LIBSVM 2.9

ChihChung Chang, ChihJen Lin — Sat, 27 Feb 2010 01:09:23 -0000

LIBSVM is an integrated software for support vector classification, (C-SVC, nu-SVC ), regression (epsilon-SVR, nu-SVR) and distribution estimation (one-class SVM). It supports multi-class classification.

Our goal is to help users from other fields to easily use SVM as a tool. LIBSVM provides a simple interface where users can easily link it with their own programs. Main features of LIBSVM include

o Different SVM formulations

o Efficient multi-class classification

o Cross validation for model selection

o Probability estimates

o Weighted SVM for unbalanced data

o Both C++ and Java sources

o GUI demonstrating SVM classification and regression

o Python, R (also Splus), MATLAB, Perl, Ruby, Weka, CLISP and LabVIEW interfaces. C# .NET code is available. It's also included in some learning environments: YALE and PCP.

o Automatic model selection which can generate contour of cross valiation accuracy.

mloss.org LIBSVM

LIBSVM 2.9