An analytical comparison of locally-connected reconfigurable neural network architectures using a c. Elegans locomotive model

Research output: Contribution to journalArticle

Abstract

The scale of modern neural networks is growing rapidly, with direct hardware implementations providing significant speed and energy improvements over their software counterparts. However, these hardware implementations frequently assume global connectivity between neurons and thus suffer from communication bottlenecks. Such issues are not found in biological neural networks. It should therefore be possible to develop new architectures to reduce the dependence on global communications by considering the connectivity of biological networks. This paper introduces two reconfigurable locally-connected architectures for implementing biologically inspired neural networks in real time. Both proposed architectures are validated using the segmented locomotive model of the C. elegans, performing a demonstration of forwards, backwards serpentine motion and coiling behaviours. Local connectivity is discovered to offer up to a 17.5× speed improvement over hybrid systems that use combinations of local and global infrastructure. Furthermore, the concept of locality of connections is considered in more detail, highlighting the importance of dimensionality when designing neuromorphic architectures. Convolutional Neural Networks are shown to map poorly to locally connected architectures despite their apparent local structure, and both the locality and dimensionality of new neural processing systems is demonstrated as a critical component for matching the function and efficiency seen in biological networks.

Original languageEnglish
Article number43
JournalComputers
Volume7
Issue number3
DOIs
Publication statusPublished - 1 Sep 2018

Keywords

  • Architecture
  • C. Elegans
  • FPGA
  • Neural-network
  • Neuromorphic
  • Reconfigurable

ASJC Scopus subject areas

  • Human-Computer Interaction
  • Computer Networks and Communications

Cite this

@article{c5c4f073d77745eda7633e2b44ebd536,
title = "An analytical comparison of locally-connected reconfigurable neural network architectures using a c. Elegans locomotive model",
abstract = "The scale of modern neural networks is growing rapidly, with direct hardware implementations providing significant speed and energy improvements over their software counterparts. However, these hardware implementations frequently assume global connectivity between neurons and thus suffer from communication bottlenecks. Such issues are not found in biological neural networks. It should therefore be possible to develop new architectures to reduce the dependence on global communications by considering the connectivity of biological networks. This paper introduces two reconfigurable locally-connected architectures for implementing biologically inspired neural networks in real time. Both proposed architectures are validated using the segmented locomotive model of the C. elegans, performing a demonstration of forwards, backwards serpentine motion and coiling behaviours. Local connectivity is discovered to offer up to a 17.5× speed improvement over hybrid systems that use combinations of local and global infrastructure. Furthermore, the concept of locality of connections is considered in more detail, highlighting the importance of dimensionality when designing neuromorphic architectures. Convolutional Neural Networks are shown to map poorly to locally connected architectures despite their apparent local structure, and both the locality and dimensionality of new neural processing systems is demonstrated as a critical component for matching the function and efficiency seen in biological networks.",
keywords = "Architecture, C. Elegans, FPGA, Neural-network, Neuromorphic, Reconfigurable",
author = "Jonathan Graham-Harper-Cater and Benjamin Metcalfe and Peter Wilson",
year = "2018",
month = "9",
day = "1",
doi = "10.3390/computers7030043",
language = "English",
volume = "7",
journal = "Computers",
issn = "2073-431X",
publisher = "MDPI",
number = "3",

}

TY - JOUR

T1 - An analytical comparison of locally-connected reconfigurable neural network architectures using a c. Elegans locomotive model

AU - Graham-Harper-Cater, Jonathan

AU - Metcalfe, Benjamin

AU - Wilson, Peter

PY - 2018/9/1

Y1 - 2018/9/1

N2 - The scale of modern neural networks is growing rapidly, with direct hardware implementations providing significant speed and energy improvements over their software counterparts. However, these hardware implementations frequently assume global connectivity between neurons and thus suffer from communication bottlenecks. Such issues are not found in biological neural networks. It should therefore be possible to develop new architectures to reduce the dependence on global communications by considering the connectivity of biological networks. This paper introduces two reconfigurable locally-connected architectures for implementing biologically inspired neural networks in real time. Both proposed architectures are validated using the segmented locomotive model of the C. elegans, performing a demonstration of forwards, backwards serpentine motion and coiling behaviours. Local connectivity is discovered to offer up to a 17.5× speed improvement over hybrid systems that use combinations of local and global infrastructure. Furthermore, the concept of locality of connections is considered in more detail, highlighting the importance of dimensionality when designing neuromorphic architectures. Convolutional Neural Networks are shown to map poorly to locally connected architectures despite their apparent local structure, and both the locality and dimensionality of new neural processing systems is demonstrated as a critical component for matching the function and efficiency seen in biological networks.

AB - The scale of modern neural networks is growing rapidly, with direct hardware implementations providing significant speed and energy improvements over their software counterparts. However, these hardware implementations frequently assume global connectivity between neurons and thus suffer from communication bottlenecks. Such issues are not found in biological neural networks. It should therefore be possible to develop new architectures to reduce the dependence on global communications by considering the connectivity of biological networks. This paper introduces two reconfigurable locally-connected architectures for implementing biologically inspired neural networks in real time. Both proposed architectures are validated using the segmented locomotive model of the C. elegans, performing a demonstration of forwards, backwards serpentine motion and coiling behaviours. Local connectivity is discovered to offer up to a 17.5× speed improvement over hybrid systems that use combinations of local and global infrastructure. Furthermore, the concept of locality of connections is considered in more detail, highlighting the importance of dimensionality when designing neuromorphic architectures. Convolutional Neural Networks are shown to map poorly to locally connected architectures despite their apparent local structure, and both the locality and dimensionality of new neural processing systems is demonstrated as a critical component for matching the function and efficiency seen in biological networks.

KW - Architecture

KW - C. Elegans

KW - FPGA

KW - Neural-network

KW - Neuromorphic

KW - Reconfigurable

UR - http://www.scopus.com/inward/record.url?scp=85056777484&partnerID=8YFLogxK

U2 - 10.3390/computers7030043

DO - 10.3390/computers7030043

M3 - Article

VL - 7

JO - Computers

JF - Computers

SN - 2073-431X

IS - 3

M1 - 43

ER -