A KFCM Algorithm Based on Improved Artificial Bee Colony Algorithm

Kernel fuzzy C-mean clustering (KFCM) algorithm is effective for high-dimensional data, but this algorithm has some defects of sensitivity to initialization and local optima. Artificial Bee Colony (ABC) algorithm is based on intelligent behaviors of honey bee swarm. It has the properties of strong global optimization and fast convergence speed. A KFCM algorithm based on improved ABC is proposed in this paper. In order to improve search efficiency and reduce local optima, small interval generation method is used to make initial colony more symmetrical, and roulette is replaced by Boltzmann selection mechanism. The experimental results show the proposed algorithm is more accurate in clustering and less iterations than FCM and KFCM clustering algorithm for data of large cluster number and high dimension.