Enhanced subspace clustering through combining Minkowski distance and Cosine dissimilarity

Abstract

In view of intelligent Minkowski metric Weighted K-means (iMWK) sensitive to feature weighting, a novel clustering technique called intelligent Minkowski metric feature weights subspace clustering algorithms through hybrid dissimilarity measure (iMWK-HD) is presented. First, a new optimization objective function is constructed by incorporating the Minkowski distance and Cosine dissimilarity in the subspace. Based on this objective function, the corresponding update rules for clustering are then derived, followed by the development of the novel iMWK-HD algorithm. The properties of this algorithm are investigated and the performance is evaluated experimentally using synthetic and UCI datasets. The experimental studies demonstrate that the accuracy of the proposed iMWK-HD algorithm outperforms three existing clustering algorithms, i.e., iK-means, iWK-means and iMWK-means. In addition, the proposed algorithms are immune to irrelevant features in cluster subspace.