The SpectACl of nonconvex clustering: a spectral approach to density-based clustering.

Sibylle Hess, Wouter Duivesteijn, Philipp Honysz, Katharina Morik

Onderzoeksoutput: Hoofdstuk in Boek/Rapport/CongresprocedureHoofdstukAcademicpeer review

12 Downloads (Pure)

Uittreksel

When it comes to clustering nonconvex shapes, two paradigms are used to find the most suitable clustering: minimum cut and maximum density. The most popular algorithms incorporating these paradigms are Spectral Clustering and DBSCAN.Both paradigms have their pros and cons. While minimum cut clusterings are sensitive to noise, density-based clusterings have trouble handling clusters with varying densities. In this paper, we propose SPECTACL: a method combining the ad-vantages of both approaches, while solving the two mentioned drawbacks. Our method is easy to implement, such as spectral clustering, and theoretically founded to optimize a proposed density criterion of clusterings. Through experiments on synthetic and real-world data, we demonstrate that our approach provides robust and reliable clusterings.
Originele taal-2Engels
TitelProceedings of 33rd AAAI Conference on Artificial IntelligenceAAAI
UitgeverijAssociation for the Advancement of Artificial Intelligence
Aantal pagina's27
StatusGepubliceerd - 2019
Evenement33rd AAAI Conference on Artificial Intelligence - Hawaii, Honolulu, Verenigde Staten van Amerika
Duur: 27 jan 20191 feb 2019
Congresnummer: 33
https://aaai.org/Conferences/AAAI-19/

Congres

Congres33rd AAAI Conference on Artificial Intelligence
Verkorte titelAAAI-19
LandVerenigde Staten van Amerika
StadHonolulu
Periode27/01/191/02/19
Internet adres

Vingerafdruk

Experiments

Citeer dit

Hess, S., Duivesteijn, W., Honysz, P., & Morik, K. (2019). The SpectACl of nonconvex clustering: a spectral approach to density-based clustering. In Proceedings of 33rd AAAI Conference on Artificial IntelligenceAAAI Association for the Advancement of Artificial Intelligence.
Hess, Sibylle ; Duivesteijn, Wouter ; Honysz, Philipp ; Morik, Katharina. / The SpectACl of nonconvex clustering: a spectral approach to density-based clustering. Proceedings of 33rd AAAI Conference on Artificial IntelligenceAAAI. Association for the Advancement of Artificial Intelligence, 2019.
@inbook{a986d4e9a35145f1a9e1cc9abaf9a846,
title = "The SpectACl of nonconvex clustering: a spectral approach to density-based clustering.",
abstract = "When it comes to clustering nonconvex shapes, two paradigms are used to find the most suitable clustering: minimum cut and maximum density. The most popular algorithms incorporating these paradigms are Spectral Clustering and DBSCAN.Both paradigms have their pros and cons. While minimum cut clusterings are sensitive to noise, density-based clusterings have trouble handling clusters with varying densities. In this paper, we propose SPECTACL: a method combining the ad-vantages of both approaches, while solving the two mentioned drawbacks. Our method is easy to implement, such as spectral clustering, and theoretically founded to optimize a proposed density criterion of clusterings. Through experiments on synthetic and real-world data, we demonstrate that our approach provides robust and reliable clusterings.",
keywords = "spectral clustering, Eigendecomposition, clustering, nonconvex clustering, minimum cut",
author = "Sibylle Hess and Wouter Duivesteijn and Philipp Honysz and Katharina Morik",
year = "2019",
language = "English",
booktitle = "Proceedings of 33rd AAAI Conference on Artificial IntelligenceAAAI",
publisher = "Association for the Advancement of Artificial Intelligence",

}

Hess, S, Duivesteijn, W, Honysz, P & Morik, K 2019, The SpectACl of nonconvex clustering: a spectral approach to density-based clustering. in Proceedings of 33rd AAAI Conference on Artificial IntelligenceAAAI. Association for the Advancement of Artificial Intelligence, Honolulu, Verenigde Staten van Amerika, 27/01/19.

The SpectACl of nonconvex clustering: a spectral approach to density-based clustering. / Hess, Sibylle; Duivesteijn, Wouter; Honysz, Philipp ; Morik, Katharina.

Proceedings of 33rd AAAI Conference on Artificial IntelligenceAAAI. Association for the Advancement of Artificial Intelligence, 2019.

Onderzoeksoutput: Hoofdstuk in Boek/Rapport/CongresprocedureHoofdstukAcademicpeer review

TY - CHAP

T1 - The SpectACl of nonconvex clustering: a spectral approach to density-based clustering.

AU - Hess, Sibylle

AU - Duivesteijn, Wouter

AU - Honysz, Philipp

AU - Morik, Katharina

PY - 2019

Y1 - 2019

N2 - When it comes to clustering nonconvex shapes, two paradigms are used to find the most suitable clustering: minimum cut and maximum density. The most popular algorithms incorporating these paradigms are Spectral Clustering and DBSCAN.Both paradigms have their pros and cons. While minimum cut clusterings are sensitive to noise, density-based clusterings have trouble handling clusters with varying densities. In this paper, we propose SPECTACL: a method combining the ad-vantages of both approaches, while solving the two mentioned drawbacks. Our method is easy to implement, such as spectral clustering, and theoretically founded to optimize a proposed density criterion of clusterings. Through experiments on synthetic and real-world data, we demonstrate that our approach provides robust and reliable clusterings.

AB - When it comes to clustering nonconvex shapes, two paradigms are used to find the most suitable clustering: minimum cut and maximum density. The most popular algorithms incorporating these paradigms are Spectral Clustering and DBSCAN.Both paradigms have their pros and cons. While minimum cut clusterings are sensitive to noise, density-based clusterings have trouble handling clusters with varying densities. In this paper, we propose SPECTACL: a method combining the ad-vantages of both approaches, while solving the two mentioned drawbacks. Our method is easy to implement, such as spectral clustering, and theoretically founded to optimize a proposed density criterion of clusterings. Through experiments on synthetic and real-world data, we demonstrate that our approach provides robust and reliable clusterings.

KW - spectral clustering

KW - Eigendecomposition

KW - clustering

KW - nonconvex clustering

KW - minimum cut

M3 - Chapter

BT - Proceedings of 33rd AAAI Conference on Artificial IntelligenceAAAI

PB - Association for the Advancement of Artificial Intelligence

ER -

Hess S, Duivesteijn W, Honysz P, Morik K. The SpectACl of nonconvex clustering: a spectral approach to density-based clustering. In Proceedings of 33rd AAAI Conference on Artificial IntelligenceAAAI. Association for the Advancement of Artificial Intelligence. 2019