A Folded Architecture for Hardware Implementation of a Neural Structure Using Izhikevich Model

Serhat Çağdaş; Neslihan Serap Şengör

doi:10.1007/978-3-031-15934-3_42

A Folded Architecture for Hardware Implementation of a Neural Structure Using Izhikevich Model

Serhat Çağdaş^*, Neslihan Serap Şengör

^*Corresponding author for this work

Department of Electronics and Communication Engineering

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

1 Citation (Scopus)

Abstract

Neuromorphic systems are expected to equip a new paradigm in computation so that energy efficient, intelligent systems could be implemented easily. One way of fulfilling this aim is to design processes with Spiking Neural Networks (SNN). Here, we introduce an architecture to realize Izhikevich neuron model which ease the hardware implementation of large scale neural models. By using a folding method, we ensure that multiple operations of the same type are performed by one computing unit in a time multiplexed manner. In this way, we have achieved a design that uses hardware resources more efficiently, especially by saving multiplication, and allows more neurons to be implemented on the hardware. Finally, this architecture eliminates the necessity to allocate additional resources for implementing the synaptic dynamics of the neurons. Also, to present the effectiveness of the proposed architecture, a simple cerebellar granular layer structure is implemented on FPGA.

Original language	English
Title of host publication	Artificial Neural Networks and Machine Learning – ICANN 2022 - 31st International Conference on Artificial Neural Networks, 2022, Proceedings
Editors	Elias Pimenidis, Plamen Angelov, Chrisina Jayne, Antonios Papaleonidas, Mehmet Aydin
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	508-518
Number of pages	11
ISBN (Print)	9783031159336
DOIs	https://doi.org/10.1007/978-3-031-15934-3_42
Publication status	Published - 2022
Event	31st International Conference on Artificial Neural Networks, ICANN 2022 - Bristol, United Kingdom Duration: 6 Sept 2022 → 9 Sept 2022

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	13531 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	31st International Conference on Artificial Neural Networks, ICANN 2022
Country/Territory	United Kingdom
City	Bristol
Period	6/09/22 → 9/09/22

Bibliographical note

Publisher Copyright:
© 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.

Keywords

FPGA
Folded architecture
Izhikevich
Neuromorphic circuits
SNN

Access to Document

10.1007/978-3-031-15934-3_42

Cite this

Çağdaş, S., & Şengör, N. S. (2022). A Folded Architecture for Hardware Implementation of a Neural Structure Using Izhikevich Model. In E. Pimenidis, P. Angelov, C. Jayne, A. Papaleonidas, & M. Aydin (Eds.), Artificial Neural Networks and Machine Learning – ICANN 2022 - 31st International Conference on Artificial Neural Networks, 2022, Proceedings (pp. 508-518). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 13531 LNCS). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-031-15934-3_42

Çağdaş, Serhat ; Şengör, Neslihan Serap. / A Folded Architecture for Hardware Implementation of a Neural Structure Using Izhikevich Model. Artificial Neural Networks and Machine Learning – ICANN 2022 - 31st International Conference on Artificial Neural Networks, 2022, Proceedings. editor / Elias Pimenidis ; Plamen Angelov ; Chrisina Jayne ; Antonios Papaleonidas ; Mehmet Aydin. Springer Science and Business Media Deutschland GmbH, 2022. pp. 508-518 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{b7d823f27060404ea490fdd1683a0793,

title = "A Folded Architecture for Hardware Implementation of a Neural Structure Using Izhikevich Model",

abstract = "Neuromorphic systems are expected to equip a new paradigm in computation so that energy efficient, intelligent systems could be implemented easily. One way of fulfilling this aim is to design processes with Spiking Neural Networks (SNN). Here, we introduce an architecture to realize Izhikevich neuron model which ease the hardware implementation of large scale neural models. By using a folding method, we ensure that multiple operations of the same type are performed by one computing unit in a time multiplexed manner. In this way, we have achieved a design that uses hardware resources more efficiently, especially by saving multiplication, and allows more neurons to be implemented on the hardware. Finally, this architecture eliminates the necessity to allocate additional resources for implementing the synaptic dynamics of the neurons. Also, to present the effectiveness of the proposed architecture, a simple cerebellar granular layer structure is implemented on FPGA.",

keywords = "FPGA, Folded architecture, Izhikevich, Neuromorphic circuits, SNN",

author = "Serhat {\c C}ağda{\c s} and {\c S}eng{\"o}r, {Neslihan Serap}",

note = "Publisher Copyright: {\textcopyright} 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.; 31st International Conference on Artificial Neural Networks, ICANN 2022 ; Conference date: 06-09-2022 Through 09-09-2022",

year = "2022",

doi = "10.1007/978-3-031-15934-3_42",

language = "English",

isbn = "9783031159336",

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

publisher = "Springer Science and Business Media Deutschland GmbH",

pages = "508--518",

editor = "Elias Pimenidis and Plamen Angelov and Chrisina Jayne and Antonios Papaleonidas and Mehmet Aydin",

booktitle = "Artificial Neural Networks and Machine Learning – ICANN 2022 - 31st International Conference on Artificial Neural Networks, 2022, Proceedings",

address = "Germany",

}

Çağdaş, S & Şengör, NS 2022, A Folded Architecture for Hardware Implementation of a Neural Structure Using Izhikevich Model. in E Pimenidis, P Angelov, C Jayne, A Papaleonidas & M Aydin (eds), Artificial Neural Networks and Machine Learning – ICANN 2022 - 31st International Conference on Artificial Neural Networks, 2022, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 13531 LNCS, Springer Science and Business Media Deutschland GmbH, pp. 508-518, 31st International Conference on Artificial Neural Networks, ICANN 2022, Bristol, United Kingdom, 6/09/22. https://doi.org/10.1007/978-3-031-15934-3_42

A Folded Architecture for Hardware Implementation of a Neural Structure Using Izhikevich Model. / Çağdaş, Serhat; Şengör, Neslihan Serap.
Artificial Neural Networks and Machine Learning – ICANN 2022 - 31st International Conference on Artificial Neural Networks, 2022, Proceedings. ed. / Elias Pimenidis; Plamen Angelov; Chrisina Jayne; Antonios Papaleonidas; Mehmet Aydin. Springer Science and Business Media Deutschland GmbH, 2022. p. 508-518 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 13531 LNCS).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - A Folded Architecture for Hardware Implementation of a Neural Structure Using Izhikevich Model

AU - Çağdaş, Serhat

AU - Şengör, Neslihan Serap

PY - 2022

Y1 - 2022

N2 - Neuromorphic systems are expected to equip a new paradigm in computation so that energy efficient, intelligent systems could be implemented easily. One way of fulfilling this aim is to design processes with Spiking Neural Networks (SNN). Here, we introduce an architecture to realize Izhikevich neuron model which ease the hardware implementation of large scale neural models. By using a folding method, we ensure that multiple operations of the same type are performed by one computing unit in a time multiplexed manner. In this way, we have achieved a design that uses hardware resources more efficiently, especially by saving multiplication, and allows more neurons to be implemented on the hardware. Finally, this architecture eliminates the necessity to allocate additional resources for implementing the synaptic dynamics of the neurons. Also, to present the effectiveness of the proposed architecture, a simple cerebellar granular layer structure is implemented on FPGA.

AB - Neuromorphic systems are expected to equip a new paradigm in computation so that energy efficient, intelligent systems could be implemented easily. One way of fulfilling this aim is to design processes with Spiking Neural Networks (SNN). Here, we introduce an architecture to realize Izhikevich neuron model which ease the hardware implementation of large scale neural models. By using a folding method, we ensure that multiple operations of the same type are performed by one computing unit in a time multiplexed manner. In this way, we have achieved a design that uses hardware resources more efficiently, especially by saving multiplication, and allows more neurons to be implemented on the hardware. Finally, this architecture eliminates the necessity to allocate additional resources for implementing the synaptic dynamics of the neurons. Also, to present the effectiveness of the proposed architecture, a simple cerebellar granular layer structure is implemented on FPGA.

KW - FPGA

KW - Folded architecture

KW - Izhikevich

KW - Neuromorphic circuits

KW - SNN

UR - http://www.scopus.com/inward/record.url?scp=85138011621&partnerID=8YFLogxK

U2 - 10.1007/978-3-031-15934-3_42

DO - 10.1007/978-3-031-15934-3_42

M3 - Conference contribution

AN - SCOPUS:85138011621

SN - 9783031159336

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

SP - 508

EP - 518

BT - Artificial Neural Networks and Machine Learning – ICANN 2022 - 31st International Conference on Artificial Neural Networks, 2022, Proceedings

A2 - Pimenidis, Elias

A2 - Angelov, Plamen

A2 - Jayne, Chrisina

A2 - Papaleonidas, Antonios

A2 - Aydin, Mehmet

PB - Springer Science and Business Media Deutschland GmbH

T2 - 31st International Conference on Artificial Neural Networks, ICANN 2022

Y2 - 6 September 2022 through 9 September 2022

ER -

Çağdaş S, Şengör NS. A Folded Architecture for Hardware Implementation of a Neural Structure Using Izhikevich Model. In Pimenidis E, Angelov P, Jayne C, Papaleonidas A, Aydin M, editors, Artificial Neural Networks and Machine Learning – ICANN 2022 - 31st International Conference on Artificial Neural Networks, 2022, Proceedings. Springer Science and Business Media Deutschland GmbH. 2022. p. 508-518. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-031-15934-3_42