Quantization Robustness of Monotone Operator Equilibrium Networks
IEEE Control Systems Letters · 2026 · 10 · 1651-1656
DOI: 10.1109/lcsys.2026.3703783 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Control Systems Letters, 2026. Verfasst von James Li und 2 weiteren Autor:innen. Die Arbeit nennt 23 Literaturangaben.
Abstract (Originalsprache)
Monotone operator equilibrium networks (MonDEQs) are implicit models whose well-posedness and convergence rely on a single scalar margin computed from the weight matrix, which has made them attractive for learned controllers with closed-loop stability guarantees. When such models are deployed on low-precision hardware, weights are quantized, and these structural guarantees can be destroyed. We provide an explicit quantization-preservation certificate: a single computable check on the weight perturbation determines whether the quantized network retains existence, uniqueness, linear convergence, and a bounded displacement of its equilibrium. The same threshold certifies the backward solve as well as the forward solve, so it governs both training and inference. MNIST experiments confirm a phase transition at the predicted threshold: three- and four-bit post-training quantization diverge, while five-bit and above converge; quantization-aware training recovers provable convergence at four bits.
Zitieren
APA
Li, J., Leong, P. H. W., & Chaffey, T. (2026). Quantization Robustness of Monotone Operator Equilibrium Networks. IEEE Control Systems Letters, 10, 1651-1656. https://doi.org/10.1109/lcsys.2026.3703783
Vancouver
Li J, Leong PHW, Chaffey T. Quantization Robustness of Monotone Operator Equilibrium Networks. IEEE Control Systems Letters. 2026;10:1651-1656. doi:10.1109/lcsys.2026.3703783
BibTeX
@article{Li2026,
author = {Li, James and Leong, Philip H. W. and Chaffey, Thomas},
title = {Quantization Robustness of Monotone Operator Equilibrium Networks},
journal = {IEEE Control Systems Letters},
year = {2026},
volume = {10},
pages = {1651--1656},
doi = {10.1109/lcsys.2026.3703783}
}
EUVIMED