Abstract
This paper presents a parameterised family of neural ordinary differential equations (neuralODEs) that fit the steady-state system response in a least-squares sense. The family of neuralODEs is cast in the form of recurrent equilibrium networks (NodeRENs). One of the main advantages of the proposed approach is that it uses only linear least-squares optimisation tools. As such, the solution to the steady-state fitting problem is given in a closed-form expression. Furthermore, the NodeREN family leaves a subset of parameters free. This is useful for enforcing robustness or for fitting transient data in addition to steady-state data.
| Original language | English |
|---|---|
| Title of host publication | 2024 American Control Conference, ACC 2024 |
| Publisher | Institute of Electrical and Electronics Engineers |
| Pages | 3359-3364 |
| Number of pages | 6 |
| ISBN (Electronic) | 979-8-3503-8265-5 |
| DOIs | |
| Publication status | Published - 5 Sept 2024 |
| Externally published | Yes |
| Event | 2024 American Control Conference, ACC 2024 - Westin Harbour Castle, Toronto, Canada Duration: 8 Jul 2024 → 12 Jul 2024 https://acc2024.a2c2.org/ |
Conference
| Conference | 2024 American Control Conference, ACC 2024 |
|---|---|
| Abbreviated title | ACC 2024 |
| Country/Territory | Canada |
| City | Toronto |
| Period | 8/07/24 → 12/07/24 |
| Internet address |
Funding
This work has been partially supported by the European Union's Horizon 2020 Research and Innovation Programme under grant agreement No. 739551 (KIOS CoE); by the Italian Ministry for Research in the framework of the 2020 Program for Research Projects of National Interest (PRIN), Grant 2020RTWES4; and by the EPSRC grants EP/W005557 and EP/X033546.