资源描述：
《matlab接口中libsvm的使用方法和可以使用的函数》由会员上传分享，免费在线阅读，更多相关内容在工程资料-天天文库。
1、[LIBSVM资料]matlab接口中libsvm的使用方法和可以使用的函数matlab,libsvm,接口在C版的libsvm中有很多函数，朋友们都认为matlab接口中也有相应的，其实不是！matlab接口中，libsvm可以使用的函数好像不多，具体在接口文件夹的README中有说明，可以使用的函数有如下几个，可以在matlab命令行直接使用（这是我的理解）：Usage=====matlab>model=svmtrain(training_label_vector,training_instance_matrix[,'libsvm_options']); -traini
2、ng_label_vector: Anmby1vectoroftraininglabels(typemustbedouble). -training_instance_matrix: Anmbynmatrixofmtraininginstanceswithnfeatures. Itcanbedenseorsparse(typemustbedouble). -libsvm_options: AstringoftrainingoptionsinthesameformatasthatofLIBSVM.matlab>
3、[predicted_label,accuracy,decision_values/prob_estimates]=svmpredict(testing_label_vector,testing_instance_matrix,model[,'libsvm_options']); -testing_label_vector: Anmby1vectorofpredictionlabels.Iflabelsoftest dataareunknown,simplyuseanyrandomvalues.(typemustbedouble) -t
4、esting_instance_matrix: Anmbynmatrixofmtestinginstanceswithnfeatures. Itcanbedenseorsparse.(typemustbedouble) -model: Theoutputofsvmtrain. -libsvm_options: AstringoftestingoptionsinthesameformatasthatofLIBSVM.ReturnedModelStructure========================Th
5、e'svmtrain'functionreturnsamodelwhichcanbeusedforfutureprediction. Itisastructureandisorganizedas[Parameters,nr_class,totalSV,rho,Label,ProbA,ProbB,nSV,sv_coef,SVs]: -Parameters:parameters -nr_class:numberofclasses;=2forregression/one-classsvm -totalSV:total#SV -rho:-bofthe
6、decisionfunction(s)wx+b -Label:labelofeachclass;emptyforregression/one-classSVM -ProbA:pairwiseprobabilityinformation;emptyif-b0orinone-classSVM -ProbB:pairwiseprobabilityinformation;emptyif-b0orinone-classSVM -nSV:numberofSVsforeachclass;emptyforregression/one-classSVM
7、 -sv_coef:coefficientsforSVsindecisionfunctions -SVs:supportvectorsIfyoudonotusetheoption'-b1',ProbAandProbBareemptymatrices.Ifthe'-v'optionisspecified,crossvalidationisconductedandthereturnedmodelisjustascalar:cross-validati
