Genetic algorithm approach to nonlinear adaptive filtering

The genetic algorithm (GA) which exploits the feedforward neural networks for the concise representation of nonlinear mappings has been developed to search for the suboptimal nonlinear filters. The GA does not introduce the error caused by approximation of the unit step function by the continuous one, as is the case when the least mean absolute (LMA) neural filtering algorithms are used. The developed GA approach enables also the search for minimal filter structure