To carry out their activities, Research Teams of the Frédéric Joliot Institute for Life Sciences have developed high-profile technological platforms in many areas : biomedical imaging, structural biology, metabolomics, High-Throughput screening, level 3 microbiological safety laboratory...
Within the Institute, the "Funding Research and Technology Transfer" team is at your disposal to identify the scientists and the skills you need to set up a joint project, to define the terms of a collaboration contract or study.
Whether you are an academic, a SME or an industrialist, our team informs and advices you about the possibilities of consortium assembly, technology transfer, patent licensing or use of our platforms.
The team is also at the disposal of the researchers of the institute to accompany them in achieving their valorization objectives.
Vincent Frouin is in charge of this project.
The use of machine-learning in neuroimaging offers new perspectives in early diagnosis and prognosis of brain diseases. Although such multivariate methods can capture complex relationships in the data, traditional approaches provide irregular (l2 penalty, see SVM in first row in the figure below) or scattered (l1 penalty, see Lasso in third row in the figure below) predictive pattern with a very limited relevance. A penalty like Total Variation (TV) that exploits the natural 3D structure of the images can increase the spatial coherence of the weight map (see SVM+TV and Lasso+TV in second and fourth rows in the figure below). The versatility of our framework authorizes its application to all kind of structured input data. Without any modification of the algorithm we have analysed the 2D surface-based, cortical thickness (see right columns if the figure). However, TV penalization leads to non-smooth optimization problems which disables classical gradient descent methods.
Within the BrainOmics team at NeuroSpin, we developed an optimization framework that minimizes any combination of l1, l2, and TV penalties while preserving the exact l1 penalty. This algorithm uses Nesterov's smoothing technique to approximate the TV penalty with a smooth function such that the loss and the penalties are minimized with an exact accelerated proximal gradient algorithm. We propose an original continuation algorithm that uses successively smaller values of the smoothing parameter to reach a prescribed precision while achieving the best possible convergence rate. This algorithm (CONESTA: COntinuation with NEsterov smoothing in a Shrinkage Thresholding Algorithm) can be used with other losses or penalties.
With The BrainOmics team, we produced a library (https://github.com/neurospin/pylearn-parsimony) based on the python language that implements many structured machine learning algorithms. The first stable release is scheduled by mid May 2014.
CEA is a French government-funded technological research organisation in four main areas: low-carbon energies, defense and security, information technologies and health technologies. A prominent player in the European Research Area, it is involved in setting up collaborative projects with many partners around the world.