An application of the self-organizing map to multiple view unsupervised learning

Abstract

In various data mining applications performing the task of extracting information from large databases is serious problem, which occurs in many fields e.g.: bioinformatics, commercial behaviour of Internet users, social networks analysis, management and investigation of various databases in static or dynamic states. In recent years many techniques discovering hidden structures in the data set like clustering and projection of data from high-dimensional spaces have been developed. In this paper, we propose a model for multiple view unsupervised clustering based on Kohonen self-organizing-map algorithm. The results of simulations in two dimensional space using three views of training sets having different statistical properties have been presented. © 2012 Springer-Verlag Berlin Heidelberg.