A DISCREPANCY MEASURE FOR IMPROVED CLUSTERING

A discrepancy measure is proposed to improve the clustering of pattern s 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 betwee n the patterns. The K-means clustering algorithm is modified to use th e discrepancy measure to compute the similarity between patterns and t he cluster centers. A series of clustering experiments were conducted on identical data using the modified and standard K-means algorithm. T he results obtained show that clustering performance of the modified a lgorithm is significantly superior to that of the standard algorithm.