Quantifying Robustness and Capacity of Reservoir Computers with Consistency Profiles

Thomas Lymburn; Thomas Jüngling; Michael Small

doi:10.1007/978-3-030-61616-8_36

Quantifying Robustness and Capacity of Reservoir Computers with Consistency Profiles

Thomas Lymburn, Thomas Jüngling, Michael Small

Research output: Chapter in Book/Conference paper › Conference paper

82 Downloads (Pure)

Abstract

We study the consistency property in reservoir computers with noise. Consistency quantifies the functional dependence of a driven dynamical system on its input via replica tests. We characterise the high-dimensional profile of consistency in typical reservoirs subject to intrinsic and measurement noise. An integral of the consistency is introduced to measure capacity and act as an effective size of the reservoir. We observe a scaling law in the dependency of the consistency capacity on the noise amplitude and reservoir size, and demonstrate how this measure of capacity explains performance.

Original language	English
Title of host publication	Artificial Neural Networks and Machine Learning – ICANN 2020 - 29th International Conference on Artificial Neural Networks, Proceedings
Editors	Igor Farkaš, Paolo Masulli, Stefan Wermter
Publisher	Springer Science + Business Media
Pages	447-458
Number of pages	12
ISBN (Print)	9783030616151
DOIs	https://doi.org/10.1007/978-3-030-61616-8_36
Publication status	Published - 2020
Event	29th International Conference on Artificial Neural Networks, ICANN 2020 - Bratislava, Slovakia Duration: 15 Sep 2020 → 18 Sep 2020

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	12397 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	29th International Conference on Artificial Neural Networks, ICANN 2020
Country	Slovakia
City	Bratislava
Period	15/09/20 → 18/09/20

Access to Document

10.1007/978-3-030-61616-8_36

Lymburn et al., 2020 Quantifying robustness and capacity of reservoir computers with consistency profilesAccepted author manuscript, 674 KB

Link to publication in Scopus

Fingerprint Dive into the research topics of 'Quantifying Robustness and Capacity of Reservoir Computers with Consistency Profiles'. Together they form a unique fingerprint.

1 Finished

Navigating tipping points in complex dynamical systems
Small, M., Lesterhuis, W., Bosco, A. & Zaitouny, A.
Australian Research Council
1/01/18 → 31/12/20
Project: Research

Cite this

Lymburn, T., Jüngling, T., & Small, M. (2020). Quantifying Robustness and Capacity of Reservoir Computers with Consistency Profiles. In I. Farkaš, P. Masulli, & S. Wermter (Eds.), Artificial Neural Networks and Machine Learning – ICANN 2020 - 29th International Conference on Artificial Neural Networks, Proceedings (pp. 447-458). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 12397 LNCS). Springer Science + Business Media. https://doi.org/10.1007/978-3-030-61616-8_36

Lymburn, Thomas ; Jüngling, Thomas ; Small, Michael. / Quantifying Robustness and Capacity of Reservoir Computers with Consistency Profiles. Artificial Neural Networks and Machine Learning – ICANN 2020 - 29th International Conference on Artificial Neural Networks, Proceedings. editor / Igor Farkaš ; Paolo Masulli ; Stefan Wermter. Springer Science + Business Media, 2020. pp. 447-458 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{b2542532b4a54a9296f64d7133da3f88,

title = "Quantifying Robustness and Capacity of Reservoir Computers with Consistency Profiles",

abstract = "We study the consistency property in reservoir computers with noise. Consistency quantifies the functional dependence of a driven dynamical system on its input via replica tests. We characterise the high-dimensional profile of consistency in typical reservoirs subject to intrinsic and measurement noise. An integral of the consistency is introduced to measure capacity and act as an effective size of the reservoir. We observe a scaling law in the dependency of the consistency capacity on the noise amplitude and reservoir size, and demonstrate how this measure of capacity explains performance.",

keywords = "Consistency correlation, Principal component analysis, Reservoir computing, Signal-to-noise ratio",

author = "Thomas Lymburn and Thomas J{\"u}ngling and Michael Small",

year = "2020",

doi = "10.1007/978-3-030-61616-8_36",

language = "English",

isbn = "9783030616151",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Science + Business Media",

pages = "447--458",

editor = "Igor Farka{\v s} and Paolo Masulli and Stefan Wermter",

booktitle = "Artificial Neural Networks and Machine Learning – ICANN 2020 - 29th International Conference on Artificial Neural Networks, Proceedings",

address = "United States",

note = "29th International Conference on Artificial Neural Networks, ICANN 2020 ; Conference date: 15-09-2020 Through 18-09-2020",

}

Lymburn, T, Jüngling, T & Small, M 2020, Quantifying Robustness and Capacity of Reservoir Computers with Consistency Profiles. in I Farkaš, P Masulli & S Wermter (eds), Artificial Neural Networks and Machine Learning – ICANN 2020 - 29th International Conference on Artificial Neural Networks, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 12397 LNCS, Springer Science + Business Media, pp. 447-458, 29th International Conference on Artificial Neural Networks, ICANN 2020, Bratislava, Slovakia, 15/09/20. https://doi.org/10.1007/978-3-030-61616-8_36

Quantifying Robustness and Capacity of Reservoir Computers with Consistency Profiles. / Lymburn, Thomas; Jüngling, Thomas ; Small, Michael.

Artificial Neural Networks and Machine Learning – ICANN 2020 - 29th International Conference on Artificial Neural Networks, Proceedings. ed. / Igor Farkaš; Paolo Masulli; Stefan Wermter. Springer Science + Business Media, 2020. p. 447-458 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 12397 LNCS).

Research output: Chapter in Book/Conference paper › Conference paper

TY - GEN

T1 - Quantifying Robustness and Capacity of Reservoir Computers with Consistency Profiles

AU - Lymburn, Thomas

AU - Jüngling, Thomas

AU - Small, Michael

PY - 2020

Y1 - 2020

N2 - We study the consistency property in reservoir computers with noise. Consistency quantifies the functional dependence of a driven dynamical system on its input via replica tests. We characterise the high-dimensional profile of consistency in typical reservoirs subject to intrinsic and measurement noise. An integral of the consistency is introduced to measure capacity and act as an effective size of the reservoir. We observe a scaling law in the dependency of the consistency capacity on the noise amplitude and reservoir size, and demonstrate how this measure of capacity explains performance.

AB - We study the consistency property in reservoir computers with noise. Consistency quantifies the functional dependence of a driven dynamical system on its input via replica tests. We characterise the high-dimensional profile of consistency in typical reservoirs subject to intrinsic and measurement noise. An integral of the consistency is introduced to measure capacity and act as an effective size of the reservoir. We observe a scaling law in the dependency of the consistency capacity on the noise amplitude and reservoir size, and demonstrate how this measure of capacity explains performance.

KW - Consistency correlation

KW - Principal component analysis

KW - Reservoir computing

KW - Signal-to-noise ratio

UR - http://www.scopus.com/inward/record.url?scp=85094155340&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-61616-8_36

DO - 10.1007/978-3-030-61616-8_36

M3 - Conference paper

AN - SCOPUS:85094155340

SN - 9783030616151

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 447

EP - 458

BT - Artificial Neural Networks and Machine Learning – ICANN 2020 - 29th International Conference on Artificial Neural Networks, Proceedings

A2 - Farkaš, Igor

A2 - Masulli, Paolo

A2 - Wermter, Stefan

PB - Springer Science + Business Media

T2 - 29th International Conference on Artificial Neural Networks, ICANN 2020

Y2 - 15 September 2020 through 18 September 2020

ER -

Lymburn T, Jüngling T , Small M. Quantifying Robustness and Capacity of Reservoir Computers with Consistency Profiles. In Farkaš I, Masulli P, Wermter S, editors, Artificial Neural Networks and Machine Learning – ICANN 2020 - 29th International Conference on Artificial Neural Networks, Proceedings. Springer Science + Business Media. 2020. p. 447-458. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-030-61616-8_36

Quantifying Robustness and Capacity of Reservoir Computers with Consistency Profiles

Abstract

Publication series

Conference

Access to Document

Navigating tipping points in complex dynamical systems

Cite this