python-picard

Preconditoned ICA for Real Data


Keywords
independent-component-analysis
License
BSD-3-Clause-Clear
Install
pip install python-picard==0.3

Documentation

L-BFGS ICA

CircleCI Travis Codecov

This repository hosts code to run a faster ICA solver based on L-BFGS combined a preconditioning strategy.

Dependencies

These are the dependencies to use lbfgsica:

  • numpy (>=1.8)
  • matplotlib (>=1.3)
  • scipy (>=0.16)
  • numexpr (>= 2.0)

Cite

If you use this code in your project, please cite:

Pierre Ablin, Jean-Francois Cardoso, and Alexandre Gramfort
Faster ICA by preconditioning with Hessian approximations
ArXiv Preprint, June 2017