K-means algorithm in the optimal initial centroids based on dissimilarity
dissimilarity to reflect the degree of correlation between data objects, and then uses Huffman tree to find the initial centroids. Many experiments confirm that the proposed algorithm is an efficient

