We motivate and develop a natural bicriteria measure for assessing the quality of a clustering that avoids the drawbacks of existing measures. A simple recursive heuristic is shown to have poly-logarithmic worst-case guarantees under the new measure. The main result of the article is the analysis of a popular spectral algorithm. One variant of spectral clustering turns out to have effective worst-case guarantees; another finds a "good" clustering, if one exists.
%0 Journal Article
%1 kannan2004clusterings
%A Kannan, Ravi
%A Vempala, Santosh
%A Vetta, Adrian
%C New York, NY, USA
%D 2004
%I ACM
%J J. ACM
%K clustering community evaluation
%N 3
%P 497--515
%R 10.1145/990308.990313
%T On clusterings: Good, bad and spectral
%U http://portal.acm.org/citation.cfm?id=990313
%V 51
%X We motivate and develop a natural bicriteria measure for assessing the quality of a clustering that avoids the drawbacks of existing measures. A simple recursive heuristic is shown to have poly-logarithmic worst-case guarantees under the new measure. The main result of the article is the analysis of a popular spectral algorithm. One variant of spectral clustering turns out to have effective worst-case guarantees; another finds a "good" clustering, if one exists.
@article{kannan2004clusterings,
abstract = {We motivate and develop a natural bicriteria measure for assessing the quality of a clustering that avoids the drawbacks of existing measures. A simple recursive heuristic is shown to have poly-logarithmic worst-case guarantees under the new measure. The main result of the article is the analysis of a popular spectral algorithm. One variant of spectral clustering turns out to have effective worst-case guarantees; another finds a "good" clustering, if one exists.},
added-at = {2010-07-26T18:43:13.000+0200},
address = {New York, NY, USA},
author = {Kannan, Ravi and Vempala, Santosh and Vetta, Adrian},
biburl = {https://www.bibsonomy.org/bibtex/2e861c1d63015e10120909af68d34f766/folke},
description = {On clusterings},
doi = {10.1145/990308.990313},
interhash = {a1b52b044b404dec2ffc8f1089760d88},
intrahash = {e861c1d63015e10120909af68d34f766},
issn = {0004-5411},
journal = {J. ACM},
keywords = {clustering community evaluation},
number = 3,
pages = {497--515},
publisher = {ACM},
timestamp = {2011-09-13T22:35:47.000+0200},
title = {On clusterings: Good, bad and spectral},
url = {http://portal.acm.org/citation.cfm?id=990313},
volume = 51,
year = 2004
}