SVMlin 1.0

Vikas Sindhwani — Tue, 27 Nov 2007 08:04:48 -0000

SVMlin: Fast Linear SVMs for Supervised and Semi-supervised Learning

SVMlin is software package for linear SVMs. It is well-suited to classification problems involving a large number of examples and features. It is primarily written for sparse datasets (number of non-zero features in an example is typically small). It is written in C/C++. A mex wrapper is available for MATLAB users.

SVMlin can also utilize unlabeled data, in addition to labeled examples. It currently implements two extensions of standard SVMs to incorporate unlabeled examples.

SVMlin (version 1.0) implements the following algorithms:

Fully supervised (using only labeled examples)

Linear Regularized Least Squares (RLS) Classification
Modified Finite Newton Linear L2-SVMs (Keerthi and DeCoste, JMLR, 2005)

Semi-supervised (Large Scale Semi-supervised Linear SVMs, Keerthi and Sindhwani, SIGIR 2006)

Multi-switch linear Transductive L2-SVMs
Deterministic Annealing (DA) for Semi-supervised Linear L2-SVMs

mloss.org SVMlin

SVMlin 1.0