Blind separation of sources from their linear mixtures is a well understood problem. However, if the mixtures are nonlinear, this problem becomes generally very difficult. This is because both the nonlinear mapping and the underlying sources must be learned from the data in a blind manner, and the problem is highly ill-posed without a suitable regularization. In our approach, multilayer perceptrons are used as nonlinear generative models for the data, and variational Bayesian (ensemble) learning is applied for finding the sources. The variational Bayesian technique automatically provides a reasonable regularization of the nonlinear blind separation problem. In this paper, we first consider a static nonlinear mixing model, with a successful application to real-world speech data compression. Then we discuss extraction of sources from nonlinear dynamic processes, and detection of abrupt changes in the process dynamics. In a difficult test problem with chaotic data, our approach clearly outperforms currently available nonlinear prediction and change detection techniques. The proposed methods are computationally demanding, but they can be applied to blind nonlinear problems of higher dimensions than other existing approaches.
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copy
Harri VALPOLA, Erkki OJA, Alexander ILIN, Antti HONKELA, Juha KARHUNEN, "Nonlinear Blind Source Separation by Variational Bayesian Learning" in IEICE TRANSACTIONS on Fundamentals,
vol. E86-A, no. 3, pp. 532-541, March 2003, doi: .
Abstract: Blind separation of sources from their linear mixtures is a well understood problem. However, if the mixtures are nonlinear, this problem becomes generally very difficult. This is because both the nonlinear mapping and the underlying sources must be learned from the data in a blind manner, and the problem is highly ill-posed without a suitable regularization. In our approach, multilayer perceptrons are used as nonlinear generative models for the data, and variational Bayesian (ensemble) learning is applied for finding the sources. The variational Bayesian technique automatically provides a reasonable regularization of the nonlinear blind separation problem. In this paper, we first consider a static nonlinear mixing model, with a successful application to real-world speech data compression. Then we discuss extraction of sources from nonlinear dynamic processes, and detection of abrupt changes in the process dynamics. In a difficult test problem with chaotic data, our approach clearly outperforms currently available nonlinear prediction and change detection techniques. The proposed methods are computationally demanding, but they can be applied to blind nonlinear problems of higher dimensions than other existing approaches.
URL: https://global.ieice.org/en_transactions/fundamentals/10.1587/e86-a_3_532/_p
Copy
@ARTICLE{e86-a_3_532,
author={Harri VALPOLA, Erkki OJA, Alexander ILIN, Antti HONKELA, Juha KARHUNEN, },
journal={IEICE TRANSACTIONS on Fundamentals},
title={Nonlinear Blind Source Separation by Variational Bayesian Learning},
year={2003},
volume={E86-A},
number={3},
pages={532-541},
abstract={Blind separation of sources from their linear mixtures is a well understood problem. However, if the mixtures are nonlinear, this problem becomes generally very difficult. This is because both the nonlinear mapping and the underlying sources must be learned from the data in a blind manner, and the problem is highly ill-posed without a suitable regularization. In our approach, multilayer perceptrons are used as nonlinear generative models for the data, and variational Bayesian (ensemble) learning is applied for finding the sources. The variational Bayesian technique automatically provides a reasonable regularization of the nonlinear blind separation problem. In this paper, we first consider a static nonlinear mixing model, with a successful application to real-world speech data compression. Then we discuss extraction of sources from nonlinear dynamic processes, and detection of abrupt changes in the process dynamics. In a difficult test problem with chaotic data, our approach clearly outperforms currently available nonlinear prediction and change detection techniques. The proposed methods are computationally demanding, but they can be applied to blind nonlinear problems of higher dimensions than other existing approaches.},
keywords={},
doi={},
ISSN={},
month={March},}
Copy
TY - JOUR
TI - Nonlinear Blind Source Separation by Variational Bayesian Learning
T2 - IEICE TRANSACTIONS on Fundamentals
SP - 532
EP - 541
AU - Harri VALPOLA
AU - Erkki OJA
AU - Alexander ILIN
AU - Antti HONKELA
AU - Juha KARHUNEN
PY - 2003
DO -
JO - IEICE TRANSACTIONS on Fundamentals
SN -
VL - E86-A
IS - 3
JA - IEICE TRANSACTIONS on Fundamentals
Y1 - March 2003
AB - Blind separation of sources from their linear mixtures is a well understood problem. However, if the mixtures are nonlinear, this problem becomes generally very difficult. This is because both the nonlinear mapping and the underlying sources must be learned from the data in a blind manner, and the problem is highly ill-posed without a suitable regularization. In our approach, multilayer perceptrons are used as nonlinear generative models for the data, and variational Bayesian (ensemble) learning is applied for finding the sources. The variational Bayesian technique automatically provides a reasonable regularization of the nonlinear blind separation problem. In this paper, we first consider a static nonlinear mixing model, with a successful application to real-world speech data compression. Then we discuss extraction of sources from nonlinear dynamic processes, and detection of abrupt changes in the process dynamics. In a difficult test problem with chaotic data, our approach clearly outperforms currently available nonlinear prediction and change detection techniques. The proposed methods are computationally demanding, but they can be applied to blind nonlinear problems of higher dimensions than other existing approaches.
ER -