Ana Estela Antunes da Silva, Karina E. Barreto Kinderé Cunha, Fernando J. Von Zuben, Luiz Eduardo Galvão Martins.
This paper analysis the representation power of the U_Matrix for data clustering basing on the Kohonen Self-Organizing Map (SOM). The application of this algorithm preserves topological relations from a set of input sample data when these are mapped into a neural network with neighborhood preservation. Although there may be a decrease of the dimension from the input space to the output space, the goal of the mapping is minimize the information loss and indicate the formation of classes in which the input data can be classified. The function of the U_Matrix is to graphically demonstrate which are such clusters and the number of clusters. Clique no link abaixo para buscar o texto completo deste trabalho na Web: Buscar na Web