Weight partitioning for dynamic fixed-point neuromorphic computing systems

Yongshin Kang; Joon Sung Yang; Jaeyong Chung

doi:10.1109/TCAD.2018.2878167

Weight partitioning for dynamic fixed-point neuromorphic computing systems

Yongshin Kang, Joon Sung Yang, Jaeyong Chung

College of Engineering

Research output: Contribution to journal › Article › peer-review

3 Citations (Scopus)

Abstract

Neuromorphic computing systems consist of neurons and synapses with limited programmability, and neural networks are modified to be mapped for such a system. In order to map a perceptron with a large number of connections into a hardware neuron with a fixed, small number of synapses, it is decomposed into a tree of perceptrons, which substantially affects the neuron usage and predictive performance. In this paper, we propose two decomposition algorithms that take advantage of plastic connections and the dynamic scaling capability of neurons. One algorithm based on sorting considers the neuron usage first, and the other algorithm based on packing considers the predictive performance first. Our experimental results on two popular deep convolutional neural networks showed that the sorting-based algorithm substantially improved the accuracy at no cost of neurons compared to a previous work, and the packing-based algorithm improved it even further at a small cost of neurons.

Original language	English
Article number	8509191
Pages (from-to)	2167-2171
Number of pages	5
Journal	IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
Volume	38
Issue number	11
DOIs	https://doi.org/10.1109/TCAD.2018.2878167
Publication status	Published - 2019 Nov

Bibliographical note

Funding Information:
Manuscript received October 13, 2017; revised December 26, 2017; accepted February 6, 2018. Date of publication October 25, 2018; date of current version October 16, 2019. This work was supported in part by the Basic Science Research Program through the National Research Foundation of Korea funded by the Ministry of Education under Grant NRF-2017R1D1A1B03029103, and in part by the Institute for Information and Communications Technology Promotion funded by the Korea Government under Grant 1711073912. This paper was recommended by Associate Editor C. H. Chang.

Publisher Copyright:
© 1982-2012 IEEE.

All Science Journal Classification (ASJC) codes

Software
Computer Graphics and Computer-Aided Design
Electrical and Electronic Engineering

Access to Document

10.1109/TCAD.2018.2878167

Cite this

@article{bce689d545084c079854e701793dc0e9,

title = "Weight partitioning for dynamic fixed-point neuromorphic computing systems",

abstract = "Neuromorphic computing systems consist of neurons and synapses with limited programmability, and neural networks are modified to be mapped for such a system. In order to map a perceptron with a large number of connections into a hardware neuron with a fixed, small number of synapses, it is decomposed into a tree of perceptrons, which substantially affects the neuron usage and predictive performance. In this paper, we propose two decomposition algorithms that take advantage of plastic connections and the dynamic scaling capability of neurons. One algorithm based on sorting considers the neuron usage first, and the other algorithm based on packing considers the predictive performance first. Our experimental results on two popular deep convolutional neural networks showed that the sorting-based algorithm substantially improved the accuracy at no cost of neurons compared to a previous work, and the packing-based algorithm improved it even further at a small cost of neurons.",

author = "Yongshin Kang and Yang, {Joon Sung} and Jaeyong Chung",

note = "Funding Information: Manuscript received October 13, 2017; revised December 26, 2017; accepted February 6, 2018. Date of publication October 25, 2018; date of current version October 16, 2019. This work was supported in part by the Basic Science Research Program through the National Research Foundation of Korea funded by the Ministry of Education under Grant NRF-2017R1D1A1B03029103, and in part by the Institute for Information and Communications Technology Promotion funded by the Korea Government under Grant 1711073912. This paper was recommended by Associate Editor C. H. Chang. Publisher Copyright: {\textcopyright} 1982-2012 IEEE.",

year = "2019",

month = nov,

doi = "10.1109/TCAD.2018.2878167",

language = "English",

volume = "38",

pages = "2167--2171",

journal = "IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems",

issn = "0278-0070",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

number = "11",

}

TY - JOUR

T1 - Weight partitioning for dynamic fixed-point neuromorphic computing systems

AU - Kang, Yongshin

AU - Yang, Joon Sung

AU - Chung, Jaeyong

N1 - Funding Information: Manuscript received October 13, 2017; revised December 26, 2017; accepted February 6, 2018. Date of publication October 25, 2018; date of current version October 16, 2019. This work was supported in part by the Basic Science Research Program through the National Research Foundation of Korea funded by the Ministry of Education under Grant NRF-2017R1D1A1B03029103, and in part by the Institute for Information and Communications Technology Promotion funded by the Korea Government under Grant 1711073912. This paper was recommended by Associate Editor C. H. Chang. Publisher Copyright: © 1982-2012 IEEE.

PY - 2019/11

Y1 - 2019/11

N2 - Neuromorphic computing systems consist of neurons and synapses with limited programmability, and neural networks are modified to be mapped for such a system. In order to map a perceptron with a large number of connections into a hardware neuron with a fixed, small number of synapses, it is decomposed into a tree of perceptrons, which substantially affects the neuron usage and predictive performance. In this paper, we propose two decomposition algorithms that take advantage of plastic connections and the dynamic scaling capability of neurons. One algorithm based on sorting considers the neuron usage first, and the other algorithm based on packing considers the predictive performance first. Our experimental results on two popular deep convolutional neural networks showed that the sorting-based algorithm substantially improved the accuracy at no cost of neurons compared to a previous work, and the packing-based algorithm improved it even further at a small cost of neurons.

AB - Neuromorphic computing systems consist of neurons and synapses with limited programmability, and neural networks are modified to be mapped for such a system. In order to map a perceptron with a large number of connections into a hardware neuron with a fixed, small number of synapses, it is decomposed into a tree of perceptrons, which substantially affects the neuron usage and predictive performance. In this paper, we propose two decomposition algorithms that take advantage of plastic connections and the dynamic scaling capability of neurons. One algorithm based on sorting considers the neuron usage first, and the other algorithm based on packing considers the predictive performance first. Our experimental results on two popular deep convolutional neural networks showed that the sorting-based algorithm substantially improved the accuracy at no cost of neurons compared to a previous work, and the packing-based algorithm improved it even further at a small cost of neurons.

UR - http://www.scopus.com/inward/record.url?scp=85055690555&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85055690555&partnerID=8YFLogxK

U2 - 10.1109/TCAD.2018.2878167

DO - 10.1109/TCAD.2018.2878167

M3 - Article

AN - SCOPUS:85055690555

SN - 0278-0070

VL - 38

SP - 2167

EP - 2171

JO - IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems

JF - IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems

IS - 11

M1 - 8509191

ER -

Weight partitioning for dynamic fixed-point neuromorphic computing systems

Abstract

Bibliographical note

All Science Journal Classification (ASJC) codes

Access to Document

Other files and links

Fingerprint

Cite this