Samy Jelassi

PhD student, Princeton University

Email sjelassi(at)princeton.edu
Curriculum Vitae
Google Scholar


I am a final-year PhD student in the ORFE department at Princeton University. I am fortunate to be advised by Profs. Yuanzhi Li, Joan Bruna and Boris Hanin. Previously, I obtained a B.Sc. in computer science from Ecole Normale Superieure de Lyon in 2015 and a M.Sc. in applied math from Ecole Normale Superieure Paris-Saclay where I was advised by Prof. Francis Bach. During my PhD, I was fortunate to do research internships at Facebook AI Research (with Aaron Defazio), Deepmind (with Bernardo Avila Pires and Remi Munos) and Google Research (with Srinadh Bhojanapalli and Sashank Reddi).

My current primary interest is in understanding the abilities of attention models. During my PhD, I have focused on questions in the intersection between deep learning and optimization.



Publications

Vision Transformers provably learn spatial structure,
Samy Jelassi, Michael E. Sander, Yuanzhi Li.
Neural Information Processing Systems (NeurIPS), 2022.
[Link]

Dissecting adaptive methods in GANs,
Samy Jelassi, David Dobre, Arthur Mensch, Yuanzhi Li, Gauthier Gidel.
Preprint, 2022.
[Link]

Towards understanding how momentum improves generalization in deep learning,
Samy Jelassi, Yuanzhi Li.
International Conference on Machine Learning (ICML), 2022.
Oral presentation at ICML 2021 workshop "Overparameterization: Pitfalls & Opportunities".
[Link]

Depth separation beyond radial functions,
Luca Venturi, Samy Jelassi, Tristan Ozuch, Joan Bruna.
Journal of Machine Learning Research (JMLR).
[Link]

Adaptivity without Compromise: A Momentumized, Adaptive, Dual Averaged Gradient Method for Stochastic Optimization,
Aaron Defazio, Samy Jelassi.
Journal of Machine Learning Research (JMLR).
[Link] [Facebook announcement]

Auction learning as a two-player game,
Jad Rahme, Samy Jelassi, Matt Weinberg.
International Conference on Learning Representations (ICLR), 2021.
[Link] [Blog post]

A Permutation-Equivariant Neural Network Architecture For Auction Design,
Jad Rahme, Samy Jelassi, Joan Bruna, Matt Weinberg.
AAAI 2021.
[Link]

Extragradient with player sampling for faster Nash equilibrium finding,
Samy Jelassi, Carles Domingo-Enrich, Damien Scieur, Arthur Mensch, Joan Bruna.
International Conference on Machine Learning (ICML), 2020.
[Link]

A mean-field analysis of two-player zero-sum games,
Carles Domingo-Enrich, Samy Jelassi, Arthur Mensch, Grant M. Rotskoff, Joan Bruna.
Neural Information Processing Systems (NeurIPS), 2019.
[Link]

Towards closing the gap between the theory and practice of SVRG,
Othmane Sebbouh, Nidham Gazagnadou, Samy Jelassi, Francis Bach, Robert M. Gower.
Neural Information Processing Systems (NeurIPS), 2019.
[Link]

Global convergence of neuron birth-death dynamics,
Grant M. Rotskoff, Samy Jelassi, Joan Bruna, Eric Vanden-Eijnden.
International Conference on Machine Learning (ICML), 2019.
[Link]

Smoothed analysis of the low-rank approach for smooth semidefinite programs,
Thomas Pumir*, Samy Jelassi*, Nicolas Boumal.
Neural Information Processing Systems (NeurIPS), 2018.
Oral presentation (one of 30 among 1100 accepted papers).
[Link]


Academic Service

Teaching Experience