Title: Application of Self-Organizing Maps for clustering DJIA and NASDAQ100 portfolios

Abstract: In this paper we apply the Self-Organized Map (SOM) method for clustering the DJIA and NASDAQ100 portfolios for determination of non-linear correlations between stocks. We represent the application of this method as alternative to ultrametric spaces method. We have found, that SOM method is more relevant and perspective for clustering ill-structured large databases and, in particular, NASDAQ100, where nonlinear processing of the large data samples is required.