High speed k-winner-take-all competitive learning in reconfigurable hardware

Hui Ya Li, Yao Jung Yeh, Wen-Jyi Hwang, Cheng Tsun Yang

研究成果: 書貢獻/報告類型會議貢獻

4 引文 (Scopus)

摘要

A novel hardware architecture of the competitive learning (CL) algorithm with k-winners-take-all activation is presented in this paper. It is used as a custom logic block in the arithmetic logic unit (ALU) of the softcore NIOS processor for CL training. Both the partial distance search (PDS) module and hardware divider adopt finite precision calculation for area cost reduction at the expense of slight degradation in training performance. The PDS module also employs subspace search and multiple-coefficient accumulation techniques for effective reduction of the computation latency for the PDS search. Experiment results show that the CPU time is lower than that of Pentium IV processors running the CL training program without the support of custom hardware.

原文英語
主出版物標題Next-Generation Applied Intelligence - 22nd International Conference on Industrial, Engineering and Other Applications of Applied Intelligent Systems, IEA/AIE 2009, Proceedings
頁面594-603
頁數10
DOIs
出版狀態已發佈 - 2009 十一月 16
事件22nd International Conference on Industrial, Engineering and Other Applications of Applied Intelligent Systems, IEA/AIE 2009 - Tainan, 臺灣
持續時間: 2009 六月 242009 六月 27

出版系列

名字Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
5579 LNAI
ISSN(列印)0302-9743
ISSN(電子)1611-3349

其他

其他22nd International Conference on Industrial, Engineering and Other Applications of Applied Intelligent Systems, IEA/AIE 2009
國家臺灣
城市Tainan
期間09/6/2409/6/27

指紋

Winner-take-all
Competitive Learning
Reconfigurable hardware
Reconfigurable Hardware
High Speed
Hardware
Cost reduction
Learning algorithms
Computer hardware
Program processors
Partial
Chemical activation
Degradation
Logic
Module
Hardware Architecture
CPU Time
Experiments
Latency
Activation

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

引用此文

Li, H. Y., Yeh, Y. J., Hwang, W-J., & Yang, C. T. (2009). High speed k-winner-take-all competitive learning in reconfigurable hardware. 於 Next-Generation Applied Intelligence - 22nd International Conference on Industrial, Engineering and Other Applications of Applied Intelligent Systems, IEA/AIE 2009, Proceedings (頁 594-603). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); 卷 5579 LNAI). https://doi.org/10.1007/978-3-642-02568-6_60

High speed k-winner-take-all competitive learning in reconfigurable hardware. / Li, Hui Ya; Yeh, Yao Jung; Hwang, Wen-Jyi; Yang, Cheng Tsun.

Next-Generation Applied Intelligence - 22nd International Conference on Industrial, Engineering and Other Applications of Applied Intelligent Systems, IEA/AIE 2009, Proceedings. 2009. p. 594-603 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); 卷 5579 LNAI).

研究成果: 書貢獻/報告類型會議貢獻

Li, HY, Yeh, YJ, Hwang, W-J & Yang, CT 2009, High speed k-winner-take-all competitive learning in reconfigurable hardware. 於 Next-Generation Applied Intelligence - 22nd International Conference on Industrial, Engineering and Other Applications of Applied Intelligent Systems, IEA/AIE 2009, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 卷 5579 LNAI, 頁 594-603, 22nd International Conference on Industrial, Engineering and Other Applications of Applied Intelligent Systems, IEA/AIE 2009, Tainan, 臺灣, 09/6/24. https://doi.org/10.1007/978-3-642-02568-6_60
Li HY, Yeh YJ, Hwang W-J, Yang CT. High speed k-winner-take-all competitive learning in reconfigurable hardware. 於 Next-Generation Applied Intelligence - 22nd International Conference on Industrial, Engineering and Other Applications of Applied Intelligent Systems, IEA/AIE 2009, Proceedings. 2009. p. 594-603. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-642-02568-6_60
Li, Hui Ya ; Yeh, Yao Jung ; Hwang, Wen-Jyi ; Yang, Cheng Tsun. / High speed k-winner-take-all competitive learning in reconfigurable hardware. Next-Generation Applied Intelligence - 22nd International Conference on Industrial, Engineering and Other Applications of Applied Intelligent Systems, IEA/AIE 2009, Proceedings. 2009. 頁 594-603 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{c40299932bd847c683fb580fe1199424,
title = "High speed k-winner-take-all competitive learning in reconfigurable hardware",
abstract = "A novel hardware architecture of the competitive learning (CL) algorithm with k-winners-take-all activation is presented in this paper. It is used as a custom logic block in the arithmetic logic unit (ALU) of the softcore NIOS processor for CL training. Both the partial distance search (PDS) module and hardware divider adopt finite precision calculation for area cost reduction at the expense of slight degradation in training performance. The PDS module also employs subspace search and multiple-coefficient accumulation techniques for effective reduction of the computation latency for the PDS search. Experiment results show that the CPU time is lower than that of Pentium IV processors running the CL training program without the support of custom hardware.",
author = "Li, {Hui Ya} and Yeh, {Yao Jung} and Wen-Jyi Hwang and Yang, {Cheng Tsun}",
year = "2009",
month = "11",
day = "16",
doi = "10.1007/978-3-642-02568-6_60",
language = "English",
isbn = "3642025676",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
pages = "594--603",
booktitle = "Next-Generation Applied Intelligence - 22nd International Conference on Industrial, Engineering and Other Applications of Applied Intelligent Systems, IEA/AIE 2009, Proceedings",

}

TY - GEN

T1 - High speed k-winner-take-all competitive learning in reconfigurable hardware

AU - Li, Hui Ya

AU - Yeh, Yao Jung

AU - Hwang, Wen-Jyi

AU - Yang, Cheng Tsun

PY - 2009/11/16

Y1 - 2009/11/16

N2 - A novel hardware architecture of the competitive learning (CL) algorithm with k-winners-take-all activation is presented in this paper. It is used as a custom logic block in the arithmetic logic unit (ALU) of the softcore NIOS processor for CL training. Both the partial distance search (PDS) module and hardware divider adopt finite precision calculation for area cost reduction at the expense of slight degradation in training performance. The PDS module also employs subspace search and multiple-coefficient accumulation techniques for effective reduction of the computation latency for the PDS search. Experiment results show that the CPU time is lower than that of Pentium IV processors running the CL training program without the support of custom hardware.

AB - A novel hardware architecture of the competitive learning (CL) algorithm with k-winners-take-all activation is presented in this paper. It is used as a custom logic block in the arithmetic logic unit (ALU) of the softcore NIOS processor for CL training. Both the partial distance search (PDS) module and hardware divider adopt finite precision calculation for area cost reduction at the expense of slight degradation in training performance. The PDS module also employs subspace search and multiple-coefficient accumulation techniques for effective reduction of the computation latency for the PDS search. Experiment results show that the CPU time is lower than that of Pentium IV processors running the CL training program without the support of custom hardware.

UR - http://www.scopus.com/inward/record.url?scp=71049188555&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=71049188555&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-02568-6_60

DO - 10.1007/978-3-642-02568-6_60

M3 - Conference contribution

AN - SCOPUS:71049188555

SN - 3642025676

SN - 9783642025679

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 594

EP - 603

BT - Next-Generation Applied Intelligence - 22nd International Conference on Industrial, Engineering and Other Applications of Applied Intelligent Systems, IEA/AIE 2009, Proceedings

ER -