SVMstruct 
 http://svmlight.joachims.org/svm_struct.html 
SVMstruct,by Joachims, is an SVM implementation that can model complex(multivariate) output data y, such as trees, sequences, or sets. Thesecomplex output SVM models can be applied to natural language parsing,sequence alignment in protein homology detection, and Markov models forpart-of-speech tagging. Several implementations exist: SVMmulticlass,for multi-class classification; SVMcfg, learns a weighted context freegrammar from examples; SVMalign, learns to align protein sequences fromtraining alignments; SVMhmm, learns a Markov model from examples. Thesemodules have straightforward applications in bioinformatics, but one canimagine significant implementations for cheminformatics, when thechemical structure is represented as trees or sequences.

 mySVM 
 http://www-ai.cs.uni-dortmund.de/SOFTWARE/MYSVM/index.html 
mySVM,by Stefan Rüping, is a C++ implementation of SVM classification andregression. Available as C++ source code and Windows binaries. Kernels:linear, polynomial, radial basis function, neural (tanh), anova.

 JmySVM 
 http://www-ai.cs.uni-dortmund.de/SOFTWARE/YALE/index.html 
JmySVM,a Java version of mySVM is part of the YaLE (Yet Another LearningEnvironment) learning environment.

 mySVM/db 
 http://www-ai.cs.uni-dortmund.de/SOFTWARE/MYSVMDB/index.html 
mySVM/dbis an efficient extension of mySVM which is designed to run directlyinside a relational database using an internal JAVA engine. It wastested with an Oracle database, but with small modifications it shouldalso run on any database offering a JDBC interface. It is especiallyuseful for large datasets available as relational databases.

 LIBSVM 
 http://www.csie.ntu.edu.tw/~cjlin/libsvm/ 
LIBSVM(Library for Support Vector Machines), is developed by Chang and Linand contains C-classification, ν-classification, ε-regression, andν-regression. Developed in C++ and Java, it supports also multi-classclassification, weighted SVM for unbalanced data, cross-validation andautomatic model selection. It has interfaces for Python, R, Splus,MATLAB, Perl, Ruby, and LabVIEW. Kernels: linear, polynomial, radialbasis function, and neural (tanh).

 looms 
 http://www.csie.ntu.edu.tw/~cjlin/looms/ 
looms,by Lee and Lin, is a very efficient leave-one-out model selection forSVM two-class classification. While LOO cross-validation is usually tootime consuming to be performed for large datasets, looms implementsnumerical procedures that make LOO accessible. Given a range ofparameters, looms automatically returns the parameter and model with thebest LOO statistics. Available as C source code and Windows binaries.

 BSVM 
 http://www.csie.ntu.edu.tw/~cjlin/bsvm/ 
BSVM,authored by of Hsu and Lin, provides two implementations of multi-classclassification, together with SVM regression. Available as source codefor UNIX/Linux and as binaries for Windows.

 SVMTorch 
 http://www.idiap.ch/learning/SVMTorch.html 
SVMTorch,by Collobert and Bengio, is part of the Torch machine learning libraryand implements SVM classification and regression. Distributed as C++source code or binaries for Linux and Solaris.

 Weka 
 http://www.cs.waikato.ac.nz/ml/weka/ 
Wekais a collection of machine learning algorithms for data mining tasks.The algorithms can either be applied directly to a dataset or calledfrom a Java code. Contains an SVM implementation.

 SVM in R 
 http://cran.r-project.org/src/contrib/Descriptions/e1071.html 
ThisSVM implementation in R (http://www.r-project.org/) containsC-classification, n-classification, e-regression, and n-regression.Kernels: linear, polynomial, radial basis, neural (tanh).

 M-SVM 
 http://www.loria.fr/~guermeur/ 
Multi-classSVM implementation in C by Guermeur.

 Gist 
 http://microarray.cpmc.columbia.edu/gist/ 
Gistis a C implementation of support vector machine classification andkernel principal components analysis. The SVM part of Gist is availableas an interactive web server at http://svm.sdsc.edu and it is a veryconvenient option for users that want to experiment with small datasets(several hundreds patterns). Kernels: linear, polynomial, radial.

 MATLAB SVM Toolbox 
 http://www.isis.ecs.soton.ac.uk/resources/svminfo/ 
ThisSVM MATLAB toolbox, by Gunn, implements SVM classification andregression with various kernels: linear, polynomial, Gaussian radialbasis function, exponential radial basis function, neural (tanh),Fourier series, spline, and B spline.

 TinySVM 
 http://chasen.org/~taku/software/TinySVM/ 
TinySVMis a C++ implementation of C-classification and C-regression which usessparse vector representation and can handle several ten-thousands oftraining examples, and hundred-thousands of feature dimensions.Distributed as binary/source for Linux and binary for Windows.

 SmartLab 
 http://www.smartlab.dibe.unige.it/ 
SmartLabprovides several support vector machines implementations: cSVM, Windowsand Linux implementation of two-classes classification; mcSVM, Windowsand Linux implementation of multi-classes classification; rSVM, Windowsand Linux implementation of regression; javaSVM1 and javaSVM2, Javaapplets for SVM classification.

 Gini-SVM 
 http://bach.ece.jhu.edu/svm/ginisvm/ 
Gini-SVM,by Chakrabartty and Cauwenberghs, is a multi-class probabilityregression engine that generates conditional probability distribution asa solution. Available as source code.

 GPDT 
 http://dm.unife.it/gpdt/ 
GPDT,by Serafini, Zanni, and Zanghirati, is a C++ implementation forlarge-scale SVM classification in both scalar and distributed memoryparallel environments. Available as C++ source code and Windowsbinaries.

 HeroSvm 
 http://www.cenparmi.concordia.ca/~people/jdong/HeroSvm.html 
HeroSvm,by Dong, is developed in C++, implements SVM classification, and isdistributed as a dynamic link library for Windows. Kernels: linear,polynomial, radial basis function.

 Spider 
 http://www.kyb.tuebingen.mpg.de/bs/people/spider/ 
Spideris an object orientated environment for machine learning in MATLAB, forunsupervised, supervised or semi-supervised machine learning problems,and includes training, testing, model selection, cross-validation, andstatistical tests. Implements SVM multi-class classification andregression.

 Java applets 
 http://svm.dcs.rhbnc.ac.uk/ 
TheseSVM classification and SVM regression Java applets were developed bymembers of Royal Holloway, University of London and AT&T Speech andImage Processing Services Research Lab.

 LEARNSC 
 http://www.support-vector.ws/html/downloads.html 
MATLABscripts for the book Learning and Soft Computing by Kecman,implementing SVM classification and regression.

 Tree Kernels 
 http://ai-nlp.info.uniroma2.it/moschitti/Tree-Kernel.htm 
TreeKernels, by Moschitti, is an extension of SVMlight, obtained byencoding tree kernels. Available as binaries for Windows, Linux,Mac-OSx, and Solaris. Tree kernels are suitable for encoding chemicalstructures, and thus this package brings significant capabilities forcheminformatics applications.

 LS-SVMlab 
 http://www.esat.kuleuven.ac.be/sista/lssvmlab/ 
LS-SVMlab,by Suykens, is a MATLAB implementation of least squares support vectormachines (LS-SVM) which reformulates the standard SVM leading to solvinglinear KKT systems. LS-SVM alike primal-dual formulations have beengiven to kernel PCA, kernel CCA and kernel PLS, thereby extending theclass of primal-dual kernel machines. Links between kernel versions ofclassical pattern recognition algorithms such as kernel Fisherdiscriminant analysis and extensions to unsupervised learning, recurrentnetworks and control are available.

 MATLAB SVM Toolbox 
 http://www.igi.tugraz.at/aschwaig/software.html 
Thisis a MATLAB SVM classification implementation which can handle 1-normand 2-norm SVM (linear or quadratic loss functions).

 SVM/LOO 
 http://bach.ece.jhu.edu/pub/gert/svm/incremental/ 
SVM/LOO,by Cauwenberghs, has a very efficient MATLAB implementation of theleave-one-out cross-validation.

 SVMsequel 
 http://www.isi.edu/~hdaume/SVMsequel/ 
SVMsequel,by Daume III, is a SVM multi-class classification package, distributedas C source or binaries for Linux or Solaris. Kernels: linear,polynomial, radial basis function, sigmoid, string, tree, informationdiffusion on discrete manifolds.

 LSVM 
 http://www.cs.wisc.edu/dmi/lsvm/ 
LSVM(Lagrangian Support Vector Machine) is a very fast SVM implementationin MATLAB by Mangasarian and Musicant. It can classify datasets withseveral millions patterns.

 ASVM 
 http://www.cs.wisc.edu/dmi/asvm/ 
ASVM(Active Support Vector Machine) is a very fast linear SVM script forMATLAB, by Musicant and Mangasarian, developed for large datasets.

 PSVM 
 http://www.cs.wisc.edu/dmi/svm/psvm/ 
PSVM(Proximal Support Vector Machine) is a MATLAB script by Fung andMangasarian which classifies patterns by assigning them to the closestof two parallel planes.

 OSU SVM Classifier MatlabToolbox 
 http://www.ece.osu.edu/~maj/osu_svm/ 
This MATLABtoolbox is based on LIBSVM.

 SimpleSVM Toolbox 
 http://asi.insa-rouen.fr/~gloosli/simpleSVM.html 
SimpleSVMToolbox is a MATLAB implementation of the SimpleSVM algorithm.

 SVM Toolbox 
 http://asi.insa-rouen.fr/%7Earakotom/toolbox/index 
Afairly complex MATLAB toolbox, containing many algorithms:classification using linear and quadratic penalization, multi-classclassification, ε-regression, ν-regression, wavelet kernel, SVM featureselection.

 MATLAB SVM Toolbox 
 http://theoval.sys.uea.ac.uk/~gcc/svm/toolbox/ 
Developedby Cawley, has standard SVM features, together with multi-classclassification and leave-one-out cross-validation.

 R-SVM 
 http://www.biostat.harvard.edu/~xzhang/R-SVM/R-SVM.html 
R-SVM,by Zhang and Wong, is based on SVMTorch and is specially designed forthe classification of microarray gene expression data. R-SVM uses SVMfor classification and for selecting a subset of relevant genesaccording to their relative contribution in the classification. Thisprocess is done recursively in such a way that a series of gene subsetsand classification models can be obtained in a recursive manner, atdifferent levels of gene selection. The performance of theclassification can be evaluated either on an independent test data setor by cross-validation on the same data set. Distributed as Linuxbinary.

 jSVM 
 http://www-cad.eecs.berkeley.edu/~hwawen/research/projects/jsvm/doc/manual/index.html 
jSVMis a Java wrapper for SVMlight.

 SvmFu 
 http://five-percent-nation.mit.edu/SvmFu/ 
SvmFu,by Rifkin, is a C++ package for SVM classification. Kernels: linear,polynomial, and Gaussian radial basis function.

 PyML 
 http://pyml.sourceforge.net/ 
PyMLis an interactive object oriented framework for machine learning inPython. It contains a wrapper for LIBSVM, and procedures for optimizing aclassifier: multi-class methods, descriptor selection, model selection,jury of classifiers, cross-validation, ROC curves.