Blind signal separation through cooperating ANNs

Authors:
Francisco Bellas;Richard J. Duro;Fernando López-Peña
Affiliations:
Grupo Integrado de Ingeniería, Universidade da Coruña;Grupo Integrado de Ingeniería, Universidade da Coruña;Grupo Integrado de Ingeniería, Universidade da Coruña
Venue:
KES'05 Proceedings of the 9th international conference on Knowledge-Based Intelligent Information and Engineering Systems - Volume Part I
Year:
2005

Citing 7
Cited 0

Genetic algorithms with sharing for multimodal function optimization

Proceedings of the Second International Conference on Genetic Algorithms on Genetic algorithms and their application
Blind separation of sources, Part 1: an adaptive algorithm based on neuromimetic architecture

Signal Processing
Combining evolution with credit apportionment: a new learning algorithm for neural nets

Neural Networks
Evolutionary Algorithms for Solving Multi-Objective Problems

Evolutionary Algorithms for Solving Multi-Objective Problems
Forming neural networks through efficient and adaptive coevolution

Evolutionary Computation
A family of frequency- and time-domain contrasts for blind separation of convolutive mixtures of temporally dependent signals

IEEE Transactions on Signal Processing
Speciation as automatic categorical modularization

IEEE Transactions on Evolutionary Computation

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper is devoted to proposing and testing a strategy for decomposing compound signals obtained in remote sensing applications through the automatic generation of cooperating ANNs that model it. Each ANN will specialize in one of the primitives that make up the whole. The evolutionary based algorithm that is proposed for this purpose implies that the combination of networks takes place at a phenotypic operational level, this is, the architecture of the networks is not the issue, but rather function they implement. This way, a population of networks that are automatically classified into different species depending on the performance of their phenotype, and individuals of each species cooperate forming a group to obtain a complex output, in this case the signal that is required. The magnitude that reflects the difference between ANNs is their affinity vector, which must be automatically created and modified depending on the actuation of the phenotype of each individual. The main objective of this approach is to model complex functions such as multidimensional signals, which are typical of remote sensing application, providing a decomposition of them into primitive functions.