I've observed that SVMs fit with sklearn consistently get around 5 percentage points lower accuracy than equivalent SVMs fit with Adam Coates' SVM implementation based on minFunc. Am I overlooking some basic usage issue (eg too loose of a default convergence criterion), or is this likely to be a defect in the underlying libsvm implementation?
To demonstrate, run svm_comparison.m in matlab then svm_comparison.py in python. You'll need Adam Coates' code from http://www.stanford.edu/~acoates/papers/sc_vq_demo.tgz for train_svm to work. To be clear, SVC and Adam Coates' code are supposedly minimizing the exact same convex loss function, so this really shouldn't happen.
svm_comparison.m
Description: application/vnd.wolfram.mathematica.package
svm_comparison.py
Description: Binary data
------------------------------------------------------------------------------ Keep Your Developer Skills Current with LearnDevNow! The most comprehensive online learning library for Microsoft developers is just $99.99! Visual Studio, SharePoint, SQL - plus HTML5, CSS3, MVC3, Metro Style Apps, more. Free future releases when you subscribe now! http://p.sf.net/sfu/learndevnow-d2d
_______________________________________________ Scikit-learn-general mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
