The Laboratoire de Mathématiques de Lens (LML) at the University of Artois is a mathematics laboratory that is pioneering work on Artificial Intelligence, building on a track record in geometry, algebra, analysis and mathematical physics. The members of the team have been teaching deep learning for the last four years at the Institut Technologique Universitaire de Lens, run a working group on deep learning and supervise internships at different levels on the subject. They have produced a signature recognition system, a system that can detect heartbeats signals from videos of people, and are collaborating with two companies, HNL (using this latter system to predict emotions) and Unchartech (for work on the link between the generative adversarial networks (GANs) and reinforcement learning for modelling of time series).
Moreover, they have introduced a new technique enabling transfer for GANs, useful for situations where data or computing power is lacking. LML’s work within REVERT focuses mainly on mathematical/deep learning models for the prediction of NPH treatment responders and it involves careful analysis of the data produced by the project and an exploration of the existing neural network architectures to identify models that fit the best the problem of NPH diagnosis.