Extracting noise contaminated information in multiple sources

Abstract

This article is focused on the problem of recognition of information which might be masked by other signal components, artefacts and noise or totally embedded in those, but which is shared by a set of inputs. It is not uncommon that even the completely imperceptible information is of high importance in very different contexts. We have developed a method of partial linear dependence, PLD to deal with this problem, improving available methods and techniques involved; it is especially useful for the analysis of acoustic and brain signals where it can be used to extend current concepts of connectivity, as well. This method is well applicable in other application domains, especially in the analysis of variety of biological signals.