Nicolas Berkouk

AI Scientific Expert - CNIL (French DPA) - AI Department

Research

Publications & Preprint related to Topology and Machine Learning

  1. Singular Value Representation: A New Graph Perspective On Neural Networks, with Dan Meller. Accepted to the conference AISTAT23.
  2. giotto-deep: A Python Package for Topological Deep Learning, with Raphael Reinauer and Matteo Caorsi. Published in The Journal of Open Source Software
  3. Persformer: A Transformer Architecture for Topological Machine Learning, with Raphael Reinauer and Matteo Caorsi. Submitted.

Publications & Preprint related to Applied Algebraic Topology

  1. One Diamond to Rule Them All: Old and new topics about zigzag, levelsets and extended persistence, with Luca Nyckees. Submitted.
  2. Persistence and the Sheaf-Function Correspondence. Submitted.
  3. Projected distances for multi-parameter persistence modules, with François Petit. Submitted.
  4. Level-sets Persistence and Sheaf Theory, with Grégory Ginot and Steve Oudot. Submitted.
  5. Ephemeral Persistence modules and Distance Comparisons, with François Petit. Published in Algebraic and Geometric Topology.
  6. Algebraic Homotopy Interleaving Distance. Published in the Proceedings of the International Conference on Geometric Science of Information 2021
  7. A Derived Isometry for Constructible Sheaves on R, with Grégory Ginot. Published in Advances in Mathematics.

Publications & Preprint related to Sociology

  1. Are competitive entrance exams neutral? Competition and sponsorship in access to the École polytechnique, with Pierre François. Published in Sociologie.