Skip to main content
European Commission logo print header

Blind Source Separation and Applications

Ziel

OBJECTIVES
The subject of this project is Blind Source Separation (BSS) and Independent Component Analysis (ICA).
The main objectives of the project are to achieve significant advances in:
1. Development of theory and algorithms for linear ICA (in particular instantaneous and convolutive ICA, noisy and adaptive ICA), for non-linear ICA, and for separation of non-independent signals.
2. Two main application areas: biomedical signals and acoustic mixtures

DESCRIPTION OF WORK
The main item of the work to be performed are:
1. Theory and algorithms
- Analysis of open issues in linear ICA.
- Development of better techniques for noisy and/or adaptive ICA
- Development of theory and algorithms for nonlinear ICA.
- Development of theory and algorithms for separation of non-independent signals.
In all these aspects, theoretical issues will be addressed, side by side with the development of algorithms.
2. Applications
- Applications of these techniques and algorithms to biomedical signals (especially MEG and MNG) for artefact extraction and elimination and for a better interpretation of these signals.
- Application for decomposing evoked fields in biomedical signals, enabling direct access to the underlying brain functioning.
- Application to the separation of acoustic mixtures, first in well controlled situations and progressively proceeding to less controlled ones (actual persons as speaker, noise, moving sources and microphones).
- Application of the separation of acoustic mixtures to the development of a new generation of hearing aids with a much better handling of noisy environments, taking into account the technological constraints for implementing the processing within a wearable device.

Aufforderung zur Vorschlagseinreichung

Data not available

Koordinator

INSTITUT NATIONAL POLYTECHNIQUE DE GRENOBLE
EU-Beitrag
Keine Daten
Adresse
46 AVENUE FELIX VIALLET
38031 GRENOBLE CEDEX 1
Frankreich

Auf der Karte ansehen

Gesamtkosten
Keine Daten

Beteiligte (3)