Kernel based rough fuzzy c-Means clustering optimized using particle swarm optimization

This article presents a new kernelized rough fuzzy c-Means clustering algorithm optimized using particle swarm optimization. The proposed algorithm involves the amalgamation of concepts of rough sets, fuzzy sets, c-Means clustering algorithm with kernel trick optimized using particle swarm optimization (PSO). While the fuzzy set enables efficient handling of overlapping partitions, the concept of rough set deals with uncertainty, vagueness, incompleteness, and indiscernibility in class definition. Whereas, the kernel trick (by projecting the feature space into a higher dimension using an appropriate non-linear mapping function) ensures linear separability of the complex clusters which are otherwise not linearly separable in its original feature space. Finally, the PSO finds the (near) optimum values of the different parameters used in the proposed method. Kernelized Xie-Beni index has been used as the fitness (evaluation) function for the PSO. The effectiveness of the proposed algorithm is evaluated using a number of real life benchmark datasets. Experimental results justify the superiority of the proposed method in comparison to other traditional rough-/fuzzy clustering techniques.