Research

Research interests

• Probability Theory: growth-fragmentation processes, population genetics

• Machine Learning Theory: deep learning, reinforcement learning, denoising diffusions

Publications and prepublications

Denoising diffusions:

1. Kernel-Smoothed Scores for Denoising Diffusion: A Bias-Variance Study

Franck Gabriel, François Ged, Maria Han Veiga, Emmanuel Schertzer

Under review, 2025

Reinforcement learning:

2. Matryoshka Policy Gradient for Entropy-Regularized RL: Convergence and Global Optimality

François Ged, Maria Han Veiga

Journal of Machine Learning Research, 2024

Deep learning:

3. The Loss Landscape of Shallow ReLU-like Neural Networks: Stationary Points, Saddle Escaping and Network Embedding
   Zhengqing Wu, Berfin Şimşek, François Ged

International Confererence on Learning Representation, 2025

4. Freeze and Chaos: NTK views on DNN Normalization, Checkerboard and Boundary Artifacts

Arthur Jacot, Franck Gabriel, François Ged, Clément Hongler

Mathematical and Scientific Machine Learning, 2022

5. Saddle-to-Saddle Dynamics in Deep Linear Networks: Small Initialization Training, Symmetry, and Sparsity

Arthur Jacot, François Ged, Berfin Şimşek, Clément Hongler, Franck Gabriel

submitted

6. Geometry of the loss landscape in overparameterized neural networks: Symmetries and invariances

Berfin Şimşek, François Ged, Arthur Jacot, Francesco Spadaro, Clément Hongler, Wulfram Gerstner, Johanni Brea

International Conference on Machine Learning, 2021

Probability theory:

7. Moran model with simultaneous strong and weak selections: convergence towards a Lambda-Wright-Fisher SDE

François Ged 

MathematicS In Action, 2023

8. Intrinsic Area Near the Origin for Self-Similar Growth-Fragmentations and Related Random Surfaces

François Ged

Annales de l'Institut Henri Poincaré, Probabilités et Statistiques, 2022

9. Profile of a Self-Similar Growth-Fragmentation

François Ged

Electronic Journal of Probability, 2019