Abstract
Mixing matrix estimation in instantaneous blind source separation (BSS) can be performed by exploiting the sparsity and disjoint orthogonality of source signals. As a result, approaches for estimating the unknown mixing process typically employ clustering algorithms on the mixtures in a parametric domain, where the signals can be sparsely represented. In this paper, we propose two algorithms to perform discriminative clustering of the mixture signals for estimating the mixing matrix. For the case of overdetermined BSS, we develop an algorithm to perform linear discriminant analysis based on similarity measures and combine it with K-hyperline clustering. Furthermore, we propose to perform discriminative clustering in a high-dimensional feature space obtained by an implicit mapping, using the kernel trick, for the case of underdetermined source separation. Using simulations on synthetic data, we demonstrate the improvements in mixing matrix estimation performance obtained using the proposed algorithms in comparison to other clustering methods. Finally we perform mixing matrix estimation from speech mixtures, by clustering single source points in the time-frequency domain, and show that the proposed algorithms achieve higher signal to interference ratio when compared to other baseline algorithms.
| Original language | English (US) |
|---|---|
| Pages (from-to) | 9-18 |
| Number of pages | 10 |
| Journal | Digital Signal Processing: A Review Journal |
| Volume | 23 |
| Issue number | 1 |
| DOIs | |
| State | Published - Jan 2013 |
Fingerprint
Keywords
- Discriminative clustering
- Instantaneous blind source separation
- K-hyperline clustering
- Kernel trick
- Mixing matrix estimation
ASJC Scopus subject areas
- Signal Processing
- Electrical and Electronic Engineering
Cite this
Mixing matrix estimation using discriminative clustering for blind source separation. / J. Thiagarajan, Jayaraman; Natesan Ramamurthy, Karthikeyan; Spanias, Andreas.
In: Digital Signal Processing: A Review Journal, Vol. 23, No. 1, 01.2013, p. 9-18.Research output: Contribution to journal › Article
}
TY - JOUR
T1 - Mixing matrix estimation using discriminative clustering for blind source separation
AU - J. Thiagarajan, Jayaraman
AU - Natesan Ramamurthy, Karthikeyan
AU - Spanias, Andreas
PY - 2013/1
Y1 - 2013/1
N2 - Mixing matrix estimation in instantaneous blind source separation (BSS) can be performed by exploiting the sparsity and disjoint orthogonality of source signals. As a result, approaches for estimating the unknown mixing process typically employ clustering algorithms on the mixtures in a parametric domain, where the signals can be sparsely represented. In this paper, we propose two algorithms to perform discriminative clustering of the mixture signals for estimating the mixing matrix. For the case of overdetermined BSS, we develop an algorithm to perform linear discriminant analysis based on similarity measures and combine it with K-hyperline clustering. Furthermore, we propose to perform discriminative clustering in a high-dimensional feature space obtained by an implicit mapping, using the kernel trick, for the case of underdetermined source separation. Using simulations on synthetic data, we demonstrate the improvements in mixing matrix estimation performance obtained using the proposed algorithms in comparison to other clustering methods. Finally we perform mixing matrix estimation from speech mixtures, by clustering single source points in the time-frequency domain, and show that the proposed algorithms achieve higher signal to interference ratio when compared to other baseline algorithms.
AB - Mixing matrix estimation in instantaneous blind source separation (BSS) can be performed by exploiting the sparsity and disjoint orthogonality of source signals. As a result, approaches for estimating the unknown mixing process typically employ clustering algorithms on the mixtures in a parametric domain, where the signals can be sparsely represented. In this paper, we propose two algorithms to perform discriminative clustering of the mixture signals for estimating the mixing matrix. For the case of overdetermined BSS, we develop an algorithm to perform linear discriminant analysis based on similarity measures and combine it with K-hyperline clustering. Furthermore, we propose to perform discriminative clustering in a high-dimensional feature space obtained by an implicit mapping, using the kernel trick, for the case of underdetermined source separation. Using simulations on synthetic data, we demonstrate the improvements in mixing matrix estimation performance obtained using the proposed algorithms in comparison to other clustering methods. Finally we perform mixing matrix estimation from speech mixtures, by clustering single source points in the time-frequency domain, and show that the proposed algorithms achieve higher signal to interference ratio when compared to other baseline algorithms.
KW - Discriminative clustering
KW - Instantaneous blind source separation
KW - K-hyperline clustering
KW - Kernel trick
KW - Mixing matrix estimation
UR - http://www.scopus.com/inward/record.url?scp=84869504065&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84869504065&partnerID=8YFLogxK
U2 - 10.1016/j.dsp.2012.08.002
DO - 10.1016/j.dsp.2012.08.002
M3 - Article
AN - SCOPUS:84869504065
VL - 23
SP - 9
EP - 18
JO - Digital Signal Processing: A Review Journal
JF - Digital Signal Processing: A Review Journal
SN - 1051-2004
IS - 1
ER -