Connectionists: PhD Position in artificial intelligence/machine learning with focus on Brain Computer Interface at CEA Grenoble, France

[Tetiana AKSENOVA 218551]

Applications are invited for the research position in the field of artificial intelligence/machine learning with the focus on motor ECoG-driven Brain Computer Interface (BCI) with multiple degrees of freedom, at CEA Grenoble, France, funded by an international mobility program.
Context
The thesis will be carried out within the frame of the multidisciplinary project "Brain Computer Interface" (BCI) at LETI-CLINATEC® which is located at the research technological center CEA (Commissariat à l'Energie Atomique et aux Energies Alternatives), Grenoble, in collaboration with CEA LETI - DTBS (Grenoble) and CEA LIST-DM2I (Paris-Saclay), France. The goal of the BCI project is the proof of concept that it is possible to control complex effectors thanks to brain activity decoding. Motor BCI raises the hope that limb mobility may be restored for severely motor-impaired patients to regain autonomy, providing them with control over orthosis or prostheses.
The BCI project at LETI-CLINATEC® is based on the recording of neuronal activity at the level of cerebral motor cortex thanks to the WIMAGINE® implant dedicated to record ElectroCorticoGrams, ECoG. The real-time ECoG signal processing allows controlling complex effectors with multiple degrees of freedom. A clinical research protocol « BCI and Tetraplegia » at CLINATEC® includes several tetraplegic subjects and is in progress.
CEA-LETI teams possess significant experience in motor ECoG-based BCI. A set of decoding algorithms were developed to solve the problem of stable ECoG signal decoding. They are published in articles, defended by patents, tested in preclinical experiments in ECoG non-human primate data, and in healthy subjects using MEG (MagnetoEncephaloGraph) experiments and, finally, in tetraplegic subject in the frame of the clinical research protocol providing large number of degrees of freedom (DoF) to control complex effector such as 4-limb exoskeleton. CEA-LIST-DM2I team possesses significant experience in the field of Artificial Intelligence non-conventional applications under constraints, computing power limitations and efficient model re-training in potentially non-stationary context.
Mission
The thesis is addressing an ambitious scientific breakthrough - taking neuroprosthetics out-of-the-lab. The objective is to explore challenges of motor BCI medical applications and to achieve neuroprosthetics control performance superior to the current state-of-art and sufficient for the medical use. To achieve objectives of robust and accurate BCI control of complex effectors with large number of DoF in natural environment without or with minimal assistance, innovative Artificial Intelligence (AI) methods will be developed and applied.
The missions of the PhD fellow will be:
-              to explore specific challenges of multi-limb motor BCI addressing real life tasks, taking an advantage of long-term clinical trial and bilateral implantation of chronic wireless ECoG recording devices.
-              to develop and to apply innovative decoding/control algorithms using AI approaches.
The candidate will be integrated into the signal processing team and will collaborate with a multidisciplinary team such as software and electrical engineers, biologists and medical doctors.
Candidate Profile:
MSc or equivalent with strong knowledge in Signal processing, Applied mathematics, Machine learning, Deep learning with skills in Matlab, C/C++, Python.
Eligibility
The thesis is supported by MSCA (Marie Sklodowska Curie Actions) from the EU framework Program for Research and Innovation Horizon 2020. The NUMERICS program is a new international PhD program launched by CEA in the field of scientific computing (please refer to: http://numerics.cea.fr/). We are looking for a candidate of excellence satisfying European mobility rules. To be eligible, applicants shall not have resided or carried out their main activity in France more than 12 months in the three years prior to the call deadline (mobility condition).
Candidates should contact the persons below by April 1st 2019.
Contacts
Dr. Tetiana AKSENOVA / Téléphone : 04 38 78 03 20 /  Email : tetiana.aksenova at cea.fr
Dr. Antoine SOULOUMIAC / Téléphone : 01 69 08 49 76/  Email :  antoine.souloumiac at cea.fr
References
-              Eliseyev, A., Mestais, C., Charvet, G., Sauter, F., Abroug, N., Arizumi, N., ... & Benabid, A. L. (2014, August). CLINATEC® BCI platform based on the ECoG-recording implant WIMAGINE® and the innovative signal-processing: Preclinical results. In Engineering in Medicine and Biology Society (EMBC), 2014 36th Annual International Conference of the IEEE (pp. 1222-1225). IEEE.
-              Mestais, C. S., Charvet, G., Sauter-Starace, F., Foerster, M., Ratel, D., & Benabid, A. L. (2015). WIMAGINE: Wireless 64-Channel ECoG Recording Implant for Long Term Clinical Applications. Neural Systems and Rehabilitation Engineering, IEEE Transactions on, 23(1), 10-21.
-              Eliseyev, A., Auboiroux, V., Costecalde, T., Langar, L., Charvet, G., Mestais, C., Aksenova, T & Benabid, A. L. (2017). Recursive Exponentially Weighted N-way Partial Least Squares Regression with Recursive-Validation of Hyper-Parameters in Brain-Computer Interface Applications. Scientific reports, 7(1), 16281.
-              Schaeffer, M. C., & Aksenova, T. (2016). Switching Markov decoders for asynchronous trajectory reconstruction from ECoG signals in monkeys for BCI applications. Journal of Physiology-Paris 110.4 (2016): 348-360.
-              Pinot, Rafael, Laurent Meunier, Alexandre Araujo, Hisashi Kashima, Florian Yger, Cédric Gouy-Pailler, and Jamal Atif. 2019. "Theoretical Evidence for Adversarial Robustness through Randomization: The Case of the Exponential Family." ArXiv:1902.01148 [Cs, Stat], February. http://arxiv.org/abs/1902.01148.
-              Andrey Besedin, Pierre Blanchart, Michel Crucianu, Marin Ferecatu: Evolutive Deep Models for Online Learning on Datastreams with no Storage. IOTSTREAMING at PKDD/ECML 2017
-              Rivet, Bertrand, Antoine Souloumiac, Virginie Attina, and Guillaume Gibert. 2009. "XDAWN Algorithm to Enhance Evoked Potentials: Application to Brain-computer Interface." IEEE Transactions on Biomedical Engineering 56 (8): 2035-2043.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/connectionists/attachments/20190222/9c86a7c9/attachment.html>