Artificial Colony Clustering Algorithm based on Global Information

An improved artificial bee colony algorithm was proposed for data clustering aiming at overcoming the shortcomings of being trapped in local optimum and slow convergence rate in this paper, called artificial colonyclustering algorithm based on global information (GI-ABC). On the one hand, the modified algorithm added the average richness of food to the updating equation of food source, taking advantage of the intermediate clustering consequence to update the food source better. On the other hand, it utilized global information to improve the search efficiency of the onlooker bees, which contributed to getting the global optimum. Meanwhile, abundant experiments were conducted to evaluate it. In the experiments, four of topical real data sets selected from the UCI Machine Learning Repository were used to test the performance of the strategies compared with the other clustering algorithms, such as ABC algorithm, Particle Swarm Optimization (PSO) algorithm and K-means algorithm. The results indicate that the modified algorithm can generate better results than other algorithms and has better general performance.