Academic Bibliography

 Search 200 years of publications by Ghent University researchers.
Advanced search
1 file | 1.19 MB
Add to list
  1. Search results
  2. Exploiting signal propagation delays ...

Exploiting signal propagation delays to match task memory requirements in reservoir computing

Stefan-Teodor Iacob (UGent) and Joni Dambre (UGent)
(2024) BIOMIMETICS. 9(6).
Author
Organization
Project
Abstract
Recurrent neural networks (RNNs) transmit information over time through recurrent connections. In contrast, biological neural networks use many other temporal processing mechanisms. One of these mechanisms is the inter-neuron delays caused by varying axon properties. Recently, this feature was implemented in echo state networks (ESNs), a type of RNN, by assigning spatial locations to neurons and introducing distance-dependent inter-neuron delays. These delays were shown to significantly improve ESN task performance. However, thus far, it is still unclear why distance-based delay networks (DDNs) perform better than ESNs. In this paper, we show that by optimizing inter-node delays, the memory capacity of the network matches the memory requirements of the task. As such, networks concentrate their memory capabilities to the points in the past which contain the most information for the task at hand. Moreover, we show that DDNs have a greater total linear memory capacity, with the same amount of non-linear processing power.
Keywords
distance-based delays, inter-neuron delays, echo state networks, recurrent neural networks, reservoir computing, memory capacity, PLASTICITY

Downloads

  • Download
    DS773.pdf
    • full text (Published version)
    • |
    • open access
    • |
    • PDF
    • |
    • 1.19 MB

Citation

Please use this url to cite or link to this publication:

MLA
Iacob, Stefan-Teodor, and Joni Dambre. “Exploiting Signal Propagation Delays to Match Task Memory Requirements in Reservoir Computing.” BIOMIMETICS, vol. 9, no. 6, 2024, doi:10.3390/biomimetics9060355.
APA
Iacob, S.-T., & Dambre, J. (2024). Exploiting signal propagation delays to match task memory requirements in reservoir computing. BIOMIMETICS, 9(6). https://doi.org/10.3390/biomimetics9060355
Chicago author-date
Iacob, Stefan-Teodor, and Joni Dambre. 2024. “Exploiting Signal Propagation Delays to Match Task Memory Requirements in Reservoir Computing.” BIOMIMETICS 9 (6). https://doi.org/10.3390/biomimetics9060355.
Chicago author-date (all authors)
Iacob, Stefan-Teodor, and Joni Dambre. 2024. “Exploiting Signal Propagation Delays to Match Task Memory Requirements in Reservoir Computing.” BIOMIMETICS 9 (6). doi:10.3390/biomimetics9060355.
Vancouver
1.
Iacob S-T, Dambre J. Exploiting signal propagation delays to match task memory requirements in reservoir computing. BIOMIMETICS. 2024;9(6).
IEEE
[1]
S.-T. Iacob and J. Dambre, “Exploiting signal propagation delays to match task memory requirements in reservoir computing,” BIOMIMETICS, vol. 9, no. 6, 2024.
@article{01J20WSP4V3YZQ404QBKW249B0,
  abstract     = {{Recurrent neural networks (RNNs) transmit information over time through recurrent connections. In contrast, biological neural networks use many other temporal processing mechanisms. One of these mechanisms is the inter-neuron delays caused by varying axon properties. Recently, this feature was implemented in echo state networks (ESNs), a type of RNN, by assigning spatial locations to neurons and introducing distance-dependent inter-neuron delays. These delays were shown to significantly improve ESN task performance. However, thus far, it is still unclear why distance-based delay networks (DDNs) perform better than ESNs. In this paper, we show that by optimizing inter-node delays, the memory capacity of the network matches the memory requirements of the task. As such, networks concentrate their memory capabilities to the points in the past which contain the most information for the task at hand. Moreover, we show that DDNs have a greater total linear memory capacity, with the same amount of non-linear processing power.}},
  articleno    = {{355}},
  author       = {{Iacob, Stefan-Teodor and Dambre, Joni}},
  issn         = {{2313-7673}},
  journal      = {{BIOMIMETICS}},
  keywords     = {{distance-based delays,inter-neuron delays,echo state networks,recurrent neural networks,reservoir computing,memory capacity,PLASTICITY}},
  language     = {{eng}},
  number       = {{6}},
  pages        = {{18}},
  title        = {{Exploiting signal propagation delays to match task memory requirements in reservoir computing}},
  url          = {{http://doi.org/10.3390/biomimetics9060355}},
  volume       = {{9}},
  year         = {{2024}},
}
Altmetric
View in Altmetric
Web of Science
Times cited: