Neural networks for independent range and depth discrimination in passive acoustic localization

Two feedforward neural networks (NNs) with one hidden layer were trained using a fast backpropagation algorithm to determine the position of an acoustic source in a waveguide. One network was trained to localize the source in depth, and the other was trained independently to localize it in range. The output layer consisted of one unit for each possible range or depth of the source. The networks were trained with a signal-to-noise ratio (S/N) of 50 dB and tested with patterns generated with S/N ranging from 0 to 20 dB. The performance of the NNs was compared with that of a nearest-neighbor classifier in the context of an estimation problem. The NNs were less resistant to noise than the conventional processor, but were faster. It is explained why multilayered feedforward NNs cannot achieve the performances of optimum classifiers