Home page of Grégoire Sergeant-Perthuis

Email: gregoire[d.o.t]sergeant-perthuis[a.t]inria[d.o.t]fr, CV , Google Scholar .

I am a postdoctoral researcher @ INRIA team Ouragan which is a joint team with the team of combinatorics and optimization of the mathematics departement (IMJ-PRG) of Sorbonne Université.

Before I did my studies at Ecole Normale Supérieure (rue d'Ulm) as an "élève normalien" in Theoretical Physics and then in Mathematics.

I did my PhD in maths at Université Paris Cité under the supervision of Daniel Bennequin and I defended it in May 2022. My aim was to formalize the Critical Brain Hypothesis, a hypothesis inspired by statitical mechanics on how the brain functions coming from neurosciences, in order to derive new classes of models that are 'critical brains'. To do so I propose to use computational tool coming from algebraic topology to construct and describe objects of interest in Rigourous Statistical Mechanics.

I am interested in theoretical results on sampling and learning probabilistic models over hyper-graphs and their applications in bioinformatics and robotics. In ongoing work I implement an extension of the Belief propagation in that direction for inference on networks of proteins and work on theoretical results on it's long time behaviour. I also look at what happens when there is an additional notion of symmetry for these models over hyper-graphs and some applications to multi-agent POMDPs.

My PhD thesis Full text , Slides of Defence