Divergence-Free Neural Networks with Application to Image Denoising
Abstract
We introduce a resource-efficient neural network architecture with zero divergence by design, adapted for high-dimensional problems. Our method is directly applicable to image denoising, for which divergence-free estimators are particularly well-suited for self-supervised learning, in accordance with Stein's unbiased risk estimation theory. Comparisons of our parameterization on popular denoising datasets demonstrate that it retains sufficient expressivity to remain competitive with other divergence-based approaches, while outperforming its counterparts when the noise level is unknown and varies across the training data.
Cite
Text
Herbreteau and Meunier. "Divergence-Free Neural Networks with Application to Image Denoising." International Conference on Learning Representations, 2026.Markdown
[Herbreteau and Meunier. "Divergence-Free Neural Networks with Application to Image Denoising." International Conference on Learning Representations, 2026.](https://mlanthology.org/iclr/2026/herbreteau2026iclr-divergencefree/)BibTeX
@inproceedings{herbreteau2026iclr-divergencefree,
title = {{Divergence-Free Neural Networks with Application to Image Denoising}},
author = {Herbreteau, Sébastien and Meunier, Etienne},
booktitle = {International Conference on Learning Representations},
year = {2026},
url = {https://mlanthology.org/iclr/2026/herbreteau2026iclr-divergencefree/}
}