VladoUzunangelov/SPICER: SPICER: SpicyMKL in R

This is a port of Tomioka and Suzuki's SpicyMKL, expanded for multiclass problems.SpicyMKL solves both binary classification (hinge & logitstic losses) and regression (squared error loss) problems.Spicer solves binary classification (logistic only), regression (squared error loss) and multiclass problems (logistic loss). The latter is done by expanding the binary SpicyMKL solver via one-versus-one class pairs training approach. In addition, Spicer outputs class or probability predictions for binary/multiclass classification tasks.

install.packages("remotes")
remotes::install_github("VladoUzunangelov/SPICER")