An efficient k-means clustering filtering algorithm using density based initial cluster centers

摘要

Highlights ? A density based initialization method of linear time complexity is proposed. ? Proposed method improves the performance of k-means filtering method for high dimensional data and poorly separated cluster centers. ? Our method is deterministic and robust to noise in the dataset. ? Ideally suited to filtering algorithm as kd-tree constructed by filtering method is exploited for finding density areas. Abstract 展开▼