A Bregman Proximal Viewpoint on Neural Operators
Abdel-Rahim Mezidi, Jordan Patracone, Saverio Salzo, Amaury Habrard, Massimiliano Pontil, Rémi Emonet, Marc Sebban
Abstract
We present several advances on neural operators by viewing the action of operator layers as the minimizers of Bregman regularized optimization problems over Banach function spaces. The proposed framework allows interpreting the activation operators as Bregman proximity operators from dual to primal space. This novel viewpoint is general enough to recover classical neural operators as well as a new variant, coined Bregman neural operators, which includes the inverse activation operator and features the same expressivity of standard neural operators. Numerical experiments support the added benefits of the Bregman variant of Fourier neural operators for training deeper and more accurate models.
BibTeX
@inproceedings{
mezidi2025a,
title={A Bregman Proximal Viewpoint on Neural Operators},
author={Abdel-Rahim Mezidi and Jordan Patracone and Saverio Salzo and Amaury Habrard and Massimiliano Pontil and R{\'e}mi Emonet and Marc Sebban},
booktitle={Forty-second International Conference on Machine Learning},
year={2025},
url={https://openreview.net/forum?id=lzzPAQ1TxA}
}