/BoostingExp

Reproduction of the experiments of Boosting Pathwise Coordinate Optimization in High Dimensions

Primary LanguageC++

BoostingExp

The experiments in Boosting Pathwise Coordinate Optimization in High Dimensions The proposed ASP-Newton method is implemented in the lastest version of picasso on CRAN: install.package

Dependency

Install old picasso The old picasso implements the greedy/cyclic selection rule and can be installed by

install.packages("oldpicasso", repos = NULL, type="source")

Install spams The accelerated proximal gradient method requires spams-R which can be installed by

install.packages("spams-R/spams", repos = NULL, type="source")

Linear Regression Experiments (Table 2)

source('linear/x.R')

Logistic Regression Experiments

source('logistic_regression/x.R')

Group Lasso

source('group_lasso/x.R')

Graph Estimation

source('graph_estimation/x.R')