问题的研究[D].北京:中国农业大学经济管理学院,2004.ZHANG Chun-hua.Optimization problems in supportvector machines[D].Beijing:College of Economics&Management,China Agricultural University,2004.(inChinese) [3] WESTON J,COLLOBERT R,SINZ F,et al.Inferencewith the Universum[C]∥Proceedings of the 23rdInternational Conference on Machine Learning,Pittsburge,Pennsylvania,June 25-29,2006:1009-1016.
[4] KING I.Maximum margin semi-supervised learning withirrelevant data[C/OL]∥The 7th Chinese Worshop onMachine Learning and Applications.[2010-12-12].http:∥lamda.nju.edu.cn/conf/MLA09/program.htm,Nan-jing,November 6-8,2009.
[5] CHERKASSKY V,DAI W Y.Empirical study of theuniversum SVM learning for high-dimensional data[C]∥Proceedings of the 19th International Conference onArtificial Neural Networks:PartⅠ,Limassol Cyprus,September 14-17,2009:932-941.
[6] HUANG Kai-zhu,XU Zeng-lin,KING I,et al.Semi-supervised learning from general unlabeled data[C]∥The8th IEEE International Conference on Data Mining,Pisa,Italy,December 15-19,2008:273-282.
[7] CHEN Shuo,ZHANG Chang-shui.Selecting informativeuniversum sample for semi-supervised learning[C]∥Proceedings of the 21st International Jont Conference,Pasadena,July 11-17,2009:1016-1021.
[8] ZHANG Dan,WANG Jing-dong,WANG Fei,et al.Semi-supervised classification with universum[C]∥In SIAMInternational Conference on Data Mining(SDM),Atlanta,April 24-26,2008:323-333.
[9] JIAO Yue,ZHANG Xin-feng,ZHUO Li,et al.Tongueimage classification based on universum SVM[C]∥The3rd International Conference on BioMedical Engineeringand Informatics,Yantai,October 16-18,2010,2:657-660.
[10] 邵信光,杨慧中,陈刚.基于粒子群优化算法的支持向量机参数选择及其应[J].控制理论与应用,2006,23(5):740-743,748.SHAO Xin-guang,YANG Hui-zhong,CHEN Gang.Parameters selection and application of support vectormachines based on particle swarm optimization algorithm[J].Control Theory&Applications,2006,23(5):740-743,748.(in Chinese) [11] 袁小艳,刘爱伦.基于PSO算法的支持向量机核参数选择问题研究[J].自动化技术与应用,2007,26(5):5-8.YUAN Xiao-yan,LIU Ai-lun.Kernel parameter selectionof the support vector machine based on particle swarmoptimization[J].Techniques of Automation andApplication,2007,26(5):5-8.(in Chinese) [12] 姚全珠,蔡婕.基于PSO的LS-SVM特征选择与参数优化算法[J].计算机工程与应用,2010,46(1):134-136,229. YAO Quan-zhu,CAI Jie.Feature selection and LS-SVMparameters optimization algorithm based on PSO[J].Computer Engineering and Applications,2010,46(1):134-136,229.
[13] SHI Y H,EBERHART R.A modified particle swarmoptimizer[C]∥Proceedings of IEEE World Congress onComputational Intelligence,New York,May 4-9,1998:69-73.
[14] 吴涛.核函数的性质、方法及其在障碍检测中的应用[D].长沙:国防科学技术大学机电工程与自动化学院,2003:18-23.WU Tao.Kernels'properties,tricks and its applicationson obstacle detection[D].Changsha:College ofMechatronic Engineering and Automation,NationalUniversity of Defense Technology,2003:18-23.(inChinese) [15] 王东,吴湘滨.利用粒子群算法优化SVM分类器的超参数[J].计算机应用,2008,28(1):134-135,139.WANG Dong,WU Xiang-bin.Utilizing particle swarmoptimization to optimize hyper-parameters of SVMclassifier[J].Computer Applications,2008,28(1):134-135,139.(in Chinese) [16] 陈帅,朱建宁,潘俊,等.最小二乘支持向量机的参数优化及其应用[J].华东理工大学学报:自然科学版,2008,34(2):278-282. CHEN Shuai,ZHU Jian-ning,PAN Jun,et al.Parameters optimization of LS-SVM and its application[J].Journal of East China University of Science andTechnology:Natural Science Edition,2008,34(2):278-282.
计量
- 文章访问数: 0
- HTML全文浏览量: 0
- PDF下载量: 0
下载:
