REGULARIZED GRADIENT ALGORITHM FOR NON-NEGATIVE INDEPENDENT COMPONENT ANALYSIS

Signal Separation

Presented by: Wendyam Serge Boris Ouedraogo, Author(s): Wendyam Serge Boris Ouedraogo, Commissariat à l'Energie Atomique et aux Energies Alternatives, France; Meriem Jaidane, Ecole Nationale d'Ingénieurs de Tunis, Tunisia; Antoine Souloumiac, Commissariat à l'Energie Atomique et aux Energies Alternatives, France; Christian Jutten, Université Joseph Fourier / Grenoble et Institut Universitaire de France, France

Independent Component Analysis (ICA) is a well-known technique for solving blind source separation (BSS) problem. However "classical" ICA algorithms seem not suited for non-negative sources. This paper proposes a gradient descent approach for solving the Non-Negative Independent Component Analysis problem (NNICA). NNICA original separation criterion contains the discontinuous sign function whose minimization may lead to ill convergence (local minima) especially for sparse sources. Replacing the discontinuous function by a continuous one tanh, we propose a more accurate regularized Gradient algorithm called "Exact" Regularized Gradient (ERG) for NNICA. Experiments on synthetic data with different sparsity degrees illustrate the efficiency of the proposedmethod and a comparison shows that the proposed ERG outperforms existing methods.

You need the Flash Player.

Share:

Download subtitles | Enlarge video

Recorded:	2011-05-27 10:10 - 10:30, Club E
Added:	20. 6. 2011 00:26
Number of views:	28
Video resolution:	1024x576 px, 512x288 px
Video length:	0:18:00
Audio track:	MP3 [6.08 MB], 0:18:00