At MIND, we believe that high-quality open-source scientific software is an important aspect of research in computational methods. For these reasons, we invest heavily in community-driven projects, some directly linked to Inria, and others living their own life. The different members of MIND are contributors to various open-source projects and we hire research engineers and programmers dedicated to the projects.
NiLearn is a Python package for fast and easy statistical learning on NeuroImaging data focusing on fMRI data. It leverages the scikit-learn Python toolbox for multivariate statistics with such as predictive modeling, classification, decoding, or connectivity analysis. Parietal is actively supporting the project with the full-time involvement of Kshitij Chawla.
.
NEUROLANG (todo Demian)
PySAP is a Python package related to sparsity and its application in astronomical or medical data analysis. This package binds the ‘sparse2d’ C++ library that allows sparse decomposition, denoising, and deconvolution.
.
MRI-NUFFT is a Python package that extends various NUFFT (Non-Uniform Fast Fourier Transform) python bindings used for MRI reconstruction. In particular, it provides a unified interface for all the methods, with extra features such as coil sensitivity, density compensated adjoint and off-resonance corrections (for static B0 inhomogeneities).
alphaCSC is a Python package to learn patterns from time series. It performs shift-invariant sparse dictionary learning, also known as convolutional sparse coding (CSC) to find repeating patterns linked to events in the signals.
.
.
MIND team members also contribute regularly to the Open Source projects they use daily, for instance:
Joblib is a set of tools to provide lightweight pipelining in Python. In particular, joblib offers (i) transparent disk-caching of the output values and lazy re-evaluation (memoize pattern). (ii) easy simple parallel computing. It is used internally by other libraries such as scikit-learn, nilearn and MNE-Python.
.
Benchopt is a benchmarking suite for optimization algorithms. It is built for simplicity, transparency, and reproducibility. It is implemented in Python but can run algorithms written in many programming languages.
Past collaborations
MIND team members have also been involved in the development of:
- OpenMEEG MEG/EEG forward problem.
- cogspaces multi-study decoding for fMRI.
- modl large-scale matrix factorization, with a sklearn API + fMRI usecases.
- MNE-Python processing and visualizing MEG and EEG data.
