Back to my home page.

An Abstract Weighting Framework for Clustering Algorithms

Richard Nock and Frank Nielsen

Abstract:
Recent works in unsupervised learning have emphasized the need to understand a new trend in algorithmic design, which is to influence the clustering via weights on the instance points. In this paper, we handle clustering as a constrained minimization of a Bregman divergence. Theoretical results show benefits resembling those of boosting algorithms, and bring new modified weighted versions of clustering algorithms such as k-means, expectation-maximization (EM) and k-harmonic means. Experiments display the quality of the results obtained, and corroborate the advantages that subtle data reweightings may bring to clustering.
 
Key words: Statistical/optimization methods, clustering algorithms.
Download the PDF paper here (10 pages, 5 figures) © SIAM.
 
Bibtex entry:
@InProceedings{nn-awfca-2004,
Author = {Richard Nock and Frank Nielsen},
Title = {An Abstract Weighting Framework for Clustering Algorithms},
Booktitle = {Proceedings of the Fourth International SIAM Conference on Data Mining},
Year = 2004,
Month = {April},
Publisher = {SIAM},
Address = "Orlando, FL, USA",
Pages = "200-209"
}


© Copyright notice.
Last updated, 2004.