Artificial Bee Colony Clustering Algorithm With K-harmonic Means

Data mining is the process of a large number of incomplete, noise and random data access potentially useful information and knowledge. As a big part in data mining, cluster analysis is an unsupervised learning method does not require a priori knowledge of the relevant data sets. Aggregated into categories on the basis of the clustering algorithm, something of the nature of things, making things in a different class, different things in the same category as far as possible, as similar as possible. Cluster analysis has been widely used in all areas of life.K-means algorithm is a typical clustering method, due to its simple, efficient, and thus is widely used. The algorithm is heavily dependent on the choice of the initial focal point, it is easy to fall into local optimal results and create other problems. Some scholars have pointed to reconcile the K-means clustering (KHM) algorithm in1999, although it reduces the dependence of the K-means algorithm to select the initial center point, but the algorithm is easy to fall into local optimal results, and prior to a given cluster number. Take into account the above problems, this paper presents a new algorithm ABCKHM. Human workers swarm the main features of the algorithm is not necessary to know the specific advantages and disadvantages, is a heuristic global optimization algorithm is robust, easy combination with other algorithms.The artificial bee colony algorithm of the new algorithm makes full use of the advantages of the KHM clustering algorithm. Firstly, with global searching the artificial bee colony algorithm have done, the new algorithm combining ABC and KHM can better avoid trapping into local optimization. In ABC Algorithm part, a new destination function is design for single food source to guide searching process.Experiments indicate that our new combined algorithm outperform some classic clustering algorithm such as the K-means, KHM.