Hierarchical fast clustering algorithm using partial maximum array

摘要

PURPOSE: A method for the fast hierarchical clustering using a partial maximum array is provided to obtain an optimal solution at high speed by using the partial maximum array and rewriting a similarity matrix for a column and a row belonging to a changed cluster from the similarity matrix. CONSTITUTION: The 'i'th cluster is changed by the incorporation or the deletion(301). The similarity of the similarity matrix to be updated is changed(302). In case that the updating value of each similarity matrix is less than the previous maximum, the previous maximum is compared with the value of the updating cluster(304). If the previous maximum is the value of the updating cluster, the new maximum is obtained by searching the corresponding similarity column(305). If not, the maximum is not changed and the present partial maximum array is maintained(315). In case that the updating value of each similarity matrix is equal or larger than the previous maximum, the newly updating value is changed to the maximum(323).