Soft kernel spectral clustering

Rocco Langone, RaghvenPhDa Mall, Johan A.K. Suykens

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

14 Citations (Scopus)

Abstract

In this paper we propose an algorithm for soft (or fuzzy) clustering. In soft clustering each point is not assigned to a single cluster (like in hard clustering), but it can belong to every cluster with a different degree of membership. Generally speaking, this property is desirable in order to improve the interpretability of the results. Our starting point is a state-of-the art technique called kernel spectral clustering (KSC). Instead of using the hard assignment method present therein, we suggest a fuzzy assignment based on the cosine distance from the cluster prototypes. We then call the new method soft kernel spectral clustering (SKSC). We also introduce a related model selection technique, called average membership strength criterion, which solves the drawbacks of the previously proposed method (namely balanced linefit). We apply the new algorithm to synthetic and real datasets, for image segmentation and community detection on networks. We show that in many cases SKSC outperforms KSC.

Original languageEnglish
Title of host publication2013 International Joint Conference on Neural Networks, IJCNN 2013
DOIs
Publication statusPublished - 1 Dec 2013
Externally publishedYes
Event2013 International Joint Conference on Neural Networks, IJCNN 2013 - Dallas, TX, United States
Duration: 4 Aug 20139 Aug 2013

Other

Other2013 International Joint Conference on Neural Networks, IJCNN 2013
CountryUnited States
CityDallas, TX
Period4/8/139/8/13

Fingerprint

Fuzzy clustering
Image segmentation

ASJC Scopus subject areas

  • Software
  • Artificial Intelligence

Cite this

Langone, R., Mall, R., & Suykens, J. A. K. (2013). Soft kernel spectral clustering. In 2013 International Joint Conference on Neural Networks, IJCNN 2013 [6706850] https://doi.org/10.1109/IJCNN.2013.6706850

Soft kernel spectral clustering. / Langone, Rocco; Mall, RaghvenPhDa; Suykens, Johan A.K.

2013 International Joint Conference on Neural Networks, IJCNN 2013. 2013. 6706850.

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

Langone, R, Mall, R & Suykens, JAK 2013, Soft kernel spectral clustering. in 2013 International Joint Conference on Neural Networks, IJCNN 2013., 6706850, 2013 International Joint Conference on Neural Networks, IJCNN 2013, Dallas, TX, United States, 4/8/13. https://doi.org/10.1109/IJCNN.2013.6706850
Langone R, Mall R, Suykens JAK. Soft kernel spectral clustering. In 2013 International Joint Conference on Neural Networks, IJCNN 2013. 2013. 6706850 https://doi.org/10.1109/IJCNN.2013.6706850
Langone, Rocco ; Mall, RaghvenPhDa ; Suykens, Johan A.K. / Soft kernel spectral clustering. 2013 International Joint Conference on Neural Networks, IJCNN 2013. 2013.
@inproceedings{a21d574517d344e09c1f20760d09722f,
title = "Soft kernel spectral clustering",
abstract = "In this paper we propose an algorithm for soft (or fuzzy) clustering. In soft clustering each point is not assigned to a single cluster (like in hard clustering), but it can belong to every cluster with a different degree of membership. Generally speaking, this property is desirable in order to improve the interpretability of the results. Our starting point is a state-of-the art technique called kernel spectral clustering (KSC). Instead of using the hard assignment method present therein, we suggest a fuzzy assignment based on the cosine distance from the cluster prototypes. We then call the new method soft kernel spectral clustering (SKSC). We also introduce a related model selection technique, called average membership strength criterion, which solves the drawbacks of the previously proposed method (namely balanced linefit). We apply the new algorithm to synthetic and real datasets, for image segmentation and community detection on networks. We show that in many cases SKSC outperforms KSC.",
author = "Rocco Langone and RaghvenPhDa Mall and Suykens, {Johan A.K.}",
year = "2013",
month = "12",
day = "1",
doi = "10.1109/IJCNN.2013.6706850",
language = "English",
isbn = "9781467361293",
booktitle = "2013 International Joint Conference on Neural Networks, IJCNN 2013",

}

TY - GEN

T1 - Soft kernel spectral clustering

AU - Langone, Rocco

AU - Mall, RaghvenPhDa

AU - Suykens, Johan A.K.

PY - 2013/12/1

Y1 - 2013/12/1

N2 - In this paper we propose an algorithm for soft (or fuzzy) clustering. In soft clustering each point is not assigned to a single cluster (like in hard clustering), but it can belong to every cluster with a different degree of membership. Generally speaking, this property is desirable in order to improve the interpretability of the results. Our starting point is a state-of-the art technique called kernel spectral clustering (KSC). Instead of using the hard assignment method present therein, we suggest a fuzzy assignment based on the cosine distance from the cluster prototypes. We then call the new method soft kernel spectral clustering (SKSC). We also introduce a related model selection technique, called average membership strength criterion, which solves the drawbacks of the previously proposed method (namely balanced linefit). We apply the new algorithm to synthetic and real datasets, for image segmentation and community detection on networks. We show that in many cases SKSC outperforms KSC.

AB - In this paper we propose an algorithm for soft (or fuzzy) clustering. In soft clustering each point is not assigned to a single cluster (like in hard clustering), but it can belong to every cluster with a different degree of membership. Generally speaking, this property is desirable in order to improve the interpretability of the results. Our starting point is a state-of-the art technique called kernel spectral clustering (KSC). Instead of using the hard assignment method present therein, we suggest a fuzzy assignment based on the cosine distance from the cluster prototypes. We then call the new method soft kernel spectral clustering (SKSC). We also introduce a related model selection technique, called average membership strength criterion, which solves the drawbacks of the previously proposed method (namely balanced linefit). We apply the new algorithm to synthetic and real datasets, for image segmentation and community detection on networks. We show that in many cases SKSC outperforms KSC.

UR - http://www.scopus.com/inward/record.url?scp=84893555918&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84893555918&partnerID=8YFLogxK

U2 - 10.1109/IJCNN.2013.6706850

DO - 10.1109/IJCNN.2013.6706850

M3 - Conference contribution

SN - 9781467361293

BT - 2013 International Joint Conference on Neural Networks, IJCNN 2013

ER -