Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Ensembles of Cluster Validation Indices for Label Noise Filtering
acs Plus GmbH, DEU.
Blekinge Institute of Technology, Faculty of Computing, Department of Computer Science.ORCID iD: 0000-0003-3128-191x
Blekinge Institute of Technology, Faculty of Computing, Department of Computer Science.
Technical University Sofia, BGR.
2020 (English)In: Studies in Computational Intelligence, Springer, 2020, 864, p. 71-98Chapter in book (Refereed)
Abstract [en]

Cluster validation measures are designed to find the partitioning that best fits the underlying data. In this study, we show that these measures can be used for identifying mislabeled instances or class outliers prior to training in supervised learning problems. We introduce an ensemble technique, entitled CVI-based Outlier Filtering, which identifies and eliminates mislabeled instances from the training set, and then builds a classification hypothesis from the set of remaining instances. Our approach assigns to each instance in the training set several cluster validation scores representing its potential of being a class outlier with respect to the clustering properties the used validation measures assess. In this respect, the proposed approach may be referred to a multi-criteria outlier filtering measure. In this work, we specifically study and evaluate valued-based ensembles of cluster validation indices. The added value of this approach in comparison to the logical and rank-based ensemble solutions are discussed and further demonstrated. © 2020, Springer Nature Switzerland AG.

Place, publisher, year, edition, pages
Springer, 2020, 864. p. 71-98
Series
Studies in Computational Intelligence, ISSN 1860949X ; 864
National Category
Computer Sciences Information Systems
Identifiers
URN: urn:nbn:se:bth-19346DOI: 10.1007/978-3-030-38704-4_4Scopus ID: 2-s2.0-85081586798OAI: oai:DiVA.org:bth-19346DiVA, id: diva2:1417619
Projects
Scalable resource efficient systems for big data analytics
Funder
Knowledge Foundation, 20140032Available from: 2020-03-30 Created: 2020-03-30 Last updated: 2020-04-30Bibliographically approved

Open Access in DiVA

No full text in DiVA

Other links

Publisher's full textScopus

Authority records BETA

Boeva, VeselkaLundberg, Lars

Search in DiVA

By author/editor
Boeva, VeselkaLundberg, Lars
By organisation
Department of Computer Science
Computer SciencesInformation Systems

Search outside of DiVA

GoogleGoogle Scholar

doi
urn-nbn

Altmetric score

doi
urn-nbn
Total: 35 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf