A Primal-Dual Algorithm for Image Reconstruction with Input-Convex Neural Network Regularizers
SIAM Journal on Imaging Sciences · 30.06.2026 · 19 (2) · 782-806
DOI: 10.1137/25m1751839 ↗Kurz gefasst
Fachartikel in der Zeitschrift SIAM Journal on Imaging Sciences, online erschienen am 23.04.2026. Verfasst von Matthias J. Ehrhardt und 2 weiteren Autor:innen. Die Arbeit nennt 67 Literaturangaben. Gefördert von: Engineering and Physical Sciences Research Council.
Abstract (Originalsprache)
We address the optimization problem in a data-driven variational reconstruction framework, where the regularizer is parametrized by an input-convex neural network. While gradient-based methods are commonly used to solve such problems, they struggle to effectively handle nonsmooth problems, which often leads to slow convergence. Moreover, the nested structure of the neural network complicates the application of standard nonsmooth optimization techniques, such as proximal algorithms. To overcome these challenges, we reformulate the problem and eliminate the network’s nested structure. By relating this reformulation to epigraphical projections of the activation functions, we transform the problem into a convex optimization problem that can be efficiently solved using a primal-dual algorithm. We also prove that this reformulation is equivalent to the original variational problem. Through experiments on several imaging tasks, we show that the proposed approach not only outperforms subgradient methods and even accelerated methods in the smooth setting but also facilitates the training of the regularizer itself.
Förderung
- Engineering and Physical Sciences Research Council
Zitieren
APA
Ehrhardt, M. J., Mukherjee, S., & Wong, H. S. (2026). A Primal-Dual Algorithm for Image Reconstruction with Input-Convex Neural Network Regularizers. SIAM Journal on Imaging Sciences, 19(2), 782-806. https://doi.org/10.1137/25m1751839
Vancouver
Ehrhardt MJ, Mukherjee S, Wong HS. A Primal-Dual Algorithm for Image Reconstruction with Input-Convex Neural Network Regularizers. SIAM Journal on Imaging Sciences. 2026;19(2):782-806. doi:10.1137/25m1751839
BibTeX
@article{Ehrhardt2026,
author = {Ehrhardt, Matthias J. and Mukherjee, Subhadip and Wong, Hok Shing},
title = {A Primal-Dual Algorithm for Image Reconstruction with Input-Convex Neural Network Regularizers},
journal = {SIAM Journal on Imaging Sciences},
year = {2026},
volume = {19},
number = {2},
pages = {782--806},
doi = {10.1137/25m1751839}
}
EUVIMED