Title :
Dimension reduction as a deflation method in ICA
Author :
Zhang, Kun ; Chan, Lai-Wan
Author_Institution :
Dept. of Comput. Sci. & Eng., Chinese Univ. of Hong Kong, China
Abstract :
In independent component analysis (ICA), when using the one-unit contrast function optimization approach to estimate independent components one by one, the constraint of uncorrelatedness between independent components prevents the algorithm from converging to the previously found components. A popular way to achieve uncorrelatedness is the Gram-Schmidt-like decorrelation scheme. In fact, uncorrelatedness between independent components can be achieved by reducing the degree of freedom in the unknown parameter set of the de-mixing matrix. In this letter, we propose to exploit the dimension-reduction technique to exactly enforce uncorrelatedness between difference independent components. The advantage of this method is that dimension reduction of the observations and de-mixing weight vectors makes the computation complexity lower and produces a faster convergence. Hence, our method results in a faster algorithm in computation of ICA.
Keywords :
convergence; decorrelation; independent component analysis; signal processing; Gram-Schmidt-like decorrelation scheme; ICA; convergence; demixing weight vector; dimension-reduction technique; independent component analysis; one-unit objective function; optimization approach; Closed-form solution; Constraint optimization; Convergence; Councils; Decorrelation; Entropy; Independent component analysis; Mutual information; Optimization methods; Signal processing algorithms; Decorrelation; Gram–Schmidt-like decorrelation; dimension-reduction; independent component analysis (ICA); one-unit objective function;
Journal_Title :
Signal Processing Letters, IEEE
DOI :
10.1109/LSP.2005.860541