Newton-Like Methods for Nonparametric Independent Component Analysis
The performance of ICA algorithms significantly depends on the choice of the contrast function and the optimisation algorithm used in obtaining the demixing matrix. In this paper we focus on the standard linear nonparametric ICA problem from an optimisation point of view. It is well known that after a pre-whitening process, the problem can be solved via an optimisation approach on a suitable manifold. We propose an approximate Newton’s method on the unit sphere to solve the one-unit linear nonparametric ICA problem. The local convergence properties are discussed. The performance of the proposed algorithms is investigated by numerical experiments.
KeywordsIndependent Component Analysis Independent Component Analysis Blind Source Separation Contrast Function Independent Component Analysis Algorithm
Unable to display preview. Download preview PDF.
- 6.Hüper, K., Shen, H., Seghouane, A.-K.: Local convergence properties of Fast- ICA and some generalisations. In: Proceedings of the 31st IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP 2006), pp. V1009–V1012 (2006)Google Scholar
- 7.Shen, H., Hüper, K.: Newton-like methods for parallel independent component analysis. To appear in: LSP 2006, Maynooth, Ireland, September 6-8 (2006)Google Scholar
- 9.Shen, H., Hüper, K., Seghouane, A.-K.: Geometric optimisation and FastICA algorithms. To appear in: MTNS 2006, Kyoto, Japan, July 24-28 (2006)Google Scholar
- 10.Smith, S.: Optimization techniques on riemannian manifolds. Hamiltonian and gradientflows, algorithms and control, Fields Institute Communications 3, 113–136 (1994)Google Scholar
- 11.Cardoso, J.F.: Blind source separation: statistical principles. Proceedings of the IEEE (90), 2099–2026 (1998)Google Scholar
- 13.Amari, S., Cichocki, A., Yang, H.H.: A new learning algorithm for blind signal separation. In: Advances in Neural Information Processing Systems, vol. 8, pp. 757–763 (1996)Google Scholar