High-Performance ELM for Memory Constrained Edge Computing Devices with Metal Performance Shaders

Anton Akusok; Leonardo Espinosa Leal; Kaj-Mikael Björk; Amaury Lendasse

doi:10.1007/978-3-030-58989-9_9

High-Performance ELM for Memory Constrained Edge Computing Devices with Metal Performance Shaders

Anton Akusok, Leonardo Espinosa Leal, Kaj-Mikael Björk, Amaury Lendasse

Finance, Helsinki

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

Abstract

This paper proposes a block solution method for the Extreme Learning Machine. It combines the speed of a direct non-iterative solver with minimal memory requirements. The method is suitable for edge computing scenarios running on a mobile device with GPU acceleration. The implementation tested on the GPU of iPad Pro outperforms a laptop CPU, and trains a 19,000-neuron model using under one gigabyte of memory. It confirms the feasibility of Big Data analysis on modern mobile devices.

Original language	English
Title of host publication	Proceedings of ELM2019
Editors	Jiuwen Cao, Chi Man Vong, Yoan Miche, Amaury Lendasse
Place of Publication	Cham
Publisher	Springer
Publication date	2021
Pages	79-88
ISBN (Print)	978-3-030-58988-2, 978-3-030-59049-9
ISBN (Electronic)	978-3-030-58989-9
DOIs	https://doi.org/10.1007/978-3-030-58989-9_9
Publication status	Published - 2021
MoE publication type	A4 Article in conference proceedings

Publication series

Name	Proceedings in Adaptation, Learning and Optimization
Volume	14
ISSN (Print)	2363-6084
ISSN (Electronic)	2363-6092

Keywords

113 Computer and information sciences

Access to Document

10.1007/978-3-030-58989-9_9

OpenUrl availability

Full text

Cite this

Akusok, A., Leal, L. E., Björk, K.-M., & Lendasse, A. (2021). High-Performance ELM for Memory Constrained Edge Computing Devices with Metal Performance Shaders. In J. Cao, C. M. Vong, Y. Miche, & A. Lendasse (Eds.), Proceedings of ELM2019 (pp. 79-88). (Proceedings in Adaptation, Learning and Optimization; Vol. 14). Springer. https://doi.org/10.1007/978-3-030-58989-9_9

@inproceedings{1aaa98d59ce446cb8f0ffabc91af7973,

title = "High-Performance ELM for Memory Constrained Edge Computing Devices with Metal Performance Shaders",

abstract = "This paper proposes a block solution method for the Extreme Learning Machine. It combines the speed of a direct non-iterative solver with minimal memory requirements. The method is suitable for edge computing scenarios running on a mobile device with GPU acceleration. The implementation tested on the GPU of iPad Pro outperforms a laptop CPU, and trains a 19,000-neuron model using under one gigabyte of memory. It confirms the feasibility of Big Data analysis on modern mobile devices.",

keywords = "113 Computer and information sciences",

author = "Anton Akusok and Leal, {Leonardo Espinosa} and Kaj-Mikael Bj{\"o}rk and Amaury Lendasse",

year = "2021",

doi = "10.1007/978-3-030-58989-9_9",

language = "English",

isbn = "978-3-030-58988-2",

series = "Proceedings in Adaptation, Learning and Optimization",

publisher = "Springer",

pages = "79--88",

editor = "Jiuwen Cao and Vong, {Chi Man} and Yoan Miche and Amaury Lendasse",

booktitle = "Proceedings of ELM2019",

address = "International",

}

High-Performance ELM for Memory Constrained Edge Computing Devices with Metal Performance Shaders. / Akusok, Anton; Leal, Leonardo Espinosa; Björk, Kaj-Mikael et al.
Proceedings of ELM2019. ed. / Jiuwen Cao; Chi Man Vong; Yoan Miche; Amaury Lendasse. Cham: Springer, 2021. p. 79-88 (Proceedings in Adaptation, Learning and Optimization; Vol. 14).

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

TY - GEN

T1 - High-Performance ELM for Memory Constrained Edge Computing Devices with Metal Performance Shaders

AU - Akusok, Anton

AU - Leal, Leonardo Espinosa

AU - Björk, Kaj-Mikael

AU - Lendasse, Amaury

PY - 2021

Y1 - 2021

N2 - This paper proposes a block solution method for the Extreme Learning Machine. It combines the speed of a direct non-iterative solver with minimal memory requirements. The method is suitable for edge computing scenarios running on a mobile device with GPU acceleration. The implementation tested on the GPU of iPad Pro outperforms a laptop CPU, and trains a 19,000-neuron model using under one gigabyte of memory. It confirms the feasibility of Big Data analysis on modern mobile devices.

AB - This paper proposes a block solution method for the Extreme Learning Machine. It combines the speed of a direct non-iterative solver with minimal memory requirements. The method is suitable for edge computing scenarios running on a mobile device with GPU acceleration. The implementation tested on the GPU of iPad Pro outperforms a laptop CPU, and trains a 19,000-neuron model using under one gigabyte of memory. It confirms the feasibility of Big Data analysis on modern mobile devices.

KW - 113 Computer and information sciences

UR - https://www.mendeley.com/catalogue/63e91091-ebf1-3432-9ef0-4cfcc89e74e4/

U2 - 10.1007/978-3-030-58989-9_9

DO - 10.1007/978-3-030-58989-9_9

M3 - Conference contribution

SN - 978-3-030-58988-2

SN - 978-3-030-59049-9

T3 - Proceedings in Adaptation, Learning and Optimization

SP - 79

EP - 88

BT - Proceedings of ELM2019

A2 - Cao, Jiuwen

A2 - Vong, Chi Man

A2 - Miche, Yoan

A2 - Lendasse, Amaury

PB - Springer

CY - Cham

ER -

High-Performance ELM for Memory Constrained Edge Computing Devices with Metal Performance Shaders

Abstract

Publication series

Keywords

Access to Document

OpenUrl availability

Other files and links

Fingerprint

Cite this