Neural networks for blind separation with unknown number of sources

Cichocki, Andrzej; Karhunen, Juha; Kasprzak, Włodzimierz; Vigário, Ricardo

doi:10.1016/s0925-2312(98)00091-5

Cited by 86 publications

(60 citation statements)

References 38 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Cichocki et al claimed that the enhanced nonlinear PCA with a whitening process was able to extract less number of ICs than the sources [20]. However, the necessary preprocessing (whitening) stage results in failure of separation due to data distortion when ill-conditioned mixing matrices or weak sources are involved.…”

Section: Less-complete Icamentioning

confidence: 99%

“…Also, the gradient of (18) is obtained (19) where the terms containing the vector and the matrix are as given in (12) and the additional term with elements, . The Hessian matrix is approximated as Vec (20) where is a diagonal matrix having diagonal elements given by the vector with elements . By applying the relational properties same as in Section III, the Newton-like learning in (5) is then obtained.…”

Section: Ica-rmentioning

confidence: 99%

“…The nonlinear PCA algorithm [20] and fastICA algorithm [16] provided possible ways to extract the subspace in parallel manner. However, the extraction of signals at the global optimum by using these algorithms is always determined by the contrast function adopted.…”

mentioning

confidence: 99%

See 2 more Smart Citations

Approach and Applications of Constrained ICA

Lü

Rajapakse

2005

IEEE Trans. Neural Netw.

269

240

View full text Add to dashboard Cite

Abstract-This paper presents the technique of constrained independent component analysis (cICA) and demonstrates two applications, less-complete ICA, and ICA with reference (ICA-R). The cICA is proposed as a general framework to incorporate additional requirements and prior information in the form of constraints into the ICA contrast function. The adaptive solutions using the Newton-like learning are proposed to solve the constrained optimization problem. The applications illustrate the versatility of the cICA by separating subspaces of independent components according to density types and extracting a set of desired sources when rough templates are available. The experiments using face images and functional MR images demonstrate the usage and efficacy of the cICA.Index Terms-Constrained optimization, ICA with reference (ICA-R), independent component analysis (ICA), less-complete ICA.

show abstract

Section: Less-complete Icamentioning

confidence: 99%

Section: Ica-rmentioning

confidence: 99%

See 1 more Smart Citation

Approach and Applications of Constrained ICA

Lü

Rajapakse

2005

IEEE Trans. Neural Netw.

269

240

View full text Add to dashboard Cite

show abstract

“…When the number of the components in the observed signal is larger than that of ICs (n ¿ m), the learning rule becomes much complex as the pseudo-inverse of the matrix is involved [30]. When the number of the observed signal components is less than that of the essential ICs (n ¡ m), the ICs are not fully identiÿable in most cases because A is non-invertible [27,28]; the known information to the system is inadequate to represent the full space of the ICs [13]. For simplicity and without loosing generality in this manuscript, we consider the case of the complete ICA where n = m and, ideally, the demixing matrix W = A −1 .…”

Section: Introductionmentioning

confidence: 99%

Eliminating indeterminacy in ICA

Lü

Rajapakse

2003

Neurocomputing

View full text Add to dashboard Cite

“…As a special case, separation of instantaneous mixtures is very successful so far and many approaches have been proposed [1]- [5]. However, a more challenging situation is the separation of convolutive mixtures with long mixing channels [6]- [10].…”

mentioning

confidence: 99%

Convolutive Blind Source Separation Based on Disjointness Maximization of Subband Signals

Mei

Mertins

2008

IEEE Signal Process. Lett.

View full text Add to dashboard Cite

Abstract-The concept of disjoint component analysis (DCA) is based on the fact that different speech or audio signals are typically more disjoint than mixtures of them. This letter studies the problem of blind separation of convolutive mixtures through the subband-wise maximization of the disjointness of time-frequency representations of the signals. In our approach, we first define a frequency-dependent measure representing the closeness to disjointness of a group of subband signals. Then, this frequency-dependent measure is integrated to form an objective function that only depends on the time-domain parameters of the separation system. Lastly, an efficient natural-gradient-based learning rule is developed for the update of the separation-system coefficients.Index Terms-Convolutive blind source separation, disjointness maximization.

show abstract

Neural networks for blind separation with unknown number of sources

Cited by 86 publications

References 38 publications

Approach and Applications of Constrained ICA

Approach and Applications of Constrained ICA

Eliminating indeterminacy in ICA

Convolutive Blind Source Separation Based on Disjointness Maximization of Subband Signals

Contact Info

Product

Resources

About