Understanding Deep Learning

Towards the unification of deep neural networks and iterative proximal methods.

(Coming soon… )

Related Publications

1. D. Pinto-Suarez, J. Patracone and M. Sebban,

   Beyond Gaussian Initialization - Activation-Induced Gibbs States for Critical and Width-Efficient Networks,

   EurIPS Workshop - Foundations of Efficient Deep Learning (AXIOM), 2026.
   Article Page   PDF   BibTeX  
2. D. Pinto-Suarez, J. Patracone and M. Sebban,

   Principled Backpropagation-Free Training via Layerwise Fenchel-Young Lifting,

   preprint, 2026.
   Article Page   PDF   BibTeX  
3. D. Pinto-Suarez, J. Patracone and M. Sebban,

   Activation-Matched Gibbs Initialization for Neural Networks,

   preprint, 2026.
   Article Page   PDF   BibTeX  
4. V. Mitarchuk, A. Habrard, Q. Bertrand, J. Patracone, G. Gasso and R. Emonet,

   The Metric Matters: Certifying Generative Models with Validation and PAC-Bayes Bounds,

   preprint, 2026.
   Article Page   PDF   BibTeX  
5. B. Gao, J. Patracone and O. Alata,

   Cautious Adaptation under Persistent–Transient Ambiguity for Online Forecasting,

   preprint, 2026.
   Article Page   PDF   BibTeX  
6. J. Frecon, G. Gasso, M. Pontil and S. Salzo,

   Bregman Neural Networks,

   International Conference on Machine Learning (ICML), 2022.
   Article Page   PDF   BibTeX   Toolbox  

Back