An investigation on the compression quality of aiNet

Thomas Stibor*, Jonathan Timmis

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference Proceeding (Non-Journal item)

22 Citations (SciVal)

Abstract

AiNet Is an Immune-inspired algorithm for data compression, i.e. the reduction of redundancy in data sets. In this paper we investigate the compression quality of aiNet. Therefore, a similarity measure between input set and reduced output set is presented which is based on the Parzen window estimation and the Kullback-Leibler divergence. Four different artificially generated data sets are created and the compression quality is investigated. Experiments reveal that aiNet produced reasonable results on an uniformly distributed data set, but poor results on non-uniformly distributed data sets, i.e. data sets which contain dense point regions. This effect is caused by the optimization criterion of aiNet.

Original languageEnglish
Title of host publicationProceedings of the 2007 IEEE Symposium on Foundations of Computational Intelligence, FOCI 2007
PublisherIEEE Press
Pages495-502
Number of pages8
ISBN (Print)1424407036, 9781424407033
DOIs
Publication statusPublished - 2007
Event2007 IEEE Symposium on Foundations of Computational Intelligence, FOCI 2007 - Honolulu, HI, United States of America
Duration: 01 Apr 200705 Apr 2007

Publication series

NameProceedings of the 2007 IEEE Symposium on Foundations of Computational Intelligence, FOCI 2007

Conference

Conference2007 IEEE Symposium on Foundations of Computational Intelligence, FOCI 2007
Country/TerritoryUnited States of America
CityHonolulu, HI
Period01 Apr 200705 Apr 2007

Keywords

  • clustering algorithms
  • immune system
  • computer science
  • computational intelligence
  • density measurement
  • continuous production
  • mirrors
  • helium
  • data compression
  • pattern recognition

Fingerprint

Dive into the research topics of 'An investigation on the compression quality of aiNet'. Together they form a unique fingerprint.

Cite this