Blind source recovery: a framework in the state space

Authors:
Khurram Waheed;Fathi M. Salem
Affiliations:
Department of Electrical and Computer Engineering, San Diego State University, San Diego, CA;Department of Electrical and Computer Engineering, Michigan State University, East Lansing, MI
Venue:
The Journal of Machine Learning Research
Year:
2003

Citing 11
Cited 1

Space or time adaptive signal processing by neural network models

AIP Conference Proceedings 151 on Neural Networks for Computing
An information-maximization approach to blind separation and blind deconvolution

Neural Computation
Adaptive filter theory (3rd ed.)

Adaptive filter theory (3rd ed.)
A fast fixed-point algorithm for independent component analysis

Neural Computation
Stability analysis of learning algorithms for blind source separation

Neural Networks
Natural gradient works efficiently in learning

Neural Computation
Advances in Independent Component Analysis

Advances in Independent Component Analysis
Blind Separation of Multiple Speakers in a Multipath Environment

ICASSP '97 Proceedings of the 1997 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP '97) -Volume 1 - Volume 1
Adaptive Blind Signal and Image Processing: Learning Algorithms and Applications

Adaptive Blind Signal and Image Processing: Learning Algorithms and Applications
Equivariant adaptive source separation

IEEE Transactions on Signal Processing
Blind source separation-semiparametric statistical approach

IEEE Transactions on Signal Processing

Linear State Space Feedforward and Feedback Structures for Blind Source Recovery in Dynamic Environments

Neural Processing Letters

Quantified Score

Hi-index	0.00

Visualization

Abstract

Blind Source Recovery (BSR) denotes recovery of originalsources/signals from environments that may include convolution,temporal variation, and even nonlinearity. It also infers therecovery of sources even in the absence of precise environmentidentifiability. This paper describes, in a comprehensive fashion,a generalized BSR formulation achieved by the application ofstochastic optimization principles to the Kullback-Lieblerdivergence as a performance functional subject to the constraintsof the general (i.e., nonlinear and time-varying) state spacerepresentation. This technique is used to derive update laws fornonlinear time-varying dynamical systems, which are subsequentlyspecialized to time-invariant and linear systems. Further, thestate space demixing network structures have been exploited todevelop learning rules, capable of handling most filteringparadigms, which can be conveniently extended to nonlinear models.In the special cases, distinct linear state-space algorithms arepresented for the minimum phase and non-minimum phase mixingenvironment models. Conventional (FIR/IIR) filtering models aresubsequently derived from this general structure and are comparedwith material in the recent literature. Illustrative simulationexamples are presented to demonstrate the online adaptationcapabilities of the developed algorithms. Some of this reportedwork has also been implemented in dedicated hardware/softwareplatforms.