Heterogeneous Activation Function Extraction for Training and Optimization of SNN Systems

Amir Zjajo; Sumeet Kumar; Rene Van Leuken

doi:10.1109/AICAS.2019.8771619

Heterogeneous Activation Function Extraction for Training and Optimization of SNN Systems

Amir Zjajo, Sumeet Kumar, Rene Van Leuken

Signal Processing Systems

Research output: Chapter in Book/Conference proceedings/Edited volume › Conference contribution › Scientific › peer-review

Abstract

Energy-efficiency and computation capability characteristics of analog/mixed-signal spiking neural networks offer capable platform for implementation of cognitive tasks on resource-limited embedded platforms. However, inherent mismatch in analog devices severely influence accuracy and reliability of the computing system. In this paper, we devise efficient algorithm for extracting of heterogeneous activation functions of analog hardware neurons as a set of constraints in an off-line training and optimization process, and examine how compensation of the mismatch effects influence synchronicity and information processing capabilities of the system.

Original language	English
Title of host publication	2019 IEEE International Conference on Artificial Intelligence Circuits and Systems (AICAS)
Publisher	IEEE
Pages	244-245
Number of pages	2
ISBN (Electronic)	978-1-5386-7884-8
ISBN (Print)	978-1-5386-7885-5
DOIs	https://doi.org/10.1109/AICAS.2019.8771619
Publication status	Published - 2019
Event	1st IEEE International Conference on Artificial Intelligence Circuits and Systems, AICAS 2019 - Hsinchu, Taiwan Duration: 18 Mar 2019 → 20 Mar 2019

Conference

Conference	1st IEEE International Conference on Artificial Intelligence Circuits and Systems, AICAS 2019
Country/Territory	Taiwan
City	Hsinchu
Period	18/03/19 → 20/03/19

Keywords

activation functions
deep neural network
extraction
mismatch
spiking neural network
training

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.1109/AICAS.2019.8771619

Cite this

@inproceedings{7bf409737fe14109a4149917c2861814,

title = "Heterogeneous Activation Function Extraction for Training and Optimization of SNN Systems",

abstract = "Energy-efficiency and computation capability characteristics of analog/mixed-signal spiking neural networks offer capable platform for implementation of cognitive tasks on resource-limited embedded platforms. However, inherent mismatch in analog devices severely influence accuracy and reliability of the computing system. In this paper, we devise efficient algorithm for extracting of heterogeneous activation functions of analog hardware neurons as a set of constraints in an off-line training and optimization process, and examine how compensation of the mismatch effects influence synchronicity and information processing capabilities of the system.",

keywords = "activation functions, deep neural network, extraction, mismatch, spiking neural network, training",

author = "Amir Zjajo and Sumeet Kumar and {Van Leuken}, Rene",

year = "2019",

doi = "10.1109/AICAS.2019.8771619",

language = "English",

isbn = "978-1-5386-7885-5",

pages = "244--245",

booktitle = "2019 IEEE International Conference on Artificial Intelligence Circuits and Systems (AICAS)",

publisher = "IEEE",

address = "United States",

note = "1st IEEE International Conference on Artificial Intelligence Circuits and Systems, AICAS 2019 ; Conference date: 18-03-2019 Through 20-03-2019",

}

Zjajo, A, Kumar, S & Van Leuken, R 2019, Heterogeneous Activation Function Extraction for Training and Optimization of SNN Systems. in 2019 IEEE International Conference on Artificial Intelligence Circuits and Systems (AICAS)., 8771619, IEEE, pp. 244-245, 1st IEEE International Conference on Artificial Intelligence Circuits and Systems, AICAS 2019, Hsinchu, Taiwan, 18/03/19. https://doi.org/10.1109/AICAS.2019.8771619

Heterogeneous Activation Function Extraction for Training and Optimization of SNN Systems. / Zjajo, Amir; Kumar, Sumeet; Van Leuken, Rene.
2019 IEEE International Conference on Artificial Intelligence Circuits and Systems (AICAS). IEEE, 2019. p. 244-245 8771619.

Research output: Chapter in Book/Conference proceedings/Edited volume › Conference contribution › Scientific › peer-review

TY - GEN

T1 - Heterogeneous Activation Function Extraction for Training and Optimization of SNN Systems

AU - Zjajo, Amir

AU - Kumar, Sumeet

AU - Van Leuken, Rene

PY - 2019

Y1 - 2019

N2 - Energy-efficiency and computation capability characteristics of analog/mixed-signal spiking neural networks offer capable platform for implementation of cognitive tasks on resource-limited embedded platforms. However, inherent mismatch in analog devices severely influence accuracy and reliability of the computing system. In this paper, we devise efficient algorithm for extracting of heterogeneous activation functions of analog hardware neurons as a set of constraints in an off-line training and optimization process, and examine how compensation of the mismatch effects influence synchronicity and information processing capabilities of the system.

AB - Energy-efficiency and computation capability characteristics of analog/mixed-signal spiking neural networks offer capable platform for implementation of cognitive tasks on resource-limited embedded platforms. However, inherent mismatch in analog devices severely influence accuracy and reliability of the computing system. In this paper, we devise efficient algorithm for extracting of heterogeneous activation functions of analog hardware neurons as a set of constraints in an off-line training and optimization process, and examine how compensation of the mismatch effects influence synchronicity and information processing capabilities of the system.

KW - activation functions

KW - deep neural network

KW - extraction

KW - mismatch

KW - spiking neural network

KW - training

UR - http://www.scopus.com/inward/record.url?scp=85070461228&partnerID=8YFLogxK

U2 - 10.1109/AICAS.2019.8771619

DO - 10.1109/AICAS.2019.8771619

M3 - Conference contribution

SN - 978-1-5386-7885-5

SP - 244

EP - 245

BT - 2019 IEEE International Conference on Artificial Intelligence Circuits and Systems (AICAS)

PB - IEEE

T2 - 1st IEEE International Conference on Artificial Intelligence Circuits and Systems, AICAS 2019

Y2 - 18 March 2019 through 20 March 2019

ER -

Heterogeneous Activation Function Extraction for Training and Optimization of SNN Systems

Abstract

Conference

Keywords

UN SDGs

Access to Document

Other files and links

Fingerprint

Cite this