Analysis And Research On Clustering Algorithm Based On Ant Colony Optimization

With the rapid development of computer science and technology, especially the wideapplication of database, data mining has become one of the hot research fields. In the massiveof data field, there is much important valuable information, so that effective data analysis isnecessary for useful information extraction.Clustering analysis is an important tool of datamining and has a profound influence on the social economic development and people’s dailylife.This paper mainly discusses clustering algorithms based on ant colony optimization, andresearch aspects are as followed:Firstly, it analyzes the advantages and disadvantages of the K-medoids clusteringalgorithm and proposes a K-medoids clustering algorithm based on ant colony algorithm(ACO). As a bionic optimization algorithm, ACO has strong global searching ability and highsolution efficiency, as well as strong robustness. Therefore, the algorithm based on ACOproposed this paper improves the accuracy and steadiness of clustering. Simulationexperiments show that this algorithm is efficient and feasible.Secondly, a K-means algorithm based on rough set and ACO is put forward with thefusion of rough set theory, ACO and K-means algorithm. It optimizes the K-means with ACO,then K-means cluster number and initial cluster centers can be obtained dynamically with theprinciple of maximum minimum, meanwhile processes boundary objects with upper andlower approximation of rough set theory. Compared with other algorithms, this algorithm hashigher accuracy rate, faster execution speed and much more steady performance.Thirdly, a kind of K-medoids clustering algorithm based on differential evolution isproposed to improve K-medoids clustering with the problem of slow convergence. Asdifferential evolution is of heuristic global search ability and strong robustness, this papercombines with K-medoids clustering of high efficiency and DE of global optimization ability,so that it can overcome the defects of K-medoids clustering, improve the global search abilityand formal clustering quality, as well as shorten the convergence time. Experimental resultsindicate that the algorithm is of high stability and convergence speed is improved.Finally, according to the characteristics analysis of the cloud computing environment,this paper applies the algorithm of ACO-K medoids into cloud computing environment withresource allocation and optimization. The algorithm is able to find routes effectively andproperly in cloud computing, reduce load dynamicly with consideration of global loadbalancing to obtain the optimal computing resources and improve efficiency of cloud computing. Based on stimulation experiments in the end of paper, it analyzes the networkbandwidth, node time delay and other factors to affect the results of resource allocation, andverifies the algorithm of high efficiency in the cloud computing.