FCM Clustering from the View Point of Iteratively Reweighted Least Squares

Author

Ichihashi, Hidetomo ; Honda, Katsuhiro

Author_Institution

Dept. of Ind. Eng., Electr. Eng. & Inf. Sci., Osaka Prefecture Univ.

fYear

2005

fDate

25-25 May 2005

Firstpage

873

Lastpage

878

Abstract

By alleviating theoretical strictness, a broad class of membership functions can be used in fuzzy c-means (FCM) clustering from the viewpoint of iteratively reweighted least-squares (IRLS) techniques. Clustering characteristics of regular FCM, entropy regularized FCM or deterministic annealing by Rose and our proposed IRLS approaches are compared by using 3D graphics and contour maps. Though an in-depth analysis of theoretical aspect is beyond the scope of this paper, numerical comparisons reveal that IRLS algorithm using different membership functions do not share the same property with the regular FCM and entropy regularized FCM or DA

Keywords

entropy; fuzzy set theory; least squares approximations; pattern clustering; simulated annealing; unsupervised learning; 3D graphics; FCM clustering; IRLS algorithm; contour maps; deterministic annealing; entropy regularized FCM; fuzzy c-means clustering; iteratively reweighted least squares; membership functions; theoretical strictness; Algorithm design and analysis; Annealing; Clustering algorithms; Entropy; Graphics; Industrial engineering; Information science; Least squares methods; Noise robustness; Unsupervised learning;

fLanguage

English

Publisher

ieee

Conference_Titel

Fuzzy Systems, 2005. FUZZ '05. The 14th IEEE International Conference on

Conference_Location

Reno, NV

Print_ISBN

0-7803-9159-4

Type

conf

DOI

10.1109/FUZZY.2005.1452509

Filename

1452509