Modelling of nanowire FETs based neural network for tactile pattern recognition in E-skin

William Taube; Fengyuan Liu; Anastasios Vilouras; Dhayalan Shakthivel; Carlos García Núñez; Hadi Heidari; Fabrice Labeau; Duncan Gregory; Ravinder Dahiya

doi:10.1109/BioCAS.2016.7833859

Modelling of nanowire FETs based neural network for tactile pattern recognition in E-skin

William Taube, Fengyuan Liu, Anastasios Vilouras, Dhayalan Shakthivel, Carlos García Núñez, Hadi Heidari, Fabrice Labeau, Duncan Gregory, Ravinder Dahiya

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

2 Citations (Scopus)

Abstract

This paper presents device, circuit and system modelling to validate the use of neural nanowire FETs (u-NWFETs) towards a hardware-realizable Neural Network. Hardware neural networks are promising for neuromorphic computing and have many prospective applications for bi-directional interface in prosthetics, and electroceuticals etc. Device simulation of a u-NWFET has been carried out followed by circuit implementation to validate the use of silicon nanowires (Si-NWs) as neuronal elements. A system level simulation of 258 neurons (225 sensor neurons, 50 hidden layer neurons and 3 output layer neurons) has been performed to demonstrate tactile pattern recognition. Training has been carried out and validation of the trained network gives an accurate classification of a database of 50 tactile images into 3 classifiers.

Original language	English
Title of host publication	2016 IEEE Biomedical Circuits and Systems Conference (BioCAS)
Publisher	IEEE
Pages	572-575
Number of pages	4
ISBN (Electronic)	978-1-5090-2959-4
ISBN (Print)	978-1-5090-2960-0
DOIs	https://doi.org/10.1109/BioCAS.2016.7833859
Publication status	Published - 26 Jan 2017
Externally published	Yes

Keywords

silicon neurons
silicon nanowires (Si-NWs)
artificial neural network (ANN)
physical neural networks (PNN)
pattern recognition

Access to Document

10.1109/BioCAS.2016.7833859

Cite this

@inproceedings{b94b53e3987a43879559a0f2ba8adaa8,

title = "Modelling of nanowire FETs based neural network for tactile pattern recognition in E-skin",

abstract = "This paper presents device, circuit and system modelling to validate the use of neural nanowire FETs (u-NWFETs) towards a hardware-realizable Neural Network. Hardware neural networks are promising for neuromorphic computing and have many prospective applications for bi-directional interface in prosthetics, and electroceuticals etc. Device simulation of a u-NWFET has been carried out followed by circuit implementation to validate the use of silicon nanowires (Si-NWs) as neuronal elements. A system level simulation of 258 neurons (225 sensor neurons, 50 hidden layer neurons and 3 output layer neurons) has been performed to demonstrate tactile pattern recognition. Training has been carried out and validation of the trained network gives an accurate classification of a database of 50 tactile images into 3 classifiers.",

keywords = "silicon neurons, silicon nanowires (Si-NWs), artificial neural network (ANN), physical neural networks (PNN), pattern recognition",

author = "William Taube and Fengyuan Liu and Anastasios Vilouras and Dhayalan Shakthivel and {Garc{\'i}a N{\'u}{\~n}ez}, Carlos and Hadi Heidari and Fabrice Labeau and Duncan Gregory and Ravinder Dahiya",

year = "2017",

month = jan,

day = "26",

doi = "10.1109/BioCAS.2016.7833859",

language = "English",

isbn = "978-1-5090-2960-0",

pages = "572--575",

booktitle = "2016 IEEE Biomedical Circuits and Systems Conference (BioCAS)",

publisher = "IEEE",

address = "United States",

}

TY - GEN

T1 - Modelling of nanowire FETs based neural network for tactile pattern recognition in E-skin

AU - Taube, William

AU - Liu, Fengyuan

AU - Vilouras, Anastasios

AU - Shakthivel, Dhayalan

AU - García Núñez, Carlos

AU - Heidari, Hadi

AU - Labeau, Fabrice

AU - Gregory, Duncan

AU - Dahiya, Ravinder

PY - 2017/1/26

Y1 - 2017/1/26

N2 - This paper presents device, circuit and system modelling to validate the use of neural nanowire FETs (u-NWFETs) towards a hardware-realizable Neural Network. Hardware neural networks are promising for neuromorphic computing and have many prospective applications for bi-directional interface in prosthetics, and electroceuticals etc. Device simulation of a u-NWFET has been carried out followed by circuit implementation to validate the use of silicon nanowires (Si-NWs) as neuronal elements. A system level simulation of 258 neurons (225 sensor neurons, 50 hidden layer neurons and 3 output layer neurons) has been performed to demonstrate tactile pattern recognition. Training has been carried out and validation of the trained network gives an accurate classification of a database of 50 tactile images into 3 classifiers.

AB - This paper presents device, circuit and system modelling to validate the use of neural nanowire FETs (u-NWFETs) towards a hardware-realizable Neural Network. Hardware neural networks are promising for neuromorphic computing and have many prospective applications for bi-directional interface in prosthetics, and electroceuticals etc. Device simulation of a u-NWFET has been carried out followed by circuit implementation to validate the use of silicon nanowires (Si-NWs) as neuronal elements. A system level simulation of 258 neurons (225 sensor neurons, 50 hidden layer neurons and 3 output layer neurons) has been performed to demonstrate tactile pattern recognition. Training has been carried out and validation of the trained network gives an accurate classification of a database of 50 tactile images into 3 classifiers.

KW - silicon neurons

KW - silicon nanowires (Si-NWs)

KW - artificial neural network (ANN)

KW - physical neural networks (PNN)

KW - pattern recognition

U2 - 10.1109/BioCAS.2016.7833859

DO - 10.1109/BioCAS.2016.7833859

M3 - Conference contribution

SN - 978-1-5090-2960-0

SP - 572

EP - 575

BT - 2016 IEEE Biomedical Circuits and Systems Conference (BioCAS)

PB - IEEE

ER -

Modelling of nanowire FETs based neural network for tactile pattern recognition in E-skin

Abstract

Keywords

Access to Document

Fingerprint

Cite this