Better solvers for SLOPE package
- Mentors
- Jonas Wallin, Johan Larsson
- Organization
- R Project for Statistical Computing
SLOPE package offers implementations that solve the Sorted L-One Penalized Estimation (SLOPE) model for various objective functions. However when predictors are correlated, current implementation is slow because it uses FISTA. As a part of this project, we aim to create a testing suite which contains implementations different algorithms, namely ADMM, proximal-Newton methods etc. and benchmark their performance. The best one will finally be incorporate into the SLOPE package.