SVM21norm
所属分类:matlab编程
开发工具:matlab
文件大小:19KB
下载次数:9
上传日期:2018-11-29 17:17:02
上 传 者:
arafet
说明: A matlab implementation of SVM for multiclass learning. For more information: Xiao Cai, Feiping Nie, Heng Huang, and Chris Ding. Multi-Class L2,1-Norms Support Vector Machine. IEEE International Conference on Data Mining (ICDM), Vancouver, Canada, 2011.
文件列表:
SVM21norm (0, 2014-11-09)
SVM21norm\callLiblinear.m (237, 2011-09-11)
SVM21norm\callMSVM_WW.m (185, 2011-09-11)
SVM21norm\libsvmwrite.mexa64 (9714, 2011-06-14)
SVM21norm\Loop.m (1837, 2011-09-14)
SVM21norm\MCSVM21norm.m (215, 2011-09-14)
SVM21norm\quadsolve.m (11154, 2004-11-18)
SVM21norm\readLiblinearW.m (512, 2011-09-07)
SVM21norm\slinearsolve.m (12321, 2003-05-12)
SVM21norm\SVMCrossValidation.m (7881, 2011-10-13)
SVM21norm\svm_multi_init.m (2101, 2003-05-12)
SVM21norm\svm_multi_predK.m (10234, 2003-05-12)
This code is running in Linux OS for the ICDM 2011 paper "Multi-class l2, 1-norm support vector machine",
which includes two multi-class SVM solvers, 'CS' (from Liblinear) and 'WW' (from J Weston, C Watkins)
You can refer the following way to call the code,
fprintf('Multi-class SVM21norm WW\n');
para.c = 0.5;
NIter = 10;
t_ic = cputime;
[rank_list, fea_rank_list_value, num_iter] = MCSVM21norm(trainFeature, trainLabel, para, NIter, 1);
t_oc = cputime;
t_elapse_1 = t_oc - t_ic;
fprintf('Multi-class SVM21norm CS\n');
para.c = 0.5;
para.beta = 0.5;
para.tol = 1e-3;
NIter = 30;
t_ic = cputime;
[rank_list, fea_rank_list_value, num_iter] = MCSVM21norm(trainFeature, trainLabel, para, NIter, 2);
t_oc = cputime;
t_elapse_2 = t_oc - t_ic;
近期下载者:
相关文件:
收藏者: