Automated Neural Network Accelerator Generation Framework for Multiple Neural Network Applications

Inho Lee, Seongmin Hong, Giha Ryu, Yongjun Park

Research output: Chapter in Book/Report/Conference proceedingConference contributionResearchpeer-review

Abstract

Neural networks are widely used in various applications, but general neural network accelerators support only one application at a time. Therefore, information for each application, such as synaptic weights and bias data, must be loaded quickly to use multiple neural network applications. Field-programmable gate array (FPGA)-based implementation has huge performance overhead owing to low data transmission bandwidth. In order to solve this problem, this paper presents an automated FPGA-based multi-neural network accelerator generation framework that can quickly support several applications by storing neural network application data in an on-chip memory inside the FPGA. To do this, we first design a shared custom hardware accelerator that can support rapid changes in multiple target neural network applications. Then, we introduce an automated multi-neural network accelerator generation framework that performs training, weight quantization, and neural accelerator synthesis.

Original languageEnglish
Title of host publicationProceedings of TENCON 2018 - 2018 IEEE Region 10 Conference
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages2287-2290
Number of pages4
ISBN (Electronic)9781538654576
DOIs
StatePublished - 2019 Feb 22
Event2018 IEEE Region 10 Conference, TENCON 2018 - Jeju, Korea, Republic of
Duration: 2018 Oct 282018 Oct 31

Publication series

NameIEEE Region 10 Annual International Conference, Proceedings/TENCON
Volume2018-October
ISSN (Print)2159-3442
ISSN (Electronic)2159-3450

Conference

Conference2018 IEEE Region 10 Conference, TENCON 2018
CountryKorea, Republic of
CityJeju
Period18/10/2818/10/31

Fingerprint

Particle accelerators
Neural networks
Field programmable gate arrays (FPGA)
Data communication systems
Hardware
Bandwidth
Data storage equipment

Keywords

  • FPGA
  • accelerator
  • neural network

Cite this

Lee, I., Hong, S., Ryu, G., & Park, Y. (2019). Automated Neural Network Accelerator Generation Framework for Multiple Neural Network Applications. In Proceedings of TENCON 2018 - 2018 IEEE Region 10 Conference (pp. 2287-2290). [8650190] (IEEE Region 10 Annual International Conference, Proceedings/TENCON; Vol. 2018-October). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/TENCON.2018.8650190
Lee, Inho ; Hong, Seongmin ; Ryu, Giha ; Park, Yongjun. / Automated Neural Network Accelerator Generation Framework for Multiple Neural Network Applications. Proceedings of TENCON 2018 - 2018 IEEE Region 10 Conference. Institute of Electrical and Electronics Engineers Inc., 2019. pp. 2287-2290 (IEEE Region 10 Annual International Conference, Proceedings/TENCON).
@inproceedings{e89708d24a1844c18985deb0f537595f,
title = "Automated Neural Network Accelerator Generation Framework for Multiple Neural Network Applications",
abstract = "Neural networks are widely used in various applications, but general neural network accelerators support only one application at a time. Therefore, information for each application, such as synaptic weights and bias data, must be loaded quickly to use multiple neural network applications. Field-programmable gate array (FPGA)-based implementation has huge performance overhead owing to low data transmission bandwidth. In order to solve this problem, this paper presents an automated FPGA-based multi-neural network accelerator generation framework that can quickly support several applications by storing neural network application data in an on-chip memory inside the FPGA. To do this, we first design a shared custom hardware accelerator that can support rapid changes in multiple target neural network applications. Then, we introduce an automated multi-neural network accelerator generation framework that performs training, weight quantization, and neural accelerator synthesis.",
keywords = "FPGA, accelerator, neural network",
author = "Inho Lee and Seongmin Hong and Giha Ryu and Yongjun Park",
year = "2019",
month = "2",
day = "22",
doi = "10.1109/TENCON.2018.8650190",
language = "English",
series = "IEEE Region 10 Annual International Conference, Proceedings/TENCON",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
pages = "2287--2290",
booktitle = "Proceedings of TENCON 2018 - 2018 IEEE Region 10 Conference",

}

Lee, I, Hong, S, Ryu, G & Park, Y 2019, Automated Neural Network Accelerator Generation Framework for Multiple Neural Network Applications. in Proceedings of TENCON 2018 - 2018 IEEE Region 10 Conference., 8650190, IEEE Region 10 Annual International Conference, Proceedings/TENCON, vol. 2018-October, Institute of Electrical and Electronics Engineers Inc., pp. 2287-2290, 2018 IEEE Region 10 Conference, TENCON 2018, Jeju, Korea, Republic of, 18/10/28. https://doi.org/10.1109/TENCON.2018.8650190

Automated Neural Network Accelerator Generation Framework for Multiple Neural Network Applications. / Lee, Inho; Hong, Seongmin; Ryu, Giha; Park, Yongjun.

Proceedings of TENCON 2018 - 2018 IEEE Region 10 Conference. Institute of Electrical and Electronics Engineers Inc., 2019. p. 2287-2290 8650190 (IEEE Region 10 Annual International Conference, Proceedings/TENCON; Vol. 2018-October).

Research output: Chapter in Book/Report/Conference proceedingConference contributionResearchpeer-review

TY - GEN

T1 - Automated Neural Network Accelerator Generation Framework for Multiple Neural Network Applications

AU - Lee, Inho

AU - Hong, Seongmin

AU - Ryu, Giha

AU - Park, Yongjun

PY - 2019/2/22

Y1 - 2019/2/22

N2 - Neural networks are widely used in various applications, but general neural network accelerators support only one application at a time. Therefore, information for each application, such as synaptic weights and bias data, must be loaded quickly to use multiple neural network applications. Field-programmable gate array (FPGA)-based implementation has huge performance overhead owing to low data transmission bandwidth. In order to solve this problem, this paper presents an automated FPGA-based multi-neural network accelerator generation framework that can quickly support several applications by storing neural network application data in an on-chip memory inside the FPGA. To do this, we first design a shared custom hardware accelerator that can support rapid changes in multiple target neural network applications. Then, we introduce an automated multi-neural network accelerator generation framework that performs training, weight quantization, and neural accelerator synthesis.

AB - Neural networks are widely used in various applications, but general neural network accelerators support only one application at a time. Therefore, information for each application, such as synaptic weights and bias data, must be loaded quickly to use multiple neural network applications. Field-programmable gate array (FPGA)-based implementation has huge performance overhead owing to low data transmission bandwidth. In order to solve this problem, this paper presents an automated FPGA-based multi-neural network accelerator generation framework that can quickly support several applications by storing neural network application data in an on-chip memory inside the FPGA. To do this, we first design a shared custom hardware accelerator that can support rapid changes in multiple target neural network applications. Then, we introduce an automated multi-neural network accelerator generation framework that performs training, weight quantization, and neural accelerator synthesis.

KW - FPGA

KW - accelerator

KW - neural network

UR - http://www.scopus.com/inward/record.url?scp=85063208887&partnerID=8YFLogxK

U2 - 10.1109/TENCON.2018.8650190

DO - 10.1109/TENCON.2018.8650190

M3 - Conference contribution

T3 - IEEE Region 10 Annual International Conference, Proceedings/TENCON

SP - 2287

EP - 2290

BT - Proceedings of TENCON 2018 - 2018 IEEE Region 10 Conference

PB - Institute of Electrical and Electronics Engineers Inc.

ER -

Lee I, Hong S, Ryu G, Park Y. Automated Neural Network Accelerator Generation Framework for Multiple Neural Network Applications. In Proceedings of TENCON 2018 - 2018 IEEE Region 10 Conference. Institute of Electrical and Electronics Engineers Inc. 2019. p. 2287-2290. 8650190. (IEEE Region 10 Annual International Conference, Proceedings/TENCON). https://doi.org/10.1109/TENCON.2018.8650190