Output details

Minkowski metric, feature weighting and anomalous cluster initializing in K-Means clustering

<12> In this paper we present an extension of K-Means addressing one of its major drawbacks: its lack of defence against noisy features. We do so by adding cluster dependent feature weights to its criterion and introducing the use of the Minkowski metric in clustering. The latter allows feature weights to become intuitively appealing feature rescaling factors, which could be used in the data pre-processing for other algorithms. We demonstrate the superiority of our method by comparing it with others on popular datasets and generated sets of Gaussian clusters, both as they are and with additional uniform random noise features.