A COMMON NEURAL-NETWORK MODEL FOR UNSUPERVISED EXPLORATORY DATA-ANALYSIS AND INDEPENDENT COMPONENT ANALYSIS

This paper presents the derivation of an unsupervised learning algorit hm, which enables the identification and visualization of latent struc ture within ensembles of high-dimensional data. This provides a linear projection of the data onto a lower dimensional subspace to identify the characteristic structure of the observations independent latent ca uses. The algorithm is shown to be a very promising tool for unsupervi sed exploratory data analysis and data visualization. Experimental res ults confirm the attractiveness of this technique for exploratory data analysis and an empirical comparison is made with the recently propos ed generative topographic mapping (GTM) and standard principal compone nt analysis (PCA), Based on standard probability density models a gene ric nonlinearity is developed which allows both 1) identification and visualization of dichotomised clusters inherent in the observed data a nd 2) separation of sources with arbitrary distributions from mixtures , whose dimensionality may be greater than that of number of sources. The resulting algorithm is therefore also a generalized neural approac h to independent component analysis (ICA) and it is considered to be a promising method for analysis of real-world data that will consist of sub- and super-Gaussian components such as biomedical signals.