GDCluster: A General Decentralized Clustering Algorithm

Hoda Mashayekhi; Jafar Habibi; Tania Khalafbeigi; Spyros Voulgaris; Maarten van Steen

doi:10.1109/TKDE.2015.2391123

GDCluster: A General Decentralized Clustering Algorithm

Hoda Mashayekhi^*, Jafar Habibi, Tania Khalafbeigi, Spyros Voulgaris, Maarten van Steen

^*Corresponding author for this work

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

28 Citations (Scopus)

151 Downloads (Pure)

Abstract

In many popular applications like peer-to-peer systems, large amounts of data are distributed among multiple sources. Analysis of this data and identifying clusters is challenging due to processing, storage, and transmission costs. In this paper, we propose GDCluster, a general fully decentralized clustering method, which is capable of clustering dynamic and distributed data sets. Nodes continuously cooperate through decentralized gossip-based communication to maintain summarized views of the data set. We customize GDCluster for execution of the partition-based and density-based clustering methods on the summarized views, and also offer enhancements to the basic algorithm. Coping with dynamic data is made possible by gradually adapting the clustering model. Our experimental evaluations show that GDCluster can discover the clusters efficiently with scalable transmission cost, and also expose its supremacy in comparison to the popular method LSP2P.

Original language	English
Article number	7006742
Pages (from-to)	1892-1905
Number of pages	14
Journal	IEEE transactions on knowledge and data engineering
Volume	27
Issue number	7
DOIs	https://doi.org/10.1109/TKDE.2015.2391123
Publication status	Published - 1 Jul 2015

Keywords

Clustering
Distributed systems
Partition-based clustering
Density-based clustering
Dynamic systems
n/a OA procedure

Access to Document

10.1109/TKDE.2015.2391123

ArticleAccepted author version, 774 KB

Cite this

@article{61ae904be083446cb42a051d9830b6df,

title = "GDCluster: A General Decentralized Clustering Algorithm",

abstract = "In many popular applications like peer-to-peer systems, large amounts of data are distributed among multiple sources. Analysis of this data and identifying clusters is challenging due to processing, storage, and transmission costs. In this paper, we propose GDCluster, a general fully decentralized clustering method, which is capable of clustering dynamic and distributed data sets. Nodes continuously cooperate through decentralized gossip-based communication to maintain summarized views of the data set. We customize GDCluster for execution of the partition-based and density-based clustering methods on the summarized views, and also offer enhancements to the basic algorithm. Coping with dynamic data is made possible by gradually adapting the clustering model. Our experimental evaluations show that GDCluster can discover the clusters efficiently with scalable transmission cost, and also expose its supremacy in comparison to the popular method LSP2P.",

keywords = "Clustering, Distributed systems, Partition-based clustering, Density-based clustering, Dynamic systems, n/a OA procedure",

author = "Hoda Mashayekhi and Jafar Habibi and Tania Khalafbeigi and Spyros Voulgaris and {van Steen}, Maarten",

note = "Publisher Copyright: {\textcopyright} 2015 IEEE.",

year = "2015",

month = jul,

day = "1",

doi = "10.1109/TKDE.2015.2391123",

language = "English",

volume = "27",

pages = "1892--1905",

journal = "IEEE transactions on knowledge and data engineering",

issn = "1041-4347",

publisher = "IEEE",

number = "7",

}

TY - JOUR

T1 - GDCluster

T2 - A General Decentralized Clustering Algorithm

AU - Mashayekhi, Hoda

AU - Habibi, Jafar

AU - Khalafbeigi, Tania

AU - Voulgaris, Spyros

AU - van Steen, Maarten

PY - 2015/7/1

Y1 - 2015/7/1

N2 - In many popular applications like peer-to-peer systems, large amounts of data are distributed among multiple sources. Analysis of this data and identifying clusters is challenging due to processing, storage, and transmission costs. In this paper, we propose GDCluster, a general fully decentralized clustering method, which is capable of clustering dynamic and distributed data sets. Nodes continuously cooperate through decentralized gossip-based communication to maintain summarized views of the data set. We customize GDCluster for execution of the partition-based and density-based clustering methods on the summarized views, and also offer enhancements to the basic algorithm. Coping with dynamic data is made possible by gradually adapting the clustering model. Our experimental evaluations show that GDCluster can discover the clusters efficiently with scalable transmission cost, and also expose its supremacy in comparison to the popular method LSP2P.

AB - In many popular applications like peer-to-peer systems, large amounts of data are distributed among multiple sources. Analysis of this data and identifying clusters is challenging due to processing, storage, and transmission costs. In this paper, we propose GDCluster, a general fully decentralized clustering method, which is capable of clustering dynamic and distributed data sets. Nodes continuously cooperate through decentralized gossip-based communication to maintain summarized views of the data set. We customize GDCluster for execution of the partition-based and density-based clustering methods on the summarized views, and also offer enhancements to the basic algorithm. Coping with dynamic data is made possible by gradually adapting the clustering model. Our experimental evaluations show that GDCluster can discover the clusters efficiently with scalable transmission cost, and also expose its supremacy in comparison to the popular method LSP2P.

KW - Clustering

KW - Distributed systems

KW - Partition-based clustering

KW - Density-based clustering

KW - Dynamic systems

KW - n/a OA procedure

UR - http://www.scopus.com/inward/record.url?scp=84959489386&partnerID=8YFLogxK

U2 - 10.1109/TKDE.2015.2391123

DO - 10.1109/TKDE.2015.2391123

M3 - Article

SN - 1041-4347

VL - 27

SP - 1892

EP - 1905

JO - IEEE transactions on knowledge and data engineering

JF - IEEE transactions on knowledge and data engineering

IS - 7

M1 - 7006742

ER -

GDCluster: A General Decentralized Clustering Algorithm

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this