A new k-winners-take-all neural network and its array architecture

Jui Cheng Yen; Jiun-In  Guo; Hun Chen Chen

doi:10.1109/72.712163

A new k-winners-take-all neural network and its array architecture

Jui Cheng Yen^*, Jiun-In Guo, Hun Chen Chen

^*Corresponding author for this work

Institute of Electronics

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

51 Scopus citations

Abstract

In this paper, a new neural-network model called WINSTRON and its novel array architecture are proposed. Based on a competitive learning algorithm that is originated from the coarse-fine competition, WINSTRON can identify the k larger elements or the k smaller ones in a data set. We will then prove that WINSTRON converges to the correct state in any situation. In addition, the convergence rates of WINSTRON for three special data distributions will be derived. In order to realize WINSTRON, its array architecture with low hardware complexity and high computing speed is also detailed. Finally, simulation results are included to demonstrate its effectiveness and its advantages over three existing networks.

Original language	English
Pages (from-to)	901-912
Number of pages	12
Journal	IEEE Transactions on Neural Networks
Volume	9
Issue number	5
DOIs	https://doi.org/10.1109/72.712163
State	Published - 1 Dec 1998

Keywords

Artificial neural network
Competitive learning
Distributed arithmetic
k-winners-take-all
Memory-based architecture
Systolic array

Access to Document

10.1109/72.712163

Link to publication in Scopus

Fingerprint Dive into the research topics of 'A new k-winners-take-all neural network and its array architecture'. Together they form a unique fingerprint.

Cite this

@article{7d26fac823d2402d9018770db51ba5d5,

title = "A new k-winners-take-all neural network and its array architecture",

abstract = "In this paper, a new neural-network model called WINSTRON and its novel array architecture are proposed. Based on a competitive learning algorithm that is originated from the coarse-fine competition, WINSTRON can identify the k larger elements or the k smaller ones in a data set. We will then prove that WINSTRON converges to the correct state in any situation. In addition, the convergence rates of WINSTRON for three special data distributions will be derived. In order to realize WINSTRON, its array architecture with low hardware complexity and high computing speed is also detailed. Finally, simulation results are included to demonstrate its effectiveness and its advantages over three existing networks.",

keywords = "Artificial neural network, Competitive learning, Distributed arithmetic, k-winners-take-all, Memory-based architecture, Systolic array",

author = "Yen, {Jui Cheng} and Jiun-In Guo and Chen, {Hun Chen}",

year = "1998",

month = dec,

day = "1",

doi = "10.1109/72.712163",

language = "English",

volume = "9",

pages = "901--912",

journal = "IEEE Transactions on Neural Networks and Learning Systems",

issn = "2162-237X",

publisher = "IEEE Computational Intelligence Society",

number = "5",

}

TY - JOUR

T1 - A new k-winners-take-all neural network and its array architecture

AU - Yen, Jui Cheng

AU - Guo, Jiun-In

AU - Chen, Hun Chen

PY - 1998/12/1

Y1 - 1998/12/1

N2 - In this paper, a new neural-network model called WINSTRON and its novel array architecture are proposed. Based on a competitive learning algorithm that is originated from the coarse-fine competition, WINSTRON can identify the k larger elements or the k smaller ones in a data set. We will then prove that WINSTRON converges to the correct state in any situation. In addition, the convergence rates of WINSTRON for three special data distributions will be derived. In order to realize WINSTRON, its array architecture with low hardware complexity and high computing speed is also detailed. Finally, simulation results are included to demonstrate its effectiveness and its advantages over three existing networks.

AB - In this paper, a new neural-network model called WINSTRON and its novel array architecture are proposed. Based on a competitive learning algorithm that is originated from the coarse-fine competition, WINSTRON can identify the k larger elements or the k smaller ones in a data set. We will then prove that WINSTRON converges to the correct state in any situation. In addition, the convergence rates of WINSTRON for three special data distributions will be derived. In order to realize WINSTRON, its array architecture with low hardware complexity and high computing speed is also detailed. Finally, simulation results are included to demonstrate its effectiveness and its advantages over three existing networks.

KW - Artificial neural network

KW - Competitive learning

KW - Distributed arithmetic

KW - k-winners-take-all

KW - Memory-based architecture

KW - Systolic array

UR - http://www.scopus.com/inward/record.url?scp=0032164161&partnerID=8YFLogxK

U2 - 10.1109/72.712163

DO - 10.1109/72.712163

M3 - Article

C2 - 18255775

AN - SCOPUS:0032164161

VL - 9

SP - 901

EP - 912

JO - IEEE Transactions on Neural Networks and Learning Systems

JF - IEEE Transactions on Neural Networks and Learning Systems

SN - 2162-237X

IS - 5

ER -