My name is Jason Hénault. I graduated with a Research Master’s degree in Robotics from the University of Clermont-Ferrand (France). I’m interested in domains linking engineering sciences and biology, like neurons modelization, DNA mining, EEG interfaces, etc.
I have studied deep learning during my first year of Master through CNN applied to road images, thanks to Caffe. I worked with a team of two Ph.D. students, an intern student and my supervisor, Prof Thierry Chateau, co-director of the unit COMSEE. That made me face many challenges, both scientific and technical. In that context, I learned how neural networks with deep layers work. I applied these technics on image sets (which I have supervised the creation and given the necessary instructions). In the end, I concluded by writing a master thesis and a team meeting to explain how deep learning works.
My master second year's internship was done at the neurocybernetics unit of ETIS lab at Cergy (France). That team works on bio-mimetics brain structures in order to explore a wide spectrum of artificial intelligence and cognition. Under the supevision of Dr. Nicolas Cuperlier and Dr. Marwen Belkaid, I studied autonomous navigation via hippocampus and adjacent structures imitation. More precisely, it was about, thanks to the middleware design tool called « Promethe » (developed by the lab), improving localization scripts working on the principle of panorama recognition by re-identification of points of interest. To do this, experiments were done in a simulated environment via Webots (because of the difficulty to do experiments on distances of 1 km). Many hypotheses have been formulated and so, in addition to improvement and the proof of the working on big environment, a model based on transitions was formulated, opening up new perspectives.