Results of the Call for Projects “Mathematics of Deep Learning”

news |

Publié le

13 January 2026

Discover the winners of the “Deep Learning Mathematics” Call for Projects launched by PEPR IA.

Following the call for projects on “Mathematics of Deep Learning” launched in November 2024, the PEPR IA is pleased to announce the winning projects that have been selected.

After a review and evaluation phase conducted by an international jury, eight projects were selected for their quality, innovative nature, and impact on one of the main areas of focus of the PEPR IA: “the mathematical foundations of AI.”

Check out the winning projects:

Géné-Pi project : Generative artificial intelligence and mathematics
MacLeOD project : Machine learning on geometries and distributions
MadLearning project : Deep learning mathematics: from theory to practice
MAGICALL project : Mathematics of generative models: an interdisciplinary analysis of loss function landscapes
PERSNET project : PERsistent Structures in Neural NETworks: Topological and Statistical Approaches to Improve Deep Learning
PRODIGE-AI project : Probability, Random Matrix Theory, Geometry, and Generalization for Generative AI
TENSOR4ML project : Tensor methods for mastering modern machine learning
THEOREM project : Theory for more efficient generative models