Performance Comparison of Various Robust Data Clustering Algorithms
Shashank Sharma, Megha Goel, Prabhjot Kaur · International Journal of Intelligent Systems and Applications · 2013
Robust clustering techniques are real life clustering techniques for noisy data.They work efficiently in the presence of noise.Fuzzy C-means (FCM) is the first clustering algorith m, based upon fuzzy sets, proposed by J C Bezdek but it does not give accurate results in the presence of noise.In this paper, FCM and various robust clustering algorith ms namely: Possibilistic C-Means (PCM ), Possibilistic Fuzzy Cmeans (PFCM), Credib ilistic Fuzzy C-means (CFCM), Noise Clustering (NC) and Density Oriented Fu zzy C-Means (DOFCM) are studied and compared based upon robust characteristics of a clustering algorith m.For the performance analysis of these algorithms in noisy environment, they are applied on various noisy synthetic data sets, standard data sets like DUNN dataset, Bensaid data set.In co mparison to FCM , PCM, PFCM, CFCM, and NC, DOFCM clustering method identified outliers very well and selected more desirable cluster centroids.