Title: Fuzzy clustering of networks
Authors: Ilaria Bombelli - Sapienza University of Rome (Italy) [presenting]
Ichcha Manipur - ICAR-CNR (Italy)
Mario Guarracino - ICAR-CNR, Naples (Italy)
Maria Brigida Ferraro - Sapienza University of Rome (Italy)
Abstract: Networks represent a powerful model to describe problems and applications in various fields, such as economics, science and technology. The focus is on the fuzzy clustering of networks. In detail, we provide computational procedures to look for clusters of networks, where each network represents an object. Our proposal is based on the Non-Euclidean Fuzzy Relational Clustering (NEFRC) algorithm. Since the algorithm requires as input a distance matrix, we need some specific measures of distance between networks. First of all, we represent each network using probability distributions (Node Distance Distribution and Transition Matrices) to obtain a matrix representation. Then we use a measure of dissimilarity between probability distributions, and we get the networks distance matrix the NEFRC algorithm requires as input. We check the adequacy of the proposals through simulations and real-case studies.