This is an implementation of a structured sparsity regularization method with total variation and l2 constraints. It can be used to obtain sparse and smooth linear models.
The method is described in https://hal.inria.fr/hal-01167861/file/ConvexECML2015.pdf
To run the example in python:
-
Make sure you have installed scipy,numpy,sklearn, and nilearn python packages Note: these can all be easily installed with the python pip utility
-
Run the Synthetic_Example.py
-
Run the FMRI_Example.py this might take 5-10 minutes to complete.
The matlab example is very basic.
For any questions email eugene.belilovsky@inria.fr