Eigenvalue based spectral classification
This paper describes a new method of classification based on spectral analysis. The motivations behind developing the new model were the failures of the classical spectral cluster analysis based on combinatorial and normalized Laplacian for a set of real-world datasets of textual documents. Reasons...
Main Authors: | , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
Public Library of Science (PLoS)
2023-01-01
|
Series: | PLoS ONE |
Online Access: | https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10079090/?tool=EBI |
_version_ | 1797850302119936000 |
---|---|
author | Piotr Borkowski Mieczysław A. Kłopotek Bartłomiej Starosta Sławomir T. Wierzchoń Marcin Sydow |
author_facet | Piotr Borkowski Mieczysław A. Kłopotek Bartłomiej Starosta Sławomir T. Wierzchoń Marcin Sydow |
author_sort | Piotr Borkowski |
collection | DOAJ |
description | This paper describes a new method of classification based on spectral analysis. The motivations behind developing the new model were the failures of the classical spectral cluster analysis based on combinatorial and normalized Laplacian for a set of real-world datasets of textual documents. Reasons of the failures are analysed. While the known methods are all based on usage of eigenvectors of graph Laplacians, a new classification method based on eigenvalues of graph Laplacians is proposed and studied. |
first_indexed | 2024-04-09T18:59:16Z |
format | Article |
id | doaj.art-6b95fa7bbc5944c49c27cbc56b4f85e4 |
institution | Directory Open Access Journal |
issn | 1932-6203 |
language | English |
last_indexed | 2024-04-09T18:59:16Z |
publishDate | 2023-01-01 |
publisher | Public Library of Science (PLoS) |
record_format | Article |
series | PLoS ONE |
spelling | doaj.art-6b95fa7bbc5944c49c27cbc56b4f85e42023-04-09T05:32:13ZengPublic Library of Science (PLoS)PLoS ONE1932-62032023-01-01184Eigenvalue based spectral classificationPiotr BorkowskiMieczysław A. KłopotekBartłomiej StarostaSławomir T. WierzchońMarcin SydowThis paper describes a new method of classification based on spectral analysis. The motivations behind developing the new model were the failures of the classical spectral cluster analysis based on combinatorial and normalized Laplacian for a set of real-world datasets of textual documents. Reasons of the failures are analysed. While the known methods are all based on usage of eigenvectors of graph Laplacians, a new classification method based on eigenvalues of graph Laplacians is proposed and studied.https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10079090/?tool=EBI |
spellingShingle | Piotr Borkowski Mieczysław A. Kłopotek Bartłomiej Starosta Sławomir T. Wierzchoń Marcin Sydow Eigenvalue based spectral classification PLoS ONE |
title | Eigenvalue based spectral classification |
title_full | Eigenvalue based spectral classification |
title_fullStr | Eigenvalue based spectral classification |
title_full_unstemmed | Eigenvalue based spectral classification |
title_short | Eigenvalue based spectral classification |
title_sort | eigenvalue based spectral classification |
url | https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10079090/?tool=EBI |
work_keys_str_mv | AT piotrborkowski eigenvaluebasedspectralclassification AT mieczysławakłopotek eigenvaluebasedspectralclassification AT bartłomiejstarosta eigenvaluebasedspectralclassification AT sławomirtwierzchon eigenvaluebasedspectralclassification AT marcinsydow eigenvaluebasedspectralclassification |