Distance-based delays in echo state networks
- Author
- Stefan-Teodor Iacob (UGent) , Matthias Freiberger (UGent) and Joni Dambre (UGent)
- Organization
- Project
- Abstract
- Physical reservoir computing, a paradigm bearing the promise of energy-efficient high-performance computing, has raised much attention in recent years. We argue though, that the effect of signal propagation delay on reservoir task performance, one of the most central aspects of physical reservoirs, is still insufficiently understood in a more general learning context. Such physically imposed delay has been found to play a crucial role in some specific physical realizations, such as integrated photonic reservoirs. While delays at the readout layer and input of Echo State Networks (ESNs) have been successfully exploited before to improve performance, to our knowledge this feature has not been studied in a more general setting. We introduce inter-node delays, based on physical distances, into ESNs as model systems for physical reservoir computing. We propose a novel ESN design that includes variable signal delays along the connections between neurons, comparable to varying axon lengths in biological neural networks or varying length delay lines in physical systems. We study the impact of the resulting variable inter-node delays in this setup in comparison with conventional ESNs and find that incorporating variable delays significantly improves reservoir performance on the NARMA-10 benchmark task.
- Keywords
- Variable delays, Evolutionary algorithms, Bio-inspired computing, Echo state networks
Downloads
-
(...).pdf
- full text (Published version)
- |
- UGent only
- |
- |
- 976.41 KB
-
DS579 acc.pdf
- full text (Accepted manuscript)
- |
- open access
- |
- |
- 482.97 KB
Citation
Please use this url to cite or link to this publication: http://hdl.handle.net/1854/LU-01GJYSMWMAEH0FGKRTEJR0KNRB
- MLA
- Iacob, Stefan-Teodor, et al. “Distance-Based Delays in Echo State Networks.” Intelligent Data Engineering and Automated Learning (IDEAL 2022), edited by Hujun Yin et al., vol. 13756, Springer, 2022, pp. 211–22, doi:10.1007/978-3-031-21753-1_21.
- APA
- Iacob, S.-T., Freiberger, M., & Dambre, J. (2022). Distance-based delays in echo state networks. In H. Yin, D. Camacho, & P. Tino (Eds.), Intelligent Data Engineering and Automated Learning (IDEAL 2022) (Vol. 13756, pp. 211–222). https://doi.org/10.1007/978-3-031-21753-1_21
- Chicago author-date
- Iacob, Stefan-Teodor, Matthias Freiberger, and Joni Dambre. 2022. “Distance-Based Delays in Echo State Networks.” In Intelligent Data Engineering and Automated Learning (IDEAL 2022), edited by Hujun Yin, David Camacho, and Peter Tino, 13756:211–22. Springer. https://doi.org/10.1007/978-3-031-21753-1_21.
- Chicago author-date (all authors)
- Iacob, Stefan-Teodor, Matthias Freiberger, and Joni Dambre. 2022. “Distance-Based Delays in Echo State Networks.” In Intelligent Data Engineering and Automated Learning (IDEAL 2022), ed by. Hujun Yin, David Camacho, and Peter Tino, 13756:211–222. Springer. doi:10.1007/978-3-031-21753-1_21.
- Vancouver
- 1.Iacob S-T, Freiberger M, Dambre J. Distance-based delays in echo state networks. In: Yin H, Camacho D, Tino P, editors. Intelligent Data Engineering and Automated Learning (IDEAL 2022). Springer; 2022. p. 211–22.
- IEEE
- [1]S.-T. Iacob, M. Freiberger, and J. Dambre, “Distance-based delays in echo state networks,” in Intelligent Data Engineering and Automated Learning (IDEAL 2022), Manchester, UK, 2022, vol. 13756, pp. 211–222.
@inproceedings{01GJYSMWMAEH0FGKRTEJR0KNRB,
abstract = {{Physical reservoir computing, a paradigm bearing the promise of energy-efficient high-performance computing, has raised much attention in recent years. We argue though, that the effect of signal propagation delay on reservoir task performance, one of the most central aspects of physical reservoirs, is still insufficiently understood in a more general learning context. Such physically imposed delay has been found to play a crucial role in some specific physical realizations, such as integrated photonic reservoirs. While delays at the readout layer and input of Echo State Networks (ESNs) have been successfully exploited before to improve performance, to our knowledge this feature has not been studied in a more general setting. We introduce inter-node delays, based on physical distances, into ESNs as model systems for physical reservoir computing. We propose a novel ESN design that includes variable signal delays along the connections between neurons, comparable to varying axon lengths in biological neural networks or varying length delay lines in physical systems. We study the impact of the resulting variable inter-node delays in this setup in comparison with conventional ESNs and find that incorporating variable delays significantly improves reservoir performance on the NARMA-10 benchmark task.}},
author = {{Iacob, Stefan-Teodor and Freiberger, Matthias and Dambre, Joni}},
booktitle = {{Intelligent Data Engineering and Automated Learning (IDEAL 2022)}},
editor = {{Yin, Hujun and Camacho, David and Tino, Peter}},
isbn = {{9783031217524}},
issn = {{0302-9743}},
keywords = {{Variable delays,Evolutionary algorithms,Bio-inspired computing,Echo state networks}},
language = {{eng}},
location = {{Manchester, UK}},
pages = {{211--222}},
publisher = {{Springer}},
title = {{Distance-based delays in echo state networks}},
url = {{http://doi.org/10.1007/978-3-031-21753-1_21}},
volume = {{13756}},
year = {{2022}},
}
- Altmetric
- View in Altmetric
- Web of Science
- Times cited: