Self-organizing maps clustering based on continuous multiresolution entropy

TORRES, H.M, GURLEKIAN, J.A., RUFINER, H.L. Y TORRES, M.E.

PHYSICA A - STATISTICAL AND THEORETICAL PHYSICS

Elsevier

Lugar: Amsterdam; Año: 2006 vol. 359C p. 659 - 676

0378-4371

The detection of changes in the parameter values of a non-lineal dynamic system is a branch of study with multiple applications. In this paper, we explored a variant of an automatic detector of slight parameters variations in  non-lineal dynamic systems proposed by Torres et al. The new method takes the advantages of the continuous multiresolution entropy in order to show slight changes in the parameters and uses self organized maps to quantify and cluster such changes. We discuss the performance of this method while applied to automatic segmentation of natural and synthetic dipthongs in the presence of additive noise. Our results shows the potentiality of the proposed method.