The SHOGUN machine learning toolbox's focus is on large scale learning methods with focus on Support Vector Machines (SVM), providing interfaces to python, octave, matlab, r and the command line.
This release features 8 successful Google Summer of Code projects and it is the result of an incredible effort by our students. All projects come with very cool ipython-notebooks that contain background, code examples and visualizations. These can be found on our webpage!
In addition, the following features have been added:
Added method to importance sample the (true) marginal likelihood of a
Gaussian Process using a posterior approximation.
Added a new class for classical probability distribution that can be
sampled and whose log-pdf can be evaluated. Added the multivariate
Gaussian with various numerical flavours.
Cross-validation framework works now with Gaussian Processes
Added nu-SVR for LibSVR class
Modelselection is now supported for parameters of sub-kernels of
combined kernels in the MKL context. Thanks to Evangelos Anagnostopoulos
Probability output for multi-class SVMs is now supported using various
heuristics. Thanks to Shell Xu Hu.
Added an "equals" method to all Shogun objects that recursively
compares all registered parameters with those of another instance --
up to a specified accuracy.
Added a "clone" method to all Shogun objects that creates a deep copy
Multiclass LDA. Thanks to Kevin Hughes.
Added a new datatype, complex128_t, for complex numbers. Math functions,
support for SGVector/Matrix, SGSparseVector/Matrix, and serialization
with Ascii and Xml files added. [Soumyajit De].
Added mini-framework for numerical integration in one variable. Implemented
Gauss-Kronrod and Gauss-Hermite quadrature formulas.
Changed from configure script to CMake by Viktor Gal.
Add C++0x and C++11 cmake detection scripts
ND-Array typmap support for python and octave modular.
Fix json serialization.
Fixed bugs in FITC inference method that caused wrong posterior results.
Fixed bugs in GP Regression that caused negative values for the variances.
Fixed two memory errors in the streaming-features framework.
Fixed bug in the Kernel Mean Matching implementation (thanks to Meghana Kshirsagar).
Bugfixes Cleanups and API Changes:
Switch compile system to cmake
SGSparseVector/Matrix are now derived from SGReferenceData and thus refcounted.
Move README and INSTALL files to top level directory.
Use common RefCount class for ReferencedData and CSGObjects.
Rename HMSVMLabels to SequenceLabels
Refactored method to fit a sigmoid to SVM scores, now in CStatistics,
still called from CBinaryLabels.
Use Dynamic arrays to hold preprocessors in features instead of raw pointers.
Use Dynamic arrays to hold Features in CombinedFeatures.
Use Dynamic arrays to hold Kernels in CombinedKernels/ProductKernels.
Use Eigen3 for GPs, LDA
- Operating System:
- Data Formats:
- JMLR-MLOSS Publication:
Multiple Kernel Learning,