An empirical study of applying ensembles of heterogeneous classifiers on imperfect data

Kuo Wei Hsu, Jaideep Srivastava

Research output: Chapter in Book/Report/Conference proceedingConference contribution

6 Citations (Scopus)

Abstract

Two factors that slow down the deployment of classification or supervised learning in real-world situations. One is the reality that data are not perfect in practice, while the other is the fact that every technique has its own limits. Although there have been techniques developed to resolve issues about imperfectness of real-world data, there is no single one that outperforms all others and each such technique focuses on some types of imperfectness. Furthermore, quite a few works apply ensembles of heterogeneous classifiers to such situations. In this paper, we report a work on progress that studies the impact of heterogeneity on ensemble, especially focusing on the following aspects: diversity and classification quality for imbalanced data. Our goal is to evaluate how introducing heterogeneity into ensemble influences its behavior and performance.

Original languageEnglish
Title of host publicationNew Frontiers in Applied Data Mining - PAKDD 2009 International Workshops, Revised Selected Papers
Pages28-39
Number of pages12
DOIs
Publication statusPublished - 30 Sep 2010
Event13th Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2009 - Bangkok, Thailand
Duration: 27 Apr 200930 Apr 2009

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume5669 LNAI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other13th Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2009
CountryThailand
CityBangkok
Period27/4/0930/4/09

    Fingerprint

Keywords

  • AdaBoost
  • bagging
  • diversity
  • heterogeneity
  • imbalanced data

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Hsu, K. W., & Srivastava, J. (2010). An empirical study of applying ensembles of heterogeneous classifiers on imperfect data. In New Frontiers in Applied Data Mining - PAKDD 2009 International Workshops, Revised Selected Papers (pp. 28-39). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 5669 LNAI). https://doi.org/10.1007/978-3-642-14640-4_3