PARAMETRIC SIGNAL RESTORATION USING ARTIFICIAL NEURAL NETWORKS

The problem of parametric signal restoration given its blurred/nonline arly distorted version contaminated by additive noise is discussed, It is postulated that feedforward artificial neural networks can be used to find a solution to this problem, The proposed estimator does not r equire iterative calculations that are normally performed using numeri cal methods for signal parameter estimation. Thus high speed is the ma in advantage of this approach, A two-stage neural network-based estima tor architecture is considered in which the vector of measurements is projected on the signal subspace and the resulting features form the i nput to a feedforward neural network, The effect of noise on the estim ator performance is analyzed and compared to the least-squares techniq ue. It is shown, for low and moderate noise levels, that the two estim ators are similar to each other in terms of their noise performance, p rovided the neural network approximates the inverse mapping from the m easurement space to the parameter space with a negligible error, Howev er, if the neural network is trained on noisy signal observations, the proposed technique is superior to the least-squares estimate (LSE) mo del fitting, Numerical examples are presented to support the analytica l results, Problems for future research are addressed.