Jérémy Scanvic
/ʒeʁemi skãvik/, /d͡ʒɛɹəmi skɑnvik/
Welcome here 👋 I’m a PhD candidate at ENSL and Hirsch working on the spatial robustness of vision networks. I’ve also designed a self-supervised learning method for image deconvolution and super-resolution combining tools from equivariant imaging and spectral analysis. My main interests are deep learning, image processing and spectral analysis. Don’t hesitate to reach out!
“Aren’t programs simpler than logical formulas? The answer is no. Logic is the formalization of everyday mathematics, and everyday mathematics is simpler than programs.”
— Leslie Lamport (1994)
News
- Mar 11'26 - Release of UNet-AF on arXiv (!)
- Feb 2'26 - Scale-Equivariant Imaging accepted in IEEE TCI
- Jan 25'26 - Equivariant Splitting accepted to ICLR 2026
- Dec 8'25 - Release of the Equational Theories Project on arXiv
Posts
Projects