Romanian Reports in Physics vol:55 issue:1 pages:43-67
Abstract. The paper is an overview of the most frequently used neural network algorithms for implementing Independent Component Analysis (ICA). The performance of six structurally different algorithms was ranked in blind separation of independent artificially generated signals using the stationary linear ICA model. Ranking of the estimated components was also carried out and compared among different ICA approaches. All algorithms were run with different contrast functions, which were optimally selected on the basis of maximizing the sum of individual negentropies of the network outputs or minimizing their mutual information. Both subgaussian and supergaussian onedimensional time series were employed throughout the numerical simulations.