Generalized Concomitant Lasso

This repository hosts the implementation of solvers for multi-task Generalized Concomitant Lasso

The algorithms are in ./shcl/*_fast.pyx. Currently implemented are:

• multi-task Generalized Concomitant Lasso (multitask_generalized_solver())
• multi-task Block Homoscedastic Concomitant Lasso (multitask_blockhomo_solver())
• Block Homoscedastic Concomitant Lasso (one task)

For optimal time performance, the algorithms are written in Cython, using calls to BLAS/LAPACK when possible.

Installation

Clone the repository:

$git clone https://github.com/mathurinm/SHCL.git
$cd SHCL/
$conda env create --file environment.yml
$source activate shcl-env
$pip install --no-deps -e .  # do not forget the . at the end

Examples

Once you have created the conda environment with the previous command, you can run the examples with:

$source activate shcl-env
$ipython -i examples/blockhomo_example.py

Dependencies

All dependencies are in ./environment.yml

Cite

If you use this code, please cite this paper:

Mathurin Massias, Olivier Fercoq, Alexandre Gramfort and Joseph Salmon

Generalized Concomitant Multi-Task Lasso for sparse multimodal regression

to appear in AISTATS 2018