OPTIMIZACIÓN BASADA EN ENJAMBRES DE PARTÍCULAS PARA AJUSTAR LOS PARÁMETROS DE LOS MÉTODOS SCAN

Abstract

Classic and fuzzy Scan methods are widely used for cluster detection over a lineal or circular data sequence. This sequence has been previously processed and transformed into a binary sequence. The value number one represents the category that is of interest and the zero value represents the rest. The objective of these methods is to detect a cluster of ones. These methods depend on some parameters: the width of the mobile window that covers the entrance sequence, the step of the window's movement and the size of the fuzzy part in the Fuzzy Scan methods. In this paper it is shown how with the combination of a bioinspired algorithm (particle swarm optimization) is useful to find the adequate values of the parameters of the Scan techniques. Also, a Bioinformatics problem is solved as an example.