High performance relevance vector machine on HMPSoC

Yongfu He; Shaojun Wang; Yu Peng; Yeyong Pang; Ning Ma; Jingyue Pang

doi:10.1109/FPT.2014.7082812

High performance relevance vector machine on HMPSoC

Yongfu He^*
, Shaojun Wang
, Yu Peng
, Yeyong Pang
, Ning Ma
, Jingyue Pang

^*Corresponding author for this work

School of Electronics and Information Engineering

Harbin Institute of Technology

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

3 Scopus citations

Abstract

Relevance Vector Machine (RVM) with the uncertainty expressing ability has spawned broad applications in Prognostic and Health Management (PHM). However computationally intensive intrinsic nature of RVM greatly limits its usage. This paper presents a software and hardware co-design approach based on HMPSoC technology, which efficiently exploited sequential and parallel nature of RVM. Multi-channel and pipelined hardware architecture for the acceleration of kernel formulation and intermediate values calculation is proposed. The hardware that wrapped with AXI-Stream interface is integrated into HMPSoC as an acceleration engine. We implement the design on an on-board PHM prototype platform with a Xilinx Zynq XC7Z020 AP SoC. The experiment results show 5.3x and 46.8x speed up in terms of the time cost than the RVM running on PC with a Xeon 5620 processor and ARM Cortex A9 processor. The energy consumption is reduced by 153.0x and 37.3, respectively.

Original language	English
Title of host publication	Proceedings of the 2014 International Conference on Field-Programmable Technology, FPT 2014
Editors	Jialin Chen, Wenbo Yin, Yuichiro Shibata, Lingli Wang, Hayden Kwok-Hay So, Yuchun Ma
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	334-337
Number of pages	4
ISBN (Electronic)	9781479962457
DOIs	https://doi.org/10.1109/FPT.2014.7082812
State	Published - 2014
Event	13th International Conference on Field-Programmable Technology, FPT 2014 - Shanghai, China Duration: 10 Dec 2014 → 12 Dec 2014

Publication series

Name	Proceedings of the 2014 International Conference on Field-Programmable Technology, FPT 2014

Conference

Conference	13th International Conference on Field-Programmable Technology, FPT 2014
Country/Territory	China
City	Shanghai
Period	10/12/14 → 12/12/14

UN SDGs

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

SDG 7 Affordable and Clean Energy

Access to Document

10.1109/FPT.2014.7082812

Cite this

He, Y., Wang, S., Peng, Y., Pang, Y., Ma, N., & Pang, J. (2014). High performance relevance vector machine on HMPSoC. In J. Chen, W. Yin, Y. Shibata, L. Wang, H. K.-H. So, & Y. Ma (Eds.), Proceedings of the 2014 International Conference on Field-Programmable Technology, FPT 2014 (pp. 334-337). Article 7082812 (Proceedings of the 2014 International Conference on Field-Programmable Technology, FPT 2014). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/FPT.2014.7082812

He, Yongfu ; Wang, Shaojun ; Peng, Yu et al. / High performance relevance vector machine on HMPSoC. Proceedings of the 2014 International Conference on Field-Programmable Technology, FPT 2014. editor / Jialin Chen ; Wenbo Yin ; Yuichiro Shibata ; Lingli Wang ; Hayden Kwok-Hay So ; Yuchun Ma. Institute of Electrical and Electronics Engineers Inc., 2014. pp. 334-337 (Proceedings of the 2014 International Conference on Field-Programmable Technology, FPT 2014).

@inproceedings{03004a5ea0b94b1c953d4251f09e2952,

title = "High performance relevance vector machine on HMPSoC",

abstract = "Relevance Vector Machine (RVM) with the uncertainty expressing ability has spawned broad applications in Prognostic and Health Management (PHM). However computationally intensive intrinsic nature of RVM greatly limits its usage. This paper presents a software and hardware co-design approach based on HMPSoC technology, which efficiently exploited sequential and parallel nature of RVM. Multi-channel and pipelined hardware architecture for the acceleration of kernel formulation and intermediate values calculation is proposed. The hardware that wrapped with AXI-Stream interface is integrated into HMPSoC as an acceleration engine. We implement the design on an on-board PHM prototype platform with a Xilinx Zynq XC7Z020 AP SoC. The experiment results show 5.3x and 46.8x speed up in terms of the time cost than the RVM running on PC with a Xeon 5620 processor and ARM Cortex A9 processor. The energy consumption is reduced by 153.0x and 37.3, respectively.",

author = "Yongfu He and Shaojun Wang and Yu Peng and Yeyong Pang and Ning Ma and Jingyue Pang",

note = "Publisher Copyright: {\textcopyright} 2014 IEEE.; 13th International Conference on Field-Programmable Technology, FPT 2014 ; Conference date: 10-12-2014 Through 12-12-2014",

year = "2014",

doi = "10.1109/FPT.2014.7082812",

language = "英语",

series = "Proceedings of the 2014 International Conference on Field-Programmable Technology, FPT 2014",

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

pages = "334--337",

editor = "Jialin Chen and Wenbo Yin and Yuichiro Shibata and Lingli Wang and So, \{Hayden Kwok-Hay\} and Yuchun Ma",

booktitle = "Proceedings of the 2014 International Conference on Field-Programmable Technology, FPT 2014",

address = "美国",

}

He, Y, Wang, S, Peng, Y, Pang, Y, Ma, N & Pang, J 2014, High performance relevance vector machine on HMPSoC. in J Chen, W Yin, Y Shibata, L Wang, HK-H So & Y Ma (eds), Proceedings of the 2014 International Conference on Field-Programmable Technology, FPT 2014., 7082812, Proceedings of the 2014 International Conference on Field-Programmable Technology, FPT 2014, Institute of Electrical and Electronics Engineers Inc., pp. 334-337, 13th International Conference on Field-Programmable Technology, FPT 2014, Shanghai, China, 10/12/14. https://doi.org/10.1109/FPT.2014.7082812

High performance relevance vector machine on HMPSoC. / He, Yongfu; Wang, Shaojun; Peng, Yu et al.
Proceedings of the 2014 International Conference on Field-Programmable Technology, FPT 2014. ed. / Jialin Chen; Wenbo Yin; Yuichiro Shibata; Lingli Wang; Hayden Kwok-Hay So; Yuchun Ma. Institute of Electrical and Electronics Engineers Inc., 2014. p. 334-337 7082812 (Proceedings of the 2014 International Conference on Field-Programmable Technology, FPT 2014).

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

TY - GEN

T1 - High performance relevance vector machine on HMPSoC

AU - He, Yongfu

AU - Wang, Shaojun

AU - Peng, Yu

AU - Pang, Yeyong

AU - Ma, Ning

AU - Pang, Jingyue

PY - 2014

Y1 - 2014

N2 - Relevance Vector Machine (RVM) with the uncertainty expressing ability has spawned broad applications in Prognostic and Health Management (PHM). However computationally intensive intrinsic nature of RVM greatly limits its usage. This paper presents a software and hardware co-design approach based on HMPSoC technology, which efficiently exploited sequential and parallel nature of RVM. Multi-channel and pipelined hardware architecture for the acceleration of kernel formulation and intermediate values calculation is proposed. The hardware that wrapped with AXI-Stream interface is integrated into HMPSoC as an acceleration engine. We implement the design on an on-board PHM prototype platform with a Xilinx Zynq XC7Z020 AP SoC. The experiment results show 5.3x and 46.8x speed up in terms of the time cost than the RVM running on PC with a Xeon 5620 processor and ARM Cortex A9 processor. The energy consumption is reduced by 153.0x and 37.3, respectively.

AB - Relevance Vector Machine (RVM) with the uncertainty expressing ability has spawned broad applications in Prognostic and Health Management (PHM). However computationally intensive intrinsic nature of RVM greatly limits its usage. This paper presents a software and hardware co-design approach based on HMPSoC technology, which efficiently exploited sequential and parallel nature of RVM. Multi-channel and pipelined hardware architecture for the acceleration of kernel formulation and intermediate values calculation is proposed. The hardware that wrapped with AXI-Stream interface is integrated into HMPSoC as an acceleration engine. We implement the design on an on-board PHM prototype platform with a Xilinx Zynq XC7Z020 AP SoC. The experiment results show 5.3x and 46.8x speed up in terms of the time cost than the RVM running on PC with a Xeon 5620 processor and ARM Cortex A9 processor. The energy consumption is reduced by 153.0x and 37.3, respectively.

UR - https://www.scopus.com/pages/publications/84928741906

U2 - 10.1109/FPT.2014.7082812

DO - 10.1109/FPT.2014.7082812

M3 - 会议稿件

AN - SCOPUS:84928741906

T3 - Proceedings of the 2014 International Conference on Field-Programmable Technology, FPT 2014

SP - 334

EP - 337

BT - Proceedings of the 2014 International Conference on Field-Programmable Technology, FPT 2014

A2 - Chen, Jialin

A2 - Yin, Wenbo

A2 - Shibata, Yuichiro

A2 - Wang, Lingli

A2 - So, Hayden Kwok-Hay

A2 - Ma, Yuchun

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 13th International Conference on Field-Programmable Technology, FPT 2014

Y2 - 10 December 2014 through 12 December 2014

ER -

He Y, Wang S, Peng Y, Pang Y, Ma N, Pang J. High performance relevance vector machine on HMPSoC. In Chen J, Yin W, Shibata Y, Wang L, So HKH, Ma Y, editors, Proceedings of the 2014 International Conference on Field-Programmable Technology, FPT 2014. Institute of Electrical and Electronics Engineers Inc. 2014. p. 334-337. 7082812. (Proceedings of the 2014 International Conference on Field-Programmable Technology, FPT 2014). doi: 10.1109/FPT.2014.7082812

High performance relevance vector machine on HMPSoC

Abstract

Publication series

Conference

UN SDGs

Access to Document

Other files and links

Fingerprint

Cite this