Divide to Federate Clustering Concept for Unsupervised Learning

Atiq Ur Rehman*, Samir Brahim Belhaouari, Tanya Stanko, Vladimir Gorovoy

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

In this paper, the concept of divide and federate is evaluated to find the clusters that are different in densities and shapes and are contaminated with noise. The proposed divide-and-federate clustering method is based on the density and distance evaluation of the data. Wherein, the first phase of the algorithm divides the data into different sub-clusters based on the density evaluation with respect to all the data dimensions and, in the second phase, the small sub-clusters are federated with large sub-clusters to create the actual data clusters. The federation phase of the proposed clustering method is based on the distance evaluation of clusters and is merged based on the close proximity of neighbors. The proposed clustering algorithm is capable of handling noisy data through the integration of an outlier detection preprocessing method. The usefulness of the proposed algorithm is demonstrated with some examples of complex synthetic benchmark functions.

Original languageEnglish
Title of host publicationProceedings of 7th International Congress on Information and Communication Technology - ICICT 2022
EditorsXin-She Yang, Simon Sherratt, Nilanjan Dey, Amit Joshi
PublisherSpringer Science and Business Media Deutschland GmbH
Pages19-29
Number of pages11
ISBN (Print)9789811923968
DOIs
Publication statusPublished - 2023
Event7th International Congress on Information and Communication Technology, ICICT 2022 - Virtual, Online
Duration: 21 Feb 202224 Feb 2022

Publication series

NameLecture Notes in Networks and Systems
Volume465
ISSN (Print)2367-3370
ISSN (Electronic)2367-3389

Conference

Conference7th International Congress on Information and Communication Technology, ICICT 2022
CityVirtual, Online
Period21/02/2224/02/22

Keywords

  • Clustering
  • Subspace learning
  • Unsupervised learning

Fingerprint

Dive into the research topics of 'Divide to Federate Clustering Concept for Unsupervised Learning'. Together they form a unique fingerprint.

Cite this