A Novel Kernel Based Fuzzy C Means Clustering With Cluster Validity Measures?

Clustering algorithms are an integral part of both computational intelligence and pattern recognition. It is unsupervised methods for classifying data into subgroups with similarity based inter cluster and intra cluster. In fuzzy clustering algorithms, mainly used algorithm is Fuzzy c-means (FCM) algorithm. This FCM algorithm is efficient only for spherical data when the input of the data structure is not spherical or complex this method is unsuccessful. For this, modification of the FCM is done by the labeling of a pixel to be partial by the labels in its immediate neighborhood and this modification is called BCFCM (Bias-Corrected FCM). Since it is computationally time taking and lacks enough robustness to noise for that kernel versions of FCM with spatial constraints, such as KFCM, were proposed to solve the drawbacks of BCFCM. In this paper, a novel kernel-based fuzzy C-means clustering algorithm (KFCM) is proposed for clustering. It is recognized by replacing the kernel-induced distance metric over the original Euclidean distance, and the corresponding algorithms are called kernel fuzzy c-means (KFCM) algorithm. The experimental results shows that proposed clustering technique provides better accuracy with less error rate than the BCFCM algorithm.