The fuzzy mega-cluster: robustifying FCM by scaling down memberships

Authors:
Amit Banerjee;Rajesh N. Davé
Affiliations:
Department of Mechanical Engineering, New Jersey Institute of Technology, Newark, NJ;Department of Mechanical Engineering, New Jersey Institute of Technology, Newark, NJ
Venue:
FSKD'05 Proceedings of the Second international conference on Fuzzy Systems and Knowledge Discovery - Volume Part I
Year:
2005

Citing 8
Cited 0

Characterization and detection of noise in clustering

Pattern Recognition Letters
Application of the least trimmed squares technique to prototype-based clustering

Pattern Recognition Letters - Special issue on fuzzy set technology in pattern recognition
Pattern Recognition with Fuzzy Objective Function Algorithms

Pattern Recognition with Fuzzy Objective Function Algorithms
A Least Biased Fuzzy Clustering Method

IEEE Transactions on Pattern Analysis and Machine Intelligence
Suppressed fuzzy c-means clustering algorithm

Pattern Recognition Letters
The possibilistic C-means algorithm: insights and recommendations

IEEE Transactions on Fuzzy Systems
Robust clustering methods: a unified view

IEEE Transactions on Fuzzy Systems
Fuzzy order statistics and their application to fuzzy clustering

IEEE Transactions on Fuzzy Systems

Quantified Score

Hi-index	0.00

Visualization

Abstract

A new robust clustering scheme based on fuzzy c-means is proposed and the concept of a fuzzy mega-cluster is introduced in this paper. The fuzzy mega-cluster is conceptually similar to the noise cluster, designed to group outliers in a separate cluster. This proposed scheme, called the mega-clustering algorithm is shown to be robust against outliers. Another interesting property is its ability to distinguish between true outliers and non-outliers (vectors that are neither part of any particular cluster nor can be considered true noise). Robustness is achieved by scaling down the fuzzy memberships, as generated by FCM so that the infamous unity constraint of FCM is relaxed with the intensity of scaling differing across datum. The mega-clustering algorithm is tested on noisy data sets from literature and the results presented.