Single-state distributed k-winners-take-all neural network model

Yinyan Zhang; Shuai Li; Xuefeng Zhou; Jian Weng; Guanggang Geng

doi:10.1016/j.ins.2023.119528

Single-state distributed k-winners-take-all neural network model

Yinyan Zhang, Shuai Li^*, Xuefeng Zhou, Jian Weng, Guanggang Geng

^*Corresponding author for this work

BA56 Cognitive production industry

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

8 Citations (Scopus)

Abstract

Distributed k-winners-takes-all (k-WTA) neural network (k-WTANN) models have better scalability than centralized ones. In this work, a distributed k-WTANN model with a simple structure is designed for the efficient selection of k winners among a group of more than k agents via competition based on their inputs. Unlike an existing distributed k-WTANN model, the proposed model does not rely on consensus filters, and only has one state variable. We prove that under mild conditions, the proposed distributed k-WTANN model has global asymptotic convergence. The theoretical conclusions are validated via numerical examples, which also show that our model is of better convergence speed than the existing distributed k-WTANN model.

Original language	English
Article number	119528
Journal	Information Sciences
Volume	647
DOIs	https://doi.org/10.1016/j.ins.2023.119528
Publication status	Published - Nov 2023
MoE publication type	A1 Journal article-refereed

Funding

This work is supported in part by the National Natural Science Foundation of China under Grant 62206109 , the Guangdong Key Laboratory of Data Security and Privacy Preserving under Grant 2017B030301004 , the Guangdong Basic and Applied Basic Research Foundation under Grant 2022A1515010976 , the Young Scholar Program of Pazhou Lab under Grant PZL2021KF0022 , the Science and Technology Program of Guangzhou under Grant 202201010457 . G. Geng is supported by the Pearl River Talents Plan . We would like to thank the reviewers for their valuable comments.

Keywords

Distributed competition
k-winners-take-all
Recurrent neural network

Access to Document

10.1016/j.ins.2023.119528

Cite this

@article{41b07ec7832c40c3818fdce5e2d10da4,

title = "Single-state distributed k-winners-take-all neural network model",

abstract = "Distributed k-winners-takes-all (k-WTA) neural network (k-WTANN) models have better scalability than centralized ones. In this work, a distributed k-WTANN model with a simple structure is designed for the efficient selection of k winners among a group of more than k agents via competition based on their inputs. Unlike an existing distributed k-WTANN model, the proposed model does not rely on consensus filters, and only has one state variable. We prove that under mild conditions, the proposed distributed k-WTANN model has global asymptotic convergence. The theoretical conclusions are validated via numerical examples, which also show that our model is of better convergence speed than the existing distributed k-WTANN model.",

keywords = "Distributed competition, k-winners-take-all, Recurrent neural network",

author = "Yinyan Zhang and Shuai Li and Xuefeng Zhou and Jian Weng and Guanggang Geng",

year = "2023",

month = nov,

doi = "10.1016/j.ins.2023.119528",

language = "English",

volume = "647",

journal = "Information Sciences",

issn = "0020-0255",

publisher = "Elsevier",

}

TY - JOUR

T1 - Single-state distributed k-winners-take-all neural network model

AU - Zhang, Yinyan

AU - Li, Shuai

AU - Zhou, Xuefeng

AU - Weng, Jian

AU - Geng, Guanggang

PY - 2023/11

Y1 - 2023/11

N2 - Distributed k-winners-takes-all (k-WTA) neural network (k-WTANN) models have better scalability than centralized ones. In this work, a distributed k-WTANN model with a simple structure is designed for the efficient selection of k winners among a group of more than k agents via competition based on their inputs. Unlike an existing distributed k-WTANN model, the proposed model does not rely on consensus filters, and only has one state variable. We prove that under mild conditions, the proposed distributed k-WTANN model has global asymptotic convergence. The theoretical conclusions are validated via numerical examples, which also show that our model is of better convergence speed than the existing distributed k-WTANN model.

AB - Distributed k-winners-takes-all (k-WTA) neural network (k-WTANN) models have better scalability than centralized ones. In this work, a distributed k-WTANN model with a simple structure is designed for the efficient selection of k winners among a group of more than k agents via competition based on their inputs. Unlike an existing distributed k-WTANN model, the proposed model does not rely on consensus filters, and only has one state variable. We prove that under mild conditions, the proposed distributed k-WTANN model has global asymptotic convergence. The theoretical conclusions are validated via numerical examples, which also show that our model is of better convergence speed than the existing distributed k-WTANN model.

KW - Distributed competition

KW - k-winners-take-all

KW - Recurrent neural network

UR - http://www.scopus.com/inward/record.url?scp=85168805261&partnerID=8YFLogxK

U2 - 10.1016/j.ins.2023.119528

DO - 10.1016/j.ins.2023.119528

M3 - Article

AN - SCOPUS:85168805261

SN - 0020-0255

VL - 647

JO - Information Sciences

JF - Information Sciences

M1 - 119528

ER -

Single-state distributed k-winners-take-all neural network model

Abstract

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this