Mehdi Khamassi
Research Director at Sorbonne Université, Sorbonne Center for Artificial Intelligence
Paris, Ile-de-France
Summary
👤
Senior🎓
Top SchoolMehdi Khamassi is a Research Director at CNRS in Paris who bridges cognitive science, neuroscience, AI and robotics to study decision-making and reinforcement learning in mammals and to translate those insights into more adaptive social robots. With over a decade of research leadership and visiting positions at institutions including Oxford, Trento and Montréal, he investigates neural mechanisms in prefrontal cortex, basal ganglia, hippocampus and dopamine systems to explain flexible adaptation and dysfunction in neuropsychiatric and neurodegenerative conditions. His lab applies biologically inspired algorithms to autonomous and social robotics, notably using small interactive robots in autism therapy to probe how social versus non-social reward processing shapes learning. He combines theoretical computational models, open-source algorithmic code and experimental work, making papers and resources openly available from his professional webpage. A less obvious strength is his focus on whether common reinforcement principles underlie social and non-social learning—an angle that drives both basic neuroscience hypotheses and practical advances in human-robot interaction.
11 years of coding experience
12 years of employment as a software developer
Habilitation (Entitlement) to Direct Researches (HDR), Biology, Doctor of Philosophy (PhD), Cognitive Sciences (Artificial Intelligence and Neurobiology), Master of Science (MSc), Cognitive Sciences - CogMaster, Master, Computer Science, Artificial Intelligence, Statistical Modelling, Maths Sup / Maths Spé, Mathematics and Physics, Habilitation (Entitlement) to Direct Researches (HDR), Biology, Doctor of Philosophy (PhD), Cognitive Sciences (Artificial Intelligence and Neurobiology), Master of Science (MSc), Cognitive Sciences - CogMaster, Master, Computer Science, Artificial Intelligence, Statistical Modelling, Maths Sup / Maths Spé, Mathematics and Physics at Université Pierre et Marie Curie (Paris VI)
Université Pierre et Marie Curie (Paris VI)
Université Pierre et Marie Curie (Paris VI)
Lycée Charlemagne
French, English, arabic (spoken), german (a little bit forgotten), italian (basics), japanese (basics), greek (basics)
Github Skills (8)
grid-world9
journal8
algorithms7
replication6
reinforcement-learning6
computational-science6
reproducible-science4
meta-learning2
Programming languages (2)
Github contributions (5)
MehdiKhamassi/SocialMetaLearning
Aug 2019 - May 2022
Code for active exploration in RL during HRI with a social reward function (human engagement). The model uses meta-learning for active exploration in a parameterized action space (combining discrete actions with continuous parameters of actions).
Contributions:2 PRs, 6 pushes, 2 branches in 2 years 9 months
meta-learning
MehdiKhamassi/RLwithReplay
Jun 2018 - Dec 2022
This is the source code to simulate model-based (MB) and model-free (MF) reinforcement learning algorithms with replays in grid worlds.
Contributions:23 commits, 1 PR, 20 pushes in 4 years 6 months
grid-worldreinforcement-learning
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.