Scilab Home Page | Wiki | Bug Tracker | Forge | Mailing List Archives | Scilab Online Help | File Exchange
ATOMS : libsvm and liblinear details
Please login or create an account

libsvm and liblinear

Libraries for SVM and large-scale linear classification
(1343 downloads for this version - 14724 downloads for all versions)
Holger Nahrstaedt
Chih-Chung Chang
Chih-Jen Lin
Technische Universitaet Berlin
Package maintainer
Holger Nahrstaedt
Supported Scilab Versions
>= 5.4
Creation Date
March 25, 2015
ATOMS packaging system
Available on
Install command
--> atomsInstall('libsvm')
This tool provides a simple interface to LIBSVM, a library for support vector machines ( It is very easy to use as the usage and the way of specifying parameters are the same as that of LIBSVM. This tool provides also a simple interface to LIBLINEAR, a library for large-scale regularized linear classification ( It is very easy to use as the usage and the way of specifying parameters are the same as that of LIBLINEAR. This Toolbox is compatible with the NaN-toolbox! Changelog ============ 1.4.5 - libsvm_loadmodel and libsvm_savemodel fixed - st_deviation renamed to stdev 1.4.4 - 2nu-SVM added - LIBLINEAR is updated to 1.94 - LIBSVM is updated to 3.20 - some bugfixes - The crossvalidation (libsvm_svmtrain with "-v 5") result is now a vector with [Cross Validation Accuracy, Positive Cross Validation Accuracy, Negative Cross Validation Accuracy] 1.4.3 - Depreated stack-c function were removed 1.4.2 - new functions libsvm_savemodel and libsvm_loadmodel 1.4.1 - buxfix for getpath - help files fixed for libsvm_linpredict and libsvm_lintrain - path operations are replaced by fullfile 1.4.0 - unit tests for libsvmwrite and libsvmread - fix issues 805, 806, 808, 809, 813, 814, - renaming of the following functions: *svmtrain > libsvm_svmtrain *svmpredict > libsvm_svmpredict *train > libsvm_lintrain *predict > libsvm_linpredict *svmconfmat > libsvm_confmat *svmgrid > libsvm_grid *svmgridlinear > libsvm_gridlinear *svmnormalize > libsvm_normalize *svmpartest > libsvm_partest *svmrocplot > libsvm_rocplot *svmscale > libsvm_scale *svmtoy > libsvm_toy 1.3.1 - compatible with scilab-5.4.0-beta-1 and scilab-5.4.0-alpha-1 or lower 1.3 - compatible with scilab-5.4.0-beta-1 - incompatible with scilab-5.4.0-alpha-1 and lower - fix several bugs in examples - fix precomputed kernel bug in svmtrain - LIBLINEAR is updated to 1.91 - LIBSVM is updated to 3.12 1.2.2 - some bug fixes - help files improved 1.2.1 - svmtoy added - improved error handling in sci_gateway - improved help files - bug in performance demo removed 1.2 - the Nan-Toolbox 1.3 is compatible to this toolbox now! - improved help-files - improved demos - LIBLINEAR with optional instance weight support 1.1 - improved demos - works under Windows - new function: svmnormalize 1.0 - first release of libsvm - toolbox 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. It was converted to Scilab 5.3 by Holger Nahrstaedt from TU Berlin. 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. ACM Transactions on Intelligent Systems and Technology, 2:27:1--27:27, 2011. Software available at Please cite LIBLINEAR as follows R.-E. Fan, K.-W. Chang, C.-J. Hsieh, X.-R. Wang, and C.-J. Lin. LIBLINEAR: A Library for Large Linear Classification, Journal of Machine Learning Research 9(2008), 1871-1874.Software available at
Files (6)
[217.06 Ko] libsvm_1.4.5-3.bin.i686.linux.tar.gz
Linux 32-bit
Automatically generated by the ATOMS compilation chain

[213.20 Ko] libsvm_1.4.5-3.bin.x86_64.darwin.tar.gz
MacOSX version
Automatically generated by the ATOMS compilation chain

[224.66 Ko] libsvm_1.4.5-3.bin.x86_64.linux.tar.gz
Linux 64-bit
Automatically generated by the ATOMS compilation chain

[245.42 Ko]
Windows 32-bit
Automatically generated by the ATOMS compilation chain

[257.82 Ko]
Windows 64-bit
Automatically generated by the ATOMS compilation chain

News (0)
Comments (1)
    Leave a comment 
Comment from Marc Albertelli -- April 21, 2015, 10:54:13 AM    
Unlike to classication, the "probability estimates" options doesn't work with
The svm_predict function returns a null array for the "decision_values"
Thanks a lot for your help,

ps : see below a very simple example (derived from the demos) that illustrates the problem

N = 20;
M = 1;
t = rand(N,1,'norm');

m = 1//:10:100;
x = [t];
for ii=1:M-1
    x = [x  t+ii*rand(N,1,'norm')/2];

x=libsvm_scale(x,[0 1]);
y = 2*t + rand(N,1,"norm")/2 + 7;

model = libsvm_svmtrain(y(:),x(:,:),'-s 4 -t 2 -n 0.5 -c 1 -b 1');
[predicted_label, accuracy, decision_values] = libsvm_svmpredict(y(:),x(:,:), model, '-b

Leave a comment
You need to log in before you can leave a comment.