Federated Learning with Spiking Neural Networks in Heterogeneous Systems

Sadia Anjum Tumpa; Sonali Singh; Md Fahim Faysal Khan; Mahmut Tylan Kandemir; Vijaykrishnan Narayanan; Chita R. Das

doi:10.1109/ISVLSI59464.2023.10238618

Federated Learning with Spiking Neural Networks in Heterogeneous Systems

Sadia Anjum Tumpa, Sonali Singh, Md Fahim Faysal Khan, Mahmut Tylan Kandemir, Vijaykrishnan Narayanan, Chita R. Das

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

5 Scopus citations

Abstract

With the advances in IoT and edge-computing, Federated Learning is ever more popular as it offers data privacy. Low-power spiking neural networks (SNN) are ideal candidates for local nodes in such federated setup. Most prior works assume that the participating nodes have uniform compute resources, which may not be practical. In this work, we propose a federated SNN learning framework for a realistic heterogeneous environment, consisting of nodes with diverse memory-compute capabilities through activation-checkpointing and time-skipping that offers 4times reduction in effective memory requirement for low-memory nodes while improving the accuracy upto 10% for non-independent and identically-distributed data.

Original language	English (US)
Title of host publication	2023 IEEE Computer Society Annual Symposium on VLSI, ISVLSI 2023 - Proceedings
Editors	Fernanda Kastensmidt, Ricardo Reis, Aida Todri-Sanial, Hai Li, Carolina Metzler
Publisher	IEEE Computer Society
ISBN (Electronic)	9798350327694
DOIs	https://doi.org/10.1109/ISVLSI59464.2023.10238618
State	Published - 2023
Event	26th IEEE Computer Society Annual Symposium on VLSI, ISVLSI 2023 - Foz do Iguacu, Brazil Duration: Jun 20 2023 → Jun 23 2023

Publication series

Name	Proceedings of IEEE Computer Society Annual Symposium on VLSI, ISVLSI
Volume	2023-June
ISSN (Print)	2159-3469
ISSN (Electronic)	2159-3477

Conference

Conference	26th IEEE Computer Society Annual Symposium on VLSI, ISVLSI 2023
Country/Territory	Brazil
City	Foz do Iguacu
Period	6/20/23 → 6/23/23

All Science Journal Classification (ASJC) codes

Hardware and Architecture
Control and Systems Engineering
Electrical and Electronic Engineering

Access to Document

10.1109/ISVLSI59464.2023.10238618

Cite this

Tumpa, S. A., Singh, S., Khan, M. F. F., Kandemir, M. T., Narayanan, V., & Das, C. R. (2023). Federated Learning with Spiking Neural Networks in Heterogeneous Systems. In F. Kastensmidt, R. Reis, A. Todri-Sanial, H. Li, & C. Metzler (Eds.), 2023 IEEE Computer Society Annual Symposium on VLSI, ISVLSI 2023 - Proceedings (Proceedings of IEEE Computer Society Annual Symposium on VLSI, ISVLSI; Vol. 2023-June). IEEE Computer Society. https://doi.org/10.1109/ISVLSI59464.2023.10238618

Tumpa, Sadia Anjum ; Singh, Sonali ; Khan, Md Fahim Faysal et al. / Federated Learning with Spiking Neural Networks in Heterogeneous Systems. 2023 IEEE Computer Society Annual Symposium on VLSI, ISVLSI 2023 - Proceedings. editor / Fernanda Kastensmidt ; Ricardo Reis ; Aida Todri-Sanial ; Hai Li ; Carolina Metzler. IEEE Computer Society, 2023. (Proceedings of IEEE Computer Society Annual Symposium on VLSI, ISVLSI).

@inproceedings{f3e67dd9c03d4981ae3871a4653e01be,

title = "Federated Learning with Spiking Neural Networks in Heterogeneous Systems",

abstract = "With the advances in IoT and edge-computing, Federated Learning is ever more popular as it offers data privacy. Low-power spiking neural networks (SNN) are ideal candidates for local nodes in such federated setup. Most prior works assume that the participating nodes have uniform compute resources, which may not be practical. In this work, we propose a federated SNN learning framework for a realistic heterogeneous environment, consisting of nodes with diverse memory-compute capabilities through activation-checkpointing and time-skipping that offers 4times reduction in effective memory requirement for low-memory nodes while improving the accuracy upto 10% for non-independent and identically-distributed data.",

author = "Tumpa, {Sadia Anjum} and Sonali Singh and Khan, {Md Fahim Faysal} and Kandemir, {Mahmut Tylan} and Vijaykrishnan Narayanan and Das, {Chita R.}",

note = "Publisher Copyright: {\textcopyright} 2023 IEEE.; 26th IEEE Computer Society Annual Symposium on VLSI, ISVLSI 2023 ; Conference date: 20-06-2023 Through 23-06-2023",

year = "2023",

doi = "10.1109/ISVLSI59464.2023.10238618",

language = "English (US)",

series = "Proceedings of IEEE Computer Society Annual Symposium on VLSI, ISVLSI",

publisher = "IEEE Computer Society",

editor = "Fernanda Kastensmidt and Ricardo Reis and Aida Todri-Sanial and Hai Li and Carolina Metzler",

booktitle = "2023 IEEE Computer Society Annual Symposium on VLSI, ISVLSI 2023 - Proceedings",

address = "United States",

}

Tumpa, SA, Singh, S, Khan, MFF, Kandemir, MT , Narayanan, V & Das, CR 2023, Federated Learning with Spiking Neural Networks in Heterogeneous Systems. in F Kastensmidt, R Reis, A Todri-Sanial, H Li & C Metzler (eds), 2023 IEEE Computer Society Annual Symposium on VLSI, ISVLSI 2023 - Proceedings. Proceedings of IEEE Computer Society Annual Symposium on VLSI, ISVLSI, vol. 2023-June, IEEE Computer Society, 26th IEEE Computer Society Annual Symposium on VLSI, ISVLSI 2023, Foz do Iguacu, Brazil, 6/20/23. https://doi.org/10.1109/ISVLSI59464.2023.10238618

Federated Learning with Spiking Neural Networks in Heterogeneous Systems. / Tumpa, Sadia Anjum; Singh, Sonali; Khan, Md Fahim Faysal et al.
2023 IEEE Computer Society Annual Symposium on VLSI, ISVLSI 2023 - Proceedings. ed. / Fernanda Kastensmidt; Ricardo Reis; Aida Todri-Sanial; Hai Li; Carolina Metzler. IEEE Computer Society, 2023. (Proceedings of IEEE Computer Society Annual Symposium on VLSI, ISVLSI; Vol. 2023-June).

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

TY - GEN

T1 - Federated Learning with Spiking Neural Networks in Heterogeneous Systems

AU - Tumpa, Sadia Anjum

AU - Singh, Sonali

AU - Khan, Md Fahim Faysal

AU - Kandemir, Mahmut Tylan

AU - Narayanan, Vijaykrishnan

AU - Das, Chita R.

PY - 2023

Y1 - 2023

N2 - With the advances in IoT and edge-computing, Federated Learning is ever more popular as it offers data privacy. Low-power spiking neural networks (SNN) are ideal candidates for local nodes in such federated setup. Most prior works assume that the participating nodes have uniform compute resources, which may not be practical. In this work, we propose a federated SNN learning framework for a realistic heterogeneous environment, consisting of nodes with diverse memory-compute capabilities through activation-checkpointing and time-skipping that offers 4times reduction in effective memory requirement for low-memory nodes while improving the accuracy upto 10% for non-independent and identically-distributed data.

AB - With the advances in IoT and edge-computing, Federated Learning is ever more popular as it offers data privacy. Low-power spiking neural networks (SNN) are ideal candidates for local nodes in such federated setup. Most prior works assume that the participating nodes have uniform compute resources, which may not be practical. In this work, we propose a federated SNN learning framework for a realistic heterogeneous environment, consisting of nodes with diverse memory-compute capabilities through activation-checkpointing and time-skipping that offers 4times reduction in effective memory requirement for low-memory nodes while improving the accuracy upto 10% for non-independent and identically-distributed data.

UR - http://www.scopus.com/inward/record.url?scp=85172120060&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85172120060&partnerID=8YFLogxK

U2 - 10.1109/ISVLSI59464.2023.10238618

DO - 10.1109/ISVLSI59464.2023.10238618

M3 - Conference contribution

AN - SCOPUS:85172120060

T3 - Proceedings of IEEE Computer Society Annual Symposium on VLSI, ISVLSI

BT - 2023 IEEE Computer Society Annual Symposium on VLSI, ISVLSI 2023 - Proceedings

A2 - Kastensmidt, Fernanda

A2 - Reis, Ricardo

A2 - Todri-Sanial, Aida

A2 - Li, Hai

A2 - Metzler, Carolina

PB - IEEE Computer Society

T2 - 26th IEEE Computer Society Annual Symposium on VLSI, ISVLSI 2023

Y2 - 20 June 2023 through 23 June 2023

ER -

Tumpa SA, Singh S, Khan MFF, Kandemir MT , Narayanan V , Das CR. Federated Learning with Spiking Neural Networks in Heterogeneous Systems. In Kastensmidt F, Reis R, Todri-Sanial A, Li H, Metzler C, editors, 2023 IEEE Computer Society Annual Symposium on VLSI, ISVLSI 2023 - Proceedings. IEEE Computer Society. 2023. (Proceedings of IEEE Computer Society Annual Symposium on VLSI, ISVLSI). doi: 10.1109/ISVLSI59464.2023.10238618

Federated Learning with Spiking Neural Networks in Heterogeneous Systems

Abstract

Publication series

Conference

All Science Journal Classification (ASJC) codes

Access to Document

Other files and links

Fingerprint

Cite this