Adaptive Clustering Method Based On The Rank Constraint

Clustering analysis now become one of the important research content in the field of data mining and plays an important role in the identification data,in addition clustering is also applied to economics,biology,psychology,geography,marketing and so on.Therefore,clustering technology in recent years has become a focus on the content of academia,get the attention of scholars.This paper introduces the background and the research status of the data mining technique and clustering technology,summarized the common similarity measurement method and the classical clustering method and analyzed their advantages and disadvantages,In-depth in the similarity matrix and clustering partition,the main content of this paper is as follows:First,clustering divides the data set based on the data of similarity matrix,so the clustering result is highly dependent on the data of similarity matrix,there are many ways of building similarity matrix,the use of more construction method is the Euclidean distance measurement method,but in many cases the Euclidean may cause the measurement result in great error and the final clustering result is not ideal.A density sensitive distance similarity measure is introduced to acquire the similar matrix which can enlarge the distance between the different classes and reduce the distance between the same classes effectively.Second,there will benk possibilities for dividing a clustering with n samples,clustering as a kind of unsupervised learning algorithm,researchers have been concerning how to make the result of clustering better and more accurate,rank constraint is imposed to the Laplacian matrix of the similarity matrix,thus the number of connected area of the similar matrix is equal to the number of clustering,the data can be directly into the right class and take the final clustering result.Experimental results show that the approach can improve the performance of clustering.Each kind of clustering algorithm integrates the researcher's efforts and to some extent changed some shortages of the existing clustering algorithms,although many clustering algorithm has been widely applied,but there are still some shortage,so to explore clustering algorithm with better performance has always been the goal of the researchers.