Research And Implementation Of KFCM Algorithm Based On Bat Algorithm

With the advent of the era of data explosion,how to efficiently and quickly mine out the valuable information from redundant data has become more and more important.Cluster analysis is a data mining technique with no label and no supervised learning,which can find valuable information in the data at low cost.Fuzzy C-means algorithm(FCM)is a typical algorithm in cluster analysis.It adds fuzzy set theory on the basis of K-means algorithm and has been widely used in various fields.However,Fuzzy C-means algorithm is sensitive to the initial clustering center and not robust.To solve these problems,this paper proposes an improved kernel fuzzy c-means algorithm(IBA-KFCM).This algorithm is based on the optimized bat algorithm(IBA),which uses the improved bat algorithm IBA's global optimization characteristics to find the most location and use it as the initial clustering center of KFCM algorithm to reduce the impact of clustering results on the initial center.At the same time,based on the objective function of KFCM algorithm,sample weight is added in this paper to reduce the influence of noise points and outliers on the clustering effect and improve the robustness of KFCM algorithm.Then,FCM,KFCM and IBA-KFCM algorithm proposed in this paper were tested on UCI data set and the experimental results were analyzed and compared.Finally,the improved IBA-KFCM algorithm is applied to text clustering,and the feasibility and effectiveness of the improved algorithm are verified by comparing and analyzing the experimental results.The main research and work of this paper is as follows:(1)The standard bat algorithm is improved by adding the local search strategy of speed weight and curve decline,so that the basic bat algorithm can overcome the problems that it is easy to fall into the local extreme value and the low convergence accuracy in the later stage.The improved IBA algorithm has better global optimization ability and higher post-convergence accuracy than the basic algorithm.(2)The optimal position obtained by IBA algorithm is regarded as the initial clustering center point of KFCM algorithm,which effectively improves the dependence of clustering results of KFCM algorithm on the selection of initial clustering center;At the same time,the weight coefficient of sample points is added to the objective function of KFCM algorithm to reduce the influence of the algorithm on the sample outliers and noise points and improve the robustness of the algorithm.(3)IBA-KFCM algorithm was applied to UCI data set and text clustering,and the experimental results were analyzed and compared to verify the feasibility and effectiveness of the improved algorithm proposed in this paper,which improved the clustering quality and algorithm performance of KFCM algorithm.