Blind Source Separation of Linear Mixtures with Singular Matrices
We consider the Blind Source Separation problem of linear mixtures with singular matrices and show that it can be solved if the sources are sufficiently sparse. More generally, we consider the problem of identifying the source matrix S ∈ IR nxN if a linear mixture X = AS is known only, where A∈ IR mxn , m ≤ n and the rank of A is less than m. A sufficient condition for solving this problem is that the level of sparsity of S is bigger than m–rank(A) in sense that the number of zeros in each column of S is bigger than m–rank(A). We present algorithms for such identification and illustrate them by examples.
KeywordsIndependent Component Analysis Independent Component Analysis Blind Source Separation Scaling Matrice Estimate Source
Unable to display preview. Download preview PDF.
- 3.Eriksson, J., Koivunen, V.: Identifiability and separability of linear ica models revisited. In: Proc. of ICA 2003, pp. 23–27 (2003)Google Scholar
- 4.Georgiev, P., Theis, F.J., Cichocki, A.: Blind source separation and sparse component analysis of overcomplete mixtures. In: Proc. of ICASSP 2004, Montreal, Canada (2004)Google Scholar
- 7.Theis, F.J., Lang, E.W., Puntonet, C.G.: A geometric algorithm for overcomplete linear ICA. Neurocomputing (2003) (in print)Google Scholar
- 8.Waheed, K., Salem, F.: Algebraic Overcomplete Independent Component Analysis. In: Proc. Int. Conf. ICA 2003, Nara, Japan, pp. 1077–1082 (2003)Google Scholar