Probability of large-scale data set EM clustering algorithms based on partial information constraints

The current situation, the need for clustering of data is very large, and the use of traditional algorithm for clustering process often tedious and time consuming is very long, the effect is not obvious. Based on this, this paper proposes a data sets EM probability based on some constraint information clustering algorithm, the detailed implementation process of the whole algorithm is described. Through experiment contrast scalable EM, positive_PC_SEM and full_PC_SEM clustering quality and efficiency of execution of the algorithm, the results show that the positive_PC_SEM algorithm and scalable EM algorithm compared to the clustering quality and efficiency is higher, although full_PC_SEM clustering quality is very high, but requires a lot of time.