Interval type-2 approach to kernel possibilistic C-means clustering

摘要: Kernel based fuzzy clustering has been extensively used for pattern sets that have clusters overlap and of different volume. The kernel approach adds additional degree freedom by implicitly mapping input patterns into higher dimensional space known as space. shown to produce improved results over conventional algorithms such C-means (FCM), possibilistic c-means (PCM) (PFCM) not only spherical data but also non sets. However, in the case (KPCM) well PCM, cluster coincidence drawback still exist which poor locations prototypes. In this paper, we propose an interval type-2 (IT2) KPCM overcome problem PCM KPCM. Although choice function can be dependent, use Gaussian our experiments. Using same value variance proposed method outperforms Experimental show validity method.