Possibilistic clustering using non-Euclidean distance

This paper presents a novel fuzzy clustering algorithm called novel possibilistic c-means (NPCM) clustering algorithm. Possibilistic c-means model (PCM) has been proposed by Krishnapuram and Keller to resist noises. It is claimed that NPCM is the extension of PCM by introducing a non-Euclidean distance into PCM to replace the Euclidean distance used in PCM. Based on robust statistical point of view and influence function, the non-Euclidean distance is more robust than the Euclidean distance. So the NPCM algorithm is more robust than PCM. Moreover, with the new distance NPCM can deal with noises or outliers better than PCM and fuzzy c-means (FCM). The experimental results show the better performance of NPCM.