I am Clément Vignac, a PhD student at EPFL, Switzerland. I work in LTS4 under the supervision of Pascal Frossard. I study ways to make neural networks more data efficient by incorporating in our models prior knowledge about symmetries of a problem. I mostly focus on neural networks for sets and graphs, for which the symmetry group is the set of permutations of the elements.
Before joining EPFL, I graduated from École Polytechnique and École Normale Supérieure de Paris-Saclay (master MVA - Mathematics, Vision, Learning) in 2018. I completed my master thesis on structured prediction in LCSL, Genoa (Italy) under the supervision of Alessandro Rudi and Lorenzo Rosasco.
Clément Vignac, Andreas Loukas and Pascal Frossard - Building powerful and equivariant graph neural networks with structural message-passing (Neurips 2020)
For a full list of publications, look at my Google scholar page.