Research On Improvement Of Clustering By Fast Search And Find Of Density Peaks And Differential Privacy Protection

Clustering by fast search and find of density peaks(DPC)is a density-based clustering algorithm with unique parameters,simple algorithm without iteration,and suitable for arbitrary shape of clusters.However,DPC still has some shortcomings,such as the selection of clustering center requires human participation.In addition,people began to pay attention to privacy,and privacy protection technology was proposed.Differential privacy model has the advantages of strict mathematical theoretical basis,and has become a hot research direction of privacy protection technology.At present,in the research of clustering algorithm based on differential privacy,the performance of clustering algorithm is reduced due to the added noise and the sensitivity of algorithm parameters.How to make the algorithm not only protect sensitive information but also ensure the availability of data is the key problem to be solved.In view of the above problems,the main research work of this paper is as follows:(1)Adaptive clustering by fast search and find of density peaks(AdDPC)is proposed.According to the characteristic that the tradictional DPC needs human participation to select the clustering center,this paper improves the clustering center selection strategy of the algorithm.The weighted factor is introduced to select clustering centers by using the slope change of decision metric,so that the algorithm can select clustering centers adaptively.Experimental results show that AdDPC can adapt and select the clustering center correctly,and the clustering performance is improved.(2)A differential privacy protection scheme based on adaptive clustering by fast search and find of density peaks is prosposed.AdDPC is analyzed for privacy leakage,and the differential privacy protection technology is introduced into the calculation of AdDPC local density and distance,so that the algorithm can meet the requirements of differential privacy,and the algorithm is theoretically analyzed.Finally,through the experimental comparison,it is found that the security and clustering availability of the algorithm are guaranteed.(3)An improved adaptive clustering by fast search and find of density peaks based on differential privacy with rho is proposed.To solve the problem of low clustering accuracy of differential privacy protection scheme based on adaptive clustering by fast search and find of density peaks,this paper introduced the definition of reachability and improves the allocation stragety of non-cluster centers of algorithm.The points with reachability but not assigned to the same cluster are classified into one cluster.Experiments show that the clustering performance of the algorithm has been improved.