(more than 6000 downloads on mloss.org)
MSVMpack is an open source package dedicated to multi-class support vector machines: SVMs which can handle more than two classes without relying on decomposition methods. The aim is to provide a unified framework and implementation for all the different M-SVM models in a single package.The package provides a set of command-line tools for training and testing M-SVMs together with a C API.
MSVMpack is available for Linux and Mac OS X (uses POSIX threads for parallelization).
k-LinReg is an open source software dedicated to switched linear regression with large data sets. It is based on a simple k-means like algorithm which provides both speed and accuracy.
Applications of k-LinReg include for instance hybrid (switched linear or piecewe affine) dynamical system identification.
k-LinReg is available as a platform-independent Matlab implementation or a parallel implementation in C for Linux.
COFSR is a Matlab implementation of the continuous optimization framework described in
F. Lauer, G. Bloch, and R. Vidal, A continuous optimization framework for hybrid system identification, Automatica, 47(3):608-613, 2011,
for switched regression and hybrid system identification.
Compared to k-LinReg, it offers multiple choices for the form of the objective function (smooth product-of-error type or nonsmooth minimum-of-error type), the regularization and the loss function, including losses that are robust to outliers.
The package also includes the code by Huyer and Neumaier for the MCS algorithm.