DPC: Screening Methods for Sparse Models via Dual Projection onto Convex Set

This package implements the highly effective screening rules — developed under the DPC framework — for many popular sparse learning models, like Lasso, non-negative Lasso, group Lasso, sparse-group Lasso etc.

Features including
- highly effective in reducing the data size by discarding the irrelevant variables
- highly effective in speeding up any existing algorithms for solving the sparse learning models
- extremely efficient in dealing with big data problems

Software Download

DPC package 2.1.1
TLFre rule for Sparse-Group Lasso is added.