2009
Conference article  Restricted

ICA by maximizing non-stability

Wang B., Kuruoglu E. E., Zhang J.

Non-stability  Pattern Recognition  Models. Statistical  Source separation  Impulsive signals  60F05 Central limit and other weak theorems  60G52 Stable processes  ICA  Alpha-stable negentropy  Probability and Statistics. Time series analysis 

We propose a new approach for ICA by maximizing the non-stability contrast function in this paper. This new version of ICA is motivated by the Generalized Central Limit Theorem (GCLT), an important extension of classical CLT. We demonstrate that the classical ICA based on maximization of non-Gaussianity is a special case of the new approach of ICA we introduce here which is based on maximization of non-Stability with certain constraints. To be able to quantify non-stability, we introduce a new measure of stability namely Alpha-stable negentropy. A numerical gradient ascent algorithm for the maximization of the alpha-stable negentropy with the objective of source separation is also introduced in this paper. Experiments show that ICA by maximum of non-stability performs very successfully in impulsive source separation problems.

Source: Independent Component Analysis and Signal Separation. 8th International Conference, pp. 179–186, Paraty, Rio Janerio, Brasile, 15-18 Marzo 2009


Metrics



Back to previous page
BibTeX entry
@inproceedings{oai:it.cnr:prodotti:44285,
	title = {ICA by maximizing non-stability},
	author = {Wang B. and Kuruoglu E.  E. and Zhang J.},
	doi = {10.1007/978-3-642-00599-2_23},
	booktitle = {Independent Component Analysis and Signal Separation. 8th International Conference, pp. 179–186, Paraty, Rio Janerio, Brasile, 15-18 Marzo 2009},
	year = {2009}
}