Kernelized Fuzzy Fisher Criterion based Clustering Algorithm

Fuzzy Fisher Criterion(FFC) based clustering method uses the fuzzy Fisher´s linear discriminant(FLD) as its clustering objective function and is more robust to noises and outliers than fuzzy c-means clustering(FCM). But FFC can only be used in linear separable dataset. In this paper, a novel fuzzy clustering algorithm, called Kernelized Fuzzy Fisher Criterion(KFFC) based clustering algorithm, is proposed. With kernel methods KFFC can perform clustering in kernel feature space while FFC makes clustering in Euclidean space. The experimental results show that the proposed algorithm can deal with the linear non-separable problem better than FFC.