This content refers to an unreleased development version of PyMVPA

To provide the most recent news and documentation www.pymvpa.org reflects the development 0.6 series of PyMVPA. If you are interested in the documentation of the previous stable 0.4 series of PyMVPA, please visit v04.pymvpa.org.

mvpa.clfs.svm

Importer for the available SVM and SVR machines.

Multiple external libraries implementing Support Vector Machines (Classification) and Regressions are available: LIBSVM, and shogun. This module is just a helper to provide default implementation for SVM depending on the availability of external libraries. By default LIBSVM implementation is chosen by default, but in any case both libraries are available through importing from this module:

> from mvpa.clfs.svm import sg, libsvm
> help(sg.SVM)
> help(libsvm.SVM)

Please refer to particular interface for more documentation about parameterization and available kernels and implementations.

Classes

LinearCSVMC(**kwargs[, C])	C-SVM classifier using linear kernel.
LinearNuSVMC(**kwargs[, nu])	Nu-SVM classifier using linear kernel.
LinearSVMKernel	A simple Linear kernel: K(a,b) = a*b.T
RbfCSVMC(**kwargs[, C])	C-SVM classifier using a radial basis function kernel
RbfNuSVMC(**kwargs[, nu])	Nu-SVM classifier using a radial basis function kernel
RbfSVMKernel	Radial Basis Function kernel (aka Gaussian):
SVM(**kwargs)	Support Vector Machine Classifier.