About me

I am a PhD student in Applied Mathematics under the joint supervision of Eric Moulines, professor at the CMAP laboratory, Ecole Polytechnique, and Aymeric Dieuleveut, associate professor at the CMAP laboratory, Ecole Polytechnique.

I am working on Federated Learning, analyzing the trade-offs between data privacy, communication cost and data heterogeneity in the context of distributed learning. Here is a short presentation of Federated Learning where I try to give the key concepts.

I am graduated from Ensimag, Grenoble with a focus on Computer Science, Applied Mathematics and 3D Animation. I also had the opportunity to do a six-months academic exchange with the MIPT, Moscow to study Statistics, Machine Learning and Image recognition under the supervision of Vadim Strijov.

News!

12/2021: I am very proud to announce that my paper Preserved central model for faster bidirectional compression in distributed settings has been accepted at Neurips 2021! This paper has been written with Aymeric Dieuleveut (my Ph.D. supervisor at Polytechnique) in collaboration with Laetitia Kameni and Richard VIDAL from Accenture France. In this work, we developed a new algorithm - MCM - that tackles communication constraints in a Federated Learning framework.

10/2021: The Congress of Young Researchers in Applied Mathematics (CJC-MA) has ended and was a collective success! It was 3 high level plenary speakers, 39 scientific presentations, 32 posters and about 120 participants! The event was organized at École Polytechnique, and its goal was to promote exchanges between PhD and post-doc students from French laboratories. It was a great moment of exchange and sharing.

01/2021: As part of my Ph.D., I help with the organization of the CJC-MA 2021 (Congress of young research in Applied Mathematics) as President of the Organizing Committee. This congress will take place at Ecole Polytechnique in October 2021 and aim to gather young researchers from the whole of France.

06/2020: I released my first paper on Arxiv: Bidirectional compression in heterogeneous settings for federated learning with partial participation: tight convergence guarantees. This paper has been written with Aymeric Dieuleveut (my Ph.D. supervisor at Polytechnique) in collaboration with Laetitia Kameni and Richard VIDAL from Accenture France.

12/2019: I am starting my Ph.D at Ecole Polytechnique!