Um Algoritmo Multi­Protótipo Rápido para Agrupamentos com Diferentes Formatos

Murilo C. NaldiAndré C. P. L. F. de Carvalho

Clustering are advantageous techniques for being able to obtain groups of similar data in an unsupervised fashion. However, different results can be obtained for the same data set, which can result in clusters of different qualities. In addition, practical applications demand scalability and good com­putational performance. The usage of multiple prototypes per cluster is studied in this work, integrated by four distinct metrics. To do this, a low computational cost algorithm was developed and its performance is compared with other te­chniques capable of obtaining groups of different shapes. The obtained results show that the algorithm is fast and robust, being capable to obtain clustering of reasonable quality to distinct types of data sets in low computational time.

Caso o link acima esteja inválido, faça uma busca pelo texto completo na Web: Buscar na Web

Biblioteca Digital Brasileira de Computação - Contato:
     Mantida por: