A discrepancy measure for improved clustering

摘要

A discrepancy measure is proposed to improve the clustering of patterns which experience non-linear distortions. The discrepancy measure is an outcome of a non-linear alignment procedure which optimally aligns the elements of patterns in order to minimize the dissimilarity between the patterns. The K-means clustering algorithm is modified to use the discrepancy measure to compute the similarity between patterns and the cluster centers. A series of clustering experiments were conducted on identical data using the modified and standard K-means algorithm. The results obtained show that clustering performance of the modified algorithm is significantly superior to that of the standard algorithm.