On the stability properties of Gated Recurrent Units neural networks
Systems and Control
2021-10-12 v6 Machine Learning
Systems and Control
Abstract
The goal of this paper is to provide sufficient conditions for guaranteeing the Input-to-State Stability (ISS) and the Incremental Input-to-State Stability ({\delta}ISS) of Gated Recurrent Units (GRUs) neural networks. These conditions, devised for both single-layer and multi-layer architectures, consist of nonlinear inequalities on network's weights. They can be employed to check the stability of trained networks, or can be enforced as constraints during the training procedure of a GRU. The resulting training procedure is tested on a Quadruple Tank nonlinear benchmark system, showing satisfactory modeling performances.
Keywords
Cite
@article{arxiv.2011.06806,
title = {On the stability properties of Gated Recurrent Units neural networks},
author = {Fabio Bonassi and Marcello Farina and Riccardo Scattolini},
journal= {arXiv preprint arXiv:2011.06806},
year = {2021}
}
Comments
Copyright 2021. This manuscript version is made available under the CC-BY-NC-ND 4.0