Nonlinear Independent Component Analysis Using Ensemble Learning: Experiments and Discussion

Harri Valpola¹, Xavier Giannakopoulos², Antti Honkela¹ and Juha Karhunen¹

¹Helsinki University of Technology, Neural Networks Research Centre
P.O.Box 5400, FIN-02015 HUT, Espoo, Finland
²IDSIA, Galleria 2, CH-6928 Manno, Switzerland
E-mail: Harri.Valpola@hut.fi, Xavier@idsia.ch, Antti.Honkela@hut.fi,
Juha.Karhunen@hut.fi URL: http://www.cis.hut.fi/

Abstract:

In this paper, we present experimental results on a nonlinear independent component analysis approach based on Bayesian ensemble learning. The theory of the method is discussed in a companion paper. Simulations with artificial and natural data demonstrate the feasibility and good performance of the proposed approach. We also discuss the relationships of the method to other existing methods.