Blind Source Separation and Independent Component Analysis
Thursday 10 July 2008, 1130-1230
The seminar will present theoretical developments in the area of blind source separation (BSS), also known as independent component analysis (ICA).
BSS and ICA address those situations where mixtures of a set of unknown source signals are available and the problem is to extract the original source signals from these observed mixtures without the knowledge of the mixing matrix.
In the case of linear, instantaneous mixtures, the received signals, y, can be written as y = A*x, where x represents the unknown original source signals and A is the unknown mixing matrix. It will be demonstrated that independent source signals can be extracted using higher-order statistics (HOS).
BSS and ICA have many potential applications in diverse fields. The seminar will present theoretical developments as well as examples of simulation, demonstration, and real-world examples.
Prof. Asoke K Nandi
David Jardine Chair of Signal Processing
Department of Electrical Engineering & Electronics
The University of Liverpool
Brownlow Hill, Liverpool, L69 3GJ, UK