HP: A light-weight hybrid algorithm for accurate data partitioning

Peter Olukanmi, Fulufhelo Nelwamondo, Tshilidzi Marwala

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

1 Citation (Scopus)

Abstract

This paper introduces a hybridization of the k-means and k-medoids paradigms. The new algorithms is named HP (hybrid partitioning) algorithm. Specifically, we improve on a recently developed scalable version of k-means (k-means-lite), by introducing the PAM algorithm into it in such a way that the high accuracy of the latter is absorbed without inheriting its high inefficiency. K-means-lite runs standard k-means on the combination of intermediate centroids obtained by initially feeding n samples into k-means. In HP, instead of k-means, PAM is used to cluster the combination of centroids obtained from the samples. This PAM component is fast because it is run on very small data, precisely of size nk, Experiments show that this modification improves not only the accuracy of k-means-lite but also outperforms the accuracy of k-means, without losing much k-means-lite's efficiency.

Original languageEnglish
Title of host publication2020 International Conference on Artificial Intelligence, Big Data, Computing and Data Communication Systems, icABCD 2020 - Proceedings
EditorsSameerchand Pudaruth, Upasana Singh
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9781728167701
DOIs
Publication statusPublished - Aug 2020
Event2020 International Conference on Artificial Intelligence, Big Data, Computing and Data Communication Systems, icABCD 2020 - Durban, KwaZulu Natal, South Africa
Duration: 6 Aug 20207 Aug 2020

Publication series

Name2020 International Conference on Artificial Intelligence, Big Data, Computing and Data Communication Systems, icABCD 2020 - Proceedings

Conference

Conference2020 International Conference on Artificial Intelligence, Big Data, Computing and Data Communication Systems, icABCD 2020
Country/TerritorySouth Africa
CityDurban, KwaZulu Natal
Period6/08/207/08/20

Keywords

  • Accurate
  • Clustering
  • Efficient
  • K-means
  • K-means-lite
  • K-medoids
  • PAM

ASJC Scopus subject areas

  • Artificial Intelligence
  • Computer Networks and Communications
  • Computer Science Applications
  • Computer Vision and Pattern Recognition
  • Hardware and Architecture
  • Information Systems and Management

Fingerprint

Dive into the research topics of 'HP: A light-weight hybrid algorithm for accurate data partitioning'. Together they form a unique fingerprint.

Cite this