A large-scale performance study of cluster-based high-dimensional indexing

Gylfi Gudmundsson Pór*, Björn Jónsson Pór, Laurent Amsaleg

*Corresponding author for this work

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

25 Citations (Scopus)

Abstract

High-dimensional clustering is used by some content-based image retrieval systems to partition the data into groups; the groups (clusters) are then indexed to accelerate processing of queries. Recently, the Cluster Pruning approach was proposed as a simple way to produce such clusters. While the original evaluation of the algorithm was performed within a text indexing context at a rather small scale, its simplicity motivated us to study its behavior in an image indexing context at a much larger scale. This paper summarizes the results of this study and shows that while the basic algorithm works fairly well, three extensions dramatically improve its performance and scalability, accelerating both query processing and the construction of clusters, making Cluster Pruning a promising basis for building large-scale systems that require a clustering algorithm.

Original languageEnglish
Title of host publicationVLS-MCMR'10 - Proceedings of the 2010 ACM International Workshop on Very-Large-Scale Multimedia Corpus, Mining and Retrieval, Co-located with ACM Multimedia 2010
Pages31-36
Number of pages6
DOIs
Publication statusPublished - 2010
Event2010 ACM International Workshop on Very-Large-Scale Multimedia Corpus, Mining and Retrieval, VLS-MCMR'10, Co-located with ACM Multimedia 2010 - Firenze, Italy
Duration: 29 Oct 201029 Oct 2010

Publication series

NameVLS-MCMR'10 - Proceedings of the 2010 ACM International Workshop on Very-Large-Scale Multimedia Corpus, Mining and Retrieval, Co-located with ACM Multimedia 2010

Conference

Conference2010 ACM International Workshop on Very-Large-Scale Multimedia Corpus, Mining and Retrieval, VLS-MCMR'10, Co-located with ACM Multimedia 2010
Country/TerritoryItaly
CityFirenze
Period29/10/1029/10/10

Other keywords

  • Algorithms
  • Performance

Fingerprint

Dive into the research topics of 'A large-scale performance study of cluster-based high-dimensional indexing'. Together they form a unique fingerprint.

Cite this