Geodesic-based and projection-based neural blind deconvolution algorithms

Authors:
Simone Fiori
Affiliations:
Dipartimento di Elettronica, Intelligenza Artificiale e Telecomunicazioni (DEIT), Facoltí di Ingegneria, Universití Politecnica delle Marche, Via Brecce Bianche, Ancona I-60131, Italy
Venue:
Signal Processing
Year:
2008

Citing 12
Cited 3

Bayesian deconvolution of Bernoulli-Gaussian processes

Signal Processing
A stochastic algorithm for parametric and non-parametric estimation in the case of incomplete data

Signal Processing
An information-maximization approach to blind separation and blind deconvolution

Neural Computation
The Geometry of Algorithms with Orthogonality Constraints

SIAM Journal on Matrix Analysis and Applications
Blind Channel Equalization and Identification

Blind Channel Equalization and Identification
Generalized mutual information approach to multichannel blind deconvolution

Signal Processing
Editorial: Geometrical methods in neural networks and learning

Neurocomputing
Blind adaptation of stable discrete-time IIR filters in state-space form

IEEE Transactions on Signal Processing
Blind source separation-semiparametric statistical approach

IEEE Transactions on Signal Processing
Super-exponential methods for blind deconvolution

IEEE Transactions on Information Theory
Monotonic convergence of fixed-point algorithms for ICA

IEEE Transactions on Neural Networks
Fast fixed-point neural blind-deconvolution algorithm

IEEE Transactions on Neural Networks

On vector averaging over the unit hypersphere

Digital Signal Processing
On approximation of orientation distributions by means of spherical ridgelets

IEEE Transactions on Image Processing
Visualization of Riemannian-manifold-valued elements by multidimensional scaling

Neurocomputing

Quantified Score

Hi-index	0.08

Visualization

Abstract

The present contribution studies a geodesic-based and a projection-based learning algorithm over a curved parameter space for blind deconvolution (BD) application. The chosen deconvolving structure appears as a single neuron model whose learning rules naturally arise from criterion-function minimization over a smooth manifold. We consider the BD performances of the two classes of algorithms as well as their computational burden. Also, numerical comparisons with seven BD algorithms known from the scientific literature are illustrated and discussed.