The Adaptive Soft Subspace Clustering Algorithm About The Distance Between Clusters

Because the soft subspace clustering algorithm based on k-means framework is sensitive to the initial clustering centers,and the effect of the distance between clusters is uncertain,this paper proposes the adaptive soft subspace clustering algorithm about the distance between clusters.The algorithm firstly combines the principle of the maximum and minimum and the method of the minimum sum of square distance in data subsets to obtain the more accurate initial clustering centers.Then,according to the relationship of the compactness within a cluster measuring the correlation between the samples in the cluster and the distance between clusters measuring the independence of the samples in different clusters,define the control coefficient using the ratio of the distance between clusters and the sum of the distance between clusters and the compactness within a cluster,so it can adjust itself according to the last clustering result,and the algorithm realize adoptive operation.At last,The two indexes of mutual information NMI and RI index are used to analyze the results of cluster classification,and experiments show that the improved algorithm has better clustering performance in this paper.