I am Omar Chehab, a Ph.D. student in the Parietal team at Inria, Université Paris-Saclay, supervised by Alexandre Gramfort and Aapo Hyvärinen. I currently work on self-supervised learning from brain data.

More generally, I am interested in the inferential and dynamics-with-control viewpoints on learning and computation in the brain, and what they suggest about how the brain encodes and processes (visual, auditory, linguistic) stimuli. We use neuroimaging data as a proxy for the neural code, investigating its structure in a top-down approach, using unsupervised representation learning, or conversely emulating it using mechanistic biophysical models in a bottom-up approach.

I am also interested in optimization (sparse, structured, accelerated, robust) and crossing different formalisms (ex: dynamical systems, optimization, information theory) within Machine Learning.

I’ve previously interned at Facebook AI Research, ENS Paris, U of Toronto with Jeff Rosenthal, Inria Parietal and AXA France. I completed my graduate studies @ ENS Paris-Saclay and Institut Polytechnique de Paris (ENSTA) in Applied Mathematics, Machine Learning and Engineering.

My work is funded by European (ERC) and French (ANR) grants “Signal And Learning Applied To Brain Data” and “Bridging Artificial Intelligence and Neuroscience”. I am also affiliated to NeuroSpin.