Kernel Statistics Toolbox

Developing Tools

• KGaussian: for building Gaussian kernel data.
• SVMs: smooth SVM solved in the primal; Lagrangian SVM solved in the dual.
• KPCA: kernel principal component analysis for dimension reduction.
• KSIR: kernel sliced inverse regression for dimension reduction.
• KFDA: kernel Fisher linear discriminant analysis for multiclass classification.
• KCCA: kernel canonical correlation analysis.
• SVRs: regularized LS SVR; smooth epsilon-insensitive SVR.
• kernel cluster analysis

Auxiliary Tools

• Reduced Kernel
• hibiscus-plus: a collection of UD-based model selection programs for SVM and SVR. It contains 3 main codes: hibiscus4LSSVR (for regularized least square SVR), hibiscus4mSSVR (for multi-response epsilon-SVR. It now only supports epsilon-SSVR. Support for LIBSVM for regression is in development.) and hibiscus4SVM (for SVM, it supports SSVM, LIBSVM and Lagrangian SVM).

Reference