Research interests
I am a researcher at Inria Scool and my research interest is to understand intelligence. As a computer scientist, my best bet is to try reproducing intelligence by having computer programs solve (a priori) complex tasks. So far, these programs are predominantly based on machine learning but my goal since joining Scool is to add elements of symbolic AI to the mix of predominantly continuous machine learning tools. Indeed, it seems to me that discretization is an important part of, if not intelligent behavior then at the very least, describing intelligent behavior.
In keywords, my work mostly centers around reinforcement learning, with elements of computer vision, Bayesian optimization and convex optimization. In terms of applications, I am interested in robotics and more recently on learning human understandable decision making policies.
Background and collaborations
As part of my PhD, I worked with my advisors Michèle Sebag and Marc Schoenauer on reducing the expertise requirements of policy learning algorithms, to allow uninitiated users teach robots new tasks. We did so by proposing a learning framework (Preference-based Reinforcement Learning) where the user gives binary feedback (better/worse) to trajectories demonstrated by the robot; hence reducing the role of the user to that of a mere critic.
After that, I joined TU Darmstadt as a postdoc in the lab of Jan Peters to work on robot learning, where I also had the pleasure of being advised by Gerhard Neumann. Together with the great team at Darmstadt we worked on various topics such as reinforcement learning, Bayesian optimization, deep RL or convex optimization, with robotics applications such as in-hand object manipulation or ball catching.
Before joining Inria, I also worked as a research fellow at Aalto University for 6 short but sweet months, collaborating with Joni Pajarinen, Alexander Ilin and Juho Kannala on object-centric image decomposition for and with reinforcement learning. This is still a work in progress and I hope to visit Finland soon again!
Prior to my PhD, I received a diploma in Computer Engineering from École Nationale Supérieure d'Informatique (Algiers, Algeria) and an MSc in Artificial Intelligence and Decision from Sorbonne Université (Paris, France).