----------------------------------------- --- MATLAB/OCTAVE interface of LIBSVM --- ----------------------------------------- Table of Contents ================= - Introduction - Installation - Usage - Returned Model Structure - Examples - Other Utilities - Additional Information Introduction ============ This tool provides a simple interface to LIBSVM, a library for support vector machines (http://www.csie.ntu.edu.tw/~cjlin/libsvm). It is very easy to use as the usage and the way of specifying parameters are the same as that of LIBSVM. Installation ============ On Unix systems, we recommend using GNU g++ as your compiler and type 'make' to build 'svmtrain.mexglx' and 'svmpredict.mexglx'. Note that we assume your MATLAB is installed in '/usr/local/matlab', if not, please change MATLABDIR in Makefile. Example: linux> make To use Octave, type 'make octave': Example: linux> make octave On Windows systems, pre-built 'svmtrain.mexw32' and 'svmpredict.mexw32' are included in this package, so no need to conduct installation. If you have modified the sources and would like to re-build the package, type 'mex -setup' in MATLAB to choose a compiler for mex first. Then type 'make' to start the installation. Starting from MATLAB 7.1 (R14SP3), the default MEX file extension is changed from .dll to .mexw32 or .mexw64 (depends on 32-bit or 64-bit Windows). If your MATLAB is older than 7.1, you have to build these files yourself. Example: matlab> mex -setup (ps: MATLAB will show the following messages to setup default compiler.) Please choose your compiler for building external interface (MEX) files: Would you like mex to locate installed compilers [y]/n? y Select a compiler: [1] Microsoft Visual C/C++ version 7.1 in C:\Program Files\Microsoft Visual Studio [0] None Compiler: 1 Please verify your choices: Compiler: Microsoft Visual C/C++ 7.1 Location: C:\Program Files\Microsoft Visual Studio Are these correct?([y]/n): y matlab> make Under 64-bit Windows, Visual Studio 2005 user will need "X64 Compiler and Tools". The package won't be installed by default, but you can find it in customized installation options. For list of supported/compatible compilers for MATLAB, please check the following page: http://www.mathworks.com/support/compilers/current_release/ Usage ===== matlab> model = svmtrain(training_label_vector, training_instance_matrix [, 'libsvm_options']); -training_label_vector: An m by 1 vector of training labels (type must be double). -training_instance_matrix: An m by n matrix of m training instances with n features. It can be dense or sparse (type must be double). -libsvm_options: A string of training options in the same format as that of LIBSVM. matlab> [predicted_label, accuracy, decision_values/prob_estimates] = svmpredict(testing_label_vector, testing_instance_matrix, model [, 'libsvm_options']); -testing_label_vector: An m by 1 vector of prediction labels. If labels of test data are unknown, simply use any random values. (type must be double) -testing_instance_matrix: An m by n matrix of m testing instances with n features. It can be dense or sparse. (type must be double) -model: The output of svmtrain. -libsvm_options: A string of testing options in the same format as that of LIBSVM. Returned Model Structure ======================== The 'svmtrain' function returns a model which can be used for future prediction. It is a structure and is organized as [Parameters, nr_class, totalSV, rho, Label, ProbA, ProbB, nSV, sv_coef, SVs]: -Parameters: parameters -nr_class: number of classes; = 2 for regression/one-class svm -totalSV: total #SV -rho: -b of the decision function(s) wx+b -Label: label of each class; empty for regression/one-class SVM -ProbA: pairwise probability information; empty if -b 0 or in one-class SVM -ProbB: pairwise probability information; empty if -b 0 or in one-class SVM -nSV: number of SVs for each class; empty for regression/one-class SVM -sv_coef: coefficients for SVs in decision functions -SVs: support vectors If you do not use the option '-b 1', ProbA and ProbB are empty matrices. If the '-v' option is specified, cross validation is conducted and the returned model is just a scalar: cross-validation accuracy for classification and mean-squared error for regression. More details about this model can be found in LIBSVM FAQ (http://www.csie.ntu.edu.tw/~cjlin/libsvm/faq.html) and LIBSVM implementation document (http://www.csie.ntu.edu.tw/~cjlin/papers/libsvm.pdf). Result of Prediction ==================== The function 'svmpredict' has three outputs. The first one, predictd_label, is a vector of predicted labels. The second output, accuracy, is a vector including accuracy (for classification), mean squared error, and squared correlation coefficient (for regression). The third is a matrix containing decision values or probability estimates (if '-b 1' is specified). If k is the number of classes, for decision values, each row includes results of predicting k(k-1/2) binary-class SVMs. For probabilities, each row contains k values indicating the probability that the testing instance is in each class. Note that the order of classes here is the same as 'Label' field in the model structure. Examples ======== Train and test on the provided data heart_scale: matlab> load heart_scale.mat matlab> model = svmtrain(heart_scale_label, heart_scale_inst, '-c 1 -g 0.07'); matlab> [predict_label, accuracy, dec_values] = svmpredict(heart_scale_label, heart_scale_inst, model); % test the training data For probability estimates, you need '-b 1' for training and testing: matlab> load heart_scale.mat matlab> model = svmtrain(heart_scale_label, heart_scale_inst, '-c 1 -g 0.07 -b 1'); matlab> load heart_scale.mat matlab> [predict_label, accuracy, prob_estimates] = svmpredict(heart_scale_label, heart_scale_inst, model, '-b 1'); To use precomputed kernel, you must include sample serial number as the first column of the training and testing data (assume your kernel matrix is K, # of instances is n): matlab> K1 = [(1:n)', K]; % include sample serial number as first column matlab> model = svmtrain(label_vector, K1, '-t 4'); matlab> [predict_label, accuracy, dec_values] = svmpredict(label_vector, K1, model); % test the training data Take linear kernel for example, the following precomputed kernel example gives exactly same training error as LIBSVM built-in linear kernel matlab> load heart_scale.mat matlab> n = size(heart_scale_inst,1); matlab> K = heart_scale_inst*heart_scale_inst'; matlab> K1 = [(1:n)', K]; matlab> model = svmtrain(heart_scale_label, K1, '-t 4'); matlab> [predict_label, accuracy, dec_values] = svmpredict(heart_scale_label, K1, model); Note that for testing, you can put anything in the testing_label_vector. For details of precomputed kernels, please read the section ``Precomputed Kernels'' in the README of the LIBSVM package. Other Utilities =============== A matlab function read_sparse reads files in LIBSVM format: [label_vector, instance_matrix] = read_sparse('data.txt'); Two outputs are labels and instances, which can then be used as inputs of svmtrain or svmpredict. This code is derived from svm-train.c in LIBSVM by Rong-En Fan from National Taiwan University. Additional Information ====================== This interface was initially written by Jun-Cheng Chen, Kuan-Jen Peng, Chih-Yuan Yang and Chih-Huai Cheng from Department of Computer Science, National Taiwan University. The current version was prepared by Rong-En Fan and Ting-Fan Wu. If you find this tool useful, please cite LIBSVM as follows Chih-Chung Chang and Chih-Jen Lin, LIBSVM : a library for support vector machines, 2001. Software available at http://www.csie.ntu.edu.tw/~cjlin/libsvm For any question, please contact Chih-Jen Lin , or check the FAQ page: http://www.csie.ntu.edu.tw/~cjlin/libsvm/faq.html#/Q9:_MATLAB_interface