Skip to main content

Classifier tuned for neuroimaging based on SpaRSA solver

Project description

BrainOwl

This is a classifier tuned for neuroimaging. In particular for task-related fMRI. It is meant to be used with the OWL norm (also called Ordered $l_1$ norm) and uses a solver based on SpaRSA.

The OWL norm should identify features relevant for the learning problem, even if they are correlated. Weight maps based on OWL tend to be sparse, but not so sparse like the solutions from LASSO, for example.

Install

Install using pip (be sure you have python3>=3.5):

pip install brainowl

And done.

If you want to have the source code, you can clone the repository using git:

git clone https://github.com/jpvaldes/brainowl.git

and then install it:

cd brainowl
pip install -e .

Usage example

The included Jupyter notebook contains an example usage of the BrainOwl classifier showing how to decode two categories of the classic neuroimaging Haxby dataset.

The dataset will be downloaded automatically if it is not found.

Acknowledgments

The scikit-learn library for making it easier to develop new ideas, the pyowl implementation, and the nilearn project (in particular, the SpaceNet learners).

This project contains code from pyowl.

References

X Zeng, M A T Figueiredo, The Ordered Weighted $l_1$ Norm:
Atomic Formulation, Projections, and Algorithms.

J. Bogdan, E. Berg, W. Su, and E. Candes, Statistical Estimation and
Testing via the Ordered $l_1$ Norm.

Stephen Wright, Robert Nowak, and Mario Figueiredo. Sparse Reconstruction
by Separable Approximation. IEEE Transactions on Signal Processing, 2009,
Vol. 52, No. 7, 2479-2493.

Marcos Raydan. The Barzilai and Borwein Gradient Method for the Large Scale
Unconstrained Minimization Problem. SIAM J. Optim., 1997, Vol. 7, No. 1,
26-33.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

brainowl-0.1.tar.gz (9.2 kB view hashes)

Uploaded Source

Built Distribution

brainowl-0.1-py3-none-any.whl (10.0 kB view hashes)

Uploaded Python 3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page