A1376
Title: Fuzzy clustering approaches for star-shaped sets: A comparative study
Authors: Ana Belen Ramos-Guajardo - Fundacion Universidad de Oviedo (Spain) [presenting]
Maria Brigida Ferraro - Sapienza University of Rome (Italy)
Gil Gonzalez-Rodriguez - University of Oviedo (Spain)
Jose Grana Colubi - University of Oviedo (Spain)
Abstract: The aim is to develop a novel method to identify groups of star-shaped sets in Rp. These sets are characterized by a central point and a radial function that accounts for directional inaccuracy around that center. To achieve this, a new fuzzy clustering algorithm based on the Mahalanobis distance is proposed, incorporating the covariance matrices associated with each cluster. The use of the Mahalanobis distance in clustering has been demonstrated to effectively identify non-spherical clusters, which are often overlooked when employing a Euclidean-type distance. A comparative analysis is conducted between this method and a previously proposed generalization of the fuzzy k-means algorithm for star-shaped sets, focusing on their performance in a real-life application involving clasts from the Cantabrian Coast.
